The whole DevOps platform: projects, merge requests, CI, security, packages, and much more. 1,429 tables: the stress test. Far past the explorer's comfort zone, kept in on purpose. Domains here are by table-name prefix, not by judgment.
1429 tables · 14379 columns · 1861 foreign keys · 111 domains | 39 errors · 161 warnings · 1668 notes
Generated from gitlabhq@v19.3.1.sql on 2026-09-05.
Conventions are claims. Findings below are where the schema breaks them.
agent_organization_authorizations looks like a many-to-many link table but has no unique constraint across (agent_id, organization_id). The same pair can be inserted twice; every join through it will double-count.
Fix Use the FK pair as the primary key (or add a UNIQUE).
ALTER TABLE agent_organization_authorizations ADD PRIMARY KEY (agent_id, organization_id);
namespace_id, value_stream_idjunction-uniquenessanalytics_cycle_analytics_value_stream_settings looks like a many-to-many link table but has no unique constraint across (namespace_id, value_stream_id). The same pair can be inserted twice; every join through it will double-count.
Fix Use the FK pair as the primary key (or add a UNIQUE).
ALTER TABLE analytics_cycle_analytics_value_stream_settings ADD PRIMARY KEY (namespace_id, value_stream_id);
namespace_storage_forks_cost_factor is binary floating point. 0.1 + 0.2 ≠ 0.3; sums drift; reconciliation against the ledger will be off by cents.
Fix Use NUMERIC(p, s) or integer minor units.
ALTER TABLE application_settings ALTER COLUMN namespace_storage_forks_cost_factor TYPE numeric(14,2);
sentry_clientside_traces_sample_rate is binary floating point. 0.1 + 0.2 ≠ 0.3; sums drift; reconciliation against the ledger will be off by cents.
Fix Use NUMERIC(p, s) or integer minor units.
ALTER TABLE application_settings ALTER COLUMN sentry_clientside_traces_sample_rate TYPE numeric(14,2);
external_audit_event_destination_id, namespace_idjunction-uniquenessaudit_events_streaming_http_group_namespace_filters looks like a many-to-many link table but has no unique constraint across (external_audit_event_destination_id, namespace_id). The same pair can be inserted twice; every join through it will double-count.
Fix Use the FK pair as the primary key (or add a UNIQUE).
ALTER TABLE audit_events_streaming_http_group_namespace_filters ADD PRIMARY KEY (external_audit_event_destination_id, namespace_id);
audit_events_instance_external_audit_event_destination_id, namespace_idjunction-uniquenessaudit_events_streaming_http_instance_namespace_filters looks like a many-to-many link table but has no unique constraint across (audit_events_instance_external_audit_event_destination_id, namespace_id). The same pair can be inserted twice; every join through it will double-count.
Fix Use the FK pair as the primary key (or add a UNIQUE).
ALTER TABLE audit_events_streaming_http_instance_namespace_filters ADD PRIMARY KEY (audit_events_instance_external_audit_event_destination_id, namespace_id);
public_projects_minutes_cost_factor is binary floating point. 0.1 + 0.2 ≠ 0.3; sums drift; reconciliation against the ledger will be off by cents.
Fix Use NUMERIC(p, s) or integer minor units.
ALTER TABLE ci_runners ALTER COLUMN public_projects_minutes_cost_factor TYPE numeric(14,2);
private_projects_minutes_cost_factor is binary floating point. 0.1 + 0.2 ≠ 0.3; sums drift; reconciliation against the ledger will be off by cents.
Fix Use NUMERIC(p, s) or integer minor units.
ALTER TABLE ci_runners ALTER COLUMN private_projects_minutes_cost_factor TYPE numeric(14,2);
cluster_projects looks like a many-to-many link table but has no unique constraint across (cluster_id, project_id). The same pair can be inserted twice; every join through it will double-count.
Fix Use the FK pair as the primary key (or add a UNIQUE).
ALTER TABLE cluster_projects ADD PRIMARY KEY (cluster_id, project_id);
csv_issue_imports looks like a many-to-many link table but has no unique constraint across (project_id, user_id). The same pair can be inserted twice; every join through it will double-count.
Fix Use the FK pair as the primary key (or add a UNIQUE).
ALTER TABLE csv_issue_imports ADD PRIMARY KEY (project_id, user_id);
duo_workflows_workloads looks like a many-to-many link table but has no unique constraint across (project_id, workflow_id). The same pair can be inserted twice; every join through it will double-count.
Fix Use the FK pair as the primary key (or add a UNIQUE).
ALTER TABLE duo_workflows_workloads ADD PRIMARY KEY (project_id, workflow_id);
group_deletion_schedules looks like a many-to-many link table but has no unique constraint across (group_id, user_id). The same pair can be inserted twice; every join through it will double-count.
Fix Use the FK pair as the primary key (or add a UNIQUE).
ALTER TABLE group_deletion_schedules ADD PRIMARY KEY (group_id, user_id);
public_projects_minutes_cost_factor is binary floating point. 0.1 + 0.2 ≠ 0.3; sums drift; reconciliation against the ledger will be off by cents.
Fix Use NUMERIC(p, s) or integer minor units.
ALTER TABLE group_type_ci_runners ALTER COLUMN public_projects_minutes_cost_factor TYPE numeric(14,2);
private_projects_minutes_cost_factor is binary floating point. 0.1 + 0.2 ≠ 0.3; sums drift; reconciliation against the ledger will be off by cents.
Fix Use NUMERIC(p, s) or integer minor units.
ALTER TABLE group_type_ci_runners ALTER COLUMN private_projects_minutes_cost_factor TYPE numeric(14,2);
insights looks like a many-to-many link table but has no unique constraint across (namespace_id, project_id). The same pair can be inserted twice; every join through it will double-count.
Fix Use the FK pair as the primary key (or add a UNIQUE).
ALTER TABLE insights ADD PRIMARY KEY (namespace_id, project_id);
public_projects_minutes_cost_factor is binary floating point. 0.1 + 0.2 ≠ 0.3; sums drift; reconciliation against the ledger will be off by cents.
Fix Use NUMERIC(p, s) or integer minor units.
ALTER TABLE instance_type_ci_runners ALTER COLUMN public_projects_minutes_cost_factor TYPE numeric(14,2);
private_projects_minutes_cost_factor is binary floating point. 0.1 + 0.2 ≠ 0.3; sums drift; reconciliation against the ledger will be off by cents.
Fix Use NUMERIC(p, s) or integer minor units.
ALTER TABLE instance_type_ci_runners ALTER COLUMN private_projects_minutes_cost_factor TYPE numeric(14,2);
lfs_file_locks looks like a many-to-many link table but has no unique constraint across (project_id, user_id). The same pair can be inserted twice; every join through it will double-count.
Fix Use the FK pair as the primary key (or add a UNIQUE).
ALTER TABLE lfs_file_locks ADD PRIMARY KEY (project_id, user_id);
merge_request_merge_schedules looks like a many-to-many link table but has no unique constraint across (merge_request_id, project_id). The same pair can be inserted twice; every join through it will double-count.
Fix Use the FK pair as the primary key (or add a UNIQUE).
ALTER TABLE merge_request_merge_schedules ADD PRIMARY KEY (merge_request_id, project_id);
namespace_import_users looks like a many-to-many link table but has no unique constraint across (namespace_id, user_id). The same pair can be inserted twice; every join through it will double-count.
Fix Use the FK pair as the primary key (or add a UNIQUE).
ALTER TABLE namespace_import_users ADD PRIMARY KEY (namespace_id, user_id);
namespace_secret_counts looks like a many-to-many link table but has no unique constraint across (namespace_id, root_namespace_id). The same pair can be inserted twice; every join through it will double-count.
Fix Use the FK pair as the primary key (or add a UNIQUE).
ALTER TABLE namespace_secret_counts ADD PRIMARY KEY (namespace_id, root_namespace_id);
oauth_openid_requests looks like a many-to-many link table but has no unique constraint across (access_grant_id, organization_id). The same pair can be inserted twice; every join through it will double-count.
Fix Use the FK pair as the primary key (or add a UNIQUE).
ALTER TABLE oauth_openid_requests ADD PRIMARY KEY (access_grant_id, organization_id);
observability_traces_issues_connections looks like a many-to-many link table but has no unique constraint across (issue_id, project_id). The same pair can be inserted twice; every join through it will double-count.
Fix Use the FK pair as the primary key (or add a UNIQUE).
ALTER TABLE observability_traces_issues_connections ADD PRIMARY KEY (issue_id, project_id);
p_ci_job_definition_instances looks like a many-to-many link table but has no unique constraint across (job_definition_id, job_id, partition_id). The same pair can be inserted twice; every join through it will double-count.
Fix Use the FK pair as the primary key (or add a UNIQUE).
ALTER TABLE p_ci_job_definition_instances ADD PRIMARY KEY (job_definition_id, job_id, partition_id);
p_generated_ref_commits looks like a many-to-many link table but has no unique constraint across (merge_request_iid, project_id). The same pair can be inserted twice; every join through it will double-count.
Fix Use the FK pair as the primary key (or add a UNIQUE).
ALTER TABLE p_generated_ref_commits ADD PRIMARY KEY (merge_request_iid, project_id);
packages_pypi_file_metadata looks like a many-to-many link table but has no unique constraint across (package_file_id, project_id). The same pair can be inserted twice; every join through it will double-count.
Fix Use the FK pair as the primary key (or add a UNIQUE).
ALTER TABLE packages_pypi_file_metadata ADD PRIMARY KEY (package_file_id, project_id);
organization_id, personal_access_token_idjunction-uniquenesspersonal_access_token_last_used_ips looks like a many-to-many link table but has no unique constraint across (organization_id, personal_access_token_id). The same pair can be inserted twice; every join through it will double-count.
Fix Use the FK pair as the primary key (or add a UNIQUE).
ALTER TABLE personal_access_token_last_used_ips ADD PRIMARY KEY (organization_id, personal_access_token_id);
public_projects_minutes_cost_factor is binary floating point. 0.1 + 0.2 ≠ 0.3; sums drift; reconciliation against the ledger will be off by cents.
Fix Use NUMERIC(p, s) or integer minor units.
ALTER TABLE project_type_ci_runners ALTER COLUMN public_projects_minutes_cost_factor TYPE numeric(14,2);
private_projects_minutes_cost_factor is binary floating point. 0.1 + 0.2 ≠ 0.3; sums drift; reconciliation against the ledger will be off by cents.
Fix Use NUMERIC(p, s) or integer minor units.
ALTER TABLE project_type_ci_runners ALTER COLUMN private_projects_minutes_cost_factor TYPE numeric(14,2);
projects_branch_rules_squash_options looks like a many-to-many link table but has no unique constraint across (project_id, protected_branch_id). The same pair can be inserted twice; every join through it will double-count.
Fix Use the FK pair as the primary key (or add a UNIQUE).
ALTER TABLE projects_branch_rules_squash_options ADD PRIMARY KEY (project_id, protected_branch_id);
security_policy_schedule_pipelines looks like a many-to-many link table but has no unique constraint across (project_id, security_policy_id). The same pair can be inserted twice; every join through it will double-count.
Fix Use the FK pair as the primary key (or add a UNIQUE).
ALTER TABLE security_policy_schedule_pipelines ADD PRIMARY KEY (project_id, security_policy_id);
user_admin_roles looks like a many-to-many link table but has no unique constraint across (admin_role_id, user_id). The same pair can be inserted twice; every join through it will double-count.
Fix Use the FK pair as the primary key (or add a UNIQUE).
ALTER TABLE user_admin_roles ADD PRIMARY KEY (admin_role_id, user_id);
user_member_roles looks like a many-to-many link table but has no unique constraint across (member_role_id, user_id). The same pair can be inserted twice; every join through it will double-count.
Fix Use the FK pair as the primary key (or add a UNIQUE).
ALTER TABLE user_member_roles ADD PRIMARY KEY (member_role_id, user_id);
ascp_component_id, vulnerability_occurrence_idjunction-uniquenessvulnerability_finding_ascp_component_links looks like a many-to-many link table but has no unique constraint across (ascp_component_id, vulnerability_occurrence_id). The same pair can be inserted twice; every join through it will double-count.
Fix Use the FK pair as the primary key (or add a UNIQUE).
ALTER TABLE vulnerability_finding_ascp_component_links ADD PRIMARY KEY (ascp_component_id, vulnerability_occurrence_id);
work_item_colors looks like a many-to-many link table but has no unique constraint across (issue_id, namespace_id). The same pair can be inserted twice; every join through it will double-count.
Fix Use the FK pair as the primary key (or add a UNIQUE).
ALTER TABLE work_item_colors ADD PRIMARY KEY (issue_id, namespace_id);
work_item_type_custom_lifecycles looks like a many-to-many link table but has no unique constraint across (lifecycle_id, namespace_id). The same pair can be inserted twice; every join through it will double-count.
Fix Use the FK pair as the primary key (or add a UNIQUE).
ALTER TABLE work_item_type_custom_lifecycles ADD PRIMARY KEY (lifecycle_id, namespace_id);
workspace_agentk_states looks like a many-to-many link table but has no unique constraint across (project_id, workspace_id). The same pair can be inserted twice; every join through it will double-count.
Fix Use the FK pair as the primary key (or add a UNIQUE).
ALTER TABLE workspace_agentk_states ADD PRIMARY KEY (project_id, workspace_id);
workspace_tokens looks like a many-to-many link table but has no unique constraint across (project_id, workspace_id). The same pair can be inserted twice; every join through it will double-count.
Fix Use the FK pair as the primary key (or add a UNIQUE).
ALTER TABLE workspace_tokens ADD PRIMARY KEY (project_id, workspace_id);
zoekt_replicas looks like a many-to-many link table but has no unique constraint across (namespace_id, zoekt_enabled_namespace_id). The same pair can be inserted twice; every join through it will double-count.
Fix Use the FK pair as the primary key (or add a UNIQUE).
ALTER TABLE zoekt_replicas ADD PRIMARY KEY (namespace_id, zoekt_enabled_namespace_id);
model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE abuse_report_uploads ADD CONSTRAINT abuse_report_uploads_model_type_check CHECK (model_type IN (...));
model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE achievement_uploads ADD CONSTRAINT achievement_uploads_model_type_check CHECK (model_type IN (...));
model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
ai_catalog_item_versions → ai_catalog_items → ai_catalog_item_versions. Rows must be inserted with a deferred constraint or a NULL-then-update dance; backups/restores and truncation have no valid order; ON DELETE CASCADE can loop.
Fix Break the cycle (move one FK to a link table) or mark one FK DEFERRABLE INITIALLY DEFERRED.
ai_catalog_items soft-deletes via deleted_at, but (organization_id, identifier, item_type) is unique across live AND deleted rows. A user who deletes their account can never sign up again with the same value, and ON CONFLICT upserts will resurrect ghosts.
Fix Replace with a partial unique index scoped to live rows.
DROP CONSTRAINT/INDEX ...; CREATE UNIQUE INDEX ai_catalog_items_organization_id_identifier_item_type_live ON ai_catalog_items(organization_id, identifier, item_type) WHERE deleted_at IS NULL;
model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE ai_vectorizable_file_uploads ADD CONSTRAINT ai_vectorizable_file_uploads_model_type_check CHECK (model_type IN (...));
model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE alert_management_alert_metric_image_uploads ADD CONSTRAINT alert_management_alert_metric_image_uploads_model_type_check CHECK (model_type IN (...));
model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE appearance_uploads ADD CONSTRAINT appearance_uploads_model_type_check CHECK (model_type IN (...));
model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
entity_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE audit_events ADD CONSTRAINT audit_events_entity_type_check CHECK (entity_type IN (...));
entity_id points at different tables depending on entity_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
target_id points at different tables depending on target_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
awardable_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE award_emoji ADD CONSTRAINT award_emoji_awardable_type_check CHECK (awardable_type IN (...));
awardable_id points at different tables depending on awardable_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE badges ADD CONSTRAINT badges_type_check CHECK (type IN (...));
model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE bulk_import_export_upload_uploads ADD CONSTRAINT bulk_import_export_upload_uploads_model_type_check CHECK (model_type IN (...));
model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_builds must scan ci_build_trace_chunks to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to ci_build_trace_chunks maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_builds rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.
Fix Add an index on the FK column(s).
CREATE INDEX CONCURRENTLY idx_ci_build_trace_chunks_partition_id_build_id ON ci_build_trace_chunks(partition_id, build_id);
PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_pipelines must scan ci_daily_build_group_report_results to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to ci_daily_build_group_report_results maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_pipelines rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.
Fix Add an index on the FK column(s).
CREATE INDEX CONCURRENTLY idx_ci_daily_build_group_report_results_partition_id_last_pipeline_id ON ci_daily_build_group_report_results(partition_id, last_pipeline_id);
PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_builds must scan ci_job_variables to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to ci_job_variables maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_builds rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.
Fix Add an index on the FK column(s).
CREATE INDEX CONCURRENTLY idx_ci_job_variables_partition_id_job_id ON ci_job_variables(partition_id, job_id);
PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_pipelines must scan ci_pipeline_artifacts to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to ci_pipeline_artifacts maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_pipelines rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.
Fix Add an index on the FK column(s).
CREATE INDEX CONCURRENTLY idx_ci_pipeline_artifacts_partition_id_pipeline_id ON ci_pipeline_artifacts(partition_id, pipeline_id);
PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_pipelines must scan ci_pipeline_chat_data to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to ci_pipeline_chat_data maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_pipelines rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.
Fix Add an index on the FK column(s).
CREATE INDEX CONCURRENTLY idx_ci_pipeline_chat_data_partition_id_pipeline_id ON ci_pipeline_chat_data(partition_id, pipeline_id);
PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_pipelines must scan ci_pipeline_messages to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to ci_pipeline_messages maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_pipelines rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.
Fix Add an index on the FK column(s).
CREATE INDEX CONCURRENTLY idx_ci_pipeline_messages_partition_id_pipeline_id ON ci_pipeline_messages(partition_id, pipeline_id);
PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_pipelines must scan ci_pipeline_metadata to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to ci_pipeline_metadata maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_pipelines rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.
Fix Add an index on the FK column(s).
CREATE INDEX CONCURRENTLY idx_ci_pipeline_metadata_partition_id_pipeline_id ON ci_pipeline_metadata(partition_id, pipeline_id);
PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_builds must scan ci_resources to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to ci_resources maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_builds rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.
Fix Add a partial index, scoped to the rows that have a value. partition_id, build_id is nullable, so the NULL rows are not indexed at all and the write cost is a fraction of a full index, while the constraint check still uses it.
CREATE INDEX CONCURRENTLY idx_ci_resources_partition_id_build_id ON ci_resources(partition_id, build_id) WHERE partition_id IS NOT NULL AND build_id IS NOT NULL;
runner_type, runner_idpolymorphic-referencerunner_id points at different tables depending on runner_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
runner_id points at different tables depending on runner_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
runner_id points at different tables depending on runner_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
runner_id points at different tables depending on runner_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_builds must scan ci_sources_pipelines to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to ci_sources_pipelines maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_builds rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.
Fix Add a partial index, scoped to the rows that have a value. source_partition_id, source_job_id is nullable, so the NULL rows are not indexed at all and the write cost is a fraction of a full index, while the constraint check still uses it.
CREATE INDEX CONCURRENTLY idx_ci_sources_pipelines_source_partition_id_source_job_id ON ci_sources_pipelines(source_partition_id, source_job_id) WHERE source_partition_id IS NOT NULL AND source_job_id IS NOT NULL;
PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_pipelines must scan ci_sources_pipelines to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to ci_sources_pipelines maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_pipelines rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.
Fix Add a partial index, scoped to the rows that have a value. source_partition_id, source_pipeline_id is nullable, so the NULL rows are not indexed at all and the write cost is a fraction of a full index, while the constraint check still uses it.
CREATE INDEX CONCURRENTLY idx_ci_sources_pipelines_source_partition_id_source_pipeline_id ON ci_sources_pipelines(source_partition_id, source_pipeline_id) WHERE source_partition_id IS NOT NULL AND source_pipeline_id IS NOT NULL;
PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_pipelines must scan ci_sources_pipelines to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to ci_sources_pipelines maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_pipelines rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.
Fix Add a partial index, scoped to the rows that have a value. partition_id, pipeline_id is nullable, so the NULL rows are not indexed at all and the write cost is a fraction of a full index, while the constraint check still uses it.
CREATE INDEX CONCURRENTLY idx_ci_sources_pipelines_partition_id_pipeline_id ON ci_sources_pipelines(partition_id, pipeline_id) WHERE partition_id IS NOT NULL AND pipeline_id IS NOT NULL;
PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_pipelines must scan ci_sources_projects to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to ci_sources_projects maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_pipelines rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.
Fix Add an index on the FK column(s).
CREATE INDEX CONCURRENTLY idx_ci_sources_projects_partition_id_pipeline_id ON ci_sources_projects(partition_id, pipeline_id);
PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_builds must scan ci_unit_test_failures to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to ci_unit_test_failures maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_builds rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.
Fix Add an index on the FK column(s).
CREATE INDEX CONCURRENTLY idx_ci_unit_test_failures_partition_id_build_id ON ci_unit_test_failures(partition_id, build_id);
environment_scope is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE ci_variables ADD CONSTRAINT ci_variables_environment_scope_check CHECK (environment_scope IN (...));
machine_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE cluster_providers_gcp ADD CONSTRAINT cluster_providers_gcp_machine_type_check CHECK (machine_type IN (...));
environment_scope is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE clusters ADD CONSTRAINT clusters_environment_scope_check CHECK (environment_scope IN (...));
model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE dependency_list_export_part_uploads ADD CONSTRAINT dependency_list_export_part_uploads_model_type_check CHECK (model_type IN (...));
model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE dependency_list_export_uploads ADD CONSTRAINT dependency_list_export_uploads_model_type_check CHECK (model_type IN (...));
model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
deployable_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE deployments ADD CONSTRAINT deployments_deployable_type_check CHECK (deployable_type IN (...));
deployable_id points at different tables depending on deployable_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE design_management_action_uploads ADD CONSTRAINT design_management_action_uploads_model_type_check CHECK (model_type IN (...));
model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
environment_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE environments ADD CONSTRAINT environments_environment_type_check CHECK (environment_type IN (...));
state is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE environments ADD CONSTRAINT environments_state_check CHECK (state IN (...));
target_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE events ADD CONSTRAINT events_target_type_check CHECK (target_type IN (...));
target_id points at different tables depending on target_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
selective_sync_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE geo_nodes ADD CONSTRAINT geo_nodes_selective_sync_type_check CHECK (selective_sync_type IN (...));
target_id points at different tables depending on target_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
entity_id points at different tables depending on entity_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE import_export_upload_uploads ADD CONSTRAINT import_export_upload_uploads_model_type_check CHECK (model_type IN (...));
model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
target_id points at different tables depending on target_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
category is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE integrations ADD CONSTRAINT integrations_category_check CHECK (category IN (...));
model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE issuable_metric_image_uploads ADD CONSTRAINT issuable_metric_image_uploads_model_type_check CHECK (model_type IN (...));
model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
issues → epics → work_item_parent_links → issues. Rows must be inserted with a deferred constraint or a NULL-then-update dance; backups/restores and truncation have no valid order; ON DELETE CASCADE can loop.
Fix Break the cycle (move one FK to a link table) or mark one FK DEFERRABLE INITIALLY DEFERRED.
issues → epics → issues. Rows must be inserted with a deferred constraint or a NULL-then-update dance; backups/restores and truncation have no valid order; ON DELETE CASCADE can loop.
Fix Break the cycle (move one FK to a link table) or mark one FK DEFERRABLE INITIALLY DEFERRED.
type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE keys ADD CONSTRAINT keys_type_check CHECK (type IN (...));
target_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE label_links ADD CONSTRAINT label_links_target_type_check CHECK (target_type IN (...));
target_id points at different tables depending on target_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE labels ADD CONSTRAINT labels_type_check CHECK (type IN (...));
source_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE members ADD CONSTRAINT members_source_type_check CHECK (source_type IN (...));
type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE members ADD CONSTRAINT members_type_check CHECK (type IN (...));
source_id points at different tables depending on source_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
a_mode is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE merge_request_diff_files ADD CONSTRAINT merge_request_diff_files_a_mode_check CHECK (a_mode IN (...));
b_mode is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE merge_request_diff_files ADD CONSTRAINT merge_request_diff_files_b_mode_check CHECK (b_mode IN (...));
a_mode is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE merge_request_diff_files_99208b8fac ADD CONSTRAINT merge_request_diff_files_99208b8fac_a_mode_check CHECK (a_mode IN (...));
b_mode is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE merge_request_diff_files_99208b8fac ADD CONSTRAINT merge_request_diff_files_99208b8fac_b_mode_check CHECK (b_mode IN (...));
state is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE merge_request_diffs ADD CONSTRAINT merge_request_diffs_state_check CHECK (state IN (...));
merge_requests → merge_request_diffs → merge_requests. Rows must be inserted with a deferred constraint or a NULL-then-update dance; backups/restores and truncation have no valid order; ON DELETE CASCADE can loop.
Fix Break the cycle (move one FK to a link table) or mark one FK DEFERRABLE INITIALLY DEFERRED.
merge_status is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE merge_requests ADD CONSTRAINT merge_requests_merge_status_check CHECK (merge_status IN (...));
state is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE milestones ADD CONSTRAINT milestones_state_check CHECK (state IN (...));
model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE namespace_uploads ADD CONSTRAINT namespace_uploads_model_type_check CHECK (model_type IN (...));
model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE namespaces ADD CONSTRAINT namespaces_type_check CHECK (type IN (...));
ldap_sync_status is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE namespaces ADD CONSTRAINT namespaces_ldap_sync_status_check CHECK (ldap_sync_status IN (...));
a_mode is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE note_diff_files ADD CONSTRAINT note_diff_files_a_mode_check CHECK (a_mode IN (...));
b_mode is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE note_diff_files ADD CONSTRAINT note_diff_files_b_mode_check CHECK (b_mode IN (...));
type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE notes ADD CONSTRAINT notes_type_check CHECK (type IN (...));
noteable_id points at different tables depending on noteable_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
source_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE notification_settings ADD CONSTRAINT notification_settings_source_type_check CHECK (source_type IN (...));
source_id points at different tables depending on source_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
owner_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE oauth_applications ADD CONSTRAINT oauth_applications_owner_type_check CHECK (owner_type IN (...));
owner_id points at different tables depending on owner_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE organization_detail_uploads ADD CONSTRAINT organization_detail_uploads_model_type_check CHECK (model_type IN (...));
model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_builds must scan p_ci_build_needs to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to p_ci_build_needs maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_builds rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.
Fix Add an index on the FK column(s).
CREATE INDEX CONCURRENTLY idx_p_ci_build_needs_partition_id_build_id ON p_ci_build_needs(partition_id, build_id);
PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_job_artifacts must scan p_ci_build_trace_metadata to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to p_ci_build_trace_metadata maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_job_artifacts rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.
Fix Add a partial index, scoped to the rows that have a value. partition_id, trace_artifact_id is nullable, so the NULL rows are not indexed at all and the write cost is a fraction of a full index, while the constraint check still uses it.
CREATE INDEX CONCURRENTLY idx_p_ci_build_trace_metadata_partition_id_trace_artifact_id ON p_ci_build_trace_metadata(partition_id, trace_artifact_id) WHERE partition_id IS NOT NULL AND trace_artifact_id IS NOT NULL;
PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_stages must scan p_ci_builds to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to p_ci_builds maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_stages rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.
Fix Add a partial index, scoped to the rows that have a value. partition_id, stage_id is nullable, so the NULL rows are not indexed at all and the write cost is a fraction of a full index, while the constraint check still uses it.
CREATE INDEX CONCURRENTLY idx_p_ci_builds_partition_id_stage_id ON p_ci_builds(partition_id, stage_id) WHERE partition_id IS NOT NULL AND stage_id IS NOT NULL;
PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_pipelines must scan p_ci_builds to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to p_ci_builds maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_pipelines rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.
Fix Add a partial index, scoped to the rows that have a value. upstream_pipeline_partition_id, upstream_pipeline_id is nullable, so the NULL rows are not indexed at all and the write cost is a fraction of a full index, while the constraint check still uses it.
CREATE INDEX CONCURRENTLY idx_p_ci_builds_upstream_pipeline_partition_id_upstream_pipeline_id ON p_ci_builds(upstream_pipeline_partition_id, upstream_pipeline_id) WHERE upstream_pipeline_partition_id IS NOT NULL AND upstream_pipeline_id IS NOT NULL;
PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_pipelines must scan p_ci_builds to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to p_ci_builds maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_pipelines rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.
Fix Add a partial index, scoped to the rows that have a value. auto_canceled_by_partition_id, auto_canceled_by_id is nullable, so the NULL rows are not indexed at all and the write cost is a fraction of a full index, while the constraint check still uses it.
CREATE INDEX CONCURRENTLY idx_p_ci_builds_auto_canceled_by_partition_id_auto_canceled_by_id ON p_ci_builds(auto_canceled_by_partition_id, auto_canceled_by_id) WHERE auto_canceled_by_partition_id IS NOT NULL AND auto_canceled_by_id IS NOT NULL;
PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_pipelines must scan p_ci_builds to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to p_ci_builds maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_pipelines rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.
Fix Add a partial index, scoped to the rows that have a value. partition_id, commit_id is nullable, so the NULL rows are not indexed at all and the write cost is a fraction of a full index, while the constraint check still uses it.
CREATE INDEX CONCURRENTLY idx_p_ci_builds_partition_id_commit_id ON p_ci_builds(partition_id, commit_id) WHERE partition_id IS NOT NULL AND commit_id IS NOT NULL;
status is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE p_ci_builds ADD CONSTRAINT p_ci_builds_status_check CHECK (status IN (...));
type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE p_ci_builds ADD CONSTRAINT p_ci_builds_type_check CHECK (type IN (...));
PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_builds must scan p_ci_builds_partition_overrides to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to p_ci_builds_partition_overrides maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_builds rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.
Fix Add an index on the FK column(s).
CREATE INDEX CONCURRENTLY idx_p_ci_builds_partition_overrides_partition_id_build_id ON p_ci_builds_partition_overrides(partition_id, build_id);
PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_builds must scan p_ci_job_artifacts to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to p_ci_job_artifacts maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_builds rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.
Fix Add an index on the FK column(s).
CREATE INDEX CONCURRENTLY idx_p_ci_job_artifacts_partition_id_job_id ON p_ci_job_artifacts(partition_id, job_id);
PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_job_definitions must scan p_ci_job_definition_instances to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to p_ci_job_definition_instances maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_job_definitions rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.
Fix Add an index on the FK column(s).
CREATE INDEX CONCURRENTLY idx_p_ci_job_definition_instances_partition_id_job_definition_id ON p_ci_job_definition_instances(partition_id, job_definition_id);
PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_builds must scan p_ci_job_inputs to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to p_ci_job_inputs maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_builds rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.
Fix Add an index on the FK column(s).
CREATE INDEX CONCURRENTLY idx_p_ci_job_inputs_partition_id_job_id ON p_ci_job_inputs(partition_id, job_id);
PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_builds must scan p_ci_job_messages to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to p_ci_job_messages maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_builds rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.
Fix Add an index on the FK column(s).
CREATE INDEX CONCURRENTLY idx_p_ci_job_messages_partition_id_job_id ON p_ci_job_messages(partition_id, job_id);
PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_pipelines must scan p_ci_pipeline_variables to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to p_ci_pipeline_variables maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_pipelines rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.
Fix Add an index on the FK column(s).
CREATE INDEX CONCURRENTLY idx_p_ci_pipeline_variables_partition_id_pipeline_id ON p_ci_pipeline_variables(partition_id, pipeline_id);
PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_pipelines must scan p_ci_pipelines to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to p_ci_pipelines maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_pipelines rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.
Fix Add a partial index, scoped to the rows that have a value. auto_canceled_by_partition_id, auto_canceled_by_id is nullable, so the NULL rows are not indexed at all and the write cost is a fraction of a full index, while the constraint check still uses it.
CREATE INDEX CONCURRENTLY idx_p_ci_pipelines_auto_canceled_by_partition_id_auto_canceled_by_id ON p_ci_pipelines(auto_canceled_by_partition_id, auto_canceled_by_id) WHERE auto_canceled_by_partition_id IS NOT NULL AND auto_canceled_by_id IS NOT NULL;
status is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE p_ci_pipelines ADD CONSTRAINT p_ci_pipelines_status_check CHECK (status IN (...));
PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_pipelines must scan p_ci_stages to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to p_ci_stages maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_pipelines rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.
Fix Add a partial index, scoped to the rows that have a value. partition_id, pipeline_id is nullable, so the NULL rows are not indexed at all and the write cost is a fraction of a full index, while the constraint check still uses it.
CREATE INDEX CONCURRENTLY idx_p_ci_stages_partition_id_pipeline_id ON p_ci_stages(partition_id, pipeline_id) WHERE partition_id IS NOT NULL AND pipeline_id IS NOT NULL;
noteable_id points at different tables depending on noteable_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
state is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE pool_repositories ADD CONSTRAINT pool_repositories_state_check CHECK (state IN (...));
target_id points at different tables depending on target_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE project_import_export_relation_export_upload_uploads ADD CONSTRAINT project_import_export_relation_export_upload_uploads_model_type_check CHECK (model_type IN (...));
model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
status is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE project_mirror_data ADD CONSTRAINT project_mirror_data_status_check CHECK (status IN (...));
model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE project_topic_uploads ADD CONSTRAINT project_topic_uploads_model_type_check CHECK (model_type IN (...));
model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE project_uploads ADD CONSTRAINT project_uploads_model_type_check CHECK (model_type IN (...));
model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
import_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE projects ADD CONSTRAINT projects_import_type_check CHECK (import_type IN (...));
source_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE redirect_routes ADD CONSTRAINT redirect_routes_source_type_check CHECK (source_type IN (...));
source_id points at different tables depending on source_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
update_status is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE remote_mirrors ADD CONSTRAINT remote_mirrors_update_status_check CHECK (update_status IN (...));
source_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE routes ADD CONSTRAINT routes_source_type_check CHECK (source_type IN (...));
source_id points at different tables depending on source_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE snippet_uploads ADD CONSTRAINT snippet_uploads_model_type_check CHECK (model_type IN (...));
model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE snippets ADD CONSTRAINT snippets_type_check CHECK (type IN (...));
noteable_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE spam_logs ADD CONSTRAINT spam_logs_noteable_type_check CHECK (noteable_type IN (...));
item_id points at different tables depending on item_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
subscribable_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE subscriptions ADD CONSTRAINT subscriptions_subscribable_type_check CHECK (subscribable_type IN (...));
subscribable_id points at different tables depending on subscribable_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
terraform_states soft-deletes via deleted_at, but (project_id, name) is unique across live AND deleted rows. A user who deletes their account can never sign up again with the same value, and ON CONFLICT upserts will resurrect ghosts.
Fix Replace with a partial unique index scoped to live rows.
DROP CONSTRAINT/INDEX ...; CREATE UNIQUE INDEX terraform_states_project_id_name_live ON terraform_states(project_id, name) WHERE deleted_at IS NULL;
terraform_states soft-deletes via deleted_at, but (uuid) is unique across live AND deleted rows. A user who deletes their account can never sign up again with the same value, and ON CONFLICT upserts will resurrect ghosts.
Fix Replace with a partial unique index scoped to live rows.
DROP CONSTRAINT/INDEX ...; CREATE UNIQUE INDEX terraform_states_uuid_live ON terraform_states(uuid) WHERE deleted_at IS NULL;
target_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE todos ADD CONSTRAINT todos_target_type_check CHECK (target_type IN (...));
state is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE todos ADD CONSTRAINT todos_state_check CHECK (state IN (...));
target_id points at different tables depending on target_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE uploads ADD CONSTRAINT uploads_model_type_check CHECK (model_type IN (...));
model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
subject_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE user_agent_details ADD CONSTRAINT user_agent_details_subject_type_check CHECK (subject_type IN (...));
subject_id points at different tables depending on subject_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
target_id points at different tables depending on target_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE user_permission_export_upload_uploads ADD CONSTRAINT user_permission_export_upload_uploads_model_type_check CHECK (model_type IN (...));
model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE user_uploads ADD CONSTRAINT user_uploads_model_type_check CHECK (model_type IN (...));
model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
state is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE users ADD CONSTRAINT users_state_check CHECK (state IN (...));
vulnerabilities → vulnerability_occurrences → vulnerabilities. Rows must be inserted with a deferred constraint or a NULL-then-update dance; backups/restores and truncation have no valid order; ON DELETE CASCADE can loop.
Fix Break the cycle (move one FK to a link table) or mark one FK DEFERRABLE INITIALLY DEFERRED.
model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE vulnerability_archive_export_uploads ADD CONSTRAINT vulnerability_archive_export_uploads_model_type_check CHECK (model_type IN (...));
model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE vulnerability_export_part_uploads ADD CONSTRAINT vulnerability_export_part_uploads_model_type_check CHECK (model_type IN (...));
model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE vulnerability_export_uploads ADD CONSTRAINT vulnerability_export_uploads_model_type_check CHECK (model_type IN (...));
model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
external_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE vulnerability_identifiers ADD CONSTRAINT vulnerability_identifiers_external_type_check CHECK (external_type IN (...));
external_id points at different tables depending on external_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE vulnerability_remediation_uploads ADD CONSTRAINT vulnerability_remediation_uploads_model_type_check CHECK (model_type IN (...));
model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
response_status is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE web_hook_logs_daily ADD CONSTRAINT web_hook_logs_daily_response_status_check CHECK (response_status IN (...));
type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.
Fix Add a CHECK constraint (cheap, easy to evolve) or a lookup table if values need metadata. Avoid native ENUM types unless the set is truly frozen.
ALTER TABLE web_hooks ADD CONSTRAINT web_hooks_type_check CHECK (type IN (...));
item_id points at different tables depending on item_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.
Fix Either one nullable FK per target with a CHECK that exactly one is set, or a supertype table that the targets reference.
abuse_events.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
abuse_events.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
abuse_events.abuse_report_id may be NULL, so the relationship to abuse_reports is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
abuse_events has several nullable FKs (organization_id, user_id, abuse_report_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE abuse_events ADD CONSTRAINT abuse_events_one_target CHECK ((organization_id IS NOT NULL)::int + (user_id IS NOT NULL)::int + (abuse_report_id IS NOT NULL)::int = 1);
abuse_report_events.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
abuse_report_events.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
abuse_report_events has several nullable FKs (user_id, organization_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE abuse_report_events ADD CONSTRAINT abuse_report_events_one_target CHECK ((user_id IS NOT NULL)::int + (organization_id IS NOT NULL)::int = 1);
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE abuse_report_uploads ALTER COLUMN created_at TYPE timestamptz;
abuse_report_user_mentions.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
abuse_reports.assignee_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
abuse_reports.reporter_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
abuse_reports.resolved_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
abuse_reports.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE abuse_reports ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE abuse_reports ALTER COLUMN updated_at TYPE timestamptz;
abuse_reports has several nullable FKs (assignee_id, reporter_id, resolved_by_id, organization_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE abuse_reports ADD CONSTRAINT abuse_reports_one_target CHECK ((assignee_id IS NOT NULL)::int + (reporter_id IS NOT NULL)::int + (resolved_by_id IS NOT NULL)::int + (organization_id IS NOT NULL)::int = 1);
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE achievement_uploads ALTER COLUMN created_at TYPE timestamptz;
activation_metrics.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
admin_roles.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
agent_activity_events.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
agent_activity_events.merge_request_id may be NULL, so the relationship to merge_requests is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
agent_activity_events.agent_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
agent_activity_events.agent_token_id may be NULL, so the relationship to cluster_agent_tokens is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
agent_activity_events.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
project_id, merge_request_id, agent_project_id, agent_token_id, user_idexclusive-arcagent_activity_events has several nullable FKs (project_id, merge_request_id, agent_project_id, agent_token_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE agent_activity_events ADD CONSTRAINT agent_activity_events_one_target CHECK ((project_id IS NOT NULL)::int + (merge_request_id IS NOT NULL)::int + (agent_project_id IS NOT NULL)::int + (agent_token_id IS NOT NULL)::int + (user_id IS NOT NULL)::int = 1);
ai_active_context_tasks.depends_on_id may be NULL, so the relationship to ai_active_context_tasks is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ai_audit_events references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
ai_catalog_item_consumers.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ai_catalog_item_consumers.parent_item_consumer_id may be NULL, so the relationship to ai_catalog_item_consumers is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ai_catalog_item_consumers.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ai_catalog_item_consumers.service_account_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ai_catalog_item_consumers.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
group_id, parent_item_consumer_id, project_id, service_account_id, organization_idexclusive-arcai_catalog_item_consumers has several nullable FKs (group_id, parent_item_consumer_id, project_id, service_account_id, organization_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE ai_catalog_item_consumers ADD CONSTRAINT ai_catalog_item_consumers_one_target CHECK ((group_id IS NOT NULL)::int + (parent_item_consumer_id IS NOT NULL)::int + (project_id IS NOT NULL)::int + (service_account_id IS NOT NULL)::int + (organization_id IS NOT NULL)::int = 1);
ai_catalog_item_versions.created_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ai_catalog_items.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ai_catalog_items.latest_released_version_id may be NULL, so the relationship to ai_catalog_item_versions is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ai_catalog_items.latest_version_id may be NULL, so the relationship to ai_catalog_item_versions is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ai_catalog_items has several nullable FKs (project_id, latest_released_version_id, latest_version_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE ai_catalog_items ADD CONSTRAINT ai_catalog_items_one_target CHECK ((project_id IS NOT NULL)::int + (latest_released_version_id IS NOT NULL)::int + (latest_version_id IS NOT NULL)::int = 1);
ai_catalog_mcp_server_blocks.created_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ai_catalog_mcp_servers.created_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ai_feature_settings.ai_self_hosted_model_id may be NULL, so the relationship to ai_self_hosted_models is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ai_flow_triggers.ai_catalog_item_consumer_id may be NULL, so the relationship to ai_catalog_item_consumers is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ai_instance_accessible_entity_rules references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
ai_namespace_feature_access_rules.through_namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ai_settings.amazon_q_oauth_application_id may be NULL, so the relationship to oauth_applications is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ai_settings.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ai_settings.duo_workflow_oauth_application_id may be NULL, so the relationship to oauth_applications is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ai_settings.duo_workflow_service_account_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ai_settings.amazon_q_service_account_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
amazon_q_oauth_application_id, organization_id, duo_workflow_oauth_application_id, duo_workflow_service_account_user_id, amazon_q_service_account_user_idexclusive-arcai_settings has several nullable FKs (amazon_q_oauth_application_id, organization_id, duo_workflow_oauth_application_id, duo_workflow_service_account_user_id, amazon_q_service_account_user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE ai_settings ADD CONSTRAINT ai_settings_one_target CHECK ((amazon_q_oauth_application_id IS NOT NULL)::int + (organization_id IS NOT NULL)::int + (duo_workflow_oauth_application_id IS NOT NULL)::int + (duo_workflow_service_account_user_id IS NOT NULL)::int + (amazon_q_service_account_user_id IS NOT NULL)::int = 1);
ai_testing_terms_acceptances references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
ai_tool_rules.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ai_usage_events.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE ai_vectorizable_file_uploads ALTER COLUMN created_at TYPE timestamptz;
alert_management_alert_assignees.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE alert_management_alert_metric_image_uploads ALTER COLUMN created_at TYPE timestamptz;
alert_management_alert_metric_images.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
alert_management_alert_user_mentions.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
alert_management_alert_user_mentions.note_id may be NULL, so the relationship to notes is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
alert_management_alert_user_mentions has several nullable FKs (project_id, note_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE alert_management_alert_user_mentions ADD CONSTRAINT alert_management_alert_user_mentions_one_target CHECK ((project_id IS NOT NULL)::int + (note_id IS NOT NULL)::int = 1);
alert_management_alerts.issue_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
alert_management_alerts.environment_id may be NULL, so the relationship to environments is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
alert_management_alerts has several nullable FKs (issue_id, environment_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE alert_management_alerts ADD CONSTRAINT alert_management_alerts_one_target CHECK ((issue_id IS NOT NULL)::int + (environment_id IS NOT NULL)::int = 1);
analytics_cycle_analytics_group_stages.stage_event_hash_id may be NULL, so the relationship to analytics_cycle_analytics_stage_event_hashes is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
analytics_cycle_analytics_group_stages.start_event_label_id may be NULL, so the relationship to labels is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
analytics_cycle_analytics_group_stages.end_event_label_id may be NULL, so the relationship to labels is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
stage_event_hash_id, start_event_label_id, end_event_label_idexclusive-arcanalytics_cycle_analytics_group_stages has several nullable FKs (stage_event_hash_id, start_event_label_id, end_event_label_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE analytics_cycle_analytics_group_stages ADD CONSTRAINT analytics_cycle_analytics_group_stages_one_target CHECK ((stage_event_hash_id IS NOT NULL)::int + (start_event_label_id IS NOT NULL)::int + (end_event_label_id IS NOT NULL)::int = 1);
analytics_cycle_analytics_issue_stage_events references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_00 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_01 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_02 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_03 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_04 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_05 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_06 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_07 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_08 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_09 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_10 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_11 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_12 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_13 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_14 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_15 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_16 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_17 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_18 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_19 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_20 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_21 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_22 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_23 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_24 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_25 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_26 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_27 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_28 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_29 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_30 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_issue_stage_events_31 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_00 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_01 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_02 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_03 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_04 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_05 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_06 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_07 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_08 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_09 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_10 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_11 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_12 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_13 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_14 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_15 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_16 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_17 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_18 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_19 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_20 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_21 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_22 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_23 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_24 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_25 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_26 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_27 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_28 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_29 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_30 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_cycle_analytics_merge_request_stage_events_31 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analytics_dashboards_pointers.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
analytics_dashboards_pointers.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
analytics_devops_adoption_segments.display_namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
analytics_devops_adoption_segments.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
analytics_devops_adoption_segments has several nullable FKs (display_namespace_id, namespace_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE analytics_devops_adoption_segments ADD CONSTRAINT analytics_devops_adoption_segments_one_target CHECK ((display_namespace_id IS NOT NULL)::int + (namespace_id IS NOT NULL)::int = 1);
analytics_devops_adoption_snapshots.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
analytics_usage_trends_measurements references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analyzer_namespace_statuses references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
analyzer_project_statuses references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE appearance_uploads ALTER COLUMN created_at TYPE timestamptz;
appearances references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
application_settings.duo_template_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
application_settings.o11y_oauth_application_id may be NULL, so the relationship to oauth_applications is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
application_settings.workspaces_oauth_application_id may be NULL, so the relationship to oauth_applications is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
application_settings.web_ide_oauth_application_id may be NULL, so the relationship to oauth_applications is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE application_settings ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE application_settings ALTER COLUMN updated_at TYPE timestamptz;
Tables this wide usually hide several entities (or a JSON column that wants to be one). Every row update rewrites the whole tuple; TOAST kicks in; indexes bloat.
Fix Look for column groups that always change together and split them out.
duo_template_project_id, o11y_oauth_application_id, workspaces_oauth_application_id, web_ide_oauth_application_idexclusive-arcapplication_settings has several nullable FKs (duo_template_project_id, o11y_oauth_application_id, workspaces_oauth_application_id, web_ide_oauth_application_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE application_settings ADD CONSTRAINT application_settings_one_target CHECK ((duo_template_project_id IS NOT NULL)::int + (o11y_oauth_application_id IS NOT NULL)::int + (workspaces_oauth_application_id IS NOT NULL)::int + (web_ide_oauth_application_id IS NOT NULL)::int = 1);
approval_group_rules.scan_result_policy_id may be NULL, so the relationship to scan_result_policies is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
approval_group_rules.approval_policy_rule_id may be NULL, so the relationship to approval_policy_rules is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
approval_group_rules.security_orchestration_policy_configuration_id may be NULL, so the relationship to security_orchestration_policy_configurations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
scan_result_policy_id, approval_policy_rule_id, security_orchestration_policy_configuration_idexclusive-arcapproval_group_rules has several nullable FKs (scan_result_policy_id, approval_policy_rule_id, security_orchestration_policy_configuration_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE approval_group_rules ADD CONSTRAINT approval_group_rules_one_target CHECK ((scan_result_policy_id IS NOT NULL)::int + (approval_policy_rule_id IS NOT NULL)::int + (security_orchestration_policy_configuration_id IS NOT NULL)::int = 1);
approval_group_rules_protected_branches.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
approval_group_rules_users.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
approval_merge_request_rule_sources.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
approval_merge_request_rules.security_orchestration_policy_configuration_id may be NULL, so the relationship to security_orchestration_policy_configurations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
approval_merge_request_rules.scan_result_policy_id may be NULL, so the relationship to scan_result_policies is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
approval_merge_request_rules.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
security_orchestration_policy_configuration_id, scan_result_policy_id, project_idexclusive-arcapproval_merge_request_rules has several nullable FKs (security_orchestration_policy_configuration_id, scan_result_policy_id, project_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE approval_merge_request_rules ADD CONSTRAINT approval_merge_request_rules_one_target CHECK ((security_orchestration_policy_configuration_id IS NOT NULL)::int + (scan_result_policy_id IS NOT NULL)::int + (project_id IS NOT NULL)::int = 1);
approval_merge_request_rules_approved_approvers.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
approval_merge_request_rules_users.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
approval_policy_merge_request_bypass_events.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
approval_project_rules.approval_policy_rule_id may be NULL, so the relationship to approval_policy_rules is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
approval_project_rules.scan_result_policy_id may be NULL, so the relationship to scan_result_policies is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
approval_project_rules.security_orchestration_policy_configuration_id may be NULL, so the relationship to security_orchestration_policy_configurations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
approval_policy_rule_id, scan_result_policy_id, security_orchestration_policy_configuration_idexclusive-arcapproval_project_rules has several nullable FKs (approval_policy_rule_id, scan_result_policy_id, security_orchestration_policy_configuration_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE approval_project_rules ADD CONSTRAINT approval_project_rules_one_target CHECK ((approval_policy_rule_id IS NOT NULL)::int + (scan_result_policy_id IS NOT NULL)::int + (security_orchestration_policy_configuration_id IS NOT NULL)::int = 1);
approval_project_rules_protected_branches.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
approval_project_rules_users.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE approvals ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE approvals ALTER COLUMN updated_at TYPE timestamptz;
approvals.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE ar_internal_metadata ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE ar_internal_metadata ALTER COLUMN updated_at TYPE timestamptz;
ar_internal_metadata references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
Tables this wide usually hide several entities (or a JSON column that wants to be one). Every row update rewrites the whole tuple; TOAST kicks in; indexes bloat.
Fix Look for column groups that always change together and split them out.
ascp_scans.base_scan_id may be NULL, so the relationship to ascp_scans is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE audit_events ALTER COLUMN created_at TYPE timestamptz;
audit_events references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
audit_events_amazon_s3_configurations.stream_destination_id may be NULL, so the relationship to audit_events_group_external_streaming_destinations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
audit_events_external_audit_event_destinations.stream_destination_id may be NULL, so the relationship to audit_events_group_external_streaming_destinations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
audit_events_google_cloud_logging_configurations.stream_destination_id may be NULL, so the relationship to audit_events_group_external_streaming_destinations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
audit_events_instance_amazon_s3_configurations.stream_destination_id may be NULL, so the relationship to audit_events_instance_external_streaming_destinations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
audit_events_instance_external_audit_event_destinations.stream_destination_id may be NULL, so the relationship to audit_events_instance_external_streaming_destinations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
audit_events_instance_google_cloud_logging_configurations.stream_destination_id may be NULL, so the relationship to audit_events_instance_external_streaming_destinations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
audit_events_streaming_event_type_filters.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
audit_events_streaming_headers.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
authentication_events.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE award_emoji ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE award_emoji ALTER COLUMN updated_at TYPE timestamptz;
award_emoji.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
award_emoji.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
background_operation_jobs references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
background_operation_jobs_cell_local references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
background_operation_workers references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
background_operation_workers_cell_local references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
backup_finding_evidences references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
backup_finding_flags references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
backup_finding_identifiers references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
backup_finding_links references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
backup_finding_remediations references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
backup_finding_signatures references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
backup_findings references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
backup_vulnerabilities references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
backup_vulnerability_external_issue_links references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
backup_vulnerability_issue_links references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
backup_vulnerability_merge_request_links references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
backup_vulnerability_reads references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
backup_vulnerability_severity_overrides references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
backup_vulnerability_state_transitions references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
backup_vulnerability_user_mentions references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
badges.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
badges.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
board_assignees.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
board_assignees.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
board_group_recent_visits.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
board_group_recent_visits.board_id may be NULL, so the relationship to boards is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
board_group_recent_visits.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
board_group_recent_visits has several nullable FKs (user_id, board_id, group_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE board_group_recent_visits ADD CONSTRAINT board_group_recent_visits_one_target CHECK ((user_id IS NOT NULL)::int + (board_id IS NOT NULL)::int + (group_id IS NOT NULL)::int = 1);
board_labels.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
board_labels.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
board_project_recent_visits.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
board_project_recent_visits.board_id may be NULL, so the relationship to boards is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
board_project_recent_visits.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
board_project_recent_visits has several nullable FKs (project_id, board_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE board_project_recent_visits ADD CONSTRAINT board_project_recent_visits_one_target CHECK ((project_id IS NOT NULL)::int + (board_id IS NOT NULL)::int + (user_id IS NOT NULL)::int = 1);
board_user_preferences.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
board_user_preferences.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE boards ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE boards ALTER COLUMN updated_at TYPE timestamptz;
boards.iteration_cadence_id may be NULL, so the relationship to iterations_cadences is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
boards.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
boards.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
boards has several nullable FKs (iteration_cadence_id, project_id, group_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE boards ADD CONSTRAINT boards_one_target CHECK ((iteration_cadence_id IS NOT NULL)::int + (project_id IS NOT NULL)::int + (group_id IS NOT NULL)::int = 1);
boards_epic_board_labels.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
boards_epic_board_positions.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
boards_epic_list_user_preferences.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
boards_epic_lists.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
boards_epic_lists.label_id may be NULL, so the relationship to labels is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
boards_epic_lists has several nullable FKs (group_id, label_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE boards_epic_lists ADD CONSTRAINT boards_epic_lists_one_target CHECK ((group_id IS NOT NULL)::int + (label_id IS NOT NULL)::int = 1);
boards_epic_user_preferences.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
starts_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE broadcast_messages ALTER COLUMN starts_at TYPE timestamptz;
ends_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE broadcast_messages ALTER COLUMN ends_at TYPE timestamptz;
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE broadcast_messages ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE broadcast_messages ALTER COLUMN updated_at TYPE timestamptz;
bulk_import_batch_trackers.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
bulk_import_batch_trackers.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
bulk_import_batch_trackers.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
bulk_import_configurations.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
bulk_import_entities.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
bulk_import_entities.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
bulk_import_entities.parent_id may be NULL, so the relationship to bulk_import_entities is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
bulk_import_entities.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
bulk_import_entities has several nullable FKs (organization_id, namespace_id, parent_id, project_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE bulk_import_entities ADD CONSTRAINT bulk_import_entities_one_target CHECK ((organization_id IS NOT NULL)::int + (namespace_id IS NOT NULL)::int + (parent_id IS NOT NULL)::int + (project_id IS NOT NULL)::int = 1);
bulk_import_export_batches.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
bulk_import_export_batches.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE bulk_import_export_upload_uploads ALTER COLUMN created_at TYPE timestamptz;
bulk_import_export_uploads.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
bulk_import_export_uploads.batch_id may be NULL, so the relationship to bulk_import_export_batches is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
bulk_import_export_uploads.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
bulk_import_export_uploads has several nullable FKs (group_id, batch_id, project_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE bulk_import_export_uploads ADD CONSTRAINT bulk_import_export_uploads_one_target CHECK ((group_id IS NOT NULL)::int + (batch_id IS NOT NULL)::int + (project_id IS NOT NULL)::int = 1);
bulk_import_exports.offline_export_id may be NULL, so the relationship to import_offline_exports is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
bulk_import_exports.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
bulk_import_exports.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
bulk_import_exports.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
bulk_import_exports has several nullable FKs (offline_export_id, project_id, group_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE bulk_import_exports ADD CONSTRAINT bulk_import_exports_one_target CHECK ((offline_export_id IS NOT NULL)::int + (project_id IS NOT NULL)::int + (group_id IS NOT NULL)::int + (user_id IS NOT NULL)::int = 1);
bulk_import_failures.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
bulk_import_failures.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
bulk_import_failures.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
bulk_import_trackers.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
bulk_import_trackers.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
bulk_import_trackers.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
catalog_resource_versions.published_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cd_applications.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cd_artifact_sources.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cd_artifact_sources.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cd_artifact_sources has several nullable FKs (group_id, organization_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE cd_artifact_sources ADD CONSTRAINT cd_artifact_sources_one_target CHECK ((group_id IS NOT NULL)::int + (organization_id IS NOT NULL)::int = 1);
cd_deployments.service_id may be NULL, so the relationship to cd_services is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cd_deployments.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cd_deployments.rollout_environment_id may be NULL, so the relationship to cd_rollout_environments is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cd_deployments.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cd_deployments has several nullable FKs (service_id, group_id, rollout_environment_id, organization_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE cd_deployments ADD CONSTRAINT cd_deployments_one_target CHECK ((service_id IS NOT NULL)::int + (group_id IS NOT NULL)::int + (rollout_environment_id IS NOT NULL)::int + (organization_id IS NOT NULL)::int = 1);
cd_environments.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cd_rollout_environments.previous_version_set_id may be NULL, so the relationship to cd_version_sets is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cd_rollout_steps.rollout_environment_id may be NULL, so the relationship to cd_rollout_environments is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cd_rollouts.application_flow_definition_id may be NULL, so the relationship to cd_application_flow_definitions is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cd_rollouts.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cd_rollouts.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cd_rollouts.application_id may be NULL, so the relationship to cd_applications is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
application_flow_definition_id, organization_id, group_id, application_idexclusive-arccd_rollouts has several nullable FKs (application_flow_definition_id, organization_id, group_id, application_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE cd_rollouts ADD CONSTRAINT cd_rollouts_one_target CHECK ((application_flow_definition_id IS NOT NULL)::int + (organization_id IS NOT NULL)::int + (group_id IS NOT NULL)::int + (application_id IS NOT NULL)::int = 1);
cd_services.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cd_services.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cd_services has several nullable FKs (group_id, organization_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE cd_services ADD CONSTRAINT cd_services_one_target CHECK ((group_id IS NOT NULL)::int + (organization_id IS NOT NULL)::int = 1);
cd_version_set_entries.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cd_version_set_entries.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cd_version_set_entries.artifact_source_id may be NULL, so the relationship to cd_artifact_sources is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cd_version_set_entries has several nullable FKs (group_id, organization_id, artifact_source_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE cd_version_set_entries ADD CONSTRAINT cd_version_set_entries_one_target CHECK ((group_id IS NOT NULL)::int + (organization_id IS NOT NULL)::int + (artifact_source_id IS NOT NULL)::int = 1);
cd_version_sets.created_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cd_version_sets.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cd_version_sets.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cd_version_sets has several nullable FKs (created_by_id, group_id, organization_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE cd_version_sets ADD CONSTRAINT cd_version_sets_one_target CHECK ((created_by_id IS NOT NULL)::int + (group_id IS NOT NULL)::int + (organization_id IS NOT NULL)::int = 1);
cd_versions.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cd_versions.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cd_versions has several nullable FKs (organization_id, group_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE cd_versions ADD CONSTRAINT cd_versions_one_target CHECK ((organization_id IS NOT NULL)::int + (group_id IS NOT NULL)::int = 1);
cells_outstanding_leases references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
last_used_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE chat_names ALTER COLUMN last_used_at TYPE timestamptz;
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE chat_names ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE chat_names ALTER COLUMN updated_at TYPE timestamptz;
chat_names references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE chat_teams ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE chat_teams ALTER COLUMN updated_at TYPE timestamptz;
ci_build_runtime_environments references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
ci_deleted_objects references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
ci_freeze_periods references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
ci_gitlab_hosted_runner_monthly_usages references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
ci_group_variables references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
ci_instance_runner_monthly_usages references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
ci_instance_variables references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
ci_job_token_authorizations references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
ci_job_token_group_scope_links references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
ci_job_token_project_scope_links references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
ci_minutes_additional_packs references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
ci_namespace_mirrors references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
ci_namespace_monthly_usages references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
ci_partitions references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
next_run_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE ci_pipeline_schedules ALTER COLUMN next_run_at TYPE timestamptz;
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE ci_pipeline_schedules ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE ci_pipeline_schedules ALTER COLUMN updated_at TYPE timestamptz;
ci_project_metrics references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
ci_project_mirrors references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
ci_project_monthly_usages references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
ci_resources.partition_id.build_id may be NULL, so the relationship to p_ci_builds is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ci_runner_controller_runner_level_scopings_instance_type references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
ci_runner_machines references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
ci_runner_namespaces.runner_id may be NULL, so the relationship to group_type_ci_runners is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE ci_runner_projects ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE ci_runner_projects ALTER COLUMN updated_at TYPE timestamptz;
ci_runner_taggings references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
ci_runner_versions references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
ci_secure_file_states references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
ci_secure_files references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
ci_sources_pipelines.source_partition_id.source_job_id may be NULL, so the relationship to p_ci_builds is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ci_sources_pipelines.source_partition_id.source_pipeline_id may be NULL, so the relationship to p_ci_pipelines is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ci_sources_pipelines.partition_id.pipeline_id may be NULL, so the relationship to p_ci_pipelines is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ci_subscriptions_projects references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE ci_triggers ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE ci_triggers ALTER COLUMN updated_at TYPE timestamptz;
ci_triggers references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
ci_variables references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
cloud_connector_access references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
cloud_connector_keys references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
cluster_agent_migrations.issue_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cluster_agent_tokens.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cluster_agent_tokens.created_by_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cluster_agent_tokens has several nullable FKs (project_id, created_by_user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE cluster_agent_tokens ADD CONSTRAINT cluster_agent_tokens_one_target CHECK ((project_id IS NOT NULL)::int + (created_by_user_id IS NOT NULL)::int = 1);
cluster_agent_url_configurations.created_by_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cluster_agents.created_by_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE cluster_platforms_kubernetes ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE cluster_platforms_kubernetes ALTER COLUMN updated_at TYPE timestamptz;
cluster_platforms_kubernetes.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cluster_platforms_kubernetes.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cluster_platforms_kubernetes.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE cluster_projects ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE cluster_projects ALTER COLUMN updated_at TYPE timestamptz;
cluster_providers_aws.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cluster_providers_aws.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cluster_providers_aws.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE cluster_providers_gcp ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE cluster_providers_gcp ALTER COLUMN updated_at TYPE timestamptz;
cluster_providers_gcp.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cluster_providers_gcp.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
cluster_providers_gcp.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE clusters ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE clusters ALTER COLUMN updated_at TYPE timestamptz;
clusters.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
clusters.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
clusters.management_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
clusters.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
clusters.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
clusters has several nullable FKs (organization_id, group_id, management_project_id, project_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE clusters ADD CONSTRAINT clusters_one_target CHECK ((organization_id IS NOT NULL)::int + (group_id IS NOT NULL)::int + (management_project_id IS NOT NULL)::int + (project_id IS NOT NULL)::int + (user_id IS NOT NULL)::int = 1);
clusters_kubernetes_namespaces.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
clusters_kubernetes_namespaces.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
clusters_kubernetes_namespaces.sharding_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
clusters_kubernetes_namespaces.cluster_project_id may be NULL, so the relationship to cluster_projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
clusters_kubernetes_namespaces.environment_id may be NULL, so the relationship to environments is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
clusters_kubernetes_namespaces.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
organization_id, group_id, sharding_project_id, cluster_project_id, environment_id, project_idexclusive-arcclusters_kubernetes_namespaces has several nullable FKs (organization_id, group_id, sharding_project_id, cluster_project_id, environment_id, project_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE clusters_kubernetes_namespaces ADD CONSTRAINT clusters_kubernetes_namespaces_one_target CHECK ((organization_id IS NOT NULL)::int + (group_id IS NOT NULL)::int + (sharding_project_id IS NOT NULL)::int + (cluster_project_id IS NOT NULL)::int + (environment_id IS NOT NULL)::int + (project_id IS NOT NULL)::int = 1);
commit_user_mentions.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
compliance_framework_security_policies.security_policy_id may be NULL, so the relationship to security_policies is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
compliance_framework_security_policies.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
compliance_framework_security_policies.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
security_policy_id, namespace_id, project_idexclusive-arccompliance_framework_security_policies has several nullable FKs (security_policy_id, namespace_id, project_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE compliance_framework_security_policies ADD CONSTRAINT compliance_framework_security_policies_one_target CHECK ((security_policy_id IS NOT NULL)::int + (namespace_id IS NOT NULL)::int + (project_id IS NOT NULL)::int = 1);
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE container_repositories ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE container_repositories ALTER COLUMN updated_at TYPE timestamptz;
container_repository_states.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
content_blocked_states references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE conversational_development_index_metrics ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE conversational_development_index_metrics ALTER COLUMN updated_at TYPE timestamptz;
conversational_development_index_metrics references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
Tables this wide usually hide several entities (or a JSON column that wants to be one). Every row update rewrites the whole tuple; TOAST kicks in; indexes bloat.
Fix Look for column groups that always change together and split them out.
country_access_logs references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
coverage_fuzzing_corpuses.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
custom_dashboard_versions.updated_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
custom_dashboards.created_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
custom_dashboards.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
custom_dashboards.updated_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
custom_dashboards has several nullable FKs (created_by_id, namespace_id, updated_by_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE custom_dashboards ADD CONSTRAINT custom_dashboards_one_target CHECK ((created_by_id IS NOT NULL)::int + (namespace_id IS NOT NULL)::int + (updated_by_id IS NOT NULL)::int = 1);
custom_fields.created_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
custom_fields.updated_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
custom_fields has several nullable FKs (created_by_id, updated_by_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE custom_fields ADD CONSTRAINT custom_fields_one_target CHECK ((created_by_id IS NOT NULL)::int + (updated_by_id IS NOT NULL)::int = 1);
customer_relations_contacts.organization_id may be NULL, so the relationship to customer_relations_organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
dast_sites.dast_site_validation_id may be NULL, so the relationship to dast_site_validations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
dependency_firewall_activity_stats.dependency_firewall_policy_rule_id may be NULL, so the relationship to dependency_firewall_policy_rules is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE dependency_list_export_part_uploads ALTER COLUMN created_at TYPE timestamptz;
dependency_list_export_upload_states.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
dependency_list_export_upload_states.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
dependency_list_export_upload_states.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE dependency_list_export_uploads ALTER COLUMN created_at TYPE timestamptz;
dependency_proxy_blob_states references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
dependency_proxy_manifest_states.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE deploy_keys_projects ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE deploy_keys_projects ALTER COLUMN updated_at TYPE timestamptz;
deploy_tokens.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
deploy_tokens.creator_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
deploy_tokens.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
deploy_tokens has several nullable FKs (project_id, creator_id, group_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE deploy_tokens ADD CONSTRAINT deploy_tokens_one_target CHECK ((project_id IS NOT NULL)::int + (creator_id IS NOT NULL)::int + (group_id IS NOT NULL)::int = 1);
deployment_approvals.approval_rule_id may be NULL, so the relationship to protected_environment_approval_rules is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
deployment_approvals.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
deployment_approvals has several nullable FKs (approval_rule_id, project_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE deployment_approvals ADD CONSTRAINT deployment_approvals_one_target CHECK ((approval_rule_id IS NOT NULL)::int + (project_id IS NOT NULL)::int = 1);
deployment_clusters.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
deployment_merge_requests.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
deployment_merge_requests.environment_id may be NULL, so the relationship to environments is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
deployment_merge_requests has several nullable FKs (project_id, environment_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE deployment_merge_requests ADD CONSTRAINT deployment_merge_requests_one_target CHECK ((project_id IS NOT NULL)::int + (environment_id IS NOT NULL)::int = 1);
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE deployments ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE deployments ALTER COLUMN updated_at TYPE timestamptz;
description_versions.merge_request_id may be NULL, so the relationship to merge_requests is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
description_versions.issue_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
description_versions.epic_id may be NULL, so the relationship to epics is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE design_management_action_uploads ALTER COLUMN created_at TYPE timestamptz;
design_management_designs.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
design_management_designs.issue_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
design_management_designs has several nullable FKs (namespace_id, issue_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE design_management_designs ADD CONSTRAINT design_management_designs_one_target CHECK ((namespace_id IS NOT NULL)::int + (issue_id IS NOT NULL)::int = 1);
design_management_designs_versions.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
design_management_repositories.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
design_management_repository_states.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
design_management_versions.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
design_management_versions.author_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
design_management_versions.issue_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
design_management_versions has several nullable FKs (namespace_id, author_id, issue_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE design_management_versions ADD CONSTRAINT design_management_versions_one_target CHECK ((namespace_id IS NOT NULL)::int + (author_id IS NOT NULL)::int + (issue_id IS NOT NULL)::int = 1);
design_user_mentions.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
detached_partitions references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
diff_note_positions.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
dora_daily_metrics.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
draft_notes.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
duo_workflow_session_artifacts.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
duo_workflow_session_artifacts.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
duo_workflow_session_artifacts has several nullable FKs (project_id, namespace_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE duo_workflow_session_artifacts ADD CONSTRAINT duo_workflow_session_artifacts_one_target CHECK ((project_id IS NOT NULL)::int + (namespace_id IS NOT NULL)::int = 1);
duo_workflows_checkpoint_writes.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
duo_workflows_checkpoint_writes.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
duo_workflows_events.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
duo_workflows_events.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
duo_workflows_workflow_merge_requests.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
duo_workflows_workflow_merge_requests.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
duo_workflows_workflow_notes.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
duo_workflows_workflow_notes.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
duo_workflows_workflow_pipelines.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
duo_workflows_workflow_pipelines.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
duo_workflows_workflow_work_items.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
duo_workflows_workflow_work_items.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
Tables this wide usually hide several entities (or a JSON column that wants to be one). Every row update rewrites the whole tuple; TOAST kicks in; indexes bloat.
Fix Look for column groups that always change together and split them out.
duo_workflows_workflows.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
duo_workflows_workflows.ai_catalog_item_version_id may be NULL, so the relationship to ai_catalog_item_versions is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
duo_workflows_workflows.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
duo_workflows_workflows.issue_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
duo_workflows_workflows.ai_catalog_item_id may be NULL, so the relationship to ai_catalog_items is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
duo_workflows_workflows.service_account_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
duo_workflows_workflows.agent_identity_id may be NULL, so the relationship to ai_agent_identities is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
duo_workflows_workflows.merge_request_id may be NULL, so the relationship to merge_requests is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
project_id, ai_catalog_item_version_id, namespace_id, issue_id, ai_catalog_item_id, service_account_id, agent_identity_id, merge_request_idexclusive-arcduo_workflows_workflows has several nullable FKs (project_id, ai_catalog_item_version_id, namespace_id, issue_id, ai_catalog_item_id, service_account_id, agent_identity_id, merge_request_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE duo_workflows_workflows ADD CONSTRAINT duo_workflows_workflows_one_target CHECK ((project_id IS NOT NULL)::int + (ai_catalog_item_version_id IS NOT NULL)::int + (namespace_id IS NOT NULL)::int + (issue_id IS NOT NULL)::int + (ai_catalog_item_id IS NOT NULL)::int + (service_account_id IS NOT NULL)::int + (agent_identity_id IS NOT NULL)::int + (merge_request_id IS NOT NULL)::int = 1);
elastic_index_settings references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE emails ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE emails ALTER COLUMN updated_at TYPE timestamptz;
confirmed_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE emails ALTER COLUMN confirmed_at TYPE timestamptz;
confirmation_sent_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE emails ALTER COLUMN confirmation_sent_at TYPE timestamptz;
enabled_foundational_flows.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
enabled_foundational_flows.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE environments ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE environments ALTER COLUMN updated_at TYPE timestamptz;
environments.merge_request_id may be NULL, so the relationship to merge_requests is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
environments.cluster_agent_id may be NULL, so the relationship to cluster_agents is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
environments has several nullable FKs (merge_request_id, cluster_agent_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE environments ADD CONSTRAINT environments_one_target CHECK ((merge_request_id IS NOT NULL)::int + (cluster_agent_id IS NOT NULL)::int = 1);
epic_issues.work_item_parent_link_id may be NULL, so the relationship to work_item_parent_links is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
epic_issues.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
epic_issues has several nullable FKs (work_item_parent_link_id, namespace_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE epic_issues ADD CONSTRAINT epic_issues_one_target CHECK ((work_item_parent_link_id IS NOT NULL)::int + (namespace_id IS NOT NULL)::int = 1);
epic_user_mentions.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
epic_user_mentions.note_id may be NULL, so the relationship to notes is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
epic_user_mentions has several nullable FKs (group_id, note_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE epic_user_mentions ADD CONSTRAINT epic_user_mentions_one_target CHECK ((group_id IS NOT NULL)::int + (note_id IS NOT NULL)::int = 1);
last_edited_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE epics ALTER COLUMN last_edited_at TYPE timestamptz;
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE epics ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE epics ALTER COLUMN updated_at TYPE timestamptz;
closed_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE epics ALTER COLUMN closed_at TYPE timestamptz;
Tables this wide usually hide several entities (or a JSON column that wants to be one). Every row update rewrites the whole tuple; TOAST kicks in; indexes bloat.
Fix Look for column groups that always change together and split them out.
epics.due_date_sourcing_epic_id may be NULL, so the relationship to epics is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
epics.start_date_sourcing_milestone_id may be NULL, so the relationship to milestones is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
epics.due_date_sourcing_milestone_id may be NULL, so the relationship to milestones is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
epics.work_item_parent_link_id may be NULL, so the relationship to work_item_parent_links is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
epics.start_date_sourcing_epic_id may be NULL, so the relationship to epics is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
epics.closed_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
epics.assignee_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
epics.issue_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
epics.parent_id may be NULL, so the relationship to epics is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
due_date_sourcing_epic_id, start_date_sourcing_milestone_id, due_date_sourcing_milestone_id, work_item_parent_link_id, start_date_sourcing_epic_id, closed_by_id, assignee_id, issue_id, parent_idexclusive-arcepics has several nullable FKs (due_date_sourcing_epic_id, start_date_sourcing_milestone_id, due_date_sourcing_milestone_id, work_item_parent_link_id, start_date_sourcing_epic_id, closed_by_id, assignee_id, issue_id, parent_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE epics ADD CONSTRAINT epics_one_target CHECK ((due_date_sourcing_epic_id IS NOT NULL)::int + (start_date_sourcing_milestone_id IS NOT NULL)::int + (due_date_sourcing_milestone_id IS NOT NULL)::int + (work_item_parent_link_id IS NOT NULL)::int + (start_date_sourcing_epic_id IS NOT NULL)::int + (closed_by_id IS NOT NULL)::int + (assignee_id IS NOT NULL)::int + (issue_id IS NOT NULL)::int + (parent_id IS NOT NULL)::int = 1);
error_tracking_error_events.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
events.personal_namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
events.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
events.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
evidences.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
external_status_checks_protected_branches.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE feature_gates ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE feature_gates ALTER COLUMN updated_at TYPE timestamptz;
feature_gates references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE features ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE features ALTER COLUMN updated_at TYPE timestamptz;
features references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
fork_network_members.forked_from_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
fork_networks.root_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
geo_ci_job_artifact_verification_summaries references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE geo_event_log ALTER COLUMN created_at TYPE timestamptz;
geo_event_log.cache_invalidation_event_id may be NULL, so the relationship to geo_cache_invalidation_events is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
geo_event_log.geo_event_id may be NULL, so the relationship to geo_events is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
geo_event_log has several nullable FKs (cache_invalidation_event_id, geo_event_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE geo_event_log ADD CONSTRAINT geo_event_log_one_target CHECK ((cache_invalidation_event_id IS NOT NULL)::int + (geo_event_id IS NOT NULL)::int = 1);
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE geo_node_namespace_links ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE geo_node_namespace_links ALTER COLUMN updated_at TYPE timestamptz;
last_event_date stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE geo_node_statuses ALTER COLUMN last_event_date TYPE timestamptz;
cursor_last_event_date stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE geo_node_statuses ALTER COLUMN cursor_last_event_date TYPE timestamptz;
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE geo_node_statuses ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE geo_node_statuses ALTER COLUMN updated_at TYPE timestamptz;
last_successful_status_check_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE geo_node_statuses ALTER COLUMN last_successful_status_check_at TYPE timestamptz;
gitlab_subscription_histories references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
gitlab_subscriptions.hosted_plan_id may be NULL, so the relationship to plans is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
gitlab_subscriptions.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
gitlab_subscriptions has several nullable FKs (hosted_plan_id, namespace_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE gitlab_subscriptions ADD CONSTRAINT gitlab_subscriptions_one_target CHECK ((hosted_plan_id IS NOT NULL)::int + (namespace_id IS NOT NULL)::int = 1);
gpg_key_subkeys.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
gpg_keys.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
gpg_signatures.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
gpg_signatures.gpg_key_subkey_id may be NULL, so the relationship to gpg_key_subkeys is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
gpg_signatures.gpg_key_id may be NULL, so the relationship to gpg_keys is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
gpg_signatures has several nullable FKs (project_id, gpg_key_subkey_id, gpg_key_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE gpg_signatures ADD CONSTRAINT gpg_signatures_one_target CHECK ((project_id IS NOT NULL)::int + (gpg_key_subkey_id IS NOT NULL)::int + (gpg_key_id IS NOT NULL)::int = 1);
granular_scopes.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
group_audit_events references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
group_crm_settings.source_group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
group_group_links.member_role_id may be NULL, so the relationship to member_roles is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
group_import_states.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
group_secrets_manager_maintenance_tasks references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
group_secrets_managers.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
group_security_exclusions references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
group_wiki_repository_states.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
groups_visits references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE historical_data ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE historical_data ALTER COLUMN updated_at TYPE timestamptz;
historical_data references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE identities ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE identities ALTER COLUMN updated_at TYPE timestamptz;
identities.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
identities.saml_provider_id may be NULL, so the relationship to saml_providers is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
identities has several nullable FKs (user_id, saml_provider_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE identities ADD CONSTRAINT identities_one_target CHECK ((user_id IS NOT NULL)::int + (saml_provider_id IS NOT NULL)::int = 1);
import_export_upload_upload_states.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
import_export_upload_upload_states.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE import_export_upload_uploads ALTER COLUMN created_at TYPE timestamptz;
import_export_uploads.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
import_export_uploads.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
import_export_uploads.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
import_export_uploads has several nullable FKs (user_id, group_id, project_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE import_export_uploads ADD CONSTRAINT import_export_uploads_one_target CHECK ((user_id IS NOT NULL)::int + (group_id IS NOT NULL)::int + (project_id IS NOT NULL)::int = 1);
import_failures.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
import_failures.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
import_failures.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
import_failures.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
import_failures has several nullable FKs (group_id, organization_id, user_id, project_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE import_failures ADD CONSTRAINT import_failures_one_target CHECK ((group_id IS NOT NULL)::int + (organization_id IS NOT NULL)::int + (user_id IS NOT NULL)::int + (project_id IS NOT NULL)::int = 1);
import_offline_configurations.bulk_import_id may be NULL, so the relationship to bulk_imports is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
import_offline_configurations.offline_export_id may be NULL, so the relationship to import_offline_exports is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
import_placeholder_memberships.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
import_placeholder_memberships.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
import_placeholder_memberships has several nullable FKs (group_id, project_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE import_placeholder_memberships ADD CONSTRAINT import_placeholder_memberships_one_target CHECK ((group_id IS NOT NULL)::int + (project_id IS NOT NULL)::int = 1);
import_placeholder_user_details.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
reassigned_by_user_id, reassign_to_user_id, placeholder_user_idexclusive-arcimport_source_users has several nullable FKs (reassigned_by_user_id, reassign_to_user_id, placeholder_user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE import_source_users ADD CONSTRAINT import_source_users_one_target CHECK ((reassigned_by_user_id IS NOT NULL)::int + (reassign_to_user_id IS NOT NULL)::int + (placeholder_user_id IS NOT NULL)::int = 1);
import_source_users.reassigned_by_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
import_source_users.reassign_to_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
import_source_users.placeholder_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
incident_management_escalation_rules has several nullable FKs (user_id, project_id, oncall_schedule_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE incident_management_escalation_rules ADD CONSTRAINT incident_management_escalation_rules_one_target CHECK ((user_id IS NOT NULL)::int + (project_id IS NOT NULL)::int + (oncall_schedule_id IS NOT NULL)::int = 1);
incident_management_escalation_rules.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
incident_management_escalation_rules.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
incident_management_escalation_rules.oncall_schedule_id may be NULL, so the relationship to incident_management_oncall_schedules is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
incident_management_issuable_escalation_statuses has several nullable FKs (namespace_id, policy_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE incident_management_issuable_escalation_statuses ADD CONSTRAINT incident_management_issuable_escalation_statuses_one_target CHECK ((namespace_id IS NOT NULL)::int + (policy_id IS NOT NULL)::int = 1);
incident_management_issuable_escalation_statuses.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
incident_management_issuable_escalation_statuses.policy_id may be NULL, so the relationship to incident_management_escalation_policies is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
incident_management_oncall_participants.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
incident_management_oncall_rotations.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
incident_management_oncall_shifts.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
incident_management_pending_alert_escalations.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
incident_management_pending_issue_escalations.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
incident_management_timeline_event_tag_links.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
author_id, updated_by_user_id, promoted_from_note_idexclusive-arcincident_management_timeline_events has several nullable FKs (author_id, updated_by_user_id, promoted_from_note_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE incident_management_timeline_events ADD CONSTRAINT incident_management_timeline_events_one_target CHECK ((author_id IS NOT NULL)::int + (updated_by_user_id IS NOT NULL)::int + (promoted_from_note_id IS NOT NULL)::int = 1);
incident_management_timeline_events.author_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
incident_management_timeline_events.updated_by_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
incident_management_timeline_events.promoted_from_note_id may be NULL, so the relationship to notes is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
indexed_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE index_statuses ALTER COLUMN indexed_at TYPE timestamptz;
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE index_statuses ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE index_statuses ALTER COLUMN updated_at TYPE timestamptz;
instance_audit_events references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
instance_model_selection_feature_settings references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
integrations has several nullable FKs (project_id, organization_id, group_id, inherit_from_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE integrations ADD CONSTRAINT integrations_one_target CHECK ((project_id IS NOT NULL)::int + (organization_id IS NOT NULL)::int + (group_id IS NOT NULL)::int + (inherit_from_id IS NOT NULL)::int = 1);
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE integrations ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE integrations ALTER COLUMN updated_at TYPE timestamptz;
Tables this wide usually hide several entities (or a JSON column that wants to be one). Every row update rewrites the whole tuple; TOAST kicks in; indexes bloat.
Fix Look for column groups that always change together and split them out.
integrations.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
integrations.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
integrations.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
integrations.inherit_from_id may be NULL, so the relationship to integrations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
internal_ids.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
internal_ids.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE issuable_metric_image_uploads ALTER COLUMN created_at TYPE timestamptz;
issuable_metric_images.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
issuable_resource_links.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
issuable_severities.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
issuable_slas.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
issue_assignees.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
issue_assignment_events has several nullable FKs (namespace_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE issue_assignment_events ADD CONSTRAINT issue_assignment_events_one_target CHECK ((namespace_id IS NOT NULL)::int + (user_id IS NOT NULL)::int = 1);
issue_assignment_events.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
issue_assignment_events.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
issue_customer_relations_contacts.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
issue_email_participants.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
issue_emails.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE issue_links ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE issue_links ALTER COLUMN updated_at TYPE timestamptz;
issue_links.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
first_mentioned_in_commit_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE issue_metrics ALTER COLUMN first_mentioned_in_commit_at TYPE timestamptz;
first_associated_with_milestone_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE issue_metrics ALTER COLUMN first_associated_with_milestone_at TYPE timestamptz;
first_added_to_board_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE issue_metrics ALTER COLUMN first_added_to_board_at TYPE timestamptz;
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE issue_metrics ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE issue_metrics ALTER COLUMN updated_at TYPE timestamptz;
issue_metrics.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
issue_search_data.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
issue_search_data_00 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_01 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_02 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_03 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_04 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_05 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_06 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_07 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_08 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_09 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_10 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_11 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_12 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_13 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_14 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_15 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_16 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_17 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_18 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_19 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_20 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_21 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_22 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_23 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_24 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_25 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_26 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_27 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_28 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_29 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_30 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_31 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_32 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_33 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_34 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_35 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_36 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_37 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_38 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_39 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_40 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_41 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_42 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_43 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_44 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_45 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_46 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_47 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_48 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_49 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_50 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_51 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_52 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_53 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_54 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_55 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_56 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_57 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_58 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_59 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_60 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_61 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_62 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_search_data_63 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
issue_tracker_data has several nullable FKs (integration_id, group_id, organization_id, project_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE issue_tracker_data ADD CONSTRAINT issue_tracker_data_one_target CHECK ((integration_id IS NOT NULL)::int + (group_id IS NOT NULL)::int + (organization_id IS NOT NULL)::int + (project_id IS NOT NULL)::int = 1);
issue_tracker_data.integration_id may be NULL, so the relationship to integrations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
issue_tracker_data.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
issue_tracker_data.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
issue_tracker_data.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
issue_user_mentions has several nullable FKs (namespace_id, note_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE issue_user_mentions ADD CONSTRAINT issue_user_mentions_one_target CHECK ((namespace_id IS NOT NULL)::int + (note_id IS NOT NULL)::int = 1);
issue_user_mentions.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
issue_user_mentions.note_id may be NULL, so the relationship to notes is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
author_id, sprint_id, project_id, milestone_id, duplicated_to_id, moved_to_id, closed_by_id, namespace_id, promoted_to_epic_id, updated_by_idexclusive-arcissues has several nullable FKs (author_id, sprint_id, project_id, milestone_id, duplicated_to_id, moved_to_id, closed_by_id, namespace_id, promoted_to_epic_id, updated_by_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE issues ADD CONSTRAINT issues_one_target CHECK ((author_id IS NOT NULL)::int + (sprint_id IS NOT NULL)::int + (project_id IS NOT NULL)::int + (milestone_id IS NOT NULL)::int + (duplicated_to_id IS NOT NULL)::int + (moved_to_id IS NOT NULL)::int + (closed_by_id IS NOT NULL)::int + (namespace_id IS NOT NULL)::int + (promoted_to_epic_id IS NOT NULL)::int + (updated_by_id IS NOT NULL)::int = 1);
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE issues ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE issues ALTER COLUMN updated_at TYPE timestamptz;
last_edited_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE issues ALTER COLUMN last_edited_at TYPE timestamptz;
Tables this wide usually hide several entities (or a JSON column that wants to be one). Every row update rewrites the whole tuple; TOAST kicks in; indexes bloat.
Fix Look for column groups that always change together and split them out.
issues.author_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
issues.sprint_id may be NULL, so the relationship to sprints is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
issues.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
issues.milestone_id may be NULL, so the relationship to milestones is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
issues.duplicated_to_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
issues.moved_to_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
issues.closed_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
issues.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
issues.promoted_to_epic_id may be NULL, so the relationship to epics is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
issues.updated_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
jira_connect_installations.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
jira_imports has several nullable FKs (label_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE jira_imports ADD CONSTRAINT jira_imports_one_target CHECK ((label_id IS NOT NULL)::int + (user_id IS NOT NULL)::int = 1);
jira_imports.label_id may be NULL, so the relationship to labels is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
jira_imports.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
jira_tracker_data has several nullable FKs (organization_id, group_id, integration_id, project_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE jira_tracker_data ADD CONSTRAINT jira_tracker_data_one_target CHECK ((organization_id IS NOT NULL)::int + (group_id IS NOT NULL)::int + (integration_id IS NOT NULL)::int + (project_id IS NOT NULL)::int = 1);
Tables this wide usually hide several entities (or a JSON column that wants to be one). Every row update rewrites the whole tuple; TOAST kicks in; indexes bloat.
Fix Look for column groups that always change together and split them out.
jira_tracker_data.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
jira_tracker_data.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
jira_tracker_data.integration_id may be NULL, so the relationship to integrations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
jira_tracker_data.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
job_environments.deployment_id may be NULL, so the relationship to deployments is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE keys ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE keys ALTER COLUMN updated_at TYPE timestamptz;
last_used_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE keys ALTER COLUMN last_used_at TYPE timestamptz;
keys.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
label_links has several nullable FKs (namespace_id, label_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE label_links ADD CONSTRAINT label_links_one_target CHECK ((namespace_id IS NOT NULL)::int + (label_id IS NOT NULL)::int = 1);
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE label_links ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE label_links ALTER COLUMN updated_at TYPE timestamptz;
label_links.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
label_links.label_id may be NULL, so the relationship to labels is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE label_priorities ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE label_priorities ALTER COLUMN updated_at TYPE timestamptz;
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE labels ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE labels ALTER COLUMN updated_at TYPE timestamptz;
labels.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
labels.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
labels.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ldap_admin_role_links.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE ldap_group_links ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE ldap_group_links ALTER COLUMN updated_at TYPE timestamptz;
ldap_group_links.member_role_id may be NULL, so the relationship to member_roles is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE lfs_file_locks ALTER COLUMN created_at TYPE timestamptz;
lfs_object_states references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE lfs_objects ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE lfs_objects ALTER COLUMN updated_at TYPE timestamptz;
lfs_objects references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE lfs_objects_projects ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE lfs_objects_projects ALTER COLUMN updated_at TYPE timestamptz;
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE licenses ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE licenses ALTER COLUMN updated_at TYPE timestamptz;
licenses references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
custom_status_id, iteration_id, project_id, label_id, user_id, group_id, milestone_idexclusive-arclists has several nullable FKs (custom_status_id, iteration_id, project_id, label_id, user_id, group_id, milestone_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE lists ADD CONSTRAINT lists_one_target CHECK ((custom_status_id IS NOT NULL)::int + (iteration_id IS NOT NULL)::int + (project_id IS NOT NULL)::int + (label_id IS NOT NULL)::int + (user_id IS NOT NULL)::int + (group_id IS NOT NULL)::int + (milestone_id IS NOT NULL)::int = 1);
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE lists ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE lists ALTER COLUMN updated_at TYPE timestamptz;
lists.custom_status_id may be NULL, so the relationship to work_item_custom_statuses is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
lists.iteration_id may be NULL, so the relationship to sprints is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
lists.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
lists.label_id may be NULL, so the relationship to labels is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
lists.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
lists.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
lists.milestone_id may be NULL, so the relationship to milestones is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
loose_foreign_keys_deleted_records references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
loose_foreign_keys_namespace_deleted_records references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
loose_foreign_keys_organization_deleted_records references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
loose_foreign_keys_project_deleted_records references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
loose_foreign_keys_user_deleted_records references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
member_approvals has several nullable FKs (member_role_id, member_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE member_approvals ADD CONSTRAINT member_approvals_one_target CHECK ((member_role_id IS NOT NULL)::int + (member_id IS NOT NULL)::int = 1);
member_approvals.member_role_id may be NULL, so the relationship to member_roles is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
member_approvals.member_id may be NULL, so the relationship to members is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
member_roles.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
member_roles.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
members has several nullable FKs (member_namespace_id, member_role_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE members ADD CONSTRAINT members_one_target CHECK ((member_namespace_id IS NOT NULL)::int + (member_role_id IS NOT NULL)::int = 1);
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE members ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE members ALTER COLUMN updated_at TYPE timestamptz;
invite_accepted_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE members ALTER COLUMN invite_accepted_at TYPE timestamptz;
requested_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE members ALTER COLUMN requested_at TYPE timestamptz;
members.member_namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
members.member_role_id may be NULL, so the relationship to member_roles is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
merge_request_assignees.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
merge_request_assignment_events has several nullable FKs (project_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE merge_request_assignment_events ADD CONSTRAINT merge_request_assignment_events_one_target CHECK ((project_id IS NOT NULL)::int + (user_id IS NOT NULL)::int = 1);
merge_request_assignment_events.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
merge_request_assignment_events.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
merge_request_blocks.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
merge_request_cleanup_schedules.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
authored_date stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE merge_request_commits_metadata ALTER COLUMN authored_date TYPE timestamptz;
committed_date stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE merge_request_commits_metadata ALTER COLUMN committed_date TYPE timestamptz;
merge_request_commits_metadata references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
merge_request_context_commit_diff_files.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
merge_request_context_commits has several nullable FKs (project_id, merge_request_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE merge_request_context_commits ADD CONSTRAINT merge_request_context_commits_one_target CHECK ((project_id IS NOT NULL)::int + (merge_request_id IS NOT NULL)::int = 1);
merge_request_context_commits.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
merge_request_context_commits.merge_request_id may be NULL, so the relationship to merge_requests is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
authored_date stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE merge_request_diff_commits ALTER COLUMN authored_date TYPE timestamptz;
committed_date stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE merge_request_diff_commits ALTER COLUMN committed_date TYPE timestamptz;
merge_request_diff_commits references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
merge_request_diff_commits_b5377a7a34 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
merge_request_diff_details.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
merge_request_diff_files.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
merge_request_diff_files_99208b8fac references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE merge_request_diffs ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE merge_request_diffs ALTER COLUMN updated_at TYPE timestamptz;
merge_request_diffs.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
merge_request_metrics has several nullable FKs (target_project_id, merged_by_id, latest_closed_by_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE merge_request_metrics ADD CONSTRAINT merge_request_metrics_one_target CHECK ((target_project_id IS NOT NULL)::int + (merged_by_id IS NOT NULL)::int + (latest_closed_by_id IS NOT NULL)::int = 1);
latest_build_started_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE merge_request_metrics ALTER COLUMN latest_build_started_at TYPE timestamptz;
latest_build_finished_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE merge_request_metrics ALTER COLUMN latest_build_finished_at TYPE timestamptz;
first_deployed_to_production_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE merge_request_metrics ALTER COLUMN first_deployed_to_production_at TYPE timestamptz;
merged_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE merge_request_metrics ALTER COLUMN merged_at TYPE timestamptz;
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE merge_request_metrics ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE merge_request_metrics ALTER COLUMN updated_at TYPE timestamptz;
merge_request_metrics.target_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
merge_request_metrics.merged_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
merge_request_metrics.latest_closed_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
merge_request_predictions.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
merge_request_requested_changes references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
merge_request_reviewers.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
merge_request_user_mentions has several nullable FKs (project_id, note_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE merge_request_user_mentions ADD CONSTRAINT merge_request_user_mentions_one_target CHECK ((project_id IS NOT NULL)::int + (note_id IS NOT NULL)::int = 1);
merge_request_user_mentions.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
merge_request_user_mentions.note_id may be NULL, so the relationship to notes is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
latest_merge_request_diff_id, assignee_id, updated_by_id, milestone_id, merge_user_id, author_id, source_project_idexclusive-arcmerge_requests has several nullable FKs (latest_merge_request_diff_id, assignee_id, updated_by_id, milestone_id, merge_user_id, author_id, source_project_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE merge_requests ADD CONSTRAINT merge_requests_one_target CHECK ((latest_merge_request_diff_id IS NOT NULL)::int + (assignee_id IS NOT NULL)::int + (updated_by_id IS NOT NULL)::int + (milestone_id IS NOT NULL)::int + (merge_user_id IS NOT NULL)::int + (author_id IS NOT NULL)::int + (source_project_id IS NOT NULL)::int = 1);
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE merge_requests ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE merge_requests ALTER COLUMN updated_at TYPE timestamptz;
last_edited_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE merge_requests ALTER COLUMN last_edited_at TYPE timestamptz;
Tables this wide usually hide several entities (or a JSON column that wants to be one). Every row update rewrites the whole tuple; TOAST kicks in; indexes bloat.
Fix Look for column groups that always change together and split them out.
merge_requests.latest_merge_request_diff_id may be NULL, so the relationship to merge_request_diffs is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
merge_requests.assignee_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
merge_requests.updated_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
merge_requests.milestone_id may be NULL, so the relationship to milestones is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
merge_requests.merge_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
merge_requests.author_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
merge_requests.source_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
merge_requests_approval_rules has several nullable FKs (group_id, project_id, source_rule_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE merge_requests_approval_rules ADD CONSTRAINT merge_requests_approval_rules_one_target CHECK ((group_id IS NOT NULL)::int + (project_id IS NOT NULL)::int + (source_rule_id IS NOT NULL)::int = 1);
merge_requests_approval_rules.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
merge_requests_approval_rules.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
merge_requests_approval_rules.source_rule_id may be NULL, so the relationship to merge_requests_approval_rules is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
merge_requests_approval_rules_approver_users.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
merge_requests_approval_rules_approver_users.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE merge_requests_closing_issues ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE merge_requests_closing_issues ALTER COLUMN updated_at TYPE timestamptz;
merge_requests_closing_issues.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
merge_requests_compliance_violations.target_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
merge_requests_merge_data.merge_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
milestone_releases.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE milestones ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE milestones ALTER COLUMN updated_at TYPE timestamptz;
milestones.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
milestones.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ml_candidate_metadata.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ml_candidate_metrics has several nullable FKs (project_id, candidate_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE ml_candidate_metrics ADD CONSTRAINT ml_candidate_metrics_one_target CHECK ((project_id IS NOT NULL)::int + (candidate_id IS NOT NULL)::int = 1);
ml_candidate_metrics.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ml_candidate_metrics.candidate_id may be NULL, so the relationship to ml_candidates is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ml_candidate_params has several nullable FKs (project_id, candidate_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE ml_candidate_params ADD CONSTRAINT ml_candidate_params_one_target CHECK ((project_id IS NOT NULL)::int + (candidate_id IS NOT NULL)::int = 1);
ml_candidate_params.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ml_candidate_params.candidate_id may be NULL, so the relationship to ml_candidates is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ml_candidates has several nullable FKs (project_id, package_id, model_version_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE ml_candidates ADD CONSTRAINT ml_candidates_one_target CHECK ((project_id IS NOT NULL)::int + (package_id IS NOT NULL)::int + (model_version_id IS NOT NULL)::int + (user_id IS NOT NULL)::int = 1);
ml_candidates.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ml_candidates.package_id may be NULL, so the relationship to packages_packages is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ml_candidates.model_version_id may be NULL, so the relationship to ml_model_versions is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ml_candidates.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ml_experiment_metadata.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ml_experiments has several nullable FKs (user_id, model_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE ml_experiments ADD CONSTRAINT ml_experiments_one_target CHECK ((user_id IS NOT NULL)::int + (model_id IS NOT NULL)::int = 1);
ml_experiments.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ml_experiments.model_id may be NULL, so the relationship to ml_models is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ml_model_metadata.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ml_model_versions.package_id may be NULL, so the relationship to packages_packages is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ml_models.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
namespace_cluster_agent_mappings.creator_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
namespace_descendants references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_00 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_01 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_02 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_03 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_04 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_05 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_06 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_07 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_08 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_09 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_10 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_11 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_12 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_13 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_14 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_15 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_16 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_17 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_18 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_19 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_20 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_21 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_22 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_23 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_24 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_25 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_26 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_27 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_28 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_29 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_30 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
namespace_descendants_31 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
Tables this wide usually hide several entities (or a JSON column that wants to be one). Every row update rewrites the whole tuple; TOAST kicks in; indexes bloat.
Fix Look for column groups that always change together and split them out.
namespace_settings.default_compliance_framework_id may be NULL, so the relationship to compliance_management_frameworks is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
shared_runners_seconds_last_reset stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE namespace_statistics ALTER COLUMN shared_runners_seconds_last_reset TYPE timestamptz;
custom_project_templates_group_id, file_template_project_id, duo_template_project_idexclusive-arcnamespace_template_settings has several nullable FKs (custom_project_templates_group_id, file_template_project_id, duo_template_project_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE namespace_template_settings ADD CONSTRAINT namespace_template_settings_one_target CHECK ((custom_project_templates_group_id IS NOT NULL)::int + (file_template_project_id IS NOT NULL)::int + (duo_template_project_id IS NOT NULL)::int = 1);
namespace_template_settings.custom_project_templates_group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
namespace_template_settings.file_template_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
namespace_template_settings.duo_template_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE namespace_uploads ALTER COLUMN created_at TYPE timestamptz;
namespaces.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
namespaces.parent_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE namespaces ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE namespaces ALTER COLUMN updated_at TYPE timestamptz;
ldap_sync_last_update_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE namespaces ALTER COLUMN ldap_sync_last_update_at TYPE timestamptz;
ldap_sync_last_successful_update_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE namespaces ALTER COLUMN ldap_sync_last_successful_update_at TYPE timestamptz;
ldap_sync_last_sync_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE namespaces ALTER COLUMN ldap_sync_last_sync_at TYPE timestamptz;
Any change to its key, its delete semantics, or its partitioning touches every dependent. Migrations on hub tables need the longest lock windows and the most careful rollout.
Fix Treat schema changes here as breaking changes with a written rollout plan.
Tables this wide usually hide several entities (or a JSON column that wants to be one). Every row update rewrites the whole tuple; TOAST kicks in; indexes bloat.
Fix Look for column groups that always change together and split them out.
namespaces has several nullable FKs (organization_id, parent_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE namespaces ADD CONSTRAINT namespaces_one_target CHECK ((organization_id IS NOT NULL)::int + (parent_id IS NOT NULL)::int = 1);
note_diff_files.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
note_metadata.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
notes has several nullable FKs (review_id, namespace_id, project_id, organization_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE notes ADD CONSTRAINT notes_one_target CHECK ((review_id IS NOT NULL)::int + (namespace_id IS NOT NULL)::int + (project_id IS NOT NULL)::int + (organization_id IS NOT NULL)::int = 1);
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE notes ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE notes ALTER COLUMN updated_at TYPE timestamptz;
resolved_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE notes ALTER COLUMN resolved_at TYPE timestamptz;
Tables this wide usually hide several entities (or a JSON column that wants to be one). Every row update rewrites the whole tuple; TOAST kicks in; indexes bloat.
Fix Look for column groups that always change together and split them out.
notes.review_id may be NULL, so the relationship to reviews is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
notes.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
notes.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
notes.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE notification_settings ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE notification_settings ALTER COLUMN updated_at TYPE timestamptz;
notification_settings references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
Tables this wide usually hide several entities (or a JSON column that wants to be one). Every row update rewrites the whole tuple; TOAST kicks in; indexes bloat.
Fix Look for column groups that always change together and split them out.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE oauth_access_grants ALTER COLUMN created_at TYPE timestamptz;
revoked_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE oauth_access_grants ALTER COLUMN revoked_at TYPE timestamptz;
revoked_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE oauth_access_tokens ALTER COLUMN revoked_at TYPE timestamptz;
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE oauth_access_tokens ALTER COLUMN created_at TYPE timestamptz;
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE oauth_applications ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE oauth_applications ALTER COLUMN updated_at TYPE timestamptz;
oauth_applications.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
observability_metrics_issues_connections.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
operations_feature_flags_issues.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
operations_scopes.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
operations_strategies.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
operations_strategies_user_lists.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
organization_cluster_agent_mappings.creator_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE organization_detail_uploads ALTER COLUMN created_at TYPE timestamptz;
p_ai_active_context_code_repositories references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_batched_git_ref_updates_deletions references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_catalog_resource_sync_events references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_build_trace_metadata.partition_id.trace_artifact_id may be NULL, so the relationship to p_ci_job_artifacts is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
p_ci_builds.partition_id.stage_id may be NULL, so the relationship to p_ci_stages is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
p_ci_builds.resource_group_id may be NULL, so the relationship to ci_resource_groups is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
p_ci_builds.upstream_pipeline_partition_id.upstream_pipeline_id may be NULL, so the relationship to p_ci_pipelines is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
p_ci_builds.auto_canceled_by_partition_id.auto_canceled_by_id may be NULL, so the relationship to p_ci_pipelines is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
p_ci_builds.partition_id.commit_id may be NULL, so the relationship to p_ci_pipelines is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
finished_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE p_ci_builds ALTER COLUMN finished_at TYPE timestamptz;
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE p_ci_builds ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE p_ci_builds ALTER COLUMN updated_at TYPE timestamptz;
started_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE p_ci_builds ALTER COLUMN started_at TYPE timestamptz;
erased_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE p_ci_builds ALTER COLUMN erased_at TYPE timestamptz;
artifacts_expire_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE p_ci_builds ALTER COLUMN artifacts_expire_at TYPE timestamptz;
queued_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE p_ci_builds ALTER COLUMN queued_at TYPE timestamptz;
Tables this wide usually hide several entities (or a JSON column that wants to be one). Every row update rewrites the whole tuple; TOAST kicks in; indexes bloat.
Fix Look for column groups that always change together and split them out.
p_ci_builds_partition_overrides_0 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_builds_partition_overrides_1 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_builds_partition_overrides_2 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
build_finished_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE p_ci_finished_build_ch_sync_events ALTER COLUMN build_finished_at TYPE timestamptz;
p_ci_finished_build_ch_sync_events references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
pipeline_finished_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE p_ci_finished_pipeline_ch_sync_events ALTER COLUMN pipeline_finished_at TYPE timestamptz;
p_ci_finished_pipeline_ch_sync_events references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_job_runtime_environments references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_artifact_states references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_00 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_01 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_02 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_03 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_04 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_05 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_06 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_07 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_08 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_09 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_10 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_11 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_12 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_13 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_14 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_15 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_16 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_17 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_18 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_19 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_20 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_21 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_22 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_23 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_24 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_25 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_26 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_27 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_28 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_29 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_30 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_31 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_32 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_33 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_34 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_35 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_36 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_37 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_38 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_39 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_40 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_41 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_42 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_43 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_44 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_45 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_46 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_47 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_48 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_49 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_50 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_51 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_52 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_53 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_54 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_55 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_56 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_57 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_58 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_59 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_60 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_61 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_62 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipeline_iids_63 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_pipelines.external_pull_request_id may be NULL, so the relationship to external_pull_requests is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
p_ci_pipelines.auto_canceled_by_partition_id.auto_canceled_by_id may be NULL, so the relationship to p_ci_pipelines is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
p_ci_pipelines.pipeline_schedule_id may be NULL, so the relationship to ci_pipeline_schedules is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
p_ci_pipelines.ci_ref_id may be NULL, so the relationship to ci_refs is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE p_ci_pipelines ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE p_ci_pipelines ALTER COLUMN updated_at TYPE timestamptz;
committed_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE p_ci_pipelines ALTER COLUMN committed_at TYPE timestamptz;
started_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE p_ci_pipelines ALTER COLUMN started_at TYPE timestamptz;
finished_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE p_ci_pipelines ALTER COLUMN finished_at TYPE timestamptz;
Tables this wide usually hide several entities (or a JSON column that wants to be one). Every row update rewrites the whole tuple; TOAST kicks in; indexes bloat.
Fix Look for column groups that always change together and split them out.
p_ci_pipelines has several nullable FKs (external_pull_request_id, pipeline_schedule_id, ci_ref_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE p_ci_pipelines ADD CONSTRAINT p_ci_pipelines_one_target CHECK ((external_pull_request_id IS NOT NULL)::int + (pipeline_schedule_id IS NOT NULL)::int + (ci_ref_id IS NOT NULL)::int = 1);
p_ci_runtime_environments references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_ci_stages.partition_id.pipeline_id may be NULL, so the relationship to p_ci_pipelines is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE p_ci_stages ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE p_ci_stages ALTER COLUMN updated_at TYPE timestamptz;
p_ci_workload_variable_inclusions.partition_id.workload_id may be NULL, so the relationship to p_ci_workloads is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
p_knowledge_graph_code_indexing_tasks references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
p_sent_notifications.issue_email_participant_id may be NULL, so the relationship to issue_email_participants is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_build_infos.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_composer_metadata.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_composer_packages.creator_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
project_id, package_revision_id, package_reference_id, recipe_revision_idexclusive-arcpackages_conan_file_metadata has several nullable FKs (project_id, package_revision_id, package_reference_id, recipe_revision_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE packages_conan_file_metadata ADD CONSTRAINT packages_conan_file_metadata_one_target CHECK ((project_id IS NOT NULL)::int + (package_revision_id IS NOT NULL)::int + (package_reference_id IS NOT NULL)::int + (recipe_revision_id IS NOT NULL)::int = 1);
packages_conan_file_metadata.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_conan_file_metadata.package_revision_id may be NULL, so the relationship to packages_conan_package_revisions is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_conan_file_metadata.package_reference_id may be NULL, so the relationship to packages_conan_package_references is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_conan_file_metadata.recipe_revision_id may be NULL, so the relationship to packages_conan_recipe_revisions is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_conan_jwt_signing_keys references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
packages_conan_metadata.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_conan_package_references.recipe_revision_id may be NULL, so the relationship to packages_conan_recipe_revisions is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_debian_file_metadata.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_debian_group_architectures.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_debian_group_component_files has several nullable FKs (group_id, architecture_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE packages_debian_group_component_files ADD CONSTRAINT packages_debian_group_component_files_one_target CHECK ((group_id IS NOT NULL)::int + (architecture_id IS NOT NULL)::int = 1);
packages_debian_group_component_files.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_debian_group_component_files.architecture_id may be NULL, so the relationship to packages_debian_group_architectures is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_debian_group_components.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_debian_group_distribution_keys.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_debian_group_distributions.creator_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_debian_project_architectures.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_debian_project_component_files has several nullable FKs (project_id, architecture_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE packages_debian_project_component_files ADD CONSTRAINT packages_debian_project_component_files_one_target CHECK ((project_id IS NOT NULL)::int + (architecture_id IS NOT NULL)::int = 1);
packages_debian_project_component_files.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_debian_project_component_files.architecture_id may be NULL, so the relationship to packages_debian_project_architectures is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_debian_project_components.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_debian_project_distribution_keys.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_debian_project_distributions.creator_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_debian_publications.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_dependencies.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_dependency_links.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_helm_file_metadata.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_helm_metadata_cache_states.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_maven_metadata.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_npm_metadata.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_npm_metadata_caches references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
packages_nuget_dependency_link_metadata.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_nuget_metadata.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_nuget_symbol_states references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
packages_nuget_symbols.package_id may be NULL, so the relationship to packages_packages is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_package_file_build_infos.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_package_file_states references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
packages_packages.creator_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_pypi_metadata.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_rpm_metadata.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_rubygems_metadata.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
packages_rubygems_spec_files references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
packages_tags.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
pages_deployment_states.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
pages_domain_acme_orders.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
pages_domains.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
path_locks has several nullable FKs (project_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE path_locks ADD CONSTRAINT path_locks_one_target CHECK ((project_id IS NOT NULL)::int + (user_id IS NOT NULL)::int = 1);
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE path_locks ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE path_locks ALTER COLUMN updated_at TYPE timestamptz;
path_locks.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
path_locks.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
personal_access_tokens has several nullable FKs (group_id, previous_personal_access_token_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE personal_access_tokens ADD CONSTRAINT personal_access_tokens_one_target CHECK ((group_id IS NOT NULL)::int + (previous_personal_access_token_id IS NOT NULL)::int = 1);
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE personal_access_tokens ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE personal_access_tokens ALTER COLUMN updated_at TYPE timestamptz;
personal_access_tokens.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
personal_access_tokens.previous_personal_access_token_id may be NULL, so the relationship to personal_access_tokens is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
Tables this wide usually hide several entities (or a JSON column that wants to be one). Every row update rewrites the whole tuple; TOAST kicks in; indexes bloat.
Fix Look for column groups that always change together and split them out.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE plans ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE plans ALTER COLUMN updated_at TYPE timestamptz;
pm_checkpoints references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
pm_cve_enrichment references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
pool_repositories.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
postgres_async_foreign_key_validations references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
postgres_async_indexes references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
postgres_reindex_actions references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
postgres_reindex_queued_actions references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
programming_languages references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
project_audit_events references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
project_compliance_framework_settings.framework_id may be NULL, so the relationship to compliance_management_frameworks is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
project_compliance_standards_adherence.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
project_control_compliance_statuses.requirement_status_id may be NULL, so the relationship to project_requirement_compliance_statuses is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
project_data_transfers references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
project_export_jobs.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
jira_dvcs_cloud_last_sync_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE project_feature_usages ALTER COLUMN jira_dvcs_cloud_last_sync_at TYPE timestamptz;
jira_dvcs_server_last_sync_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE project_feature_usages ALTER COLUMN jira_dvcs_server_last_sync_at TYPE timestamptz;
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE project_features ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE project_features ALTER COLUMN updated_at TYPE timestamptz;
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE project_group_links ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE project_group_links ALTER COLUMN updated_at TYPE timestamptz;
project_group_links.member_role_id may be NULL, so the relationship to member_roles is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE project_import_export_relation_export_upload_uploads ALTER COLUMN created_at TYPE timestamptz;
last_update_started_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE project_mirror_data ALTER COLUMN last_update_started_at TYPE timestamptz;
last_update_scheduled_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE project_mirror_data ALTER COLUMN last_update_scheduled_at TYPE timestamptz;
next_execution_timestamp stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE project_mirror_data ALTER COLUMN next_execution_timestamp TYPE timestamptz;
project_relation_export_uploads.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
project_relation_exports.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
organization_id, project_secrets_manager_idexclusive-arcproject_secrets_manager_maintenance_tasks has several nullable FKs (organization_id, project_secrets_manager_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE project_secrets_manager_maintenance_tasks ADD CONSTRAINT project_secrets_manager_maintenance_tasks_one_target CHECK ((organization_id IS NOT NULL)::int + (project_secrets_manager_id IS NOT NULL)::int = 1);
project_secrets_manager_maintenance_tasks.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
project_secrets_manager_maintenance_tasks.project_secrets_manager_id may be NULL, so the relationship to project_secrets_managers is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
project_secrets_managers.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
project_security_exclusions references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
project_security_statistics references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
pipeline_execution_policy_bot_access_group_id, duo_dependency_bump_breaking_changes_enabled_by_id, push_rule_idexclusive-arcproject_settings has several nullable FKs (pipeline_execution_policy_bot_access_group_id, duo_dependency_bump_breaking_changes_enabled_by_id, push_rule_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE project_settings ADD CONSTRAINT project_settings_one_target CHECK ((pipeline_execution_policy_bot_access_group_id IS NOT NULL)::int + (duo_dependency_bump_breaking_changes_enabled_by_id IS NOT NULL)::int + (push_rule_id IS NOT NULL)::int = 1);
Tables this wide usually hide several entities (or a JSON column that wants to be one). Every row update rewrites the whole tuple; TOAST kicks in; indexes bloat.
Fix Look for column groups that always change together and split them out.
project_settings.pipeline_execution_policy_bot_access_group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
project_settings.duo_dependency_bump_breaking_changes_enabled_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
project_settings.push_rule_id may be NULL, so the relationship to push_rules is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
project_states references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
shared_runners_seconds_last_reset stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE project_statistics ALTER COLUMN shared_runners_seconds_last_reset TYPE timestamptz;
project_statistics.root_namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE project_topic_uploads ALTER COLUMN created_at TYPE timestamptz;
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE project_uploads ALTER COLUMN created_at TYPE timestamptz;
projects.project_namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
projects.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE projects ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE projects ALTER COLUMN updated_at TYPE timestamptz;
last_activity_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE projects ALTER COLUMN last_activity_at TYPE timestamptz;
mirror_last_update_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE projects ALTER COLUMN mirror_last_update_at TYPE timestamptz;
mirror_last_successful_update_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE projects ALTER COLUMN mirror_last_successful_update_at TYPE timestamptz;
last_repository_check_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE projects ALTER COLUMN last_repository_check_at TYPE timestamptz;
last_repository_updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE projects ALTER COLUMN last_repository_updated_at TYPE timestamptz;
Any change to its key, its delete semantics, or its partitioning touches every dependent. Migrations on hub tables need the longest lock windows and the most careful rollout.
Fix Treat schema changes here as breaking changes with a written rollout plan.
Tables this wide usually hide several entities (or a JSON column that wants to be one). Every row update rewrites the whole tuple; TOAST kicks in; indexes bloat.
Fix Look for column groups that always change together and split them out.
projects has several nullable FKs (project_namespace_id, organization_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE projects ADD CONSTRAINT projects_one_target CHECK ((project_namespace_id IS NOT NULL)::int + (organization_id IS NOT NULL)::int = 1);
projects_visits references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
protected_branch_project_id, group_id, member_role_id, protected_branch_namespace_id, user_idexclusive-arcprotected_branch_merge_access_levels has several nullable FKs (protected_branch_project_id, group_id, member_role_id, protected_branch_namespace_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE protected_branch_merge_access_levels ADD CONSTRAINT protected_branch_merge_access_levels_one_target CHECK ((protected_branch_project_id IS NOT NULL)::int + (group_id IS NOT NULL)::int + (member_role_id IS NOT NULL)::int + (protected_branch_namespace_id IS NOT NULL)::int + (user_id IS NOT NULL)::int = 1);
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE protected_branch_merge_access_levels ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE protected_branch_merge_access_levels ALTER COLUMN updated_at TYPE timestamptz;
protected_branch_merge_access_levels.protected_branch_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
protected_branch_merge_access_levels.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
protected_branch_merge_access_levels.member_role_id may be NULL, so the relationship to member_roles is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
protected_branch_merge_access_levels.protected_branch_namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
protected_branch_merge_access_levels.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
deploy_key_id, protected_branch_namespace_id, group_id, member_role_id, protected_branch_project_id, user_idexclusive-arcprotected_branch_push_access_levels has several nullable FKs (deploy_key_id, protected_branch_namespace_id, group_id, member_role_id, protected_branch_project_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE protected_branch_push_access_levels ADD CONSTRAINT protected_branch_push_access_levels_one_target CHECK ((deploy_key_id IS NOT NULL)::int + (protected_branch_namespace_id IS NOT NULL)::int + (group_id IS NOT NULL)::int + (member_role_id IS NOT NULL)::int + (protected_branch_project_id IS NOT NULL)::int + (user_id IS NOT NULL)::int = 1);
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE protected_branch_push_access_levels ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE protected_branch_push_access_levels ALTER COLUMN updated_at TYPE timestamptz;
protected_branch_push_access_levels.deploy_key_id may be NULL, so the relationship to keys is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
protected_branch_push_access_levels.protected_branch_namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
protected_branch_push_access_levels.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
protected_branch_push_access_levels.member_role_id may be NULL, so the relationship to member_roles is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
protected_branch_push_access_levels.protected_branch_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
protected_branch_push_access_levels.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
protected_branch_project_id, protected_branch_namespace_id, member_role_id, user_id, group_idexclusive-arcprotected_branch_unprotect_access_levels has several nullable FKs (protected_branch_project_id, protected_branch_namespace_id, member_role_id, user_id, group_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE protected_branch_unprotect_access_levels ADD CONSTRAINT protected_branch_unprotect_access_levels_one_target CHECK ((protected_branch_project_id IS NOT NULL)::int + (protected_branch_namespace_id IS NOT NULL)::int + (member_role_id IS NOT NULL)::int + (user_id IS NOT NULL)::int + (group_id IS NOT NULL)::int = 1);
protected_branch_unprotect_access_levels.protected_branch_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
protected_branch_unprotect_access_levels.protected_branch_namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
protected_branch_unprotect_access_levels.member_role_id may be NULL, so the relationship to member_roles is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
protected_branch_unprotect_access_levels.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
protected_branch_unprotect_access_levels.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE protected_branches ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE protected_branches ALTER COLUMN updated_at TYPE timestamptz;
protected_branches.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
protected_branches.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
protected_environment_group_id, group_id, user_id, protected_environment_project_idexclusive-arcprotected_environment_approval_rules has several nullable FKs (protected_environment_group_id, group_id, user_id, protected_environment_project_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE protected_environment_approval_rules ADD CONSTRAINT protected_environment_approval_rules_one_target CHECK ((protected_environment_group_id IS NOT NULL)::int + (group_id IS NOT NULL)::int + (user_id IS NOT NULL)::int + (protected_environment_project_id IS NOT NULL)::int = 1);
protected_environment_approval_rules.protected_environment_group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
protected_environment_approval_rules.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
protected_environment_approval_rules.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
protected_environment_approval_rules.protected_environment_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
protected_environment_group_id, protected_environment_project_id, group_id, user_idexclusive-arcprotected_environment_deploy_access_levels has several nullable FKs (protected_environment_group_id, protected_environment_project_id, group_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE protected_environment_deploy_access_levels ADD CONSTRAINT protected_environment_deploy_access_levels_one_target CHECK ((protected_environment_group_id IS NOT NULL)::int + (protected_environment_project_id IS NOT NULL)::int + (group_id IS NOT NULL)::int + (user_id IS NOT NULL)::int = 1);
protected_environment_deploy_access_levels.protected_environment_group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
protected_environment_deploy_access_levels.protected_environment_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
protected_environment_deploy_access_levels.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
protected_environment_deploy_access_levels.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
protected_environments.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
protected_environments.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
protected_tag_create_access_levels has several nullable FKs (deploy_key_id, project_id, group_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE protected_tag_create_access_levels ADD CONSTRAINT protected_tag_create_access_levels_one_target CHECK ((deploy_key_id IS NOT NULL)::int + (project_id IS NOT NULL)::int + (group_id IS NOT NULL)::int + (user_id IS NOT NULL)::int = 1);
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE protected_tag_create_access_levels ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE protected_tag_create_access_levels ALTER COLUMN updated_at TYPE timestamptz;
protected_tag_create_access_levels.deploy_key_id may be NULL, so the relationship to keys is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
protected_tag_create_access_levels.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
protected_tag_create_access_levels.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
protected_tag_create_access_levels.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE protected_tags ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE protected_tags ALTER COLUMN updated_at TYPE timestamptz;
push_event_payloads.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE push_rules ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE push_rules ALTER COLUMN updated_at TYPE timestamptz;
push_rules.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE redirect_routes ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE redirect_routes ALTER COLUMN updated_at TYPE timestamptz;
redirect_routes.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
related_epic_links has several nullable FKs (group_id, issue_link_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE related_epic_links ADD CONSTRAINT related_epic_links_one_target CHECK ((group_id IS NOT NULL)::int + (issue_link_id IS NOT NULL)::int = 1);
related_epic_links.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
related_epic_links.issue_link_id may be NULL, so the relationship to issue_links is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
release_links.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
releases has several nullable FKs (project_id, author_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE releases ADD CONSTRAINT releases_one_target CHECK ((project_id IS NOT NULL)::int + (author_id IS NOT NULL)::int = 1);
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE releases ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE releases ALTER COLUMN updated_at TYPE timestamptz;
releases.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
releases.author_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
last_update_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE remote_mirrors ALTER COLUMN last_update_at TYPE timestamptz;
last_successful_update_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE remote_mirrors ALTER COLUMN last_successful_update_at TYPE timestamptz;
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE remote_mirrors ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE remote_mirrors ALTER COLUMN updated_at TYPE timestamptz;
last_update_started_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE remote_mirrors ALTER COLUMN last_update_started_at TYPE timestamptz;
remote_mirrors.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
required_code_owners_sections.protected_branch_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
required_code_owners_sections.protected_branch_namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
requirements.issue_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
requirements_management_test_reports has several nullable FKs (project_id, issue_id, author_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE requirements_management_test_reports ADD CONSTRAINT requirements_management_test_reports_one_target CHECK ((project_id IS NOT NULL)::int + (issue_id IS NOT NULL)::int + (author_id IS NOT NULL)::int = 1);
requirements_management_test_reports.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
requirements_management_test_reports.issue_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
requirements_management_test_reports.author_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
triggered_by_id, merge_request_id, issue_id, iteration_idexclusive-arcresource_iteration_events has several nullable FKs (triggered_by_id, merge_request_id, issue_id, iteration_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE resource_iteration_events ADD CONSTRAINT resource_iteration_events_one_target CHECK ((triggered_by_id IS NOT NULL)::int + (merge_request_id IS NOT NULL)::int + (issue_id IS NOT NULL)::int + (iteration_id IS NOT NULL)::int = 1);
resource_iteration_events.triggered_by_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
resource_iteration_events.merge_request_id may be NULL, so the relationship to merge_requests is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
resource_iteration_events.issue_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
resource_iteration_events.iteration_id may be NULL, so the relationship to sprints is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
resource_label_events has several nullable FKs (issue_id, epic_id, merge_request_id, label_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE resource_label_events ADD CONSTRAINT resource_label_events_one_target CHECK ((issue_id IS NOT NULL)::int + (epic_id IS NOT NULL)::int + (merge_request_id IS NOT NULL)::int + (label_id IS NOT NULL)::int + (user_id IS NOT NULL)::int = 1);
resource_label_events.issue_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
resource_label_events.epic_id may be NULL, so the relationship to epics is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
resource_label_events.merge_request_id may be NULL, so the relationship to merge_requests is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
resource_label_events.label_id may be NULL, so the relationship to labels is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
resource_label_events.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
resource_link_events has several nullable FKs (system_note_metadata_id, namespace_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE resource_link_events ADD CONSTRAINT resource_link_events_one_target CHECK ((system_note_metadata_id IS NOT NULL)::int + (namespace_id IS NOT NULL)::int = 1);
resource_link_events.system_note_metadata_id may be NULL, so the relationship to system_note_metadata is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
resource_link_events.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
resource_milestone_events has several nullable FKs (merge_request_id, issue_id, milestone_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE resource_milestone_events ADD CONSTRAINT resource_milestone_events_one_target CHECK ((merge_request_id IS NOT NULL)::int + (issue_id IS NOT NULL)::int + (milestone_id IS NOT NULL)::int + (user_id IS NOT NULL)::int = 1);
resource_milestone_events.merge_request_id may be NULL, so the relationship to merge_requests is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
resource_milestone_events.issue_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
resource_milestone_events.milestone_id may be NULL, so the relationship to milestones is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
resource_milestone_events.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
issue_id, merge_request_id, source_merge_request_id, epic_id, user_idexclusive-arcresource_state_events has several nullable FKs (issue_id, merge_request_id, source_merge_request_id, epic_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE resource_state_events ADD CONSTRAINT resource_state_events_one_target CHECK ((issue_id IS NOT NULL)::int + (merge_request_id IS NOT NULL)::int + (source_merge_request_id IS NOT NULL)::int + (epic_id IS NOT NULL)::int + (user_id IS NOT NULL)::int = 1);
resource_state_events.issue_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
resource_state_events.merge_request_id may be NULL, so the relationship to merge_requests is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
resource_state_events.source_merge_request_id may be NULL, so the relationship to merge_requests is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
resource_state_events.epic_id may be NULL, so the relationship to epics is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
resource_state_events.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
resource_weight_events has several nullable FKs (namespace_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE resource_weight_events ADD CONSTRAINT resource_weight_events_one_target CHECK ((namespace_id IS NOT NULL)::int + (user_id IS NOT NULL)::int = 1);
resource_weight_events.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
resource_weight_events.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
reviews.author_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE routes ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE routes ALTER COLUMN updated_at TYPE timestamptz;
routes.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
saml_group_links.member_role_id may be NULL, so the relationship to member_roles is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
saml_providers.member_role_id may be NULL, so the relationship to member_roles is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
saved_views has several nullable FKs (created_by_id, updated_by_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE saved_views ADD CONSTRAINT saved_views_one_target CHECK ((created_by_id IS NOT NULL)::int + (updated_by_id IS NOT NULL)::int = 1);
saved_views.created_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
saved_views.updated_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
sbom_occurrences has several nullable FKs (component_version_id, source_package_id, source_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE sbom_occurrences ADD CONSTRAINT sbom_occurrences_one_target CHECK ((component_version_id IS NOT NULL)::int + (source_package_id IS NOT NULL)::int + (source_id IS NOT NULL)::int = 1);
sbom_occurrences.component_version_id may be NULL, so the relationship to sbom_component_versions is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
sbom_occurrences.source_package_id may be NULL, so the relationship to sbom_source_packages is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
sbom_occurrences.source_id may be NULL, so the relationship to sbom_sources is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
sbom_occurrences_vulnerabilities.sbom_occurrence_ref_id may be NULL, so the relationship to sbom_occurrence_refs is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
sbom_vulnerability_scans.sbom_vulnerability_scan_result_id may be NULL, so the relationship to sbom_vulnerability_scan_results is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
scan_result_policies has several nullable FKs (approval_policy_rule_id, project_id, namespace_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE scan_result_policies ADD CONSTRAINT scan_result_policies_one_target CHECK ((approval_policy_rule_id IS NOT NULL)::int + (project_id IS NOT NULL)::int + (namespace_id IS NOT NULL)::int = 1);
scan_result_policies.approval_policy_rule_id may be NULL, so the relationship to approval_policy_rules is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
scan_result_policies.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
scan_result_policies.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
scan_result_policy_violations.approval_policy_rule_id may be NULL, so the relationship to approval_policy_rules is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
scan_result_policy_violations.scan_result_policy_id may be NULL, so the relationship to scan_result_policies is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
schema_migrations references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
scim_identities.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
scim_oauth_access_tokens has several nullable FKs (organization_id, group_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE scim_oauth_access_tokens ADD CONSTRAINT scim_oauth_access_tokens_one_target CHECK ((organization_id IS NOT NULL)::int + (group_id IS NOT NULL)::int = 1);
scim_oauth_access_tokens.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
scim_oauth_access_tokens.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
secrets_management_recovery_keys references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
security_finding_enrichments.vulnerability_id may be NULL, so the relationship to vulnerabilities is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
security_finding_token_statuses references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
security_inventory_filters references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
security_orchestration_policy_configurations.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
security_orchestration_policy_configurations.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
security_orchestration_policy_rule_schedules has several nullable FKs (namespace_id, project_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE security_orchestration_policy_rule_schedules ADD CONSTRAINT security_orchestration_policy_rule_schedules_one_target CHECK ((namespace_id IS NOT NULL)::int + (project_id IS NOT NULL)::int + (user_id IS NOT NULL)::int = 1);
security_orchestration_policy_rule_schedules.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
security_orchestration_policy_rule_schedules.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
security_orchestration_policy_rule_schedules.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
security_policy_dismissals has several nullable FKs (security_policy_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE security_policy_dismissals ADD CONSTRAINT security_policy_dismissals_one_target CHECK ((security_policy_id IS NOT NULL)::int + (user_id IS NOT NULL)::int = 1);
security_policy_dismissals.security_policy_id may be NULL, so the relationship to security_policies is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
security_policy_dismissals.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
security_policy_settings.csp_namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
security_scan_profile_triggers.security_scan_profile_configuration_id may be NULL, so the relationship to security_scan_profile_configurations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
sentry_issues.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
service_access_tokens references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
service_desk_custom_email_verifications.triggerer_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
service_desk_settings.file_template_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
slack_api_scopes.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
slack_integrations has several nullable FKs (project_id, group_id, integration_id, organization_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE slack_integrations ADD CONSTRAINT slack_integrations_one_target CHECK ((project_id IS NOT NULL)::int + (group_id IS NOT NULL)::int + (integration_id IS NOT NULL)::int + (organization_id IS NOT NULL)::int = 1);
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE slack_integrations ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE slack_integrations ALTER COLUMN updated_at TYPE timestamptz;
slack_integrations.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
slack_integrations.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
slack_integrations.integration_id may be NULL, so the relationship to integrations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
slack_integrations.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
slack_integrations_scopes.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
slack_integrations_scopes.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
slack_integrations_scopes.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
snippet_repositories.snippet_organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
snippet_repositories.snippet_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
snippet_repository_states.snippet_organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
snippet_repository_states.snippet_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
snippet_repository_storage_moves.snippet_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
snippet_repository_storage_moves.snippet_organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
snippet_statistics.snippet_organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
snippet_statistics.snippet_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE snippet_uploads ALTER COLUMN created_at TYPE timestamptz;
snippet_user_mentions has several nullable FKs (snippet_project_id, snippet_organization_id, note_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE snippet_user_mentions ADD CONSTRAINT snippet_user_mentions_one_target CHECK ((snippet_project_id IS NOT NULL)::int + (snippet_organization_id IS NOT NULL)::int + (note_id IS NOT NULL)::int = 1);
snippet_user_mentions.snippet_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
snippet_user_mentions.snippet_organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
snippet_user_mentions.note_id may be NULL, so the relationship to notes is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE snippets ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE snippets ALTER COLUMN updated_at TYPE timestamptz;
snippets.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
custom_software_license_id, approval_policy_rule_id, scan_result_policy_idexclusive-arcsoftware_license_policies has several nullable FKs (custom_software_license_id, approval_policy_rule_id, scan_result_policy_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE software_license_policies ADD CONSTRAINT software_license_policies_one_target CHECK ((custom_software_license_id IS NOT NULL)::int + (approval_policy_rule_id IS NOT NULL)::int + (scan_result_policy_id IS NOT NULL)::int = 1);
software_license_policies.custom_software_license_id may be NULL, so the relationship to custom_software_licenses is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
software_license_policies.approval_policy_rule_id may be NULL, so the relationship to approval_policy_rules is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
software_license_policies.scan_result_policy_id may be NULL, so the relationship to scan_result_policies is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
spam_logs has several nullable FKs (user_id, organization_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE spam_logs ADD CONSTRAINT spam_logs_one_target CHECK ((user_id IS NOT NULL)::int + (organization_id IS NOT NULL)::int = 1);
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE spam_logs ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE spam_logs ALTER COLUMN updated_at TYPE timestamptz;
spam_logs.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
spam_logs.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
sprints has several nullable FKs (iterations_cadence_id, group_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE sprints ADD CONSTRAINT sprints_one_target CHECK ((iterations_cadence_id IS NOT NULL)::int + (group_id IS NOT NULL)::int = 1);
sprints.iterations_cadence_id may be NULL, so the relationship to iterations_cadences is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
sprints.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ssh_signatures has several nullable FKs (user_id, key_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE ssh_signatures ADD CONSTRAINT ssh_signatures_one_target CHECK ((user_id IS NOT NULL)::int + (key_id IS NOT NULL)::int = 1);
ssh_signatures.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
ssh_signatures.key_id may be NULL, so the relationship to keys is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
status_check_responses.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
status_page_published_incidents.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
subscription_add_on_purchases.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
subscription_seat_assignments.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
subscription_user_add_on_assignments.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
subscriptions has several nullable FKs (user_id, project_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE subscriptions ADD CONSTRAINT subscriptions_one_target CHECK ((user_id IS NOT NULL)::int + (project_id IS NOT NULL)::int = 1);
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE subscriptions ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE subscriptions ALTER COLUMN updated_at TYPE timestamptz;
subscriptions.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
subscriptions.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
suggestions.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
supply_chain_attestation_states.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
system_access_group_microsoft_application_idfk-nullablesystem_access_group_microsoft_graph_access_tokens.system_access_group_microsoft_application_id may be NULL, so the relationship to system_access_group_microsoft_applications is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
system_access_microsoft_applications.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
system_access_microsoft_graph_access_tokens.system_access_microsoft_application_id may be NULL, so the relationship to system_access_microsoft_applications is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
system_note_metadata has several nullable FKs (namespace_id, description_version_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE system_note_metadata ADD CONSTRAINT system_note_metadata_one_target CHECK ((namespace_id IS NOT NULL)::int + (description_version_id IS NOT NULL)::int = 1);
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE system_note_metadata ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE system_note_metadata ALTER COLUMN updated_at TYPE timestamptz;
system_note_metadata.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
system_note_metadata.description_version_id may be NULL, so the relationship to description_versions is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
tag_gpg_signatures has several nullable FKs (gpg_key_subkey_id, gpg_key_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE tag_gpg_signatures ADD CONSTRAINT tag_gpg_signatures_one_target CHECK ((gpg_key_subkey_id IS NOT NULL)::int + (gpg_key_id IS NOT NULL)::int = 1);
tag_gpg_signatures.gpg_key_subkey_id may be NULL, so the relationship to gpg_key_subkeys is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
tag_gpg_signatures.gpg_key_id may be NULL, so the relationship to gpg_keys is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
tag_ssh_signatures.key_id may be NULL, so the relationship to keys is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
tags references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
terraform_state_version_states references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
terraform_state_versions has several nullable FKs (project_id, created_by_user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE terraform_state_versions ADD CONSTRAINT terraform_state_versions_one_target CHECK ((project_id IS NOT NULL)::int + (created_by_user_id IS NOT NULL)::int = 1);
terraform_state_versions.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
terraform_state_versions.created_by_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
terraform_states.locked_by_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
timelogs has several nullable FKs (project_id, timelog_category_id, issue_id, merge_request_id, note_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE timelogs ADD CONSTRAINT timelogs_one_target CHECK ((project_id IS NOT NULL)::int + (timelog_category_id IS NOT NULL)::int + (issue_id IS NOT NULL)::int + (merge_request_id IS NOT NULL)::int + (note_id IS NOT NULL)::int = 1);
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE timelogs ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE timelogs ALTER COLUMN updated_at TYPE timestamptz;
spent_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE timelogs ALTER COLUMN spent_at TYPE timestamptz;
timelogs.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
timelogs.timelog_category_id may be NULL, so the relationship to timelog_categories is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
timelogs.issue_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
timelogs.merge_request_id may be NULL, so the relationship to merge_requests is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
timelogs.note_id may be NULL, so the relationship to notes is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
todos has several nullable FKs (project_id, organization_id, note_id, group_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE todos ADD CONSTRAINT todos_one_target CHECK ((project_id IS NOT NULL)::int + (organization_id IS NOT NULL)::int + (note_id IS NOT NULL)::int + (group_id IS NOT NULL)::int = 1);
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE todos ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE todos ALTER COLUMN updated_at TYPE timestamptz;
todos.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
todos.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
todos.note_id may be NULL, so the relationship to notes is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
todos.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
upcoming_reconciliations.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
upload_states references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
uploads.uploaded_by_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
uploads.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
uploads.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
uploads.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE uploads ALTER COLUMN created_at TYPE timestamptz;
uploads has several nullable FKs (uploaded_by_user_id, project_id, namespace_id, organization_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE uploads ADD CONSTRAINT uploads_one_target CHECK ((uploaded_by_user_id IS NOT NULL)::int + (project_id IS NOT NULL)::int + (namespace_id IS NOT NULL)::int + (organization_id IS NOT NULL)::int = 1);
user_achievements has several nullable FKs (namespace_id, awarded_by_user_id, revoked_by_user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE user_achievements ADD CONSTRAINT user_achievements_one_target CHECK ((namespace_id IS NOT NULL)::int + (awarded_by_user_id IS NOT NULL)::int + (revoked_by_user_id IS NOT NULL)::int = 1);
user_achievements.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
user_achievements.awarded_by_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
user_achievements.revoked_by_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE user_agent_details ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE user_agent_details ALTER COLUMN updated_at TYPE timestamptz;
user_agent_details.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
user_audit_events references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE user_custom_attributes ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE user_custom_attributes ALTER COLUMN updated_at TYPE timestamptz;
Tables this wide usually hide several entities (or a JSON column that wants to be one). Every row update rewrites the whole tuple; TOAST kicks in; indexes bloat.
Fix Look for column groups that always change together and split them out.
user_details.bot_namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
user_group_member_roles.shared_with_group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE user_permission_export_upload_uploads ALTER COLUMN created_at TYPE timestamptz;
default_duo_add_on_assignment_id, duo_default_namespace_id, knowledge_graph_governing_namespace_idexclusive-arcuser_preferences has several nullable FKs (default_duo_add_on_assignment_id, duo_default_namespace_id, knowledge_graph_governing_namespace_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE user_preferences ADD CONSTRAINT user_preferences_one_target CHECK ((default_duo_add_on_assignment_id IS NOT NULL)::int + (duo_default_namespace_id IS NOT NULL)::int + (knowledge_graph_governing_namespace_id IS NOT NULL)::int = 1);
Tables this wide usually hide several entities (or a JSON column that wants to be one). Every row update rewrites the whole tuple; TOAST kicks in; indexes bloat.
Fix Look for column groups that always change together and split them out.
user_preferences.default_duo_add_on_assignment_id may be NULL, so the relationship to subscription_user_add_on_assignments is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
user_preferences.duo_default_namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
user_preferences.knowledge_graph_governing_namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
user_project_member_roles.shared_with_group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE user_uploads ALTER COLUMN created_at TYPE timestamptz;
reset_password_sent_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE users ALTER COLUMN reset_password_sent_at TYPE timestamptz;
remember_created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE users ALTER COLUMN remember_created_at TYPE timestamptz;
current_sign_in_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE users ALTER COLUMN current_sign_in_at TYPE timestamptz;
last_sign_in_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE users ALTER COLUMN last_sign_in_at TYPE timestamptz;
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE users ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE users ALTER COLUMN updated_at TYPE timestamptz;
locked_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE users ALTER COLUMN locked_at TYPE timestamptz;
password_expires_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE users ALTER COLUMN password_expires_at TYPE timestamptz;
last_credential_check_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE users ALTER COLUMN last_credential_check_at TYPE timestamptz;
confirmed_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE users ALTER COLUMN confirmed_at TYPE timestamptz;
confirmation_sent_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE users ALTER COLUMN confirmation_sent_at TYPE timestamptz;
admin_email_unsubscribed_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE users ALTER COLUMN admin_email_unsubscribed_at TYPE timestamptz;
otp_grace_period_started_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE users ALTER COLUMN otp_grace_period_started_at TYPE timestamptz;
Any change to its key, its delete semantics, or its partitioning touches every dependent. Migrations on hub tables need the longest lock windows and the most careful rollout.
Fix Treat schema changes here as breaking changes with a written rollout plan.
Tables this wide usually hide several entities (or a JSON column that wants to be one). Every row update rewrites the whole tuple; TOAST kicks in; indexes bloat.
Fix Look for column groups that always change together and split them out.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE users_star_projects ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE users_star_projects ALTER COLUMN updated_at TYPE timestamptz;
users_statistics references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
value_stream_dashboard_counts references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
verification_codes references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_entries_00 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_entries_01 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_entries_02 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_entries_03 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_entries_04 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_entries_05 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_entries_06 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_entries_07 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_entries_08 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_entries_09 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_entries_10 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_entries_11 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_entries_12 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_entries_13 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_entries_14 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_entries_15 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_remote_entries_00 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_remote_entries_01 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_remote_entries_02 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_remote_entries_03 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_remote_entries_04 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_remote_entries_05 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_remote_entries_06 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_remote_entries_07 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_remote_entries_08 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_remote_entries_09 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_remote_entries_10 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_remote_entries_11 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_remote_entries_12 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_remote_entries_13 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_remote_entries_14 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_container_cache_remote_entries_15 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_entries references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_entries_00 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_entries_01 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_entries_02 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_entries_03 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_entries_04 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_entries_05 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_entries_06 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_entries_07 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_entries_08 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_entries_09 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_entries_10 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_entries_11 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_entries_12 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_entries_13 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_entries_14 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_entries_15 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_local_entries_00 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_local_entries_01 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_local_entries_02 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_local_entries_03 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_local_entries_04 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_local_entries_05 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_local_entries_06 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_local_entries_07 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_local_entries_08 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_local_entries_09 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_local_entries_10 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_local_entries_11 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_local_entries_12 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_local_entries_13 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_local_entries_14 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_local_entries_15 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_remote_entries_00 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_remote_entries_01 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_remote_entries_02 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_remote_entries_03 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_remote_entries_04 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_remote_entries_05 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_remote_entries_06 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_remote_entries_07 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_remote_entries_08 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_remote_entries_09 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_remote_entries_10 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_remote_entries_11 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_remote_entries_12 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_remote_entries_13 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_remote_entries_14 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_cache_remote_entries_15 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_maven_registry_upstreams.upstream_id may be NULL, so the relationship to virtual_registries_packages_maven_upstreams is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
virtual_registries_packages_maven_registry_upstreams.local_upstream_id may be NULL, so the relationship to virtual_registries_packages_maven_local_upstreams is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
virtual_registries_packages_npm_cache_local_entries_00 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_local_entries_01 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_local_entries_02 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_local_entries_03 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_local_entries_04 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_local_entries_05 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_local_entries_06 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_local_entries_07 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_local_entries_08 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_local_entries_09 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_local_entries_10 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_local_entries_11 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_local_entries_12 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_local_entries_13 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_local_entries_14 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_local_entries_15 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_remote_entries_00 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_remote_entries_01 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_remote_entries_02 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_remote_entries_03 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_remote_entries_04 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_remote_entries_05 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_remote_entries_06 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_remote_entries_07 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_remote_entries_08 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_remote_entries_09 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_remote_entries_10 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_remote_entries_11 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_remote_entries_12 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_remote_entries_13 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_remote_entries_14 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
virtual_registries_packages_npm_cache_remote_entries_15 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
vulnerabilities.finding_id may be NULL, so the relationship to vulnerability_occurrences is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE vulnerability_archive_export_uploads ALTER COLUMN created_at TYPE timestamptz;
vulnerability_archive_exports references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
vulnerability_archived_records references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
vulnerability_archives references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE vulnerability_export_part_uploads ALTER COLUMN created_at TYPE timestamptz;
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE vulnerability_export_uploads ALTER COLUMN created_at TYPE timestamptz;
vulnerability_external_issue_links.vulnerability_occurrence_id may be NULL, so the relationship to vulnerability_occurrences is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
vulnerability_feedback references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
vulnerability_occurrence_id, vulnerability_remediation_idexclusive-arcvulnerability_findings_remediations has several nullable FKs (vulnerability_occurrence_id, vulnerability_remediation_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE vulnerability_findings_remediations ADD CONSTRAINT vulnerability_findings_remediations_one_target CHECK ((vulnerability_occurrence_id IS NOT NULL)::int + (vulnerability_remediation_id IS NOT NULL)::int = 1);
vulnerability_findings_remediations.vulnerability_occurrence_id may be NULL, so the relationship to vulnerability_occurrences is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
vulnerability_findings_remediations.vulnerability_remediation_id may be NULL, so the relationship to vulnerability_remediations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
vulnerability_historical_statistics references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
vulnerability_namespace_historical_statistics references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
vulnerability_namespace_statistics references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
vulnerability_occurrence_identifiers.partition_id.identifier_id may be NULL, so the relationship to vulnerability_identifiers is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
vulnerability_occurrences.vulnerability_id may be NULL, so the relationship to vulnerabilities is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
vulnerability_occurrences.partition_id.primary_identifier_id may be NULL, so the relationship to vulnerability_identifiers is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
vulnerability_reads.vulnerability_occurrence_id may be NULL, so the relationship to vulnerability_occurrences is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE vulnerability_remediation_uploads ALTER COLUMN created_at TYPE timestamptz;
vulnerability_statistics references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE web_hook_logs_daily ALTER COLUMN updated_at TYPE timestamptz;
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE web_hook_logs_daily ALTER COLUMN created_at TYPE timestamptz;
web_hook_logs_daily references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
web_hooks has several nullable FKs (project_id, integration_id, organization_id, group_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE web_hooks ADD CONSTRAINT web_hooks_one_target CHECK ((project_id IS NOT NULL)::int + (integration_id IS NOT NULL)::int + (organization_id IS NOT NULL)::int + (group_id IS NOT NULL)::int = 1);
created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE web_hooks ALTER COLUMN created_at TYPE timestamptz;
updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.
Fix Use TIMESTAMPTZ.
ALTER TABLE web_hooks ALTER COLUMN updated_at TYPE timestamptz;
Tables this wide usually hide several entities (or a JSON column that wants to be one). Every row update rewrites the whole tuple; TOAST kicks in; indexes bloat.
Fix Look for column groups that always change together and split them out.
web_hooks.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
web_hooks.integration_id may be NULL, so the relationship to integrations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
web_hooks.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
web_hooks.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
wiki_page_meta.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
wiki_page_meta.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
wiki_page_slugs.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
wiki_page_slugs.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
wiki_repository_states references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
work_item_current_statuses.custom_status_id may be NULL, so the relationship to work_item_custom_statuses is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
work_item_custom_lifecycles has several nullable FKs (updated_by_id, created_by_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE work_item_custom_lifecycles ADD CONSTRAINT work_item_custom_lifecycles_one_target CHECK ((updated_by_id IS NOT NULL)::int + (created_by_id IS NOT NULL)::int = 1);
work_item_custom_lifecycles.updated_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
work_item_custom_lifecycles.created_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
work_item_custom_statuses has several nullable FKs (updated_by_id, created_by_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE work_item_custom_statuses ADD CONSTRAINT work_item_custom_statuses_one_target CHECK ((updated_by_id IS NOT NULL)::int + (created_by_id IS NOT NULL)::int = 1);
work_item_custom_statuses.updated_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
work_item_custom_statuses.created_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
work_item_custom_types.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
work_item_custom_types.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
start_date_sourcing_milestone_id, start_date_sourcing_work_item_id, due_date_sourcing_work_item_id, due_date_sourcing_milestone_idexclusive-arcwork_item_dates_sources has several nullable FKs (start_date_sourcing_milestone_id, start_date_sourcing_work_item_id, due_date_sourcing_work_item_id, due_date_sourcing_milestone_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE work_item_dates_sources ADD CONSTRAINT work_item_dates_sources_one_target CHECK ((start_date_sourcing_milestone_id IS NOT NULL)::int + (start_date_sourcing_work_item_id IS NOT NULL)::int + (due_date_sourcing_work_item_id IS NOT NULL)::int + (due_date_sourcing_milestone_id IS NOT NULL)::int = 1);
work_item_dates_sources.start_date_sourcing_milestone_id may be NULL, so the relationship to milestones is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
work_item_dates_sources.start_date_sourcing_work_item_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
work_item_dates_sources.due_date_sourcing_work_item_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
work_item_dates_sources.due_date_sourcing_milestone_id may be NULL, so the relationship to milestones is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
work_item_parent_links.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
work_item_progresses.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
work_item_settings.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
work_item_settings.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
work_item_transitions has several nullable FKs (promoted_to_epic_id, duplicated_to_id, moved_to_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE work_item_transitions ADD CONSTRAINT work_item_transitions_one_target CHECK ((promoted_to_epic_id IS NOT NULL)::int + (duplicated_to_id IS NOT NULL)::int + (moved_to_id IS NOT NULL)::int = 1);
work_item_transitions.promoted_to_epic_id may be NULL, so the relationship to epics is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
work_item_transitions.duplicated_to_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
work_item_transitions.moved_to_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
work_item_type_visibility_defaults.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
work_item_type_visibility_defaults.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
workspace_variables.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
workspaces.personal_access_token_id may be NULL, so the relationship to personal_access_tokens is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
x509_certificates.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
x509_issuers.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
zentao_tracker_data.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
zentao_tracker_data.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
zentao_tracker_data.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
zoekt_indices has several nullable FKs (zoekt_enabled_namespace_id, zoekt_replica_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.
Fix Add a CHECK that exactly one is non-null, or restructure with a supertype.
ALTER TABLE zoekt_indices ADD CONSTRAINT zoekt_indices_one_target CHECK ((zoekt_enabled_namespace_id IS NOT NULL)::int + (zoekt_replica_id IS NOT NULL)::int = 1);
zoekt_indices.zoekt_enabled_namespace_id may be NULL, so the relationship to zoekt_enabled_namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
zoekt_indices.zoekt_replica_id may be NULL, so the relationship to zoekt_replicas is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
zoekt_repositories.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.
Fix Confirm the optionality is a domain rule; document it in the column comment.
zoekt_tasks references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.
Fix Document its purpose or drop it.
1429 tables · 1861 foreign keys · 111 domains. Open the 3D explorer: rotate, zoom, click any table or relationship.
Tables whose names start with namespaces_.
namespaces.organization_id → organizations.id — one organizations, many namespaces · optional · ON DELETE CASCADE · indexednamespaces.parent_id → namespaces.id — one namespaces, many namespaces · optional · ON DELETE RESTRICT · indexednamespaces_consents.namespace_id → namespaces.id — one namespaces, many namespaces_consents · required · ON DELETE CASCADE · indexednamespaces_storage_limit_exclusions.namespace_id → namespaces.id — one namespaces, at most one namespaces_storage_limit_exclusions · required · ON DELETE CASCADE · indexednamespaces_sync_events.namespace_id → namespaces.id — one namespaces, many namespaces_sync_events · required · ON DELETE CASCADE · indexedorganization_id — namespaces.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.parent_id — namespaces.parent_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.type — type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.ldap_sync_status — ldap_sync_status is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.organization_id, parent_id — namespaces has several nullable FKs (organization_id, parent_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.ldap_sync_last_update_at — ldap_sync_last_update_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.ldap_sync_last_successful_update_at — ldap_sync_last_successful_update_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.ldap_sync_last_sync_at — ldap_sync_last_sync_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
name | varchar | NOT NULL | |||
path | varchar | NOT NULL | |||
owner_id | bigint | null | |||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
type | varchar | NOT NULL | CAST('User' AS varchar) | ||
avatar | varchar | null | |||
membership_lock | boolean | null | FALSE | ||
share_with_group_lock | boolean | null | FALSE | ||
visibility_level | integer | NOT NULL | 20 | ||
request_access_enabled | boolean | NOT NULL | TRUE | ||
ldap_sync_status | varchar | NOT NULL | CAST('ready' AS varchar) | ||
ldap_sync_error | varchar | null | |||
ldap_sync_last_update_at | timestamp | null | |||
ldap_sync_last_successful_update_at | timestamp | null | |||
ldap_sync_last_sync_at | timestamp | null | |||
lfs_enabled | boolean | null | |||
parent_id | bigint | null | namespaces.id 1:N · RESTRICT | ||
shared_runners_minutes_limit | integer | null | |||
repository_size_limit | bigint | null | |||
require_two_factor_authentication | boolean | NOT NULL | FALSE | ||
two_factor_grace_period | integer | NOT NULL | 48 | ||
project_creation_level | integer | null | |||
runners_token | varchar | null | |||
file_template_project_id | bigint | null | |||
saml_discovery_token | varchar | null | |||
runners_token_encrypted | varchar | null | |||
custom_project_templates_group_id | bigint | null | |||
auto_devops_enabled | boolean | null | |||
extra_shared_runners_minutes_limit | integer | null | |||
last_ci_minutes_notification_at | timestamptz | null | |||
last_ci_minutes_usage_notification_level | integer | null | |||
subgroup_creation_level | integer | null | 1 | ||
max_pages_size | integer | null | |||
max_artifacts_size | integer | null | |||
mentions_disabled | boolean | null | |||
default_branch_protection | smallint | null | |||
max_personal_access_token_lifetime | integer | null | |||
shared_runners_enabled | boolean | NOT NULL | TRUE | ||
allow_descendants_override_disabled_shared_runners | boolean | NOT NULL | FALSE | ||
traversal_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE | CHECK organization_id IS NOT NULL | |
state | smallint | null | 0 | CHECK state IS NOT NULL |
Table checks: organization_id IS NOT NULL; state IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
user_id | bigint | null | |||
feature_name | smallint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
reason | text | NOT NULL | CHECK char_length(reason) <= 255 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
Table checks: char_length(reason) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE |
Tables whose names start with projects_.
projects.project_namespace_id → namespaces.id — one namespaces, at most one projects · optional · ON DELETE CASCADE · indexedprojects.organization_id → organizations.id — one organizations, many projects · optional · ON DELETE CASCADE · indexedprojects.namespace_id → namespaces.id — one namespaces, many projects · required · ON DELETE RESTRICT · indexedprojects_branch_rules_merge_request_approval_settings.protected_branch_id → protected_branches.id — one protected_branches, at most one projects_branch_rules_merge_request_approval_settings · required · ON DELETE CASCADE · indexedprojects_branch_rules_merge_request_approval_settings.project_id → projects.id — one projects, many projects_branch_rules_merge_request_approval_settings · required · ON DELETE CASCADE · indexedprojects_branch_rules_squash_options.protected_branch_id → protected_branches.id — one protected_branches, at most one projects_branch_rules_squash_options · required · ON DELETE CASCADE · indexedprojects_branch_rules_squash_options.project_id → projects.id — one projects, many projects_branch_rules_squash_options · required · ON DELETE CASCADE · indexedprojects_sync_events.project_id → projects.id — one projects, many projects_sync_events · required · ON DELETE CASCADE · indexedprojects_with_pipeline_variables.project_id → projects.id — one projects, at most one projects_with_pipeline_variables · required · ON DELETE CASCADE · indexedproject_namespace_id — projects.project_namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id — projects.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.import_type — import_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.project_namespace_id, organization_id — projects has several nullable FKs (project_namespace_id, organization_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.last_activity_at — last_activity_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.mirror_last_update_at — mirror_last_update_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.mirror_last_successful_update_at — mirror_last_successful_update_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.last_repository_check_at — last_repository_check_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.last_repository_updated_at — last_repository_updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
name | varchar | null | |||
path | varchar | null | |||
description | text | null | |||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
creator_id | bigint | null | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · RESTRICT | ||
last_activity_at | timestamp | null | |||
import_url | varchar | null | |||
visibility_level | integer | NOT NULL | 0 | ||
archived | boolean | NOT NULL | FALSE | ||
avatar | varchar | null | |||
merge_requests_template | text | null | |||
star_count | integer | NOT NULL | 0 | CHECK star_count >= 0 | |
merge_requests_rebase_enabled | boolean | null | FALSE | ||
import_type | varchar | null | |||
import_source | varchar | null | |||
approvals_before_merge | integer | NOT NULL | 0 | ||
reset_approvals_on_push | boolean | null | TRUE | ||
merge_requests_ff_only_enabled | boolean | null | FALSE | ||
issues_template | text | null | |||
mirror | boolean | NOT NULL | FALSE | ||
mirror_last_update_at | timestamp | null | |||
mirror_last_successful_update_at | timestamp | null | |||
mirror_user_id | bigint | null | |||
shared_runners_enabled | boolean | NOT NULL | TRUE | ||
runners_token | varchar | null | |||
build_allow_git_fetch | boolean | NOT NULL | TRUE | ||
build_timeout | integer | NOT NULL | 3600 | ||
mirror_trigger_builds | boolean | NOT NULL | FALSE | ||
pending_delete | boolean | null | FALSE | ||
public_builds | boolean | NOT NULL | TRUE | ||
last_repository_check_failed | boolean | null | |||
last_repository_check_at | timestamp | null | |||
only_allow_merge_if_pipeline_succeeds | boolean | NOT NULL | FALSE | ||
has_external_issue_tracker | boolean | null | |||
repository_storage | varchar | NOT NULL | CAST('default' AS varchar) | ||
repository_read_only | boolean | null | |||
request_access_enabled | boolean | NOT NULL | TRUE | ||
has_external_wiki | boolean | null | |||
ci_config_path | varchar | null | |||
lfs_enabled | boolean | null | |||
description_html | text | null | |||
only_allow_merge_if_all_discussions_are_resolved | boolean | null | |||
repository_size_limit | bigint | null | |||
printing_merge_request_link_enabled | boolean | NOT NULL | TRUE | ||
auto_cancel_pending_pipelines | integer | NOT NULL | 1 | ||
service_desk_enabled | boolean | null | TRUE | ||
cached_markdown_version | integer | null | |||
last_repository_updated_at | timestamp | null | |||
disable_overriding_approvers_per_merge_request | boolean | null | |||
storage_version | smallint | null | |||
resolve_outdated_diff_discussions | boolean | null | |||
remote_mirror_available_overridden | boolean | null | |||
only_mirror_protected_branches | boolean | null | |||
pull_mirror_available_overridden | boolean | null | |||
jobs_cache_index | integer | null | |||
external_authorization_classification_label | varchar | null | |||
mirror_overwrites_diverged_branches | boolean | null | |||
pages_https_only | boolean | null | TRUE | ||
external_webhook_token | varchar | null | |||
packages_enabled | boolean | null | |||
merge_requests_author_approval | boolean | null | FALSE | ||
pool_repository_id | bigint | null | |||
runners_token_encrypted | varchar | null | |||
bfg_object_map | varchar | null | |||
detected_repository_languages | boolean | null | |||
merge_requests_disable_committers_approval | boolean | null | |||
require_password_to_approve | boolean | null | |||
emails_disabled | boolean | null | |||
max_pages_size | integer | null | |||
max_artifacts_size | integer | null | |||
pull_mirror_branch_prefix | varchar(50) | null | |||
remove_source_branch_after_merge | boolean | null | |||
marked_for_deletion_at | date | null | |||
marked_for_deletion_by_user_id | bigint | null | |||
autoclose_referenced_issues | boolean | null | |||
suggestion_commit_message | varchar(255) | null | |||
project_namespace_id | bigint | null | namespaces.id 1:1 · CASCADE | CHECK project_namespace_id IS NOT NULL | |
hidden | boolean | NOT NULL | FALSE | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE | CHECK organization_id IS NOT NULL |
Table checks: organization_id IS NOT NULL; project_namespace_id IS NOT NULL; star_count >= 0
projects_visits references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
entity_id | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | |||
visited_at PK | timestamptz | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
protected_branch_id | bigint | NOT NULL | protected_branches.id 1:1 · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
prevent_author_approval | boolean | NOT NULL | FALSE | ||
prevent_committer_approval | boolean | NOT NULL | FALSE | ||
prevent_editing_approval_rules | boolean | NOT NULL | FALSE | ||
require_reauthentication_to_approve | boolean | NOT NULL | FALSE | ||
approval_removals | smallint | NOT NULL | 1 |
project_id, protected_branch_id — projects_branch_rules_squash_options looks like a many-to-many link table but has no unique constraint across (project_id, protected_branch_id). The same pair can be inserted twice; every join through it will double-count.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
protected_branch_id | bigint | NOT NULL | protected_branches.id 1:1 · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
squash_option | smallint | NOT NULL | 3 |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | projects.id 1:1 · CASCADE |
Tables whose names start with users_.
users.organization_id → organizations.id — one organizations, many users · required · ON DELETE NO ACTION · indexedusers_ops_dashboard_projects.user_id → users.id — one users, many users_ops_dashboard_projects · required · ON DELETE CASCADE · indexedusers_ops_dashboard_projects.project_id → projects.id — one projects, many users_ops_dashboard_projects · required · ON DELETE CASCADE · indexedusers_security_dashboard_projects.project_id → projects.id — one projects, many users_security_dashboard_projects · required · ON DELETE CASCADE · indexedusers_security_dashboard_projects.user_id → users.id — one users, many users_security_dashboard_projects · required · ON DELETE CASCADE · indexedusers_star_projects.project_id → projects.id — one projects, many users_star_projects · required · ON DELETE CASCADE · indexedstate — state is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.reset_password_sent_at — reset_password_sent_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.remember_created_at — remember_created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.current_sign_in_at — current_sign_in_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.last_sign_in_at — last_sign_in_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.locked_at — locked_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.password_expires_at — password_expires_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.last_credential_check_at — last_credential_check_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.confirmed_at — confirmed_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.confirmation_sent_at — confirmation_sent_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.admin_email_unsubscribed_at — admin_email_unsubscribed_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.otp_grace_period_started_at — otp_grace_period_started_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
email | varchar | NOT NULL | CAST('' AS varchar) | ||
encrypted_password | varchar | NOT NULL | CAST('' AS varchar) | ||
reset_password_token | varchar | null | |||
reset_password_sent_at | timestamp | null | |||
remember_created_at | timestamp | null | |||
sign_in_count | integer | null | 0 | ||
current_sign_in_at | timestamp | null | |||
last_sign_in_at | timestamp | null | |||
current_sign_in_ip | varchar | null | |||
last_sign_in_ip | varchar | null | |||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
name | varchar | null | |||
admin | boolean | NOT NULL | FALSE | ||
projects_limit | integer | NOT NULL | |||
failed_attempts | integer | null | 0 | ||
locked_at | timestamp | null | |||
username | varchar | null | |||
can_create_group | boolean | NOT NULL | TRUE | ||
can_create_team | boolean | NOT NULL | TRUE | ||
state | varchar | null | |||
color_scheme_id | bigint | NOT NULL | 1 | ||
password_expires_at | timestamp | null | |||
created_by_id | bigint | null | |||
last_credential_check_at | timestamp | null | |||
avatar | varchar | null | |||
confirmation_token | varchar | null | |||
confirmed_at | timestamp | null | |||
confirmation_sent_at | timestamp | null | |||
unconfirmed_email | varchar | null | |||
hide_no_ssh_key | boolean | null | FALSE | CHECK hide_no_ssh_key IS NOT NULL | |
admin_email_unsubscribed_at | timestamp | null | |||
notification_email | varchar | null | |||
hide_no_password | boolean | null | FALSE | CHECK hide_no_password IS NOT NULL | |
password_automatically_set | boolean | null | FALSE | ||
encrypted_otp_secret | varchar | null | |||
encrypted_otp_secret_iv | varchar | null | |||
encrypted_otp_secret_salt | varchar | null | |||
otp_required_for_login | boolean | NOT NULL | FALSE | ||
otp_backup_codes | text | null | |||
public_email | varchar | null | |||
dashboard | integer | null | 0 | ||
project_view | integer | null | 2 | CHECK project_view IS NOT NULL | |
consumed_timestep | integer | null | |||
layout | integer | null | 0 | ||
hide_project_limit | boolean | null | FALSE | ||
note | text | null | |||
unlock_token | varchar | null | |||
otp_grace_period_started_at | timestamp | null | |||
external | boolean | null | FALSE | ||
incoming_email_token | varchar | null | |||
auditor | boolean | NOT NULL | FALSE | ||
require_two_factor_authentication_from_group | boolean | NOT NULL | FALSE | ||
two_factor_grace_period | integer | NOT NULL | 48 | ||
last_activity_on | date | null | |||
notified_of_own_activity | boolean | null | FALSE | CHECK notified_of_own_activity IS NOT NULL | |
preferred_language | varchar | null | |||
theme_id | smallint | null | |||
accepted_term_id | bigint | null | |||
feed_token | varchar | null | |||
private_profile | boolean | NOT NULL | FALSE | ||
roadmap_layout | smallint | null | |||
include_private_contributions | boolean | null | |||
commit_email | varchar | null | |||
group_view | integer | null | |||
managing_group_id | bigint | null | |||
first_name | varchar(255) | null | |||
last_name | varchar(255) | null | |||
static_object_token | varchar(255) | null | |||
user_type | smallint | null | 0 | CHECK user_type IS NOT NULL | |
static_object_token_encrypted | text | null | CHECK char_length(static_object_token_encrypted) <= 255 | ||
otp_secret_expires_at | timestamptz | null | |||
onboarding_in_progress | boolean | NOT NULL | FALSE | ||
color_mode_id | smallint | NOT NULL | 1 | ||
composite_identity_enforced | boolean | NOT NULL | FALSE | ||
organization_id | bigint | NOT NULL | organizations.id 1:N | ||
otp_secret | text | null | CHECK char_length(otp_secret) <= 255 |
Table checks: project_view IS NOT NULL; user_type IS NOT NULL; hide_no_password IS NOT NULL; hide_no_ssh_key IS NOT NULL; char_length(static_object_token_encrypted) <= 255; notified_of_own_activity IS NOT NULL; char_length(otp_secret) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
user_id PK | bigint | NOT NULL | users.id 1:N · CASCADE | ||
project_id PK | bigint | NOT NULL | projects.id 1:N · CASCADE |
created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
user_id | bigint | NOT NULL | |||
created_at | timestamp | null | |||
updated_at | timestamp | null |
users_statistics references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
without_groups_and_projects | integer | NOT NULL | 0 | ||
with_highest_role_guest | integer | NOT NULL | 0 | ||
with_highest_role_reporter | integer | NOT NULL | 0 | ||
with_highest_role_developer | integer | NOT NULL | 0 | ||
with_highest_role_maintainer | integer | NOT NULL | 0 | ||
with_highest_role_owner | integer | NOT NULL | 0 | ||
bots | integer | NOT NULL | 0 | ||
blocked | integer | NOT NULL | 0 | ||
with_highest_role_minimal_access | integer | NOT NULL | 0 | ||
with_highest_role_guest_with_custom_role | integer | NOT NULL | 0 | ||
with_highest_role_planner | integer | NOT NULL | 0 | ||
with_highest_role_security_manager | integer | NOT NULL | 0 |
Tables whose names start with ai_.
ai_events_counts.organization_id → organizations.id — one organizations, many ai_events_counts · required · ON DELETE CASCADE · indexedai_usage_events.namespace_id → namespaces.id — one namespaces, many ai_usage_events · optional · ON DELETE SET NULL · indexedai_usage_events.organization_id → organizations.id — one organizations, many ai_usage_events · required · ON DELETE CASCADE · indexedai_active_context_collections.connection_id → ai_active_context_connections.id — one ai_active_context_connections, many ai_active_context_collections · required · ON DELETE CASCADE · indexedai_active_context_migrations.connection_id → ai_active_context_connections.id — one ai_active_context_connections, many ai_active_context_migrations · required · ON DELETE CASCADE · indexedai_active_context_tasks.depends_on_id → ai_active_context_tasks.id — one ai_active_context_tasks, many ai_active_context_tasks · optional · ON DELETE SET NULL · indexedai_active_context_tasks.connection_id → ai_active_context_connections.id — one ai_active_context_connections, many ai_active_context_tasks · required · ON DELETE CASCADE · indexedai_agent_identities.user_id → users.id — one users, many ai_agent_identities · required · ON DELETE CASCADE · indexedai_agent_identities.project_id → projects.id — one projects, many ai_agent_identities · required · ON DELETE CASCADE · indexedai_catalog_item_consumers.group_id → namespaces.id — one namespaces, many ai_catalog_item_consumers · optional · ON DELETE CASCADE · indexedai_catalog_item_consumers.parent_item_consumer_id → ai_catalog_item_consumers.id — one ai_catalog_item_consumers, many ai_catalog_item_consumers · optional · ON DELETE CASCADE · indexedai_catalog_item_consumers.project_id → projects.id — one projects, many ai_catalog_item_consumers · optional · ON DELETE CASCADE · indexedai_catalog_item_consumers.service_account_id → users.id — one users, at most one ai_catalog_item_consumers · optional · ON DELETE SET NULL · indexedai_catalog_item_consumers.organization_id → organizations.id — one organizations, many ai_catalog_item_consumers · optional · ON DELETE CASCADE · indexedai_catalog_item_consumers.ai_catalog_item_id → ai_catalog_items.id — one ai_catalog_items, many ai_catalog_item_consumers · required · ON DELETE RESTRICT · indexedai_catalog_item_stars.ai_catalog_item_id → ai_catalog_items.id — one ai_catalog_items, many ai_catalog_item_stars · required · ON DELETE CASCADE · indexedai_catalog_item_stars.user_id → users.id — one users, many ai_catalog_item_stars · required · ON DELETE CASCADE · indexedai_catalog_item_stars.organization_id → organizations.id — one organizations, many ai_catalog_item_stars · required · ON DELETE CASCADE · indexedai_catalog_item_version_dependencies.dependency_id → ai_catalog_items.id — one ai_catalog_items, many ai_catalog_item_version_dependencies · required · ON DELETE CASCADE · indexedai_catalog_item_version_dependencies.organization_id → organizations.id — one organizations, many ai_catalog_item_version_dependencies · required · ON DELETE CASCADE · indexedai_catalog_item_version_dependencies.ai_catalog_item_version_id → ai_catalog_item_versions.id — one ai_catalog_item_versions, many ai_catalog_item_version_dependencies · required · ON DELETE CASCADE · indexedai_catalog_item_versions.created_by_id → users.id — one users, many ai_catalog_item_versions · optional · ON DELETE SET NULL · indexedai_catalog_item_versions.ai_catalog_item_id → ai_catalog_items.id — one ai_catalog_items, many ai_catalog_item_versions · required · ON DELETE CASCADE · indexedai_catalog_item_versions.organization_id → organizations.id — one organizations, many ai_catalog_item_versions · required · ON DELETE CASCADE · indexedai_catalog_items.project_id → projects.id — one projects, many ai_catalog_items · optional · ON DELETE CASCADE · indexedai_catalog_items.organization_id → organizations.id — one organizations, many ai_catalog_items · required · ON DELETE CASCADE · indexedai_catalog_items.latest_released_version_id → ai_catalog_item_versions.id — one ai_catalog_item_versions, many ai_catalog_items · optional · ON DELETE NO ACTION · indexedai_catalog_items.latest_version_id → ai_catalog_item_versions.id — one ai_catalog_item_versions, many ai_catalog_items · optional · ON DELETE NO ACTION · indexedai_catalog_mcp_server_blocks.ai_catalog_mcp_server_id → ai_catalog_mcp_servers.id — one ai_catalog_mcp_servers, many ai_catalog_mcp_server_blocks · required · ON DELETE CASCADE · indexedai_catalog_mcp_server_blocks.namespace_id → namespaces.id — one namespaces, many ai_catalog_mcp_server_blocks · required · ON DELETE CASCADE · indexedai_catalog_mcp_server_blocks.organization_id → organizations.id — one organizations, many ai_catalog_mcp_server_blocks · required · ON DELETE CASCADE · indexedai_catalog_mcp_server_blocks.created_by_id → users.id — one users, many ai_catalog_mcp_server_blocks · optional · ON DELETE SET NULL · indexedai_catalog_mcp_servers.created_by_id → users.id — one users, many ai_catalog_mcp_servers · optional · ON DELETE SET NULL · indexedai_catalog_mcp_servers.organization_id → organizations.id — one organizations, many ai_catalog_mcp_servers · required · ON DELETE CASCADE · indexedai_catalog_mcp_servers_users.user_id → users.id — one users, many ai_catalog_mcp_servers_users · required · ON DELETE CASCADE · indexedai_catalog_mcp_servers_users.ai_catalog_mcp_server_id → ai_catalog_mcp_servers.id — one ai_catalog_mcp_servers, many ai_catalog_mcp_servers_users · required · ON DELETE CASCADE · indexedai_catalog_mcp_servers_users.organization_id → organizations.id — one organizations, many ai_catalog_mcp_servers_users · required · ON DELETE CASCADE · indexedai_conversation_messages.organization_id → organizations.id — one organizations, many ai_conversation_messages · required · ON DELETE CASCADE · indexedai_conversation_threads.organization_id → organizations.id — one organizations, many ai_conversation_threads · required · ON DELETE CASCADE · indexedai_conversation_threads.user_id → users.id — one users, many ai_conversation_threads · required · ON DELETE CASCADE · indexedai_feature_settings.ai_self_hosted_model_id → ai_self_hosted_models.id — one ai_self_hosted_models, many ai_feature_settings · optional · ON DELETE CASCADE · indexedai_flow_triggers.ai_catalog_item_consumer_id → ai_catalog_item_consumers.id — one ai_catalog_item_consumers, many ai_flow_triggers · optional · ON DELETE CASCADE · indexedai_namespace_feature_access_rules.root_namespace_id → namespaces.id — one namespaces, many ai_namespace_feature_access_rules · required · ON DELETE CASCADE · indexedai_namespace_feature_access_rules.through_namespace_id → namespaces.id — one namespaces, many ai_namespace_feature_access_rules · optional · ON DELETE CASCADE · indexedai_namespace_feature_settings.namespace_id → namespaces.id — one namespaces, many ai_namespace_feature_settings · required · ON DELETE CASCADE · indexedai_settings.amazon_q_oauth_application_id → oauth_applications.id — one oauth_applications, many ai_settings · optional · ON DELETE SET NULL · indexedai_settings.organization_id → organizations.id — one organizations, at most one ai_settings · optional · ON DELETE CASCADE · indexedai_settings.duo_workflow_oauth_application_id → oauth_applications.id — one oauth_applications, many ai_settings · optional · ON DELETE SET NULL · indexedai_settings.duo_workflow_service_account_user_id → users.id — one users, many ai_settings · optional · ON DELETE SET NULL · indexedai_settings.amazon_q_service_account_user_id → users.id — one users, many ai_settings · optional · ON DELETE SET NULL · indexedai_tool_rules.project_id → projects.id — one projects, many ai_tool_rules · optional · ON DELETE CASCADE · indexedai_tool_rules.namespace_id → namespaces.id — one namespaces, many ai_tool_rules · required · ON DELETE CASCADE · indexedai_user_metrics.user_id → users.id — one users, at most one ai_user_metrics · required · ON DELETE CASCADE · indexedai_vectorizable_file_upload_states.project_id → projects.id — one projects, many ai_vectorizable_file_upload_states · required · ON DELETE CASCADE · indexedai_vectorizable_file_upload_states.ai_vectorizable_file_upload_id → ai_vectorizable_file_uploads.id — one ai_vectorizable_file_uploads, at most one ai_vectorizable_file_upload_states · required · ON DELETE CASCADE · indexedai_vectorizable_files.project_id → projects.id — one projects, many ai_vectorizable_files · required · ON DELETE CASCADE · indexedai_audit_events references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at PK | timestamptz | NOT NULL | |||
cloud_event_id | uuid | NOT NULL | |||
author_id | bigint | NOT NULL | |||
project_id | bigint | null | |||
namespace_id | bigint | null | |||
event_name | text | NOT NULL | CHECK char_length(event_name) <= 255 | ||
ip_address | inet | null | |||
workflow_id | bigint | null | |||
details | text | null |
Table checks: char_length(event_name) <= 255; num_nonnulls(namespace_id, project_id) = 1
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
events_date PK | date | NOT NULL | |||
namespace_id | bigint | null | |||
user_id | bigint | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
event | smallint | NOT NULL | |||
total_occurrences | integer | NOT NULL | 0 |
namespace_id — ai_usage_events.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
timestamp PK | timestamptz | NOT NULL | |||
user_id | bigint | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
event | smallint | NOT NULL | |||
extras | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
namespace_id | bigint | null | namespaces.id 1:N · SET NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
metadata | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
number_of_partitions | integer | NOT NULL | 1 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
connection_id | bigint | NOT NULL | ai_active_context_connections.id 1:N · CASCADE | ||
options | jsonb | NOT NULL | CAST('{}' AS jsonb) |
Table checks: char_length(name) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
active | boolean | NOT NULL | FALSE | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
prefix | text | null | CHECK char_length(prefix) <= 255 | ||
adapter_class | text | NOT NULL | CHECK char_length(adapter_class) <= 255 | ||
options | jsonb | NOT NULL |
Table checks: char_length(name) <= 255; char_length(adapter_class) <= 255; char_length(prefix) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
connection_id | bigint | NOT NULL | ai_active_context_connections.id 1:N · CASCADE | ||
started_at | timestamptz | null | |||
completed_at | timestamptz | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
retries_left | smallint | NOT NULL | |||
version | text | NOT NULL | CHECK version ~ CAST('^[0-9]{14}$' AS text) | ||
metadata | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
error_message | text | null | CHECK char_length(error_message) <= 1024 |
Table checks: retries_left > 0 OR (retries_left = 0 AND status = 255); version ~ CAST('^[0-9]{14}$' AS text); char_length(error_message) <= 1024; char_length(version) <= 255
depends_on_id — ai_active_context_tasks.depends_on_id may be NULL, so the relationship to ai_active_context_tasks is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
connection_id | bigint | NOT NULL | ai_active_context_connections.id 1:N · CASCADE | ||
depends_on_id | bigint | null | ai_active_context_tasks.id 1:N · SET NULL | ||
started_at | timestamptz | null | |||
completed_at | timestamptz | null | |||
status | smallint | NOT NULL | 0 | ||
retries_left | smallint | NOT NULL | 3 | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
error_message | text | null | CHECK char_length(error_message) <= 1024 | ||
params | jsonb | NOT NULL | CAST('{}' AS jsonb) |
Table checks: retries_left > 0 OR (retries_left = 0 AND status = 255); char_length(error_message) <= 1024; char_length(name) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
revoked_at | timestamptz | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
agent_type | text | NOT NULL | CHECK char_length(agent_type) <= 50 | ||
machine_fingerprint | text | NOT NULL | CHECK char_length(machine_fingerprint) <= 64 |
Table checks: char_length(machine_fingerprint) <= 64; char_length(agent_type) <= 50
group_id — ai_catalog_item_consumers.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.parent_item_consumer_id — ai_catalog_item_consumers.parent_item_consumer_id may be NULL, so the relationship to ai_catalog_item_consumers is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — ai_catalog_item_consumers.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.service_account_id — ai_catalog_item_consumers.service_account_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id — ai_catalog_item_consumers.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id, parent_item_consumer_id, project_id, service_account_id, organization_id — ai_catalog_item_consumers has several nullable FKs (group_id, parent_item_consumer_id, project_id, service_account_id, organization_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
ai_catalog_item_id | bigint | NOT NULL | ai_catalog_items.id 1:N · RESTRICT | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
enabled | boolean | NOT NULL | FALSE | ||
locked | boolean | NOT NULL | TRUE | ||
pinned_version_prefix | text | null | CHECK char_length(pinned_version_prefix) <= 50 | ||
service_account_id | bigint | null | users.id 1:1 · SET NULL | ||
parent_item_consumer_id | bigint | null | ai_catalog_item_consumers.id 1:N · CASCADE |
Table checks: num_nonnulls(group_id, project_id) = 1; char_length(pinned_version_prefix) <= 50
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
ai_catalog_item_id | bigint | NOT NULL | ai_catalog_items.id 1:N · CASCADE | ||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
ai_catalog_item_version_id | bigint | NOT NULL | ai_catalog_item_versions.id 1:N · CASCADE | ||
dependency_id | bigint | NOT NULL | ai_catalog_items.id 1:N · CASCADE | ||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE |
created_by_id — ai_catalog_item_versions.created_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
release_date | timestamptz | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
ai_catalog_item_id | bigint | NOT NULL | ai_catalog_items.id 1:N · CASCADE | ||
schema_version | smallint | NOT NULL | |||
version | text | NOT NULL | CHECK char_length(version) <= 50 | ||
definition | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
created_by_id | bigint | null | users.id 1:N · SET NULL | ||
yaml_definition_file | text | null | CHECK char_length(yaml_definition_file) <= 255 | ||
yaml_definition_file_store | smallint | NOT NULL | 1 | ||
deprecated | boolean | NOT NULL | FALSE |
Table checks: char_length(yaml_definition_file) <= 255; char_length(version) <= 50
project_id — ai_catalog_items.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.latest_released_version_id — ai_catalog_items.latest_released_version_id may be NULL, so the relationship to ai_catalog_item_versions is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.latest_version_id — ai_catalog_items.latest_version_id may be NULL, so the relationship to ai_catalog_item_versions is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id, identifier, item_type — ai_catalog_items soft-deletes via deleted_at, but (organization_id, identifier, item_type) is unique across live AND deleted rows. A user who deletes their account can never sign up again with the same value, and ON CONFLICT upserts will resurrect ghosts.project_id, latest_released_version_id, latest_version_id — ai_catalog_items has several nullable FKs (project_id, latest_released_version_id, latest_version_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
item_type | smallint | NOT NULL | |||
description | text | NOT NULL | CHECK char_length(description) <= 1024 | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
public | boolean | NOT NULL | FALSE | ||
deleted_at | timestamptz | null | |||
latest_version_id | bigint | null | ai_catalog_item_versions.id 1:N | ||
latest_released_version_id | bigint | null | ai_catalog_item_versions.id 1:N | ||
verification_level | smallint | NOT NULL | 0 | ||
identifier | text | null | CHECK char_length(identifier) <= 255 | ||
foundational_flow_reference | text | null | CHECK char_length(foundational_flow_reference) <= 255 | ||
last_30_day_usage_count | integer | NOT NULL | 0 | ||
last_30_day_usage_count_updated_at | timestamptz | NOT NULL | CAST('1970-01-01 00:00:00' AS timestamp) | ||
star_count | integer | NOT NULL | 0 | CHECK star_count >= 0 | |
visibility | smallint | NOT NULL | 0 |
Table checks: char_length(identifier) <= 255; char_length(description) <= 1024; char_length(foundational_flow_reference) <= 255; star_count >= 0; char_length(name) <= 255
created_by_id — ai_catalog_mcp_server_blocks.created_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
ai_catalog_mcp_server_id | bigint | NOT NULL | ai_catalog_mcp_servers.id 1:N · CASCADE | ||
created_by_id | bigint | null | users.id 1:N · SET NULL |
created_by_id — ai_catalog_mcp_servers.created_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
created_by_id | bigint | null | users.id 1:N · SET NULL | ||
transport | smallint | NOT NULL | 0 | ||
auth_type | smallint | NOT NULL | 0 | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
description | text | null | CHECK char_length(description) <= 2048 | ||
url | text | NOT NULL | CHECK char_length(url) <= 2048 | ||
homepage_url | text | null | CHECK char_length(homepage_url) <= 2048 | ||
oauth_client_xid | text | null | CHECK char_length(oauth_client_xid) <= 255 | ||
oauth_client_secret | jsonb | null |
Table checks: char_length(description) <= 2048; char_length(oauth_client_xid) <= 255; char_length(homepage_url) <= 2048; char_length(url) <= 2048; char_length(name) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
ai_catalog_mcp_server_id | bigint | NOT NULL | ai_catalog_mcp_servers.id 1:N · CASCADE | ||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
token | jsonb | null | |||
refresh_token | jsonb | null | |||
expires_at | timestamptz | null |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
thread_id | bigint | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
role | smallint | NOT NULL | |||
has_feedback | boolean | null | FALSE | ||
extras | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
error_details | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
content | text | NOT NULL | CHECK char_length(content) <= 524288 | ||
request_xid | text | null | CHECK char_length(request_xid) <= 255 | ||
message_xid | text | null | CHECK char_length(message_xid) <= 255 | ||
referer_url | text | null | CHECK char_length(referer_url) <= 255 |
Table checks: char_length(content) <= 524288; char_length(request_xid) <= 255; char_length(message_xid) <= 255; char_length(referer_url) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
last_updated_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
conversation_type | smallint | NOT NULL |
ai_self_hosted_model_id — ai_feature_settings.ai_self_hosted_model_id may be NULL, so the relationship to ai_self_hosted_models is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
ai_self_hosted_model_id | bigint | null | ai_self_hosted_models.id 1:N · CASCADE | ||
feature | smallint | NOT NULL | |||
provider | smallint | NOT NULL |
ai_catalog_item_consumer_id — ai_flow_triggers.ai_catalog_item_consumer_id may be NULL, so the relationship to ai_catalog_item_consumers is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
user_id | bigint | null | |||
config_path | text | null | CHECK char_length(config_path) <= 255 | ||
description | text | NOT NULL | CHECK char_length(description) <= 255 | ||
event_types | smallint[] | NOT NULL | CAST('{}' AS smallint[]) | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
ai_catalog_item_consumer_id | bigint | null | ai_catalog_item_consumers.id 1:N · CASCADE | ||
filter | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(filter) = CAST('object' AS text) |
Table checks: char_length(description) <= 255; jsonb_typeof(filter) = CAST('object' AS text); num_nonnulls(ai_catalog_item_consumer_id, user_id) <= 1; char_length(config_path) <= 255
ai_instance_accessible_entity_rules references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
through_namespace_id | bigint | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
accessible_entity | text | NOT NULL | CHECK char_length(accessible_entity) <= 255 |
Table checks: char_length(accessible_entity) <= 255
through_namespace_id — ai_namespace_feature_access_rules.through_namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
root_namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
through_namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
accessible_entity | text | NOT NULL | CHECK char_length(accessible_entity) <= 255 |
Table checks: char_length(accessible_entity) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
feature | smallint | NOT NULL | |||
offered_model_ref | text | null | CHECK char_length(offered_model_ref) <= 255 | ||
offered_model_name | text | null | CHECK char_length(offered_model_name) <= 255 | ||
model_allowlist_enabled | boolean | NOT NULL | FALSE | ||
model_allowlist_gitlab_model_refs | text[] | NOT NULL | CAST('{}' AS text[]) | CHECK cardinality(model_allowlist_gitlab_model_refs) <= 100 |
Table checks: cardinality(model_allowlist_gitlab_model_refs) <= 100; char_length(offered_model_ref) <= 255; char_length(offered_model_name) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
model | smallint | NOT NULL | |||
endpoint | text | null | CHECK char_length(endpoint) <= 2048 | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
encrypted_api_token | bytea | null | |||
encrypted_api_token_iv | bytea | null | |||
identifier | text | null | CHECK char_length(identifier) <= 255 | ||
provider | smallint | NOT NULL | 0 |
Table checks: char_length(endpoint) <= 2048; char_length(name) <= 255; char_length(identifier) <= 255
amazon_q_oauth_application_id — ai_settings.amazon_q_oauth_application_id may be NULL, so the relationship to oauth_applications is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id — ai_settings.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.duo_workflow_oauth_application_id — ai_settings.duo_workflow_oauth_application_id may be NULL, so the relationship to oauth_applications is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.duo_workflow_service_account_user_id — ai_settings.duo_workflow_service_account_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.amazon_q_service_account_user_id — ai_settings.amazon_q_service_account_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.amazon_q_oauth_application_id, organization_id, duo_workflow_oauth_application_id, duo_workflow_service_account_user_id, amazon_q_service_account_user_id — ai_settings has several nullable FKs (amazon_q_oauth_application_id, organization_id, duo_workflow_oauth_application_id, duo_workflow_service_account_user_id, amazon_q_service_account_user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
ai_gateway_url | text | null | CHECK char_length(ai_gateway_url) <= 2048 | ||
singleton | boolean | NOT NULL | TRUE | Always true, used for singleton enforcement | |
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
amazon_q_oauth_application_id | bigint | null | oauth_applications.id 1:N · SET NULL | ||
amazon_q_service_account_user_id | bigint | null | users.id 1:N · SET NULL | ||
amazon_q_ready | boolean | NOT NULL | FALSE | ||
amazon_q_role_arn | text | null | CHECK char_length(amazon_q_role_arn) <= 2048 | ||
duo_workflow_service_account_user_id | bigint | null | users.id 1:N · SET NULL | ||
duo_workflow_oauth_application_id | bigint | null | oauth_applications.id 1:N · SET NULL | ||
enabled_instance_verbose_ai_logs | boolean | null | |||
duo_core_features_enabled | boolean | null | |||
duo_agent_platform_service_url | text | null | CHECK char_length(duo_agent_platform_service_url) <= 2048 | ||
foundational_agents_default_enabled | boolean | null | TRUE | ||
ai_gateway_timeout_seconds | integer | null | 60 | ||
minimum_access_level_execute | smallint | null | |||
minimum_access_level_manage | smallint | null | |||
minimum_access_level_enable_on_projects | smallint | null | |||
minimum_access_level_execute_async | smallint | null | |||
feature_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(feature_settings) = CAST('object' AS text) | |
self_hosted_duo_agent_platform_service_secure | boolean | NOT NULL | TRUE | ||
include_recommended_allowed | boolean | NOT NULL | FALSE | ||
allow_all_unix_sockets | boolean | NOT NULL | FALSE | ||
enforce_on_local_clients | boolean | NOT NULL | FALSE | ||
allow_project_extension | boolean | NOT NULL | TRUE | ||
allowed_domains | text[] | NOT NULL | CAST('{}' AS text[]) | ||
denied_domains | text[] | NOT NULL | CAST('{}' AS text[]) | ||
organization_id | bigint | null | organizations.id 1:1 · CASCADE | CHECK organization_id IS NOT NULL |
Table checks: char_length(ai_gateway_url) <= 2048; organization_id IS NOT NULL; char_length(duo_agent_platform_service_url) <= 2048; char_length(amazon_q_role_arn) <= 2048; jsonb_typeof(feature_settings) = CAST('object' AS text)
ai_testing_terms_acceptances references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
user_id PK | bigint | NOT NULL | |||
user_email | text | NOT NULL | CHECK char_length(user_email) <= 255 |
Table checks: char_length(user_email) <= 255
project_id — ai_tool_rules.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
web_access | smallint | null | CHECK web_access IS NULL OR web_access = ANY(ARRAY[0, 1, 2]) | ||
local_access | smallint | null | CHECK local_access IS NULL OR local_access = ANY(ARRAY[0, 1, 2]) | ||
tool_name | text | NOT NULL | CHECK char_length(tool_name) <= 255 | ||
tool_source | text | null | CHECK char_length(tool_source) <= 255 | ||
tool_arguments | jsonb | null | |||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
background_access | smallint | null | CHECK background_access IS NULL OR background_access = ANY(ARRAY[0, 2]) |
Table checks: char_length(tool_source) <= 255; char_length(tool_name) <= 255; background_access IS NULL OR background_access = ANY(ARRAY[0, 2]); web_access IS NOT NULL OR local_access IS NOT NULL OR background_access IS NOT NULL; local_access IS NULL OR local_access = ANY(ARRAY[0, 1, 2]); web_access IS NULL OR web_access = ANY(ARRAY[0, 1, 2])
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
user_id PK | bigint | NOT NULL | users.id 1:1 · CASCADE | ||
last_duo_activity_on | date | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
ai_vectorizable_file_upload_id | bigint | NOT NULL | ai_vectorizable_file_uploads.id 1:1 · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 |
Table checks: char_length(verification_failure) <= 255
model_type — model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.model_type, model_id — model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | nextval(CAST('uploads_id_seq' AS regclass)) | ||
size | bigint | NOT NULL | |||
model_id | bigint | NOT NULL | |||
uploaded_by_user_id | bigint | null | |||
organization_id | bigint | null | |||
namespace_id | bigint | null | |||
project_id | bigint | null | CHECK project_id IS NOT NULL | ||
created_at | timestamp | null | |||
store | integer | NOT NULL | 1 | ||
version | integer | null | 1 | ||
path | text | NOT NULL | CHECK char_length(path) <= 511 | ||
checksum | text | null | CHECK char_length(checksum) <= 64 | ||
model_type PK | text | NOT NULL | |||
uploader | text | NOT NULL | |||
mount_point | text | null | |||
secret | text | null |
Table checks: char_length(path) <= 511; project_id IS NOT NULL; char_length(checksum) <= 64
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
file | text | NOT NULL | CHECK char_length(file) <= 255 |
Table checks: char_length(file) <= 255; char_length(name) <= 255
Tables whose names start with audit_.
audit_events_amazon_s3_configurations.stream_destination_id → audit_events_group_external_streaming_destinations.id — one audit_events_group_external_streaming_destinations, many audit_events_amazon_s3_configurations · optional · ON DELETE SET NULL · indexedaudit_events_amazon_s3_configurations.namespace_id → namespaces.id — one namespaces, many audit_events_amazon_s3_configurations · required · ON DELETE CASCADE · indexedaudit_events_external_audit_event_destinations.stream_destination_id → audit_events_group_external_streaming_destinations.id — one audit_events_group_external_streaming_destinations, many audit_events_external_audit_event_destinations · optional · ON DELETE SET NULL · indexedaudit_events_external_audit_event_destinations.namespace_id → namespaces.id — one namespaces, many audit_events_external_audit_event_destinations · required · ON DELETE CASCADE · indexedaudit_events_google_cloud_logging_configurations.stream_destination_id → audit_events_group_external_streaming_destinations.id — one audit_events_group_external_streaming_destinations, many audit_events_google_cloud_logging_configurations · optional · ON DELETE SET NULL · indexedaudit_events_google_cloud_logging_configurations.namespace_id → namespaces.id — one namespaces, many audit_events_google_cloud_logging_configurations · required · ON DELETE CASCADE · indexedaudit_events_group_external_streaming_destinations.group_id → namespaces.id — one namespaces, many audit_events_group_external_streaming_destinations · required · ON DELETE CASCADE · indexedaudit_events_group_streaming_event_type_filters.namespace_id → namespaces.id — one namespaces, many audit_events_group_streaming_event_type_filters · required · ON DELETE CASCADE · indexedaudit_events_group_streaming_event_type_filters.external_streaming_destination_id → audit_events_group_external_streaming_destinations.id — one audit_events_group_external_streaming_destinations, many audit_events_group_streaming_event_type_filters · required · ON DELETE CASCADE · indexedaudit_events_instance_amazon_s3_configurations.stream_destination_id → audit_events_instance_external_streaming_destinations.id — one audit_events_instance_external_streaming_destinations, many audit_events_instance_amazon_s3_configurations · optional · ON DELETE SET NULL · indexedaudit_events_instance_external_audit_event_destinations.stream_destination_id → audit_events_instance_external_streaming_destinations.id — one audit_events_instance_external_streaming_destinations, many audit_events_instance_external_audit_event_destinations · optional · ON DELETE SET NULL · indexedaudit_events_instance_google_cloud_logging_configurations.stream_destination_id → audit_events_instance_external_streaming_destinations.id — one audit_events_instance_external_streaming_destinations, many audit_events_instance_google_cloud_logging_configurations · optional · ON DELETE SET NULL · indexedaudit_events_instance_streaming_event_type_filters.external_streaming_destination_id → audit_events_instance_external_streaming_destinations.id — one audit_events_instance_external_streaming_destinations, many audit_events_instance_streaming_event_type_filters · required · ON DELETE CASCADE · indexedaudit_events_streaming_event_type_filters.group_id → namespaces.id — one namespaces, many audit_events_streaming_event_type_filters · optional · ON DELETE CASCADE · indexedaudit_events_streaming_event_type_filters.external_audit_event_destination_id → audit_events_external_audit_event_destinations.id — one audit_events_external_audit_event_destinations, many audit_events_streaming_event_type_filters · required · ON DELETE CASCADE · indexedaudit_events_streaming_group_namespace_filters.external_streaming_destination_id → audit_events_group_external_streaming_destinations.id — one audit_events_group_external_streaming_destinations, many audit_events_streaming_group_namespace_filters · required · ON DELETE CASCADE · indexedaudit_events_streaming_group_namespace_filters.namespace_id → namespaces.id — one namespaces, many audit_events_streaming_group_namespace_filters · required · ON DELETE CASCADE · indexedaudit_events_streaming_headers.group_id → namespaces.id — one namespaces, many audit_events_streaming_headers · optional · ON DELETE CASCADE · indexedaudit_events_streaming_headers.external_audit_event_destination_id → audit_events_external_audit_event_destinations.id — one audit_events_external_audit_event_destinations, many audit_events_streaming_headers · required · ON DELETE CASCADE · indexedaudit_events_streaming_http_group_namespace_filters.namespace_id → namespaces.id — one namespaces, at most one audit_events_streaming_http_group_namespace_filters · required · ON DELETE CASCADE · indexedaudit_events_streaming_http_group_namespace_filters.external_audit_event_destination_id → audit_events_external_audit_event_destinations.id — one audit_events_external_audit_event_destinations, at most one audit_events_streaming_http_group_namespace_filters · required · ON DELETE CASCADE · indexedaudit_events_streaming_http_instance_namespace_filters.namespace_id → namespaces.id — one namespaces, many audit_events_streaming_http_instance_namespace_filters · required · ON DELETE CASCADE · indexedaudit_events_streaming_http_instance_namespace_filters.audit_events_instance_external_audit_event_destination_id → audit_events_instance_external_audit_event_destinations.id — one audit_events_instance_external_audit_event_destinations, at most one audit_events_streaming_http_instance_namespace_filters · required · ON DELETE CASCADE · indexedaudit_events_streaming_instance_event_type_filters.instance_external_audit_event_destination_id → audit_events_instance_external_audit_event_destinations.id — one audit_events_instance_external_audit_event_destinations, many audit_events_streaming_instance_event_type_filters · required · ON DELETE CASCADE · indexedaudit_events_streaming_instance_namespace_filters.external_streaming_destination_id → audit_events_instance_external_streaming_destinations.id — one audit_events_instance_external_streaming_destinations, many audit_events_streaming_instance_namespace_filters · required · ON DELETE CASCADE · indexedaudit_events_streaming_instance_namespace_filters.namespace_id → namespaces.id — one namespaces, many audit_events_streaming_instance_namespace_filters · required · ON DELETE CASCADE · indexedentity_type — entity_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.entity_type, entity_id — entity_id points at different tables depending on entity_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.target_type, target_id — target_id points at different tables depending on target_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.audit_events references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | nextval(CAST('audit_events_id_seq' AS regclass)) | ||
author_id | bigint | NOT NULL | |||
entity_id | bigint | NOT NULL | |||
entity_type | varchar | NOT NULL | |||
details | text | null | |||
ip_address | inet | null | |||
author_name | text | null | CHECK char_length(author_name) <= 255 | ||
entity_path | text | null | CHECK char_length(entity_path) <= 5500 | ||
target_details | text | null | CHECK char_length(target_details) <= 5500 | ||
created_at PK | timestamp | NOT NULL | |||
target_type | text | null | CHECK char_length(target_type) <= 255 | ||
target_id | bigint | null |
Table checks: char_length(entity_path) <= 5500; char_length(author_name) <= 255; char_length(target_type) <= 255; char_length(target_details) <= 5500
stream_destination_id — audit_events_amazon_s3_configurations.stream_destination_id may be NULL, so the relationship to audit_events_group_external_streaming_destinations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
access_key_xid | text | NOT NULL | CHECK char_length(access_key_xid) <= 128 | ||
name | text | NOT NULL | CHECK char_length(name) <= 72 | ||
bucket_name | text | NOT NULL | CHECK char_length(bucket_name) <= 63 | ||
aws_region | text | NOT NULL | CHECK char_length(aws_region) <= 50 | ||
encrypted_secret_access_key | bytea | NOT NULL | |||
encrypted_secret_access_key_iv | bytea | NOT NULL | |||
stream_destination_id | bigint | null | audit_events_group_external_streaming_destinations.id 1:N · SET NULL | ||
active | boolean | NOT NULL | TRUE |
Table checks: char_length(bucket_name) <= 63; char_length(aws_region) <= 50; char_length(name) <= 72; char_length(access_key_xid) <= 128
stream_destination_id — audit_events_external_audit_event_destinations.stream_destination_id may be NULL, so the relationship to audit_events_group_external_streaming_destinations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
destination_url | text | NOT NULL | CHECK char_length(destination_url) <= 255 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
verification_token | text | null | CHECK char_length(verification_token) <= 24 | ||
name | text | NOT NULL | CHECK char_length(name) <= 72 | ||
stream_destination_id | bigint | null | audit_events_group_external_streaming_destinations.id 1:N · SET NULL | ||
active | boolean | NOT NULL | TRUE |
Table checks: char_length(destination_url) <= 255; char_length(verification_token) <= 24; char_length(name) <= 72
stream_destination_id — audit_events_google_cloud_logging_configurations.stream_destination_id may be NULL, so the relationship to audit_events_group_external_streaming_destinations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
google_project_id_name | text | NOT NULL | CHECK char_length(google_project_id_name) <= 30 | ||
client_email | text | NOT NULL | CHECK char_length(client_email) <= 254 | ||
log_id_name | text | null | CAST('audit_events' AS text) | CHECK char_length(log_id_name) <= 511 | |
encrypted_private_key | bytea | NOT NULL | |||
encrypted_private_key_iv | bytea | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 72 | ||
stream_destination_id | bigint | null | audit_events_group_external_streaming_destinations.id 1:N · SET NULL | ||
active | boolean | NOT NULL | TRUE |
Table checks: char_length(client_email) <= 254; char_length(google_project_id_name) <= 30; char_length(log_id_name) <= 511; char_length(name) <= 72
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
category | smallint | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 72 | ||
config | jsonb | NOT NULL | |||
encrypted_secret_token | bytea | NOT NULL | CHECK octet_length(encrypted_secret_token) <= 4112 | ||
encrypted_secret_token_iv | bytea | NOT NULL | |||
legacy_destination_ref | bigint | null | |||
active | boolean | NOT NULL | TRUE |
Table checks: char_length(name) <= 72; octet_length(encrypted_secret_token) <= 4112
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
external_streaming_destination_id | bigint | NOT NULL | audit_events_group_external_streaming_destinations.id 1:N · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
audit_event_type | text | NOT NULL | CHECK char_length(audit_event_type) <= 255 | ||
kind | smallint | NOT NULL | 0 |
Table checks: char_length(audit_event_type) <= 255
stream_destination_id — audit_events_instance_amazon_s3_configurations.stream_destination_id may be NULL, so the relationship to audit_events_instance_external_streaming_destinations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
access_key_xid | text | NOT NULL | CHECK char_length(access_key_xid) <= 128 | ||
name | text | NOT NULL | CHECK char_length(name) <= 72 | ||
bucket_name | text | NOT NULL | CHECK char_length(bucket_name) <= 63 | ||
aws_region | text | NOT NULL | CHECK char_length(aws_region) <= 50 | ||
encrypted_secret_access_key | bytea | NOT NULL | |||
encrypted_secret_access_key_iv | bytea | NOT NULL | |||
stream_destination_id | bigint | null | audit_events_instance_external_streaming_destinations.id 1:N · SET NULL | ||
active | boolean | NOT NULL | TRUE |
Table checks: char_length(name) <= 72; char_length(bucket_name) <= 63; char_length(aws_region) <= 50; char_length(access_key_xid) <= 128
stream_destination_id — audit_events_instance_external_audit_event_destinations.stream_destination_id may be NULL, so the relationship to audit_events_instance_external_streaming_destinations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
destination_url | text | NOT NULL | CHECK char_length(destination_url) <= 255 | ||
encrypted_verification_token | bytea | NOT NULL | |||
encrypted_verification_token_iv | bytea | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 72 | ||
stream_destination_id | bigint | null | audit_events_instance_external_streaming_destinations.id 1:N · SET NULL | ||
active | boolean | NOT NULL | TRUE |
Table checks: char_length(name) <= 72; char_length(destination_url) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
category | smallint | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 72 | ||
config | jsonb | NOT NULL | |||
encrypted_secret_token | bytea | NOT NULL | CHECK octet_length(encrypted_secret_token) <= 4112 | ||
encrypted_secret_token_iv | bytea | NOT NULL | |||
legacy_destination_ref | bigint | null | |||
active | boolean | NOT NULL | TRUE |
Table checks: char_length(name) <= 72; octet_length(encrypted_secret_token) <= 4112
stream_destination_id — audit_events_instance_google_cloud_logging_configurations.stream_destination_id may be NULL, so the relationship to audit_events_instance_external_streaming_destinations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
google_project_id_name | text | NOT NULL | CHECK char_length(google_project_id_name) <= 30 | ||
client_email | text | NOT NULL | CHECK char_length(client_email) <= 254 | ||
log_id_name | text | null | CAST('audit_events' AS text) | CHECK char_length(log_id_name) <= 511 | |
name | text | NOT NULL | CHECK char_length(name) <= 72 | ||
encrypted_private_key | bytea | NOT NULL | |||
encrypted_private_key_iv | bytea | NOT NULL | |||
stream_destination_id | bigint | null | audit_events_instance_external_streaming_destinations.id 1:N · SET NULL | ||
active | boolean | NOT NULL | TRUE |
Table checks: char_length(client_email) <= 254; char_length(log_id_name) <= 511; char_length(google_project_id_name) <= 30; char_length(name) <= 72
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
external_streaming_destination_id | bigint | NOT NULL | audit_events_instance_external_streaming_destinations.id 1:N · CASCADE | ||
audit_event_type | text | NOT NULL | CHECK char_length(audit_event_type) <= 255 |
Table checks: char_length(audit_event_type) <= 255
group_id — audit_events_streaming_event_type_filters.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
external_audit_event_destination_id | bigint | NOT NULL | audit_events_external_audit_event_destinations.id 1:N · CASCADE | ||
audit_event_type | text | NOT NULL | CHECK char_length(audit_event_type) <= 255 | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK group_id IS NOT NULL |
Table checks: group_id IS NOT NULL; char_length(audit_event_type) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
external_streaming_destination_id | bigint | NOT NULL | audit_events_group_external_streaming_destinations.id 1:N · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE |
group_id — audit_events_streaming_headers.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
external_audit_event_destination_id | bigint | NOT NULL | audit_events_external_audit_event_destinations.id 1:N · CASCADE | ||
key | text | NOT NULL | CHECK char_length(key) <= 255 | ||
value | text | NOT NULL | CHECK char_length(value) <= 2000 | ||
active | boolean | NOT NULL | TRUE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK group_id IS NOT NULL |
Table checks: group_id IS NOT NULL; char_length(key) <= 255; char_length(value) <= 2000
external_audit_event_destination_id, namespace_id — audit_events_streaming_http_group_namespace_filters looks like a many-to-many link table but has no unique constraint across (external_audit_event_destination_id, namespace_id). The same pair can be inserted twice; every join through it will double-count.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
external_audit_event_destination_id | bigint | NOT NULL | audit_events_external_audit_event_destinations.id 1:1 · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:1 · CASCADE |
audit_events_instance_external_audit_event_destination_id, namespace_id — audit_events_streaming_http_instance_namespace_filters looks like a many-to-many link table but has no unique constraint across (audit_events_instance_external_audit_event_destination_id, namespace_id). The same pair can be inserted twice; every join through it will double-count.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
audit_events_instance_external_audit_event_destination_id | bigint | NOT NULL | audit_events_instance_external_audit_event_destinations.id 1:1 · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
instance_external_audit_event_destination_id | bigint | NOT NULL | audit_events_instance_external_audit_event_destinations.id 1:N · CASCADE | ||
audit_event_type | text | NOT NULL | CHECK char_length(audit_event_type) <= 255 |
Table checks: char_length(audit_event_type) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
external_streaming_destination_id | bigint | NOT NULL | audit_events_instance_external_streaming_destinations.id 1:N · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE |
Tables whose names start with background_.
No foreign keys in this domain.
background_operation_jobs references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | uuid | NOT NULL | gen_random_uuid() | ||
worker_id | uuid | NOT NULL | |||
organization_id | bigint | NOT NULL | |||
partition PK | bigint | NOT NULL | 1 | ||
worker_partition | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
started_at | timestamptz | null | |||
finished_at | timestamptz | null | |||
batch_size | integer | NOT NULL | |||
sub_batch_size | integer | NOT NULL | |||
pause_ms | integer | NOT NULL | 100 | CHECK pause_ms >= 100 | |
status | smallint | NOT NULL | 0 | ||
attempts | smallint | NOT NULL | 0 | ||
metrics | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
min_cursor | jsonb | null | |||
max_cursor | jsonb | null |
Table checks: jsonb_typeof(min_cursor) = CAST('array' AS text) AND jsonb_typeof(max_cursor) = CAST('array' AS text); pause_ms >= 100; num_nonnulls(min_cursor, max_cursor) = 2
background_operation_jobs_cell_local references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
partition PK | bigint | NOT NULL | 1 | ||
worker_id | bigint | NOT NULL | |||
worker_partition | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
started_at | timestamptz | null | |||
finished_at | timestamptz | null | |||
batch_size | integer | NOT NULL | |||
sub_batch_size | integer | NOT NULL | |||
pause_ms | integer | NOT NULL | 100 | CHECK pause_ms >= 100 | |
status | smallint | NOT NULL | 0 | ||
attempts | smallint | NOT NULL | 0 | ||
metrics | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
min_cursor | jsonb | null | |||
max_cursor | jsonb | null |
Table checks: pause_ms >= 100; num_nonnulls(min_cursor, max_cursor) = 2; jsonb_typeof(min_cursor) = CAST('array' AS text) AND jsonb_typeof(max_cursor) = CAST('array' AS text)
background_operation_workers references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | uuid | NOT NULL | gen_random_uuid() | ||
organization_id | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | |||
total_tuple_count | bigint | null | |||
partition PK | bigint | NOT NULL | 1 | ||
started_at | timestamptz | null | |||
on_hold_until | timestamptz | null | |||
created_at | timestamptz | NOT NULL | |||
finished_at | timestamptz | null | |||
batch_size | integer | NOT NULL | |||
sub_batch_size | integer | NOT NULL | CHECK sub_batch_size > 0 | ||
pause_ms | integer | NOT NULL | 100 | ||
max_batch_size | integer | null | |||
priority | smallint | NOT NULL | 0 | ||
status | smallint | NOT NULL | 0 | ||
interval | smallint | NOT NULL | |||
job_class_name | text | NOT NULL | CHECK char_length(job_class_name) <= 100 | ||
batch_class_name | text | NOT NULL | CHECK char_length(batch_class_name) <= 100 | ||
table_name | text | NOT NULL | CHECK char_length(table_name) <= 63 | ||
column_name | text | NOT NULL | CHECK char_length(column_name) <= 63 | ||
gitlab_schema | text | NOT NULL | CHECK char_length(gitlab_schema) <= 255 | ||
job_arguments | jsonb | null | CAST('[]' AS jsonb) | ||
min_cursor | jsonb | null | |||
max_cursor | jsonb | null |
Table checks: char_length(column_name) <= 63; char_length(gitlab_schema) <= 255; sub_batch_size > 0; char_length(job_class_name) <= 100; char_length(batch_class_name) <= 100; char_length(table_name) <= 63; batch_size >= sub_batch_size; num_nonnulls(min_cursor, max_cursor) = 2; jsonb_typeof(min_cursor) = CAST('array' AS text) AND jsonb_typeof(max_cursor) = CAST('array' AS text)
background_operation_workers_cell_local references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
total_tuple_count | bigint | null | |||
partition PK | bigint | NOT NULL | 1 | ||
started_at | timestamptz | null | |||
on_hold_until | timestamptz | null | |||
created_at | timestamptz | NOT NULL | |||
finished_at | timestamptz | null | |||
batch_size | integer | NOT NULL | |||
sub_batch_size | integer | NOT NULL | CHECK sub_batch_size > 0 | ||
pause_ms | integer | NOT NULL | 100 | ||
max_batch_size | integer | null | |||
priority | smallint | NOT NULL | 0 | ||
status | smallint | NOT NULL | 0 | ||
interval | smallint | NOT NULL | |||
job_class_name | text | NOT NULL | CHECK char_length(job_class_name) <= 100 | ||
batch_class_name | text | NOT NULL | CHECK char_length(batch_class_name) <= 100 | ||
table_name | text | NOT NULL | CHECK char_length(table_name) <= 63 | ||
column_name | text | NOT NULL | CHECK char_length(column_name) <= 63 | ||
gitlab_schema | text | NOT NULL | CHECK char_length(gitlab_schema) <= 255 | ||
job_arguments | jsonb | null | CAST('[]' AS jsonb) | ||
min_cursor | jsonb | null | |||
max_cursor | jsonb | null |
Table checks: char_length(table_name) <= 63; char_length(column_name) <= 63; char_length(gitlab_schema) <= 255; char_length(batch_class_name) <= 100; batch_size >= sub_batch_size; char_length(job_class_name) <= 100; num_nonnulls(min_cursor, max_cursor) = 2; sub_batch_size > 0; jsonb_typeof(min_cursor) = CAST('array' AS text) AND jsonb_typeof(max_cursor) = CAST('array' AS text)
Tables whose names start with backup_.
No foreign keys in this domain.
backup_finding_evidences references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
original_record_identifier PK | bigint | NOT NULL | |||
finding_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
date PK | date | NOT NULL | |||
data | jsonb | NOT NULL |
backup_finding_flags references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
original_record_identifier PK | bigint | NOT NULL | |||
finding_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
date PK | date | NOT NULL | |||
data | jsonb | NOT NULL |
backup_finding_identifiers references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
original_record_identifier PK | bigint | NOT NULL | |||
finding_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
date PK | date | NOT NULL | |||
data | jsonb | NOT NULL |
backup_finding_links references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
original_record_identifier PK | bigint | NOT NULL | |||
finding_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
date PK | date | NOT NULL | |||
data | jsonb | NOT NULL |
backup_finding_remediations references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
original_record_identifier PK | bigint | NOT NULL | |||
finding_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
date PK | date | NOT NULL | |||
data | jsonb | NOT NULL |
backup_finding_signatures references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
original_record_identifier PK | bigint | NOT NULL | |||
finding_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
date PK | date | NOT NULL | |||
data | jsonb | NOT NULL |
backup_findings references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
original_record_identifier PK | bigint | NOT NULL | |||
vulnerability_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
date PK | date | NOT NULL | |||
data | jsonb | NOT NULL |
backup_vulnerabilities references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
original_record_identifier PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
date PK | date | NOT NULL | |||
data | jsonb | NOT NULL | |||
traversal_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
backup_vulnerability_external_issue_links references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
original_record_identifier PK | bigint | NOT NULL | |||
vulnerability_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
date PK | date | NOT NULL | |||
data | jsonb | NOT NULL |
backup_vulnerability_issue_links references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
original_record_identifier PK | bigint | NOT NULL | |||
vulnerability_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
date PK | date | NOT NULL | |||
data | jsonb | NOT NULL |
backup_vulnerability_merge_request_links references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
original_record_identifier PK | bigint | NOT NULL | |||
vulnerability_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
date PK | date | NOT NULL | |||
data | jsonb | NOT NULL |
backup_vulnerability_reads references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
original_record_identifier PK | bigint | NOT NULL | |||
vulnerability_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
date PK | date | NOT NULL | |||
data | jsonb | NOT NULL |
backup_vulnerability_severity_overrides references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
original_record_identifier PK | bigint | NOT NULL | |||
vulnerability_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
date PK | date | NOT NULL | |||
data | jsonb | NOT NULL |
backup_vulnerability_state_transitions references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
original_record_identifier PK | bigint | NOT NULL | |||
vulnerability_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
date PK | date | NOT NULL | |||
data | jsonb | NOT NULL |
backup_vulnerability_user_mentions references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
original_record_identifier PK | bigint | NOT NULL | |||
vulnerability_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
date PK | date | NOT NULL | |||
data | jsonb | NOT NULL |
Tables whose names start with batched_.
batched_background_migration_job_transition_logs.batched_background_migration_job_id → batched_background_migration_jobs.id — one batched_background_migration_jobs, many batched_background_migration_job_transition_logs · required · ON DELETE CASCADE · indexedbatched_background_migration_jobs.batched_background_migration_id → batched_background_migrations.id — one batched_background_migrations, many batched_background_migration_jobs · required · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
batched_background_migration_job_id | bigint | NOT NULL | batched_background_migration_jobs.id 1:N · CASCADE | ||
created_at PK | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
previous_status | smallint | NOT NULL | |||
next_status | smallint | NOT NULL | |||
exception_class | text | null | CHECK char_length(exception_class) <= 100 | ||
exception_message | text | null | CHECK char_length(exception_message) <= 1000 |
Table checks: char_length(exception_message) <= 1000; char_length(exception_class) <= 100
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
started_at | timestamptz | null | |||
finished_at | timestamptz | null | |||
batched_background_migration_id | bigint | NOT NULL | batched_background_migrations.id 1:N · CASCADE | ||
min_value | bigint | null | |||
max_value | bigint | null | |||
batch_size | integer | NOT NULL | |||
sub_batch_size | integer | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
attempts | smallint | NOT NULL | 0 | ||
metrics | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
pause_ms | integer | NOT NULL | 100 | CHECK pause_ms >= 100 | |
min_cursor | jsonb | null | |||
max_cursor | jsonb | null |
Table checks: jsonb_typeof(min_cursor) = CAST('array' AS text) AND jsonb_typeof(max_cursor) = CAST('array' AS text); pause_ms >= 100; num_nonnulls(min_value, max_value) = 2 OR num_nonnulls(min_cursor, max_cursor) = 2
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
min_value | bigint | null | 1 | CHECK min_value > 0 | |
max_value | bigint | null | |||
batch_size | integer | NOT NULL | |||
sub_batch_size | integer | NOT NULL | CHECK sub_batch_size > 0 | ||
interval | smallint | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
job_class_name | text | NOT NULL | CHECK char_length(job_class_name) <= 100 | ||
batch_class_name | text | NOT NULL | CAST('PrimaryKeyBatchingStrategy' AS text) | CHECK char_length(batch_class_name) <= 100 | |
table_name | text | NOT NULL | CHECK char_length(table_name) <= 63 | ||
column_name | text | NOT NULL | CHECK char_length(column_name) <= 63 | ||
job_arguments | jsonb | NOT NULL | CAST('[]' AS jsonb) | ||
total_tuple_count | bigint | null | |||
pause_ms | integer | NOT NULL | 100 | CHECK pause_ms >= 100 | |
max_batch_size | integer | null | |||
started_at | timestamptz | null | |||
on_hold_until | timestamptz | null | execution of this migration is on hold until this time | ||
gitlab_schema | text | NOT NULL | CHECK char_length(gitlab_schema) <= 255 | ||
finished_at | timestamptz | null | |||
queued_migration_version | text | null | CHECK char_length(queued_migration_version) <= 14 | ||
min_cursor | jsonb | null | |||
max_cursor | jsonb | null |
Table checks: char_length(gitlab_schema) <= 255; jsonb_typeof(min_cursor) = CAST('array' AS text) AND jsonb_typeof(max_cursor) = CAST('array' AS text); char_length(column_name) <= 63; pause_ms >= 100; char_length(table_name) <= 63; char_length(queued_migration_version) <= 14; batch_size >= sub_batch_size; char_length(job_class_name) <= 100; num_nonnulls(min_value, max_value) = 2 OR num_nonnulls(min_cursor, max_cursor) = 2; char_length(batch_class_name) <= 100; max_value >= min_value; min_value > 0; sub_batch_size > 0
Tables whose names start with p_.
p_ci_build_names.partition_id, build_id → p_ci_builds.partition_id, id — one p_ci_builds, at most one p_ci_build_names · required · ON DELETE CASCADE · indexedp_ci_build_needs.partition_id, build_id → p_ci_builds.partition_id, id — one p_ci_builds, many p_ci_build_needs · required · ON DELETE CASCADE · not indexedp_ci_build_sources.partition_id, build_id → p_ci_builds.partition_id, id — one p_ci_builds, at most one p_ci_build_sources · required · ON DELETE CASCADE · indexedp_ci_build_trace_metadata.partition_id, trace_artifact_id → p_ci_job_artifacts.partition_id, id — one p_ci_job_artifacts, many p_ci_build_trace_metadata · optional · ON DELETE CASCADE · not indexedp_ci_build_trace_metadata.partition_id, build_id → p_ci_builds.partition_id, id — one p_ci_builds, at most one p_ci_build_trace_metadata · required · ON DELETE CASCADE · indexedp_ci_builds.partition_id, stage_id → p_ci_stages.partition_id, id — one p_ci_stages, many p_ci_builds · optional · ON DELETE CASCADE · not indexedp_ci_builds.resource_group_id → ci_resource_groups.id — one ci_resource_groups, many p_ci_builds · optional · ON DELETE SET NULL · indexedp_ci_builds.upstream_pipeline_partition_id, upstream_pipeline_id → p_ci_pipelines.partition_id, id — one p_ci_pipelines, many p_ci_builds · optional · ON DELETE CASCADE · not indexedp_ci_builds.auto_canceled_by_partition_id, auto_canceled_by_id → p_ci_pipelines.partition_id, id — one p_ci_pipelines, many p_ci_builds · optional · ON DELETE SET NULL · not indexedp_ci_builds.partition_id, commit_id → p_ci_pipelines.partition_id, id — one p_ci_pipelines, many p_ci_builds · optional · ON DELETE CASCADE · not indexedp_ci_job_annotations.partition_id, job_id → p_ci_builds.partition_id, id — one p_ci_builds, many p_ci_job_annotations · required · ON DELETE CASCADE · indexedp_ci_job_artifact_reports.partition_id, job_artifact_id → p_ci_job_artifacts.partition_id, id — one p_ci_job_artifacts, at most one p_ci_job_artifact_reports · required · ON DELETE CASCADE · indexedp_ci_job_artifacts.partition_id, job_id → p_ci_builds.partition_id, id — one p_ci_builds, many p_ci_job_artifacts · required · ON DELETE CASCADE · not indexedp_ci_job_definition_instances.partition_id, job_definition_id → p_ci_job_definitions.partition_id, id — one p_ci_job_definitions, many p_ci_job_definition_instances · required · ON DELETE RESTRICT · not indexedp_ci_job_definition_instances.partition_id, job_id → p_ci_builds.partition_id, id — one p_ci_builds, at most one p_ci_job_definition_instances · required · ON DELETE CASCADE · indexedp_ci_job_inputs.partition_id, job_id → p_ci_builds.partition_id, id — one p_ci_builds, many p_ci_job_inputs · required · ON DELETE CASCADE · not indexedp_ci_job_messages.partition_id, job_id → p_ci_builds.partition_id, id — one p_ci_builds, many p_ci_job_messages · required · ON DELETE CASCADE · not indexedp_ci_pipeline_variables.partition_id, pipeline_id → p_ci_pipelines.partition_id, id — one p_ci_pipelines, many p_ci_pipeline_variables · required · ON DELETE CASCADE · not indexedp_ci_pipelines.external_pull_request_id → external_pull_requests.id — one external_pull_requests, many p_ci_pipelines · optional · ON DELETE SET NULL · indexedp_ci_pipelines.auto_canceled_by_partition_id, auto_canceled_by_id → p_ci_pipelines.partition_id, id — one p_ci_pipelines, many p_ci_pipelines · optional · ON DELETE SET NULL · not indexedp_ci_pipelines.pipeline_schedule_id → ci_pipeline_schedules.id — one ci_pipeline_schedules, many p_ci_pipelines · optional · ON DELETE SET NULL · indexedp_ci_pipelines.ci_ref_id → ci_refs.id — one ci_refs, many p_ci_pipelines · optional · ON DELETE SET NULL · indexedp_ci_runner_machine_builds.partition_id, build_id → p_ci_builds.partition_id, id — one p_ci_builds, at most one p_ci_runner_machine_builds · required · ON DELETE CASCADE · indexedp_ci_stages.partition_id, pipeline_id → p_ci_pipelines.partition_id, id — one p_ci_pipelines, many p_ci_stages · optional · ON DELETE CASCADE · not indexedp_ci_workload_variable_inclusions.partition_id, workload_id → p_ci_workloads.partition_id, id — one p_ci_workloads, many p_ci_workload_variable_inclusions · optional · ON DELETE CASCADE · indexedp_ci_workloads.partition_id, pipeline_id → p_ci_pipelines.partition_id, id — one p_ci_pipelines, at most one p_ci_workloads · required · ON DELETE CASCADE · indexedp_ai_active_context_code_enabled_namespaces.namespace_id → namespaces.id — one namespaces, many p_ai_active_context_code_enabled_namespaces · required · ON DELETE CASCADE · indexedp_duo_workflows_checkpoint_blobs.workflow_id → duo_workflows_workflows.id — one duo_workflows_workflows, many p_duo_workflows_checkpoint_blobs · required · ON DELETE CASCADE · indexedp_duo_workflows_checkpoint_headers.workflow_id → duo_workflows_workflows.id — one duo_workflows_workflows, many p_duo_workflows_checkpoint_headers · required · ON DELETE CASCADE · indexedp_duo_workflows_checkpoints.workflow_id → duo_workflows_workflows.id — one duo_workflows_workflows, many p_duo_workflows_checkpoints · required · ON DELETE CASCADE · indexedp_generated_ref_commits.project_id, merge_request_iid → merge_requests.target_project_id, iid — one merge_requests, many p_generated_ref_commits · required · ON DELETE CASCADE · indexedp_generated_ref_commits.project_id → projects.id — one projects, many p_generated_ref_commits · required · ON DELETE CASCADE · indexedp_sent_notifications.issue_email_participant_id → issue_email_participants.id — one issue_email_participants, many p_sent_notifications · optional · ON DELETE CASCADE · indexedp_ci_builds_partition_overrides.partition_id, build_id → p_ci_builds.partition_id, id — one p_ci_builds, many p_ci_builds_partition_overrides · required · ON DELETE CASCADE · not indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
build_id PK | bigint | NOT NULL | p_ci_builds.partition_id 1:1 · CASCADE | ||
partition_id PK | bigint | NOT NULL | p_ci_builds.partition_id 1:1 · CASCADE | ||
project_id | bigint | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
search_vector | tsvector | null |
Table checks: char_length(name) <= 255
partition_id, build_id — PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_builds must scan p_ci_build_needs to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to p_ci_build_needs maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_builds rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
name | text | NOT NULL | |||
artifacts | boolean | NOT NULL | TRUE | ||
optional | boolean | NOT NULL | FALSE | ||
build_id | bigint | NOT NULL | p_ci_builds.partition_id 1:N · CASCADE | ||
partition_id PK | bigint | NOT NULL | p_ci_builds.partition_id 1:N · CASCADE | ||
id PK | bigint | NOT NULL | |||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
build_id PK | bigint | NOT NULL | p_ci_builds.partition_id 1:1 · CASCADE | ||
partition_id PK | bigint | NOT NULL | p_ci_builds.partition_id 1:1 · CASCADE | ||
project_id | bigint | NOT NULL | |||
source | smallint | null | |||
pipeline_source | smallint | null |
partition_id, trace_artifact_id — PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_job_artifacts must scan p_ci_build_trace_metadata to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to p_ci_build_trace_metadata maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_job_artifacts rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.partition_id, trace_artifact_id — p_ci_build_trace_metadata.partition_id.trace_artifact_id may be NULL, so the relationship to p_ci_job_artifacts is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
build_id PK | bigint | NOT NULL | p_ci_builds.partition_id 1:1 · CASCADE | ||
partition_id PK | bigint | NOT NULL | p_ci_builds.partition_id 1:1 · CASCADE | ||
trace_artifact_id | bigint | null | p_ci_job_artifacts.partition_id 1:N · CASCADE | ||
last_archival_attempt_at | timestamptz | null | |||
archived_at | timestamptz | null | |||
archival_attempts | smallint | NOT NULL | 0 | ||
checksum | bytea | null | |||
remote_checksum | bytea | null | |||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
partition_id, stage_id — PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_stages must scan p_ci_builds to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to p_ci_builds maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_stages rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.upstream_pipeline_partition_id, upstream_pipeline_id — PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_pipelines must scan p_ci_builds to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to p_ci_builds maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_pipelines rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.auto_canceled_by_partition_id, auto_canceled_by_id — PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_pipelines must scan p_ci_builds to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to p_ci_builds maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_pipelines rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.partition_id, commit_id — PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_pipelines must scan p_ci_builds to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to p_ci_builds maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_pipelines rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.partition_id, stage_id — p_ci_builds.partition_id.stage_id may be NULL, so the relationship to p_ci_stages is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.resource_group_id — p_ci_builds.resource_group_id may be NULL, so the relationship to ci_resource_groups is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.upstream_pipeline_partition_id, upstream_pipeline_id — p_ci_builds.upstream_pipeline_partition_id.upstream_pipeline_id may be NULL, so the relationship to p_ci_pipelines is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.auto_canceled_by_partition_id, auto_canceled_by_id — p_ci_builds.auto_canceled_by_partition_id.auto_canceled_by_id may be NULL, so the relationship to p_ci_pipelines is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.partition_id, commit_id — p_ci_builds.partition_id.commit_id may be NULL, so the relationship to p_ci_pipelines is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.status — status is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.type — type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.finished_at — finished_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.started_at — started_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.erased_at — erased_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.artifacts_expire_at — artifacts_expire_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.queued_at — queued_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
status | varchar | null | |||
finished_at | timestamp | null | |||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
started_at | timestamp | null | |||
coverage | double precision | null | |||
name | varchar | null | |||
allow_failure | boolean | NOT NULL | FALSE | ||
stage_idx | integer | null | |||
tag | boolean | null | |||
ref | varchar | null | |||
type | varchar | null | |||
target_url | varchar | null | |||
description | varchar | null | |||
erased_at | timestamp | null | |||
artifacts_expire_at | timestamp | null | |||
environment | varchar | null | |||
when | varchar | null | |||
queued_at | timestamp | null | |||
lock_version | integer | null | 0 | CHECK lock_version IS NOT NULL | |
coverage_regex | varchar | null | |||
retried | boolean | null | |||
protected | boolean | null | |||
failure_reason | integer | null | |||
scheduled_at | timestamptz | null | |||
token_encrypted | varchar | null | |||
resource_group_id | bigint | null | ci_resource_groups.id 1:N · SET NULL | ||
waiting_for_resource_at | timestamptz | null | |||
processed | boolean | null | |||
scheduling_type | smallint | null | |||
id PK | bigint | NOT NULL | |||
stage_id | bigint | null | p_ci_stages.partition_id 1:N · CASCADE | ||
partition_id PK | bigint | NOT NULL | p_ci_pipelines.partition_id 1:N · CASCADE | ||
auto_canceled_by_partition_id | bigint | null | p_ci_pipelines.partition_id 1:N · SET NULL | ||
auto_canceled_by_id | bigint | null | p_ci_pipelines.partition_id 1:N · SET NULL | ||
commit_id | bigint | null | p_ci_pipelines.partition_id 1:N · CASCADE | ||
erased_by_id | bigint | null | |||
project_id | bigint | null | CHECK project_id IS NOT NULL | ||
runner_id | bigint | null | |||
upstream_pipeline_id | bigint | null | p_ci_pipelines.partition_id 1:N · CASCADE | ||
user_id | bigint | null | |||
upstream_pipeline_partition_id | bigint | null | p_ci_pipelines.partition_id 1:N · CASCADE | ||
scoped_user_id | bigint | null | |||
timeout | integer | null | |||
timeout_source | smallint | null | |||
exit_code | smallint | null | |||
debug_trace_enabled | boolean | null |
Table checks: lock_version IS NOT NULL; project_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
partition_id PK | bigint | NOT NULL | p_ci_builds.partition_id 1:N · CASCADE | ||
job_id | bigint | NOT NULL | p_ci_builds.partition_id 1:N · CASCADE | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
data | jsonb | NOT NULL | CAST('[]' AS jsonb) | CHECK jsonb_typeof(data) = CAST('array' AS text) | |
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL; char_length(name) <= 255; jsonb_typeof(data) = CAST('array' AS text)
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
job_artifact_id PK | bigint | NOT NULL | p_ci_job_artifacts.partition_id 1:1 · CASCADE | ||
partition_id PK | bigint | NOT NULL | p_ci_job_artifacts.partition_id 1:1 · CASCADE | ||
project_id | bigint | NOT NULL | |||
status | smallint | NOT NULL | |||
validation_error | text | null | CHECK char_length(validation_error) <= 255 |
Table checks: char_length(validation_error) <= 255
partition_id, job_id — PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_builds must scan p_ci_job_artifacts to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to p_ci_job_artifacts maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_builds rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id | bigint | NOT NULL | |||
file_type | integer | NOT NULL | |||
size | bigint | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
expire_at | timestamptz | null | |||
file | varchar | null | |||
file_store | integer | null | 1 | CHECK file_store IS NOT NULL | |
file_sha256 | bytea | null | |||
file_format | smallint | null | |||
file_location | smallint | null | |||
id PK | bigint | NOT NULL | |||
job_id | bigint | NOT NULL | p_ci_builds.partition_id 1:N · CASCADE | ||
locked | smallint | null | 2 | ||
partition_id PK | bigint | NOT NULL | p_ci_builds.partition_id 1:N · CASCADE | ||
accessibility | smallint | NOT NULL | 0 | ||
file_final_path | text | null | CHECK char_length(file_final_path) <= 1024 | ||
exposed_as | text | null | CHECK char_length(exposed_as) <= 100 | ||
exposed_paths | text[] | null |
Table checks: file_store IS NOT NULL; char_length(file_final_path) <= 1024; char_length(exposed_as) <= 100
partition_id, job_definition_id — PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_job_definitions must scan p_ci_job_definition_instances to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to p_ci_job_definition_instances maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_job_definitions rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.job_definition_id, job_id, partition_id — p_ci_job_definition_instances looks like a many-to-many link table but has no unique constraint across (job_definition_id, job_id, partition_id). The same pair can be inserted twice; every join through it will double-count.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
job_id PK | bigint | NOT NULL | p_ci_builds.partition_id 1:1 · CASCADE | ||
job_definition_id | bigint | NOT NULL | p_ci_job_definitions.partition_id 1:N · RESTRICT | ||
partition_id PK | bigint | NOT NULL | p_ci_builds.partition_id 1:1 · CASCADE | ||
project_id | bigint | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
partition_id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | null | |||
interruptible | boolean | NOT NULL | FALSE | ||
checksum | bytea | NOT NULL | |||
config | jsonb | NOT NULL | CAST('{}' AS jsonb) |
partition_id, job_id — PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_builds must scan p_ci_job_inputs to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to p_ci_job_inputs maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_builds rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
job_id | bigint | NOT NULL | p_ci_builds.partition_id 1:N · CASCADE | ||
partition_id PK | bigint | NOT NULL | p_ci_builds.partition_id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
value | jsonb | null |
Table checks: char_length(name) <= 255
partition_id, job_id — PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_builds must scan p_ci_job_messages to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to p_ci_job_messages maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_builds rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
job_id | bigint | NOT NULL | p_ci_builds.partition_id 1:N · CASCADE | ||
partition_id PK | bigint | NOT NULL | p_ci_builds.partition_id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | |||
severity | smallint | NOT NULL | 0 | ||
content | text | null | CHECK char_length(content) <= 10000 |
Table checks: char_length(content) <= 10000
p_ci_job_runtime_environments references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
build_id PK | bigint | NOT NULL | |||
partition_id PK | bigint | NOT NULL | |||
runtime_environment_id | bigint | null | |||
runner_machine_id | bigint | null | |||
project_id | bigint | NOT NULL | |||
suspend_on_success | boolean | NOT NULL | FALSE | ||
suspend_on_failure | boolean | NOT NULL | FALSE |
p_ci_pipeline_artifact_states references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
pipeline_artifact_id PK | bigint | NOT NULL | |||
partition_id PK | bigint | NOT NULL | |||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | null | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 | ||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL; char_length(verification_failure) <= 255
partition_id, pipeline_id — PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_pipelines must scan p_ci_pipeline_variables to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to p_ci_pipeline_variables maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_pipelines rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
key | varchar | NOT NULL | |||
encrypted_value | text | null | |||
encrypted_value_salt | varchar | null | |||
encrypted_value_iv | varchar | null | |||
variable_type | smallint | NOT NULL | 1 | ||
partition_id PK | bigint | NOT NULL | p_ci_pipelines.partition_id 1:N · CASCADE | ||
raw | boolean | NOT NULL | FALSE | ||
id PK | bigint | NOT NULL | |||
pipeline_id | bigint | NOT NULL | p_ci_pipelines.partition_id 1:N · CASCADE | ||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
auto_canceled_by_partition_id, auto_canceled_by_id — PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_pipelines must scan p_ci_pipelines to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to p_ci_pipelines maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_pipelines rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.external_pull_request_id — p_ci_pipelines.external_pull_request_id may be NULL, so the relationship to external_pull_requests is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.auto_canceled_by_partition_id, auto_canceled_by_id — p_ci_pipelines.auto_canceled_by_partition_id.auto_canceled_by_id may be NULL, so the relationship to p_ci_pipelines is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.pipeline_schedule_id — p_ci_pipelines.pipeline_schedule_id may be NULL, so the relationship to ci_pipeline_schedules is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.ci_ref_id — p_ci_pipelines.ci_ref_id may be NULL, so the relationship to ci_refs is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.status — status is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.external_pull_request_id, pipeline_schedule_id, ci_ref_id — p_ci_pipelines has several nullable FKs (external_pull_request_id, pipeline_schedule_id, ci_ref_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.committed_at — committed_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.started_at — started_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.finished_at — finished_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
ref | varchar | null | |||
sha | varchar | null | |||
before_sha | varchar | null | |||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
tag | boolean | null | FALSE | ||
yaml_errors | text | null | |||
committed_at | timestamp | null | |||
project_id | bigint | null | CHECK project_id IS NOT NULL | ||
status | varchar | null | |||
started_at | timestamp | null | |||
finished_at | timestamp | null | |||
duration | integer | null | |||
user_id | bigint | null | |||
lock_version | integer | null | 0 | CHECK lock_version IS NOT NULL | |
pipeline_schedule_id | bigint | null | ci_pipeline_schedules.id 1:N · SET NULL | ||
source | integer | null | |||
config_source | integer | null | |||
protected | boolean | null | |||
failure_reason | integer | null | |||
iid | integer | null | |||
merge_request_id | bigint | null | |||
source_sha | bytea | null | |||
target_sha | bytea | null | |||
external_pull_request_id | bigint | null | external_pull_requests.id 1:N · SET NULL | ||
ci_ref_id | bigint | null | ci_refs.id 1:N · SET NULL | ||
locked | smallint | NOT NULL | 1 | ||
partition_id PK | bigint | NOT NULL | |||
id PK | bigint | NOT NULL | |||
auto_canceled_by_id | bigint | null | p_ci_pipelines.partition_id 1:N · SET NULL | ||
auto_canceled_by_partition_id | bigint | null | p_ci_pipelines.partition_id 1:N · SET NULL | ||
trigger_id | bigint | null |
Table checks: project_id IS NOT NULL; lock_version IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
partition_id PK | bigint | NOT NULL | p_ci_builds.partition_id 1:1 · CASCADE | ||
build_id PK | bigint | NOT NULL | p_ci_builds.partition_id 1:1 · CASCADE | ||
runner_machine_id | bigint | NOT NULL | |||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
partition_id, pipeline_id — PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_pipelines must scan p_ci_stages to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to p_ci_stages maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_pipelines rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.partition_id, pipeline_id — p_ci_stages.partition_id.pipeline_id may be NULL, so the relationship to p_ci_pipelines is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id | bigint | null | CHECK project_id IS NOT NULL | ||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
name | varchar | null | |||
status | integer | null | |||
lock_version | integer | null | 0 | CHECK lock_version IS NOT NULL | |
position | integer | null | |||
id PK | bigint | NOT NULL | |||
partition_id PK | bigint | NOT NULL | p_ci_pipelines.partition_id 1:N · CASCADE | ||
pipeline_id | bigint | null | p_ci_pipelines.partition_id 1:N · CASCADE |
Table checks: project_id IS NOT NULL; lock_version IS NOT NULL
partition_id, workload_id — p_ci_workload_variable_inclusions.partition_id.workload_id may be NULL, so the relationship to p_ci_workloads is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
workload_id | bigint | null | p_ci_workloads.partition_id 1:N · CASCADE | ||
partition_id PK | bigint | NOT NULL | p_ci_workloads.partition_id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
variable_name | text | NOT NULL | CHECK char_length(variable_name) <= 255 |
Table checks: char_length(variable_name) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
partition_id PK | bigint | NOT NULL | p_ci_pipelines.partition_id 1:1 · CASCADE | ||
project_id | bigint | NOT NULL | |||
pipeline_id | bigint | NOT NULL | p_ci_pipelines.partition_id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
branch_name | text | null | CHECK char_length(branch_name) <= 255 | ||
status | smallint | NOT NULL | 0 |
Table checks: char_length(branch_name) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id PK | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
connection_id | bigint | NOT NULL | |||
metadata | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
state | smallint | NOT NULL | 0 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
p_ai_active_context_code_repositories references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id PK | bigint | NOT NULL | |||
connection_id | bigint | null | |||
enabled_namespace_id | bigint | null | |||
metadata | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
last_commit | text | null | CHECK char_length(last_commit) <= 64 | ||
state | smallint | NOT NULL | 0 | ||
indexed_at | timestamptz | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
last_queried_at | timestamptz | null |
Table checks: char_length(last_commit) <= 64
p_batched_git_ref_updates_deletions references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
partition_id PK | bigint | NOT NULL | 1 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
status | smallint | NOT NULL | 1 | ||
ref | text | NOT NULL | CHECK char_length(ref) <= 1024 |
Table checks: char_length(ref) <= 1024
p_catalog_resource_sync_events references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
catalog_resource_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
partition_id PK | bigint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 1 | ||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() |
build_finished_at — build_finished_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.p_ci_finished_build_ch_sync_events references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
build_id PK | bigint | NOT NULL | |||
partition PK | bigint | NOT NULL | 1 | ||
build_finished_at | timestamp | NOT NULL | |||
processed | boolean | NOT NULL | FALSE | ||
project_id | bigint | NOT NULL |
pipeline_finished_at — pipeline_finished_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.p_ci_finished_pipeline_ch_sync_events references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
pipeline_id PK | bigint | NOT NULL | |||
project_namespace_id | bigint | NOT NULL | |||
partition PK | bigint | NOT NULL | 1 | ||
pipeline_finished_at | timestamp | NOT NULL | |||
processed | boolean | NOT NULL | FALSE |
p_ci_runtime_environments references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
partition PK | bigint | NOT NULL | 1 | ||
project_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
environment_key | text | NOT NULL | CHECK char_length(environment_key) <= 512 |
Table checks: char_length(environment_key) <= 512
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
workflow_id | bigint | NOT NULL | duo_workflows_workflows.id 1:N · CASCADE | ||
project_id | bigint | null | |||
namespace_id | bigint | null | |||
workflow_created_at PK | timestamptz | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
current_thread | integer | NOT NULL | 0 | ||
thread_ts | text | NOT NULL | CHECK char_length(thread_ts) <= 255 | ||
channel | text | NOT NULL | CHECK char_length(channel) <= 255 | ||
version | text | NOT NULL | CHECK char_length(version) <= 255 | ||
write_type | text | NOT NULL | CHECK char_length(write_type) <= 255 | ||
step_action | text | NOT NULL | CHECK char_length(step_action) <= 255 | ||
data | bytea | NOT NULL | CHECK octet_length(data) <= 1048576 |
Table checks: char_length(step_action) <= 255; char_length(thread_ts) <= 255; char_length(version) <= 255; char_length(write_type) <= 255; octet_length(data) <= 1048576; num_nonnulls(namespace_id, project_id) = 1; char_length(channel) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
workflow_id | bigint | NOT NULL | duo_workflows_workflows.id 1:N · CASCADE | ||
project_id | bigint | null | |||
namespace_id | bigint | null | |||
workflow_created_at PK | timestamptz | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
current_thread | integer | NOT NULL | 0 | ||
checkpoint | jsonb | NOT NULL | |||
metadata | jsonb | NOT NULL | |||
thread_ts | text | NOT NULL | CHECK char_length(thread_ts) <= 255 | ||
parent_ts | text | null | CHECK char_length(parent_ts) <= 255 | ||
checkpoint_ns | text | null | CHECK char_length(checkpoint_ns) <= 4096 |
Table checks: char_length(thread_ts) <= 255; char_length(parent_ts) <= 255; char_length(checkpoint_ns) <= 4096; num_nonnulls(namespace_id, project_id) = 1
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
workflow_id | bigint | NOT NULL | duo_workflows_workflows.id 1:N · CASCADE | ||
project_id | bigint | null | |||
created_at PK | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
namespace_id | bigint | null | |||
thread_ts | text | NOT NULL | CHECK char_length(thread_ts) <= 255 | ||
parent_ts | text | null | CHECK char_length(parent_ts) <= 255 | ||
checkpoint | jsonb | NOT NULL | |||
metadata | jsonb | NOT NULL | |||
ui_chat_log | jsonb | null | |||
current_thread | integer | NOT NULL | 0 | ||
checkpoint_ns | text | null | CHECK char_length(checkpoint_ns) <= 4096 |
Table checks: num_nonnulls(namespace_id, project_id) = 1; char_length(thread_ts) <= 255; char_length(checkpoint_ns) <= 4096; char_length(parent_ts) <= 255
merge_request_iid, project_id — p_generated_ref_commits looks like a many-to-many link table but has no unique constraint across (merge_request_iid, project_id). The same pair can be inserted twice; every join through it will double-count.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
merge_request_iid | bigint | NOT NULL | merge_requests.target_project_id 1:N · CASCADE | ||
project_id PK | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
commit_sha | bytea | NOT NULL |
p_knowledge_graph_code_indexing_tasks references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
created_at PK | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
ref | text | NOT NULL | CHECK char_length(ref) <= 255 | ||
commit_sha | text | NOT NULL | CHECK char_length(commit_sha) <= 40 | ||
traversal_path | text | NOT NULL | CHECK char_length(traversal_path) <= 1024 |
Table checks: char_length(ref) <= 255; char_length(traversal_path) <= 1024; char_length(commit_sha) <= 40
issue_email_participant_id — p_sent_notifications.issue_email_participant_id may be NULL, so the relationship to issue_email_participants is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.noteable_type, noteable_id — noteable_id points at different tables depending on noteable_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | nextval(CAST('sent_notifications_id_seq' AS regclass)) | ||
project_id | bigint | null | |||
noteable_id | bigint | null | |||
recipient_id | bigint | null | |||
issue_email_participant_id | bigint | null | issue_email_participants.id 1:N · CASCADE | ||
namespace_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
partition PK | integer | NOT NULL | 1 | ||
noteable_type | text | null | CHECK char_length(noteable_type) <= 255 | ||
commit_id | text | null | CHECK char_length(commit_id) <= 255 | ||
reply_key | text | NOT NULL | CHECK char_length(reply_key) <= 255 | ||
in_reply_to_discussion_id | text | null | CHECK char_length(in_reply_to_discussion_id) <= 255 |
Table checks: char_length(commit_id) <= 255; char_length(reply_key) <= 255; char_length(noteable_type) <= 255; char_length(in_reply_to_discussion_id) <= 255
partition_id, build_id — PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_builds must scan p_ci_builds_partition_overrides to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to p_ci_builds_partition_overrides maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_builds rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
build_id PK | bigint | NOT NULL | p_ci_builds.partition_id 1:N · CASCADE | ||
partition_id | bigint | NOT NULL | p_ci_builds.partition_id 1:N · CASCADE | ||
project_id | bigint | NOT NULL |
p_ci_builds_partition_overrides_0 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
build_id PK | bigint | NOT NULL | |||
partition_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL |
p_ci_builds_partition_overrides_1 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
build_id PK | bigint | NOT NULL | |||
partition_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL |
p_ci_builds_partition_overrides_2 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
build_id PK | bigint | NOT NULL | |||
partition_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL |
p_ci_pipeline_iids references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_00 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_01 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_02 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_03 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_04 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_05 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_06 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_07 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_08 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_09 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_10 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_11 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_12 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_13 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_14 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_15 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_16 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_17 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_18 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_19 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_20 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_21 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_22 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_23 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_24 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_25 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_26 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_27 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_28 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_29 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_30 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_31 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_32 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_33 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_34 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_35 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_36 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_37 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_38 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_39 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_40 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_41 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_42 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_43 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_44 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_45 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_46 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_47 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_48 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_49 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_50 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_51 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_52 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_53 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_54 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_55 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_56 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_57 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_58 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_59 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_60 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_61 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_62 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
p_ci_pipeline_iids_63 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
iid PK | integer | NOT NULL |
Tables whose names start with group_.
group_crm_settings.source_group_id → namespaces.id — one namespaces, many group_crm_settings · optional · ON DELETE CASCADE · indexedgroup_crm_settings.group_id → namespaces.id — one namespaces, at most one group_crm_settings · required · ON DELETE CASCADE · indexedgroup_custom_attributes.group_id → namespaces.id — one namespaces, many group_custom_attributes · required · ON DELETE CASCADE · indexedgroup_deletion_schedules.user_id → users.id — one users, many group_deletion_schedules · required · ON DELETE CASCADE · indexedgroup_deletion_schedules.group_id → namespaces.id — one namespaces, at most one group_deletion_schedules · required · ON DELETE CASCADE · indexedgroup_deploy_tokens.group_id → namespaces.id — one namespaces, many group_deploy_tokens · required · ON DELETE CASCADE · indexedgroup_deploy_tokens.deploy_token_id → deploy_tokens.id — one deploy_tokens, many group_deploy_tokens · required · ON DELETE CASCADE · indexedgroup_features.group_id → namespaces.id — one namespaces, at most one group_features · required · ON DELETE CASCADE · indexedgroup_group_links.member_role_id → member_roles.id — one member_roles, many group_group_links · optional · ON DELETE SET NULL · indexedgroup_group_links.shared_with_group_id → namespaces.id — one namespaces, many group_group_links · required · ON DELETE CASCADE · indexedgroup_group_links.shared_group_id → namespaces.id — one namespaces, many group_group_links · required · ON DELETE CASCADE · indexedgroup_import_states.user_id → users.id — one users, many group_import_states · optional · ON DELETE CASCADE · indexedgroup_import_states.group_id → namespaces.id — one namespaces, at most one group_import_states · required · ON DELETE CASCADE · indexedgroup_merge_request_approval_settings.group_id → namespaces.id — one namespaces, at most one group_merge_request_approval_settings · required · ON DELETE CASCADE · indexedgroup_push_rules.group_id → namespaces.id — one namespaces, at most one group_push_rules · required · ON DELETE CASCADE · indexedgroup_repository_storage_moves.group_id → namespaces.id — one namespaces, many group_repository_storage_moves · required · ON DELETE CASCADE · indexedgroup_saved_replies.group_id → namespaces.id — one namespaces, many group_saved_replies · required · ON DELETE CASCADE · indexedgroup_scim_auth_access_tokens.group_id → namespaces.id — one namespaces, many group_scim_auth_access_tokens · required · ON DELETE CASCADE · indexedgroup_scim_identities.group_id → namespaces.id — one namespaces, many group_scim_identities · required · ON DELETE CASCADE · indexedgroup_scim_identities.user_id → users.id — one users, many group_scim_identities · required · ON DELETE CASCADE · indexedgroup_secret_rotation_infos.group_id → namespaces.id — one namespaces, many group_secret_rotation_infos · required · ON DELETE CASCADE · indexedgroup_secrets_managers.group_id → namespaces.id — one namespaces, many group_secrets_managers · optional · ON DELETE CASCADE · indexedgroup_ssh_certificates.namespace_id → namespaces.id — one namespaces, many group_ssh_certificates · required · ON DELETE CASCADE · indexedgroup_type_ci_runner_machines.runner_id, runner_type → group_type_ci_runners.id, runner_type — one group_type_ci_runners, many group_type_ci_runner_machines · required · ON DELETE CASCADE · indexedgroup_upload_states.namespace_id → namespaces.id — one namespaces, many group_upload_states · required · ON DELETE CASCADE · indexedgroup_upload_states.group_upload_id → namespace_uploads.id — one namespace_uploads, at most one group_upload_states · required · ON DELETE CASCADE · indexedgroup_wiki_repositories.group_id → namespaces.id — one namespaces, at most one group_wiki_repositories · required · ON DELETE CASCADE · indexedgroup_wiki_repository_states.group_id → namespaces.id — one namespaces, many group_wiki_repository_states · optional · ON DELETE CASCADE · indexedgroup_wiki_repository_states.group_wiki_repository_id → group_wiki_repositories.group_id — one group_wiki_repositories, at most one group_wiki_repository_states · required · ON DELETE CASCADE · indexedtarget_type, target_id — target_id points at different tables depending on target_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.group_audit_events references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | nextval(CAST('audit_events_id_seq' AS regclass)) | ||
created_at PK | timestamptz | NOT NULL | |||
group_id | bigint | NOT NULL | |||
author_id | bigint | NOT NULL | |||
target_id | bigint | null | |||
event_name | text | null | CHECK char_length(event_name) <= 255 | ||
details | text | null | |||
ip_address | inet | null | |||
author_name | text | null | CHECK char_length(author_name) <= 255 | ||
entity_path | text | null | CHECK char_length(entity_path) <= 5500 | ||
target_details | text | null | CHECK char_length(target_details) <= 5500 | ||
target_type | text | null | CHECK char_length(target_type) <= 255 |
Table checks: char_length(author_name) <= 255; char_length(entity_path) <= 5500; char_length(event_name) <= 255; char_length(target_details) <= 5500; char_length(target_type) <= 255
source_group_id — group_crm_settings.source_group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id PK | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
enabled | boolean | NOT NULL | TRUE | ||
source_group_id | bigint | null | namespaces.id 1:N · CASCADE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
key | varchar | NOT NULL | |||
value | varchar | NOT NULL |
group_id, user_id — group_deletion_schedules looks like a many-to-many link table but has no unique constraint across (group_id, user_id). The same pair can be inserted twice; every join through it will double-count.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id PK | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
marked_for_deletion_on | date | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
deploy_token_id | bigint | NOT NULL | deploy_tokens.id 1:N · CASCADE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id PK | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
wiki_access_level | smallint | NOT NULL | 20 |
member_role_id — group_group_links.member_role_id may be NULL, so the relationship to member_roles is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
shared_group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
shared_with_group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
expires_at | date | null | |||
group_access | smallint | NOT NULL | 30 | ||
member_role_id | bigint | null | member_roles.id 1:N · SET NULL |
user_id — group_import_states.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id PK | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
jid | text | null | CHECK char_length(jid) <= 100 | ||
last_error | text | null | CHECK char_length(last_error) <= 255 | ||
user_id | bigint | null | users.id 1:N · CASCADE | CHECK user_id IS NOT NULL |
Table checks: char_length(last_error) <= 255; char_length(jid) <= 100; user_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
group_id PK | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
allow_author_approval | boolean | NOT NULL | FALSE | ||
allow_committer_approval | boolean | NOT NULL | FALSE | ||
allow_overrides_to_approver_list_per_merge_request | boolean | NOT NULL | FALSE | ||
retain_approvals_on_push | boolean | NOT NULL | FALSE | ||
require_password_to_approve | boolean | NOT NULL | FALSE | ||
require_saml_auth_to_approve | boolean | NOT NULL | FALSE | ||
require_reauthentication_to_approve | boolean | NOT NULL | FALSE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
max_file_size | integer | NOT NULL | 0 | ||
member_check | boolean | NOT NULL | FALSE | ||
prevent_secrets | boolean | NOT NULL | FALSE | ||
commit_committer_name_check | boolean | NOT NULL | FALSE | ||
deny_delete_tag | boolean | null | |||
reject_unsigned_commits | boolean | null | |||
commit_committer_check | boolean | null | |||
reject_non_dco_commits | boolean | null | |||
commit_message_regex | text | null | CHECK char_length(commit_message_regex) <= 511 | ||
branch_name_regex | text | null | CHECK char_length(branch_name_regex) <= 511 | ||
commit_message_negative_regex | text | null | CHECK char_length(commit_message_negative_regex) <= 2047 | ||
author_email_regex | text | null | CHECK char_length(author_email_regex) <= 511 | ||
file_name_regex | text | null | CHECK char_length(file_name_regex) <= 511 |
Table checks: char_length(commit_message_negative_regex) <= 2047; char_length(file_name_regex) <= 511; char_length(commit_message_regex) <= 511; char_length(author_email_regex) <= 511; char_length(branch_name_regex) <= 511
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
state | smallint | NOT NULL | 1 | ||
source_storage_name | text | NOT NULL | CHECK char_length(source_storage_name) <= 255 | ||
destination_storage_name | text | NOT NULL | CHECK char_length(destination_storage_name) <= 255 | ||
error_message | text | null | CHECK char_length(error_message) <= 256 |
Table checks: char_length(error_message) <= 256; char_length(destination_storage_name) <= 255; char_length(source_storage_name) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
content | text | NOT NULL | CHECK char_length(content) <= 10000 |
Table checks: char_length(name) <= 255; char_length(content) <= 10000
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
token_encrypted | bytea | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
active | boolean | null | FALSE | ||
extern_uid | text | NOT NULL | CHECK char_length(extern_uid) <= 255 |
Table checks: char_length(extern_uid) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
next_reminder_at | timestamptz | NOT NULL | |||
last_reminder_at | timestamptz | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
secret_metadata_version | integer | NOT NULL | |||
rotation_interval_days | integer | NOT NULL | |||
secret_name | text | NOT NULL | CHECK char_length(secret_name) <= 255 |
Table checks: char_length(secret_name) <= 255
group_secrets_manager_maintenance_tasks references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | null | |||
group_id | bigint | NOT NULL | |||
root_namespace_id | bigint | NOT NULL | |||
organization_id | bigint | NOT NULL | |||
last_processed_at | timestamptz | null | |||
action | smallint | NOT NULL | |||
retry_count | smallint | NOT NULL | 0 |
group_id — group_secrets_managers.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
group_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK group_id IS NOT NULL | |
status | smallint | NOT NULL | 0 | ||
group_path | text | null | CHECK char_length(group_path) <= 64 | ||
root_namespace_path | text | null | CHECK char_length(root_namespace_path) <= 64 | ||
organization_id | bigint | null | CHECK organization_id IS NOT NULL | ||
root_namespace_id | bigint | null | CHECK root_namespace_id IS NOT NULL |
Table checks: root_namespace_id IS NOT NULL; char_length(root_namespace_path) <= 64; group_id IS NOT NULL; char_length(group_path) <= 64; organization_id IS NOT NULL
group_security_exclusions references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
scanner | smallint | NOT NULL | |||
type | smallint | NOT NULL | |||
active | boolean | NOT NULL | TRUE | ||
description | text | null | CHECK char_length(description) <= 255 | ||
value | text | NOT NULL | CHECK char_length(value) <= 255 |
Table checks: char_length(description) <= 255; char_length(value) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
fingerprint | bytea | NOT NULL | |||
title | text | NOT NULL | CHECK char_length(title) <= 256 | ||
key | text | NOT NULL | CHECK char_length(key) <= 524288 |
Table checks: char_length(title) <= 256; char_length(key) <= 524288
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
runner_id | bigint | NOT NULL | group_type_ci_runners.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
contacted_at | timestamptz | null | |||
creation_state | smallint | NOT NULL | 0 | ||
executor_type | smallint | null | |||
runner_type PK | smallint | NOT NULL | group_type_ci_runners.id 1:N · CASCADE | ||
config | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
system_xid | text | NOT NULL | CHECK char_length(system_xid) <= 64 | ||
platform | text | null | CHECK char_length(platform) <= 255 | ||
architecture | text | null | CHECK char_length(architecture) <= 255 | ||
revision | text | null | CHECK char_length(revision) <= 255 | ||
ip_address | text | null | CHECK char_length(ip_address) <= 1024 | ||
version | text | null | CHECK char_length(version) <= 2048 | ||
runtime_features | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
organization_id | bigint | null | CHECK organization_id IS NOT NULL | ||
labels | jsonb | NOT NULL | CAST('{}' AS jsonb) |
Table checks: char_length(system_xid) <= 64; char_length(revision) <= 255; char_length(version) <= 2048; char_length(ip_address) <= 1024; char_length(platform) <= 255; char_length(architecture) <= 255; organization_id IS NOT NULL
public_projects_minutes_cost_factor — public_projects_minutes_cost_factor is binary floating point. 0.1 + 0.2 ≠ 0.3; sums drift; reconciliation against the ledger will be off by cents.private_projects_minutes_cost_factor — private_projects_minutes_cost_factor is binary floating point. 0.1 + 0.2 ≠ 0.3; sums drift; reconciliation against the ledger will be off by cents.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
creator_id | bigint | null | |||
created_at | timestamptz | null | |||
updated_at | timestamptz | null | |||
contacted_at | timestamptz | null | |||
token_expires_at | timestamptz | null | |||
public_projects_minutes_cost_factor | double precision | NOT NULL | 1.0 | ||
private_projects_minutes_cost_factor | double precision | NOT NULL | 1.0 | ||
access_level | integer | NOT NULL | 0 | ||
maximum_timeout | integer | null | |||
runner_type PK | smallint | NOT NULL | |||
registration_type | smallint | NOT NULL | 0 | ||
creation_state | smallint | NOT NULL | 0 | ||
active | boolean | NOT NULL | TRUE | ||
run_untagged | boolean | NOT NULL | TRUE | ||
locked | boolean | NOT NULL | FALSE | ||
name | text | null | CHECK char_length(name) <= 256 | ||
token_encrypted | text | null | CHECK char_length(token_encrypted) <= 512 | ||
description | text | null | CHECK char_length(description) <= 1024 | ||
maintainer_note | text | null | CHECK char_length(maintainer_note) <= 1024 | ||
allowed_plans | text[] | NOT NULL | CAST('{}' AS text[]) | ||
allowed_plan_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
organization_id | bigint | null | CHECK organization_id IS NOT NULL | ||
allowed_plan_name_uids | smallint[] | NOT NULL | CAST('{}' AS smallint[]) | ||
token_rotation_deadline | timestamptz | null | |||
uuid | uuid | null |
Table checks: char_length(token_encrypted) <= 512; char_length(name) <= 256; char_length(maintainer_note) <= 1024; char_length(description) <= 1024; organization_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
group_upload_id | bigint | NOT NULL | namespace_uploads.id 1:1 · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 |
Table checks: char_length(verification_failure) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
shard_id | bigint | null | |||
group_id PK | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
disk_path | text | NOT NULL | CHECK char_length(disk_path) <= 80 | ||
last_repository_updated_at | timestamptz | null |
Table checks: char_length(disk_path) <= 80
group_id — group_wiki_repository_states.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
group_wiki_repository_id | bigint | NOT NULL | group_wiki_repositories.group_id 1:1 · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK group_id IS NOT NULL |
Table checks: char_length(verification_failure) <= 255; group_id IS NOT NULL
Tables whose names start with incident_.
incident_management_pending_alert_escalations.project_id → projects.id — one projects, many incident_management_pending_alert_escalations · optional · ON DELETE CASCADE · indexedincident_management_pending_alert_escalations.alert_id → alert_management_alerts.id — one alert_management_alerts, many incident_management_pending_alert_escalations · required · ON DELETE CASCADE · indexedincident_management_pending_alert_escalations.rule_id → incident_management_escalation_rules.id — one incident_management_escalation_rules, many incident_management_pending_alert_escalations · required · ON DELETE CASCADE · indexedincident_management_pending_issue_escalations.rule_id → incident_management_escalation_rules.id — one incident_management_escalation_rules, many incident_management_pending_issue_escalations · required · ON DELETE CASCADE · indexedincident_management_pending_issue_escalations.issue_id → issues.id — one issues, many incident_management_pending_issue_escalations · required · ON DELETE CASCADE · indexedincident_management_pending_issue_escalations.namespace_id → namespaces.id — one namespaces, many incident_management_pending_issue_escalations · optional · ON DELETE CASCADE · indexedincident_management_escalation_policies.project_id → projects.id — one projects, many incident_management_escalation_policies · required · ON DELETE CASCADE · indexedincident_management_escalation_rules.user_id → users.id — one users, many incident_management_escalation_rules · optional · ON DELETE CASCADE · indexedincident_management_escalation_rules.project_id → projects.id — one projects, many incident_management_escalation_rules · optional · ON DELETE CASCADE · indexedincident_management_escalation_rules.policy_id → incident_management_escalation_policies.id — one incident_management_escalation_policies, many incident_management_escalation_rules · required · ON DELETE CASCADE · indexedincident_management_escalation_rules.oncall_schedule_id → incident_management_oncall_schedules.id — one incident_management_oncall_schedules, many incident_management_escalation_rules · optional · ON DELETE CASCADE · indexedincident_management_issuable_escalation_statuses.namespace_id → namespaces.id — one namespaces, many incident_management_issuable_escalation_statuses · optional · ON DELETE CASCADE · indexedincident_management_issuable_escalation_statuses.policy_id → incident_management_escalation_policies.id — one incident_management_escalation_policies, many incident_management_issuable_escalation_statuses · optional · ON DELETE SET NULL · indexedincident_management_issuable_escalation_statuses.issue_id → issues.id — one issues, at most one incident_management_issuable_escalation_statuses · required · ON DELETE CASCADE · indexedincident_management_oncall_participants.project_id → projects.id — one projects, many incident_management_oncall_participants · optional · ON DELETE CASCADE · indexedincident_management_oncall_participants.oncall_rotation_id → incident_management_oncall_rotations.id — one incident_management_oncall_rotations, many incident_management_oncall_participants · required · ON DELETE CASCADE · indexedincident_management_oncall_participants.user_id → users.id — one users, many incident_management_oncall_participants · required · ON DELETE CASCADE · indexedincident_management_oncall_rotations.project_id → projects.id — one projects, many incident_management_oncall_rotations · optional · ON DELETE CASCADE · indexedincident_management_oncall_rotations.oncall_schedule_id → incident_management_oncall_schedules.id — one incident_management_oncall_schedules, many incident_management_oncall_rotations · required · ON DELETE CASCADE · indexedincident_management_oncall_schedules.project_id → projects.id — one projects, many incident_management_oncall_schedules · required · ON DELETE CASCADE · indexedincident_management_oncall_shifts.project_id → projects.id — one projects, many incident_management_oncall_shifts · optional · ON DELETE CASCADE · indexedincident_management_oncall_shifts.rotation_id → incident_management_oncall_rotations.id — one incident_management_oncall_rotations, many incident_management_oncall_shifts · required · ON DELETE CASCADE · indexedincident_management_oncall_shifts.participant_id → incident_management_oncall_participants.id — one incident_management_oncall_participants, many incident_management_oncall_shifts · required · ON DELETE CASCADE · indexedincident_management_timeline_event_tag_links.project_id → projects.id — one projects, many incident_management_timeline_event_tag_links · optional · ON DELETE CASCADE · indexedincident_management_timeline_event_tag_links.timeline_event_id → incident_management_timeline_events.id — one incident_management_timeline_events, many incident_management_timeline_event_tag_links · required · ON DELETE CASCADE · indexedincident_management_timeline_event_tag_links.timeline_event_tag_id → incident_management_timeline_event_tags.id — one incident_management_timeline_event_tags, many incident_management_timeline_event_tag_links · required · ON DELETE CASCADE · indexedincident_management_timeline_event_tags.project_id → projects.id — one projects, many incident_management_timeline_event_tags · required · ON DELETE CASCADE · indexedincident_management_timeline_events.issue_id → issues.id — one issues, many incident_management_timeline_events · required · ON DELETE CASCADE · indexedincident_management_timeline_events.author_id → users.id — one users, many incident_management_timeline_events · optional · ON DELETE SET NULL · indexedincident_management_timeline_events.updated_by_user_id → users.id — one users, many incident_management_timeline_events · optional · ON DELETE SET NULL · indexedincident_management_timeline_events.project_id → projects.id — one projects, many incident_management_timeline_events · required · ON DELETE CASCADE · indexedincident_management_timeline_events.promoted_from_note_id → notes.id — one notes, many incident_management_timeline_events · optional · ON DELETE SET NULL · indexedproject_id — incident_management_pending_alert_escalations.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
rule_id | bigint | NOT NULL | incident_management_escalation_rules.id 1:N · CASCADE | ||
alert_id | bigint | NOT NULL | alert_management_alerts.id 1:N · CASCADE | ||
process_at PK | timestamptz | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
namespace_id — incident_management_pending_issue_escalations.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
rule_id | bigint | NOT NULL | incident_management_escalation_rules.id 1:N · CASCADE | ||
issue_id | bigint | NOT NULL | issues.id 1:N · CASCADE | ||
process_at PK | timestamptz | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
name | text | NOT NULL | CHECK char_length(name) <= 72 | ||
description | text | null | CHECK char_length(description) <= 160 |
Table checks: char_length(description) <= 160; char_length(name) <= 72
user_id — incident_management_escalation_rules.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — incident_management_escalation_rules.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.oncall_schedule_id — incident_management_escalation_rules.oncall_schedule_id may be NULL, so the relationship to incident_management_oncall_schedules is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id, project_id, oncall_schedule_id — incident_management_escalation_rules has several nullable FKs (user_id, project_id, oncall_schedule_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
policy_id | bigint | NOT NULL | incident_management_escalation_policies.id 1:N · CASCADE | ||
oncall_schedule_id | bigint | null | incident_management_oncall_schedules.id 1:N · CASCADE | ||
status | smallint | NOT NULL | |||
elapsed_time_seconds | integer | NOT NULL | |||
is_removed | boolean | NOT NULL | FALSE | ||
user_id | bigint | null | users.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL; num_nonnulls(oncall_schedule_id, user_id) = 1
namespace_id — incident_management_issuable_escalation_statuses.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.policy_id — incident_management_issuable_escalation_statuses.policy_id may be NULL, so the relationship to incident_management_escalation_policies is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id, policy_id — incident_management_issuable_escalation_statuses has several nullable FKs (namespace_id, policy_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
issue_id | bigint | NOT NULL | issues.id 1:1 · CASCADE | ||
policy_id | bigint | null | incident_management_escalation_policies.id 1:N · SET NULL | ||
escalations_started_at | timestamptz | null | |||
resolved_at | timestamptz | null | |||
status | smallint | NOT NULL | 0 | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL
project_id — incident_management_oncall_participants.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
oncall_rotation_id | bigint | NOT NULL | incident_management_oncall_rotations.id 1:N · CASCADE | ||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
color_palette | smallint | NOT NULL | |||
color_weight | smallint | NOT NULL | |||
is_removed | boolean | NOT NULL | FALSE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
project_id — incident_management_oncall_rotations.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
oncall_schedule_id | bigint | NOT NULL | incident_management_oncall_schedules.id 1:N · CASCADE | ||
length | integer | NOT NULL | |||
length_unit | smallint | NOT NULL | |||
starts_at | timestamptz | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 200 | ||
ends_at | timestamptz | null | |||
active_period_start | time | null | |||
active_period_end | time | null | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL; char_length(name) <= 200
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
iid | integer | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 200 | ||
description | text | null | CHECK char_length(description) <= 1000 | ||
timezone | text | null | CHECK char_length(timezone) <= 100 |
Table checks: char_length(description) <= 1000; char_length(timezone) <= 100; char_length(name) <= 200
project_id — incident_management_oncall_shifts.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
rotation_id | bigint | NOT NULL | incident_management_oncall_rotations.id 1:N · CASCADE | ||
participant_id | bigint | NOT NULL | incident_management_oncall_participants.id 1:N · CASCADE | ||
starts_at | timestamptz | NOT NULL | |||
ends_at | timestamptz | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
project_id — incident_management_timeline_event_tag_links.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
timeline_event_id | bigint | NOT NULL | incident_management_timeline_events.id 1:N · CASCADE | ||
timeline_event_tag_id | bigint | NOT NULL | incident_management_timeline_event_tags.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 |
Table checks: char_length(name) <= 255
author_id — incident_management_timeline_events.author_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.updated_by_user_id — incident_management_timeline_events.updated_by_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.promoted_from_note_id — incident_management_timeline_events.promoted_from_note_id may be NULL, so the relationship to notes is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.author_id, updated_by_user_id, promoted_from_note_id — incident_management_timeline_events has several nullable FKs (author_id, updated_by_user_id, promoted_from_note_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
occurred_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
author_id | bigint | null | users.id 1:N · SET NULL | ||
issue_id | bigint | NOT NULL | issues.id 1:N · CASCADE | ||
updated_by_user_id | bigint | null | users.id 1:N · SET NULL | ||
promoted_from_note_id | bigint | null | notes.id 1:N · SET NULL | ||
cached_markdown_version | integer | null | |||
editable | boolean | NOT NULL | FALSE | ||
note | text | NOT NULL | CHECK char_length(note) <= 10000 | ||
note_html | text | NOT NULL | CHECK char_length(note_html) <= 10000 | ||
action | text | NOT NULL | CHECK char_length(action) <= 128 |
Table checks: char_length(action) <= 128; char_length(note) <= 10000; char_length(note_html) <= 10000
Tables whose names start with instance_.
instance_audit_events_streaming_headers.instance_external_audit_event_destination_id → audit_events_instance_external_audit_event_destinations.id — one audit_events_instance_external_audit_event_destinations, many instance_audit_events_streaming_headers · required · ON DELETE CASCADE · indexedinstance_type_ci_runner_machines.runner_id, runner_type → instance_type_ci_runners.id, runner_type — one instance_type_ci_runners, many instance_type_ci_runner_machines · required · ON DELETE CASCADE · indexedtarget_type, target_id — target_id points at different tables depending on target_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.instance_audit_events references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | nextval(CAST('audit_events_id_seq' AS regclass)) | ||
created_at PK | timestamptz | NOT NULL | |||
author_id | bigint | NOT NULL | |||
target_id | bigint | null | |||
event_name | text | null | CHECK char_length(event_name) <= 255 | ||
details | text | null | |||
ip_address | inet | null | |||
author_name | text | null | CHECK char_length(author_name) <= 255 | ||
entity_path | text | null | CHECK char_length(entity_path) <= 5500 | ||
target_details | text | null | CHECK char_length(target_details) <= 5500 | ||
target_type | text | null | CHECK char_length(target_type) <= 255 |
Table checks: char_length(author_name) <= 255; char_length(entity_path) <= 5500; char_length(event_name) <= 255; char_length(target_details) <= 5500; char_length(target_type) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
instance_external_audit_event_destination_id | bigint | NOT NULL | audit_events_instance_external_audit_event_destinations.id 1:N · CASCADE | ||
key | text | NOT NULL | CHECK char_length(key) <= 255 | ||
value | text | NOT NULL | CHECK char_length(value) <= 2000 | ||
active | boolean | NOT NULL | TRUE |
Table checks: char_length(value) <= 2000; char_length(key) <= 255
instance_model_selection_feature_settings references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
feature | smallint | NOT NULL | |||
offered_model_ref | text | null | CHECK char_length(offered_model_ref) <= 255 | ||
offered_model_name | text | null | CHECK char_length(offered_model_name) <= 255 | ||
model_allowlist_enabled | boolean | NOT NULL | FALSE | ||
model_allowlist_gitlab_model_refs | text[] | NOT NULL | CAST('{}' AS text[]) | CHECK cardinality(model_allowlist_gitlab_model_refs) <= 100 |
Table checks: char_length(offered_model_ref) <= 255; char_length(offered_model_name) <= 255; cardinality(model_allowlist_gitlab_model_refs) <= 100
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
runner_id | bigint | NOT NULL | instance_type_ci_runners.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
contacted_at | timestamptz | null | |||
creation_state | smallint | NOT NULL | 0 | ||
executor_type | smallint | null | |||
runner_type PK | smallint | NOT NULL | instance_type_ci_runners.id 1:N · CASCADE | ||
config | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
system_xid | text | NOT NULL | CHECK char_length(system_xid) <= 64 | ||
platform | text | null | CHECK char_length(platform) <= 255 | ||
architecture | text | null | CHECK char_length(architecture) <= 255 | ||
revision | text | null | CHECK char_length(revision) <= 255 | ||
ip_address | text | null | CHECK char_length(ip_address) <= 1024 | ||
version | text | null | CHECK char_length(version) <= 2048 | ||
runtime_features | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
organization_id | bigint | null | CHECK organization_id IS NULL | ||
labels | jsonb | NOT NULL | CAST('{}' AS jsonb) |
Table checks: char_length(system_xid) <= 64; char_length(revision) <= 255; char_length(version) <= 2048; char_length(ip_address) <= 1024; char_length(platform) <= 255; char_length(architecture) <= 255; organization_id IS NULL
public_projects_minutes_cost_factor — public_projects_minutes_cost_factor is binary floating point. 0.1 + 0.2 ≠ 0.3; sums drift; reconciliation against the ledger will be off by cents.private_projects_minutes_cost_factor — private_projects_minutes_cost_factor is binary floating point. 0.1 + 0.2 ≠ 0.3; sums drift; reconciliation against the ledger will be off by cents.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
creator_id | bigint | null | |||
created_at | timestamptz | null | |||
updated_at | timestamptz | null | |||
contacted_at | timestamptz | null | |||
token_expires_at | timestamptz | null | |||
public_projects_minutes_cost_factor | double precision | NOT NULL | 1.0 | ||
private_projects_minutes_cost_factor | double precision | NOT NULL | 1.0 | ||
access_level | integer | NOT NULL | 0 | ||
maximum_timeout | integer | null | |||
runner_type PK | smallint | NOT NULL | |||
registration_type | smallint | NOT NULL | 0 | ||
creation_state | smallint | NOT NULL | 0 | ||
active | boolean | NOT NULL | TRUE | ||
run_untagged | boolean | NOT NULL | TRUE | ||
locked | boolean | NOT NULL | FALSE | ||
name | text | null | CHECK char_length(name) <= 256 | ||
token_encrypted | text | null | CHECK char_length(token_encrypted) <= 512 | ||
description | text | null | CHECK char_length(description) <= 1024 | ||
maintainer_note | text | null | CHECK char_length(maintainer_note) <= 1024 | ||
allowed_plans | text[] | NOT NULL | CAST('{}' AS text[]) | ||
allowed_plan_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
organization_id | bigint | null | CHECK organization_id IS NULL | ||
allowed_plan_name_uids | smallint[] | NOT NULL | CAST('{}' AS smallint[]) | ||
token_rotation_deadline | timestamptz | null | |||
uuid | uuid | null |
Table checks: char_length(token_encrypted) <= 512; char_length(name) <= 256; char_length(maintainer_note) <= 1024; char_length(description) <= 1024; organization_id IS NULL
Tables whose names start with loose_.
No foreign keys in this domain.
loose_foreign_keys_deleted_records references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
partition PK | bigint | NOT NULL | 1 | ||
primary_key_value | bigint | NOT NULL | |||
status | smallint | NOT NULL | 1 | ||
created_at | timestamptz | NOT NULL | now() | ||
fully_qualified_table_name | text | NOT NULL | CHECK char_length(fully_qualified_table_name) <= 150 | ||
consume_after | timestamptz | null | now() | ||
cleanup_attempts | smallint | null | 0 |
Table checks: char_length(fully_qualified_table_name) <= 150
loose_foreign_keys_namespace_deleted_records references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | uuid | NOT NULL | gen_random_uuid_v7() | ||
partition PK | bigint | NOT NULL | 1 | ||
primary_key_value | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | |||
consume_after | timestamptz | null | now() | ||
created_at | timestamptz | NOT NULL | now() | ||
status | smallint | NOT NULL | 1 | ||
cleanup_attempts | smallint | null | 0 | ||
fully_qualified_table_name | text | NOT NULL | CHECK char_length(fully_qualified_table_name) <= 150 |
Table checks: char_length(fully_qualified_table_name) <= 150
loose_foreign_keys_organization_deleted_records references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | uuid | NOT NULL | gen_random_uuid_v7() | ||
partition PK | bigint | NOT NULL | 1 | ||
primary_key_value | bigint | NOT NULL | |||
organization_id | bigint | NOT NULL | |||
consume_after | timestamptz | null | now() | ||
created_at | timestamptz | NOT NULL | now() | ||
status | smallint | NOT NULL | 1 | ||
cleanup_attempts | smallint | null | 0 | ||
fully_qualified_table_name | text | NOT NULL | CHECK char_length(fully_qualified_table_name) <= 150 |
Table checks: char_length(fully_qualified_table_name) <= 150
loose_foreign_keys_project_deleted_records references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | uuid | NOT NULL | gen_random_uuid_v7() | ||
partition PK | bigint | NOT NULL | 1 | ||
primary_key_value | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
consume_after | timestamptz | null | now() | ||
created_at | timestamptz | NOT NULL | now() | ||
status | smallint | NOT NULL | 1 | ||
cleanup_attempts | smallint | null | 0 | ||
fully_qualified_table_name | text | NOT NULL | CHECK char_length(fully_qualified_table_name) <= 150 |
Table checks: char_length(fully_qualified_table_name) <= 150
loose_foreign_keys_user_deleted_records references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | uuid | NOT NULL | gen_random_uuid_v7() | ||
partition PK | bigint | NOT NULL | 1 | ||
primary_key_value | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | |||
consume_after | timestamptz | null | now() | ||
created_at | timestamptz | NOT NULL | now() | ||
status | smallint | NOT NULL | 1 | ||
cleanup_attempts | smallint | null | 0 | ||
fully_qualified_table_name | text | NOT NULL | CHECK char_length(fully_qualified_table_name) <= 150 |
Table checks: char_length(fully_qualified_table_name) <= 150
Tables whose names start with merge_.
merge_requests_merge_data.merge_user_id → users.id — one users, many merge_requests_merge_data · optional · ON DELETE SET NULL · indexedmerge_requests_merge_data.project_id → projects.id — one projects, many merge_requests_merge_data · required · ON DELETE CASCADE · indexedmerge_requests_merge_data.merge_request_id → merge_requests.id — one merge_requests, at most one merge_requests_merge_data · required · ON DELETE CASCADE · indexedmerge_request_approval_metrics.target_project_id → projects.id — one projects, many merge_request_approval_metrics · required · ON DELETE CASCADE · indexedmerge_request_approval_metrics.merge_request_id → merge_requests.id — one merge_requests, at most one merge_request_approval_metrics · required · ON DELETE CASCADE · indexedmerge_request_assignees.project_id → projects.id — one projects, many merge_request_assignees · optional · ON DELETE CASCADE · indexedmerge_request_assignees.merge_request_id → merge_requests.id — one merge_requests, many merge_request_assignees · required · ON DELETE CASCADE · indexedmerge_request_assignees.user_id → users.id — one users, many merge_request_assignees · required · ON DELETE CASCADE · indexedmerge_request_assignment_events.merge_request_id → merge_requests.id — one merge_requests, many merge_request_assignment_events · required · ON DELETE CASCADE · indexedmerge_request_assignment_events.project_id → projects.id — one projects, many merge_request_assignment_events · optional · ON DELETE CASCADE · indexedmerge_request_assignment_events.user_id → users.id — one users, many merge_request_assignment_events · optional · ON DELETE SET NULL · indexedmerge_request_blocks.project_id → projects.id — one projects, many merge_request_blocks · optional · ON DELETE CASCADE · indexedmerge_request_blocks.blocked_merge_request_id → merge_requests.id — one merge_requests, many merge_request_blocks · required · ON DELETE CASCADE · indexedmerge_request_blocks.blocking_merge_request_id → merge_requests.id — one merge_requests, many merge_request_blocks · required · ON DELETE CASCADE · indexedmerge_request_cleanup_schedules.project_id → projects.id — one projects, many merge_request_cleanup_schedules · optional · ON DELETE CASCADE · indexedmerge_request_cleanup_schedules.merge_request_id → merge_requests.id — one merge_requests, at most one merge_request_cleanup_schedules · required · ON DELETE CASCADE · indexedmerge_request_context_commit_diff_files.project_id → projects.id — one projects, many merge_request_context_commit_diff_files · optional · ON DELETE CASCADE · indexedmerge_request_context_commit_diff_files.merge_request_context_commit_id → merge_request_context_commits.id — one merge_request_context_commits, many merge_request_context_commit_diff_files · required · ON DELETE CASCADE · indexedmerge_request_context_commits.project_id → projects.id — one projects, many merge_request_context_commits · optional · ON DELETE CASCADE · indexedmerge_request_context_commits.merge_request_id → merge_requests.id — one merge_requests, many merge_request_context_commits · optional · ON DELETE CASCADE · indexedmerge_request_diff_commit_users.organization_id → organizations.id — one organizations, many merge_request_diff_commit_users · required · ON DELETE CASCADE · indexedmerge_request_diff_details.project_id → projects.id — one projects, many merge_request_diff_details · optional · ON DELETE CASCADE · indexedmerge_request_diff_details.merge_request_diff_id → merge_request_diffs.id — one merge_request_diffs, at most one merge_request_diff_details · required · ON DELETE CASCADE · indexedmerge_request_diff_files.project_id → projects.id — one projects, many merge_request_diff_files · optional · ON DELETE CASCADE · indexedmerge_request_diff_files.merge_request_diff_id → merge_request_diffs.id — one merge_request_diffs, many merge_request_diff_files · required · ON DELETE CASCADE · indexedmerge_request_diffs.project_id → projects.id — one projects, many merge_request_diffs · optional · ON DELETE CASCADE · indexedmerge_request_diffs.merge_request_id → merge_requests.id — one merge_requests, many merge_request_diffs · required · ON DELETE CASCADE · indexedmerge_request_merge_schedules.project_id → projects.id — one projects, many merge_request_merge_schedules · required · ON DELETE CASCADE · indexedmerge_request_merge_schedules.merge_request_id → merge_requests.id — one merge_requests, at most one merge_request_merge_schedules · required · ON DELETE CASCADE · indexedmerge_request_metrics.target_project_id → projects.id — one projects, many merge_request_metrics · optional · ON DELETE CASCADE · indexedmerge_request_metrics.merged_by_id → users.id — one users, many merge_request_metrics · optional · ON DELETE SET NULL · indexedmerge_request_metrics.latest_closed_by_id → users.id — one users, many merge_request_metrics · optional · ON DELETE SET NULL · indexedmerge_request_metrics.merge_request_id → merge_requests.id — one merge_requests, at most one merge_request_metrics · required · ON DELETE CASCADE · indexedmerge_request_predictions.project_id → projects.id — one projects, many merge_request_predictions · optional · ON DELETE CASCADE · indexedmerge_request_predictions.merge_request_id → merge_requests.id — one merge_requests, at most one merge_request_predictions · required · ON DELETE CASCADE · indexedmerge_request_reviewers.project_id → projects.id — one projects, many merge_request_reviewers · optional · ON DELETE CASCADE · indexedmerge_request_reviewers.user_id → users.id — one users, many merge_request_reviewers · required · ON DELETE CASCADE · indexedmerge_request_reviewers.merge_request_id → merge_requests.id — one merge_requests, many merge_request_reviewers · required · ON DELETE CASCADE · indexedmerge_request_saved_views.user_id → users.id — one users, many merge_request_saved_views · required · ON DELETE CASCADE · indexedmerge_request_user_mentions.project_id → projects.id — one projects, many merge_request_user_mentions · optional · ON DELETE CASCADE · indexedmerge_request_user_mentions.merge_request_id → merge_requests.id — one merge_requests, many merge_request_user_mentions · required · ON DELETE CASCADE · indexedmerge_request_user_mentions.note_id → notes.id — one notes, many merge_request_user_mentions · optional · ON DELETE CASCADE · indexedmerge_requests.latest_merge_request_diff_id → merge_request_diffs.id — one merge_request_diffs, many merge_requests · optional · ON DELETE SET NULL · indexedmerge_requests.assignee_id → users.id — one users, many merge_requests · optional · ON DELETE SET NULL · indexedmerge_requests.updated_by_id → users.id — one users, many merge_requests · optional · ON DELETE SET NULL · indexedmerge_requests.milestone_id → milestones.id — one milestones, many merge_requests · optional · ON DELETE SET NULL · indexedmerge_requests.target_project_id → projects.id — one projects, many merge_requests · required · ON DELETE CASCADE · indexedmerge_requests.merge_user_id → users.id — one users, many merge_requests · optional · ON DELETE SET NULL · indexedmerge_requests.author_id → users.id — one users, many merge_requests · optional · ON DELETE SET NULL · indexedmerge_requests.source_project_id → projects.id — one projects, many merge_requests · optional · ON DELETE SET NULL · indexedmerge_requests_approval_rules.group_id → namespaces.id — one namespaces, many merge_requests_approval_rules · optional · ON DELETE CASCADE · indexedmerge_requests_approval_rules.project_id → projects.id — one projects, many merge_requests_approval_rules · optional · ON DELETE CASCADE · indexedmerge_requests_approval_rules.source_rule_id → merge_requests_approval_rules.id — one merge_requests_approval_rules, many merge_requests_approval_rules · optional · ON DELETE SET NULL · indexedmerge_requests_approval_rules_approver_groups.approval_rule_id → merge_requests_approval_rules.id — one merge_requests_approval_rules, many merge_requests_approval_rules_approver_groups · required · ON DELETE CASCADE · indexedmerge_requests_approval_rules_approver_groups.group_id → namespaces.id — one namespaces, many merge_requests_approval_rules_approver_groups · required · ON DELETE CASCADE · indexedmerge_requests_approval_rules_approver_users.user_id → users.id — one users, many merge_requests_approval_rules_approver_users · required · ON DELETE CASCADE · indexedmerge_requests_approval_rules_approver_users.project_id → projects.id — one projects, many merge_requests_approval_rules_approver_users · optional · ON DELETE CASCADE · indexedmerge_requests_approval_rules_approver_users.approval_rule_id → merge_requests_approval_rules.id — one merge_requests_approval_rules, many merge_requests_approval_rules_approver_users · required · ON DELETE CASCADE · indexedmerge_requests_approval_rules_approver_users.group_id → namespaces.id — one namespaces, many merge_requests_approval_rules_approver_users · optional · ON DELETE CASCADE · indexedmerge_requests_approval_rules_groups.approval_rule_id → merge_requests_approval_rules.id — one merge_requests_approval_rules, many merge_requests_approval_rules_groups · required · ON DELETE CASCADE · indexedmerge_requests_approval_rules_groups.group_id → namespaces.id — one namespaces, many merge_requests_approval_rules_groups · required · ON DELETE CASCADE · indexedmerge_requests_approval_rules_merge_requests.project_id → projects.id — one projects, many merge_requests_approval_rules_merge_requests · required · ON DELETE CASCADE · indexedmerge_requests_approval_rules_merge_requests.approval_rule_id → merge_requests_approval_rules.id — one merge_requests_approval_rules, many merge_requests_approval_rules_merge_requests · required · ON DELETE CASCADE · indexedmerge_requests_approval_rules_merge_requests.merge_request_id → merge_requests.id — one merge_requests, many merge_requests_approval_rules_merge_requests · required · ON DELETE CASCADE · indexedmerge_requests_approval_rules_projects.approval_rule_id → merge_requests_approval_rules.id — one merge_requests_approval_rules, many merge_requests_approval_rules_projects · required · ON DELETE CASCADE · indexedmerge_requests_approval_rules_projects.project_id → projects.id — one projects, many merge_requests_approval_rules_projects · required · ON DELETE CASCADE · indexedmerge_requests_closing_issues.project_id → projects.id — one projects, many merge_requests_closing_issues · optional · ON DELETE CASCADE · indexedmerge_requests_closing_issues.merge_request_id → merge_requests.id — one merge_requests, many merge_requests_closing_issues · required · ON DELETE CASCADE · indexedmerge_requests_closing_issues.issue_id → issues.id — one issues, many merge_requests_closing_issues · required · ON DELETE CASCADE · indexedmerge_requests_compliance_violations.merge_request_id → merge_requests.id — one merge_requests, many merge_requests_compliance_violations · required · ON DELETE CASCADE · indexedmerge_requests_compliance_violations.target_project_id → projects.id — one projects, many merge_requests_compliance_violations · optional · ON DELETE CASCADE · indexedmerge_requests_compliance_violations.violating_user_id → users.id — one users, many merge_requests_compliance_violations · required · ON DELETE CASCADE · indexedmerge_trains.user_id → users.id — one users, many merge_trains · required · ON DELETE CASCADE · indexedmerge_trains.merge_request_id → merge_requests.id — one merge_requests, at most one merge_trains · required · ON DELETE CASCADE · indexedmerge_trains.target_project_id → projects.id — one projects, many merge_trains · required · ON DELETE CASCADE · indexedauthored_date — authored_date stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.committed_date — committed_date stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.merge_request_commits_metadata references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
authored_date | timestamp | null | |||
committed_date | timestamp | null | |||
id PK | bigint | NOT NULL | |||
project_id PK | bigint | NOT NULL | |||
commit_author_id | bigint | NOT NULL | |||
committer_id | bigint | NOT NULL | |||
sha | bytea | NOT NULL | |||
message | text | null |
merge_request_diff_commits_b5377a7a34 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
merge_request_commits_metadata_id | bigint | NOT NULL | |||
merge_request_diff_id PK | bigint | NOT NULL | |||
project_id PK | bigint | NOT NULL | |||
relative_order PK | integer | NOT NULL |
a_mode — a_mode is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.b_mode — b_mode is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.merge_request_diff_files_99208b8fac references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
new_file | boolean | NOT NULL | |||
renamed_file | boolean | NOT NULL | |||
deleted_file | boolean | NOT NULL | |||
too_large | boolean | NOT NULL | |||
a_mode | varchar | NOT NULL | |||
b_mode | varchar | NOT NULL | |||
new_path | text | null | |||
old_path | text | NOT NULL | |||
diff | text | null | |||
binary | boolean | null | |||
external_diff_offset | integer | null | |||
external_diff_size | integer | null | |||
generated | boolean | null | |||
encoded_file_path | boolean | NOT NULL | FALSE | ||
project_id | bigint | null | CHECK project_id IS NOT NULL | ||
merge_request_diff_id PK | bigint | NOT NULL | |||
relative_order PK | integer | NOT NULL |
Table checks: project_id IS NOT NULL
merge_user_id — merge_requests_merge_data.merge_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
merge_request_id PK | bigint | NOT NULL | merge_requests.id 1:1 · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
merge_user_id | bigint | null | users.id 1:N · SET NULL | ||
merge_params | text | null | |||
merge_error | text | null | |||
merge_jid | text | null | CHECK char_length(merge_jid) <= 255 | ||
merge_commit_sha | bytea | null | |||
merged_commit_sha | bytea | null | |||
merge_ref_sha | bytea | null | |||
squash_commit_sha | bytea | null | |||
in_progress_merge_commit_sha | bytea | null | |||
merge_status | smallint | NOT NULL | 0 | ||
auto_merge_enabled | boolean | NOT NULL | FALSE | ||
squash | boolean | NOT NULL | FALSE |
Table checks: char_length(merge_jid) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
merge_request_id PK | bigint | NOT NULL | merge_requests.id 1:1 · CASCADE | ||
last_approved_at | timestamptz | NOT NULL | |||
target_project_id | bigint | NOT NULL | projects.id 1:N · CASCADE |
project_id — merge_request_assignees.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
merge_request_id | bigint | NOT NULL | merge_requests.id 1:N · CASCADE | ||
created_at | timestamptz | null | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
project_id — merge_request_assignment_events.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id — merge_request_assignment_events.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id, user_id — merge_request_assignment_events has several nullable FKs (project_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | null | users.id 1:N · SET NULL | ||
merge_request_id | bigint | NOT NULL | merge_requests.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | now() | ||
action | smallint | NOT NULL | 1 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
project_id — merge_request_blocks.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
blocking_merge_request_id | bigint | NOT NULL | merge_requests.id 1:N · CASCADE | ||
blocked_merge_request_id | bigint | NOT NULL | merge_requests.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
project_id — merge_request_cleanup_schedules.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
merge_request_id PK | bigint | NOT NULL | merge_requests.id 1:1 · CASCADE | ||
scheduled_at | timestamptz | NOT NULL | |||
completed_at | timestamptz | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
failed_count | integer | NOT NULL | 0 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
project_id — merge_request_context_commit_diff_files.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
sha | bytea | NOT NULL | |||
relative_order PK | integer | NOT NULL | |||
new_file | boolean | NOT NULL | |||
renamed_file | boolean | NOT NULL | |||
deleted_file | boolean | NOT NULL | |||
too_large | boolean | NOT NULL | |||
a_mode | varchar(255) | NOT NULL | |||
b_mode | varchar(255) | NOT NULL | |||
new_path | text | NOT NULL | |||
old_path | text | NOT NULL | |||
diff | text | null | |||
binary | boolean | null | |||
merge_request_context_commit_id PK | bigint | NOT NULL | merge_request_context_commits.id 1:N · CASCADE | ||
generated | boolean | null | |||
encoded_file_path | boolean | NOT NULL | FALSE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
project_id — merge_request_context_commits.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.merge_request_id — merge_request_context_commits.merge_request_id may be NULL, so the relationship to merge_requests is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id, merge_request_id — merge_request_context_commits has several nullable FKs (project_id, merge_request_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
authored_date | timestamptz | null | |||
committed_date | timestamptz | null | |||
relative_order | integer | NOT NULL | |||
sha | bytea | NOT NULL | |||
author_name | text | null | |||
author_email | text | null | |||
committer_name | text | null | |||
committer_email | text | null | |||
message | text | null | |||
merge_request_id | bigint | null | merge_requests.id 1:N · CASCADE | CHECK merge_request_id IS NOT NULL | |
trailers | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: merge_request_id IS NOT NULL; project_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
name | text | null | CHECK char_length(name) <= 512 | ||
email | text | null | CHECK char_length(email) <= 512 | ||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE |
Table checks: char_length(name) <= 512; char_length(email) <= 512; COALESCE(name, CAST('' AS text)) <> CAST('' AS text) OR COALESCE(email, CAST('' AS text)) <> CAST('' AS text)
authored_date — authored_date stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.committed_date — committed_date stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.merge_request_diff_commits references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
authored_date | timestamp | null | |||
committed_date | timestamp | null | |||
merge_request_diff_id PK | bigint | NOT NULL | |||
relative_order PK | integer | NOT NULL | |||
sha | bytea | null | |||
message | text | null | |||
trailers | jsonb | null | CAST('{}' AS jsonb) | ||
commit_author_id | bigint | null | |||
committer_id | bigint | null | |||
merge_request_commits_metadata_id | bigint | null | |||
project_id | bigint | null |
project_id — merge_request_diff_details.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
merge_request_diff_id PK | bigint | NOT NULL | merge_request_diffs.id 1:1 · CASCADE | ||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
verification_retry_count | smallint | null | |||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_started_at | timestamptz | null | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL; char_length(verification_failure) <= 255
project_id — merge_request_diff_files.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.a_mode — a_mode is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.b_mode — b_mode is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
merge_request_diff_id PK | bigint | NOT NULL | merge_request_diffs.id 1:N · CASCADE | ||
relative_order PK | integer | NOT NULL | |||
new_file | boolean | NOT NULL | |||
renamed_file | boolean | NOT NULL | |||
deleted_file | boolean | NOT NULL | |||
too_large | boolean | NOT NULL | |||
a_mode | varchar | NOT NULL | |||
b_mode | varchar | NOT NULL | |||
new_path | text | null | |||
old_path | text | NOT NULL | |||
diff | text | null | |||
binary | boolean | null | |||
external_diff_offset | integer | null | |||
external_diff_size | integer | null | |||
generated | boolean | null | |||
encoded_file_path | boolean | NOT NULL | FALSE | ||
project_id | bigint | null | projects.id 1:N · CASCADE |
project_id — merge_request_diffs.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.state — state is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
state | varchar | null | |||
merge_request_id | bigint | NOT NULL | merge_requests.id 1:N · CASCADE | ||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
base_commit_sha | varchar | null | |||
real_size | varchar | null | |||
head_commit_sha | varchar | null | |||
start_commit_sha | varchar | null | |||
commits_count | integer | null | |||
external_diff | varchar | null | |||
external_diff_store | integer | null | 1 | CHECK external_diff_store IS NOT NULL | |
stored_externally | boolean | null | |||
files_count | smallint | null | |||
sorted | boolean | NOT NULL | FALSE | ||
diff_type | smallint | NOT NULL | 1 | ||
patch_id_sha | bytea | null | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL | |
base_commit_sha_bytea | bytea | null | |||
start_commit_sha_bytea | bytea | null | |||
head_commit_sha_bytea | bytea | null |
Table checks: project_id IS NOT NULL; external_diff_store IS NOT NULL
merge_request_id, project_id — merge_request_merge_schedules looks like a many-to-many link table but has no unique constraint across (merge_request_id, project_id). The same pair can be inserted twice; every join through it will double-count.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
merge_request_id | bigint | NOT NULL | merge_requests.id 1:1 · CASCADE | ||
merge_after | timestamptz | null | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE |
target_project_id — merge_request_metrics.target_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.merged_by_id — merge_request_metrics.merged_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.latest_closed_by_id — merge_request_metrics.latest_closed_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.target_project_id, merged_by_id, latest_closed_by_id — merge_request_metrics has several nullable FKs (target_project_id, merged_by_id, latest_closed_by_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.latest_build_started_at — latest_build_started_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.latest_build_finished_at — latest_build_finished_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.first_deployed_to_production_at — first_deployed_to_production_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.merged_at — merged_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
merge_request_id | bigint | NOT NULL | merge_requests.id 1:1 · CASCADE | ||
latest_build_started_at | timestamp | null | |||
latest_build_finished_at | timestamp | null | |||
first_deployed_to_production_at | timestamp | null | |||
merged_at | timestamp | null | |||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL | |||
merged_by_id | bigint | null | users.id 1:N · SET NULL | ||
latest_closed_by_id | bigint | null | users.id 1:N · SET NULL | ||
latest_closed_at | timestamptz | null | |||
first_comment_at | timestamptz | null | |||
first_commit_at | timestamptz | null | |||
last_commit_at | timestamptz | null | |||
diff_size | integer | null | |||
modified_paths_size | integer | null | |||
commits_count | integer | null | |||
first_approved_at | timestamptz | null | |||
first_reassigned_at | timestamptz | null | |||
added_lines | integer | null | |||
removed_lines | integer | null | |||
target_project_id | bigint | null | projects.id 1:N · CASCADE | CHECK target_project_id IS NOT NULL | |
id PK | bigint | NOT NULL | |||
first_contribution | boolean | NOT NULL | FALSE | ||
pipeline_id | bigint | null | |||
reviewer_first_assigned_at | timestamptz | null |
Table checks: target_project_id IS NOT NULL
project_id — merge_request_predictions.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
merge_request_id PK | bigint | NOT NULL | merge_requests.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
suggested_reviewers | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
accepted_reviewers | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
merge_request_requested_changes references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | |||
merge_request_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
project_id — merge_request_reviewers.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
merge_request_id | bigint | NOT NULL | merge_requests.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
state | smallint | NOT NULL | 0 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL | |
updated_at | timestamptz | null |
Table checks: project_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
filters | jsonb | NOT NULL | CAST('{}' AS jsonb) |
Table checks: char_length(name) <= 255
project_id — merge_request_user_mentions.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.note_id — merge_request_user_mentions.note_id may be NULL, so the relationship to notes is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id, note_id — merge_request_user_mentions has several nullable FKs (project_id, note_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
merge_request_id | bigint | NOT NULL | merge_requests.id 1:N · CASCADE | ||
mentioned_users_ids | bigint[] | null | |||
mentioned_projects_ids | bigint[] | null | |||
mentioned_groups_ids | bigint[] | null | |||
note_id | bigint | null | notes.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
latest_merge_request_diff_id — merge_requests.latest_merge_request_diff_id may be NULL, so the relationship to merge_request_diffs is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.assignee_id — merge_requests.assignee_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.updated_by_id — merge_requests.updated_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.milestone_id — merge_requests.milestone_id may be NULL, so the relationship to milestones is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.merge_user_id — merge_requests.merge_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.author_id — merge_requests.author_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.source_project_id — merge_requests.source_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.merge_status — merge_status is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.latest_merge_request_diff_id, assignee_id, updated_by_id, milestone_id, merge_user_id, author_id, source_project_id — merge_requests has several nullable FKs (latest_merge_request_diff_id, assignee_id, updated_by_id, milestone_id, merge_user_id, author_id, source_project_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.last_edited_at — last_edited_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
target_branch | varchar | NOT NULL | |||
source_branch | varchar | NOT NULL | |||
source_project_id | bigint | null | projects.id 1:N · SET NULL | ||
author_id | bigint | null | users.id 1:N · SET NULL | ||
assignee_id | bigint | null | users.id 1:N · SET NULL | ||
title | varchar | null | |||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
milestone_id | bigint | null | milestones.id 1:N · SET NULL | ||
merge_status | varchar | NOT NULL | CAST('unchecked' AS varchar) | ||
target_project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
iid | integer | null | |||
description | text | null | |||
updated_by_id | bigint | null | users.id 1:N · SET NULL | ||
merge_error | text | null | |||
merge_params | text | null | |||
merge_when_pipeline_succeeds | boolean | NOT NULL | FALSE | ||
merge_user_id | bigint | null | users.id 1:N · SET NULL | ||
merge_commit_sha | varchar | null | |||
approvals_before_merge | integer | null | |||
rebase_commit_sha | varchar | null | |||
in_progress_merge_commit_sha | varchar | null | |||
lock_version | integer | null | 0 | CHECK lock_version IS NOT NULL | |
title_html | text | null | |||
description_html | text | null | |||
time_estimate | integer | null | 0 | ||
squash | boolean | NOT NULL | FALSE | ||
cached_markdown_version | integer | null | |||
last_edited_at | timestamp | null | |||
last_edited_by_id | bigint | null | |||
merge_jid | varchar | null | |||
discussion_locked | boolean | null | |||
latest_merge_request_diff_id | bigint | null | merge_request_diffs.id 1:N · SET NULL | ||
allow_maintainer_to_push | boolean | null | TRUE | ||
state_id | smallint | NOT NULL | 1 | ||
rebase_jid | varchar | null | |||
squash_commit_sha | bytea | null | |||
merge_ref_sha | bytea | null | |||
draft | boolean | NOT NULL | FALSE | ||
prepared_at | timestamptz | null | |||
merged_commit_sha | bytea | null | |||
override_requested_changes | boolean | NOT NULL | FALSE | ||
head_pipeline_id | bigint | null | |||
imported_from | smallint | NOT NULL | 0 | ||
retargeted | boolean | NOT NULL | FALSE |
Table checks: lock_version IS NOT NULL
group_id — merge_requests_approval_rules.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — merge_requests_approval_rules.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.source_rule_id — merge_requests_approval_rules.source_rule_id may be NULL, so the relationship to merge_requests_approval_rules is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id, project_id, source_rule_id — merge_requests_approval_rules has several nullable FKs (group_id, project_id, source_rule_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
approvals_required | integer | NOT NULL | 0 | ||
rule_type | smallint | NOT NULL | 0 | ||
origin | smallint | NOT NULL | 0 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
source_rule_id | bigint | null | merge_requests_approval_rules.id 1:N · SET NULL | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
Table checks: num_nonnulls(group_id, project_id) = 1; char_length(name) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
approval_rule_id | bigint | NOT NULL | merge_requests_approval_rules.id 1:N · CASCADE | ||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
project_id — merge_requests_approval_rules_approver_users.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — merge_requests_approval_rules_approver_users.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
approval_rule_id | bigint | NOT NULL | merge_requests_approval_rules.id 1:N · CASCADE | ||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
Table checks: num_nonnulls(group_id, project_id) = 1
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
approval_rule_id | bigint | NOT NULL | merge_requests_approval_rules.id 1:N · CASCADE | ||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
approval_rule_id | bigint | NOT NULL | merge_requests_approval_rules.id 1:N · CASCADE | ||
merge_request_id | bigint | NOT NULL | merge_requests.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
approval_rule_id | bigint | NOT NULL | merge_requests_approval_rules.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
project_id — merge_requests_closing_issues.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
merge_request_id | bigint | NOT NULL | merge_requests.id 1:N · CASCADE | ||
issue_id | bigint | NOT NULL | issues.id 1:N · CASCADE | ||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL | |||
from_mr_description | boolean | NOT NULL | TRUE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL | |
link_type | smallint | NOT NULL | 0 |
Table checks: project_id IS NOT NULL
target_project_id — merge_requests_compliance_violations.target_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
violating_user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
merge_request_id | bigint | NOT NULL | merge_requests.id 1:N · CASCADE | ||
reason | smallint | NOT NULL | |||
severity_level | smallint | NOT NULL | 0 | ||
merged_at | timestamptz | null | |||
target_project_id | bigint | null | projects.id 1:N · CASCADE | CHECK target_project_id IS NOT NULL | |
title | text | null | |||
target_branch | text | null |
Table checks: target_project_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
merge_request_id | bigint | NOT NULL | merge_requests.id 1:1 · CASCADE | ||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
target_project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
target_branch | text | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
merged_at | timestamptz | null | |||
duration | integer | null | |||
pipeline_id | bigint | null |
Tables whose names start with project_.
project_daily_statistics.project_id → projects.id — one projects, many project_daily_statistics · required · ON DELETE CASCADE · indexedproject_access_tokens.personal_access_token_id → personal_access_tokens.id — one personal_access_tokens, many project_access_tokens · required · ON DELETE CASCADE · indexedproject_access_tokens.project_id → projects.id — one projects, many project_access_tokens · required · ON DELETE CASCADE · indexedproject_alerting_settings.project_id → projects.id — one projects, at most one project_alerting_settings · required · ON DELETE CASCADE · indexedproject_aliases.project_id → projects.id — one projects, many project_aliases · required · ON DELETE CASCADE · indexedproject_authorizations.project_id → projects.id — one projects, many project_authorizations · required · ON DELETE CASCADE · indexedproject_authorizations_for_migration.project_id → projects.id — one projects, many project_authorizations_for_migration · required · ON DELETE CASCADE · indexedproject_auto_devops.project_id → projects.id — one projects, at most one project_auto_devops · required · ON DELETE CASCADE · indexedproject_build_artifacts_size_refreshes.project_id → projects.id — one projects, at most one project_build_artifacts_size_refreshes · required · ON DELETE CASCADE · indexedproject_ci_cd_settings.project_id → projects.id — one projects, at most one project_ci_cd_settings · required · ON DELETE CASCADE · indexedproject_ci_feature_usages.project_id → projects.id — one projects, many project_ci_feature_usages · required · ON DELETE CASCADE · indexedproject_compliance_framework_settings.framework_id → compliance_management_frameworks.id — one compliance_management_frameworks, many project_compliance_framework_settings · optional · ON DELETE CASCADE · indexedproject_compliance_framework_settings.project_id → projects.id — one projects, many project_compliance_framework_settings · required · ON DELETE CASCADE · indexedproject_compliance_standards_adherence.namespace_id → namespaces.id — one namespaces, many project_compliance_standards_adherence · optional · ON DELETE SET NULL · indexedproject_compliance_standards_adherence.project_id → projects.id — one projects, many project_compliance_standards_adherence · required · ON DELETE CASCADE · indexedproject_compliance_violations.project_id → projects.id — one projects, many project_compliance_violations · required · ON DELETE CASCADE · indexedproject_compliance_violations.namespace_id → namespaces.id — one namespaces, many project_compliance_violations · required · ON DELETE CASCADE · indexedproject_compliance_violations.compliance_requirements_control_id → compliance_requirements_controls.id — one compliance_requirements_controls, many project_compliance_violations · required · ON DELETE CASCADE · indexedproject_compliance_violations_issues.project_compliance_violation_id → project_compliance_violations.id — one project_compliance_violations, many project_compliance_violations_issues · required · ON DELETE CASCADE · indexedproject_compliance_violations_issues.project_id → projects.id — one projects, many project_compliance_violations_issues · required · ON DELETE CASCADE · indexedproject_compliance_violations_issues.issue_id → issues.id — one issues, many project_compliance_violations_issues · required · ON DELETE CASCADE · indexedproject_control_compliance_statuses.namespace_id → namespaces.id — one namespaces, many project_control_compliance_statuses · required · ON DELETE CASCADE · indexedproject_control_compliance_statuses.project_id → projects.id — one projects, many project_control_compliance_statuses · required · ON DELETE CASCADE · indexedproject_control_compliance_statuses.compliance_requirement_id → compliance_requirements.id — one compliance_requirements, many project_control_compliance_statuses · required · ON DELETE CASCADE · indexedproject_control_compliance_statuses.requirement_status_id → project_requirement_compliance_statuses.id — one project_requirement_compliance_statuses, many project_control_compliance_statuses · optional · ON DELETE SET NULL · indexedproject_control_compliance_statuses.compliance_requirements_control_id → compliance_requirements_controls.id — one compliance_requirements_controls, many project_control_compliance_statuses · required · ON DELETE CASCADE · indexedproject_custom_attributes.project_id → projects.id — one projects, many project_custom_attributes · required · ON DELETE CASCADE · indexedproject_deploy_tokens.deploy_token_id → deploy_tokens.id — one deploy_tokens, many project_deploy_tokens · required · ON DELETE CASCADE · indexedproject_deploy_tokens.project_id → projects.id — one projects, many project_deploy_tokens · required · ON DELETE CASCADE · indexedproject_error_tracking_settings.project_id → projects.id — one projects, at most one project_error_tracking_settings · required · ON DELETE CASCADE · indexedproject_export_jobs.user_id → users.id — one users, many project_export_jobs · optional · ON DELETE SET NULL · indexedproject_export_jobs.project_id → projects.id — one projects, many project_export_jobs · required · ON DELETE CASCADE · indexedproject_feature_usages.project_id → projects.id — one projects, at most one project_feature_usages · required · ON DELETE CASCADE · indexedproject_features.project_id → projects.id — one projects, at most one project_features · required · ON DELETE CASCADE · indexedproject_group_links.group_id → namespaces.id — one namespaces, many project_group_links · required · ON DELETE CASCADE · indexedproject_group_links.member_role_id → member_roles.id — one member_roles, many project_group_links · optional · ON DELETE SET NULL · indexedproject_group_links.project_id → projects.id — one projects, many project_group_links · required · ON DELETE CASCADE · indexedproject_import_data.project_id → projects.id — one projects, many project_import_data · required · ON DELETE CASCADE · indexedproject_import_export_relation_export_upload_upload_states.project_import_export_relation_export_upload_upload_id → project_import_export_relation_export_upload_uploads.id — one project_import_export_relation_export_upload_uploads, at most one project_import_export_relation_export_upload_upload_states · required · ON DELETE CASCADE · indexedproject_import_export_relation_export_upload_upload_states.project_id → projects.id — one projects, many project_import_export_relation_export_upload_upload_states · required · ON DELETE CASCADE · indexedproject_incident_management_settings.project_id → projects.id — one projects, at most one project_incident_management_settings · required · ON DELETE CASCADE · indexedproject_metrics_settings.project_id → projects.id — one projects, at most one project_metrics_settings · required · ON DELETE CASCADE · indexedproject_mirror_data.project_id → projects.id — one projects, at most one project_mirror_data · required · ON DELETE CASCADE · indexedproject_pages_metadata.project_id → projects.id — one projects, at most one project_pages_metadata · required · ON DELETE CASCADE · indexedproject_push_rules.project_id → projects.id — one projects, at most one project_push_rules · required · ON DELETE CASCADE · indexedproject_relation_export_uploads.project_id → projects.id — one projects, many project_relation_export_uploads · optional · ON DELETE CASCADE · indexedproject_relation_export_uploads.project_relation_export_id → project_relation_exports.id — one project_relation_exports, many project_relation_export_uploads · required · ON DELETE CASCADE · indexedproject_relation_exports.project_id → projects.id — one projects, many project_relation_exports · optional · ON DELETE CASCADE · indexedproject_relation_exports.project_export_job_id → project_export_jobs.id — one project_export_jobs, many project_relation_exports · required · ON DELETE CASCADE · indexedproject_repositories.project_id → projects.id — one projects, at most one project_repositories · required · ON DELETE CASCADE · indexedproject_repository_states.project_id → projects.id — one projects, many project_repository_states · required · ON DELETE CASCADE · indexedproject_repository_states.project_repository_id → project_repositories.id — one project_repositories, at most one project_repository_states · required · ON DELETE CASCADE · indexedproject_repository_storage_moves.project_id → projects.id — one projects, many project_repository_storage_moves · required · ON DELETE CASCADE · indexedproject_requirement_compliance_statuses.namespace_id → namespaces.id — one namespaces, many project_requirement_compliance_statuses · required · ON DELETE RESTRICT · indexedproject_requirement_compliance_statuses.compliance_requirement_id → compliance_requirements.id — one compliance_requirements, many project_requirement_compliance_statuses · required · ON DELETE CASCADE · indexedproject_requirement_compliance_statuses.compliance_framework_id → compliance_management_frameworks.id — one compliance_management_frameworks, many project_requirement_compliance_statuses · required · ON DELETE CASCADE · indexedproject_requirement_compliance_statuses.project_id → projects.id — one projects, many project_requirement_compliance_statuses · required · ON DELETE RESTRICT · indexedproject_saved_replies.project_id → projects.id — one projects, many project_saved_replies · required · ON DELETE CASCADE · indexedproject_saved_replies.project_id → projects.id — one projects, many project_saved_replies · required · ON DELETE NO ACTION · indexedproject_secrets_manager_maintenance_tasks.organization_id → organizations.id — one organizations, many project_secrets_manager_maintenance_tasks · optional · ON DELETE CASCADE · indexedproject_secrets_manager_maintenance_tasks.project_secrets_manager_id → project_secrets_managers.id — one project_secrets_managers, many project_secrets_manager_maintenance_tasks · optional · ON DELETE CASCADE · indexedproject_secrets_managers.project_id → projects.id — one projects, at most one project_secrets_managers · optional · ON DELETE CASCADE · indexedproject_security_settings.project_id → projects.id — one projects, at most one project_security_settings · required · ON DELETE CASCADE · indexedproject_settings.pipeline_execution_policy_bot_access_group_id → namespaces.id — one namespaces, many project_settings · optional · ON DELETE SET NULL · indexedproject_settings.duo_dependency_bump_breaking_changes_enabled_by_id → users.id — one users, many project_settings · optional · ON DELETE SET NULL · indexedproject_settings.push_rule_id → push_rules.id — one push_rules, at most one project_settings · optional · ON DELETE SET NULL · indexedproject_settings.project_id → projects.id — one projects, at most one project_settings · required · ON DELETE CASCADE · indexedproject_statistics.root_namespace_id → namespaces.id — one namespaces, many project_statistics · optional · ON DELETE SET NULL · indexedproject_statistics.project_id → projects.id — one projects, at most one project_statistics · required · ON DELETE CASCADE · indexedproject_to_security_attributes.security_attribute_id → security_attributes.id — one security_attributes, many project_to_security_attributes · required · ON DELETE CASCADE · indexedproject_topic_upload_states.organization_id → organizations.id — one organizations, many project_topic_upload_states · required · ON DELETE CASCADE · indexedproject_topic_upload_states.project_topic_upload_id → project_topic_uploads.id — one project_topic_uploads, at most one project_topic_upload_states · required · ON DELETE CASCADE · indexedproject_topics.topic_id → topics.id — one topics, many project_topics · required · ON DELETE CASCADE · indexedproject_topics.project_id → projects.id — one projects, many project_topics · required · ON DELETE CASCADE · indexedproject_type_ci_runner_machines.runner_id, runner_type → project_type_ci_runners.id, runner_type — one project_type_ci_runners, many project_type_ci_runner_machines · required · ON DELETE CASCADE · indexedproject_upload_states.project_id → projects.id — one projects, many project_upload_states · required · ON DELETE CASCADE · indexedproject_upload_states.project_upload_id → project_uploads.id — one project_uploads, at most one project_upload_states · required · ON DELETE CASCADE · indexedproject_wiki_repositories.project_id → projects.id — one projects, at most one project_wiki_repositories · required · ON DELETE CASCADE · indexedtarget_type, target_id — target_id points at different tables depending on target_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.project_audit_events references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | nextval(CAST('audit_events_id_seq' AS regclass)) | ||
created_at PK | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | |||
author_id | bigint | NOT NULL | |||
target_id | bigint | null | |||
event_name | text | null | CHECK char_length(event_name) <= 255 | ||
details | text | null | |||
ip_address | inet | null | |||
author_name | text | null | CHECK char_length(author_name) <= 255 | ||
entity_path | text | null | CHECK char_length(entity_path) <= 5500 | ||
target_details | text | null | CHECK char_length(target_details) <= 5500 | ||
target_type | text | null | CHECK char_length(target_type) <= 255 |
Table checks: char_length(author_name) <= 255; char_length(entity_path) <= 5500; char_length(event_name) <= 255; char_length(target_details) <= 5500; char_length(target_type) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
fetch_count | integer | NOT NULL | |||
date PK | date | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
personal_access_token_id PK | bigint | NOT NULL | personal_access_tokens.id 1:N · CASCADE | ||
project_id PK | bigint | NOT NULL | projects.id 1:N · CASCADE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | projects.id 1:1 · CASCADE | ||
encrypted_token | varchar | NOT NULL | |||
encrypted_token_iv | varchar | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
name | varchar | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
user_id PK | bigint | NOT NULL | |||
project_id PK | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
access_level PK | integer | NOT NULL | |||
is_unique | boolean | null |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
user_id PK | bigint | NOT NULL | |||
project_id PK | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
access_level | smallint | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
enabled | boolean | null | |||
deploy_strategy | integer | NOT NULL | 0 |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:1 · CASCADE | ||
last_job_artifact_id | bigint | null | |||
state | smallint | NOT NULL | 1 | ||
refresh_started_at | timestamptz | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
last_job_artifact_id_on_refresh_start | bigint | null | 0 |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:1 · CASCADE | ||
group_runners_enabled | boolean | NOT NULL | TRUE | ||
merge_pipelines_enabled | boolean | null | |||
default_git_depth | integer | null | |||
forward_deployment_enabled | boolean | null | |||
merge_trains_enabled | boolean | null | FALSE | ||
auto_rollback_enabled | boolean | NOT NULL | FALSE | ||
keep_latest_artifact | boolean | NOT NULL | TRUE | ||
job_token_scope_enabled | boolean | NOT NULL | FALSE | ||
runner_token_expiration_interval | integer | null | |||
separated_caches | boolean | NOT NULL | TRUE | ||
allow_fork_pipelines_to_run_in_parent_project | boolean | NOT NULL | TRUE | ||
inbound_job_token_scope_enabled | boolean | NOT NULL | TRUE | ||
forward_deployment_rollback_allowed | boolean | NOT NULL | TRUE | ||
merge_trains_skip_train_allowed | boolean | NOT NULL | FALSE | ||
restrict_pipeline_cancellation_role | smallint | NOT NULL | 0 | ||
pipeline_variables_minimum_override_role | smallint | NOT NULL | 3 | ||
push_repository_for_job_token_allowed | boolean | NOT NULL | FALSE | ||
id_token_sub_claim_components | varchar[] | NOT NULL | CAST('{project_path,ref_type,ref}' AS varchar[]) | ||
delete_pipelines_in_seconds | integer | null | |||
allow_composite_identities_to_run_pipelines | boolean | NOT NULL | FALSE | ||
display_pipeline_variables | boolean | NOT NULL | FALSE | ||
resource_group_default_process_mode | smallint | NOT NULL | 0 | ||
max_pipelines_per_merge_train | smallint | null | |||
cross_project_push_for_job_token_allowed | boolean | NOT NULL | FALSE | ||
merge_train_enforcement | smallint | NOT NULL | 0 | ||
skip_branch_pipelines_for_mrs | boolean | NOT NULL | FALSE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
feature | smallint | NOT NULL | |||
default_branch | boolean | NOT NULL | FALSE |
framework_id — project_compliance_framework_settings.framework_id may be NULL, so the relationship to compliance_management_frameworks is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
framework_id | bigint | null | compliance_management_frameworks.id 1:N · CASCADE | CHECK framework_id IS NOT NULL | |
id PK | bigint | NOT NULL | |||
created_at | timestamptz | null |
Table checks: framework_id IS NOT NULL
namespace_id — project_compliance_standards_adherence.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
namespace_id | bigint | null | namespaces.id 1:N · SET NULL | ||
status | smallint | NOT NULL | |||
check_name | smallint | NOT NULL | |||
standard | smallint | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
audit_event_id | bigint | NOT NULL | |||
compliance_requirements_control_id | bigint | NOT NULL | compliance_requirements_controls.id 1:N · CASCADE | ||
status | smallint | NOT NULL | |||
audit_event_table_name | smallint | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_compliance_violation_id | bigint | NOT NULL | project_compliance_violations.id 1:N · CASCADE | ||
issue_id | bigint | NOT NULL | issues.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE |
requirement_status_id — project_control_compliance_statuses.requirement_status_id may be NULL, so the relationship to project_requirement_compliance_statuses is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
compliance_requirements_control_id | bigint | NOT NULL | compliance_requirements_controls.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
compliance_requirement_id | bigint | NOT NULL | compliance_requirements.id 1:N · CASCADE | ||
status | smallint | NOT NULL | |||
requirement_status_id | bigint | null | project_requirement_compliance_statuses.id 1:N · SET NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
key | varchar | NOT NULL | |||
value | varchar | NOT NULL |
project_data_transfers references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | |||
repository_egress | bigint | NOT NULL | 0 | ||
artifacts_egress | bigint | NOT NULL | 0 | ||
packages_egress | bigint | NOT NULL | 0 | ||
registry_egress | bigint | NOT NULL | 0 | ||
date | date | NOT NULL | CHECK date = date_trunc(CAST('month' AS text), CAST(date AS timestamptz)) | ||
created_at | timestamptz | NOT NULL |
Table checks: date = date_trunc(CAST('month' AS text), CAST(date AS timestamptz))
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
deploy_token_id | bigint | NOT NULL | deploy_tokens.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | projects.id 1:1 · CASCADE | ||
enabled | boolean | NOT NULL | FALSE | ||
api_url | varchar | null | |||
encrypted_token | varchar | null | |||
encrypted_token_iv | varchar | null | |||
project_name | varchar | null | |||
organization_name | varchar | null | |||
integrated | boolean | NOT NULL | TRUE | ||
sentry_project_id | bigint | null |
user_id — project_export_jobs.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
jid | varchar(100) | NOT NULL | |||
user_id | bigint | null | users.id 1:N · SET NULL | ||
exported_by_admin | boolean | null | FALSE |
jira_dvcs_cloud_last_sync_at — jira_dvcs_cloud_last_sync_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.jira_dvcs_server_last_sync_at — jira_dvcs_server_last_sync_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | projects.id 1:1 · CASCADE | ||
jira_dvcs_cloud_last_sync_at | timestamp | null | |||
jira_dvcs_server_last_sync_at | timestamp | null |
created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:1 · CASCADE | ||
merge_requests_access_level | integer | null | |||
issues_access_level | integer | null | |||
wiki_access_level | integer | null | |||
snippets_access_level | integer | NOT NULL | 20 | ||
builds_access_level | integer | null | |||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
repository_access_level | integer | NOT NULL | 20 | ||
pages_access_level | integer | NOT NULL | |||
forking_access_level | integer | null | |||
metrics_dashboard_access_level | integer | null | |||
requirements_access_level | integer | NOT NULL | 20 | ||
operations_access_level | integer | NOT NULL | 20 | ||
analytics_access_level | integer | NOT NULL | 20 | ||
security_and_compliance_access_level | integer | NOT NULL | 10 | ||
container_registry_access_level | integer | NOT NULL | 0 | ||
package_registry_access_level | integer | NOT NULL | 0 | ||
monitor_access_level | integer | NOT NULL | 20 | ||
infrastructure_access_level | integer | NOT NULL | 20 | ||
feature_flags_access_level | integer | NOT NULL | 20 | ||
environments_access_level | integer | NOT NULL | 20 | ||
releases_access_level | integer | NOT NULL | 20 | ||
model_experiments_access_level | integer | NOT NULL | |||
model_registry_access_level | integer | NOT NULL |
member_role_id — project_group_links.member_role_id may be NULL, so the relationship to member_roles is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
group_access | integer | NOT NULL | 30 | ||
expires_at | date | null | |||
member_role_id | bigint | null | member_roles.id 1:N · SET NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
data | text | null | |||
encrypted_credentials | text | null | |||
encrypted_credentials_iv | varchar | null | |||
encrypted_credentials_salt | varchar | null |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
project_import_export_relation_export_upload_upload_id | bigint | NOT NULL | project_import_export_relation_export_upload_uploads.id 1:1 · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 |
Table checks: char_length(verification_failure) <= 255
model_type — model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.model_type, model_id — model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | nextval(CAST('uploads_id_seq' AS regclass)) | ||
size | bigint | NOT NULL | |||
model_id | bigint | NOT NULL | |||
uploaded_by_user_id | bigint | null | |||
organization_id | bigint | null | |||
namespace_id | bigint | null | |||
project_id | bigint | null | CHECK project_id IS NOT NULL | ||
created_at | timestamp | null | |||
store | integer | NOT NULL | 1 | ||
version | integer | null | 1 | ||
path | text | NOT NULL | CHECK char_length(path) <= 511 | ||
checksum | text | null | CHECK char_length(checksum) <= 64 | ||
model_type PK | text | NOT NULL | |||
uploader | text | NOT NULL | |||
mount_point | text | null | |||
secret | text | null |
Table checks: char_length(path) <= 511; project_id IS NOT NULL; char_length(checksum) <= 64
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | projects.id 1:1 · CASCADE | ||
create_issue | boolean | NOT NULL | FALSE | ||
send_email | boolean | NOT NULL | FALSE | ||
issue_template_key | text | null | |||
pagerduty_active | boolean | NOT NULL | FALSE | ||
encrypted_pagerduty_token | bytea | null | CHECK octet_length(encrypted_pagerduty_token) <= 255 | ||
encrypted_pagerduty_token_iv | bytea | null | CHECK octet_length(encrypted_pagerduty_token_iv) <= 12 | ||
auto_close_incident | boolean | NOT NULL | TRUE | ||
sla_timer | boolean | null | FALSE | ||
sla_timer_minutes | integer | null |
Table checks: octet_length(encrypted_pagerduty_token_iv) <= 12; octet_length(encrypted_pagerduty_token) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | projects.id 1:1 · CASCADE | ||
external_dashboard_url | varchar | null | |||
dashboard_timezone | smallint | NOT NULL | 0 |
status — status is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.last_update_started_at — last_update_started_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.last_update_scheduled_at — last_update_scheduled_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.next_execution_timestamp — next_execution_timestamp stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:1 · CASCADE | ||
retry_count | integer | NOT NULL | 0 | ||
last_update_started_at | timestamp | null | |||
last_update_scheduled_at | timestamp | null | |||
next_execution_timestamp | timestamp | null | |||
status | varchar | null | |||
jid | varchar | null | |||
last_error | text | null | |||
last_update_at | timestamptz | null | |||
last_successful_update_at | timestamptz | null | |||
correlation_id_value | varchar(128) | null | |||
checksums | jsonb | NOT NULL | CAST('{}' AS jsonb) |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | projects.id 1:1 · CASCADE | ||
onboarding_complete | boolean | NOT NULL | FALSE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:1 · CASCADE | ||
max_file_size | integer | NOT NULL | 0 | ||
member_check | boolean | NOT NULL | FALSE | ||
prevent_secrets | boolean | NOT NULL | FALSE | ||
commit_committer_name_check | boolean | NOT NULL | FALSE | ||
deny_delete_tag | boolean | null | |||
reject_unsigned_commits | boolean | null | |||
commit_committer_check | boolean | null | |||
reject_non_dco_commits | boolean | null | |||
commit_message_regex | text | null | CHECK char_length(commit_message_regex) <= 511 | ||
branch_name_regex | text | null | CHECK char_length(branch_name_regex) <= 511 | ||
commit_message_negative_regex | text | null | CHECK char_length(commit_message_negative_regex) <= 2047 | ||
author_email_regex | text | null | CHECK char_length(author_email_regex) <= 511 | ||
file_name_regex | text | null | CHECK char_length(file_name_regex) <= 511 |
Table checks: char_length(author_email_regex) <= 511; char_length(commit_message_regex) <= 511; char_length(commit_message_negative_regex) <= 2047; char_length(file_name_regex) <= 511; char_length(branch_name_regex) <= 511
project_id — project_relation_export_uploads.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_relation_export_id | bigint | NOT NULL | project_relation_exports.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
export_file | text | NOT NULL | CHECK char_length(export_file) <= 255 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: char_length(export_file) <= 255; project_id IS NOT NULL
project_id — project_relation_exports.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_export_job_id | bigint | NOT NULL | project_export_jobs.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
relation | text | NOT NULL | CHECK char_length(relation) <= 255 | ||
jid | text | null | CHECK char_length(jid) <= 255 | ||
export_error | text | null | CHECK char_length(export_error) <= 300 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: char_length(jid) <= 255; char_length(relation) <= 255; char_length(export_error) <= 300; project_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
shard_id | bigint | null | |||
disk_path | varchar | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:1 · CASCADE | ||
object_format | smallint | NOT NULL | 0 |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
project_repository_id | bigint | NOT NULL | project_repositories.id 1:1 · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 |
Table checks: char_length(verification_failure) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
state | smallint | NOT NULL | 1 | ||
source_storage_name | text | NOT NULL | CHECK char_length(source_storage_name) <= 255 | ||
destination_storage_name | text | NOT NULL | CHECK char_length(destination_storage_name) <= 255 | ||
error_message | text | null | CHECK char_length(error_message) <= 256 |
Table checks: char_length(error_message) <= 256; char_length(destination_storage_name) <= 255; char_length(source_storage_name) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · RESTRICT | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · RESTRICT | ||
compliance_requirement_id | bigint | NOT NULL | compliance_requirements.id 1:N · CASCADE | ||
compliance_framework_id | bigint | NOT NULL | compliance_management_frameworks.id 1:N · CASCADE | ||
pass_count | integer | NOT NULL | 0 | ||
fail_count | integer | NOT NULL | 0 | ||
pending_count | integer | NOT NULL | 0 |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
content | text | NOT NULL | CHECK char_length(content) <= 10000 |
Table checks: char_length(content) <= 10000; char_length(name) <= 255
organization_id — project_secrets_manager_maintenance_tasks.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_secrets_manager_id — project_secrets_manager_maintenance_tasks.project_secrets_manager_id may be NULL, so the relationship to project_secrets_managers is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id, project_secrets_manager_id — project_secrets_manager_maintenance_tasks has several nullable FKs (organization_id, project_secrets_manager_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | null | |||
project_secrets_manager_id | bigint | null | project_secrets_managers.id 1:N · CASCADE | ||
last_processed_at | timestamptz | null | |||
action | smallint | NOT NULL | |||
retry_count | smallint | NOT NULL | 0 | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE | CHECK organization_id IS NOT NULL | |
project_id | bigint | null | CHECK project_id IS NOT NULL | ||
root_namespace_id | bigint | null | CHECK root_namespace_id IS NOT NULL | ||
parent_group_id | bigint | null |
Table checks: organization_id IS NOT NULL; project_id IS NOT NULL; root_namespace_id IS NOT NULL
project_id — project_secrets_managers.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | null | projects.id 1:1 · CASCADE | CHECK project_id IS NOT NULL | |
status | smallint | NOT NULL | 0 | ||
namespace_path | text | null | CHECK char_length(namespace_path) <= 64 | ||
project_path | text | null | CHECK char_length(project_path) <= 64 | ||
organization_id | bigint | null | CHECK organization_id IS NOT NULL | ||
root_namespace_id | bigint | null | CHECK root_namespace_id IS NOT NULL |
Table checks: char_length(project_path) <= 64; project_id IS NOT NULL; organization_id IS NOT NULL; char_length(namespace_path) <= 64; root_namespace_id IS NOT NULL
project_security_exclusions references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
scanner | smallint | NOT NULL | |||
type | smallint | NOT NULL | |||
active | boolean | NOT NULL | TRUE | ||
description | text | null | CHECK char_length(description) <= 255 | ||
value | text | NOT NULL | CHECK char_length(value) <= 255 |
Table checks: char_length(description) <= 255; char_length(value) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | projects.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
auto_fix_container_scanning | boolean | NOT NULL | TRUE | ||
auto_fix_dast | boolean | NOT NULL | TRUE | ||
auto_fix_dependency_scanning | boolean | NOT NULL | TRUE | ||
auto_fix_sast | boolean | NOT NULL | TRUE | ||
continuous_vulnerability_scans_enabled | boolean | NOT NULL | FALSE | ||
container_scanning_for_registry_enabled | boolean | NOT NULL | FALSE | ||
secret_push_protection_enabled | boolean | null | FALSE | CHECK secret_push_protection_enabled IS NOT NULL | |
validity_checks_enabled | boolean | NOT NULL | FALSE | ||
license_configuration_source | smallint | NOT NULL | 0 | ||
cvs_for_container_scanning_enabled | boolean | NOT NULL | TRUE | ||
cvs_for_dependency_scanning_enabled | boolean | NOT NULL | TRUE | ||
license_scanning_for_cyclonedx_enabled | boolean | NOT NULL | TRUE | ||
fast_dependency_paths_enabled | boolean | NOT NULL | FALSE |
Table checks: secret_push_protection_enabled IS NOT NULL
project_security_statistics references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
vulnerability_count | integer | NOT NULL | 0 |
pipeline_execution_policy_bot_access_group_id — project_settings.pipeline_execution_policy_bot_access_group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.duo_dependency_bump_breaking_changes_enabled_by_id — project_settings.duo_dependency_bump_breaking_changes_enabled_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.push_rule_id — project_settings.push_rule_id may be NULL, so the relationship to push_rules is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.pipeline_execution_policy_bot_access_group_id, duo_dependency_bump_breaking_changes_enabled_by_id, push_rule_id — project_settings has several nullable FKs (pipeline_execution_policy_bot_access_group_id, duo_dependency_bump_breaking_changes_enabled_by_id, push_rule_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | projects.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
push_rule_id | bigint | null | push_rules.id 1:1 · SET NULL | ||
show_default_award_emojis | boolean | null | TRUE | CHECK show_default_award_emojis IS NOT NULL | |
allow_merge_on_skipped_pipeline | boolean | null | |||
squash_option | smallint | null | 3 | ||
has_confluence | boolean | NOT NULL | FALSE | ||
has_vulnerabilities | boolean | NOT NULL | FALSE | ||
prevent_merge_without_jira_issue | boolean | NOT NULL | FALSE | ||
cve_id_request_enabled | boolean | NOT NULL | TRUE | ||
mr_default_target_self | boolean | NOT NULL | FALSE | ||
previous_default_branch | text | null | CHECK char_length(previous_default_branch) <= 4096 | ||
warn_about_potentially_unwanted_characters | boolean | NOT NULL | TRUE | ||
merge_commit_template | text | null | CHECK char_length(merge_commit_template) <= 500 | ||
has_shimo | boolean | NOT NULL | FALSE | ||
squash_commit_template | text | null | CHECK char_length(squash_commit_template) <= 500 | ||
legacy_open_source_license_available | boolean | NOT NULL | TRUE | ||
target_platforms | varchar[] | NOT NULL | CAST('{}' AS varchar[]) | ||
enforce_auth_checks_on_uploads | boolean | NOT NULL | TRUE | ||
selective_code_owner_removals | boolean | NOT NULL | FALSE | ||
issue_branch_template | text | null | CHECK char_length(issue_branch_template) <= 255 | ||
show_diff_preview_in_email | boolean | NOT NULL | TRUE | ||
suggested_reviewers_enabled | boolean | NOT NULL | FALSE | ||
only_allow_merge_if_all_status_checks_passed | boolean | NOT NULL | FALSE | ||
mirror_branch_regex | text | null | CHECK char_length(mirror_branch_regex) <= 255 | ||
allow_pipeline_trigger_approve_deployment | boolean | NOT NULL | FALSE | ||
emails_enabled | boolean | NOT NULL | TRUE | ||
pages_unique_domain_enabled | boolean | NOT NULL | FALSE | ||
pages_unique_domain | text | null | CHECK char_length(pages_unique_domain) <= 63 | ||
runner_registration_enabled | boolean | null | TRUE | ||
product_analytics_instrumentation_key | text | null | CHECK char_length(product_analytics_instrumentation_key) <= 255 | ||
product_analytics_data_collector_host | text | null | CHECK char_length(product_analytics_data_collector_host) <= 255 | ||
cube_api_base_url | text | null | CHECK char_length(cube_api_base_url) <= 512 | ||
encrypted_cube_api_key | bytea | null | |||
encrypted_cube_api_key_iv | bytea | null | |||
encrypted_product_analytics_configurator_connection_string | bytea | null | |||
encrypted_product_analytics_configurator_connection_string_iv | bytea | null | |||
pages_multiple_versions_enabled | boolean | NOT NULL | FALSE | ||
allow_merge_without_pipeline | boolean | NOT NULL | FALSE | ||
duo_features_enabled | boolean | NOT NULL | TRUE | ||
require_reauthentication_to_approve | boolean | null | |||
observability_alerts_enabled | boolean | NOT NULL | TRUE | ||
spp_repository_pipeline_access | boolean | null | TRUE | ||
max_number_of_vulnerabilities | integer | null | |||
pages_primary_domain | text | null | CHECK char_length(pages_primary_domain) <= 255 | ||
extended_prat_expiry_webhooks_execute | boolean | NOT NULL | FALSE | ||
merge_request_title_regex | text | null | CHECK char_length(merge_request_title_regex) <= 255 | ||
protect_merge_request_pipelines | boolean | NOT NULL | TRUE | ||
auto_duo_code_review_enabled | boolean | null | |||
model_prompt_cache_enabled | boolean | null | |||
web_based_commit_signing_enabled | boolean | NOT NULL | FALSE | ||
duo_context_exclusion_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
merge_request_title_regex_description | text | null | CHECK char_length(merge_request_title_regex_description) <= 255 | ||
duo_remote_flows_enabled | boolean | null | |||
duo_foundational_flows_enabled | boolean | null | |||
duo_sast_fp_detection_enabled | boolean | NOT NULL | FALSE | ||
duo_sast_vr_workflow_enabled | boolean | NOT NULL | FALSE | ||
automatic_rebase_enabled | boolean | NOT NULL | FALSE | ||
duo_secret_detection_fp_enabled | boolean | NOT NULL | FALSE | ||
reviewer_assignment_strategy | smallint | NOT NULL | 0 | ||
pipeline_execution_policy_bot_access_enabled | boolean | NOT NULL | FALSE | ||
pipeline_execution_policy_bot_access_file_patterns | text[] | null | CAST('{}' AS text[]) | CHECK cardinality(pipeline_execution_policy_bot_access_file_patterns) <= 20 | |
pipeline_execution_policy_bot_access_group_id | bigint | null | namespaces.id 1:N · SET NULL | ||
security_policy_pipeline_must_succeed | boolean | NOT NULL | FALSE | ||
mr_default_title_template | text | null | CHECK char_length(mr_default_title_template) <= 100 | ||
tool_approval_for_session_enabled | boolean | null | |||
dap_session_tracking_enabled | boolean | NOT NULL | FALSE | ||
duo_dependency_bump_breaking_changes_enabled | boolean | NOT NULL | FALSE | ||
ai_audit_events_storage_enabled | boolean | NOT NULL | FALSE | ||
duo_dependency_bump_breaking_changes_enabled_by_id | bigint | null | users.id 1:N · SET NULL | ||
feature_flags_minimum_role | smallint | NOT NULL | 2 | ||
duo_vulnerability_context_analysis_enabled | boolean | NOT NULL | FALSE |
Table checks: char_length(pages_unique_domain) <= 63; char_length(merge_request_title_regex_description) <= 255; char_length(previous_default_branch) <= 4096; char_length(issue_branch_template) <= 255; char_length(product_analytics_data_collector_host) <= 255; char_length(mirror_branch_regex) <= 255; char_length(pages_primary_domain) <= 255; char_length(product_analytics_instrumentation_key) <= 255; char_length(squash_commit_template) <= 500; show_default_award_emojis IS NOT NULL; char_length(mr_default_title_template) <= 100; char_length(merge_commit_template) <= 500; char_length(merge_request_title_regex) <= 255; char_length(cube_api_base_url) <= 512; cardinality(pipeline_execution_policy_bot_access_file_patterns) <= 20
project_states references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
project_id | bigint | NOT NULL | |||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 |
Table checks: char_length(verification_failure) <= 255
root_namespace_id — project_statistics.root_namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.shared_runners_seconds_last_reset — shared_runners_seconds_last_reset stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:1 · CASCADE | ||
namespace_id | bigint | NOT NULL | |||
commit_count | bigint | NOT NULL | 0 | ||
storage_size | bigint | NOT NULL | 0 | ||
repository_size | bigint | NOT NULL | 0 | ||
lfs_objects_size | bigint | NOT NULL | 0 | ||
build_artifacts_size | bigint | NOT NULL | 0 | ||
shared_runners_seconds | bigint | NOT NULL | 0 | ||
shared_runners_seconds_last_reset | timestamp | null | |||
packages_size | bigint | NOT NULL | 0 | ||
wiki_size | bigint | null | |||
snippets_size | bigint | null | |||
pipeline_artifacts_size | bigint | NOT NULL | 0 | ||
uploads_size | bigint | NOT NULL | 0 | ||
container_registry_size | bigint | NOT NULL | 0 | ||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
root_namespace_id | bigint | null | namespaces.id 1:N · SET NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
security_attribute_id | bigint | NOT NULL | security_attributes.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
traversal_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
project_topic_upload_id | bigint | NOT NULL | project_topic_uploads.id 1:1 · CASCADE | ||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 |
Table checks: char_length(verification_failure) <= 255
model_type — model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.model_type, model_id — model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | nextval(CAST('uploads_id_seq' AS regclass)) | ||
size | bigint | NOT NULL | |||
model_id | bigint | NOT NULL | |||
uploaded_by_user_id | bigint | null | |||
organization_id | bigint | null | CHECK organization_id IS NOT NULL | ||
namespace_id | bigint | null | |||
project_id | bigint | null | |||
created_at | timestamp | null | |||
store | integer | NOT NULL | 1 | ||
version | integer | null | 1 | ||
path | text | NOT NULL | CHECK char_length(path) <= 511 | ||
checksum | text | null | CHECK char_length(checksum) <= 64 | ||
model_type PK | text | NOT NULL | |||
uploader | text | NOT NULL | |||
mount_point | text | null | |||
secret | text | null |
Table checks: char_length(path) <= 511; char_length(checksum) <= 64; organization_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
topic_id | bigint | NOT NULL | topics.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
runner_id | bigint | NOT NULL | project_type_ci_runners.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
contacted_at | timestamptz | null | |||
creation_state | smallint | NOT NULL | 0 | ||
executor_type | smallint | null | |||
runner_type PK | smallint | NOT NULL | project_type_ci_runners.id 1:N · CASCADE | ||
config | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
system_xid | text | NOT NULL | CHECK char_length(system_xid) <= 64 | ||
platform | text | null | CHECK char_length(platform) <= 255 | ||
architecture | text | null | CHECK char_length(architecture) <= 255 | ||
revision | text | null | CHECK char_length(revision) <= 255 | ||
ip_address | text | null | CHECK char_length(ip_address) <= 1024 | ||
version | text | null | CHECK char_length(version) <= 2048 | ||
runtime_features | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
organization_id | bigint | null | CHECK organization_id IS NOT NULL | ||
labels | jsonb | NOT NULL | CAST('{}' AS jsonb) |
Table checks: char_length(system_xid) <= 64; char_length(revision) <= 255; char_length(version) <= 2048; char_length(ip_address) <= 1024; char_length(platform) <= 255; char_length(architecture) <= 255; organization_id IS NOT NULL
public_projects_minutes_cost_factor — public_projects_minutes_cost_factor is binary floating point. 0.1 + 0.2 ≠ 0.3; sums drift; reconciliation against the ledger will be off by cents.private_projects_minutes_cost_factor — private_projects_minutes_cost_factor is binary floating point. 0.1 + 0.2 ≠ 0.3; sums drift; reconciliation against the ledger will be off by cents.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
creator_id | bigint | null | |||
created_at | timestamptz | null | |||
updated_at | timestamptz | null | |||
contacted_at | timestamptz | null | |||
token_expires_at | timestamptz | null | |||
public_projects_minutes_cost_factor | double precision | NOT NULL | 1.0 | ||
private_projects_minutes_cost_factor | double precision | NOT NULL | 1.0 | ||
access_level | integer | NOT NULL | 0 | ||
maximum_timeout | integer | null | |||
runner_type PK | smallint | NOT NULL | |||
registration_type | smallint | NOT NULL | 0 | ||
creation_state | smallint | NOT NULL | 0 | ||
active | boolean | NOT NULL | TRUE | ||
run_untagged | boolean | NOT NULL | TRUE | ||
locked | boolean | NOT NULL | FALSE | ||
name | text | null | CHECK char_length(name) <= 256 | ||
token_encrypted | text | null | CHECK char_length(token_encrypted) <= 512 | ||
description | text | null | CHECK char_length(description) <= 1024 | ||
maintainer_note | text | null | CHECK char_length(maintainer_note) <= 1024 | ||
allowed_plans | text[] | NOT NULL | CAST('{}' AS text[]) | ||
allowed_plan_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
organization_id | bigint | null | CHECK organization_id IS NOT NULL | ||
allowed_plan_name_uids | smallint[] | NOT NULL | CAST('{}' AS smallint[]) | ||
token_rotation_deadline | timestamptz | null | |||
uuid | uuid | null |
Table checks: char_length(token_encrypted) <= 512; char_length(name) <= 256; char_length(maintainer_note) <= 1024; char_length(description) <= 1024; organization_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
project_upload_id | bigint | NOT NULL | project_uploads.id 1:1 · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 |
Table checks: char_length(verification_failure) <= 255
model_type — model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.model_type, model_id — model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | nextval(CAST('uploads_id_seq' AS regclass)) | ||
size | bigint | NOT NULL | |||
model_id | bigint | NOT NULL | |||
uploaded_by_user_id | bigint | null | |||
organization_id | bigint | null | |||
namespace_id | bigint | null | |||
project_id | bigint | null | CHECK project_id IS NOT NULL | ||
created_at | timestamp | null | |||
store | integer | NOT NULL | 1 | ||
version | integer | null | 1 | ||
path | text | NOT NULL | CHECK char_length(path) <= 511 | ||
checksum | text | null | CHECK char_length(checksum) <= 64 | ||
model_type PK | text | NOT NULL | |||
uploader | text | NOT NULL | |||
mount_point | text | null | |||
secret | text | null |
Table checks: char_length(path) <= 511; char_length(checksum) <= 64; project_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
Tables whose names start with security_.
security_findings.scanner_id → vulnerability_scanners.id — one vulnerability_scanners, many security_findings · required · ON DELETE CASCADE · indexedsecurity_findings.scan_id → security_scans.id — one security_scans, many security_findings · required · ON DELETE CASCADE · indexedsecurity_attributes.security_category_id → security_categories.id — one security_categories, many security_attributes · required · ON DELETE CASCADE · indexedsecurity_finding_enrichments.vulnerability_id → vulnerabilities.id — one vulnerabilities, many security_finding_enrichments · optional · ON DELETE SET NULL · indexedsecurity_orchestration_policy_configurations.namespace_id → namespaces.id — one namespaces, many security_orchestration_policy_configurations · optional · ON DELETE CASCADE · indexedsecurity_orchestration_policy_configurations.project_id → projects.id — one projects, many security_orchestration_policy_configurations · optional · ON DELETE CASCADE · indexedsecurity_orchestration_policy_configurations.security_policy_management_project_id → projects.id — one projects, many security_orchestration_policy_configurations · required · ON DELETE CASCADE · indexedsecurity_orchestration_policy_rule_schedules.namespace_id → namespaces.id — one namespaces, many security_orchestration_policy_rule_schedules · optional · ON DELETE CASCADE · indexedsecurity_orchestration_policy_rule_schedules.project_id → projects.id — one projects, many security_orchestration_policy_rule_schedules · optional · ON DELETE CASCADE · indexedsecurity_orchestration_policy_rule_schedules.security_orchestration_policy_configuration_id → security_orchestration_policy_configurations.id — one security_orchestration_policy_configurations, many security_orchestration_policy_rule_schedules · required · ON DELETE CASCADE · indexedsecurity_orchestration_policy_rule_schedules.user_id → users.id — one users, many security_orchestration_policy_rule_schedules · optional · ON DELETE CASCADE · indexedsecurity_pipeline_execution_policy_config_links.security_policy_id → security_policies.id — one security_policies, many security_pipeline_execution_policy_config_links · required · ON DELETE CASCADE · indexedsecurity_pipeline_execution_policy_config_links.project_id → projects.id — one projects, many security_pipeline_execution_policy_config_links · required · ON DELETE CASCADE · indexedsecurity_pipeline_execution_project_schedules.project_id → projects.id — one projects, many security_pipeline_execution_project_schedules · required · ON DELETE CASCADE · indexedsecurity_pipeline_execution_project_schedules.security_policy_id → security_policies.id — one security_policies, many security_pipeline_execution_project_schedules · required · ON DELETE CASCADE · indexedsecurity_policies.security_policy_management_project_id → projects.id — one projects, many security_policies · required · ON DELETE CASCADE · indexedsecurity_policies.security_orchestration_policy_configuration_id → security_orchestration_policy_configurations.id — one security_orchestration_policy_configurations, many security_policies · required · ON DELETE CASCADE · indexedsecurity_policy_dismissals.project_id → projects.id — one projects, many security_policy_dismissals · required · ON DELETE CASCADE · indexedsecurity_policy_dismissals.merge_request_id → merge_requests.id — one merge_requests, many security_policy_dismissals · required · ON DELETE CASCADE · indexedsecurity_policy_dismissals.security_policy_id → security_policies.id — one security_policies, many security_policy_dismissals · optional · ON DELETE SET NULL · indexedsecurity_policy_dismissals.user_id → users.id — one users, many security_policy_dismissals · optional · ON DELETE SET NULL · indexedsecurity_policy_project_links.project_id → projects.id — one projects, many security_policy_project_links · required · ON DELETE CASCADE · indexedsecurity_policy_project_links.security_policy_id → security_policies.id — one security_policies, many security_policy_project_links · required · ON DELETE CASCADE · indexedsecurity_policy_requirements.namespace_id → namespaces.id — one namespaces, many security_policy_requirements · required · ON DELETE CASCADE · indexedsecurity_policy_requirements.compliance_requirement_id → compliance_requirements.id — one compliance_requirements, many security_policy_requirements · required · ON DELETE CASCADE · indexedsecurity_policy_requirements.compliance_framework_security_policy_id → compliance_framework_security_policies.id — one compliance_framework_security_policies, many security_policy_requirements · required · ON DELETE CASCADE · indexedsecurity_policy_schedule_pipelines.project_id → projects.id — one projects, many security_policy_schedule_pipelines · required · ON DELETE CASCADE · indexedsecurity_policy_schedule_pipelines.security_policy_id → security_policies.id — one security_policies, many security_policy_schedule_pipelines · required · ON DELETE CASCADE · indexedsecurity_policy_settings.organization_id → organizations.id — one organizations, at most one security_policy_settings · required · ON DELETE CASCADE · indexedsecurity_policy_settings.csp_namespace_id → namespaces.id — one namespaces, many security_policy_settings · optional · ON DELETE SET NULL · indexedsecurity_scan_execution_project_schedules.project_id → projects.id — one projects, many security_scan_execution_project_schedules · required · ON DELETE CASCADE · indexedsecurity_scan_execution_project_schedules.policy_rule_schedule_id → security_orchestration_policy_rule_schedules.id — one security_orchestration_policy_rule_schedules, many security_scan_execution_project_schedules · required · ON DELETE CASCADE · indexedsecurity_scan_execution_project_schedules.security_policy_id → security_policies.id — one security_policies, many security_scan_execution_project_schedules · required · ON DELETE CASCADE · indexedsecurity_scan_profile_configurations.security_scan_profile_id → security_scan_profiles.id — one security_scan_profiles, many security_scan_profile_configurations · required · ON DELETE CASCADE · indexedsecurity_scan_profile_project_statuses.security_scan_profile_id → security_scan_profiles.id — one security_scan_profiles, many security_scan_profile_project_statuses · required · ON DELETE CASCADE · indexedsecurity_scan_profile_triggers.security_scan_profile_configuration_id → security_scan_profile_configurations.id — one security_scan_profile_configurations, many security_scan_profile_triggers · optional · ON DELETE SET NULL · indexedsecurity_scan_profile_triggers.security_scan_profile_id → security_scan_profiles.id — one security_scan_profiles, many security_scan_profile_triggers · required · ON DELETE CASCADE · indexedsecurity_scan_profiles_projects.security_scan_profile_id → security_scan_profiles.id — one security_scan_profiles, many security_scan_profiles_projects · required · ON DELETE CASCADE · indexedsecurity_scheduled_pipeline_execution_policy_test_runs.security_policy_id → security_policies.id — one security_policies, many security_scheduled_pipeline_execution_policy_test_runs · required · ON DELETE CASCADE · indexedsecurity_scheduled_pipeline_execution_policy_test_runs.project_id → projects.id — one projects, many security_scheduled_pipeline_execution_policy_test_runs · required · ON DELETE CASCADE · indexedsecurity_trainings.project_id → projects.id — one projects, many security_trainings · required · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
scan_id | bigint | NOT NULL | security_scans.id 1:N · CASCADE | ||
scanner_id | bigint | NOT NULL | vulnerability_scanners.id 1:N · CASCADE | ||
severity | smallint | NOT NULL | |||
deduplicated | boolean | NOT NULL | FALSE | ||
uuid | uuid | null | CHECK uuid IS NOT NULL | ||
overridden_uuid | uuid | null | |||
partition_number PK | integer | NOT NULL | 1 | ||
finding_data | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
project_id | bigint | null | CHECK project_id IS NOT NULL | ||
scanner_reported_severity | smallint | null | |||
context_unaware_uuid | uuid | null |
Table checks: uuid IS NOT NULL; project_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | |||
security_category_id | bigint | NOT NULL | security_categories.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
editable_state | smallint | NOT NULL | 0 | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
description | text | null | CHECK char_length(description) <= 255 | ||
color | text | NOT NULL | CHECK char_length(color) <= 7 | ||
template_type | smallint | null | |||
deleted_at | timestamptz | null |
Table checks: char_length(color) <= 7; char_length(description) <= 255; char_length(name) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
editable_state | smallint | NOT NULL | 0 | ||
template_type | smallint | null | |||
multiple_selection | boolean | NOT NULL | FALSE | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
description | text | null | CHECK char_length(description) <= 255 | ||
deleted_at | timestamptz | null |
Table checks: char_length(name) <= 255; char_length(description) <= 255
vulnerability_id — security_finding_enrichments.vulnerability_id may be NULL, so the relationship to vulnerabilities is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
finding_uuid | uuid | NOT NULL | |||
project_id | bigint | NOT NULL | |||
cve_enrichment_id | bigint | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
cve | text | NOT NULL | CHECK char_length(cve) <= 24 | ||
epss_score | double precision | null | |||
is_known_exploit | boolean | null | |||
vulnerability_id | bigint | null | vulnerabilities.id 1:N · SET NULL |
Table checks: char_length(cve) <= 24
security_finding_token_statuses references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
security_finding_id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
raw_source_code_extract | text | null | CHECK char_length(raw_source_code_extract) <= 2048 | ||
last_verified_at | timestamptz | null |
Table checks: char_length(raw_source_code_extract) <= 2048
security_inventory_filters references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
archived | boolean | NOT NULL | FALSE | ||
sast | smallint | NOT NULL | 0 | ||
sast_advanced | smallint | NOT NULL | 0 | ||
sast_iac | smallint | NOT NULL | 0 | ||
dast | smallint | NOT NULL | 0 | ||
dependency_scanning | smallint | NOT NULL | 0 | ||
coverage_fuzzing | smallint | NOT NULL | 0 | ||
api_fuzzing | smallint | NOT NULL | 0 | ||
cluster_image_scanning | smallint | NOT NULL | 0 | ||
secret_detection_secret_push_protection | smallint | NOT NULL | 0 | ||
container_scanning_for_registry | smallint | NOT NULL | 0 | ||
secret_detection_pipeline_based | smallint | NOT NULL | 0 | ||
container_scanning_pipeline_based | smallint | NOT NULL | 0 | ||
secret_detection | smallint | NOT NULL | 0 | ||
container_scanning | smallint | NOT NULL | 0 | ||
total | integer | NOT NULL | 0 | ||
critical | integer | NOT NULL | 0 | ||
high | integer | NOT NULL | 0 | ||
medium | integer | NOT NULL | 0 | ||
low | integer | NOT NULL | 0 | ||
info | integer | NOT NULL | 0 | ||
unknown | integer | NOT NULL | 0 | ||
project_id | bigint | NOT NULL | |||
traversal_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
project_name | text | NOT NULL | CHECK char_length(project_name) <= 255 | ||
has_scanners | boolean | NOT NULL | FALSE | ||
has_failed_or_warning | boolean | NOT NULL | FALSE | ||
has_stale | boolean | NOT NULL | FALSE |
Table checks: char_length(project_name) <= 255
{"owner":"group::container security","description":"Configuration used to store relationship between project and security policy repository"}
namespace_id — security_orchestration_policy_configurations.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — security_orchestration_policy_configurations.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
security_policy_management_project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
configured_at | timestamptz | null | |||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
experiments | jsonb | NOT NULL | CAST('{}' AS jsonb) |
Table checks: project_id IS NULL <> namespace_id IS NULL
{"owner":"group::container security","description":"Schedules used to store relationship between project and security policy repository"}
namespace_id — security_orchestration_policy_rule_schedules.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — security_orchestration_policy_rule_schedules.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id — security_orchestration_policy_rule_schedules.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id, project_id, user_id — security_orchestration_policy_rule_schedules has several nullable FKs (namespace_id, project_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
next_run_at | timestamptz | null | |||
security_orchestration_policy_configuration_id | bigint | NOT NULL | security_orchestration_policy_configurations.id 1:N · CASCADE | ||
user_id | bigint | null | users.id 1:N · CASCADE | ||
policy_index | integer | NOT NULL | |||
cron | text | NOT NULL | CHECK char_length(cron) <= 255 | ||
rule_index | integer | NOT NULL | 0 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
policy_type | smallint | NOT NULL | 0 |
Table checks: char_length(cron) <= 255; num_nonnulls(namespace_id, project_id) = 1
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
security_policy_id | bigint | NOT NULL | security_policies.id 1:N · CASCADE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
next_run_at | timestamptz | NOT NULL | |||
security_policy_id | bigint | NOT NULL | security_policies.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
time_window_seconds | integer | NOT NULL | CHECK time_window_seconds > 0 | ||
cron | text | NOT NULL | CHECK char_length(cron) <= 128 | ||
cron_timezone | text | NOT NULL | CHECK char_length(cron_timezone) <= 255 | ||
snoozed_until | timestamptz | null | |||
next_run_applied_delay | integer | null |
Table checks: char_length(cron_timezone) <= 255; char_length(cron) <= 128; time_window_seconds > 0
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
security_orchestration_policy_configuration_id | bigint | NOT NULL | security_orchestration_policy_configurations.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
policy_index | smallint | NOT NULL | |||
type | smallint | NOT NULL | |||
enabled | boolean | NOT NULL | TRUE | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
description | text | null | CHECK char_length(description) <= 1000000 | ||
checksum | text | NOT NULL | CHECK char_length(checksum) <= 255 | ||
scope | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
security_policy_management_project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
content | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
metadata | jsonb | NOT NULL | CAST('{}' AS jsonb) |
Table checks: char_length(name) <= 255; char_length(checksum) <= 255; char_length(description) <= 1000000
security_policy_id — security_policy_dismissals.security_policy_id may be NULL, so the relationship to security_policies is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id — security_policy_dismissals.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.security_policy_id, user_id — security_policy_dismissals has several nullable FKs (security_policy_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
merge_request_id | bigint | NOT NULL | merge_requests.id 1:N · CASCADE | ||
security_policy_id | bigint | null | security_policies.id 1:N · SET NULL | ||
user_id | bigint | null | users.id 1:N · SET NULL | ||
security_findings_uuids | text[] | null | CAST('{}' AS text[]) | CHECK security_findings_uuids IS NOT NULL | |
dismissal_types | smallint[] | NOT NULL | CAST('{}' AS smallint[]) | ||
comment | text | null | CHECK char_length(comment) <= 255 | ||
status | smallint | NOT NULL | 0 | ||
licenses | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
license_occurrence_uuids | text[] | NOT NULL | CAST('{}' AS text[]) | CHECK cardinality(license_occurrence_uuids) <= 1000 |
Table checks: char_length(comment) <= 255; security_findings_uuids IS NOT NULL; cardinality(license_occurrence_uuids) <= 1000
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
security_policy_id | bigint | NOT NULL | security_policies.id 1:N · CASCADE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
compliance_framework_security_policy_id | bigint | NOT NULL | compliance_framework_security_policies.id 1:N · CASCADE | ||
compliance_requirement_id | bigint | NOT NULL | compliance_requirements.id 1:N · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE |
project_id, security_policy_id — security_policy_schedule_pipelines looks like a many-to-many link table but has no unique constraint across (project_id, security_policy_id). The same pair can be inserted twice; every join through it will double-count.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
security_policy_id | bigint | NOT NULL | security_policies.id 1:N · CASCADE | ||
pipeline_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
csp_namespace_id — security_policy_settings.csp_namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
csp_namespace_id | bigint | null | namespaces.id 1:N · SET NULL | ||
organization_id | bigint | NOT NULL | organizations.id 1:1 · CASCADE | ||
csp_namespace_locked_until | timestamptz | null |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
context_type | smallint | NOT NULL | 1 | ||
state | smallint | NOT NULL | 1 | ||
is_default | boolean | NOT NULL | |||
context_name | text | NOT NULL | CHECK char_length(context_name) <= 1024 | ||
traversal_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
uuid_version | smallint | NOT NULL | 1 |
Table checks: char_length(context_name) <= 1024
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
policy_rule_schedule_id | bigint | NOT NULL | security_orchestration_policy_rule_schedules.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
security_policy_id | bigint | NOT NULL | security_policies.id 1:N · CASCADE | ||
next_run_at | timestamptz | NOT NULL | |||
next_run_applied_delay | integer | NOT NULL | 0 |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
security_scan_profile_id | bigint | NOT NULL | security_scan_profiles.id 1:N · CASCADE | ||
namespace_id | bigint | NOT NULL | |||
configuration_version | smallint | NOT NULL | 1 | ||
configuration | jsonb | NOT NULL | CAST('{}' AS jsonb) |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
security_scan_profile_id | bigint | NOT NULL | security_scan_profiles.id 1:N · CASCADE | ||
build_id | bigint | null | |||
last_scan_at | timestamptz | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
consecutive_failure_count | smallint | NOT NULL | 0 | ||
consecutive_success_count | smallint | NOT NULL | 0 |
security_scan_profile_configuration_id — security_scan_profile_triggers.security_scan_profile_configuration_id may be NULL, so the relationship to security_scan_profile_configurations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
security_scan_profile_id | bigint | NOT NULL | security_scan_profiles.id 1:N · CASCADE | ||
namespace_id | bigint | NOT NULL | |||
trigger_type | smallint | NOT NULL | |||
security_scan_profile_configuration_id | bigint | null | security_scan_profile_configurations.id 1:N · SET NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
scan_type | smallint | NOT NULL | |||
gitlab_recommended | boolean | NOT NULL | FALSE | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
description | text | null | CHECK char_length(description) <= 2047 | ||
configuration | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
deleted_at | timestamptz | null |
Table checks: char_length(description) <= 2047; char_length(name) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | |||
security_scan_profile_id | bigint | NOT NULL | security_scan_profiles.id 1:N · CASCADE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
build_id | bigint | NOT NULL | |||
scan_type | smallint | NOT NULL | |||
info | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
project_id | bigint | null | CHECK project_id IS NOT NULL | ||
pipeline_id | bigint | null | |||
latest | boolean | NOT NULL | TRUE | ||
status | smallint | NOT NULL | 0 | ||
findings_partition_number | integer | NOT NULL | 1 | ||
scanner_external_id | text | null | CHECK char_length(scanner_external_id) <= 255 |
Table checks: project_id IS NOT NULL; char_length(scanner_external_id) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
security_policy_id | bigint | NOT NULL | security_policies.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
pipeline_id | bigint | null | |||
error_message | text | null | CHECK char_length(error_message) <= 255 | ||
state | smallint | NOT NULL | 0 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
Table checks: char_length(error_message) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
provider_id | bigint | null | |||
is_primary | boolean | NOT NULL | FALSE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
training_provider_id | bigint | NOT NULL | 0 |
Tables whose names start with user_.
user_achievements.namespace_id → namespaces.id — one namespaces, many user_achievements · optional · ON DELETE CASCADE · indexeduser_achievements.user_id → users.id — one users, many user_achievements · required · ON DELETE CASCADE · indexeduser_achievements.awarded_by_user_id → users.id — one users, many user_achievements · optional · ON DELETE SET NULL · indexeduser_achievements.revoked_by_user_id → users.id — one users, many user_achievements · optional · ON DELETE SET NULL · indexeduser_achievements.achievement_id → achievements.id — one achievements, many user_achievements · required · ON DELETE CASCADE · indexeduser_admin_roles.user_id → users.id — one users, at most one user_admin_roles · required · ON DELETE CASCADE · indexeduser_admin_roles.admin_role_id → admin_roles.id — one admin_roles, many user_admin_roles · required · ON DELETE CASCADE · indexeduser_agent_details.organization_id → organizations.id — one organizations, many user_agent_details · optional · ON DELETE CASCADE · indexeduser_broadcast_message_dismissals.user_id → users.id — one users, many user_broadcast_message_dismissals · required · ON DELETE CASCADE · indexeduser_broadcast_message_dismissals.broadcast_message_id → broadcast_messages.id — one broadcast_messages, many user_broadcast_message_dismissals · required · ON DELETE CASCADE · indexeduser_callouts.user_id → users.id — one users, many user_callouts · required · ON DELETE CASCADE · indexeduser_credit_card_validations.user_id → users.id — one users, at most one user_credit_card_validations · required · ON DELETE CASCADE · indexeduser_custom_attributes.user_id → users.id — one users, many user_custom_attributes · required · ON DELETE CASCADE · indexeduser_details.bot_namespace_id → namespaces.id — one namespaces, many user_details · optional · ON DELETE SET NULL · indexeduser_details.user_id → users.id — one users, at most one user_details · required · ON DELETE CASCADE · indexeduser_follow_users.followee_id → users.id — one users, many user_follow_users · required · ON DELETE CASCADE · indexeduser_follow_users.follower_id → users.id — one users, many user_follow_users · required · ON DELETE CASCADE · indexeduser_group_callouts.group_id → namespaces.id — one namespaces, many user_group_callouts · required · ON DELETE CASCADE · indexeduser_group_callouts.user_id → users.id — one users, many user_group_callouts · required · ON DELETE CASCADE · indexeduser_group_member_roles.member_role_id → member_roles.id — one member_roles, many user_group_member_roles · required · ON DELETE CASCADE · indexeduser_group_member_roles.user_id → users.id — one users, many user_group_member_roles · required · ON DELETE CASCADE · indexeduser_group_member_roles.group_id → namespaces.id — one namespaces, many user_group_member_roles · required · ON DELETE CASCADE · indexeduser_group_member_roles.shared_with_group_id → namespaces.id — one namespaces, many user_group_member_roles · optional · ON DELETE CASCADE · indexeduser_highest_roles.user_id → users.id — one users, at most one user_highest_roles · required · ON DELETE CASCADE · indexeduser_member_roles.user_id → users.id — one users, at most one user_member_roles · required · ON DELETE CASCADE · indexeduser_member_roles.member_role_id → member_roles.id — one member_roles, many user_member_roles · required · ON DELETE CASCADE · indexeduser_namespace_callouts.namespace_id → namespaces.id — one namespaces, many user_namespace_callouts · required · ON DELETE CASCADE · indexeduser_namespace_callouts.user_id → users.id — one users, many user_namespace_callouts · required · ON DELETE CASCADE · indexeduser_permission_export_upload_upload_states.user_permission_export_upload_upload_id → user_permission_export_upload_uploads.id — one user_permission_export_upload_uploads, at most one user_permission_export_upload_upload_states · required · ON DELETE CASCADE · indexeduser_permission_export_upload_upload_states.uploaded_by_user_id → users.id — one users, many user_permission_export_upload_upload_states · required · ON DELETE SET NULL · indexeduser_permission_export_uploads.user_id → users.id — one users, many user_permission_export_uploads · required · ON DELETE CASCADE · indexeduser_phone_number_validations.user_id → users.id — one users, at most one user_phone_number_validations · required · ON DELETE CASCADE · indexeduser_preferences.default_duo_add_on_assignment_id → subscription_user_add_on_assignments.id — one subscription_user_add_on_assignments, at most one user_preferences · optional · ON DELETE SET NULL · indexeduser_preferences.duo_default_namespace_id → namespaces.id — one namespaces, many user_preferences · optional · ON DELETE SET NULL · indexeduser_preferences.knowledge_graph_governing_namespace_id → namespaces.id — one namespaces, many user_preferences · optional · ON DELETE SET NULL · indexeduser_preferences.user_id → users.id — one users, at most one user_preferences · required · ON DELETE CASCADE · indexeduser_project_callouts.project_id → projects.id — one projects, many user_project_callouts · required · ON DELETE CASCADE · indexeduser_project_callouts.user_id → users.id — one users, many user_project_callouts · required · ON DELETE CASCADE · indexeduser_project_member_roles.project_id → projects.id — one projects, many user_project_member_roles · required · ON DELETE CASCADE · indexeduser_project_member_roles.member_role_id → member_roles.id — one member_roles, many user_project_member_roles · required · ON DELETE CASCADE · indexeduser_project_member_roles.shared_with_group_id → namespaces.id — one namespaces, many user_project_member_roles · optional · ON DELETE CASCADE · indexeduser_project_member_roles.user_id → users.id — one users, many user_project_member_roles · required · ON DELETE CASCADE · indexeduser_saved_views.namespace_id → namespaces.id — one namespaces, many user_saved_views · required · ON DELETE CASCADE · indexeduser_saved_views.saved_view_id → saved_views.id — one saved_views, many user_saved_views · required · ON DELETE CASCADE · indexeduser_saved_views.user_id → users.id — one users, many user_saved_views · required · ON DELETE CASCADE · indexeduser_statuses.user_id → users.id — one users, at most one user_statuses · required · ON DELETE CASCADE · indexeduser_synced_attributes_metadata.user_id → users.id — one users, at most one user_synced_attributes_metadata · required · ON DELETE CASCADE · indexeduser_upload_states.organization_id → organizations.id — one organizations, many user_upload_states · required · ON DELETE CASCADE · indexeduser_upload_states.user_upload_id → user_uploads.id — one user_uploads, at most one user_upload_states · required · ON DELETE CASCADE · indexedtarget_type, target_id — target_id points at different tables depending on target_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.user_audit_events references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | nextval(CAST('audit_events_id_seq' AS regclass)) | ||
created_at PK | timestamptz | NOT NULL | |||
user_id | bigint | NOT NULL | |||
author_id | bigint | NOT NULL | |||
target_id | bigint | null | |||
event_name | text | null | CHECK char_length(event_name) <= 255 | ||
details | text | null | |||
ip_address | inet | null | |||
author_name | text | null | CHECK char_length(author_name) <= 255 | ||
entity_path | text | null | CHECK char_length(entity_path) <= 5500 | ||
target_details | text | null | CHECK char_length(target_details) <= 5500 | ||
target_type | text | null | CHECK char_length(target_type) <= 255 |
Table checks: char_length(author_name) <= 255; char_length(entity_path) <= 5500; char_length(event_name) <= 255; char_length(target_details) <= 5500; char_length(target_type) <= 255
namespace_id — user_achievements.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.awarded_by_user_id — user_achievements.awarded_by_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.revoked_by_user_id — user_achievements.revoked_by_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id, awarded_by_user_id, revoked_by_user_id — user_achievements has several nullable FKs (namespace_id, awarded_by_user_id, revoked_by_user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
achievement_id | bigint | NOT NULL | achievements.id 1:N · CASCADE | ||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
awarded_by_user_id | bigint | null | users.id 1:N · SET NULL | ||
revoked_by_user_id | bigint | null | users.id 1:N · SET NULL | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
revoked_at | timestamptz | null | |||
priority | integer | null | |||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL | |
show_on_profile | boolean | NOT NULL | FALSE | ||
award_message | text | null | CHECK char_length(award_message) <= 200 | ||
award_message_html | text | null | |||
cached_markdown_version | integer | null |
Table checks: namespace_id IS NOT NULL; char_length(award_message) <= 200
admin_role_id, user_id — user_admin_roles looks like a many-to-many link table but has no unique constraint across (admin_role_id, user_id). The same pair can be inserted twice; every join through it will double-count.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
user_id PK | bigint | NOT NULL | users.id 1:1 · CASCADE | ||
admin_role_id | bigint | NOT NULL | admin_roles.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
ldap | boolean | NOT NULL | FALSE |
organization_id — user_agent_details.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.subject_type — subject_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.subject_type, subject_id — subject_id points at different tables depending on subject_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_agent | varchar | NOT NULL | |||
ip_address | varchar | NOT NULL | |||
subject_id | bigint | NOT NULL | |||
subject_type | varchar | NOT NULL | |||
submitted | boolean | NOT NULL | FALSE | ||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL | |||
organization_id | bigint | null | organizations.id 1:N · CASCADE | CHECK organization_id IS NOT NULL |
Table checks: organization_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
broadcast_message_id | bigint | NOT NULL | broadcast_messages.id 1:N · CASCADE | ||
expires_at | timestamptz | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
feature_name | integer | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
dismissed_at | timestamptz | null |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
user_id PK | bigint | NOT NULL | users.id 1:1 · CASCADE | ||
credit_card_validated_at | timestamptz | NOT NULL | |||
last_digits_hash | text | null | CHECK char_length(last_digits_hash) <= 44 | ||
holder_name_hash | text | null | CHECK char_length(holder_name_hash) <= 44 | ||
expiration_date_hash | text | null | CHECK char_length(expiration_date_hash) <= 44 | ||
network_hash | text | null | CHECK char_length(network_hash) <= 44 | ||
zuora_payment_method_xid | text | null | CHECK char_length(zuora_payment_method_xid) <= 50 | ||
stripe_setup_intent_xid | text | null | CHECK char_length(stripe_setup_intent_xid) <= 255 | ||
stripe_payment_method_xid | text | null | CHECK char_length(stripe_payment_method_xid) <= 255 | ||
stripe_card_fingerprint | text | null | CHECK char_length(stripe_card_fingerprint) <= 255 |
Table checks: char_length(stripe_payment_method_xid) <= 255; char_length(stripe_card_fingerprint) <= 255; char_length(stripe_setup_intent_xid) <= 255; char_length(network_hash) <= 44; char_length(expiration_date_hash) <= 44; char_length(zuora_payment_method_xid) <= 50; char_length(holder_name_hash) <= 44; char_length(last_digits_hash) <= 44
created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
key | varchar | NOT NULL | |||
value | varchar | NOT NULL |
bot_namespace_id — user_details.bot_namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
user_id PK | bigint | NOT NULL | users.id 1:1 · CASCADE | ||
job_title | varchar(200) | NOT NULL | CAST('' AS varchar) | ||
bio | varchar(255) | NOT NULL | CAST('' AS varchar) | ||
webauthn_xid | text | null | CHECK char_length(webauthn_xid) <= 100 | ||
provisioned_by_group_id | bigint | null | |||
pronouns | text | null | CHECK char_length(pronouns) <= 50 | ||
pronunciation | text | null | CHECK char_length(pronunciation) <= 255 | ||
phone | text | null | JiHu-specific columnCHECK char_length(phone) <= 50 | ||
linkedin | text | NOT NULL | CAST('' AS text) | CHECK char_length(linkedin) <= 500 | |
twitter | text | NOT NULL | CAST('' AS text) | CHECK char_length(twitter) <= 500 | |
website_url | text | NOT NULL | CAST('' AS text) | CHECK char_length(website_url) <= 500 | |
location | text | NOT NULL | CAST('' AS text) | CHECK char_length(location) <= 500 | |
password_last_changed_at | timestamptz | NOT NULL | now() | JiHu-specific column | |
discord | text | NOT NULL | CAST('' AS text) | CHECK char_length(discord) <= 500 | |
enterprise_group_id | bigint | null | |||
enterprise_group_associated_at | timestamptz | null | |||
email_reset_offered_at | timestamptz | null | |||
mastodon | text | NOT NULL | CAST('' AS text) | CHECK char_length(mastodon) <= 500 | |
project_authorizations_recalculated_at | timestamptz | NOT NULL | CAST('2010-01-01 00:00:00+00' AS timestamptz) | ||
onboarding_status | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
bluesky | text | NOT NULL | CAST('' AS text) | CHECK char_length(bluesky) <= 256 | |
bot_namespace_id | bigint | null | namespaces.id 1:N · SET NULL | ||
orcid | text | NOT NULL | CAST('' AS text) | CHECK char_length(orcid) <= 256 | |
github | text | NOT NULL | CAST('' AS text) | CHECK char_length(github) <= 500 | |
email_otp | text | null | SHA256 hash (64 hex characters)CHECK char_length(email_otp) <= 64 | ||
email_otp_last_sent_to | text | null | CHECK char_length(email_otp_last_sent_to) <= 511 | ||
email_otp_last_sent_at | timestamptz | null | |||
email_otp_required_after | timestamptz | null | |||
company | text | NOT NULL | CAST('' AS text) | CHECK char_length(company) <= 500 | |
provisioned_by_project_id | bigint | null |
Table checks: char_length(bluesky) <= 256; char_length(webauthn_xid) <= 100; char_length(twitter) <= 500; char_length(email_otp_last_sent_to) <= 511; char_length(discord) <= 500; char_length(linkedin) <= 500; char_length(website_url) <= 500; char_length(location) <= 500; char_length(orcid) <= 256; char_length(phone) <= 50; char_length(github) <= 500; char_length(email_otp) <= 64; char_length(pronouns) <= 50; char_length(mastodon) <= 500; char_length(pronunciation) <= 255; num_nonnulls(provisioned_by_group_id, provisioned_by_project_id) <= 1; char_length(company) <= 500
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
follower_id PK | bigint | NOT NULL | users.id 1:N · CASCADE | ||
followee_id PK | bigint | NOT NULL | users.id 1:N · CASCADE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
feature_name | smallint | NOT NULL | |||
dismissed_at | timestamptz | null |
shared_with_group_id — user_group_member_roles.shared_with_group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
shared_with_group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
member_role_id | bigint | NOT NULL | member_roles.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
user_id PK | bigint | NOT NULL | users.id 1:1 · CASCADE | ||
updated_at | timestamptz | NOT NULL | |||
highest_access_level | integer | null |
member_role_id, user_id — user_member_roles looks like a many-to-many link table but has no unique constraint across (member_role_id, user_id). The same pair can be inserted twice; every join through it will double-count.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:1 · CASCADE | ||
member_role_id | bigint | NOT NULL | member_roles.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
ldap | boolean | NOT NULL | FALSE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
dismissed_at | timestamptz | null | |||
feature_name | smallint | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
user_permission_export_upload_upload_id | bigint | NOT NULL | user_permission_export_upload_uploads.id 1:1 · CASCADE | ||
uploaded_by_user_id | bigint | NOT NULL | users.id 1:N · SET NULL | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 |
Table checks: char_length(verification_failure) <= 255
model_type — model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.model_type, model_id — model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | nextval(CAST('uploads_id_seq' AS regclass)) | ||
size | bigint | NOT NULL | |||
model_id | bigint | NOT NULL | |||
uploaded_by_user_id | bigint | null | CHECK uploaded_by_user_id IS NOT NULL | ||
organization_id | bigint | null | |||
namespace_id | bigint | null | |||
project_id | bigint | null | |||
created_at | timestamp | null | |||
store | integer | NOT NULL | 1 | ||
version | integer | null | 1 | ||
path | text | NOT NULL | CHECK char_length(path) <= 511 | ||
checksum | text | null | CHECK char_length(checksum) <= 64 | ||
model_type PK | text | NOT NULL | |||
uploader | text | NOT NULL | |||
mount_point | text | null | |||
secret | text | null |
Table checks: char_length(path) <= 511; uploaded_by_user_id IS NOT NULL; char_length(checksum) <= 64
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
file_store | integer | null | |||
status | smallint | NOT NULL | 0 | ||
file | text | null | CHECK char_length(file) <= 255 |
Table checks: char_length(file) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
user_id PK | bigint | NOT NULL | users.id 1:1 · CASCADE | ||
validated_at | timestamptz | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
international_dial_code | smallint | NOT NULL | |||
verification_attempts | smallint | NOT NULL | 0 | ||
risk_score | smallint | NOT NULL | 0 | ||
country | text | NOT NULL | CHECK char_length(country) <= 3 | ||
phone_number | text | NOT NULL | CHECK char_length(phone_number) <= 12 | ||
telesign_reference_xid | text | null | CHECK char_length(telesign_reference_xid) <= 255 | ||
sms_sent_at | timestamptz | null | |||
sms_send_count | smallint | NOT NULL | 0 |
Table checks: char_length(country) <= 3; char_length(phone_number) <= 12; char_length(telesign_reference_xid) <= 255
default_duo_add_on_assignment_id — user_preferences.default_duo_add_on_assignment_id may be NULL, so the relationship to subscription_user_add_on_assignments is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.duo_default_namespace_id — user_preferences.duo_default_namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.knowledge_graph_governing_namespace_id — user_preferences.knowledge_graph_governing_namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.default_duo_add_on_assignment_id, duo_default_namespace_id, knowledge_graph_governing_namespace_id — user_preferences has several nullable FKs (default_duo_add_on_assignment_id, duo_default_namespace_id, knowledge_graph_governing_namespace_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:1 · CASCADE | ||
issue_notes_filter | smallint | NOT NULL | 0 | ||
merge_request_notes_filter | smallint | NOT NULL | 0 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
epics_sort | varchar | null | |||
roadmap_epics_state | integer | null | |||
epic_notes_filter | smallint | NOT NULL | 0 | ||
issues_sort | varchar | null | |||
merge_requests_sort | varchar | null | |||
roadmaps_sort | varchar | null | |||
first_day_of_week | integer | null | |||
timezone | varchar | null | |||
time_display_relative | boolean | null | TRUE | CHECK time_display_relative IS NOT NULL | |
projects_sort | varchar(64) | null | |||
show_whitespace_in_diffs | boolean | NOT NULL | TRUE | ||
sourcegraph_enabled | boolean | null | |||
setup_for_company | boolean | null | |||
render_whitespace_in_code | boolean | null | FALSE | CHECK render_whitespace_in_code IS NOT NULL | |
tab_width | smallint | null | 8 | CHECK tab_width IS NOT NULL | |
view_diffs_file_by_file | boolean | NOT NULL | FALSE | ||
gitpod_enabled | boolean | NOT NULL | FALSE | ||
markdown_surround_selection | boolean | NOT NULL | TRUE | ||
diffs_deletion_color | text | null | CHECK char_length(diffs_deletion_color) <= 7 | ||
diffs_addition_color | text | null | CHECK char_length(diffs_addition_color) <= 7 | ||
markdown_automatic_lists | boolean | NOT NULL | TRUE | ||
use_new_navigation | boolean | null | |||
achievements_enabled | boolean | NOT NULL | TRUE | ||
pinned_nav_items | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
pass_user_identities_to_ci_jwt | boolean | NOT NULL | FALSE | ||
enabled_following | boolean | NOT NULL | TRUE | ||
visibility_pipeline_id_type | smallint | NOT NULL | 0 | ||
project_shortcut_buttons | boolean | NOT NULL | TRUE | ||
enabled_zoekt | boolean | NOT NULL | TRUE | ||
keyboard_shortcuts_enabled | boolean | NOT NULL | TRUE | ||
time_display_format | smallint | NOT NULL | 0 | ||
early_access_program_participant | boolean | NOT NULL | FALSE | ||
early_access_program_tracking | boolean | NOT NULL | FALSE | ||
extensions_marketplace_opt_in_status | smallint | NOT NULL | 0 | ||
organization_groups_projects_sort | text | null | CHECK char_length(organization_groups_projects_sort) <= 64 | ||
organization_groups_projects_display | smallint | NOT NULL | 1 | ||
dpop_enabled | boolean | NOT NULL | FALSE | ||
use_work_items_view | boolean | NOT NULL | FALSE | ||
text_editor_type | smallint | NOT NULL | 2 | ||
merge_request_dashboard_list_type | smallint | NOT NULL | 0 | ||
extensions_marketplace_opt_in_url | text | null | CHECK char_length(extensions_marketplace_opt_in_url) <= 512 | ||
dark_color_scheme_id | smallint | null | |||
work_items_display_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
default_duo_add_on_assignment_id | bigint | null | subscription_user_add_on_assignments.id 1:1 · SET NULL | ||
markdown_maintain_indentation | boolean | NOT NULL | FALSE | ||
merge_request_dashboard_show_drafts | boolean | NOT NULL | TRUE | ||
duo_default_namespace_id | bigint | null | namespaces.id 1:N · SET NULL | ||
policy_advanced_editor | boolean | NOT NULL | FALSE | ||
early_access_studio_participant | boolean | NOT NULL | FALSE | ||
wiki_use_auto_commit_message | boolean | NOT NULL | FALSE | ||
orbit_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
knowledge_graph_governing_namespace_id | bigint | null | namespaces.id 1:N · SET NULL |
Table checks: time_display_relative IS NOT NULL; char_length(diffs_deletion_color) <= 7; char_length(extensions_marketplace_opt_in_url) <= 512; char_length(organization_groups_projects_sort) <= 64; render_whitespace_in_code IS NOT NULL; char_length(diffs_addition_color) <= 7; tab_width IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
feature_name | smallint | NOT NULL | |||
dismissed_at | timestamptz | null |
shared_with_group_id — user_project_member_roles.shared_with_group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
shared_with_group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
member_role_id | bigint | NOT NULL | member_roles.id 1:N · CASCADE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
saved_view_id | bigint | NOT NULL | saved_views.id 1:N · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
relative_position | integer | null |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
user_id PK | bigint | NOT NULL | users.id 1:1 · CASCADE | ||
cached_markdown_version | integer | null | |||
emoji | varchar | NOT NULL | CAST('speech_balloon' AS varchar) | ||
message | varchar(100) | null | |||
message_html | varchar | null | |||
availability | smallint | NOT NULL | 0 | ||
clear_status_at | timestamptz | null |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
name_synced | boolean | null | FALSE | ||
email_synced | boolean | null | FALSE | ||
location_synced | boolean | null | FALSE | ||
user_id | bigint | NOT NULL | users.id 1:1 · CASCADE | ||
provider | varchar | null | |||
organization_synced | boolean | null | FALSE | ||
job_title_synced | boolean | null | FALSE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
user_upload_id | bigint | NOT NULL | user_uploads.id 1:1 · CASCADE | ||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 |
Table checks: char_length(verification_failure) <= 255
model_type — model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.model_type, model_id — model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | nextval(CAST('uploads_id_seq' AS regclass)) | ||
size | bigint | NOT NULL | |||
model_id | bigint | NOT NULL | |||
uploaded_by_user_id | bigint | null | |||
organization_id | bigint | null | CHECK organization_id IS NOT NULL | ||
namespace_id | bigint | null | |||
project_id | bigint | null | |||
created_at | timestamp | null | |||
store | integer | NOT NULL | 1 | ||
version | integer | null | 1 | ||
path | text | NOT NULL | CHECK char_length(path) <= 511 | ||
checksum | text | null | CHECK char_length(checksum) <= 64 | ||
model_type PK | text | NOT NULL | |||
uploader | text | NOT NULL | |||
mount_point | text | null | |||
secret | text | null |
Table checks: char_length(path) <= 511; char_length(checksum) <= 64; organization_id IS NOT NULL
Tables whose names start with value_.
value_stream_dashboard_aggregations.namespace_id → namespaces.id — one namespaces, at most one value_stream_dashboard_aggregations · required · ON DELETE CASCADE · indexedvalue_stream_dashboard_counts references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id PK | bigint | NOT NULL | |||
count PK | bigint | NOT NULL | |||
recorded_at PK | timestamptz | NOT NULL | |||
metric PK | smallint | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
last_run_at | timestamptz | null | |||
enabled | boolean | NOT NULL | TRUE |
Tables whose names start with vulnerability_.
vulnerability_archive_export_upload_states.vulnerability_archive_export_upload_id → vulnerability_archive_export_uploads.id — one vulnerability_archive_export_uploads, at most one vulnerability_archive_export_upload_states · required · ON DELETE CASCADE · indexedvulnerability_archive_export_upload_states.project_id → projects.id — one projects, many vulnerability_archive_export_upload_states · required · ON DELETE CASCADE · indexedvulnerability_detection_transitions.vulnerability_occurrence_id → vulnerability_occurrences.id — one vulnerability_occurrences, many vulnerability_detection_transitions · required · ON DELETE CASCADE · indexedvulnerability_export_part_upload_states.vulnerability_export_part_upload_id → vulnerability_export_part_uploads.id — one vulnerability_export_part_uploads, at most one vulnerability_export_part_upload_states · required · ON DELETE CASCADE · indexedvulnerability_export_part_upload_states.organization_id → organizations.id — one organizations, many vulnerability_export_part_upload_states · required · ON DELETE CASCADE · indexedvulnerability_export_parts.vulnerability_export_id → vulnerability_exports.id — one vulnerability_exports, many vulnerability_export_parts · required · ON DELETE CASCADE · indexedvulnerability_export_upload_states.vulnerability_export_upload_id → vulnerability_export_uploads.id — one vulnerability_export_uploads, at most one vulnerability_export_upload_states · required · ON DELETE CASCADE · indexedvulnerability_export_upload_states.organization_id → organizations.id — one organizations, many vulnerability_export_upload_states · required · ON DELETE CASCADE · indexedvulnerability_external_issue_links.vulnerability_occurrence_id → vulnerability_occurrences.id — one vulnerability_occurrences, many vulnerability_external_issue_links · optional · ON DELETE CASCADE · indexedvulnerability_external_issue_links.vulnerability_id → vulnerabilities.id — one vulnerabilities, many vulnerability_external_issue_links · required · ON DELETE CASCADE · indexedvulnerability_finding_ascp_component_links.vulnerability_occurrence_id → vulnerability_occurrences.id — one vulnerability_occurrences, at most one vulnerability_finding_ascp_component_links · required · ON DELETE CASCADE · indexedvulnerability_finding_ascp_component_links.ascp_component_id → ascp_components.id — one ascp_components, many vulnerability_finding_ascp_component_links · required · ON DELETE CASCADE · indexedvulnerability_finding_due_dates.vulnerability_occurrence_id → vulnerability_occurrences.id — one vulnerability_occurrences, at most one vulnerability_finding_due_dates · required · ON DELETE CASCADE · indexedvulnerability_finding_evidences.vulnerability_occurrence_id → vulnerability_occurrences.id — one vulnerability_occurrences, at most one vulnerability_finding_evidences · required · ON DELETE CASCADE · indexedvulnerability_finding_links.vulnerability_occurrence_id → vulnerability_occurrences.id — one vulnerability_occurrences, many vulnerability_finding_links · required · ON DELETE CASCADE · indexedvulnerability_finding_risk_scores.finding_id → vulnerability_occurrences.id — one vulnerability_occurrences, at most one vulnerability_finding_risk_scores · required · ON DELETE CASCADE · indexedvulnerability_finding_signatures.finding_id → vulnerability_occurrences.id — one vulnerability_occurrences, many vulnerability_finding_signatures · required · ON DELETE CASCADE · indexedvulnerability_findings_remediations.vulnerability_occurrence_id → vulnerability_occurrences.id — one vulnerability_occurrences, many vulnerability_findings_remediations · optional · ON DELETE CASCADE · indexedvulnerability_findings_remediations.vulnerability_remediation_id → vulnerability_remediations.id — one vulnerability_remediations, many vulnerability_findings_remediations · optional · ON DELETE CASCADE · indexedvulnerability_flags.vulnerability_occurrence_id → vulnerability_occurrences.id — one vulnerability_occurrences, many vulnerability_flags · required · ON DELETE CASCADE · indexedvulnerability_flip_guards.vulnerability_finding_id → vulnerability_occurrences.id — one vulnerability_occurrences, at most one vulnerability_flip_guards · required · ON DELETE CASCADE · indexedvulnerability_issue_links.vulnerability_id → vulnerabilities.id — one vulnerabilities, many vulnerability_issue_links · required · ON DELETE CASCADE · indexedvulnerability_management_policy_rules.security_policy_management_project_id → projects.id — one projects, many vulnerability_management_policy_rules · required · ON DELETE CASCADE · indexedvulnerability_management_policy_rules.security_policy_id → security_policies.id — one security_policies, many vulnerability_management_policy_rules · required · ON DELETE CASCADE · indexedvulnerability_merge_request_links.vulnerability_id → vulnerabilities.id — one vulnerabilities, many vulnerability_merge_request_links · required · ON DELETE CASCADE · indexedvulnerability_occurrence_identifiers.partition_id, identifier_id → vulnerability_identifiers.partition_id, id — one vulnerability_identifiers, many vulnerability_occurrence_identifiers · optional · ON DELETE CASCADE · indexedvulnerability_occurrence_identifiers.occurrence_id → vulnerability_occurrences.id — one vulnerability_occurrences, many vulnerability_occurrence_identifiers · required · ON DELETE CASCADE · indexedvulnerability_occurrences.vulnerability_id → vulnerabilities.id — one vulnerabilities, many vulnerability_occurrences · optional · ON DELETE SET NULL · indexedvulnerability_occurrences.scanner_id → vulnerability_scanners.id — one vulnerability_scanners, many vulnerability_occurrences · required · ON DELETE CASCADE · indexedvulnerability_occurrences.partition_id, primary_identifier_id → vulnerability_identifiers.partition_id, id — one vulnerability_identifiers, many vulnerability_occurrences · optional · ON DELETE CASCADE · indexedvulnerability_partial_scans.scan_id → security_scans.id — one security_scans, at most one vulnerability_partial_scans · required · ON DELETE CASCADE · indexedvulnerability_reads.vulnerability_occurrence_id → vulnerability_occurrences.id — one vulnerability_occurrences, at most one vulnerability_reads · optional · ON DELETE CASCADE · indexedvulnerability_reads.vulnerability_id → vulnerabilities.id — one vulnerabilities, at most one vulnerability_reads · required · ON DELETE CASCADE · indexedvulnerability_reads.scanner_id → vulnerability_scanners.id — one vulnerability_scanners, many vulnerability_reads · required · ON DELETE CASCADE · indexedvulnerability_remediation_upload_states.project_id → projects.id — one projects, many vulnerability_remediation_upload_states · required · ON DELETE CASCADE · indexedvulnerability_remediation_upload_states.vulnerability_remediation_upload_id → vulnerability_remediation_uploads.id — one vulnerability_remediation_uploads, at most one vulnerability_remediation_upload_states · required · ON DELETE CASCADE · indexedvulnerability_representation_information.vulnerability_id → vulnerabilities.id — one vulnerabilities, at most one vulnerability_representation_information · required · ON DELETE CASCADE · indexedvulnerability_severity_overrides.vulnerability_id → vulnerabilities.id — one vulnerabilities, many vulnerability_severity_overrides · required · ON DELETE CASCADE · indexedvulnerability_state_transitions.vulnerability_id → vulnerabilities.id — one vulnerabilities, many vulnerability_state_transitions · required · ON DELETE CASCADE · indexedvulnerability_triggered_workflows.vulnerability_occurrence_id → vulnerability_occurrences.id — one vulnerability_occurrences, many vulnerability_triggered_workflows · required · ON DELETE CASCADE · indexedvulnerability_user_mentions.vulnerability_id → vulnerabilities.id — one vulnerabilities, many vulnerability_user_mentions · required · ON DELETE CASCADE · indexedvulnerability_archive_exports references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
partition_number PK | bigint | NOT NULL | 1 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
started_at | timestamptz | null | |||
finished_at | timestamptz | null | |||
project_id | bigint | NOT NULL | |||
author_id | bigint | NOT NULL | |||
date_range | daterange | NOT NULL | |||
file_store | smallint | null | |||
format | smallint | null | |||
file | text | null | CHECK char_length(file) <= 255 | ||
status | text | null | CHECK char_length(status) <= 8 |
Table checks: char_length(file) <= 255; char_length(status) <= 8
vulnerability_archived_records references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
id PK | bigint | NOT NULL | |||
date PK | date | NOT NULL | |||
project_id | bigint | NOT NULL | |||
archive_id | bigint | NOT NULL | |||
vulnerability_identifier | bigint | NOT NULL | |||
data | jsonb | NOT NULL | CAST('{}' AS jsonb) |
vulnerability_archives references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
archived_records_count | integer | NOT NULL | 0 | CHECK archived_records_count >= 0 | |
date PK | date | NOT NULL |
Table checks: archived_records_count >= 0
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
vulnerability_archive_export_upload_id | bigint | NOT NULL | vulnerability_archive_export_uploads.id 1:1 · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 |
Table checks: char_length(verification_failure) <= 255
model_type — model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.model_type, model_id — model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | nextval(CAST('uploads_id_seq' AS regclass)) | ||
size | bigint | NOT NULL | |||
model_id | bigint | NOT NULL | |||
uploaded_by_user_id | bigint | null | |||
organization_id | bigint | null | |||
namespace_id | bigint | null | |||
project_id | bigint | null | CHECK project_id IS NOT NULL | ||
created_at | timestamp | null | |||
store | integer | NOT NULL | 1 | ||
version | integer | null | 1 | ||
path | text | NOT NULL | CHECK char_length(path) <= 511 | ||
checksum | text | null | CHECK char_length(checksum) <= 64 | ||
model_type PK | text | NOT NULL | |||
uploader | text | NOT NULL | |||
mount_point | text | null | |||
secret | text | null |
Table checks: char_length(path) <= 511; project_id IS NOT NULL; char_length(checksum) <= 64
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
vulnerability_occurrence_id | bigint | NOT NULL | vulnerability_occurrences.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
detected | boolean | NOT NULL | TRUE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
vulnerability_export_part_upload_id | bigint | NOT NULL | vulnerability_export_part_uploads.id 1:1 · CASCADE | ||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 |
Table checks: char_length(verification_failure) <= 255
model_type — model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.model_type, model_id — model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | nextval(CAST('uploads_id_seq' AS regclass)) | ||
size | bigint | NOT NULL | |||
model_id | bigint | NOT NULL | |||
uploaded_by_user_id | bigint | null | |||
organization_id | bigint | null | CHECK organization_id IS NOT NULL | ||
namespace_id | bigint | null | |||
project_id | bigint | null | |||
created_at | timestamp | null | |||
store | integer | NOT NULL | 1 | ||
version | integer | null | 1 | ||
path | text | NOT NULL | CHECK char_length(path) <= 511 | ||
checksum | text | null | CHECK char_length(checksum) <= 64 | ||
model_type PK | text | NOT NULL | |||
uploader | text | NOT NULL | |||
mount_point | text | null | |||
secret | text | null |
Table checks: char_length(path) <= 511; char_length(checksum) <= 64; organization_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
vulnerability_export_id | bigint | NOT NULL | vulnerability_exports.id 1:N · CASCADE | ||
start_id | bigint | NOT NULL | |||
end_id | bigint | NOT NULL | |||
organization_id | bigint | NOT NULL | |||
file_store | integer | null | |||
file | text | null | CHECK char_length(file) <= 255 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
Table checks: char_length(file) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
vulnerability_export_upload_id | bigint | NOT NULL | vulnerability_export_uploads.id 1:1 · CASCADE | ||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 |
Table checks: char_length(verification_failure) <= 255
model_type — model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.model_type, model_id — model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | nextval(CAST('uploads_id_seq' AS regclass)) | ||
size | bigint | NOT NULL | |||
model_id | bigint | NOT NULL | |||
uploaded_by_user_id | bigint | null | |||
organization_id | bigint | null | CHECK organization_id IS NOT NULL | ||
namespace_id | bigint | null | |||
project_id | bigint | null | |||
created_at | timestamp | null | |||
store | integer | NOT NULL | 1 | ||
version | integer | null | 1 | ||
path | text | NOT NULL | CHECK char_length(path) <= 511 | ||
checksum | text | null | CHECK char_length(checksum) <= 64 | ||
model_type PK | text | NOT NULL | |||
uploader | text | NOT NULL | |||
mount_point | text | null | |||
secret | text | null |
Table checks: organization_id IS NOT NULL; char_length(path) <= 511; char_length(checksum) <= 64
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
started_at | timestamptz | null | |||
finished_at | timestamptz | null | |||
status | varchar(255) | NOT NULL | |||
file | varchar(255) | null | |||
project_id | bigint | null | |||
author_id | bigint | NOT NULL | |||
file_store | integer | null | |||
format | smallint | NOT NULL | 0 | ||
group_id | bigint | null | |||
organization_id | bigint | NOT NULL | |||
expires_at | timestamptz | null | |||
send_email | boolean | NOT NULL | FALSE | ||
report_data | jsonb | NOT NULL | CAST('{}' AS jsonb) |
vulnerability_occurrence_id — vulnerability_external_issue_links.vulnerability_occurrence_id may be NULL, so the relationship to vulnerability_occurrences is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
author_id | bigint | NOT NULL | |||
vulnerability_id | bigint | NOT NULL | vulnerabilities.id 1:N · CASCADE | ||
link_type | smallint | NOT NULL | 1 | ||
external_type | smallint | NOT NULL | 1 | ||
external_project_key | text | NOT NULL | CHECK char_length(external_project_key) <= 255 | ||
external_issue_key | text | NOT NULL | CHECK char_length(external_issue_key) <= 255 | ||
project_id | bigint | null | CHECK project_id IS NOT NULL | ||
vulnerability_occurrence_id | bigint | null | vulnerability_occurrences.id 1:N · CASCADE |
Table checks: char_length(external_issue_key) <= 255; char_length(external_project_key) <= 255; project_id IS NOT NULL
vulnerability_feedback references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
feedback_type | smallint | NOT NULL | |||
category | smallint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
author_id | bigint | NOT NULL | |||
issue_id | bigint | null | |||
merge_request_id | bigint | null | |||
comment_author_id | bigint | null | |||
comment | text | null | |||
comment_timestamp | timestamptz | null | |||
finding_uuid | uuid | null | |||
dismissal_reason | smallint | null | |||
migrated_to_state_transition | boolean | null | FALSE | ||
pipeline_id | bigint | null |
ascp_component_id, vulnerability_occurrence_id — vulnerability_finding_ascp_component_links looks like a many-to-many link table but has no unique constraint across (ascp_component_id, vulnerability_occurrence_id). The same pair can be inserted twice; every join through it will double-count.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
vulnerability_occurrence_id | bigint | NOT NULL | vulnerability_occurrences.id 1:1 · CASCADE | ||
ascp_component_id | bigint | NOT NULL | ascp_components.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
vulnerability_occurrence_id | bigint | NOT NULL | vulnerability_occurrences.id 1:1 · CASCADE | ||
due_date | date | NOT NULL | |||
project_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
vulnerability_occurrence_id | bigint | NOT NULL | vulnerability_occurrences.id 1:1 · CASCADE | ||
data | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
vulnerability_occurrence_id | bigint | NOT NULL | vulnerability_occurrences.id 1:N · CASCADE | ||
name | text | null | CHECK char_length(name) <= 255 | ||
url | text | NOT NULL | CHECK char_length(url) <= 2048 | ||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL; char_length(name) <= 255; char_length(url) <= 2048
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
finding_id PK | bigint | NOT NULL | vulnerability_occurrences.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | |||
risk_score | double precision | NOT NULL | 0.0 |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
finding_id | bigint | NOT NULL | vulnerability_occurrences.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
algorithm_type | smallint | NOT NULL | |||
signature_sha | bytea | NOT NULL | |||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
vulnerability_occurrence_id — vulnerability_findings_remediations.vulnerability_occurrence_id may be NULL, so the relationship to vulnerability_occurrences is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.vulnerability_remediation_id — vulnerability_findings_remediations.vulnerability_remediation_id may be NULL, so the relationship to vulnerability_remediations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.vulnerability_occurrence_id, vulnerability_remediation_id — vulnerability_findings_remediations has several nullable FKs (vulnerability_occurrence_id, vulnerability_remediation_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
vulnerability_occurrence_id | bigint | null | vulnerability_occurrences.id 1:N · CASCADE | ||
vulnerability_remediation_id | bigint | null | vulnerability_remediations.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
vulnerability_occurrence_id | bigint | NOT NULL | vulnerability_occurrences.id 1:N · CASCADE | ||
flag_type | smallint | NOT NULL | 0 | ||
origin | text | NOT NULL | CHECK char_length(origin) <= 255 | ||
description | text | NOT NULL | CHECK char_length(description) <= 100000 | ||
project_id | bigint | null | CHECK project_id IS NOT NULL | ||
confidence_score | double precision | NOT NULL | 0.0 | CHECK confidence_score >= CAST(0 AS double precision) AND confidence_score <= CAST(1 AS double precision) | |
workflow_id | bigint | null | |||
status | smallint | NOT NULL | 0 |
Table checks: project_id IS NOT NULL; char_length(origin) <= 255; confidence_score >= CAST(0 AS double precision) AND confidence_score <= CAST(1 AS double precision); char_length(description) <= 100000
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
vulnerability_finding_id PK | bigint | NOT NULL | vulnerability_occurrences.id 1:1 · CASCADE | ||
project_id | bigint | NOT NULL | |||
first_automatic_transition_at | timestamptz | NOT NULL | |||
automated_transition_count | smallint | null | 1 | ||
is_guarded | boolean | NOT NULL | FALSE | ||
last_automatic_transition_at | timestamptz | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
vulnerability_historical_statistics references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | |||
total | integer | NOT NULL | 0 | ||
critical | integer | NOT NULL | 0 | ||
high | integer | NOT NULL | 0 | ||
medium | integer | NOT NULL | 0 | ||
low | integer | NOT NULL | 0 | ||
unknown | integer | NOT NULL | 0 | ||
info | integer | NOT NULL | 0 | ||
date | date | NOT NULL | |||
letter_grade | smallint | NOT NULL | |||
security_project_tracked_context_id | bigint | null |
external_type — external_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.external_type, external_id — external_id points at different tables depending on external_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | |||
fingerprint | bytea | NOT NULL | |||
external_type | varchar | NOT NULL | |||
external_id | varchar | NOT NULL | |||
name | varchar | NOT NULL | |||
url | text | null | |||
partition_id | bigint | null | 1 |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
vulnerability_id | bigint | NOT NULL | vulnerabilities.id 1:N · CASCADE | ||
issue_id | bigint | NOT NULL | |||
link_type | smallint | NOT NULL | 1 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | null | CHECK project_id IS NOT NULL | ||
vulnerability_occurrence_id | bigint | null |
Table checks: project_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
security_policy_id | bigint | NOT NULL | security_policies.id 1:N · CASCADE | ||
security_policy_management_project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
rule_index | smallint | NOT NULL | |||
type | smallint | NOT NULL | |||
content | jsonb | NOT NULL | CAST('{}' AS jsonb) |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
vulnerability_id | bigint | NOT NULL | vulnerabilities.id 1:N · CASCADE | ||
merge_request_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | null | CHECK project_id IS NOT NULL | ||
vulnerability_occurrence_id | bigint | null | |||
readiness_score | double precision | null | CHECK readiness_score >= CAST(0 AS double precision) AND readiness_score <= CAST(1 AS double precision) | ||
created_by_ai_workflow | boolean | NOT NULL | FALSE |
Table checks: project_id IS NOT NULL; readiness_score >= CAST(0 AS double precision) AND readiness_score <= CAST(1 AS double precision)
vulnerability_namespace_historical_statistics references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
traversal_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
total | integer | NOT NULL | 0 | ||
critical | integer | NOT NULL | 0 | ||
high | integer | NOT NULL | 0 | ||
medium | integer | NOT NULL | 0 | ||
low | integer | NOT NULL | 0 | ||
unknown | integer | NOT NULL | 0 | ||
info | integer | NOT NULL | 0 | ||
date | date | NOT NULL | |||
letter_grade | smallint | NOT NULL | |||
migrating | boolean | NOT NULL | FALSE |
vulnerability_namespace_statistics references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
namespace_id | bigint | NOT NULL | |||
total | integer | NOT NULL | 0 | ||
critical | integer | NOT NULL | 0 | ||
high | integer | NOT NULL | 0 | ||
medium | integer | NOT NULL | 0 | ||
low | integer | NOT NULL | 0 | ||
unknown | integer | NOT NULL | 0 | ||
info | integer | NOT NULL | 0 | ||
traversal_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
partition_id, identifier_id — vulnerability_occurrence_identifiers.partition_id.identifier_id may be NULL, so the relationship to vulnerability_identifiers is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
occurrence_id | bigint | NOT NULL | vulnerability_occurrences.id 1:N · CASCADE | ||
identifier_id | bigint | NOT NULL | vulnerability_identifiers.partition_id 1:N · CASCADE | ||
project_id | bigint | null | CHECK project_id IS NOT NULL | ||
partition_id | bigint | null | 1 | vulnerability_identifiers.partition_id 1:N · CASCADE | CHECK partition_id IS NOT NULL |
Table checks: project_id IS NOT NULL; partition_id IS NOT NULL
vulnerability_id — vulnerability_occurrences.vulnerability_id may be NULL, so the relationship to vulnerabilities is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.partition_id, primary_identifier_id — vulnerability_occurrences.partition_id.primary_identifier_id may be NULL, so the relationship to vulnerability_identifiers is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
severity | smallint | NOT NULL | |||
report_type | smallint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
scanner_id | bigint | NOT NULL | vulnerability_scanners.id 1:N · CASCADE | ||
primary_identifier_id | bigint | NOT NULL | vulnerability_identifiers.partition_id 1:N · CASCADE | ||
location_fingerprint | bytea | NOT NULL | |||
name | varchar | NOT NULL | |||
metadata_version | varchar | NOT NULL | |||
raw_metadata | text | null | |||
vulnerability_id | bigint | null | vulnerabilities.id 1:N · SET NULL | ||
details | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
description | text | null | CHECK char_length(description) <= 15000 | ||
solution | text | null | CHECK char_length(solution) <= 7000 | ||
cve | text | null | CHECK char_length(cve) <= 48400 | ||
location | jsonb | null | |||
detection_method | smallint | NOT NULL | 0 | ||
uuid | uuid | NOT NULL | CAST('00000000-0000-0000-0000-000000000000' AS uuid) | ||
initial_pipeline_id | bigint | null | |||
latest_pipeline_id | bigint | null | |||
security_project_tracked_context_id | bigint | null | |||
detected_at | timestamptz | null | now() | ||
new_uuid | uuid | null | |||
partition_id | bigint | null | 1 | vulnerability_identifiers.partition_id 1:N · CASCADE | CHECK partition_id IS NOT NULL |
Table checks: partition_id IS NOT NULL; char_length(solution) <= 7000; char_length(description) <= 15000; char_length(cve) <= 48400
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
scan_id PK | bigint | NOT NULL | security_scans.id 1:1 · CASCADE | ||
project_id | bigint | NOT NULL | |||
mode | smallint | NOT NULL | |||
pipeline_id | bigint | null | |||
scan_type | smallint | null |
vulnerability_occurrence_id — vulnerability_reads.vulnerability_occurrence_id may be NULL, so the relationship to vulnerability_occurrences is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
vulnerability_id | bigint | NOT NULL | vulnerabilities.id 1:1 · CASCADE | ||
project_id | bigint | NOT NULL | |||
scanner_id | bigint | NOT NULL | vulnerability_scanners.id 1:N · CASCADE | ||
report_type | smallint | NOT NULL | |||
severity | smallint | NOT NULL | |||
state | smallint | NOT NULL | |||
has_issues | boolean | NOT NULL | FALSE | ||
resolved_on_default_branch | boolean | NOT NULL | FALSE | ||
uuid | uuid | NOT NULL | |||
location_image | text | null | CHECK char_length(location_image) <= 2048 | ||
cluster_agent_id | text | null | CHECK char_length(cluster_agent_id) <= 10 | ||
casted_cluster_agent_id | bigint | null | |||
dismissal_reason | smallint | null | |||
has_merge_request | boolean | null | FALSE | CHECK has_merge_request IS NOT NULL | |
has_remediations | boolean | NOT NULL | FALSE | ||
owasp_top_10 | smallint | null | CAST('-1' AS integer) | ||
traversal_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK traversal_ids IS NOT NULL | |
archived | boolean | NOT NULL | FALSE | ||
identifier_names | text[] | NOT NULL | CAST('{}' AS text[]) | ||
has_vulnerability_resolution | boolean | null | FALSE | ||
auto_resolved | boolean | NOT NULL | FALSE | ||
security_project_tracked_context_id | bigint | null | |||
vulnerability_occurrence_id | bigint | null | vulnerability_occurrences.id 1:1 · CASCADE | ||
partition_id | bigint | null | 1 |
Table checks: char_length(location_image) <= 2048; has_merge_request IS NOT NULL; char_length(cluster_agent_id) <= 10; traversal_ids IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
vulnerability_remediation_upload_id | bigint | NOT NULL | vulnerability_remediation_uploads.id 1:1 · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 |
Table checks: char_length(verification_failure) <= 255
model_type — model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.model_type, model_id — model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | nextval(CAST('uploads_id_seq' AS regclass)) | ||
size | bigint | NOT NULL | |||
model_id | bigint | NOT NULL | |||
uploaded_by_user_id | bigint | null | |||
organization_id | bigint | null | |||
namespace_id | bigint | null | |||
project_id | bigint | null | CHECK project_id IS NOT NULL | ||
created_at | timestamp | null | |||
store | integer | NOT NULL | 1 | ||
version | integer | null | 1 | ||
path | text | NOT NULL | CHECK char_length(path) <= 511 | ||
checksum | text | null | CHECK char_length(checksum) <= 64 | ||
model_type PK | text | NOT NULL | |||
uploader | text | NOT NULL | |||
mount_point | text | null | |||
secret | text | null |
Table checks: project_id IS NOT NULL; char_length(path) <= 511; char_length(checksum) <= 64
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | smallint | null | |||
summary | text | NOT NULL | CHECK char_length(summary) <= 200 | ||
file | text | NOT NULL | CHECK char_length(file) <= 255 | ||
checksum | bytea | NOT NULL | Stores the SHA256 checksum of the attached diff file | ||
project_id | bigint | NOT NULL |
Table checks: char_length(summary) <= 200; char_length(file) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
vulnerability_id PK | bigint | NOT NULL | vulnerabilities.id 1:1 · CASCADE | ||
project_id | bigint | NOT NULL | |||
resolved_in_commit_sha | bytea | null | |||
vulnerability_occurrence_id | bigint | null | |||
removed_from_code | boolean | NOT NULL | FALSE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | |||
external_id | varchar | NOT NULL | |||
name | varchar | NOT NULL | |||
vendor | text | NOT NULL | CAST('GitLab' AS text) | CHECK char_length(vendor) <= 255 |
Table checks: char_length(vendor) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
vulnerability_id | bigint | NOT NULL | vulnerabilities.id 1:N · CASCADE | ||
author_id | bigint | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | |||
original_severity | smallint | NOT NULL | |||
new_severity | smallint | NOT NULL | |||
vulnerability_occurrence_id | bigint | null | |||
security_policy_id | bigint | null | |||
triggered_by_policy | boolean | NOT NULL | FALSE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
vulnerability_id | bigint | NOT NULL | vulnerabilities.id 1:N · CASCADE | ||
to_state | smallint | NOT NULL | |||
from_state | smallint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
author_id | bigint | null | |||
comment | text | null | CHECK char_length(comment) <= 50000 | ||
dismissal_reason | smallint | null | |||
project_id | bigint | null | CHECK project_id IS NOT NULL | ||
vulnerability_occurrence_id | bigint | null |
Table checks: project_id IS NOT NULL; char_length(comment) <= 50000
vulnerability_statistics references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | |||
total | integer | NOT NULL | 0 | ||
critical | integer | NOT NULL | 0 | ||
high | integer | NOT NULL | 0 | ||
medium | integer | NOT NULL | 0 | ||
low | integer | NOT NULL | 0 | ||
unknown | integer | NOT NULL | 0 | ||
info | integer | NOT NULL | 0 | ||
letter_grade | smallint | NOT NULL | |||
latest_pipeline_id | bigint | null | |||
archived | boolean | NOT NULL | FALSE | ||
traversal_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | CHECK cardinality(traversal_ids) > 0 | |
security_project_tracked_context_id | bigint | null |
Table checks: cardinality(traversal_ids) > 0
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
vulnerability_occurrence_id | bigint | NOT NULL | vulnerability_occurrences.id 1:N · CASCADE | ||
workflow_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
workflow_name | smallint | NOT NULL | 0 |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
vulnerability_id | bigint | NOT NULL | vulnerabilities.id 1:N · CASCADE | ||
mentioned_users_ids | bigint[] | null | |||
mentioned_projects_ids | bigint[] | null | |||
mentioned_groups_ids | bigint[] | null | |||
note_id | bigint | null | |||
project_id | bigint | null | CHECK project_id IS NOT NULL | ||
vulnerability_occurrence_id | bigint | null |
Table checks: project_id IS NOT NULL
Tables whose names start with web_.
web_hooks.project_id → projects.id — one projects, many web_hooks · optional · ON DELETE CASCADE · indexedweb_hooks.integration_id → integrations.id — one integrations, many web_hooks · optional · ON DELETE CASCADE · indexedweb_hooks.organization_id → organizations.id — one organizations, many web_hooks · optional · ON DELETE CASCADE · indexedweb_hooks.group_id → namespaces.id — one namespaces, many web_hooks · optional · ON DELETE CASCADE · indexedresponse_status — response_status is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.web_hook_logs_daily references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
web_hook_id | bigint | NOT NULL | |||
trigger | varchar | null | |||
url | varchar | null | |||
request_headers | text | null | |||
request_data | text | null | |||
response_headers | text | null | |||
response_body | text | null | |||
response_status | varchar | null | |||
execution_duration | double precision | null | |||
internal_error_message | varchar | null | |||
updated_at | timestamp | NOT NULL | |||
created_at PK | timestamp | NOT NULL | |||
url_hash | text | null | CHECK char_length(url_hash) <= 44 | ||
organization_id | bigint | null | |||
group_id | bigint | null | |||
project_id | bigint | null |
Table checks: char_length(url_hash) <= 44; num_nonnulls(group_id, organization_id, project_id) = 1
project_id — web_hooks.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.integration_id — web_hooks.integration_id may be NULL, so the relationship to integrations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id — web_hooks.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — web_hooks.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.type — type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.project_id, integration_id, organization_id, group_id — web_hooks has several nullable FKs (project_id, integration_id, organization_id, group_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
type | varchar | null | CAST('ProjectHook' AS varchar) | ||
push_events | boolean | NOT NULL | TRUE | ||
issues_events | boolean | NOT NULL | FALSE | ||
merge_requests_events | boolean | NOT NULL | FALSE | ||
tag_push_events | boolean | null | FALSE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
note_events | boolean | NOT NULL | FALSE | ||
enable_ssl_verification | boolean | null | TRUE | ||
wiki_page_events | boolean | NOT NULL | FALSE | ||
pipeline_events | boolean | NOT NULL | FALSE | ||
confidential_issues_events | boolean | NOT NULL | FALSE | ||
repository_update_events | boolean | NOT NULL | FALSE | ||
job_events | boolean | NOT NULL | FALSE | ||
confidential_note_events | boolean | null | |||
push_events_branch_filter | text | null | |||
encrypted_token | varchar | null | |||
encrypted_token_iv | varchar | null | |||
encrypted_url | varchar | null | |||
encrypted_url_iv | varchar | null | |||
deployment_events | boolean | NOT NULL | FALSE | ||
releases_events | boolean | NOT NULL | FALSE | ||
feature_flag_events | boolean | NOT NULL | FALSE | ||
member_events | boolean | NOT NULL | FALSE | ||
subgroup_events | boolean | NOT NULL | FALSE | ||
recent_failures | smallint | NOT NULL | 0 | ||
disabled_until | timestamptz | null | |||
encrypted_url_variables | bytea | null | |||
encrypted_url_variables_iv | bytea | null | |||
integration_id | bigint | null | integrations.id 1:N · CASCADE | ||
branch_filter_strategy | smallint | NOT NULL | 0 | ||
emoji_events | boolean | NOT NULL | FALSE | ||
name | text | null | CHECK char_length(name) <= 255 | ||
description | text | null | CHECK char_length(description) <= 2048 | ||
custom_webhook_template | text | null | CHECK char_length(custom_webhook_template) <= 4096 | ||
resource_access_token_events | boolean | NOT NULL | FALSE | ||
encrypted_custom_headers | bytea | null | |||
encrypted_custom_headers_iv | bytea | null | |||
project_events | boolean | NOT NULL | FALSE | ||
vulnerability_events | boolean | NOT NULL | FALSE | ||
member_approval_events | boolean | NOT NULL | FALSE | ||
milestone_events | boolean | NOT NULL | FALSE | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE | ||
resource_deploy_token_events | boolean | NOT NULL | FALSE | ||
filter | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(filter) = CAST('object' AS text) | |
signing_token | jsonb | null |
Table checks: char_length(name) <= 255; char_length(description) <= 2048; char_length(custom_webhook_template) <= 4096; num_nonnulls(group_id, organization_id, project_id) = 1; jsonb_typeof(filter) = CAST('object' AS text)
Tables whose names start with zoekt_.
zoekt_enabled_namespaces.root_namespace_id → namespaces.id — one namespaces, at most one zoekt_enabled_namespaces · required · ON DELETE CASCADE · indexedzoekt_indices.zoekt_enabled_namespace_id → zoekt_enabled_namespaces.id — one zoekt_enabled_namespaces, many zoekt_indices · optional · ON DELETE SET NULL · indexedzoekt_indices.zoekt_node_id → zoekt_nodes.id — one zoekt_nodes, many zoekt_indices · required · ON DELETE CASCADE · indexedzoekt_indices.zoekt_replica_id → zoekt_replicas.id — one zoekt_replicas, many zoekt_indices · optional · ON DELETE SET NULL · indexedzoekt_replicas.namespace_id → namespaces.id — one namespaces, many zoekt_replicas · required · ON DELETE CASCADE · indexedzoekt_replicas.zoekt_enabled_namespace_id → zoekt_enabled_namespaces.id — one zoekt_enabled_namespaces, many zoekt_replicas · required · ON DELETE CASCADE · indexedzoekt_repositories.project_id → projects.id — one projects, many zoekt_repositories · optional · ON DELETE SET NULL · indexedzoekt_repositories.zoekt_index_id → zoekt_indices.id — one zoekt_indices, many zoekt_repositories · required · ON DELETE RESTRICT · indexedzoekt_tasks references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
partition_id PK | bigint | NOT NULL | 1 | ||
zoekt_node_id | bigint | NOT NULL | |||
zoekt_repository_id | bigint | NOT NULL | |||
project_identifier | bigint | NOT NULL | |||
perform_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
state | smallint | NOT NULL | 0 | ||
task_type | smallint | NOT NULL | |||
retries_left | smallint | NOT NULL | 5 |
Table checks: retries_left > 0 OR (retries_left = 0 AND state = 255)
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
root_namespace_id | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
search | boolean | NOT NULL | TRUE | ||
metadata | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
last_rollout_failed_at | timestamptz | null | |||
number_of_replicas_override | integer | null |
zoekt_enabled_namespace_id — zoekt_indices.zoekt_enabled_namespace_id may be NULL, so the relationship to zoekt_enabled_namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.zoekt_replica_id — zoekt_indices.zoekt_replica_id may be NULL, so the relationship to zoekt_replicas is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.zoekt_enabled_namespace_id, zoekt_replica_id — zoekt_indices has several nullable FKs (zoekt_enabled_namespace_id, zoekt_replica_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
zoekt_enabled_namespace_id | bigint | null | zoekt_enabled_namespaces.id 1:N · SET NULL | ||
zoekt_node_id | bigint | NOT NULL | zoekt_nodes.id 1:N · CASCADE | ||
namespace_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
state | smallint | NOT NULL | 0 | ||
zoekt_replica_id | bigint | null | zoekt_replicas.id 1:N · SET NULL | ||
reserved_storage_bytes | bigint | null | CAST('10737418240' AS bigint) | ||
used_storage_bytes | bigint | NOT NULL | 0 | ||
watermark_level | smallint | NOT NULL | 0 | ||
metadata | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
used_storage_bytes_updated_at | timestamptz | NOT NULL | CAST('1970-01-01 00:00:00+00' AS timestamptz) | ||
last_indexed_at | timestamptz | NOT NULL | CAST('1970-01-01 00:00:00+00' AS timestamptz) |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
uuid | uuid | NOT NULL | |||
used_bytes | bigint | NOT NULL | 0 | ||
total_bytes | bigint | NOT NULL | 0 | ||
last_seen_at | timestamptz | NOT NULL | CAST('1970-01-01 00:00:00+00' AS timestamptz) | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
index_base_url | text | NOT NULL | CHECK char_length(index_base_url) <= 1024 | ||
search_base_url | text | NOT NULL | CHECK char_length(search_base_url) <= 1024 | ||
metadata | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
indexed_bytes | bigint | NOT NULL | 0 | ||
usable_storage_bytes | bigint | NOT NULL | 0 | ||
schema_version | smallint | NOT NULL | 0 | ||
services | smallint[] | NOT NULL | CAST('{0}' AS smallint[]) | ||
webserver_last_seen_at | timestamptz | null |
Table checks: char_length(search_base_url) <= 1024; char_length(index_base_url) <= 1024
namespace_id, zoekt_enabled_namespace_id — zoekt_replicas looks like a many-to-many link table but has no unique constraint across (namespace_id, zoekt_enabled_namespace_id). The same pair can be inserted twice; every join through it will double-count.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
zoekt_enabled_namespace_id | bigint | NOT NULL | zoekt_enabled_namespaces.id 1:N · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
state | smallint | NOT NULL | 0 |
project_id — zoekt_repositories.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
zoekt_index_id | bigint | NOT NULL | zoekt_indices.id 1:N · RESTRICT | ||
project_id | bigint | null | projects.id 1:N · SET NULL | ||
project_identifier | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
indexed_at | timestamptz | null | |||
state | smallint | NOT NULL | 0 | ||
size_bytes | bigint | NOT NULL | 0 | ||
index_file_count | integer | NOT NULL | 0 | ||
retries_left | smallint | NOT NULL | 10 | ||
schema_version | smallint | NOT NULL | 0 |
Table checks: project_id IS NULL OR project_identifier = project_id; retries_left > 0 OR (retries_left = 0 AND state >= 200)
Tables whose names start with analytics_.
analytics_cycle_analytics_aggregations.group_id → namespaces.id — one namespaces, at most one analytics_cycle_analytics_aggregations · required · ON DELETE CASCADE · indexedanalytics_cycle_analytics_group_stages.stage_event_hash_id → analytics_cycle_analytics_stage_event_hashes.id — one analytics_cycle_analytics_stage_event_hashes, many analytics_cycle_analytics_group_stages · optional · ON DELETE CASCADE · indexedanalytics_cycle_analytics_group_stages.group_value_stream_id → analytics_cycle_analytics_group_value_streams.id — one analytics_cycle_analytics_group_value_streams, many analytics_cycle_analytics_group_stages · required · ON DELETE CASCADE · indexedanalytics_cycle_analytics_group_stages.start_event_label_id → labels.id — one labels, many analytics_cycle_analytics_group_stages · optional · ON DELETE CASCADE · indexedanalytics_cycle_analytics_group_stages.group_id → namespaces.id — one namespaces, many analytics_cycle_analytics_group_stages · required · ON DELETE CASCADE · indexedanalytics_cycle_analytics_group_stages.end_event_label_id → labels.id — one labels, many analytics_cycle_analytics_group_stages · optional · ON DELETE CASCADE · indexedanalytics_cycle_analytics_group_value_streams.group_id → namespaces.id — one namespaces, many analytics_cycle_analytics_group_value_streams · required · ON DELETE CASCADE · indexedanalytics_cycle_analytics_stage_aggregations.stage_id → analytics_cycle_analytics_group_stages.id — one analytics_cycle_analytics_group_stages, at most one analytics_cycle_analytics_stage_aggregations · required · ON DELETE CASCADE · indexedanalytics_cycle_analytics_stage_aggregations.group_id → namespaces.id — one namespaces, many analytics_cycle_analytics_stage_aggregations · required · ON DELETE CASCADE · indexedanalytics_cycle_analytics_stage_event_hashes.organization_id → organizations.id — one organizations, many analytics_cycle_analytics_stage_event_hashes · required · ON DELETE CASCADE · indexedanalytics_cycle_analytics_value_stream_settings.namespace_id → namespaces.id — one namespaces, many analytics_cycle_analytics_value_stream_settings · required · ON DELETE CASCADE · indexedanalytics_cycle_analytics_value_stream_settings.value_stream_id → analytics_cycle_analytics_group_value_streams.id — one analytics_cycle_analytics_group_value_streams, at most one analytics_cycle_analytics_value_stream_settings · required · ON DELETE CASCADE · indexedanalytics_dashboards_pointers.target_project_id → projects.id — one projects, many analytics_dashboards_pointers · required · ON DELETE CASCADE · indexedanalytics_dashboards_pointers.namespace_id → namespaces.id — one namespaces, at most one analytics_dashboards_pointers · optional · ON DELETE CASCADE · indexedanalytics_dashboards_pointers.project_id → projects.id — one projects, at most one analytics_dashboards_pointers · optional · ON DELETE CASCADE · indexedanalytics_devops_adoption_segments.display_namespace_id → namespaces.id — one namespaces, many analytics_devops_adoption_segments · optional · ON DELETE CASCADE · indexedanalytics_devops_adoption_segments.namespace_id → namespaces.id — one namespaces, many analytics_devops_adoption_segments · optional · ON DELETE CASCADE · indexedanalytics_devops_adoption_snapshots.namespace_id → namespaces.id — one namespaces, many analytics_devops_adoption_snapshots · optional · ON DELETE CASCADE · indexedanalytics_cycle_analytics_issue_stage_events references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_00 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_01 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_02 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_03 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_04 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_05 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_06 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_07 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_08 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_09 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_10 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_11 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_12 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_13 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_14 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_15 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_16 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_17 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_18 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_19 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_20 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_21 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_22 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_23 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_24 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_25 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_26 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_27 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_28 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_29 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_30 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_issue_stage_events_31 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
weight | integer | null | |||
sprint_id | bigint | null | |||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_00 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_01 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_02 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_03 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_04 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_05 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_06 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_07 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_08 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_09 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_10 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_11 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_12 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_13 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_14 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_15 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_16 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_17 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_18 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_19 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_20 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_21 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_22 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_23 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_24 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_25 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_26 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_27 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_28 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_29 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_30 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
analytics_cycle_analytics_merge_request_stage_events_31 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
stage_event_hash_id PK | bigint | NOT NULL | |||
merge_request_id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
milestone_id | bigint | null | |||
author_id | bigint | null | |||
start_event_timestamp | timestamptz | NOT NULL | |||
end_event_timestamp | timestamptz | null | |||
state_id | smallint | NOT NULL | 1 | ||
duration_in_milliseconds | bigint | null |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id PK | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
incremental_runtimes_in_seconds | integer[] | NOT NULL | CAST('{}' AS integer[]) | CHECK cardinality(incremental_runtimes_in_seconds) <= 10 | |
incremental_processed_records | integer[] | NOT NULL | CAST('{}' AS integer[]) | CHECK cardinality(incremental_processed_records) <= 10 | |
last_incremental_issues_id | bigint | null | |||
last_incremental_merge_requests_id | bigint | null | |||
last_incremental_run_at | timestamptz | null | |||
last_incremental_issues_updated_at | timestamptz | null | |||
last_incremental_merge_requests_updated_at | timestamptz | null | |||
last_full_run_at | timestamptz | null | |||
last_consistency_check_updated_at | timestamptz | null | |||
enabled | boolean | NOT NULL | TRUE | ||
full_runtimes_in_seconds | integer[] | NOT NULL | CAST('{}' AS integer[]) | CHECK cardinality(full_runtimes_in_seconds) <= 10 | |
full_processed_records | integer[] | NOT NULL | CAST('{}' AS integer[]) | CHECK cardinality(full_processed_records) <= 10 | |
last_full_merge_requests_updated_at | timestamptz | null | |||
last_full_issues_updated_at | timestamptz | null | |||
last_full_issues_id | bigint | null | |||
last_full_merge_requests_id | bigint | null | |||
last_consistency_check_issues_stage_event_hash_id | bigint | null | |||
last_consistency_check_issues_start_event_timestamp | timestamptz | null | |||
last_consistency_check_issues_end_event_timestamp | timestamptz | null | |||
last_consistency_check_issues_issuable_id | bigint | null | |||
last_consistency_check_merge_requests_stage_event_hash_id | bigint | null | |||
last_consistency_check_merge_requests_start_event_timestamp | timestamptz | null | |||
last_consistency_check_merge_requests_end_event_timestamp | timestamptz | null | |||
last_consistency_check_merge_requests_issuable_id | bigint | null |
Table checks: cardinality(incremental_runtimes_in_seconds) <= 10; cardinality(incremental_processed_records) <= 10; cardinality(full_processed_records) <= 10; cardinality(full_runtimes_in_seconds) <= 10
stage_event_hash_id — analytics_cycle_analytics_group_stages.stage_event_hash_id may be NULL, so the relationship to analytics_cycle_analytics_stage_event_hashes is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.start_event_label_id — analytics_cycle_analytics_group_stages.start_event_label_id may be NULL, so the relationship to labels is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.end_event_label_id — analytics_cycle_analytics_group_stages.end_event_label_id may be NULL, so the relationship to labels is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.stage_event_hash_id, start_event_label_id, end_event_label_id — analytics_cycle_analytics_group_stages has several nullable FKs (stage_event_hash_id, start_event_label_id, end_event_label_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
relative_position | integer | null | |||
start_event_identifier | integer | NOT NULL | |||
end_event_identifier | integer | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
start_event_label_id | bigint | null | labels.id 1:N · CASCADE | ||
end_event_label_id | bigint | null | labels.id 1:N · CASCADE | ||
hidden | boolean | NOT NULL | FALSE | ||
custom | boolean | NOT NULL | TRUE | ||
name | varchar(255) | NOT NULL | |||
group_value_stream_id | bigint | NOT NULL | analytics_cycle_analytics_group_value_streams.id 1:N · CASCADE | ||
stage_event_hash_id | bigint | null | analytics_cycle_analytics_stage_event_hashes.id 1:N · CASCADE | CHECK stage_event_hash_id IS NOT NULL |
Table checks: stage_event_hash_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
name | text | NOT NULL | CHECK char_length(name) <= 100 |
Table checks: char_length(name) <= 100
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
stage_id PK | bigint | NOT NULL | analytics_cycle_analytics_group_stages.id 1:1 · CASCADE | ||
runtimes_in_seconds | integer[] | NOT NULL | CAST('{}' AS integer[]) | CHECK cardinality(runtimes_in_seconds) <= 10 | |
processed_records | integer[] | NOT NULL | CAST('{}' AS integer[]) | CHECK cardinality(processed_records) <= 10 | |
last_issues_id | bigint | null | |||
last_merge_requests_id | bigint | null | |||
last_run_at | timestamptz | null | |||
last_issues_updated_at | timestamptz | null | |||
last_merge_requests_updated_at | timestamptz | null | |||
last_completed_at | timestamptz | null | |||
enabled | boolean | NOT NULL | TRUE |
Table checks: cardinality(runtimes_in_seconds) <= 10; cardinality(processed_records) <= 10
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
hash_sha256 | bytea | null | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE |
namespace_id, value_stream_id — analytics_cycle_analytics_value_stream_settings looks like a many-to-many link table but has no unique constraint across (namespace_id, value_stream_id). The same pair can be inserted twice; every join through it will double-count.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
value_stream_id PK | bigint | NOT NULL | analytics_cycle_analytics_group_value_streams.id 1:1 · CASCADE | ||
project_ids_filter | bigint[] | null | CAST('{}' AS bigint[]) | CHECK cardinality(project_ids_filter) <= 100 AND array_position(project_ids_filter, CAST(NULL AS bigint)) IS NULL | |
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE |
Table checks: cardinality(project_ids_filter) <= 100 AND array_position(project_ids_filter, CAST(NULL AS bigint)) IS NULL
namespace_id — analytics_dashboards_pointers.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — analytics_dashboards_pointers.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | null | namespaces.id 1:1 · CASCADE | ||
project_id | bigint | null | projects.id 1:1 · CASCADE | ||
target_project_id | bigint | NOT NULL | projects.id 1:N · CASCADE |
Table checks: project_id IS NULL <> namespace_id IS NULL
display_namespace_id — analytics_devops_adoption_segments.display_namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id — analytics_devops_adoption_segments.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.display_namespace_id, namespace_id — analytics_devops_adoption_segments has several nullable FKs (display_namespace_id, namespace_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
last_recorded_at | timestamptz | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL | |
display_namespace_id | bigint | null | namespaces.id 1:N · CASCADE |
Table checks: namespace_id IS NOT NULL
namespace_id — analytics_devops_adoption_snapshots.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
recorded_at | timestamptz | NOT NULL | |||
issue_opened | boolean | NOT NULL | |||
merge_request_opened | boolean | NOT NULL | |||
merge_request_approved | boolean | NOT NULL | |||
runner_configured | boolean | NOT NULL | |||
pipeline_succeeded | boolean | NOT NULL | |||
deploy_succeeded | boolean | NOT NULL | |||
end_time | timestamptz | NOT NULL | |||
total_projects_count | integer | null | |||
code_owners_used_count | integer | null | |||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL | |
sast_enabled_count | integer | null | |||
dast_enabled_count | integer | null | |||
dependency_scanning_enabled_count | integer | null | |||
coverage_fuzzing_enabled_count | integer | null | |||
vulnerability_management_used_count | integer | null |
Table checks: namespace_id IS NOT NULL
analytics_usage_trends_measurements references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
count | bigint | NOT NULL | |||
recorded_at | timestamptz | NOT NULL | |||
identifier | smallint | NOT NULL |
Tables whose names start with issue_.
issue_search_data.namespace_id → namespaces.id — one namespaces, many issue_search_data · optional · ON DELETE CASCADE · indexedissue_search_data.issue_id → issues.id — one issues, many issue_search_data · required · ON DELETE CASCADE · indexedissue_search_data.project_id → projects.id — one projects, many issue_search_data · required · ON DELETE CASCADE · indexedissue_assignees.namespace_id → namespaces.id — one namespaces, many issue_assignees · optional · ON DELETE CASCADE · indexedissue_assignees.user_id → users.id — one users, many issue_assignees · required · ON DELETE CASCADE · indexedissue_assignees.issue_id → issues.id — one issues, many issue_assignees · required · ON DELETE CASCADE · indexedissue_assignment_events.namespace_id → namespaces.id — one namespaces, many issue_assignment_events · optional · ON DELETE CASCADE · indexedissue_assignment_events.issue_id → issues.id — one issues, many issue_assignment_events · required · ON DELETE CASCADE · indexedissue_assignment_events.user_id → users.id — one users, many issue_assignment_events · optional · ON DELETE SET NULL · indexedissue_customer_relations_contacts.issue_id → issues.id — one issues, many issue_customer_relations_contacts · required · ON DELETE CASCADE · indexedissue_customer_relations_contacts.namespace_id → namespaces.id — one namespaces, many issue_customer_relations_contacts · optional · ON DELETE CASCADE · indexedissue_customer_relations_contacts.contact_id → customer_relations_contacts.id — one customer_relations_contacts, many issue_customer_relations_contacts · required · ON DELETE CASCADE · indexedissue_email_participants.namespace_id → namespaces.id — one namespaces, many issue_email_participants · optional · ON DELETE CASCADE · indexedissue_email_participants.issue_id → issues.id — one issues, many issue_email_participants · required · ON DELETE CASCADE · indexedissue_emails.namespace_id → namespaces.id — one namespaces, many issue_emails · optional · ON DELETE CASCADE · indexedissue_emails.issue_id → issues.id — one issues, many issue_emails · required · ON DELETE CASCADE · indexedissue_links.namespace_id → namespaces.id — one namespaces, many issue_links · optional · ON DELETE CASCADE · indexedissue_links.source_id → issues.id — one issues, many issue_links · required · ON DELETE CASCADE · indexedissue_links.target_id → issues.id — one issues, many issue_links · required · ON DELETE CASCADE · indexedissue_metrics.namespace_id → namespaces.id — one namespaces, many issue_metrics · optional · ON DELETE CASCADE · indexedissue_metrics.issue_id → issues.id — one issues, at most one issue_metrics · required · ON DELETE CASCADE · indexedissue_tracker_data.integration_id → integrations.id — one integrations, many issue_tracker_data · optional · ON DELETE CASCADE · indexedissue_tracker_data.group_id → namespaces.id — one namespaces, many issue_tracker_data · optional · ON DELETE CASCADE · indexedissue_tracker_data.organization_id → organizations.id — one organizations, many issue_tracker_data · optional · ON DELETE CASCADE · indexedissue_tracker_data.project_id → projects.id — one projects, many issue_tracker_data · optional · ON DELETE CASCADE · indexedissue_user_mentions.namespace_id → namespaces.id — one namespaces, many issue_user_mentions · optional · ON DELETE CASCADE · indexedissue_user_mentions.note_id → notes.id — one notes, many issue_user_mentions · optional · ON DELETE CASCADE · indexedissue_user_mentions.issue_id → issues.id — one issues, many issue_user_mentions · required · ON DELETE CASCADE · indexednamespace_id — issue_search_data.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
issue_id PK | bigint | NOT NULL | issues.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE |
issue_search_data_00 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_01 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_02 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_03 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_04 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_05 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_06 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_07 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_08 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_09 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_10 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_11 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_12 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_13 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_14 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_15 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_16 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_17 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_18 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_19 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_20 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_21 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_22 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_23 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_24 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_25 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_26 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_27 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_28 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_29 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_30 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_31 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_32 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_33 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_34 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_35 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_36 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_37 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_38 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_39 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_40 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_41 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_42 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_43 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_44 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_45 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_46 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_47 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_48 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_49 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_50 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_51 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_52 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_53 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_54 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_55 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_56 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_57 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_58 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_59 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_60 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_61 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_62 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
issue_search_data_63 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | |||
issue_id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
search_vector | tsvector | null | |||
namespace_id | bigint | null |
namespace_id — issue_assignees.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
user_id PK | bigint | NOT NULL | users.id 1:N · CASCADE | ||
issue_id PK | bigint | NOT NULL | issues.id 1:N · CASCADE | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL
namespace_id — issue_assignment_events.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id — issue_assignment_events.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id, user_id — issue_assignment_events has several nullable FKs (namespace_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | null | users.id 1:N · SET NULL | ||
issue_id | bigint | NOT NULL | issues.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | now() | ||
action | smallint | NOT NULL | 1 | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL
namespace_id — issue_customer_relations_contacts.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
issue_id | bigint | NOT NULL | issues.id 1:N · CASCADE | ||
contact_id | bigint | NOT NULL | customer_relations_contacts.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL
namespace_id — issue_email_participants.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
issue_id | bigint | NOT NULL | issues.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
email | text | NOT NULL | CHECK char_length(email) <= 255 | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: char_length(email) <= 255; namespace_id IS NOT NULL
namespace_id — issue_emails.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
issue_id | bigint | NOT NULL | issues.id 1:N · CASCADE | ||
email_message_id | text | NOT NULL | CHECK char_length(email_message_id) <= 1000 | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL; char_length(email_message_id) <= 1000
namespace_id — issue_links.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
source_id | bigint | NOT NULL | issues.id 1:N · CASCADE | ||
target_id | bigint | NOT NULL | issues.id 1:N · CASCADE | ||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
link_type | smallint | NOT NULL | 0 | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL
namespace_id — issue_metrics.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.first_mentioned_in_commit_at — first_mentioned_in_commit_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.first_associated_with_milestone_at — first_associated_with_milestone_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.first_added_to_board_at — first_added_to_board_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
issue_id | bigint | NOT NULL | issues.id 1:1 · CASCADE | ||
first_mentioned_in_commit_at | timestamp | null | |||
first_associated_with_milestone_at | timestamp | null | |||
first_added_to_board_at | timestamp | null | |||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL | |||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL
integration_id — issue_tracker_data.integration_id may be NULL, so the relationship to integrations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — issue_tracker_data.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id — issue_tracker_data.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — issue_tracker_data.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.integration_id, group_id, organization_id, project_id — issue_tracker_data has several nullable FKs (integration_id, group_id, organization_id, project_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
encrypted_project_url | varchar | null | |||
encrypted_project_url_iv | varchar | null | |||
encrypted_issues_url | varchar | null | |||
encrypted_issues_url_iv | varchar | null | |||
encrypted_new_issue_url | varchar | null | |||
encrypted_new_issue_url_iv | varchar | null | |||
integration_id | bigint | null | integrations.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE |
Table checks: num_nonnulls(group_id, organization_id, project_id) = 1
namespace_id — issue_user_mentions.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.note_id — issue_user_mentions.note_id may be NULL, so the relationship to notes is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id, note_id — issue_user_mentions has several nullable FKs (namespace_id, note_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
issue_id | bigint | NOT NULL | issues.id 1:N · CASCADE | ||
mentioned_users_ids | bigint[] | null | |||
mentioned_projects_ids | bigint[] | null | |||
mentioned_groups_ids | bigint[] | null | |||
note_id | bigint | null | notes.id 1:N · CASCADE | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL
Tables whose names start with namespace_.
namespace_admin_notes.namespace_id → namespaces.id — one namespaces, many namespace_admin_notes · required · ON DELETE CASCADE · indexednamespace_aggregation_schedules.namespace_id → namespaces.id — one namespaces, at most one namespace_aggregation_schedules · required · ON DELETE CASCADE · indexednamespace_ai_settings.namespace_id → namespaces.id — one namespaces, at most one namespace_ai_settings · required · ON DELETE CASCADE · indexednamespace_bans.user_id → users.id — one users, many namespace_bans · required · ON DELETE CASCADE · indexednamespace_bans.namespace_id → namespaces.id — one namespaces, many namespace_bans · required · ON DELETE CASCADE · indexednamespace_ci_cd_settings.namespace_id → namespaces.id — one namespaces, at most one namespace_ci_cd_settings · required · ON DELETE CASCADE · indexednamespace_cluster_agent_mappings.namespace_id → namespaces.id — one namespaces, many namespace_cluster_agent_mappings · required · ON DELETE CASCADE · indexednamespace_cluster_agent_mappings.creator_id → users.id — one users, many namespace_cluster_agent_mappings · optional · ON DELETE SET NULL · indexednamespace_cluster_agent_mappings.cluster_agent_id → cluster_agents.id — one cluster_agents, many namespace_cluster_agent_mappings · required · ON DELETE CASCADE · indexednamespace_commit_emails.namespace_id → namespaces.id — one namespaces, many namespace_commit_emails · required · ON DELETE CASCADE · indexednamespace_commit_emails.email_id → emails.id — one emails, many namespace_commit_emails · required · ON DELETE CASCADE · indexednamespace_commit_emails.user_id → users.id — one users, many namespace_commit_emails · required · ON DELETE CASCADE · indexednamespace_details.namespace_id → namespaces.id — one namespaces, at most one namespace_details · required · ON DELETE CASCADE · indexednamespace_foundational_agent_statuses.namespace_id → namespaces.id — one namespaces, many namespace_foundational_agent_statuses · required · ON DELETE CASCADE · indexednamespace_import_users.namespace_id → namespaces.id — one namespaces, at most one namespace_import_users · required · ON DELETE CASCADE · indexednamespace_import_users.user_id → users.id — one users, at most one namespace_import_users · required · ON DELETE CASCADE · indexednamespace_isolations.namespace_id → namespaces.id — one namespaces, at most one namespace_isolations · required · ON DELETE CASCADE · indexednamespace_ldap_settings.namespace_id → namespaces.id — one namespaces, at most one namespace_ldap_settings · required · ON DELETE CASCADE · indexednamespace_limits.namespace_id → namespaces.id — one namespaces, at most one namespace_limits · required · ON DELETE CASCADE · indexednamespace_package_settings.namespace_id → namespaces.id — one namespaces, at most one namespace_package_settings · required · ON DELETE CASCADE · indexednamespace_root_storage_statistics.namespace_id → namespaces.id — one namespaces, at most one namespace_root_storage_statistics · required · ON DELETE CASCADE · indexednamespace_secret_counts.namespace_id → namespaces.id — one namespaces, at most one namespace_secret_counts · required · ON DELETE CASCADE · indexednamespace_secret_counts.root_namespace_id → namespaces.id — one namespaces, many namespace_secret_counts · required · ON DELETE CASCADE · indexednamespace_settings.default_compliance_framework_id → compliance_management_frameworks.id — one compliance_management_frameworks, at most one namespace_settings · optional · ON DELETE SET NULL · indexednamespace_settings.namespace_id → namespaces.id — one namespaces, at most one namespace_settings · required · ON DELETE CASCADE · indexednamespace_state_propagations.namespace_id → namespaces.id — one namespaces, many namespace_state_propagations · required · ON DELETE CASCADE · indexednamespace_statistics.namespace_id → namespaces.id — one namespaces, at most one namespace_statistics · required · ON DELETE CASCADE · indexednamespace_template_settings.custom_project_templates_group_id → namespaces.id — one namespaces, many namespace_template_settings · optional · ON DELETE SET NULL · indexednamespace_template_settings.file_template_project_id → projects.id — one projects, many namespace_template_settings · optional · ON DELETE SET NULL · indexednamespace_template_settings.duo_template_project_id → projects.id — one projects, many namespace_template_settings · optional · ON DELETE SET NULL · indexednamespace_template_settings.namespace_id → namespaces.id — one namespaces, at most one namespace_template_settings · required · ON DELETE CASCADE · indexednamespace_descendants references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_00 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_01 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_02 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_03 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_04 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_05 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_06 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_07 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_08 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_09 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_10 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_11 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_12 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_13 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_14 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_15 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_16 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_17 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_18 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_19 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_20 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_21 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_22 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_23 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_24 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_25 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_26 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_27 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_28 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_29 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_30 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
namespace_descendants_31 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | |||
self_and_descendant_group_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
all_project_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
traversal_ids | bigint[] | NOT NULL | CAST(ARRAY[] AS bigint[]) | ||
outdated_at | timestamptz | null | |||
calculated_at | timestamptz | null | |||
all_active_project_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
all_unarchived_project_ids | bigint[] | null | CAST('{}' AS bigint[]) | CHECK all_unarchived_project_ids IS NOT NULL | |
self_and_descendant_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
Table checks: all_unarchived_project_ids IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
note | text | null | CHECK char_length(note) <= 1000 |
Table checks: char_length(note) <= 1000
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | namespaces.id 1:1 · CASCADE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
duo_workflow_mcp_enabled | boolean | NOT NULL | FALSE | ||
foundational_agents_default_enabled | boolean | null | TRUE | ||
minimum_access_level_execute | smallint | null | |||
minimum_access_level_manage | smallint | null | |||
minimum_access_level_enable_on_projects | smallint | null | |||
minimum_access_level_execute_async | smallint | null | |||
feature_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(feature_settings) = CAST('object' AS text) | |
prompt_injection_protection_level | smallint | NOT NULL | 0 | ||
ai_usage_data_collection_enabled | boolean | NOT NULL | FALSE | ||
include_recommended_allowed | boolean | NOT NULL | FALSE | ||
allow_all_unix_sockets | boolean | NOT NULL | FALSE | ||
enforce_on_local_clients | boolean | NOT NULL | FALSE | ||
allow_project_extension | boolean | NOT NULL | TRUE | ||
allowed_domains | text[] | NOT NULL | CAST('{}' AS text[]) | ||
denied_domains | text[] | NOT NULL | CAST('{}' AS text[]) | ||
ai_catalog_restricted_to_group_hierarchy | boolean | NOT NULL | FALSE |
Table checks: jsonb_typeof(feature_settings) = CAST('object' AS text)
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
allow_stale_runner_pruning | boolean | NOT NULL | FALSE |
creator_id — namespace_cluster_agent_mappings.creator_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
cluster_agent_id | bigint | NOT NULL | cluster_agents.id 1:N · CASCADE | ||
creator_id | bigint | null | users.id 1:N · SET NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
email_id | bigint | NOT NULL | emails.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
created_at | timestamptz | null | |||
updated_at | timestamptz | null | |||
cached_markdown_version | integer | null | |||
description | text | null | |||
description_html | text | null | |||
creator_id | bigint | null | |||
state_metadata | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(state_metadata) = CAST('object' AS text) | |
deletion_scheduled_at | timestamptz | null |
Table checks: jsonb_typeof(state_metadata) = CAST('object' AS text)
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
reference | text | NOT NULL | CHECK char_length(reference) <= 255 | ||
enabled | boolean | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
Table checks: char_length(reference) <= 255
namespace_id, user_id — namespace_import_users looks like a many-to-many link table but has no unique constraint across (namespace_id, user_id). The same pair can be inserted twice; every join through it will double-count.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:1 · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:1 · CASCADE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
isolated | boolean | NOT NULL | FALSE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
sync_last_start_at | timestamptz | null | |||
sync_last_update_at | timestamptz | null | |||
sync_last_successful_at | timestamptz | null | |||
sync_status | smallint | NOT NULL | 0 | ||
sync_error | text | null | CHECK char_length(sync_error) <= 255 |
Table checks: char_length(sync_error) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
additional_purchased_storage_size | bigint | NOT NULL | 0 | ||
additional_purchased_storage_ends_on | date | null | |||
namespace_id PK | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
pre_enforcement_notification_at | timestamptz | null | |||
first_enforced_at | timestamptz | null | |||
last_enforced_at | timestamptz | null | |||
last_seat_all_used_seats_notification_at | timestamptz | null | |||
max_number_of_vulnerabilities_per_project | integer | null |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
maven_duplicates_allowed | boolean | NOT NULL | TRUE | ||
maven_duplicate_exception_regex | text | NOT NULL | CAST('' AS text) | CHECK char_length(maven_duplicate_exception_regex) <= 255 | |
generic_duplicates_allowed | boolean | NOT NULL | TRUE | ||
generic_duplicate_exception_regex | text | NOT NULL | CAST('' AS text) | CHECK char_length(generic_duplicate_exception_regex) <= 255 | |
maven_package_requests_forwarding | boolean | null | |||
lock_maven_package_requests_forwarding | boolean | NOT NULL | FALSE | ||
pypi_package_requests_forwarding | boolean | null | |||
lock_pypi_package_requests_forwarding | boolean | NOT NULL | FALSE | ||
npm_package_requests_forwarding | boolean | null | |||
lock_npm_package_requests_forwarding | boolean | NOT NULL | FALSE | ||
nuget_duplicates_allowed | boolean | NOT NULL | TRUE | ||
nuget_duplicate_exception_regex | text | NOT NULL | CAST('' AS text) | CHECK char_length(nuget_duplicate_exception_regex) <= 255 | |
nuget_symbol_server_enabled | boolean | NOT NULL | FALSE | ||
terraform_module_duplicates_allowed | boolean | NOT NULL | FALSE | ||
terraform_module_duplicate_exception_regex | text | NOT NULL | CAST('' AS text) | CHECK char_length(terraform_module_duplicate_exception_regex) <= 255 | |
audit_events_enabled | boolean | NOT NULL | FALSE | ||
rubygems_package_requests_forwarding | boolean | null | |||
lock_rubygems_package_requests_forwarding | boolean | NOT NULL | FALSE |
Table checks: char_length(generic_duplicate_exception_regex) <= 255; char_length(maven_duplicate_exception_regex) <= 255; char_length(nuget_duplicate_exception_regex) <= 255; char_length(terraform_module_duplicate_exception_regex) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
updated_at | timestamptz | NOT NULL | |||
repository_size | bigint | NOT NULL | 0 | ||
lfs_objects_size | bigint | NOT NULL | 0 | ||
wiki_size | bigint | NOT NULL | 0 | ||
build_artifacts_size | bigint | NOT NULL | 0 | ||
storage_size | bigint | NOT NULL | 0 | ||
packages_size | bigint | NOT NULL | 0 | ||
snippets_size | bigint | NOT NULL | 0 | ||
pipeline_artifacts_size | bigint | NOT NULL | 0 | ||
uploads_size | bigint | NOT NULL | 0 | ||
dependency_proxy_size | bigint | NOT NULL | 0 | ||
notification_level | smallint | NOT NULL | 100 | ||
container_registry_size | bigint | NOT NULL | 0 | ||
registry_size_estimated | boolean | NOT NULL | FALSE | ||
public_forks_storage_size | bigint | NOT NULL | 0 | ||
internal_forks_storage_size | bigint | NOT NULL | 0 | ||
private_forks_storage_size | bigint | NOT NULL | 0 |
namespace_id, root_namespace_id — namespace_secret_counts looks like a many-to-many link table but has no unique constraint across (namespace_id, root_namespace_id). The same pair can be inserted twice; every join through it will double-count.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
root_namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
count | integer | NOT NULL | 0 |
default_compliance_framework_id — namespace_settings.default_compliance_framework_id may be NULL, so the relationship to compliance_management_frameworks is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
namespace_id PK | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
prevent_forking_outside_group | boolean | NOT NULL | FALSE | ||
allow_mfa_for_subgroups | boolean | NOT NULL | TRUE | ||
default_branch_name | text | null | CHECK char_length(default_branch_name) <= 255 | ||
repository_read_only | boolean | NOT NULL | FALSE | ||
resource_access_token_creation_allowed | boolean | NOT NULL | TRUE | ||
prevent_sharing_groups_outside_hierarchy | boolean | NOT NULL | FALSE | ||
new_user_signups_cap | integer | null | |||
setup_for_company | boolean | null | |||
jobs_to_be_done | smallint | null | |||
runner_token_expiration_interval | integer | null | |||
subgroup_runner_token_expiration_interval | integer | null | |||
project_runner_token_expiration_interval | integer | null | |||
show_diff_preview_in_email | boolean | NOT NULL | TRUE | ||
enabled_git_access_protocol | smallint | NOT NULL | 0 | ||
unique_project_download_limit | smallint | NOT NULL | 0 | ||
unique_project_download_limit_interval_in_seconds | integer | NOT NULL | 0 | ||
unique_project_download_limit_allowlist | text[] | NOT NULL | CAST('{}' AS text[]) | CHECK cardinality(unique_project_download_limit_allowlist) <= 100 | |
auto_ban_user_on_excessive_projects_download | boolean | NOT NULL | FALSE | ||
only_allow_merge_if_pipeline_succeeds | boolean | NOT NULL | FALSE | ||
allow_merge_on_skipped_pipeline | boolean | NOT NULL | FALSE | ||
only_allow_merge_if_all_discussions_are_resolved | boolean | NOT NULL | FALSE | ||
default_compliance_framework_id | bigint | null | compliance_management_frameworks.id 1:1 · SET NULL | ||
runner_registration_enabled | boolean | null | TRUE | ||
allow_runner_registration_token | boolean | NOT NULL | TRUE | ||
unique_project_download_limit_alertlist | integer[] | NOT NULL | CAST('{}' AS integer[]) | CHECK cardinality(unique_project_download_limit_alertlist) <= 100 | |
emails_enabled | boolean | NOT NULL | TRUE | ||
experiment_features_enabled | boolean | NOT NULL | FALSE | ||
default_branch_protection_defaults | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK octet_length(CAST(default_branch_protection_defaults AS text)) <= 1024 | |
service_access_tokens_expiration_enforced | boolean | NOT NULL | TRUE | ||
product_analytics_enabled | boolean | NOT NULL | FALSE | ||
allow_merge_without_pipeline | boolean | NOT NULL | FALSE | ||
enforce_ssh_certificates | boolean | NOT NULL | FALSE | ||
math_rendering_limits_enabled | boolean | null | |||
lock_math_rendering_limits_enabled | boolean | NOT NULL | FALSE | ||
duo_features_enabled | boolean | null | |||
lock_duo_features_enabled | boolean | NOT NULL | FALSE | ||
disable_personal_access_tokens | boolean | NOT NULL | FALSE | ||
early_access_program_participant | boolean | NOT NULL | FALSE | ||
remove_dormant_members | boolean | NOT NULL | FALSE | ||
remove_dormant_members_period | integer | NOT NULL | 90 | ||
early_access_program_joined_by_id | bigint | null | |||
seat_control | smallint | NOT NULL | 0 | ||
last_dormant_member_review_at | timestamptz | null | |||
enterprise_users_extensions_marketplace_opt_in_status | smallint | NOT NULL | 0 | ||
spp_repository_pipeline_access | boolean | null | |||
lock_spp_repository_pipeline_access | boolean | NOT NULL | FALSE | ||
archived | boolean | NOT NULL | FALSE | ||
token_expiry_notify_inherited | boolean | NOT NULL | TRUE | ||
resource_access_token_notify_inherited | boolean | null | |||
lock_resource_access_token_notify_inherited | boolean | NOT NULL | FALSE | ||
pipeline_variables_default_role | smallint | NOT NULL | 2 | ||
force_pages_access_control | boolean | NOT NULL | FALSE | ||
extended_grat_expiry_webhooks_execute | boolean | NOT NULL | FALSE | ||
jwt_ci_cd_job_token_enabled | boolean | NOT NULL | FALSE | ||
jwt_ci_cd_job_token_opted_out | boolean | NOT NULL | FALSE | ||
require_dpop_for_manage_api_endpoints | boolean | NOT NULL | FALSE | ||
job_token_policies_enabled | boolean | NOT NULL | FALSE | ||
security_policies | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(security_policies) = CAST('object' AS text) | |
duo_nano_features_enabled | boolean | null | |||
model_prompt_cache_enabled | boolean | null | |||
lock_model_prompt_cache_enabled | boolean | NOT NULL | FALSE | ||
disable_invite_members | boolean | NOT NULL | FALSE | ||
web_based_commit_signing_enabled | boolean | null | |||
lock_web_based_commit_signing_enabled | boolean | NOT NULL | FALSE | ||
allow_enterprise_bypass_placeholder_confirmation | boolean | NOT NULL | FALSE | ||
enterprise_bypass_expires_at | timestamptz | null | |||
hide_email_on_profile | boolean | NOT NULL | FALSE | ||
allow_personal_snippets | boolean | NOT NULL | TRUE | ||
auto_duo_code_review_enabled | boolean | null | |||
lock_auto_duo_code_review_enabled | boolean | NOT NULL | FALSE | ||
step_up_auth_required_oauth_provider | text | null | CHECK char_length(step_up_auth_required_oauth_provider) <= 255 | ||
duo_remote_flows_enabled | boolean | null | |||
lock_duo_remote_flows_enabled | boolean | NOT NULL | FALSE | ||
disable_ssh_keys | boolean | NOT NULL | FALSE | ||
duo_foundational_flows_enabled | boolean | null | |||
lock_duo_foundational_flows_enabled | boolean | NOT NULL | FALSE | ||
usage_billing | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(usage_billing) = CAST('object' AS text) | |
built_in_project_templates_enabled | boolean | null | |||
lock_built_in_project_templates_enabled | boolean | NOT NULL | FALSE | ||
tool_approval_for_session_enabled | boolean | null | |||
lock_tool_approval_for_session_enabled | boolean | NOT NULL | FALSE | ||
duo_custom_flows_enabled | boolean | null | |||
lock_duo_custom_flows_enabled | boolean | NOT NULL | FALSE | ||
duo_custom_agents_enabled | boolean | null | |||
lock_duo_custom_agents_enabled | boolean | NOT NULL | FALSE | ||
mcp_server_enabled | boolean | null | |||
duo_external_agents_enabled | boolean | null | |||
lock_duo_external_agents_enabled | boolean | NOT NULL | FALSE | ||
enable_duo_code_review_by_default | smallint | NOT NULL | 0 | ||
personal_access_token_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(personal_access_token_settings) = CAST('object' AS text) | |
require_sha_for_merge | boolean | null | |||
lock_require_sha_for_merge | boolean | NOT NULL | FALSE | ||
admin_locked_duo_features_enabled | boolean | NOT NULL | FALSE | ||
ai_audit_events_storage_enabled | boolean | null | |||
lock_ai_audit_events_storage_enabled | boolean | NOT NULL | FALSE | ||
dependency_firewall_enabled | boolean | NOT NULL | FALSE | ||
policy_store_experiment_enabled | boolean | NOT NULL | FALSE | ||
enterprise_user_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(enterprise_user_settings) = CAST('object' AS text) |
Table checks: char_length(default_branch_name) <= 255; char_length(step_up_auth_required_oauth_provider) <= 255; jsonb_typeof(enterprise_user_settings) = CAST('object' AS text); jsonb_typeof(personal_access_token_settings) = CAST('object' AS text); jsonb_typeof(security_policies) = CAST('object' AS text); jsonb_typeof(usage_billing) = CAST('object' AS text); cardinality(unique_project_download_limit_alertlist) <= 100; cardinality(unique_project_download_limit_allowlist) <= 100; octet_length(CAST(default_branch_protection_defaults AS text)) <= 1024
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
started_at | timestamptz | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
source_state | smallint | NOT NULL | |||
target_state | smallint | NOT NULL | |||
status | smallint | NOT NULL | 0 |
shared_runners_seconds_last_reset — shared_runners_seconds_last_reset stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
shared_runners_seconds | integer | NOT NULL | 0 | ||
shared_runners_seconds_last_reset | timestamp | null | |||
storage_size | bigint | NOT NULL | 0 | ||
wiki_size | bigint | NOT NULL | 0 | ||
dependency_proxy_size | bigint | NOT NULL | 0 |
custom_project_templates_group_id — namespace_template_settings.custom_project_templates_group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.file_template_project_id — namespace_template_settings.file_template_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.duo_template_project_id — namespace_template_settings.duo_template_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.custom_project_templates_group_id, file_template_project_id, duo_template_project_id — namespace_template_settings has several nullable FKs (custom_project_templates_group_id, file_template_project_id, duo_template_project_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
file_template_project_id | bigint | null | projects.id 1:N · SET NULL | ||
custom_project_templates_group_id | bigint | null | namespaces.id 1:N · SET NULL | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
duo_template_project_id | bigint | null | projects.id 1:N · SET NULL |
model_type — model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.model_type, model_id — model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | nextval(CAST('uploads_id_seq' AS regclass)) | ||
size | bigint | NOT NULL | |||
model_id | bigint | NOT NULL | |||
uploaded_by_user_id | bigint | null | |||
organization_id | bigint | null | |||
namespace_id | bigint | null | CHECK namespace_id IS NOT NULL | ||
project_id | bigint | null | |||
created_at | timestamp | null | |||
store | integer | NOT NULL | 1 | ||
version | integer | null | 1 | ||
path | text | NOT NULL | CHECK char_length(path) <= 511 | ||
checksum | text | null | CHECK char_length(checksum) <= 64 | ||
model_type PK | text | NOT NULL | |||
uploader | text | NOT NULL | |||
mount_point | text | null | |||
secret | text | null |
Table checks: char_length(path) <= 511; char_length(checksum) <= 64; namespace_id IS NOT NULL
Tables whose names start with virtual_.
virtual_registries_container_cache_entries.group_id → namespaces.id — one namespaces, many virtual_registries_container_cache_entries · required · ON DELETE NO ACTION · indexedvirtual_registries_container_cache_remote_entries.group_id → namespaces.id — one namespaces, many virtual_registries_container_cache_remote_entries · required · ON DELETE NO ACTION · indexedvirtual_registries_packages_maven_cache_local_entries.package_file_id → packages_package_files.id — one packages_package_files, many virtual_registries_packages_maven_cache_local_entries · required · ON DELETE CASCADE · indexedvirtual_registries_packages_maven_cache_local_entries.group_id → namespaces.id — one namespaces, many virtual_registries_packages_maven_cache_local_entries · required · ON DELETE CASCADE · indexedvirtual_registries_packages_maven_cache_local_entries.upstream_id → virtual_registries_packages_maven_local_upstreams.id — one virtual_registries_packages_maven_local_upstreams, many virtual_registries_packages_maven_cache_local_entries · required · ON DELETE CASCADE · indexedvirtual_registries_packages_maven_cache_remote_entries.group_id → namespaces.id — one namespaces, many virtual_registries_packages_maven_cache_remote_entries · required · ON DELETE NO ACTION · indexedvirtual_registries_packages_npm_cache_local_entries.upstream_id → virtual_registries_packages_npm_upstreams.id — one virtual_registries_packages_npm_upstreams, many virtual_registries_packages_npm_cache_local_entries · required · ON DELETE CASCADE · indexedvirtual_registries_packages_npm_cache_local_entries.group_id → namespaces.id — one namespaces, many virtual_registries_packages_npm_cache_local_entries · required · ON DELETE CASCADE · indexedvirtual_registries_packages_npm_cache_local_entries.package_file_id → packages_package_files.id — one packages_package_files, many virtual_registries_packages_npm_cache_local_entries · required · ON DELETE CASCADE · indexedvirtual_registries_packages_npm_cache_remote_entries.group_id → namespaces.id — one namespaces, many virtual_registries_packages_npm_cache_remote_entries · required · ON DELETE NO ACTION · indexedvirtual_registries_cleanup_policies.group_id → namespaces.id — one namespaces, at most one virtual_registries_cleanup_policies · required · ON DELETE CASCADE · indexedvirtual_registries_container_registries.group_id → namespaces.id — one namespaces, many virtual_registries_container_registries · required · ON DELETE CASCADE · indexedvirtual_registries_container_registry_upstreams.registry_id → virtual_registries_container_registries.id — one virtual_registries_container_registries, many virtual_registries_container_registry_upstreams · required · ON DELETE CASCADE · indexedvirtual_registries_container_registry_upstreams.group_id → namespaces.id — one namespaces, many virtual_registries_container_registry_upstreams · required · ON DELETE CASCADE · indexedvirtual_registries_container_registry_upstreams.upstream_id → virtual_registries_container_upstreams.id — one virtual_registries_container_upstreams, many virtual_registries_container_registry_upstreams · required · ON DELETE CASCADE · indexedvirtual_registries_container_upstreams.group_id → namespaces.id — one namespaces, many virtual_registries_container_upstreams · required · ON DELETE CASCADE · indexedvirtual_registries_packages_maven_cache_remote_entry_states.group_id, virtual_registries_packages_maven_cache_remote_entry_iid → virtual_registries_packages_maven_cache_remote_entries.group_id, iid — one virtual_registries_packages_maven_cache_remote_entries, many virtual_registries_packages_maven_cache_remote_entry_states · required · ON DELETE CASCADE · indexedvirtual_registries_packages_maven_cache_remote_entry_states.group_id → namespaces.id — one namespaces, many virtual_registries_packages_maven_cache_remote_entry_states · required · ON DELETE NO ACTION · indexedvirtual_registries_packages_maven_local_upstreams.group_id → namespaces.id — one namespaces, many virtual_registries_packages_maven_local_upstreams · required · ON DELETE CASCADE · indexedvirtual_registries_packages_maven_registries.group_id → namespaces.id — one namespaces, many virtual_registries_packages_maven_registries · required · ON DELETE CASCADE · indexedvirtual_registries_packages_maven_registry_upstreams.upstream_id → virtual_registries_packages_maven_upstreams.id — one virtual_registries_packages_maven_upstreams, many virtual_registries_packages_maven_registry_upstreams · optional · ON DELETE CASCADE · indexedvirtual_registries_packages_maven_registry_upstreams.group_id → namespaces.id — one namespaces, many virtual_registries_packages_maven_registry_upstreams · required · ON DELETE CASCADE · indexedvirtual_registries_packages_maven_registry_upstreams.local_upstream_id → virtual_registries_packages_maven_local_upstreams.id — one virtual_registries_packages_maven_local_upstreams, many virtual_registries_packages_maven_registry_upstreams · optional · ON DELETE CASCADE · indexedvirtual_registries_packages_maven_registry_upstreams.registry_id → virtual_registries_packages_maven_registries.id — one virtual_registries_packages_maven_registries, many virtual_registries_packages_maven_registry_upstreams · required · ON DELETE CASCADE · indexedvirtual_registries_packages_maven_upstream_rules.remote_upstream_id → virtual_registries_packages_maven_upstreams.id — one virtual_registries_packages_maven_upstreams, many virtual_registries_packages_maven_upstream_rules · required · ON DELETE CASCADE · indexedvirtual_registries_packages_maven_upstream_rules.group_id → namespaces.id — one namespaces, many virtual_registries_packages_maven_upstream_rules · required · ON DELETE CASCADE · indexedvirtual_registries_packages_maven_upstreams.group_id → namespaces.id — one namespaces, many virtual_registries_packages_maven_upstreams · required · ON DELETE CASCADE · indexedvirtual_registries_packages_npm_registries.group_id → namespaces.id — one namespaces, many virtual_registries_packages_npm_registries · required · ON DELETE CASCADE · indexedvirtual_registries_packages_npm_registry_upstreams.upstream_id → virtual_registries_packages_npm_upstreams.id — one virtual_registries_packages_npm_upstreams, many virtual_registries_packages_npm_registry_upstreams · required · ON DELETE CASCADE · indexedvirtual_registries_packages_npm_registry_upstreams.registry_id → virtual_registries_packages_npm_registries.id — one virtual_registries_packages_npm_registries, many virtual_registries_packages_npm_registry_upstreams · required · ON DELETE CASCADE · indexedvirtual_registries_packages_npm_registry_upstreams.group_id → namespaces.id — one namespaces, many virtual_registries_packages_npm_registry_upstreams · required · ON DELETE CASCADE · indexedvirtual_registries_packages_npm_upstreams.group_id → namespaces.id — one namespaces, many virtual_registries_packages_npm_upstreams · required · ON DELETE CASCADE · indexedvirtual_registries_settings.group_id → namespaces.id — one namespaces, at most one virtual_registries_settings · required · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | namespaces.id 1:N | ||
upstream_id PK | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(object_storage_key) <= 1024; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20
virtual_registries_container_cache_entries_00 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(object_storage_key) <= 1024; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20
virtual_registries_container_cache_entries_01 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(object_storage_key) <= 1024; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20
virtual_registries_container_cache_entries_02 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(object_storage_key) <= 1024; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20
virtual_registries_container_cache_entries_03 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(object_storage_key) <= 1024; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20
virtual_registries_container_cache_entries_04 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(object_storage_key) <= 1024; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20
virtual_registries_container_cache_entries_05 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(object_storage_key) <= 1024; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20
virtual_registries_container_cache_entries_06 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(object_storage_key) <= 1024; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20
virtual_registries_container_cache_entries_07 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(object_storage_key) <= 1024; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20
virtual_registries_container_cache_entries_08 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(object_storage_key) <= 1024; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20
virtual_registries_container_cache_entries_09 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(object_storage_key) <= 1024; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20
virtual_registries_container_cache_entries_10 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(object_storage_key) <= 1024; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20
virtual_registries_container_cache_entries_11 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(object_storage_key) <= 1024; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20
virtual_registries_container_cache_entries_12 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(object_storage_key) <= 1024; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20
virtual_registries_container_cache_entries_13 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(object_storage_key) <= 1024; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20
virtual_registries_container_cache_entries_14 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(object_storage_key) <= 1024; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20
virtual_registries_container_cache_entries_15 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(object_storage_key) <= 1024; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_container_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | namespaces.id 1:N | ||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
digest | text | null | CHECK char_length(digest) <= 71 | ||
size | bigint | null | CHECK size IS NOT NULL |
Table checks: char_length(digest) <= 71; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; digest IS NULL OR char_length(digest) = 71; size IS NOT NULL; char_length(content_type) <= 255; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; octet_length(file_sha1) = 20
virtual_registries_container_cache_remote_entries_00 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_container_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
digest | text | null | CHECK char_length(digest) <= 71 | ||
size | bigint | null | CHECK size IS NOT NULL |
Table checks: char_length(digest) <= 71; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; digest IS NULL OR char_length(digest) = 71; size IS NOT NULL; char_length(content_type) <= 255; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; octet_length(file_sha1) = 20
virtual_registries_container_cache_remote_entries_01 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_container_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
digest | text | null | CHECK char_length(digest) <= 71 | ||
size | bigint | null | CHECK size IS NOT NULL |
Table checks: char_length(digest) <= 71; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; digest IS NULL OR char_length(digest) = 71; size IS NOT NULL; char_length(content_type) <= 255; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; octet_length(file_sha1) = 20
virtual_registries_container_cache_remote_entries_02 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_container_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
digest | text | null | CHECK char_length(digest) <= 71 | ||
size | bigint | null | CHECK size IS NOT NULL |
Table checks: char_length(digest) <= 71; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; digest IS NULL OR char_length(digest) = 71; size IS NOT NULL; char_length(content_type) <= 255; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; octet_length(file_sha1) = 20
virtual_registries_container_cache_remote_entries_03 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_container_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
digest | text | null | CHECK char_length(digest) <= 71 | ||
size | bigint | null | CHECK size IS NOT NULL |
Table checks: char_length(digest) <= 71; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; digest IS NULL OR char_length(digest) = 71; size IS NOT NULL; char_length(content_type) <= 255; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; octet_length(file_sha1) = 20
virtual_registries_container_cache_remote_entries_04 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_container_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
digest | text | null | CHECK char_length(digest) <= 71 | ||
size | bigint | null | CHECK size IS NOT NULL |
Table checks: char_length(digest) <= 71; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; digest IS NULL OR char_length(digest) = 71; size IS NOT NULL; char_length(content_type) <= 255; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; octet_length(file_sha1) = 20
virtual_registries_container_cache_remote_entries_05 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_container_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
digest | text | null | CHECK char_length(digest) <= 71 | ||
size | bigint | null | CHECK size IS NOT NULL |
Table checks: char_length(digest) <= 71; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; digest IS NULL OR char_length(digest) = 71; size IS NOT NULL; char_length(content_type) <= 255; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; octet_length(file_sha1) = 20
virtual_registries_container_cache_remote_entries_06 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_container_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
digest | text | null | CHECK char_length(digest) <= 71 | ||
size | bigint | null | CHECK size IS NOT NULL |
Table checks: char_length(digest) <= 71; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; digest IS NULL OR char_length(digest) = 71; size IS NOT NULL; char_length(content_type) <= 255; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; octet_length(file_sha1) = 20
virtual_registries_container_cache_remote_entries_07 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_container_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
digest | text | null | CHECK char_length(digest) <= 71 | ||
size | bigint | null | CHECK size IS NOT NULL |
Table checks: char_length(digest) <= 71; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; digest IS NULL OR char_length(digest) = 71; size IS NOT NULL; char_length(content_type) <= 255; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; octet_length(file_sha1) = 20
virtual_registries_container_cache_remote_entries_08 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_container_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
digest | text | null | CHECK char_length(digest) <= 71 | ||
size | bigint | null | CHECK size IS NOT NULL |
Table checks: char_length(digest) <= 71; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; digest IS NULL OR char_length(digest) = 71; size IS NOT NULL; char_length(content_type) <= 255; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; octet_length(file_sha1) = 20
virtual_registries_container_cache_remote_entries_09 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_container_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
digest | text | null | CHECK char_length(digest) <= 71 | ||
size | bigint | null | CHECK size IS NOT NULL |
Table checks: char_length(digest) <= 71; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; digest IS NULL OR char_length(digest) = 71; size IS NOT NULL; char_length(content_type) <= 255; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; octet_length(file_sha1) = 20
virtual_registries_container_cache_remote_entries_10 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_container_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
digest | text | null | CHECK char_length(digest) <= 71 | ||
size | bigint | null | CHECK size IS NOT NULL |
Table checks: char_length(digest) <= 71; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; digest IS NULL OR char_length(digest) = 71; size IS NOT NULL; char_length(content_type) <= 255; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; octet_length(file_sha1) = 20
virtual_registries_container_cache_remote_entries_11 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_container_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
digest | text | null | CHECK char_length(digest) <= 71 | ||
size | bigint | null | CHECK size IS NOT NULL |
Table checks: char_length(digest) <= 71; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; digest IS NULL OR char_length(digest) = 71; size IS NOT NULL; char_length(content_type) <= 255; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; octet_length(file_sha1) = 20
virtual_registries_container_cache_remote_entries_12 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_container_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
digest | text | null | CHECK char_length(digest) <= 71 | ||
size | bigint | null | CHECK size IS NOT NULL |
Table checks: char_length(digest) <= 71; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; digest IS NULL OR char_length(digest) = 71; size IS NOT NULL; char_length(content_type) <= 255; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; octet_length(file_sha1) = 20
virtual_registries_container_cache_remote_entries_13 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_container_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
digest | text | null | CHECK char_length(digest) <= 71 | ||
size | bigint | null | CHECK size IS NOT NULL |
Table checks: char_length(digest) <= 71; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; digest IS NULL OR char_length(digest) = 71; size IS NOT NULL; char_length(content_type) <= 255; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; octet_length(file_sha1) = 20
virtual_registries_container_cache_remote_entries_14 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_container_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
digest | text | null | CHECK char_length(digest) <= 71 | ||
size | bigint | null | CHECK size IS NOT NULL |
Table checks: char_length(digest) <= 71; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; digest IS NULL OR char_length(digest) = 71; size IS NOT NULL; char_length(content_type) <= 255; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; octet_length(file_sha1) = 20
virtual_registries_container_cache_remote_entries_15 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_container_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
digest | text | null | CHECK char_length(digest) <= 71 | ||
size | bigint | null | CHECK size IS NOT NULL |
Table checks: char_length(digest) <= 71; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; digest IS NULL OR char_length(digest) = 71; size IS NOT NULL; char_length(content_type) <= 255; char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; octet_length(file_sha1) = 20
virtual_registries_packages_maven_cache_entries references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
downloads_count | bigint | NOT NULL | 0 | ||
downloaded_at | timestamptz | null |
Table checks: char_length(content_type) <= 255; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; char_length(relative_path) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20; char_length(upstream_etag) <= 255
virtual_registries_packages_maven_cache_entries_00 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
downloads_count | bigint | NOT NULL | 0 | ||
downloaded_at | timestamptz | null |
Table checks: char_length(content_type) <= 255; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; char_length(relative_path) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20; char_length(upstream_etag) <= 255
virtual_registries_packages_maven_cache_entries_01 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
downloads_count | bigint | NOT NULL | 0 | ||
downloaded_at | timestamptz | null |
Table checks: char_length(content_type) <= 255; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; char_length(relative_path) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20; char_length(upstream_etag) <= 255
virtual_registries_packages_maven_cache_entries_02 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
downloads_count | bigint | NOT NULL | 0 | ||
downloaded_at | timestamptz | null |
Table checks: char_length(content_type) <= 255; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; char_length(relative_path) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20; char_length(upstream_etag) <= 255
virtual_registries_packages_maven_cache_entries_03 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
downloads_count | bigint | NOT NULL | 0 | ||
downloaded_at | timestamptz | null |
Table checks: char_length(content_type) <= 255; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; char_length(relative_path) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20; char_length(upstream_etag) <= 255
virtual_registries_packages_maven_cache_entries_04 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
downloads_count | bigint | NOT NULL | 0 | ||
downloaded_at | timestamptz | null |
Table checks: char_length(content_type) <= 255; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; char_length(relative_path) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20; char_length(upstream_etag) <= 255
virtual_registries_packages_maven_cache_entries_05 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
downloads_count | bigint | NOT NULL | 0 | ||
downloaded_at | timestamptz | null |
Table checks: char_length(content_type) <= 255; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; char_length(relative_path) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20; char_length(upstream_etag) <= 255
virtual_registries_packages_maven_cache_entries_06 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
downloads_count | bigint | NOT NULL | 0 | ||
downloaded_at | timestamptz | null |
Table checks: char_length(content_type) <= 255; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; char_length(relative_path) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20; char_length(upstream_etag) <= 255
virtual_registries_packages_maven_cache_entries_07 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
downloads_count | bigint | NOT NULL | 0 | ||
downloaded_at | timestamptz | null |
Table checks: char_length(content_type) <= 255; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; char_length(relative_path) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20; char_length(upstream_etag) <= 255
virtual_registries_packages_maven_cache_entries_08 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
downloads_count | bigint | NOT NULL | 0 | ||
downloaded_at | timestamptz | null |
Table checks: char_length(content_type) <= 255; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; char_length(relative_path) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20; char_length(upstream_etag) <= 255
virtual_registries_packages_maven_cache_entries_09 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
downloads_count | bigint | NOT NULL | 0 | ||
downloaded_at | timestamptz | null |
Table checks: char_length(content_type) <= 255; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; char_length(relative_path) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20; char_length(upstream_etag) <= 255
virtual_registries_packages_maven_cache_entries_10 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
downloads_count | bigint | NOT NULL | 0 | ||
downloaded_at | timestamptz | null |
Table checks: char_length(content_type) <= 255; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; char_length(relative_path) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20; char_length(upstream_etag) <= 255
virtual_registries_packages_maven_cache_entries_11 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
downloads_count | bigint | NOT NULL | 0 | ||
downloaded_at | timestamptz | null |
Table checks: char_length(content_type) <= 255; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; char_length(relative_path) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20; char_length(upstream_etag) <= 255
virtual_registries_packages_maven_cache_entries_12 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
downloads_count | bigint | NOT NULL | 0 | ||
downloaded_at | timestamptz | null |
Table checks: char_length(content_type) <= 255; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; char_length(relative_path) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20; char_length(upstream_etag) <= 255
virtual_registries_packages_maven_cache_entries_13 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
downloads_count | bigint | NOT NULL | 0 | ||
downloaded_at | timestamptz | null |
Table checks: char_length(content_type) <= 255; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; char_length(relative_path) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20; char_length(upstream_etag) <= 255
virtual_registries_packages_maven_cache_entries_14 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
downloads_count | bigint | NOT NULL | 0 | ||
downloaded_at | timestamptz | null |
Table checks: char_length(content_type) <= 255; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; char_length(relative_path) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20; char_length(upstream_etag) <= 255
virtual_registries_packages_maven_cache_entries_15 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
group_id | bigint | NOT NULL | |||
upstream_id PK | bigint | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status PK | smallint | NOT NULL | 0 | ||
relative_path PK | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
downloads_count | bigint | NOT NULL | 0 | ||
downloaded_at | timestamptz | null |
Table checks: char_length(content_type) <= 255; char_length(file) <= 1024; char_length(object_storage_key) <= 1024; char_length(relative_path) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20; char_length(upstream_etag) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
upstream_id | bigint | NOT NULL | virtual_registries_packages_maven_local_upstreams.id 1:N · CASCADE | ||
package_file_id | bigint | NOT NULL | packages_package_files.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_maven_cache_local_entries_00 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_maven_cache_local_entries_01 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_maven_cache_local_entries_02 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_maven_cache_local_entries_03 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_maven_cache_local_entries_04 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_maven_cache_local_entries_05 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_maven_cache_local_entries_06 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_maven_cache_local_entries_07 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_maven_cache_local_entries_08 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_maven_cache_local_entries_09 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_maven_cache_local_entries_10 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_maven_cache_local_entries_11 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_maven_cache_local_entries_12 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_maven_cache_local_entries_13 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_maven_cache_local_entries_14 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_maven_cache_local_entries_15 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | namespaces.id 1:N | ||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
size | integer | NOT NULL | |||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 |
Table checks: char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(upstream_etag) <= 255; char_length(object_storage_key) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20; file_md5 IS NULL OR octet_length(file_md5) = 16
virtual_registries_packages_maven_cache_remote_entries_00 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
size | integer | NOT NULL | |||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 |
Table checks: char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(upstream_etag) <= 255; char_length(object_storage_key) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20; file_md5 IS NULL OR octet_length(file_md5) = 16
virtual_registries_packages_maven_cache_remote_entries_01 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
size | integer | NOT NULL | |||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 |
Table checks: char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(upstream_etag) <= 255; char_length(object_storage_key) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20; file_md5 IS NULL OR octet_length(file_md5) = 16
virtual_registries_packages_maven_cache_remote_entries_02 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
size | integer | NOT NULL | |||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 |
Table checks: char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(upstream_etag) <= 255; char_length(object_storage_key) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20; file_md5 IS NULL OR octet_length(file_md5) = 16
virtual_registries_packages_maven_cache_remote_entries_03 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
size | integer | NOT NULL | |||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 |
Table checks: char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(upstream_etag) <= 255; char_length(object_storage_key) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20; file_md5 IS NULL OR octet_length(file_md5) = 16
virtual_registries_packages_maven_cache_remote_entries_04 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
size | integer | NOT NULL | |||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 |
Table checks: char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(upstream_etag) <= 255; char_length(object_storage_key) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20; file_md5 IS NULL OR octet_length(file_md5) = 16
virtual_registries_packages_maven_cache_remote_entries_05 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
size | integer | NOT NULL | |||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 |
Table checks: char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(upstream_etag) <= 255; char_length(object_storage_key) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20; file_md5 IS NULL OR octet_length(file_md5) = 16
virtual_registries_packages_maven_cache_remote_entries_06 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
size | integer | NOT NULL | |||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 |
Table checks: char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(upstream_etag) <= 255; char_length(object_storage_key) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20; file_md5 IS NULL OR octet_length(file_md5) = 16
virtual_registries_packages_maven_cache_remote_entries_07 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
size | integer | NOT NULL | |||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 |
Table checks: char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(upstream_etag) <= 255; char_length(object_storage_key) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20; file_md5 IS NULL OR octet_length(file_md5) = 16
virtual_registries_packages_maven_cache_remote_entries_08 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
size | integer | NOT NULL | |||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 |
Table checks: char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(upstream_etag) <= 255; char_length(object_storage_key) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20; file_md5 IS NULL OR octet_length(file_md5) = 16
virtual_registries_packages_maven_cache_remote_entries_09 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
size | integer | NOT NULL | |||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 |
Table checks: char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(upstream_etag) <= 255; char_length(object_storage_key) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20; file_md5 IS NULL OR octet_length(file_md5) = 16
virtual_registries_packages_maven_cache_remote_entries_10 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
size | integer | NOT NULL | |||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 |
Table checks: char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(upstream_etag) <= 255; char_length(object_storage_key) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20; file_md5 IS NULL OR octet_length(file_md5) = 16
virtual_registries_packages_maven_cache_remote_entries_11 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
size | integer | NOT NULL | |||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 |
Table checks: char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(upstream_etag) <= 255; char_length(object_storage_key) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20; file_md5 IS NULL OR octet_length(file_md5) = 16
virtual_registries_packages_maven_cache_remote_entries_12 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
size | integer | NOT NULL | |||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 |
Table checks: char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(upstream_etag) <= 255; char_length(object_storage_key) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20; file_md5 IS NULL OR octet_length(file_md5) = 16
virtual_registries_packages_maven_cache_remote_entries_13 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
size | integer | NOT NULL | |||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 |
Table checks: char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(upstream_etag) <= 255; char_length(object_storage_key) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20; file_md5 IS NULL OR octet_length(file_md5) = 16
virtual_registries_packages_maven_cache_remote_entries_14 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
size | integer | NOT NULL | |||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 |
Table checks: char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(upstream_etag) <= 255; char_length(object_storage_key) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20; file_md5 IS NULL OR octet_length(file_md5) = 16
virtual_registries_packages_maven_cache_remote_entries_15 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_maven_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
size | integer | NOT NULL | |||
file_store | smallint | NOT NULL | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
file | text | NOT NULL | CHECK char_length(file) <= 1024 |
Table checks: char_length(relative_path) <= 1024; char_length(file) <= 1024; char_length(upstream_etag) <= 255; char_length(object_storage_key) <= 1024; char_length(content_type) <= 255; octet_length(file_sha1) = 20; file_md5 IS NULL OR octet_length(file_md5) = 16
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
upstream_id | bigint | NOT NULL | virtual_registries_packages_npm_upstreams.id 1:N · CASCADE | ||
package_file_id | bigint | NOT NULL | packages_package_files.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_npm_cache_local_entries_00 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_npm_cache_local_entries_01 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_npm_cache_local_entries_02 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_npm_cache_local_entries_03 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_npm_cache_local_entries_04 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_npm_cache_local_entries_05 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_npm_cache_local_entries_06 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_npm_cache_local_entries_07 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_npm_cache_local_entries_08 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_npm_cache_local_entries_09 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_npm_cache_local_entries_10 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_npm_cache_local_entries_11 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_npm_cache_local_entries_12 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_npm_cache_local_entries_13 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_npm_cache_local_entries_14 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
virtual_registries_packages_npm_cache_local_entries_15 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_local_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 |
Table checks: char_length(relative_path) <= 1024
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | namespaces.id 1:N | ||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(content_type) <= 255; char_length(object_storage_key) <= 1024; char_length(file) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20
virtual_registries_packages_npm_cache_remote_entries_00 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(content_type) <= 255; char_length(object_storage_key) <= 1024; char_length(file) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20
virtual_registries_packages_npm_cache_remote_entries_01 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(content_type) <= 255; char_length(object_storage_key) <= 1024; char_length(file) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20
virtual_registries_packages_npm_cache_remote_entries_02 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(content_type) <= 255; char_length(object_storage_key) <= 1024; char_length(file) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20
virtual_registries_packages_npm_cache_remote_entries_03 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(content_type) <= 255; char_length(object_storage_key) <= 1024; char_length(file) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20
virtual_registries_packages_npm_cache_remote_entries_04 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(content_type) <= 255; char_length(object_storage_key) <= 1024; char_length(file) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20
virtual_registries_packages_npm_cache_remote_entries_05 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(content_type) <= 255; char_length(object_storage_key) <= 1024; char_length(file) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20
virtual_registries_packages_npm_cache_remote_entries_06 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(content_type) <= 255; char_length(object_storage_key) <= 1024; char_length(file) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20
virtual_registries_packages_npm_cache_remote_entries_07 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(content_type) <= 255; char_length(object_storage_key) <= 1024; char_length(file) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20
virtual_registries_packages_npm_cache_remote_entries_08 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(content_type) <= 255; char_length(object_storage_key) <= 1024; char_length(file) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20
virtual_registries_packages_npm_cache_remote_entries_09 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(content_type) <= 255; char_length(object_storage_key) <= 1024; char_length(file) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20
virtual_registries_packages_npm_cache_remote_entries_10 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(content_type) <= 255; char_length(object_storage_key) <= 1024; char_length(file) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20
virtual_registries_packages_npm_cache_remote_entries_11 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(content_type) <= 255; char_length(object_storage_key) <= 1024; char_length(file) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20
virtual_registries_packages_npm_cache_remote_entries_12 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(content_type) <= 255; char_length(object_storage_key) <= 1024; char_length(file) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20
virtual_registries_packages_npm_cache_remote_entries_13 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(content_type) <= 255; char_length(object_storage_key) <= 1024; char_length(file) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20
virtual_registries_packages_npm_cache_remote_entries_14 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(content_type) <= 255; char_length(object_storage_key) <= 1024; char_length(file) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20
virtual_registries_packages_npm_cache_remote_entries_15 references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
iid PK | bigint | NOT NULL | nextval(CAST('virtual_registries_packages_npm_cache_remote_entries_iid_seq' AS regclass)) | ||
group_id PK | bigint | NOT NULL | |||
upstream_id | bigint | NOT NULL | |||
downloads_count | bigint | NOT NULL | 0 | ||
upstream_checked_at | timestamptz | NOT NULL | now() | ||
downloaded_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | NOT NULL | 1 | ||
size | integer | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
file_md5 | bytea | null | CHECK file_md5 IS NULL OR octet_length(file_md5) = 16 | ||
file_sha1 | bytea | NOT NULL | CHECK octet_length(file_sha1) = 20 | ||
upstream_etag | text | null | CHECK char_length(upstream_etag) <= 255 | ||
content_type | text | NOT NULL | CAST('application/octet-stream' AS text) | CHECK char_length(content_type) <= 255 | |
relative_path | text | NOT NULL | CHECK char_length(relative_path) <= 1024 | ||
file | text | NOT NULL | CHECK char_length(file) <= 1024 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 1024 |
Table checks: char_length(upstream_etag) <= 255; char_length(relative_path) <= 1024; char_length(content_type) <= 255; char_length(object_storage_key) <= 1024; char_length(file) <= 1024; file_md5 IS NULL OR octet_length(file_md5) = 16; octet_length(file_sha1) = 20
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
next_run_at | timestamptz | null | |||
last_run_at | timestamptz | null | |||
last_run_deleted_size | bigint | null | 0 | CHECK last_run_deleted_size >= 0 | |
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
keep_n_days_after_download | integer | NOT NULL | 30 | CHECK keep_n_days_after_download > 0 | |
last_run_deleted_entries_count | integer | null | 0 | CHECK last_run_deleted_entries_count >= 0 | |
status | smallint | NOT NULL | 0 | ||
cadence | smallint | NOT NULL | 7 | CHECK cadence = ANY(ARRAY[1, 7, 14, 30, 90]) | |
enabled | boolean | NOT NULL | FALSE | ||
notify_on_success | boolean | NOT NULL | FALSE | ||
notify_on_failure | boolean | NOT NULL | FALSE | ||
failure_message | text | null | CHECK char_length(failure_message) <= 255 | ||
last_run_detailed_metrics | jsonb | null | CAST('{}' AS jsonb) |
Table checks: char_length(failure_message) <= 255; last_run_deleted_size >= 0; cadence = ANY(ARRAY[1, 7, 14, 30, 90]); keep_n_days_after_download > 0; last_run_deleted_entries_count >= 0
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
description | text | null | CHECK char_length(description) <= 1024 |
Table checks: char_length(description) <= 1024; char_length(name) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
registry_id | bigint | NOT NULL | virtual_registries_container_registries.id 1:N · CASCADE | ||
upstream_id | bigint | NOT NULL | virtual_registries_container_upstreams.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
position | smallint | NOT NULL | 1 | CHECK 1 <= position AND position <= 5 |
Table checks: 1 <= position AND position <= 5
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
cache_validity_hours | smallint | NOT NULL | 24 | CHECK cache_validity_hours >= 0 | |
username | jsonb | null | |||
password | jsonb | null | |||
url | text | NOT NULL | CHECK char_length(url) <= 255 | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
description | text | null | CHECK char_length(description) <= 1024 | ||
auth_url | text | null | CHECK char_length(auth_url) <= 512 |
Table checks: char_length(description) <= 1024; cache_validity_hours >= 0; char_length(name) <= 255; num_nonnulls(username, password) = 2 OR num_nulls(username, password) = 2; char_length(auth_url) <= 512; char_length(url) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
virtual_registries_packages_maven_cache_remote_entry_iid | bigint | NOT NULL | virtual_registries_packages_maven_cache_remote_entries.group_id 1:N · CASCADE | ||
group_id | bigint | NOT NULL | namespaces.id 1:N | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 |
Table checks: char_length(verification_failure) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
local_group_id | bigint | null | |||
local_project_id | bigint | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
metadata_cache_validity_hours | smallint | NOT NULL | 1 | CHECK metadata_cache_validity_hours > 0 | |
cache_validity_hours | smallint | NOT NULL | 24 | CHECK cache_validity_hours >= 0 | |
name | text | NOT NULL | CAST('' AS text) | CHECK char_length(name) <= 255 | |
description | text | null | CHECK char_length(description) <= 1024 |
Table checks: metadata_cache_validity_hours > 0; char_length(description) <= 1024; char_length(name) <= 255; num_nonnulls(local_group_id, local_project_id) = 1; cache_validity_hours >= 0
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
name | text | NOT NULL | CAST('' AS text) | CHECK char_length(name) <= 255 | |
description | text | null | CHECK char_length(description) <= 1024 |
Table checks: char_length(name) <= 255; char_length(description) <= 1024
upstream_id — virtual_registries_packages_maven_registry_upstreams.upstream_id may be NULL, so the relationship to virtual_registries_packages_maven_upstreams is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.local_upstream_id — virtual_registries_packages_maven_registry_upstreams.local_upstream_id may be NULL, so the relationship to virtual_registries_packages_maven_local_upstreams is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
registry_id | bigint | NOT NULL | virtual_registries_packages_maven_registries.id 1:N · CASCADE | ||
upstream_id | bigint | null | virtual_registries_packages_maven_upstreams.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
position | smallint | NOT NULL | 1 | CHECK 1 <= position AND position <= 20 | |
local_upstream_id | bigint | null | virtual_registries_packages_maven_local_upstreams.id 1:N · CASCADE |
Table checks: num_nonnulls(upstream_id, local_upstream_id) = 1; 1 <= position AND position <= 20
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
remote_upstream_id | bigint | NOT NULL | virtual_registries_packages_maven_upstreams.id 1:N · CASCADE | ||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
pattern_type | smallint | NOT NULL | 0 | ||
rule_type | smallint | NOT NULL | 0 | ||
target_coordinate | smallint | NOT NULL | 0 | ||
pattern | text | NOT NULL | CHECK char_length(pattern) <= 255 |
Table checks: char_length(pattern) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
url | text | NOT NULL | CHECK char_length(url) <= 255 | ||
cache_validity_hours | smallint | NOT NULL | 24 | CHECK cache_validity_hours >= 0 | |
username | jsonb | null | |||
password | jsonb | null | |||
name | text | NOT NULL | CAST('' AS text) | CHECK char_length(name) <= 255 | |
description | text | null | CHECK char_length(description) <= 1024 | ||
metadata_cache_validity_hours | smallint | NOT NULL | 24 | CHECK metadata_cache_validity_hours > 0 |
Table checks: char_length(url) <= 255; num_nonnulls(username, password) = 2 OR num_nulls(username, password) = 2; metadata_cache_validity_hours > 0; cache_validity_hours >= 0; char_length(description) <= 1024; char_length(name) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
description | text | null | CHECK char_length(description) <= 1024 |
Table checks: char_length(name) <= 255; char_length(description) <= 1024
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
registry_id | bigint | NOT NULL | virtual_registries_packages_npm_registries.id 1:N · CASCADE | ||
upstream_id | bigint | NOT NULL | virtual_registries_packages_npm_upstreams.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
position | smallint | NOT NULL | 1 | CHECK 1 <= position AND position <= 20 |
Table checks: 1 <= position AND position <= 20
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
cache_validity_hours | smallint | NOT NULL | 24 | CHECK cache_validity_hours >= 0 | |
metadata_cache_validity_hours | smallint | NOT NULL | 24 | CHECK metadata_cache_validity_hours > 0 | |
username | jsonb | null | |||
password | jsonb | null | |||
url | text | NOT NULL | CHECK char_length(url) <= 255 | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
description | text | null | CHECK char_length(description) <= 1024 | ||
auth_token | jsonb | null |
Table checks: char_length(name) <= 255; char_length(description) <= 1024; cache_validity_hours >= 0; char_length(url) <= 255; metadata_cache_validity_hours > 0
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
enabled | boolean | NOT NULL | TRUE |
Tables whose names start with abuse_.
abuse_events.organization_id → organizations.id — one organizations, many abuse_events · optional · ON DELETE CASCADE · indexedabuse_events.user_id → users.id — one users, many abuse_events · optional · ON DELETE SET NULL · indexedabuse_events.abuse_report_id → abuse_reports.id — one abuse_reports, many abuse_events · optional · ON DELETE NO ACTION · indexedabuse_report_events.user_id → users.id — one users, many abuse_report_events · optional · ON DELETE SET NULL · indexedabuse_report_events.organization_id → organizations.id — one organizations, many abuse_report_events · optional · ON DELETE CASCADE · indexedabuse_report_events.abuse_report_id → abuse_reports.id — one abuse_reports, many abuse_report_events · required · ON DELETE CASCADE · indexedabuse_report_upload_states.organization_id → organizations.id — one organizations, many abuse_report_upload_states · required · ON DELETE CASCADE · indexedabuse_report_upload_states.abuse_report_upload_id → abuse_report_uploads.id — one abuse_report_uploads, at most one abuse_report_upload_states · required · ON DELETE CASCADE · indexedabuse_report_user_mentions.abuse_report_id → abuse_reports.id — one abuse_reports, many abuse_report_user_mentions · required · ON DELETE CASCADE · indexedabuse_report_user_mentions.note_id → notes.id — one notes, many abuse_report_user_mentions · required · ON DELETE CASCADE · indexedabuse_report_user_mentions.organization_id → organizations.id — one organizations, many abuse_report_user_mentions · optional · ON DELETE CASCADE · indexedabuse_reports.assignee_id → users.id — one users, many abuse_reports · optional · ON DELETE SET NULL · indexedabuse_reports.reporter_id → users.id — one users, many abuse_reports · optional · ON DELETE CASCADE · indexedabuse_reports.resolved_by_id → users.id — one users, many abuse_reports · optional · ON DELETE SET NULL · indexedabuse_reports.organization_id → organizations.id — one organizations, many abuse_reports · optional · ON DELETE CASCADE · indexedorganization_id — abuse_events.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id — abuse_events.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.abuse_report_id — abuse_events.abuse_report_id may be NULL, so the relationship to abuse_reports is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id, user_id, abuse_report_id — abuse_events has several nullable FKs (organization_id, user_id, abuse_report_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | null | users.id 1:N · SET NULL | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
abuse_report_id | bigint | null | abuse_reports.id 1:N | ||
source | smallint | NOT NULL | |||
category | smallint | null | |||
metadata | jsonb | null | |||
organization_id | bigint | null | organizations.id 1:N · CASCADE | CHECK organization_id IS NOT NULL |
Table checks: organization_id IS NOT NULL
user_id — abuse_report_events.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id — abuse_report_events.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id, organization_id — abuse_report_events has several nullable FKs (user_id, organization_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
abuse_report_id | bigint | NOT NULL | abuse_reports.id 1:N · CASCADE | ||
user_id | bigint | null | users.id 1:N · SET NULL | ||
created_at | timestamptz | NOT NULL | |||
action | smallint | NOT NULL | 1 | ||
reason | smallint | null | |||
comment | text | null | CHECK char_length(comment) <= 1024 | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE | CHECK organization_id IS NOT NULL |
Table checks: char_length(comment) <= 1024; organization_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
abuse_report_upload_id | bigint | NOT NULL | abuse_report_uploads.id 1:1 · CASCADE | ||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 |
Table checks: char_length(verification_failure) <= 255
model_type — model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.model_type, model_id — model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | nextval(CAST('uploads_id_seq' AS regclass)) | ||
size | bigint | NOT NULL | |||
model_id | bigint | NOT NULL | |||
uploaded_by_user_id | bigint | null | |||
organization_id | bigint | null | CHECK organization_id IS NOT NULL | ||
namespace_id | bigint | null | |||
project_id | bigint | null | |||
created_at | timestamp | null | |||
store | integer | NOT NULL | 1 | ||
version | integer | null | 1 | ||
path | text | NOT NULL | CHECK char_length(path) <= 511 | ||
checksum | text | null | CHECK char_length(checksum) <= 64 | ||
model_type PK | text | NOT NULL | |||
uploader | text | NOT NULL | |||
mount_point | text | null | |||
secret | text | null |
Table checks: char_length(path) <= 511; organization_id IS NOT NULL; char_length(checksum) <= 64
organization_id — abuse_report_user_mentions.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
abuse_report_id | bigint | NOT NULL | abuse_reports.id 1:N · CASCADE | ||
note_id | bigint | NOT NULL | notes.id 1:N · CASCADE | ||
mentioned_users_ids | bigint[] | null | |||
mentioned_projects_ids | bigint[] | null | |||
mentioned_groups_ids | bigint[] | null | |||
organization_id | bigint | null | organizations.id 1:N · CASCADE | CHECK organization_id IS NOT NULL |
Table checks: organization_id IS NOT NULL
assignee_id — abuse_reports.assignee_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.reporter_id — abuse_reports.reporter_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.resolved_by_id — abuse_reports.resolved_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id — abuse_reports.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.assignee_id, reporter_id, resolved_by_id, organization_id — abuse_reports has several nullable FKs (assignee_id, reporter_id, resolved_by_id, organization_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
reporter_id | bigint | null | users.id 1:N · CASCADE | CHECK reporter_id IS NOT NULL | |
user_id | bigint | null | |||
message | text | null | CHECK char_length(message) <= 2048 | ||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
message_html | text | null | |||
cached_markdown_version | integer | null | |||
category | smallint | NOT NULL | 1 | ||
reported_from_url | text | NOT NULL | CAST('' AS text) | CHECK char_length(reported_from_url) <= 512 | |
links_to_spam | text[] | NOT NULL | CAST('{}' AS text[]) | CHECK cardinality(links_to_spam) <= 20 | |
status | smallint | NOT NULL | 1 | ||
resolved_at | timestamptz | null | |||
screenshot | text | null | CHECK char_length(screenshot) <= 255 | ||
resolved_by_id | bigint | null | users.id 1:N · SET NULL | ||
assignee_id | bigint | null | users.id 1:N · SET NULL | ||
mitigation_steps | text | null | CHECK char_length(mitigation_steps) <= 1000 | ||
evidence | jsonb | null | |||
assignee_id_convert_to_bigint | bigint | null | |||
id_convert_to_bigint | bigint | NOT NULL | 0 | ||
reporter_id_convert_to_bigint | bigint | null | |||
resolved_by_id_convert_to_bigint | bigint | null | |||
user_id_convert_to_bigint | bigint | null | |||
organization_id | bigint | null | organizations.id 1:N · CASCADE | CHECK organization_id IS NOT NULL |
Table checks: cardinality(links_to_spam) <= 20; organization_id IS NOT NULL; char_length(screenshot) <= 255; char_length(reported_from_url) <= 512; char_length(mitigation_steps) <= 1000; reporter_id IS NOT NULL; char_length(message) <= 2048
Tables whose names start with achievement_.
achievement_upload_states.namespace_id → namespaces.id — one namespaces, many achievement_upload_states · required · ON DELETE CASCADE · indexedachievement_upload_states.achievement_upload_id → achievement_uploads.id — one achievement_uploads, at most one achievement_upload_states · required · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
achievement_upload_id | bigint | NOT NULL | achievement_uploads.id 1:1 · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 |
Table checks: char_length(verification_failure) <= 255
model_type — model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.model_type, model_id — model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | nextval(CAST('uploads_id_seq' AS regclass)) | ||
size | bigint | NOT NULL | |||
model_id | bigint | NOT NULL | |||
uploaded_by_user_id | bigint | null | |||
organization_id | bigint | null | |||
namespace_id | bigint | null | CHECK namespace_id IS NOT NULL | ||
project_id | bigint | null | |||
created_at | timestamp | null | |||
store | integer | NOT NULL | 1 | ||
version | integer | null | 1 | ||
path | text | NOT NULL | CHECK char_length(path) <= 511 | ||
checksum | text | null | CHECK char_length(checksum) <= 64 | ||
model_type PK | text | NOT NULL | |||
uploader | text | NOT NULL | |||
mount_point | text | null | |||
secret | text | null |
Table checks: char_length(path) <= 511; namespace_id IS NOT NULL; char_length(checksum) <= 64
Tables whose names start with agent_.
agent_activity_events.project_id → projects.id — one projects, many agent_activity_events · optional · ON DELETE SET NULL · indexedagent_activity_events.merge_request_id → merge_requests.id — one merge_requests, many agent_activity_events · optional · ON DELETE SET NULL · indexedagent_activity_events.agent_project_id → projects.id — one projects, many agent_activity_events · optional · ON DELETE CASCADE · indexedagent_activity_events.agent_id → cluster_agents.id — one cluster_agents, many agent_activity_events · required · ON DELETE CASCADE · indexedagent_activity_events.agent_token_id → cluster_agent_tokens.id — one cluster_agent_tokens, many agent_activity_events · optional · ON DELETE SET NULL · indexedagent_activity_events.user_id → users.id — one users, many agent_activity_events · optional · ON DELETE SET NULL · indexedagent_group_authorizations.group_id → namespaces.id — one namespaces, many agent_group_authorizations · required · ON DELETE CASCADE · indexedagent_group_authorizations.agent_id → cluster_agents.id — one cluster_agents, many agent_group_authorizations · required · ON DELETE CASCADE · indexedagent_organization_authorizations.organization_id → organizations.id — one organizations, many agent_organization_authorizations · required · ON DELETE CASCADE · indexedagent_organization_authorizations.agent_id → cluster_agents.id — one cluster_agents, at most one agent_organization_authorizations · required · ON DELETE CASCADE · indexedagent_project_authorizations.project_id → projects.id — one projects, many agent_project_authorizations · required · ON DELETE CASCADE · indexedagent_project_authorizations.agent_id → cluster_agents.id — one cluster_agents, many agent_project_authorizations · required · ON DELETE CASCADE · indexedagent_user_access_group_authorizations.group_id → namespaces.id — one namespaces, many agent_user_access_group_authorizations · required · ON DELETE CASCADE · indexedagent_user_access_group_authorizations.agent_id → cluster_agents.id — one cluster_agents, many agent_user_access_group_authorizations · required · ON DELETE CASCADE · indexedagent_user_access_project_authorizations.agent_id → cluster_agents.id — one cluster_agents, many agent_user_access_project_authorizations · required · ON DELETE CASCADE · indexedagent_user_access_project_authorizations.project_id → projects.id — one projects, many agent_user_access_project_authorizations · required · ON DELETE CASCADE · indexedproject_id — agent_activity_events.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.merge_request_id — agent_activity_events.merge_request_id may be NULL, so the relationship to merge_requests is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.agent_project_id — agent_activity_events.agent_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.agent_token_id — agent_activity_events.agent_token_id may be NULL, so the relationship to cluster_agent_tokens is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id — agent_activity_events.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id, merge_request_id, agent_project_id, agent_token_id, user_id — agent_activity_events has several nullable FKs (project_id, merge_request_id, agent_project_id, agent_token_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
agent_id | bigint | NOT NULL | cluster_agents.id 1:N · CASCADE | ||
user_id | bigint | null | users.id 1:N · SET NULL | ||
project_id | bigint | null | projects.id 1:N · SET NULL | ||
merge_request_id | bigint | null | merge_requests.id 1:N · SET NULL | ||
agent_token_id | bigint | null | cluster_agent_tokens.id 1:N · SET NULL | ||
recorded_at | timestamptz | NOT NULL | |||
kind | smallint | NOT NULL | |||
level | smallint | NOT NULL | |||
detail | text | null | CHECK char_length(detail) <= 255 | ||
agent_project_id | bigint | null | projects.id 1:N · CASCADE | CHECK agent_project_id IS NOT NULL |
Table checks: char_length(detail) <= 255; agent_project_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
agent_id | bigint | NOT NULL | cluster_agents.id 1:N · CASCADE | ||
config | jsonb | NOT NULL |
agent_id, organization_id — agent_organization_authorizations looks like a many-to-many link table but has no unique constraint across (agent_id, organization_id). The same pair can be inserted twice; every join through it will double-count.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
agent_id | bigint | NOT NULL | cluster_agents.id 1:1 · CASCADE | ||
config | jsonb | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
agent_id | bigint | NOT NULL | cluster_agents.id 1:N · CASCADE | ||
config | jsonb | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
agent_id | bigint | NOT NULL | cluster_agents.id 1:N · CASCADE | ||
config | jsonb | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
agent_id | bigint | NOT NULL | cluster_agents.id 1:N · CASCADE | ||
config | jsonb | NOT NULL |
Tables whose names start with alert_.
alert_management_alert_assignees.project_id → projects.id — one projects, many alert_management_alert_assignees · optional · ON DELETE CASCADE · indexedalert_management_alert_assignees.alert_id → alert_management_alerts.id — one alert_management_alerts, many alert_management_alert_assignees · required · ON DELETE CASCADE · indexedalert_management_alert_assignees.user_id → users.id — one users, many alert_management_alert_assignees · required · ON DELETE CASCADE · indexedalert_management_alert_metric_images.project_id → projects.id — one projects, many alert_management_alert_metric_images · optional · ON DELETE CASCADE · indexedalert_management_alert_metric_images.alert_id → alert_management_alerts.id — one alert_management_alerts, many alert_management_alert_metric_images · required · ON DELETE CASCADE · indexedalert_management_alert_user_mentions.project_id → projects.id — one projects, many alert_management_alert_user_mentions · optional · ON DELETE CASCADE · indexedalert_management_alert_user_mentions.alert_management_alert_id → alert_management_alerts.id — one alert_management_alerts, many alert_management_alert_user_mentions · required · ON DELETE CASCADE · indexedalert_management_alert_user_mentions.note_id → notes.id — one notes, many alert_management_alert_user_mentions · optional · ON DELETE CASCADE · indexedalert_management_alerts.issue_id → issues.id — one issues, many alert_management_alerts · optional · ON DELETE SET NULL · indexedalert_management_alerts.project_id → projects.id — one projects, many alert_management_alerts · required · ON DELETE CASCADE · indexedalert_management_alerts.environment_id → environments.id — one environments, many alert_management_alerts · optional · ON DELETE SET NULL · indexedalert_management_http_integrations.project_id → projects.id — one projects, many alert_management_http_integrations · required · ON DELETE CASCADE · indexedalert_management_metric_image_upload_states.project_id → projects.id — one projects, many alert_management_metric_image_upload_states · required · ON DELETE CASCADE · indexedalert_management_metric_image_upload_states.alert_management_metric_image_upload_id → alert_management_alert_metric_image_uploads.id — one alert_management_alert_metric_image_uploads, at most one alert_management_metric_image_upload_states · required · ON DELETE CASCADE · indexedproject_id — alert_management_alert_assignees.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
alert_id | bigint | NOT NULL | alert_management_alerts.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
model_type — model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.model_type, model_id — model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | nextval(CAST('uploads_id_seq' AS regclass)) | ||
size | bigint | NOT NULL | |||
model_id | bigint | NOT NULL | |||
uploaded_by_user_id | bigint | null | |||
organization_id | bigint | null | |||
namespace_id | bigint | null | |||
project_id | bigint | null | CHECK project_id IS NOT NULL | ||
created_at | timestamp | null | |||
store | integer | NOT NULL | 1 | ||
version | integer | null | 1 | ||
path | text | NOT NULL | CHECK char_length(path) <= 511 | ||
checksum | text | null | CHECK char_length(checksum) <= 64 | ||
model_type PK | text | NOT NULL | |||
uploader | text | NOT NULL | |||
mount_point | text | null | |||
secret | text | null |
Table checks: char_length(path) <= 511; project_id IS NOT NULL; char_length(checksum) <= 64
project_id — alert_management_alert_metric_images.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
alert_id | bigint | NOT NULL | alert_management_alerts.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | smallint | null | |||
file | text | NOT NULL | CHECK char_length(file) <= 255 | ||
url | text | null | CHECK char_length(url) <= 255 | ||
url_text | text | null | CHECK char_length(url_text) <= 128 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: char_length(url_text) <= 128; char_length(url) <= 255; char_length(file) <= 255; project_id IS NOT NULL
project_id — alert_management_alert_user_mentions.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.note_id — alert_management_alert_user_mentions.note_id may be NULL, so the relationship to notes is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id, note_id — alert_management_alert_user_mentions has several nullable FKs (project_id, note_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
alert_management_alert_id | bigint | NOT NULL | alert_management_alerts.id 1:N · CASCADE | ||
note_id | bigint | null | notes.id 1:N · CASCADE | ||
mentioned_users_ids | bigint[] | null | |||
mentioned_projects_ids | bigint[] | null | |||
mentioned_groups_ids | bigint[] | null | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
issue_id — alert_management_alerts.issue_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.environment_id — alert_management_alerts.environment_id may be NULL, so the relationship to environments is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.issue_id, environment_id — alert_management_alerts has several nullable FKs (issue_id, environment_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
started_at | timestamptz | NOT NULL | |||
ended_at | timestamptz | null | |||
events | integer | NOT NULL | 1 | ||
iid | integer | NOT NULL | |||
severity | smallint | NOT NULL | 0 | ||
status | smallint | NOT NULL | 0 | ||
fingerprint | bytea | null | |||
issue_id | bigint | null | issues.id 1:N · SET NULL | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
title | text | NOT NULL | CHECK char_length(title) <= 200 | ||
description | text | null | CHECK char_length(description) <= 1000 | ||
service | text | null | CHECK char_length(service) <= 100 | ||
monitoring_tool | text | null | CHECK char_length(monitoring_tool) <= 100 | ||
hosts | text[] | NOT NULL | CAST('{}' AS text[]) | ||
payload | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
environment_id | bigint | null | environments.id 1:N · SET NULL | ||
domain | smallint | null | 0 |
Table checks: char_length(monitoring_tool) <= 100; char_length(description) <= 1000; char_length(service) <= 100; char_length(title) <= 200
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
active | boolean | NOT NULL | FALSE | ||
encrypted_token | text | NOT NULL | CHECK char_length(encrypted_token) <= 255 | ||
encrypted_token_iv | text | NOT NULL | CHECK char_length(encrypted_token_iv) <= 255 | ||
endpoint_identifier | text | NOT NULL | CHECK char_length(endpoint_identifier) <= 255 | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
payload_example | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
payload_attribute_mapping | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
type_identifier | smallint | NOT NULL | 0 |
Table checks: char_length(encrypted_token_iv) <= 255; char_length(name) <= 255; char_length(endpoint_identifier) <= 255; char_length(encrypted_token) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
alert_management_metric_image_upload_id | bigint | NOT NULL | alert_management_alert_metric_image_uploads.id 1:1 · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 |
Table checks: char_length(verification_failure) <= 255
Tables whose names start with analyzer_.
No foreign keys in this domain.
analyzer_namespace_statuses references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
namespace_id | bigint | NOT NULL | |||
analyzer_type | smallint | NOT NULL | |||
success | bigint | NOT NULL | 0 | ||
failure | bigint | NOT NULL | 0 | ||
traversal_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
stale | bigint | NOT NULL | 0 |
analyzer_project_statuses references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | |||
analyzer_type | smallint | NOT NULL | |||
status | smallint | NOT NULL | |||
last_call | timestamptz | NOT NULL | |||
traversal_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
build_id | bigint | null | |||
archived | boolean | NOT NULL | FALSE | ||
consecutive_absence_count | integer | NOT NULL | 0 |
Tables whose names start with appearance_.
appearance_upload_states.appearance_upload_id → appearance_uploads.id — one appearance_uploads, at most one appearance_upload_states · required · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
appearance_upload_id | bigint | NOT NULL | appearance_uploads.id 1:1 · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 |
Table checks: char_length(verification_failure) <= 255
model_type — model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.model_type, model_id — model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | nextval(CAST('uploads_id_seq' AS regclass)) | ||
size | bigint | NOT NULL | |||
model_id | bigint | NOT NULL | |||
uploaded_by_user_id | bigint | null | |||
organization_id | bigint | null | |||
namespace_id | bigint | null | |||
project_id | bigint | null | |||
created_at | timestamp | null | |||
store | integer | NOT NULL | 1 | ||
version | integer | null | 1 | ||
path | text | NOT NULL | CHECK char_length(path) <= 511 | ||
checksum | text | null | CHECK char_length(checksum) <= 64 | ||
model_type PK | text | NOT NULL | |||
uploader | text | NOT NULL | |||
mount_point | text | null | |||
secret | text | null |
Table checks: char_length(path) <= 511; num_nonnulls(namespace_id, organization_id, project_id) = 0; char_length(checksum) <= 64
Tables whose names start with application_.
application_settings.duo_template_project_id → projects.id — one projects, many application_settings · optional · ON DELETE SET NULL · indexedapplication_settings.o11y_oauth_application_id → oauth_applications.id — one oauth_applications, many application_settings · optional · ON DELETE SET NULL · indexedapplication_settings.workspaces_oauth_application_id → oauth_applications.id — one oauth_applications, many application_settings · optional · ON DELETE SET NULL · indexedapplication_settings.web_ide_oauth_application_id → oauth_applications.id — one oauth_applications, many application_settings · optional · ON DELETE SET NULL · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
cached_markdown_version | integer | null | |||
terms | text | NOT NULL | |||
terms_html | text | null |
duo_template_project_id — application_settings.duo_template_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.o11y_oauth_application_id — application_settings.o11y_oauth_application_id may be NULL, so the relationship to oauth_applications is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.workspaces_oauth_application_id — application_settings.workspaces_oauth_application_id may be NULL, so the relationship to oauth_applications is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.web_ide_oauth_application_id — application_settings.web_ide_oauth_application_id may be NULL, so the relationship to oauth_applications is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.duo_template_project_id, o11y_oauth_application_id, workspaces_oauth_application_id, web_ide_oauth_application_id — application_settings has several nullable FKs (duo_template_project_id, o11y_oauth_application_id, workspaces_oauth_application_id, web_ide_oauth_application_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.namespace_storage_forks_cost_factor — namespace_storage_forks_cost_factor is binary floating point. 0.1 + 0.2 ≠ 0.3; sums drift; reconciliation against the ledger will be off by cents.sentry_clientside_traces_sample_rate — sentry_clientside_traces_sample_rate is binary floating point. 0.1 + 0.2 ≠ 0.3; sums drift; reconciliation against the ledger will be off by cents.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
default_projects_limit | integer | null | |||
signup_enabled | boolean | null | |||
gravatar_enabled | boolean | null | |||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
home_page_url | varchar | null | |||
default_branch_protection | integer | null | 2 | ||
restricted_visibility_levels | text | null | |||
version_check_enabled | boolean | null | TRUE | ||
max_attachment_size | integer | NOT NULL | 100 | ||
default_project_visibility | integer | NOT NULL | 0 | ||
default_snippet_visibility | integer | NOT NULL | 0 | ||
user_oauth_applications | boolean | null | TRUE | ||
after_sign_out_path | varchar | null | |||
session_expire_delay | integer | NOT NULL | 10080 | ||
import_sources | text | null | |||
help_page_text | text | null | |||
require_admin_two_factor_authentication | boolean | NOT NULL | FALSE | ||
shared_runners_enabled | boolean | NOT NULL | TRUE | ||
max_artifacts_size | integer | NOT NULL | 100 | ||
max_pages_size | integer | NOT NULL | 100 | ||
require_two_factor_authentication | boolean | null | FALSE | ||
two_factor_grace_period | integer | null | 48 | ||
metrics_enabled | boolean | null | FALSE | ||
metrics_host | varchar | null | CAST('localhost' AS varchar) | ||
metrics_pool_size | integer | null | 16 | ||
metrics_timeout | integer | null | 10 | ||
metrics_method_call_threshold | integer | null | 10 | ||
recaptcha_enabled | boolean | null | FALSE | ||
metrics_port | integer | null | 8089 | ||
akismet_enabled | boolean | null | FALSE | ||
metrics_sample_interval | integer | null | 15 | ||
email_author_in_body | boolean | null | FALSE | ||
default_group_visibility | integer | null | |||
repository_checks_enabled | boolean | null | FALSE | ||
shared_runners_text | text | null | |||
metrics_packet_size | integer | null | 1 | ||
disabled_oauth_sign_in_sources | text | null | |||
health_check_access_token | varchar | null | |||
container_registry_token_expire_delay | integer | null | 5 | ||
after_sign_up_text | text | null | |||
user_default_external | boolean | NOT NULL | FALSE | ||
repository_storages | varchar | null | CAST('default' AS varchar) | ||
enabled_git_access_protocol | varchar | null | |||
usage_ping_enabled | boolean | NOT NULL | TRUE | ||
help_page_text_html | text | null | |||
shared_runners_text_html | text | null | |||
after_sign_up_text_html | text | null | |||
rsa_key_restriction | integer | NOT NULL | 0 | ||
dsa_key_restriction | integer | NOT NULL | CAST('-1' AS integer) | ||
ecdsa_key_restriction | integer | NOT NULL | 0 | ||
ed25519_key_restriction | integer | NOT NULL | 0 | ||
housekeeping_enabled | boolean | NOT NULL | TRUE | ||
housekeeping_bitmaps_enabled | boolean | NOT NULL | TRUE | ||
housekeeping_incremental_repack_period | integer | NOT NULL | 10 | ||
housekeeping_full_repack_period | integer | NOT NULL | 50 | ||
housekeeping_gc_period | integer | NOT NULL | 200 | ||
html_emails_enabled | boolean | null | TRUE | ||
plantuml_url | varchar | null | |||
plantuml_enabled | boolean | null | |||
shared_runners_minutes | integer | NOT NULL | 0 | ||
repository_size_limit | bigint | null | 0 | ||
terminal_max_session_time | integer | NOT NULL | 0 | ||
unique_ips_limit_per_user | integer | null | |||
unique_ips_limit_time_window | integer | null | |||
unique_ips_limit_enabled | boolean | NOT NULL | FALSE | ||
default_artifacts_expire_in | varchar | NOT NULL | CAST('0' AS varchar) | ||
elasticsearch_url | varchar | null | CAST('http://localhost:9200' AS varchar) | ||
geo_status_timeout | integer | null | 10 | ||
uuid | varchar | null | |||
polling_interval_multiplier | numeric | NOT NULL | 1.0 | ||
cached_markdown_version | integer | null | |||
check_namespace_plan | boolean | NOT NULL | FALSE | ||
mirror_max_delay | integer | NOT NULL | 300 | ||
mirror_max_capacity | integer | NOT NULL | 100 | ||
mirror_capacity_threshold | integer | NOT NULL | 50 | ||
prometheus_metrics_enabled | boolean | NOT NULL | TRUE | ||
authorized_keys_enabled | boolean | NOT NULL | TRUE | ||
help_page_hide_commercial_content | boolean | null | FALSE | ||
help_page_support_url | varchar | null | |||
slack_app_enabled | boolean | null | FALSE | ||
slack_app_id | varchar | null | |||
performance_bar_allowed_group_id | bigint | null | |||
allow_group_owners_to_manage_ldap | boolean | NOT NULL | TRUE | ||
hashed_storage_enabled | boolean | NOT NULL | TRUE | ||
project_export_enabled | boolean | NOT NULL | TRUE | ||
auto_devops_enabled | boolean | NOT NULL | TRUE | ||
throttle_unauthenticated_enabled | boolean | NOT NULL | FALSE | ||
throttle_unauthenticated_requests_per_period | integer | NOT NULL | 3600 | ||
throttle_unauthenticated_period_in_seconds | integer | NOT NULL | 3600 | ||
throttle_authenticated_api_enabled | boolean | NOT NULL | FALSE | ||
throttle_authenticated_api_requests_per_period | integer | NOT NULL | 7200 | ||
throttle_authenticated_api_period_in_seconds | integer | NOT NULL | 3600 | ||
throttle_authenticated_web_enabled | boolean | NOT NULL | FALSE | ||
throttle_authenticated_web_requests_per_period | integer | NOT NULL | 7200 | ||
throttle_authenticated_web_period_in_seconds | integer | NOT NULL | 3600 | ||
gitaly_timeout_default | integer | NOT NULL | 55 | ||
gitaly_timeout_medium | integer | NOT NULL | 30 | ||
gitaly_timeout_fast | integer | NOT NULL | 10 | ||
mirror_available | boolean | NOT NULL | TRUE | ||
password_authentication_enabled_for_web | boolean | null | |||
password_authentication_enabled_for_git | boolean | NOT NULL | TRUE | ||
auto_devops_domain | varchar | null | |||
external_authorization_service_enabled | boolean | NOT NULL | FALSE | ||
external_authorization_service_url | varchar | null | |||
external_authorization_service_default_label | varchar | null | |||
pages_domain_verification_enabled | boolean | NOT NULL | TRUE | ||
user_default_internal_regex | varchar | null | |||
external_authorization_service_timeout | double precision | null | 0.5 | ||
external_auth_client_cert | text | null | |||
encrypted_external_auth_client_key | text | null | |||
encrypted_external_auth_client_key_iv | varchar | null | |||
encrypted_external_auth_client_key_pass | varchar | null | |||
encrypted_external_auth_client_key_pass_iv | varchar | null | |||
email_additional_text | varchar | null | |||
enforce_terms | boolean | null | FALSE | ||
file_template_project_id | bigint | null | |||
pseudonymizer_enabled | boolean | NOT NULL | FALSE | ||
hide_third_party_offers | boolean | NOT NULL | FALSE | ||
snowplow_enabled | boolean | NOT NULL | FALSE | ||
snowplow_collector_hostname | varchar | null | |||
snowplow_cookie_domain | varchar | null | |||
user_show_add_ssh_key_message | boolean | NOT NULL | TRUE | ||
custom_project_templates_group_id | bigint | null | |||
usage_stats_set_by_user_id | bigint | null | |||
receive_max_input_size | integer | null | |||
diff_max_patch_bytes | integer | NOT NULL | 204800 | ||
archive_builds_in_seconds | integer | null | |||
commit_email_hostname | varchar | null | |||
protected_ci_variables | boolean | NOT NULL | TRUE | ||
runners_registration_token_encrypted | varchar | null | |||
local_markdown_version | integer | NOT NULL | 0 | ||
first_day_of_week | integer | NOT NULL | 0 | ||
default_project_creation | integer | NOT NULL | 2 | ||
lets_encrypt_notification_email | varchar | null | |||
lets_encrypt_terms_of_service_accepted | boolean | NOT NULL | FALSE | ||
geo_node_allowed_ips | varchar | null | CAST('0.0.0.0/0, ::/0' AS varchar) | ||
encrypted_lets_encrypt_private_key | text | null | |||
encrypted_lets_encrypt_private_key_iv | text | null | |||
dns_rebinding_protection_enabled | boolean | NOT NULL | TRUE | ||
default_project_deletion_protection | boolean | NOT NULL | FALSE | ||
grafana_enabled | boolean | NOT NULL | FALSE | ||
lock_memberships_to_ldap | boolean | NOT NULL | FALSE | ||
time_tracking_limit_to_hours | boolean | NOT NULL | FALSE | ||
grafana_url | varchar | NOT NULL | CAST('/-/grafana' AS varchar) | ||
login_recaptcha_protection_enabled | boolean | NOT NULL | FALSE | ||
outbound_local_requests_whitelist | varchar(255)[] | NOT NULL | CAST('{}' AS varchar[]) | ||
raw_blob_request_limit | integer | NOT NULL | 300 | ||
allow_local_requests_from_web_hooks_and_services | boolean | NOT NULL | FALSE | ||
allow_local_requests_from_system_hooks | boolean | NOT NULL | TRUE | ||
asset_proxy_enabled | boolean | NOT NULL | FALSE | ||
asset_proxy_url | varchar | null | |||
encrypted_asset_proxy_secret_key | text | null | |||
encrypted_asset_proxy_secret_key_iv | varchar | null | |||
static_objects_external_storage_url | varchar(255) | null | |||
max_personal_access_token_lifetime | integer | null | |||
throttle_protected_paths_enabled | boolean | NOT NULL | FALSE | ||
throttle_protected_paths_requests_per_period | integer | NOT NULL | 10 | ||
throttle_protected_paths_period_in_seconds | integer | NOT NULL | 60 | ||
protected_paths | varchar(255)[] | null | CAST('{/users/password,/users/sign_in,/api/v3/session.json,/api/v3/session,/api/v4/session.json,/api/v4/session,/users,/users/confirmation,/unsubscribes/,/import/github/personal_access_token,/admin/session}' AS varchar[]) | ||
throttle_incident_management_notification_enabled | boolean | NOT NULL | FALSE | ||
throttle_incident_management_notification_period_in_seconds | integer | null | 3600 | ||
throttle_incident_management_notification_per_period | integer | null | 3600 | ||
push_event_hooks_limit | integer | NOT NULL | 3 | ||
push_event_activities_limit | integer | NOT NULL | 3 | ||
custom_http_clone_url_root | varchar(511) | null | |||
deletion_adjourned_period | integer | NOT NULL | 30 | ||
license_trial_ends_on | date | null | |||
eks_integration_enabled | boolean | NOT NULL | FALSE | ||
eks_account_id | varchar(128) | null | |||
eks_access_key_id | varchar(128) | null | |||
encrypted_eks_secret_access_key_iv | varchar(255) | null | |||
encrypted_eks_secret_access_key | text | null | |||
snowplow_app_id | varchar | null | |||
productivity_analytics_start_date | timestamptz | null | |||
default_ci_config_path | varchar(255) | null | |||
sourcegraph_enabled | boolean | NOT NULL | FALSE | ||
sourcegraph_url | varchar(255) | null | |||
sourcegraph_public_only | boolean | NOT NULL | TRUE | ||
snippet_size_limit | bigint | NOT NULL | 52428800 | ||
minimum_password_length | integer | NOT NULL | 8 | ||
encrypted_akismet_api_key | text | null | |||
encrypted_akismet_api_key_iv | varchar(255) | null | |||
encrypted_elasticsearch_aws_secret_access_key | text | null | |||
encrypted_elasticsearch_aws_secret_access_key_iv | varchar(255) | null | |||
encrypted_recaptcha_private_key | text | null | |||
encrypted_recaptcha_private_key_iv | varchar(255) | null | |||
encrypted_recaptcha_site_key | text | null | |||
encrypted_recaptcha_site_key_iv | varchar(255) | null | |||
encrypted_slack_app_secret | text | null | |||
encrypted_slack_app_secret_iv | varchar(255) | null | |||
encrypted_slack_app_verification_token | text | null | |||
encrypted_slack_app_verification_token_iv | varchar(255) | null | |||
force_pages_access_control | boolean | NOT NULL | FALSE | ||
updating_name_disabled_for_users | boolean | NOT NULL | FALSE | ||
disable_overriding_approvers_per_merge_request | boolean | NOT NULL | FALSE | ||
prevent_merge_requests_author_approval | boolean | NOT NULL | FALSE | ||
prevent_merge_requests_committers_approval | boolean | NOT NULL | FALSE | ||
email_restrictions_enabled | boolean | NOT NULL | FALSE | ||
email_restrictions | text | null | |||
container_expiration_policies_enable_historic_entries | boolean | NOT NULL | FALSE | ||
issues_create_limit | integer | NOT NULL | 0 | ||
push_rule_id | bigint | null | |||
group_owners_can_manage_default_branch_protection | boolean | NOT NULL | TRUE | ||
container_registry_vendor | text | NOT NULL | CAST('' AS text) | CHECK char_length(container_registry_vendor) <= 255 | |
container_registry_version | text | NOT NULL | CAST('' AS text) | CHECK char_length(container_registry_version) <= 255 | |
container_registry_features | text[] | NOT NULL | CAST('{}' AS text[]) | ||
spam_check_endpoint_url | text | null | CHECK char_length(spam_check_endpoint_url) <= 255 | ||
spam_check_endpoint_enabled | boolean | NOT NULL | FALSE | ||
repository_storages_weighted | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
max_import_size | integer | NOT NULL | 0 | ||
compliance_frameworks | smallint[] | NOT NULL | CAST('{}' AS smallint[]) | ||
notify_on_unknown_sign_in | boolean | NOT NULL | TRUE | ||
default_branch_name | text | null | CHECK char_length(default_branch_name) <= 255 | ||
project_import_limit | integer | NOT NULL | 6 | ||
project_export_limit | integer | NOT NULL | 6 | ||
project_download_export_limit | integer | NOT NULL | 1 | ||
group_import_limit | integer | NOT NULL | 6 | ||
group_export_limit | integer | NOT NULL | 6 | ||
group_download_export_limit | integer | NOT NULL | 1 | ||
maintenance_mode | boolean | NOT NULL | FALSE | ||
maintenance_mode_message | text | null | CHECK char_length(maintenance_mode_message) <= 255 | ||
wiki_page_max_content_bytes | bigint | NOT NULL | 5242880 | ||
enforce_namespace_storage_limit | boolean | NOT NULL | FALSE | ||
container_registry_delete_tags_service_timeout | integer | NOT NULL | 250 | ||
kroki_url | varchar | null | CHECK char_length(CAST(kroki_url AS text)) <= 1024 | ||
kroki_enabled | boolean | null | |||
gitpod_enabled | boolean | NOT NULL | FALSE | ||
gitpod_url | text | null | CAST('https://gitpod.io/' AS text) | CHECK char_length(gitpod_url) <= 255 | |
abuse_notification_email | varchar | null | |||
require_admin_approval_after_user_signup | boolean | NOT NULL | TRUE | ||
help_page_documentation_base_url | text | null | CHECK char_length(help_page_documentation_base_url) <= 255 | ||
automatic_purchased_storage_allocation | boolean | NOT NULL | FALSE | ||
encrypted_ci_jwt_signing_key | text | null | |||
encrypted_ci_jwt_signing_key_iv | text | null | CHECK char_length(encrypted_ci_jwt_signing_key_iv) <= 255 | ||
container_registry_expiration_policies_worker_capacity | integer | NOT NULL | 4 | CHECK container_registry_expiration_policies_worker_capacity >= 0 | |
secret_detection_token_revocation_enabled | boolean | NOT NULL | FALSE | ||
secret_detection_token_revocation_url | text | null | CHECK char_length(secret_detection_token_revocation_url) <= 255 | ||
encrypted_secret_detection_token_revocation_token | text | null | |||
encrypted_secret_detection_token_revocation_token_iv | text | null | |||
domain_denylist_enabled | boolean | null | FALSE | ||
domain_denylist | text | null | |||
domain_allowlist | text | null | |||
new_user_signups_cap | integer | null | |||
encrypted_cloud_license_auth_token | text | null | |||
encrypted_cloud_license_auth_token_iv | text | null | CHECK char_length(encrypted_cloud_license_auth_token_iv) <= 255 | ||
secret_detection_revocation_token_types_url | text | null | CHECK char_length(secret_detection_revocation_token_types_url) <= 255 | ||
disable_feed_token | boolean | NOT NULL | FALSE | ||
personal_access_token_prefix | text | null | CAST('glpat-' AS text) | CHECK char_length(personal_access_token_prefix) <= 20 | |
rate_limiting_response_text | text | null | CHECK char_length(rate_limiting_response_text) <= 255 | ||
invisible_captcha_enabled | boolean | NOT NULL | FALSE | ||
container_registry_cleanup_tags_service_max_list_size | integer | NOT NULL | 200 | CHECK container_registry_cleanup_tags_service_max_list_size >= 0 | |
git_two_factor_session_expiry | integer | NOT NULL | 15 | ||
keep_latest_artifact | boolean | NOT NULL | TRUE | ||
notes_create_limit | integer | NOT NULL | 300 | ||
notes_create_limit_allowlist | text[] | NOT NULL | CAST('{}' AS text[]) | ||
kroki_formats | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
asset_proxy_whitelist | text | null | |||
admin_mode | boolean | NOT NULL | FALSE | ||
external_pipeline_validation_service_timeout | integer | null | |||
encrypted_external_pipeline_validation_service_token | text | null | |||
encrypted_external_pipeline_validation_service_token_iv | text | null | |||
external_pipeline_validation_service_url | text | null | CHECK char_length(external_pipeline_validation_service_url) <= 255 | ||
throttle_unauthenticated_packages_api_requests_per_period | integer | NOT NULL | 800 | ||
throttle_unauthenticated_packages_api_period_in_seconds | integer | NOT NULL | 15 | ||
throttle_authenticated_packages_api_requests_per_period | integer | NOT NULL | 1000 | ||
throttle_authenticated_packages_api_period_in_seconds | integer | NOT NULL | 15 | ||
throttle_unauthenticated_packages_api_enabled | boolean | NOT NULL | FALSE | ||
throttle_authenticated_packages_api_enabled | boolean | NOT NULL | FALSE | ||
deactivate_dormant_users | boolean | NOT NULL | FALSE | ||
whats_new_variant | smallint | null | 0 | ||
encrypted_spam_check_api_key | bytea | null | |||
encrypted_spam_check_api_key_iv | bytea | null | |||
floc_enabled | boolean | NOT NULL | FALSE | ||
encrypted_elasticsearch_password | bytea | null | |||
encrypted_elasticsearch_password_iv | bytea | null | |||
diff_max_lines | integer | NOT NULL | 50000 | ||
diff_max_files | integer | NOT NULL | 1000 | ||
valid_runner_registrars | varchar[] | null | CAST('{project,group}' AS varchar[]) | ||
encrypted_mailgun_signing_key | bytea | null | |||
encrypted_mailgun_signing_key_iv | bytea | null | |||
mailgun_events_enabled | boolean | NOT NULL | FALSE | ||
usage_ping_features_enabled | boolean | NOT NULL | FALSE | ||
encrypted_customers_dot_jwt_signing_key | bytea | null | |||
encrypted_customers_dot_jwt_signing_key_iv | bytea | null | |||
throttle_unauthenticated_files_api_requests_per_period | integer | NOT NULL | 125 | ||
throttle_unauthenticated_files_api_period_in_seconds | integer | NOT NULL | 15 | ||
throttle_authenticated_files_api_requests_per_period | integer | NOT NULL | 500 | ||
throttle_authenticated_files_api_period_in_seconds | integer | NOT NULL | 15 | ||
throttle_unauthenticated_files_api_enabled | boolean | NOT NULL | FALSE | ||
throttle_authenticated_files_api_enabled | boolean | NOT NULL | FALSE | ||
max_yaml_size_bytes | bigint | NOT NULL | 2097152 | CHECK max_yaml_size_bytes > 0 | |
max_yaml_depth | integer | NOT NULL | 100 | CHECK max_yaml_depth > 0 | |
throttle_authenticated_git_lfs_requests_per_period | integer | NOT NULL | 1000 | ||
throttle_authenticated_git_lfs_period_in_seconds | integer | NOT NULL | 60 | ||
throttle_authenticated_git_lfs_enabled | boolean | NOT NULL | FALSE | ||
user_deactivation_emails_enabled | boolean | NOT NULL | TRUE | ||
throttle_unauthenticated_api_enabled | boolean | NOT NULL | FALSE | ||
throttle_unauthenticated_api_requests_per_period | integer | NOT NULL | 3600 | ||
throttle_unauthenticated_api_period_in_seconds | integer | NOT NULL | 3600 | ||
jobs_per_stage_page_size | integer | NOT NULL | 200 | ||
sidekiq_job_limiter_mode | smallint | NOT NULL | 1 | ||
sidekiq_job_limiter_compression_threshold_bytes | integer | NOT NULL | 100000 | ||
sidekiq_job_limiter_limit_bytes | integer | NOT NULL | 0 | ||
suggest_pipeline_enabled | boolean | NOT NULL | TRUE | ||
throttle_unauthenticated_deprecated_api_requests_per_period | integer | NOT NULL | 1800 | ||
throttle_unauthenticated_deprecated_api_period_in_seconds | integer | NOT NULL | 3600 | ||
throttle_unauthenticated_deprecated_api_enabled | boolean | NOT NULL | FALSE | ||
throttle_authenticated_deprecated_api_requests_per_period | integer | NOT NULL | 3600 | ||
throttle_authenticated_deprecated_api_period_in_seconds | integer | NOT NULL | 3600 | ||
throttle_authenticated_deprecated_api_enabled | boolean | NOT NULL | FALSE | ||
dependency_proxy_ttl_group_policy_worker_capacity | smallint | NOT NULL | 2 | CHECK dependency_proxy_ttl_group_policy_worker_capacity >= 0 | |
content_validation_endpoint_url | text | null | JiHu-specific columnCHECK char_length(content_validation_endpoint_url) <= 255 | ||
encrypted_content_validation_api_key | bytea | null | JiHu-specific column | ||
encrypted_content_validation_api_key_iv | bytea | null | JiHu-specific column | ||
content_validation_endpoint_enabled | boolean | NOT NULL | FALSE | JiHu-specific column | |
sentry_enabled | boolean | NOT NULL | FALSE | ||
sentry_dsn | text | null | CHECK char_length(sentry_dsn) <= 255 | ||
sentry_clientside_dsn | text | null | CHECK char_length(sentry_clientside_dsn) <= 255 | ||
sentry_environment | text | null | CHECK char_length(sentry_environment) <= 255 | ||
max_ssh_key_lifetime | integer | null | |||
static_objects_external_storage_auth_token_encrypted | text | null | CHECK char_length(static_objects_external_storage_auth_token_encrypted) <= 255 | ||
future_subscriptions | jsonb | NOT NULL | CAST('[]' AS jsonb) | ||
runner_token_expiration_interval | integer | null | |||
group_runner_token_expiration_interval | integer | null | |||
project_runner_token_expiration_interval | integer | null | |||
ecdsa_sk_key_restriction | integer | NOT NULL | 0 | ||
ed25519_sk_key_restriction | integer | NOT NULL | 0 | ||
users_get_by_id_limit | integer | NOT NULL | 300 | ||
users_get_by_id_limit_allowlist | text[] | NOT NULL | CAST('{}' AS text[]) | ||
container_registry_expiration_policies_caching | boolean | NOT NULL | TRUE | ||
search_rate_limit | integer | NOT NULL | 300 | ||
search_rate_limit_unauthenticated | integer | NOT NULL | 100 | ||
encrypted_database_grafana_api_key | bytea | null | |||
encrypted_database_grafana_api_key_iv | bytea | null | |||
database_grafana_api_url | text | null | CHECK char_length(database_grafana_api_url) <= 255 | ||
database_grafana_tag | text | null | CHECK char_length(database_grafana_tag) <= 255 | ||
public_runner_releases_url | text | NOT NULL | CAST('https://gitlab.com/api/v4/projects/gitlab-org%2Fgitlab-runner/releases' AS text) | CHECK char_length(public_runner_releases_url) <= 255 | |
password_uppercase_required | boolean | NOT NULL | FALSE | ||
password_lowercase_required | boolean | NOT NULL | FALSE | ||
password_number_required | boolean | NOT NULL | FALSE | ||
password_symbol_required | boolean | NOT NULL | FALSE | ||
encrypted_arkose_labs_public_api_key | bytea | null | |||
encrypted_arkose_labs_public_api_key_iv | bytea | null | |||
encrypted_arkose_labs_private_api_key | bytea | null | |||
encrypted_arkose_labs_private_api_key_iv | bytea | null | |||
delete_inactive_projects | boolean | NOT NULL | FALSE | ||
inactive_projects_delete_after_months | integer | NOT NULL | 2 | ||
inactive_projects_min_size_mb | integer | NOT NULL | 0 | ||
inactive_projects_send_warning_email_after_months | integer | NOT NULL | 1 | ||
arkose_labs_namespace | text | NOT NULL | CAST('client' AS text) | CHECK char_length(arkose_labs_namespace) <= 255 | |
max_export_size | integer | null | 0 | ||
encrypted_slack_app_signing_secret | bytea | null | |||
encrypted_slack_app_signing_secret_iv | bytea | null | |||
pipeline_limit_per_project_user_sha | integer | NOT NULL | 0 | ||
dingtalk_integration_enabled | boolean | NOT NULL | FALSE | JiHu-specific column | |
encrypted_dingtalk_corpid | bytea | null | JiHu-specific column | ||
encrypted_dingtalk_corpid_iv | bytea | null | JiHu-specific column | ||
encrypted_dingtalk_app_key | bytea | null | JiHu-specific column | ||
encrypted_dingtalk_app_key_iv | bytea | null | JiHu-specific column | ||
encrypted_dingtalk_app_secret | bytea | null | JiHu-specific column | ||
encrypted_dingtalk_app_secret_iv | bytea | null | JiHu-specific column | ||
jira_connect_application_key | text | null | CHECK char_length(jira_connect_application_key) <= 255 | ||
globally_allowed_ips | text | NOT NULL | CAST('' AS text) | CHECK char_length(globally_allowed_ips) <= 255 | |
license_usage_data_exported | boolean | NOT NULL | FALSE | ||
phone_verification_code_enabled | boolean | NOT NULL | FALSE | JiHu-specific column | |
max_number_of_repository_downloads | smallint | NOT NULL | 0 | ||
max_number_of_repository_downloads_within_time_period | integer | NOT NULL | 0 | ||
feishu_integration_enabled | boolean | NOT NULL | FALSE | JiHu-specific column | |
encrypted_feishu_app_key | bytea | null | JiHu-specific column | ||
encrypted_feishu_app_key_iv | bytea | null | JiHu-specific column | ||
encrypted_feishu_app_secret | bytea | null | JiHu-specific column | ||
encrypted_feishu_app_secret_iv | bytea | null | JiHu-specific column | ||
error_tracking_enabled | boolean | NOT NULL | FALSE | ||
error_tracking_api_url | text | null | CHECK char_length(error_tracking_api_url) <= 255 | ||
git_rate_limit_users_allowlist | text[] | NOT NULL | CAST('{}' AS text[]) | CHECK cardinality(git_rate_limit_users_allowlist) <= 100 | |
error_tracking_access_token_encrypted | text | null | CHECK char_length(error_tracking_access_token_encrypted) <= 255 | ||
invitation_flow_enforcement | boolean | NOT NULL | FALSE | ||
deactivate_dormant_users_period | integer | NOT NULL | 90 | ||
auto_ban_user_on_excessive_projects_download | boolean | NOT NULL | FALSE | ||
max_pages_custom_domains_per_project | integer | NOT NULL | 0 | CHECK max_pages_custom_domains_per_project >= 0 | |
cube_api_base_url | text | null | CHECK char_length(cube_api_base_url) <= 512 | ||
encrypted_cube_api_key | bytea | null | |||
encrypted_cube_api_key_iv | bytea | null | |||
dashboard_limit_enabled | boolean | NOT NULL | FALSE | ||
dashboard_limit | integer | NOT NULL | 0 | ||
can_create_group | boolean | NOT NULL | TRUE | ||
jira_connect_proxy_url | text | null | CHECK char_length(jira_connect_proxy_url) <= 255 | ||
password_expiration_enabled | boolean | NOT NULL | FALSE | JiHu-specific column | |
password_expires_in_days | integer | NOT NULL | 90 | JiHu-specific column | |
password_expires_notice_before_days | integer | NOT NULL | 7 | JiHu-specific column | |
product_analytics_enabled | boolean | NOT NULL | FALSE | ||
email_confirmation_setting | smallint | null | 0 | ||
disable_admin_oauth_scopes | boolean | NOT NULL | FALSE | ||
default_preferred_language | text | NOT NULL | CAST('en' AS text) | CHECK char_length(default_preferred_language) <= 32 | |
disable_download_button | boolean | NOT NULL | FALSE | JiHu-specific column | |
encrypted_telesign_customer_xid | bytea | null | |||
encrypted_telesign_customer_xid_iv | bytea | null | |||
encrypted_telesign_api_key | bytea | null | |||
encrypted_telesign_api_key_iv | bytea | null | |||
disable_personal_access_tokens | boolean | NOT NULL | FALSE | ||
max_terraform_state_size_bytes | integer | NOT NULL | 0 | CHECK max_terraform_state_size_bytes >= 0 | |
bulk_import_enabled | boolean | NOT NULL | FALSE | ||
allow_runner_registration_token | boolean | NOT NULL | TRUE | ||
user_defaults_to_private_profile | boolean | NOT NULL | FALSE | ||
allow_possible_spam | boolean | NOT NULL | FALSE | ||
default_syntax_highlighting_theme | integer | NOT NULL | 1 | ||
search_max_shard_size_gb | integer | NOT NULL | 50 | ||
search_max_docs_denominator | integer | NOT NULL | 5000000 | ||
search_min_docs_before_rollover | integer | NOT NULL | 100000 | ||
deactivation_email_additional_text | text | null | CHECK char_length(deactivation_email_additional_text) <= 1000 | ||
jira_connect_public_key_storage_enabled | boolean | NOT NULL | FALSE | ||
git_rate_limit_users_alertlist | integer[] | NOT NULL | CAST('{}' AS integer[]) | CHECK cardinality(git_rate_limit_users_alertlist) <= 100 | |
allow_deploy_tokens_and_keys_with_external_authn | boolean | NOT NULL | FALSE | ||
security_policy_global_group_approvers_enabled | boolean | NOT NULL | TRUE | ||
projects_api_rate_limit_unauthenticated | integer | NOT NULL | 400 | ||
deny_all_requests_except_allowed | boolean | NOT NULL | FALSE | ||
product_analytics_data_collector_host | text | null | CHECK char_length(product_analytics_data_collector_host) <= 255 | ||
lock_memberships_to_saml | boolean | NOT NULL | FALSE | ||
gitlab_dedicated_instance | boolean | NOT NULL | FALSE | ||
update_runner_versions_enabled | boolean | NOT NULL | TRUE | ||
database_max_running_batched_background_migrations | integer | NOT NULL | 2 | ||
encrypted_product_analytics_configurator_connection_string | bytea | null | |||
encrypted_product_analytics_configurator_connection_string_iv | bytea | null | |||
silent_mode_enabled | boolean | NOT NULL | FALSE | ||
package_metadata_purl_types | smallint[] | null | CAST('{1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,17}' AS smallint[]) | ||
ci_max_includes | integer | NOT NULL | 150 | ||
remember_me_enabled | boolean | NOT NULL | TRUE | ||
diagramsnet_enabled | boolean | NOT NULL | TRUE | ||
diagramsnet_url | text | null | CAST('https://embed.diagrams.net' AS text) | CHECK char_length(diagramsnet_url) <= 2048 | |
allow_account_deletion | boolean | NOT NULL | TRUE | ||
wiki_asciidoc_allow_uri_includes | boolean | NOT NULL | FALSE | ||
namespace_aggregation_schedule_lease_duration_in_seconds | integer | NOT NULL | 300 | ||
vertex_ai_host | text | null | CHECK char_length(vertex_ai_host) <= 255 | ||
vertex_ai_project | text | null | CHECK char_length(vertex_ai_project) <= 255 | ||
delete_unconfirmed_users | boolean | NOT NULL | FALSE | ||
unconfirmed_users_delete_after_days | integer | NOT NULL | 7 | ||
default_branch_protection_defaults | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK octet_length(CAST(default_branch_protection_defaults AS text)) <= 1024 | |
gitlab_shell_operation_limit | integer | null | 600 | ||
protected_paths_for_get_request | text[] | NOT NULL | CAST('{/users/sign_in_path}' AS text[]) | CHECK cardinality(protected_paths_for_get_request) <= 100 | |
namespace_storage_forks_cost_factor | double precision | NOT NULL | 1.0 | CHECK namespace_storage_forks_cost_factor >= CAST(0 AS double precision) AND namespace_storage_forks_cost_factor <= CAST(1 AS double precision) | |
bulk_import_max_download_file_size | bigint | NOT NULL | 5120 | ||
max_import_remote_file_size | bigint | NOT NULL | 10240 | ||
max_decompressed_archive_size | integer | NOT NULL | 25600 | CHECK max_decompressed_archive_size >= 0 | |
sentry_clientside_traces_sample_rate | double precision | NOT NULL | 0.0 | CHECK sentry_clientside_traces_sample_rate >= CAST(0 AS double precision) AND sentry_clientside_traces_sample_rate <= CAST(1 AS double precision) | |
prometheus_alert_db_indicators_settings | jsonb | null | |||
ci_max_total_yaml_size_bytes | integer | NOT NULL | 314572800 | ||
decompress_archive_file_timeout | integer | NOT NULL | 210 | ||
search_rate_limit_allowlist | text[] | NOT NULL | CAST('{}' AS text[]) | ||
snowplow_database_collector_hostname | text | null | CHECK char_length(snowplow_database_collector_hostname) <= 255 | ||
container_registry_db_enabled | boolean | NOT NULL | FALSE | ||
failed_login_attempts_unlock_period_in_minutes | integer | null | CHECK failed_login_attempts_unlock_period_in_minutes > 0 | ||
max_login_attempts | integer | null | CHECK max_login_attempts > 0 | ||
project_jobs_api_rate_limit | integer | NOT NULL | 600 | ||
math_rendering_limits_enabled | boolean | NOT NULL | TRUE | ||
service_access_tokens_expiration_enforced | boolean | NOT NULL | TRUE | ||
make_profile_private | boolean | NOT NULL | TRUE | ||
enable_artifact_external_redirect_warning_page | boolean | NOT NULL | TRUE | ||
allow_project_creation_for_guest_and_below | boolean | NOT NULL | TRUE | ||
update_namespace_name_rate_limit | smallint | NOT NULL | 120 | ||
can_create_organization | boolean | NOT NULL | TRUE | ||
bulk_import_concurrent_pipeline_batch_limit | smallint | NOT NULL | 25 | ||
web_ide_oauth_application_id | bigint | null | oauth_applications.id 1:N · SET NULL | ||
instance_level_ai_beta_features_enabled | boolean | NOT NULL | FALSE | ||
security_txt_content | text | null | CHECK char_length(security_txt_content) <= 2048 | ||
encrypted_arkose_labs_data_exchange_key | bytea | null | |||
encrypted_arkose_labs_data_exchange_key_iv | bytea | null | |||
rate_limits | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(rate_limits) = CAST('object' AS text) | |
enable_member_promotion_management | boolean | NOT NULL | FALSE | ||
lock_math_rendering_limits_enabled | boolean | NOT NULL | FALSE | ||
security_approval_policies_limit | integer | NOT NULL | 5 | ||
encrypted_arkose_labs_client_xid | bytea | null | |||
encrypted_arkose_labs_client_xid_iv | bytea | null | |||
encrypted_arkose_labs_client_secret | bytea | null | |||
encrypted_arkose_labs_client_secret_iv | bytea | null | |||
duo_features_enabled | boolean | NOT NULL | TRUE | ||
lock_duo_features_enabled | boolean | NOT NULL | FALSE | ||
asciidoc_max_includes | smallint | NOT NULL | 32 | ||
clickhouse | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(clickhouse) = CAST('object' AS text) | |
include_optional_metrics_in_service_ping | boolean | NOT NULL | TRUE | ||
zoekt_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
service_ping_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(service_ping_settings) = CAST('object' AS text) | |
package_registry | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(package_registry) = CAST('object' AS text) | |
rate_limits_unauthenticated_git_http | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(rate_limits_unauthenticated_git_http) = CAST('object' AS text) | |
importers | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(importers) = CAST('object' AS text) | |
code_creation | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(code_creation) = CAST('object' AS text) | |
code_suggestions_api_rate_limit | integer | NOT NULL | 60 | ||
ai_action_api_rate_limit | integer | NOT NULL | 160 | ||
require_personal_access_token_expiry | boolean | NOT NULL | TRUE | ||
duo_workflow | jsonb | null | CAST('{}' AS jsonb) | CHECK jsonb_typeof(duo_workflow) = CAST('object' AS text) | |
max_artifacts_content_include_size | bigint | NOT NULL | 5242880 | ||
max_number_of_vulnerabilities_per_project | integer | null | |||
cluster_agents | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(cluster_agents) = CAST('object' AS text) | |
observability_backend_ssl_verification_enabled | boolean | NOT NULL | TRUE | ||
pages | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
security_policies | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(security_policies) = CAST('object' AS text) | |
spp_repository_pipeline_access | boolean | NOT NULL | FALSE | ||
lock_spp_repository_pipeline_access | boolean | NOT NULL | FALSE | ||
required_instance_ci_template | text | null | CHECK char_length(required_instance_ci_template) <= 1024 | ||
enforce_ci_inbound_job_token_scope_enabled | boolean | NOT NULL | TRUE | ||
allow_top_level_group_owners_to_create_service_accounts | boolean | NOT NULL | FALSE | ||
sign_in_restrictions | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(sign_in_restrictions) = CAST('object' AS text) | |
transactional_emails | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(transactional_emails) = CAST('object' AS text) | |
identity_verification_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(identity_verification_settings) = CAST('object' AS text) | |
integrations | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(integrations) = CAST('object' AS text) | |
user_seat_management | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
secret_detection_service_url | text | NOT NULL | CAST('' AS text) | CHECK char_length(secret_detection_service_url) <= 255 | |
encrypted_secret_detection_service_auth_token | bytea | null | |||
encrypted_secret_detection_service_auth_token_iv | bytea | null | |||
resource_usage_limits | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
show_migrate_from_jenkins_banner | boolean | NOT NULL | TRUE | ||
encrypted_ci_job_token_signing_key | bytea | null | |||
encrypted_ci_job_token_signing_key_iv | bytea | null | |||
elasticsearch | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(elasticsearch) = CAST('object' AS text) | |
oauth_provider | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(oauth_provider) = CAST('object' AS text) | |
observability_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(observability_settings) = CAST('object' AS text) | |
search | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
anti_abuse_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(anti_abuse_settings) = CAST('object' AS text) | |
secret_push_protection_available | boolean | null | FALSE | CHECK secret_push_protection_available IS NOT NULL | |
vscode_extension_marketplace | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(vscode_extension_marketplace) = CAST('object' AS text) | |
token_prefixes | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(token_prefixes) = CAST('object' AS text) | |
ci_cd_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(ci_cd_settings) = CAST('object' AS text) | |
database_reindexing | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(database_reindexing) = CAST('object' AS text) | |
duo_chat | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(duo_chat) = CAST('object' AS text) | |
group_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(group_settings) = CAST('object' AS text) | |
model_prompt_cache_enabled | boolean | NOT NULL | TRUE | ||
lock_model_prompt_cache_enabled | boolean | NOT NULL | FALSE | ||
response_limits | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
web_based_commit_signing_enabled | boolean | NOT NULL | FALSE | ||
lock_web_based_commit_signing_enabled | boolean | NOT NULL | FALSE | ||
tmp_asset_proxy_secret_key | jsonb | null | |||
editor_extensions | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(editor_extensions) = CAST('object' AS text) | |
security_and_compliance_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(security_and_compliance_settings) = CAST('object' AS text) | |
sdrs_url | text | null | CHECK char_length(sdrs_url) <= 255 | ||
default_profile_preferences | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
sdrs_enabled | boolean | NOT NULL | FALSE | ||
sdrs_jwt_signing_key | jsonb | null | |||
resource_access_tokens_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
auto_duo_code_review_enabled | boolean | NOT NULL | FALSE | ||
lock_auto_duo_code_review_enabled | boolean | NOT NULL | FALSE | ||
workspaces_oauth_application_id | bigint | null | oauth_applications.id 1:N · SET NULL | ||
usage_ping_generation_enabled | boolean | NOT NULL | TRUE | ||
duo_remote_flows_enabled | boolean | NOT NULL | TRUE | ||
lock_duo_remote_flows_enabled | boolean | NOT NULL | FALSE | ||
terraform_state_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(terraform_state_settings) = CAST('object' AS text) | |
duo_foundational_flows_enabled | boolean | NOT NULL | TRUE | ||
lock_duo_foundational_flows_enabled | boolean | NOT NULL | FALSE | ||
iframe_rendering_enabled | boolean | NOT NULL | FALSE | ||
iframe_rendering_allowlist | text | null | CHECK char_length(iframe_rendering_allowlist) <= 5000 | ||
database_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(database_settings) = CAST('object' AS text) | |
usage_billing | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(usage_billing) = CAST('object' AS text) | |
topology_service_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
secrets_manager_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(secrets_manager_settings) = CAST('object' AS text) | |
duo_workflows_default_image_registry | text | null | CHECK char_length(duo_workflows_default_image_registry) <= 512 | ||
diagram_proxy | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(diagram_proxy) = CAST('object' AS text) | |
built_in_project_templates_enabled | boolean | NOT NULL | TRUE | ||
lock_built_in_project_templates_enabled | boolean | NOT NULL | FALSE | ||
tool_approval_for_session_enabled | boolean | NOT NULL | FALSE | ||
lock_tool_approval_for_session_enabled | boolean | NOT NULL | FALSE | ||
personal_access_token_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(personal_access_token_settings) = CAST('object' AS text) | |
diff_limits | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(diff_limits) = CAST('object' AS text) | |
markdown_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(markdown_settings) = CAST('object' AS text) | |
active_context_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
duo_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
mcp_server_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(mcp_server_settings) = CAST('object' AS text) | |
secrets_manager_instance_enrolled | boolean | NOT NULL | FALSE | ||
duo_template_project_id | bigint | null | projects.id 1:N · SET NULL | ||
default_security_tracked_context_quota | integer | null | 2 | ||
oauth_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(oauth_settings) = CAST('object' AS text) | |
logging_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
dependency_management_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(dependency_management_settings) = CAST('object' AS text) | |
require_sha_for_merge | boolean | NOT NULL | FALSE | ||
lock_require_sha_for_merge | boolean | NOT NULL | FALSE | ||
ai_audit_events_storage_enabled | boolean | NOT NULL | FALSE | ||
lock_ai_audit_events_storage_enabled | boolean | NOT NULL | FALSE | ||
ai_gateway_url | text | null | CHECK char_length(ai_gateway_url) <= 2048 | ||
ai_gateway_timeout_seconds | integer | null | 60 | ||
enabled_instance_verbose_ai_logs | boolean | NOT NULL | FALSE | ||
duo_agent_platform_service_url | text | null | CHECK char_length(duo_agent_platform_service_url) <= 2048 | ||
self_hosted_duo_agent_platform_service_secure | boolean | NOT NULL | TRUE | ||
sidekiq_timezone_override | text | null | CHECK char_length(sidekiq_timezone_override) <= 255 | ||
nats_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(nats_settings) = CAST('object' AS text) | |
o11y_oauth_application_id | bigint | null | oauth_applications.id 1:N · SET NULL |
Table checks: container_registry_cleanup_tags_service_max_list_size >= 0; dependency_proxy_ttl_group_policy_worker_capacity >= 0; char_length(external_pipeline_validation_service_url) <= 255; failed_login_attempts_unlock_period_in_minutes > 0; cardinality(git_rate_limit_users_alertlist) <= 100; cardinality(git_rate_limit_users_allowlist) <= 100; max_decompressed_archive_size >= 0; max_login_attempts > 0; max_pages_custom_domains_per_project >= 0; max_terraform_state_size_bytes >= 0; cardinality(protected_paths_for_get_request) <= 100; container_registry_expiration_policies_worker_capacity >= 0; max_yaml_depth > 0; max_yaml_size_bytes > 0; jsonb_typeof(oauth_provider) = CAST('object' AS text); char_length(diagramsnet_url) <= 2048; char_length(vertex_ai_project) <= 255; char_length(CAST(kroki_url AS text)) <= 1024; char_length(database_grafana_tag) <= 255; char_length(gitpod_url) <= 255; char_length(static_objects_external_storage_auth_token_encrypted) <= 255; char_length(database_grafana_api_url) <= 255; char_length(snowplow_database_collector_hostname) <= 255; char_length(sentry_clientside_dsn) <= 255; char_length(jira_connect_proxy_url) <= 255; char_length(error_tracking_api_url) <= 255; char_length(sentry_dsn) <= 255; char_length(default_branch_name) <= 255; char_length(error_tracking_access_token_encrypted) <= 255; char_length(sdrs_url) <= 255; char_length(help_page_documentation_base_url) <= 255; char_length(content_validation_endpoint_url) <= 255; char_length(sentry_environment) <= 255; char_length(duo_workflows_default_image_registry) <= 512; char_length(personal_access_token_prefix) <= 20; char_length(rate_limiting_response_text) <= 255; char_length(product_analytics_data_collector_host) <= 255; char_length(globally_allowed_ips) <= 255; char_length(arkose_labs_namespace) <= 255; char_length(ai_gateway_url) <= 2048; char_length(encrypted_ci_jwt_signing_key_iv) <= 255; char_length(public_runner_releases_url) <= 255; char_length(cube_api_base_url) <= 512; char_length(iframe_rendering_allowlist) <= 5000; char_length(secret_detection_token_revocation_url) <= 255; char_length(maintenance_mode_message) <= 255; char_length(duo_agent_platform_service_url) <= 2048; char_length(secret_detection_revocation_token_types_url) <= 255; char_length(vertex_ai_host) <= 255; jsonb_typeof(anti_abuse_settings) = CAST('object' AS text); jsonb_typeof(dependency_management_settings) = CAST('object' AS text); namespace_storage_forks_cost_factor >= CAST(0 AS double precision) AND namespace_storage_forks_cost_factor <= CAST(1 AS double precision); sentry_clientside_traces_sample_rate >= CAST(0 AS double precision) AND sentry_clientside_traces_sample_rate <= CAST(1 AS double precision); jsonb_typeof(ci_cd_settings) = CAST('object' AS text); jsonb_typeof(clickhouse) = CAST('object' AS text); jsonb_typeof(cluster_agents) = CAST('object' AS text); jsonb_typeof(code_creation) = CAST('object' AS text); jsonb_typeof(database_reindexing) = CAST('object' AS text); jsonb_typeof(database_settings) = CAST('object' AS text); jsonb_typeof(diagram_proxy) = CAST('object' AS text); jsonb_typeof(diff_limits) = CAST('object' AS text); jsonb_typeof(duo_chat) = CAST('object' AS text); jsonb_typeof(duo_workflow) = CAST('object' AS text); jsonb_typeof(editor_extensions) = CAST('object' AS text); jsonb_typeof(elasticsearch) = CAST('object' AS text); jsonb_typeof(group_settings) = CAST('object' AS text); jsonb_typeof(importers) = CAST('object' AS text); jsonb_typeof(integrations) = CAST('object' AS text); jsonb_typeof(markdown_settings) = CAST('object' AS text); jsonb_typeof(mcp_server_settings) = CAST('object' AS text); jsonb_typeof(nats_settings) = CAST('object' AS text); jsonb_typeof(observability_settings) = CAST('object' AS text); jsonb_typeof(oauth_settings) = CAST('object' AS text); jsonb_typeof(package_registry) = CAST('object' AS text); jsonb_typeof(personal_access_token_settings) = CAST('object' AS text); jsonb_typeof(rate_limits) = CAST('object' AS text); jsonb_typeof(rate_limits_unauthenticated_git_http) = CAST('object' AS text); jsonb_typeof(secrets_manager_settings) = CAST('object' AS text); jsonb_typeof(security_policies) = CAST('object' AS text); jsonb_typeof(service_ping_settings) = CAST('object' AS text); jsonb_typeof(sign_in_restrictions) = CAST('object' AS text); jsonb_typeof(token_prefixes) = CAST('object' AS text); jsonb_typeof(transactional_emails) = CAST('object' AS text); jsonb_typeof(usage_billing) = CAST('object' AS text); jsonb_typeof(vscode_extension_marketplace) = CAST('object' AS text); char_length(deactivation_email_additional_text) <= 1000; char_length(secret_detection_service_url) <= 255; secret_push_protection_available IS NOT NULL; char_length(required_instance_ci_template) <= 1024; char_length(security_txt_content) <= 2048; char_length(container_registry_vendor) <= 255; char_length(spam_check_endpoint_url) <= 255; char_length(default_preferred_language) <= 32; char_length(jira_connect_application_key) <= 255; char_length(sidekiq_timezone_override) <= 255; char_length(container_registry_version) <= 255; char_length(encrypted_cloud_license_auth_token_iv) <= 255; jsonb_typeof(identity_verification_settings) = CAST('object' AS text); jsonb_typeof(security_and_compliance_settings) = CAST('object' AS text); jsonb_typeof(terraform_state_settings) = CAST('object' AS text); octet_length(CAST(default_branch_protection_defaults AS text)) <= 1024
Tables whose names start with approval_.
approval_group_rules.scan_result_policy_id → scan_result_policies.id — one scan_result_policies, many approval_group_rules · optional · ON DELETE CASCADE · indexedapproval_group_rules.approval_policy_rule_id → approval_policy_rules.id — one approval_policy_rules, many approval_group_rules · optional · ON DELETE CASCADE · indexedapproval_group_rules.security_orchestration_policy_configuration_id → security_orchestration_policy_configurations.id — one security_orchestration_policy_configurations, many approval_group_rules · optional · ON DELETE CASCADE · indexedapproval_group_rules.group_id → namespaces.id — one namespaces, many approval_group_rules · required · ON DELETE CASCADE · indexedapproval_group_rules_groups.group_id → namespaces.id — one namespaces, many approval_group_rules_groups · required · ON DELETE CASCADE · indexedapproval_group_rules_groups.approval_group_rule_id → approval_group_rules.id — one approval_group_rules, many approval_group_rules_groups · required · ON DELETE CASCADE · indexedapproval_group_rules_protected_branches.protected_branch_id → protected_branches.id — one protected_branches, many approval_group_rules_protected_branches · required · ON DELETE CASCADE · indexedapproval_group_rules_protected_branches.approval_group_rule_id → approval_group_rules.id — one approval_group_rules, many approval_group_rules_protected_branches · required · ON DELETE CASCADE · indexedapproval_group_rules_protected_branches.group_id → namespaces.id — one namespaces, many approval_group_rules_protected_branches · optional · ON DELETE CASCADE · indexedapproval_group_rules_users.group_id → namespaces.id — one namespaces, many approval_group_rules_users · optional · ON DELETE CASCADE · indexedapproval_group_rules_users.user_id → users.id — one users, many approval_group_rules_users · required · ON DELETE CASCADE · indexedapproval_group_rules_users.approval_group_rule_id → approval_group_rules.id — one approval_group_rules, many approval_group_rules_users · required · ON DELETE CASCADE · indexedapproval_merge_request_rule_sources.project_id → projects.id — one projects, many approval_merge_request_rule_sources · optional · ON DELETE CASCADE · indexedapproval_merge_request_rule_sources.approval_project_rule_id → approval_project_rules.id — one approval_project_rules, many approval_merge_request_rule_sources · required · ON DELETE CASCADE · indexedapproval_merge_request_rule_sources.approval_merge_request_rule_id → approval_merge_request_rules.id — one approval_merge_request_rules, at most one approval_merge_request_rule_sources · required · ON DELETE CASCADE · indexedapproval_merge_request_rules.security_orchestration_policy_configuration_id → security_orchestration_policy_configurations.id — one security_orchestration_policy_configurations, many approval_merge_request_rules · optional · ON DELETE CASCADE · indexedapproval_merge_request_rules.scan_result_policy_id → scan_result_policies.id — one scan_result_policies, many approval_merge_request_rules · optional · ON DELETE SET NULL · indexedapproval_merge_request_rules.project_id → projects.id — one projects, many approval_merge_request_rules · optional · ON DELETE CASCADE · indexedapproval_merge_request_rules.merge_request_id → merge_requests.id — one merge_requests, many approval_merge_request_rules · required · ON DELETE CASCADE · indexedapproval_merge_request_rules_approved_approvers.project_id → projects.id — one projects, many approval_merge_request_rules_approved_approvers · optional · ON DELETE CASCADE · indexedapproval_merge_request_rules_approved_approvers.approval_merge_request_rule_id → approval_merge_request_rules.id — one approval_merge_request_rules, many approval_merge_request_rules_approved_approvers · required · ON DELETE CASCADE · indexedapproval_merge_request_rules_approved_approvers.user_id → users.id — one users, many approval_merge_request_rules_approved_approvers · required · ON DELETE CASCADE · indexedapproval_merge_request_rules_groups.group_id → namespaces.id — one namespaces, many approval_merge_request_rules_groups · required · ON DELETE CASCADE · indexedapproval_merge_request_rules_groups.approval_merge_request_rule_id → approval_merge_request_rules.id — one approval_merge_request_rules, many approval_merge_request_rules_groups · required · ON DELETE CASCADE · indexedapproval_merge_request_rules_users.project_id → projects.id — one projects, many approval_merge_request_rules_users · optional · ON DELETE CASCADE · indexedapproval_merge_request_rules_users.approval_merge_request_rule_id → approval_merge_request_rules.id — one approval_merge_request_rules, many approval_merge_request_rules_users · required · ON DELETE CASCADE · indexedapproval_merge_request_rules_users.user_id → users.id — one users, many approval_merge_request_rules_users · required · ON DELETE CASCADE · indexedapproval_policy_merge_request_bypass_events.security_policy_id → security_policies.id — one security_policies, many approval_policy_merge_request_bypass_events · required · ON DELETE CASCADE · indexedapproval_policy_merge_request_bypass_events.user_id → users.id — one users, many approval_policy_merge_request_bypass_events · optional · ON DELETE SET NULL · indexedapproval_policy_merge_request_bypass_events.merge_request_id → merge_requests.id — one merge_requests, many approval_policy_merge_request_bypass_events · required · ON DELETE CASCADE · indexedapproval_policy_merge_request_bypass_events.project_id → projects.id — one projects, many approval_policy_merge_request_bypass_events · required · ON DELETE CASCADE · indexedapproval_policy_rule_project_links.approval_policy_rule_id → approval_policy_rules.id — one approval_policy_rules, many approval_policy_rule_project_links · required · ON DELETE CASCADE · indexedapproval_policy_rule_project_links.project_id → projects.id — one projects, many approval_policy_rule_project_links · required · ON DELETE CASCADE · indexedapproval_policy_rules.security_policy_id → security_policies.id — one security_policies, many approval_policy_rules · required · ON DELETE CASCADE · indexedapproval_policy_rules.security_policy_management_project_id → projects.id — one projects, many approval_policy_rules · required · ON DELETE CASCADE · indexedapproval_project_rules.approval_policy_rule_id → approval_policy_rules.id — one approval_policy_rules, many approval_project_rules · optional · ON DELETE CASCADE · indexedapproval_project_rules.scan_result_policy_id → scan_result_policies.id — one scan_result_policies, many approval_project_rules · optional · ON DELETE CASCADE · indexedapproval_project_rules.security_orchestration_policy_configuration_id → security_orchestration_policy_configurations.id — one security_orchestration_policy_configurations, many approval_project_rules · optional · ON DELETE CASCADE · indexedapproval_project_rules.project_id → projects.id — one projects, many approval_project_rules · required · ON DELETE CASCADE · indexedapproval_project_rules_groups.group_id → namespaces.id — one namespaces, many approval_project_rules_groups · required · ON DELETE CASCADE · indexedapproval_project_rules_groups.approval_project_rule_id → approval_project_rules.id — one approval_project_rules, many approval_project_rules_groups · required · ON DELETE CASCADE · indexedapproval_project_rules_protected_branches.project_id → projects.id — one projects, many approval_project_rules_protected_branches · optional · ON DELETE CASCADE · indexedapproval_project_rules_protected_branches.approval_project_rule_id → approval_project_rules.id — one approval_project_rules, many approval_project_rules_protected_branches · required · ON DELETE CASCADE · indexedapproval_project_rules_protected_branches.protected_branch_id → protected_branches.id — one protected_branches, many approval_project_rules_protected_branches · required · ON DELETE CASCADE · indexedapproval_project_rules_users.project_id → projects.id — one projects, many approval_project_rules_users · optional · ON DELETE CASCADE · indexedapproval_project_rules_users.approval_project_rule_id → approval_project_rules.id — one approval_project_rules, many approval_project_rules_users · required · ON DELETE CASCADE · indexedapproval_project_rules_users.user_id → users.id — one users, many approval_project_rules_users · required · ON DELETE CASCADE · indexedscan_result_policy_id — approval_group_rules.scan_result_policy_id may be NULL, so the relationship to scan_result_policies is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.approval_policy_rule_id — approval_group_rules.approval_policy_rule_id may be NULL, so the relationship to approval_policy_rules is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.security_orchestration_policy_configuration_id — approval_group_rules.security_orchestration_policy_configuration_id may be NULL, so the relationship to security_orchestration_policy_configurations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.scan_result_policy_id, approval_policy_rule_id, security_orchestration_policy_configuration_id — approval_group_rules has several nullable FKs (scan_result_policy_id, approval_policy_rule_id, security_orchestration_policy_configuration_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
approvals_required | smallint | NOT NULL | 0 | ||
report_type | smallint | null | |||
rule_type | smallint | NOT NULL | 1 | ||
security_orchestration_policy_configuration_id | bigint | null | security_orchestration_policy_configurations.id 1:N · CASCADE | ||
scan_result_policy_id | bigint | null | scan_result_policies.id 1:N · CASCADE | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
applies_to_all_protected_branches | boolean | NOT NULL | FALSE | ||
approval_policy_rule_id | bigint | null | approval_policy_rules.id 1:N · CASCADE |
Table checks: char_length(name) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
approval_group_rule_id | bigint | NOT NULL | approval_group_rules.id 1:N · CASCADE | ||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE |
group_id — approval_group_rules_protected_branches.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
approval_group_rule_id | bigint | NOT NULL | approval_group_rules.id 1:N · CASCADE | ||
protected_branch_id | bigint | NOT NULL | protected_branches.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK group_id IS NOT NULL |
Table checks: group_id IS NOT NULL
group_id — approval_group_rules_users.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
approval_group_rule_id | bigint | NOT NULL | approval_group_rules.id 1:N · CASCADE | ||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK group_id IS NOT NULL |
Table checks: group_id IS NOT NULL
project_id — approval_merge_request_rule_sources.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
approval_merge_request_rule_id | bigint | NOT NULL | approval_merge_request_rules.id 1:1 · CASCADE | ||
approval_project_rule_id | bigint | NOT NULL | approval_project_rules.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
security_orchestration_policy_configuration_id — approval_merge_request_rules.security_orchestration_policy_configuration_id may be NULL, so the relationship to security_orchestration_policy_configurations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.scan_result_policy_id — approval_merge_request_rules.scan_result_policy_id may be NULL, so the relationship to scan_result_policies is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — approval_merge_request_rules.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.security_orchestration_policy_configuration_id, scan_result_policy_id, project_id — approval_merge_request_rules has several nullable FKs (security_orchestration_policy_configuration_id, scan_result_policy_id, project_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
merge_request_id | bigint | NOT NULL | merge_requests.id 1:N · CASCADE | ||
approvals_required | smallint | NOT NULL | 0 | ||
name | varchar | NOT NULL | |||
rule_type | smallint | NOT NULL | 1 | ||
report_type | smallint | null | |||
section | text | null | CHECK char_length(section) <= 255 | ||
modified_from_project_rule | boolean | NOT NULL | FALSE | ||
orchestration_policy_idx | smallint | null | |||
vulnerabilities_allowed | smallint | NOT NULL | 0 | ||
scanners | text[] | NOT NULL | CAST('{}' AS text[]) | ||
severity_levels | text[] | NOT NULL | CAST('{}' AS text[]) | ||
vulnerability_states | text[] | NOT NULL | CAST('{new_needs_triage,new_dismissed}' AS text[]) | ||
security_orchestration_policy_configuration_id | bigint | null | security_orchestration_policy_configurations.id 1:N · CASCADE | ||
scan_result_policy_id | bigint | null | scan_result_policies.id 1:N · SET NULL | ||
applicable_post_merge | boolean | null | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL | |
approval_policy_rule_id | bigint | null | |||
role_approvers | integer[] | NOT NULL | CAST('{}' AS integer[]) | CHECK role_approvers = CAST('{}' AS integer[]) OR role_approvers <@ ARRAY[20, 30, 40, 50, 60] | |
approval_policy_action_idx | smallint | NOT NULL | 0 | ||
coverage_minimum_threshold | double precision | null | CHECK coverage_minimum_threshold IS NULL OR (coverage_minimum_threshold >= CAST(0 AS double precision) AND coverage_minimum_threshold <= CAST(100 AS double precision)) |
Table checks: coverage_minimum_threshold IS NULL OR (coverage_minimum_threshold >= CAST(0 AS double precision) AND coverage_minimum_threshold <= CAST(100 AS double precision)); char_length(section) <= 255; project_id IS NOT NULL; role_approvers = CAST('{}' AS integer[]) OR role_approvers <@ ARRAY[20, 30, 40, 50, 60]
project_id — approval_merge_request_rules_approved_approvers.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
approval_merge_request_rule_id | bigint | NOT NULL | approval_merge_request_rules.id 1:N · CASCADE | ||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
approval_merge_request_rule_id | bigint | NOT NULL | approval_merge_request_rules.id 1:N · CASCADE | ||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE |
project_id — approval_merge_request_rules_users.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
approval_merge_request_rule_id | bigint | NOT NULL | approval_merge_request_rules.id 1:N · CASCADE | ||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
user_id — approval_policy_merge_request_bypass_events.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
merge_request_id | bigint | NOT NULL | merge_requests.id 1:N · CASCADE | ||
security_policy_id | bigint | NOT NULL | security_policies.id 1:N · CASCADE | ||
user_id | bigint | null | users.id 1:N · SET NULL | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
reason | text | NOT NULL | CHECK length(pg_catalog.btrim(reason)) >= 1 AND length(pg_catalog.btrim(reason)) <= 1024 |
Table checks: length(pg_catalog.btrim(reason)) >= 1 AND length(pg_catalog.btrim(reason)) <= 1024
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
approval_policy_rule_id | bigint | NOT NULL | approval_policy_rules.id 1:N · CASCADE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
security_policy_id | bigint | NOT NULL | security_policies.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
rule_index | smallint | NOT NULL | |||
type | smallint | NOT NULL | |||
content | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
security_policy_management_project_id | bigint | NOT NULL | projects.id 1:N · CASCADE |
approval_policy_rule_id — approval_project_rules.approval_policy_rule_id may be NULL, so the relationship to approval_policy_rules is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.scan_result_policy_id — approval_project_rules.scan_result_policy_id may be NULL, so the relationship to scan_result_policies is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.security_orchestration_policy_configuration_id — approval_project_rules.security_orchestration_policy_configuration_id may be NULL, so the relationship to security_orchestration_policy_configurations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.approval_policy_rule_id, scan_result_policy_id, security_orchestration_policy_configuration_id — approval_project_rules has several nullable FKs (approval_policy_rule_id, scan_result_policy_id, security_orchestration_policy_configuration_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
approvals_required | smallint | NOT NULL | 0 | ||
name | varchar | NOT NULL | |||
rule_type | smallint | NOT NULL | 0 | ||
scanners | text[] | null | CAST('{}' AS text[]) | ||
vulnerabilities_allowed | smallint | NOT NULL | 0 | ||
severity_levels | text[] | NOT NULL | CAST('{}' AS text[]) | ||
report_type | smallint | null | |||
vulnerability_states | text[] | NOT NULL | CAST('{new_needs_triage,new_dismissed}' AS text[]) | ||
orchestration_policy_idx | smallint | null | |||
applies_to_all_protected_branches | boolean | NOT NULL | FALSE | ||
security_orchestration_policy_configuration_id | bigint | null | security_orchestration_policy_configurations.id 1:N · CASCADE | ||
scan_result_policy_id | bigint | null | scan_result_policies.id 1:N · CASCADE | ||
approval_policy_rule_id | bigint | null | approval_policy_rules.id 1:N · CASCADE | ||
approval_policy_action_idx | smallint | NOT NULL | 0 | ||
coverage_minimum_threshold | double precision | null | CHECK coverage_minimum_threshold IS NULL OR (coverage_minimum_threshold >= CAST(0 AS double precision) AND coverage_minimum_threshold <= CAST(100 AS double precision)) |
Table checks: coverage_minimum_threshold IS NULL OR (coverage_minimum_threshold >= CAST(0 AS double precision) AND coverage_minimum_threshold <= CAST(100 AS double precision))
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
approval_project_rule_id | bigint | NOT NULL | approval_project_rules.id 1:N · CASCADE | ||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE |
project_id — approval_project_rules_protected_branches.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
approval_project_rule_id PK | bigint | NOT NULL | approval_project_rules.id 1:N · CASCADE | ||
protected_branch_id PK | bigint | NOT NULL | protected_branches.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
project_id — approval_project_rules_users.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
approval_project_rule_id | bigint | NOT NULL | approval_project_rules.id 1:N · CASCADE | ||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
Tables whose names start with ascp_.
ascp_component_dependencies.component_id → ascp_components.id — one ascp_components, many ascp_component_dependencies · required · ON DELETE CASCADE · indexedascp_component_dependencies.dependency_id → ascp_components.id — one ascp_components, many ascp_component_dependencies · required · ON DELETE CASCADE · indexedascp_components.scan_id → ascp_scans.id — one ascp_scans, many ascp_components · required · ON DELETE CASCADE · indexedascp_scans.base_scan_id → ascp_scans.id — one ascp_scans, many ascp_scans · optional · ON DELETE SET NULL · indexedascp_security_contexts.component_id → ascp_components.id — one ascp_components, many ascp_security_contexts · required · ON DELETE CASCADE · indexedascp_security_contexts.scan_id → ascp_scans.id — one ascp_scans, many ascp_security_contexts · required · ON DELETE CASCADE · indexedascp_security_guidelines.scan_id → ascp_scans.id — one ascp_scans, many ascp_security_guidelines · required · ON DELETE CASCADE · indexedascp_security_guidelines.security_context_id → ascp_security_contexts.id — one ascp_security_contexts, many ascp_security_guidelines · required · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | |||
component_id | bigint | NOT NULL | ascp_components.id 1:N · CASCADE | ||
dependency_id | bigint | NOT NULL | ascp_components.id 1:N · CASCADE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | |||
scan_id | bigint | NOT NULL | ascp_scans.id 1:N · CASCADE | ||
title | text | NOT NULL | CHECK char_length(title) <= 255 | ||
sub_directory | text | NOT NULL | CHECK char_length(sub_directory) <= 1024 | ||
description | text | null | CHECK char_length(description) <= 4096 | ||
expected_user_behavior | text | null | CHECK char_length(expected_user_behavior) <= 4096 |
Table checks: char_length(expected_user_behavior) <= 4096; char_length(title) <= 255; char_length(description) <= 4096; char_length(sub_directory) <= 1024
base_scan_id — ascp_scans.base_scan_id may be NULL, so the relationship to ascp_scans is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | |||
base_scan_id | bigint | null | ascp_scans.id 1:N · SET NULL | ||
scan_sequence | integer | NOT NULL | |||
scan_type | smallint | NOT NULL | 0 | ||
commit_sha | text | NOT NULL | CHECK char_length(commit_sha) <= 64 | ||
base_commit_sha | text | null | CHECK char_length(base_commit_sha) <= 64 |
Table checks: char_length(commit_sha) <= 64; char_length(base_commit_sha) <= 64
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | |||
component_id | bigint | NOT NULL | ascp_components.id 1:N · CASCADE | ||
scan_id | bigint | NOT NULL | ascp_scans.id 1:N · CASCADE | ||
summary | text | null | CHECK char_length(summary) <= 4096 | ||
authentication_model | text | null | CHECK char_length(authentication_model) <= 1024 | ||
authorization_model | text | null | CHECK char_length(authorization_model) <= 1024 | ||
data_sensitivity | text | null | CHECK char_length(data_sensitivity) <= 255 |
Table checks: char_length(authentication_model) <= 1024; char_length(summary) <= 4096; char_length(data_sensitivity) <= 255; char_length(authorization_model) <= 1024
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | |||
security_context_id | bigint | NOT NULL | ascp_security_contexts.id 1:N · CASCADE | ||
scan_id | bigint | NOT NULL | ascp_scans.id 1:N · CASCADE | ||
severity_if_violated | smallint | NOT NULL | 0 | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
operation | text | NOT NULL | CHECK char_length(operation) <= 1024 | ||
legitimate_use | text | null | CHECK char_length(legitimate_use) <= 4096 | ||
security_boundary | text | null | CHECK char_length(security_boundary) <= 4096 | ||
business_context | text | null | CHECK char_length(business_context) <= 4096 |
Table checks: char_length(legitimate_use) <= 4096; char_length(name) <= 255; char_length(operation) <= 1024; char_length(business_context) <= 4096; char_length(security_boundary) <= 4096
Tables whose names start with board_.
board_assignees.project_id → projects.id — one projects, many board_assignees · optional · ON DELETE CASCADE · indexedboard_assignees.group_id → namespaces.id — one namespaces, many board_assignees · optional · ON DELETE CASCADE · indexedboard_assignees.assignee_id → users.id — one users, many board_assignees · required · ON DELETE CASCADE · indexedboard_assignees.board_id → boards.id — one boards, many board_assignees · required · ON DELETE CASCADE · indexedboard_group_recent_visits.user_id → users.id — one users, many board_group_recent_visits · optional · ON DELETE CASCADE · indexedboard_group_recent_visits.board_id → boards.id — one boards, many board_group_recent_visits · optional · ON DELETE CASCADE · indexedboard_group_recent_visits.group_id → namespaces.id — one namespaces, many board_group_recent_visits · optional · ON DELETE CASCADE · indexedboard_labels.group_id → namespaces.id — one namespaces, many board_labels · optional · ON DELETE CASCADE · indexedboard_labels.project_id → projects.id — one projects, many board_labels · optional · ON DELETE CASCADE · indexedboard_labels.label_id → labels.id — one labels, many board_labels · required · ON DELETE CASCADE · indexedboard_labels.board_id → boards.id — one boards, many board_labels · required · ON DELETE CASCADE · indexedboard_project_recent_visits.project_id → projects.id — one projects, many board_project_recent_visits · optional · ON DELETE CASCADE · indexedboard_project_recent_visits.board_id → boards.id — one boards, many board_project_recent_visits · optional · ON DELETE CASCADE · indexedboard_project_recent_visits.user_id → users.id — one users, many board_project_recent_visits · optional · ON DELETE CASCADE · indexedboard_user_preferences.group_id → namespaces.id — one namespaces, many board_user_preferences · optional · ON DELETE CASCADE · indexedboard_user_preferences.project_id → projects.id — one projects, many board_user_preferences · optional · ON DELETE CASCADE · indexedboard_user_preferences.user_id → users.id — one users, many board_user_preferences · required · ON DELETE CASCADE · indexedboard_user_preferences.board_id → boards.id — one boards, many board_user_preferences · required · ON DELETE CASCADE · indexedproject_id — board_assignees.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — board_assignees.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
board_id | bigint | NOT NULL | boards.id 1:N · CASCADE | ||
assignee_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE |
Table checks: num_nonnulls(group_id, project_id) = 1
user_id — board_group_recent_visits.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.board_id — board_group_recent_visits.board_id may be NULL, so the relationship to boards is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — board_group_recent_visits.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id, board_id, group_id — board_group_recent_visits has several nullable FKs (user_id, board_id, group_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
user_id | bigint | null | users.id 1:N · CASCADE | CHECK user_id IS NOT NULL | |
board_id | bigint | null | boards.id 1:N · CASCADE | CHECK board_id IS NOT NULL | |
group_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK group_id IS NOT NULL |
Table checks: user_id IS NOT NULL; group_id IS NOT NULL; board_id IS NOT NULL
group_id — board_labels.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — board_labels.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
board_id | bigint | NOT NULL | boards.id 1:N · CASCADE | ||
label_id | bigint | NOT NULL | labels.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE |
Table checks: num_nonnulls(group_id, project_id) = 1
project_id — board_project_recent_visits.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.board_id — board_project_recent_visits.board_id may be NULL, so the relationship to boards is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id — board_project_recent_visits.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id, board_id, user_id — board_project_recent_visits has several nullable FKs (project_id, board_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
user_id | bigint | null | users.id 1:N · CASCADE | CHECK user_id IS NOT NULL | |
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL | |
board_id | bigint | null | boards.id 1:N · CASCADE | CHECK board_id IS NOT NULL |
Table checks: board_id IS NOT NULL; project_id IS NOT NULL; user_id IS NOT NULL
group_id — board_user_preferences.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — board_user_preferences.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
board_id | bigint | NOT NULL | boards.id 1:N · CASCADE | ||
hide_labels | boolean | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE |
Table checks: num_nonnulls(group_id, project_id) = 1
Tables whose names start with boards_.
boards.iteration_cadence_id → iterations_cadences.id — one iterations_cadences, many boards · optional · ON DELETE CASCADE · indexedboards.project_id → projects.id — one projects, many boards · optional · ON DELETE CASCADE · indexedboards.group_id → namespaces.id — one namespaces, many boards · optional · ON DELETE CASCADE · indexedboards_epic_board_labels.group_id → namespaces.id — one namespaces, many boards_epic_board_labels · optional · ON DELETE CASCADE · indexedboards_epic_board_labels.label_id → labels.id — one labels, many boards_epic_board_labels · required · ON DELETE CASCADE · indexedboards_epic_board_labels.epic_board_id → boards_epic_boards.id — one boards_epic_boards, many boards_epic_board_labels · required · ON DELETE CASCADE · indexedboards_epic_board_positions.group_id → namespaces.id — one namespaces, many boards_epic_board_positions · optional · ON DELETE CASCADE · indexedboards_epic_board_positions.epic_id → epics.id — one epics, many boards_epic_board_positions · required · ON DELETE CASCADE · indexedboards_epic_board_positions.epic_board_id → boards_epic_boards.id — one boards_epic_boards, many boards_epic_board_positions · required · ON DELETE CASCADE · indexedboards_epic_board_recent_visits.user_id → users.id — one users, many boards_epic_board_recent_visits · required · ON DELETE CASCADE · indexedboards_epic_board_recent_visits.group_id → namespaces.id — one namespaces, many boards_epic_board_recent_visits · required · ON DELETE CASCADE · indexedboards_epic_board_recent_visits.epic_board_id → boards_epic_boards.id — one boards_epic_boards, many boards_epic_board_recent_visits · required · ON DELETE CASCADE · indexedboards_epic_boards.group_id → namespaces.id — one namespaces, many boards_epic_boards · required · ON DELETE CASCADE · indexedboards_epic_list_user_preferences.epic_list_id → boards_epic_lists.id — one boards_epic_lists, many boards_epic_list_user_preferences · required · ON DELETE CASCADE · indexedboards_epic_list_user_preferences.group_id → namespaces.id — one namespaces, many boards_epic_list_user_preferences · optional · ON DELETE CASCADE · indexedboards_epic_list_user_preferences.user_id → users.id — one users, many boards_epic_list_user_preferences · required · ON DELETE CASCADE · indexedboards_epic_lists.group_id → namespaces.id — one namespaces, many boards_epic_lists · optional · ON DELETE CASCADE · indexedboards_epic_lists.epic_board_id → boards_epic_boards.id — one boards_epic_boards, many boards_epic_lists · required · ON DELETE CASCADE · indexedboards_epic_lists.label_id → labels.id — one labels, many boards_epic_lists · optional · ON DELETE CASCADE · indexedboards_epic_user_preferences.group_id → namespaces.id — one namespaces, many boards_epic_user_preferences · optional · ON DELETE CASCADE · indexedboards_epic_user_preferences.board_id → boards.id — one boards, many boards_epic_user_preferences · required · ON DELETE CASCADE · indexedboards_epic_user_preferences.epic_id → epics.id — one epics, many boards_epic_user_preferences · required · ON DELETE CASCADE · indexedboards_epic_user_preferences.user_id → users.id — one users, many boards_epic_user_preferences · required · ON DELETE CASCADE · indexediteration_cadence_id — boards.iteration_cadence_id may be NULL, so the relationship to iterations_cadences is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — boards.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — boards.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.iteration_cadence_id, project_id, group_id — boards has several nullable FKs (iteration_cadence_id, project_id, group_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL | |||
name | varchar | NOT NULL | CAST('Development' AS varchar) | ||
milestone_id | bigint | null | |||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
weight | integer | null | |||
hide_backlog_list | boolean | NOT NULL | FALSE | ||
hide_closed_list | boolean | NOT NULL | FALSE | ||
iteration_id | bigint | null | |||
iteration_cadence_id | bigint | null | iterations_cadences.id 1:N · CASCADE |
Table checks: num_nonnulls(group_id, project_id) = 1
group_id — boards_epic_board_labels.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
epic_board_id | bigint | NOT NULL | boards_epic_boards.id 1:N · CASCADE | ||
label_id | bigint | NOT NULL | labels.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK group_id IS NOT NULL |
Table checks: group_id IS NOT NULL
group_id — boards_epic_board_positions.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
epic_board_id | bigint | NOT NULL | boards_epic_boards.id 1:N · CASCADE | ||
epic_id | bigint | NOT NULL | epics.id 1:N · CASCADE | ||
relative_position | integer | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
group_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK group_id IS NOT NULL |
Table checks: group_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
epic_board_id | bigint | NOT NULL | boards_epic_boards.id 1:N · CASCADE | ||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
hide_backlog_list | boolean | NOT NULL | FALSE | ||
hide_closed_list | boolean | NOT NULL | FALSE | ||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
name | text | NOT NULL | CAST('Development' AS text) | CHECK char_length(name) <= 255 | |
display_colors | boolean | NOT NULL | TRUE |
Table checks: char_length(name) <= 255
group_id — boards_epic_list_user_preferences.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
epic_list_id | bigint | NOT NULL | boards_epic_lists.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
collapsed | boolean | NOT NULL | FALSE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK group_id IS NOT NULL |
Table checks: group_id IS NOT NULL
group_id — boards_epic_lists.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.label_id — boards_epic_lists.label_id may be NULL, so the relationship to labels is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id, label_id — boards_epic_lists has several nullable FKs (group_id, label_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
epic_board_id | bigint | NOT NULL | boards_epic_boards.id 1:N · CASCADE | ||
label_id | bigint | null | labels.id 1:N · CASCADE | ||
position | integer | null | |||
list_type | smallint | NOT NULL | 1 | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK group_id IS NOT NULL |
Table checks: list_type <> 1 OR (position IS NOT NULL AND position >= 0); group_id IS NOT NULL
group_id — boards_epic_user_preferences.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
board_id | bigint | NOT NULL | boards.id 1:N · CASCADE | ||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
epic_id | bigint | NOT NULL | epics.id 1:N · CASCADE | ||
collapsed | boolean | NOT NULL | FALSE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK group_id IS NOT NULL |
Table checks: group_id IS NOT NULL
Tables whose names start with bulk_.
bulk_import_batch_trackers.project_id → projects.id — one projects, many bulk_import_batch_trackers · optional · ON DELETE CASCADE · indexedbulk_import_batch_trackers.organization_id → organizations.id — one organizations, many bulk_import_batch_trackers · optional · ON DELETE CASCADE · indexedbulk_import_batch_trackers.namespace_id → namespaces.id — one namespaces, many bulk_import_batch_trackers · optional · ON DELETE CASCADE · indexedbulk_import_batch_trackers.tracker_id → bulk_import_trackers.id — one bulk_import_trackers, many bulk_import_batch_trackers · required · ON DELETE CASCADE · indexedbulk_import_configurations.organization_id → organizations.id — one organizations, many bulk_import_configurations · optional · ON DELETE CASCADE · indexedbulk_import_configurations.bulk_import_id → bulk_imports.id — one bulk_imports, many bulk_import_configurations · required · ON DELETE CASCADE · indexedbulk_import_entities.organization_id → organizations.id — one organizations, many bulk_import_entities · optional · ON DELETE CASCADE · indexedbulk_import_entities.namespace_id → namespaces.id — one namespaces, many bulk_import_entities · optional · ON DELETE CASCADE · indexedbulk_import_entities.parent_id → bulk_import_entities.id — one bulk_import_entities, many bulk_import_entities · optional · ON DELETE CASCADE · indexedbulk_import_entities.bulk_import_id → bulk_imports.id — one bulk_imports, many bulk_import_entities · required · ON DELETE CASCADE · indexedbulk_import_entities.project_id → projects.id — one projects, many bulk_import_entities · optional · ON DELETE CASCADE · indexedbulk_import_export_batches.group_id → namespaces.id — one namespaces, many bulk_import_export_batches · optional · ON DELETE CASCADE · indexedbulk_import_export_batches.project_id → projects.id — one projects, many bulk_import_export_batches · optional · ON DELETE CASCADE · indexedbulk_import_export_batches.export_id → bulk_import_exports.id — one bulk_import_exports, many bulk_import_export_batches · required · ON DELETE CASCADE · indexedbulk_import_export_upload_upload_states.bulk_import_export_upload_upload_id → bulk_import_export_upload_uploads.id — one bulk_import_export_upload_uploads, at most one bulk_import_export_upload_upload_states · required · ON DELETE CASCADE · indexedbulk_import_export_upload_upload_states.project_id → projects.id — one projects, many bulk_import_export_upload_upload_states · required · ON DELETE CASCADE · indexedbulk_import_export_uploads.group_id → namespaces.id — one namespaces, many bulk_import_export_uploads · optional · ON DELETE CASCADE · indexedbulk_import_export_uploads.batch_id → bulk_import_export_batches.id — one bulk_import_export_batches, many bulk_import_export_uploads · optional · ON DELETE CASCADE · indexedbulk_import_export_uploads.project_id → projects.id — one projects, many bulk_import_export_uploads · optional · ON DELETE CASCADE · indexedbulk_import_export_uploads.export_id → bulk_import_exports.id — one bulk_import_exports, many bulk_import_export_uploads · required · ON DELETE CASCADE · indexedbulk_import_exports.offline_export_id → import_offline_exports.id — one import_offline_exports, many bulk_import_exports · optional · ON DELETE CASCADE · indexedbulk_import_exports.project_id → projects.id — one projects, many bulk_import_exports · optional · ON DELETE CASCADE · indexedbulk_import_exports.group_id → namespaces.id — one namespaces, many bulk_import_exports · optional · ON DELETE CASCADE · indexedbulk_import_exports.user_id → users.id — one users, many bulk_import_exports · optional · ON DELETE CASCADE · indexedbulk_import_failures.project_id → projects.id — one projects, many bulk_import_failures · optional · ON DELETE CASCADE · indexedbulk_import_failures.organization_id → organizations.id — one organizations, many bulk_import_failures · optional · ON DELETE CASCADE · indexedbulk_import_failures.namespace_id → namespaces.id — one namespaces, many bulk_import_failures · optional · ON DELETE CASCADE · indexedbulk_import_failures.bulk_import_entity_id → bulk_import_entities.id — one bulk_import_entities, many bulk_import_failures · required · ON DELETE CASCADE · indexedbulk_import_trackers.organization_id → organizations.id — one organizations, many bulk_import_trackers · optional · ON DELETE CASCADE · indexedbulk_import_trackers.project_id → projects.id — one projects, many bulk_import_trackers · optional · ON DELETE CASCADE · indexedbulk_import_trackers.namespace_id → namespaces.id — one namespaces, many bulk_import_trackers · optional · ON DELETE CASCADE · indexedbulk_import_trackers.bulk_import_entity_id → bulk_import_entities.id — one bulk_import_entities, many bulk_import_trackers · required · ON DELETE CASCADE · indexedbulk_imports.organization_id → organizations.id — one organizations, many bulk_imports · required · ON DELETE CASCADE · indexedbulk_imports.user_id → users.id — one users, many bulk_imports · required · ON DELETE CASCADE · indexedproject_id — bulk_import_batch_trackers.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id — bulk_import_batch_trackers.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id — bulk_import_batch_trackers.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
tracker_id | bigint | NOT NULL | bulk_import_trackers.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
batch_number | integer | NOT NULL | 0 | ||
fetched_objects_count | integer | NOT NULL | 0 | ||
imported_objects_count | integer | NOT NULL | 0 | ||
error | text | null | CHECK char_length(error) <= 255 | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE |
Table checks: num_nonnulls(namespace_id, organization_id, project_id) = 1; char_length(error) <= 255
organization_id — bulk_import_configurations.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
bulk_import_id | bigint | NOT NULL | bulk_imports.id 1:N · CASCADE | ||
encrypted_url | text | null | |||
encrypted_url_iv | text | null | |||
encrypted_access_token | text | null | |||
encrypted_access_token_iv | text | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
organization_id | bigint | null | organizations.id 1:N · CASCADE | CHECK organization_id IS NOT NULL |
Table checks: organization_id IS NOT NULL
organization_id — bulk_import_entities.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id — bulk_import_entities.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.parent_id — bulk_import_entities.parent_id may be NULL, so the relationship to bulk_import_entities is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — bulk_import_entities.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id, namespace_id, parent_id, project_id — bulk_import_entities has several nullable FKs (organization_id, namespace_id, parent_id, project_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
bulk_import_id | bigint | NOT NULL | bulk_imports.id 1:N · CASCADE | ||
parent_id | bigint | null | bulk_import_entities.id 1:N · CASCADE | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
source_type | smallint | NOT NULL | |||
source_full_path | text | NOT NULL | CHECK char_length(source_full_path) <= 255 | ||
destination_name | text | NOT NULL | CHECK char_length(destination_name) <= 255 | ||
destination_namespace | text | NOT NULL | CHECK char_length(destination_namespace) <= 255 | ||
status | smallint | NOT NULL | |||
jid | text | null | CHECK char_length(jid) <= 255 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
migrate_projects | boolean | NOT NULL | TRUE | ||
has_failures | boolean | null | FALSE | ||
migrate_memberships | boolean | NOT NULL | TRUE | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE | ||
source_xid | bigint | null |
Table checks: char_length(source_full_path) <= 255; num_nonnulls(namespace_id, organization_id, project_id) = 1; char_length(destination_name) <= 255; char_length(jid) <= 255; char_length(destination_namespace) <= 255
group_id — bulk_import_export_batches.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — bulk_import_export_batches.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
export_id | bigint | NOT NULL | bulk_import_exports.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
batch_number | integer | NOT NULL | 0 | ||
objects_count | integer | NOT NULL | 0 | ||
error | text | null | CHECK char_length(error) <= 255 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE |
Table checks: char_length(error) <= 255; num_nonnulls(group_id, project_id) = 1
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
bulk_import_export_upload_upload_id | bigint | NOT NULL | bulk_import_export_upload_uploads.id 1:1 · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 |
Table checks: char_length(verification_failure) <= 255
model_type — model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.model_type, model_id — model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | nextval(CAST('uploads_id_seq' AS regclass)) | ||
size | bigint | NOT NULL | |||
model_id | bigint | NOT NULL | |||
uploaded_by_user_id | bigint | null | |||
organization_id | bigint | null | |||
namespace_id | bigint | null | |||
project_id | bigint | null | |||
created_at | timestamp | null | |||
store | integer | NOT NULL | 1 | ||
version | integer | null | 1 | ||
path | text | NOT NULL | CHECK char_length(path) <= 511 | ||
checksum | text | null | CHECK char_length(checksum) <= 64 | ||
model_type PK | text | NOT NULL | |||
uploader | text | NOT NULL | |||
mount_point | text | null | |||
secret | text | null |
Table checks: char_length(path) <= 511; char_length(checksum) <= 64; num_nonnulls(namespace_id, project_id) = 1
group_id — bulk_import_export_uploads.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.batch_id — bulk_import_export_uploads.batch_id may be NULL, so the relationship to bulk_import_export_batches is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — bulk_import_export_uploads.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id, batch_id, project_id — bulk_import_export_uploads has several nullable FKs (group_id, batch_id, project_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
export_id | bigint | NOT NULL | bulk_import_exports.id 1:N · CASCADE | ||
updated_at | timestamptz | NOT NULL | |||
export_file | text | null | CHECK char_length(export_file) <= 255 | ||
batch_id | bigint | null | bulk_import_export_batches.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE |
Table checks: char_length(export_file) <= 255; num_nonnulls(group_id, project_id) = 1
offline_export_id — bulk_import_exports.offline_export_id may be NULL, so the relationship to import_offline_exports is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — bulk_import_exports.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — bulk_import_exports.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id — bulk_import_exports.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.offline_export_id, project_id, group_id, user_id — bulk_import_exports has several nullable FKs (offline_export_id, project_id, group_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
relation | text | NOT NULL | CHECK char_length(relation) <= 255 | ||
jid | text | null | CHECK char_length(jid) <= 255 | ||
error | text | null | CHECK char_length(error) <= 255 | ||
batched | boolean | NOT NULL | FALSE | ||
batches_count | integer | NOT NULL | 0 | ||
total_objects_count | integer | NOT NULL | 0 | ||
user_id | bigint | null | users.id 1:N · CASCADE | ||
offline_export_id | bigint | null | import_offline_exports.id 1:N · CASCADE | ||
organization_id | bigint | null |
Table checks: char_length(relation) <= 255; char_length(error) <= 255; char_length(jid) <= 255; num_nonnulls(group_id, project_id) = 1
project_id — bulk_import_failures.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id — bulk_import_failures.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id — bulk_import_failures.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
bulk_import_entity_id | bigint | NOT NULL | bulk_import_entities.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
pipeline_class | text | NOT NULL | CHECK char_length(pipeline_class) <= 255 | ||
exception_class | text | NOT NULL | CHECK char_length(exception_class) <= 255 | ||
exception_message | text | NOT NULL | CHECK char_length(exception_message) <= 255 | ||
correlation_id_value | text | null | CHECK char_length(correlation_id_value) <= 255 | ||
pipeline_step | text | null | CHECK char_length(pipeline_step) <= 255 | ||
source_url | text | null | CHECK char_length(source_url) <= 255 | ||
source_title | text | null | CHECK char_length(source_title) <= 255 | ||
subrelation | text | null | CHECK char_length(subrelation) <= 255 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE |
Table checks: char_length(pipeline_class) <= 255; char_length(exception_message) <= 255; char_length(pipeline_step) <= 255; char_length(source_title) <= 255; char_length(exception_class) <= 255; char_length(source_url) <= 255; char_length(correlation_id_value) <= 255; num_nonnulls(namespace_id, organization_id, project_id) = 1; char_length(subrelation) <= 255
organization_id — bulk_import_trackers.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — bulk_import_trackers.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id — bulk_import_trackers.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
bulk_import_entity_id | bigint | NOT NULL | bulk_import_entities.id 1:N · CASCADE | ||
relation | text | NOT NULL | CHECK char_length(relation) <= 255 | ||
next_page | text | null | CHECK char_length(next_page) <= 255 | ||
has_next_page | boolean | NOT NULL | FALSE | ||
jid | text | null | CHECK char_length(jid) <= 255 | ||
stage | smallint | NOT NULL | 0 | ||
status | smallint | NOT NULL | 0 | ||
created_at | timestamptz | null | |||
updated_at | timestamptz | null | |||
batched | boolean | null | FALSE | ||
source_objects_count | bigint | NOT NULL | 0 | ||
fetched_objects_count | bigint | NOT NULL | 0 | ||
imported_objects_count | bigint | NOT NULL | 0 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE |
Table checks: char_length(relation) <= 255; char_length(next_page) <= 255; num_nonnulls(namespace_id, organization_id, project_id) = 1; char_length(jid) <= 255; …(has_next_page) OR next_page IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
source_type | smallint | NOT NULL | |||
status | smallint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
source_version | text | null | CHECK char_length(source_version) <= 63 | ||
source_enterprise | boolean | NOT NULL | TRUE | ||
has_failures | boolean | null | FALSE | ||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE |
Table checks: char_length(source_version) <= 63
Tables whose names start with catalog_.
catalog_bundled_resource_components.catalog_bundled_version_id → catalog_bundled_resource_versions.id — one catalog_bundled_resource_versions, many catalog_bundled_resource_components · required · ON DELETE CASCADE · indexedcatalog_bundled_resource_components.catalog_bundled_resource_id → catalog_bundled_resources.id — one catalog_bundled_resources, many catalog_bundled_resource_components · required · ON DELETE CASCADE · indexedcatalog_bundled_resource_versions.catalog_bundled_resource_id → catalog_bundled_resources.id — one catalog_bundled_resources, many catalog_bundled_resource_versions · required · ON DELETE CASCADE · indexedcatalog_resource_component_last_usages.component_project_id → projects.id — one projects, many catalog_resource_component_last_usages · required · ON DELETE CASCADE · indexedcatalog_resource_component_last_usages.catalog_resource_id → catalog_resources.id — one catalog_resources, many catalog_resource_component_last_usages · required · ON DELETE CASCADE · indexedcatalog_resource_component_last_usages.component_id → catalog_resource_components.id — one catalog_resource_components, many catalog_resource_component_last_usages · required · ON DELETE CASCADE · indexedcatalog_resource_components.project_id → projects.id — one projects, many catalog_resource_components · required · ON DELETE CASCADE · indexedcatalog_resource_components.version_id → catalog_resource_versions.id — one catalog_resource_versions, many catalog_resource_components · required · ON DELETE CASCADE · indexedcatalog_resource_components.catalog_resource_id → catalog_resources.id — one catalog_resources, many catalog_resource_components · required · ON DELETE CASCADE · indexedcatalog_resource_versions.release_id → releases.id — one releases, at most one catalog_resource_versions · required · ON DELETE CASCADE · indexedcatalog_resource_versions.project_id → projects.id — one projects, many catalog_resource_versions · required · ON DELETE CASCADE · indexedcatalog_resource_versions.catalog_resource_id → catalog_resources.id — one catalog_resources, many catalog_resource_versions · required · ON DELETE CASCADE · indexedcatalog_resource_versions.published_by_id → users.id — one users, many catalog_resource_versions · optional · ON DELETE SET NULL · indexedcatalog_resources.project_id → projects.id — one projects, at most one catalog_resources · required · ON DELETE CASCADE · indexedcatalog_verified_namespaces.namespace_id → namespaces.id — one namespaces, at most one catalog_verified_namespaces · required · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
catalog_bundled_resource_id | bigint | NOT NULL | catalog_bundled_resources.id 1:N · CASCADE | ||
catalog_bundled_version_id | bigint | NOT NULL | catalog_bundled_resource_versions.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
spec | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
file_store | smallint | NOT NULL | 1 | ||
file | text | null | CHECK char_length(file) <= 1024 |
Table checks: char_length(file) <= 1024; char_length(name) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
catalog_bundled_resource_id | bigint | NOT NULL | catalog_bundled_resources.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
released_at | timestamptz | null | |||
semver_major | integer | NOT NULL | |||
semver_minor | integer | NOT NULL | |||
semver_patch | integer | NOT NULL | |||
semver_prefixed | boolean | NOT NULL | FALSE | ||
semver_prerelease | text | null | CHECK char_length(semver_prerelease) <= 255 | ||
readme | text | null | |||
readme_html | text | null |
Table checks: char_length(semver_prerelease) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
latest_released_at | timestamptz | null | |||
server_fqdn | text | NOT NULL | CHECK char_length(server_fqdn) <= 255 | ||
full_path | text | NOT NULL | CHECK char_length(full_path) <= 1024 | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
description | text | null | CHECK char_length(description) <= 1024 | ||
search_vector | tsvector | null |
Table checks: char_length(description) <= 1024; char_length(name) <= 255; char_length(server_fqdn) <= 255; char_length(full_path) <= 1024; server_fqdn = lower(server_fqdn); full_path = lower(full_path)
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
component_id | bigint | NOT NULL | catalog_resource_components.id 1:N · CASCADE | ||
catalog_resource_id | bigint | NOT NULL | catalog_resources.id 1:N · CASCADE | ||
component_project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
used_by_project_id | bigint | NOT NULL | |||
last_used_date | date | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
catalog_resource_id | bigint | NOT NULL | catalog_resources.id 1:N · CASCADE | ||
version_id | bigint | NOT NULL | catalog_resource_versions.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
spec | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
last_30_day_usage_count | integer | NOT NULL | 0 | ||
last_30_day_usage_count_updated_at | timestamptz | NOT NULL | CAST('1970-01-01 00:00:00+00' AS timestamptz) | ||
component_type | smallint | null | 1 | CHECK component_type IS NOT NULL |
Table checks: component_type IS NOT NULL; char_length(name) <= 255
published_by_id — catalog_resource_versions.published_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
release_id | bigint | NOT NULL | releases.id 1:1 · CASCADE | ||
catalog_resource_id | bigint | NOT NULL | catalog_resources.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
released_at | timestamptz | NOT NULL | CAST('1970-01-01 00:00:00+00' AS timestamptz) | ||
semver_major | integer | null | |||
semver_minor | integer | null | |||
semver_patch | integer | null | |||
semver_prerelease | text | null | CHECK char_length(semver_prerelease) <= 255 | ||
semver_prefixed | boolean | null | FALSE | ||
cached_markdown_version | integer | null | |||
readme | text | null | |||
readme_html | text | null | |||
published_by_id | bigint | null | users.id 1:N · SET NULL |
Table checks: char_length(semver_prerelease) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
state | smallint | NOT NULL | 0 | ||
latest_released_at | timestamptz | null | |||
name | varchar | null | |||
description | text | null | |||
visibility_level | integer | NOT NULL | 0 | ||
search_vector | tsvector | null | |||
verification_level | smallint | null | 0 | ||
last_30_day_usage_count | integer | NOT NULL | 0 | ||
last_30_day_usage_count_updated_at | timestamptz | NOT NULL | now() |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
verification_level | smallint | NOT NULL | 0 |
Tables whose names start with cd_.
cd_application_flow_definitions.application_id → cd_applications.id — one cd_applications, many cd_application_flow_definitions · required · ON DELETE CASCADE · indexedcd_application_flow_definitions.organization_id → organizations.id — one organizations, many cd_application_flow_definitions · required · ON DELETE CASCADE · indexedcd_application_links.application_id → cd_applications.id — one cd_applications, many cd_application_links · required · ON DELETE CASCADE · indexedcd_application_links.organization_id → organizations.id — one organizations, many cd_application_links · required · ON DELETE CASCADE · indexedcd_applications.organization_id → organizations.id — one organizations, many cd_applications · optional · ON DELETE CASCADE · indexedcd_artifact_sources.group_id → namespaces.id — one namespaces, many cd_artifact_sources · optional · ON DELETE CASCADE · indexedcd_artifact_sources.service_id → cd_services.id — one cd_services, many cd_artifact_sources · required · ON DELETE CASCADE · indexedcd_artifact_sources.organization_id → organizations.id — one organizations, many cd_artifact_sources · optional · ON DELETE CASCADE · indexedcd_deployment_transitions.organization_id → organizations.id — one organizations, many cd_deployment_transitions · required · ON DELETE CASCADE · indexedcd_deployment_transitions.deployment_id → cd_deployments.id — one cd_deployments, many cd_deployment_transitions · required · ON DELETE CASCADE · indexedcd_deployments.service_id → cd_services.id — one cd_services, many cd_deployments · optional · ON DELETE CASCADE · indexedcd_deployments.group_id → namespaces.id — one namespaces, many cd_deployments · optional · ON DELETE CASCADE · indexedcd_deployments.rollout_environment_id → cd_rollout_environments.id — one cd_rollout_environments, many cd_deployments · optional · ON DELETE CASCADE · indexedcd_deployments.organization_id → organizations.id — one organizations, many cd_deployments · optional · ON DELETE CASCADE · indexedcd_environment_driver_bindings.environment_id → cd_environments.id — one cd_environments, many cd_environment_driver_bindings · required · ON DELETE CASCADE · indexedcd_environment_driver_bindings.organization_id → organizations.id — one organizations, many cd_environment_driver_bindings · required · ON DELETE CASCADE · indexedcd_environments.organization_id → organizations.id — one organizations, many cd_environments · optional · ON DELETE CASCADE · indexedcd_rollout_environments.driver_binding_id → cd_environment_driver_bindings.id — one cd_environment_driver_bindings, many cd_rollout_environments · required · ON DELETE CASCADE · indexedcd_rollout_environments.rollout_id → cd_rollouts.id — one cd_rollouts, many cd_rollout_environments · required · ON DELETE CASCADE · indexedcd_rollout_environments.environment_id → cd_environments.id — one cd_environments, many cd_rollout_environments · required · ON DELETE CASCADE · indexedcd_rollout_environments.previous_version_set_id → cd_version_sets.id — one cd_version_sets, many cd_rollout_environments · optional · ON DELETE SET NULL · indexedcd_rollout_environments.organization_id → organizations.id — one organizations, many cd_rollout_environments · required · ON DELETE CASCADE · indexedcd_rollout_steps.organization_id → organizations.id — one organizations, many cd_rollout_steps · required · ON DELETE CASCADE · indexedcd_rollout_steps.rollout_id → cd_rollouts.id — one cd_rollouts, many cd_rollout_steps · required · ON DELETE CASCADE · indexedcd_rollout_steps.rollout_environment_id → cd_rollout_environments.id — one cd_rollout_environments, many cd_rollout_steps · optional · ON DELETE CASCADE · indexedcd_rollout_transitions.organization_id → organizations.id — one organizations, many cd_rollout_transitions · required · ON DELETE CASCADE · indexedcd_rollout_transitions.rollout_id → cd_rollouts.id — one cd_rollouts, many cd_rollout_transitions · required · ON DELETE CASCADE · indexedcd_rollouts.application_flow_definition_id → cd_application_flow_definitions.id — one cd_application_flow_definitions, many cd_rollouts · optional · ON DELETE CASCADE · indexedcd_rollouts.organization_id → organizations.id — one organizations, many cd_rollouts · optional · ON DELETE CASCADE · indexedcd_rollouts.group_id → namespaces.id — one namespaces, many cd_rollouts · optional · ON DELETE CASCADE · indexedcd_rollouts.version_set_id → cd_version_sets.id — one cd_version_sets, many cd_rollouts · required · ON DELETE CASCADE · indexedcd_rollouts.application_id → cd_applications.id — one cd_applications, many cd_rollouts · optional · ON DELETE CASCADE · indexedcd_service_environment_healths.service_id → cd_services.id — one cd_services, many cd_service_environment_healths · required · ON DELETE CASCADE · indexedcd_service_environment_healths.environment_id → cd_environments.id — one cd_environments, many cd_service_environment_healths · required · ON DELETE CASCADE · indexedcd_service_environment_healths.organization_id → organizations.id — one organizations, many cd_service_environment_healths · required · ON DELETE CASCADE · indexedcd_services.application_id → cd_applications.id — one cd_applications, many cd_services · required · ON DELETE CASCADE · indexedcd_services.group_id → namespaces.id — one namespaces, many cd_services · optional · ON DELETE CASCADE · indexedcd_services.organization_id → organizations.id — one organizations, many cd_services · optional · ON DELETE CASCADE · indexedcd_version_set_entries.version_id → cd_versions.id — one cd_versions, many cd_version_set_entries · required · ON DELETE CASCADE · indexedcd_version_set_entries.service_id → cd_services.id — one cd_services, many cd_version_set_entries · required · ON DELETE CASCADE · indexedcd_version_set_entries.version_set_id → cd_version_sets.id — one cd_version_sets, many cd_version_set_entries · required · ON DELETE CASCADE · indexedcd_version_set_entries.group_id → namespaces.id — one namespaces, many cd_version_set_entries · optional · ON DELETE CASCADE · indexedcd_version_set_entries.organization_id → organizations.id — one organizations, many cd_version_set_entries · optional · ON DELETE CASCADE · indexedcd_version_set_entries.artifact_source_id → cd_artifact_sources.id — one cd_artifact_sources, many cd_version_set_entries · optional · ON DELETE CASCADE · indexedcd_version_sets.created_by_id → users.id — one users, many cd_version_sets · optional · ON DELETE SET NULL · indexedcd_version_sets.application_id → cd_applications.id — one cd_applications, many cd_version_sets · required · ON DELETE CASCADE · indexedcd_version_sets.group_id → namespaces.id — one namespaces, many cd_version_sets · optional · ON DELETE CASCADE · indexedcd_version_sets.organization_id → organizations.id — one organizations, many cd_version_sets · optional · ON DELETE CASCADE · indexedcd_versions.organization_id → organizations.id — one organizations, many cd_versions · optional · ON DELETE CASCADE · indexedcd_versions.artifact_source_id → cd_artifact_sources.id — one cd_artifact_sources, many cd_versions · required · ON DELETE CASCADE · indexedcd_versions.group_id → namespaces.id — one namespaces, many cd_versions · optional · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
application_id | bigint | NOT NULL | cd_applications.id 1:N · CASCADE | ||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
version | integer | NOT NULL | 1 | ||
file_store | smallint | NOT NULL | 1 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file | text | NOT NULL | CHECK char_length(file) <= 255 |
Table checks: char_length(file) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
application_id | bigint | NOT NULL | cd_applications.id 1:N · CASCADE | ||
link_type | smallint | NOT NULL | 0 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
url | text | NOT NULL | CHECK char_length(url) <= 2048 |
Table checks: char_length(name) <= 255; char_length(url) <= 2048
organization_id — cd_applications.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
description | text | null | CHECK char_length(description) <= 2000 | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE | CHECK organization_id IS NOT NULL | |
last_rollout_iid | integer | NOT NULL | 0 |
Table checks: char_length(description) <= 2000; char_length(name) <= 255; organization_id IS NOT NULL
group_id — cd_artifact_sources.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id — cd_artifact_sources.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id, organization_id — cd_artifact_sources has several nullable FKs (group_id, organization_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
service_id | bigint | NOT NULL | cd_services.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
organization_id | bigint | null | organizations.id 1:N · CASCADE | CHECK organization_id IS NOT NULL | |
source_ref | text | null | CHECK source_ref IS NOT NULL | ||
source_config | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(source_config) = CAST('object' AS text) | |
name | text | null | CHECK char_length(name) <= 255 |
Table checks: source_ref IS NOT NULL; char_length(name) <= 255; jsonb_typeof(source_config) = CAST('object' AS text); char_length(source_ref) <= 255; organization_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
deployment_id | bigint | NOT NULL | cd_deployments.id 1:N · CASCADE | ||
principal_id | bigint | null | |||
created_at | timestamptz | NOT NULL | |||
from_state | smallint | NOT NULL | |||
to_state | smallint | NOT NULL | |||
event | text | NOT NULL | CHECK char_length(event) <= 72 | ||
reason | text | null | CHECK char_length(reason) <= 2000 | ||
triggered_by | text | null | CHECK char_length(triggered_by) <= 255 | ||
principal | text | null | CHECK principal IS NOT NULL | ||
on_behalf_of | text | null | CHECK char_length(on_behalf_of) <= 255 |
Table checks: char_length(event) <= 72; char_length(reason) <= 2000; principal IS NOT NULL; char_length(on_behalf_of) <= 255; char_length(principal) <= 255; char_length(triggered_by) <= 255
service_id — cd_deployments.service_id may be NULL, so the relationship to cd_services is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — cd_deployments.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.rollout_environment_id — cd_deployments.rollout_environment_id may be NULL, so the relationship to cd_rollout_environments is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id — cd_deployments.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.service_id, group_id, rollout_environment_id, organization_id — cd_deployments has several nullable FKs (service_id, group_id, rollout_environment_id, organization_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
started_at | timestamptz | null | |||
finished_at | timestamptz | null | |||
state | smallint | NOT NULL | 0 | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE | CHECK organization_id IS NOT NULL | |
rollout_environment_id | bigint | null | cd_rollout_environments.id 1:N · CASCADE | CHECK rollout_environment_id IS NOT NULL | |
service_id | bigint | null | cd_services.id 1:N · CASCADE | CHECK service_id IS NOT NULL |
Table checks: service_id IS NOT NULL; rollout_environment_id IS NOT NULL; organization_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
environment_id | bigint | NOT NULL | cd_environments.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
version | integer | NOT NULL | 1 | ||
driver_ref | text | NOT NULL | CHECK char_length(driver_ref) <= 255 | ||
driver_config | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(driver_config) = CAST('object' AS text) |
Table checks: char_length(driver_ref) <= 255; jsonb_typeof(driver_config) = CAST('object' AS text)
organization_id — cd_environments.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
organization_id | bigint | null | organizations.id 1:N · CASCADE | CHECK organization_id IS NOT NULL | |
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
description | text | null | CHECK char_length(description) <= 1024 | ||
tier | smallint | NOT NULL | 0 |
Table checks: char_length(description) <= 1024; organization_id IS NOT NULL; char_length(name) <= 255
previous_version_set_id — cd_rollout_environments.previous_version_set_id may be NULL, so the relationship to cd_version_sets is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
rollout_id | bigint | NOT NULL | cd_rollouts.id 1:N · CASCADE | ||
environment_id | bigint | NOT NULL | cd_environments.id 1:N · CASCADE | ||
driver_binding_id | bigint | NOT NULL | cd_environment_driver_bindings.id 1:N · CASCADE | ||
previous_version_set_id | bigint | null | cd_version_sets.id 1:N · SET NULL | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
started_at | timestamptz | null | |||
finished_at | timestamptz | null | |||
position | integer | NOT NULL | |||
state | smallint | NOT NULL | 0 |
rollout_environment_id — cd_rollout_steps.rollout_environment_id may be NULL, so the relationship to cd_rollout_environments is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
rollout_id | bigint | NOT NULL | cd_rollouts.id 1:N · CASCADE | ||
rollout_environment_id | bigint | null | cd_rollout_environments.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
started_at | timestamptz | null | |||
finished_at | timestamptz | null | |||
state | smallint | NOT NULL | 0 | ||
path | text | NOT NULL | CHECK char_length(path) <= 255 | ||
parent_path | text | null | CHECK char_length(parent_path) <= 255 | ||
step_type | text | NOT NULL | CHECK char_length(step_type) <= 255 | ||
name | text | null | CHECK char_length(name) <= 255 | ||
error | text | null | CHECK char_length(error) <= 2000 | ||
params | jsonb | null | CHECK params IS NULL OR jsonb_typeof(params) = CAST('object' AS text) |
Table checks: char_length(parent_path) <= 255; char_length(error) <= 2000; char_length(step_type) <= 255; char_length(path) <= 255; params IS NULL OR jsonb_typeof(params) = CAST('object' AS text); char_length(name) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
rollout_id | bigint | NOT NULL | cd_rollouts.id 1:N · CASCADE | ||
principal_id | bigint | null | |||
created_at | timestamptz | NOT NULL | |||
from_state | smallint | NOT NULL | |||
to_state | smallint | NOT NULL | |||
event | text | NOT NULL | CHECK char_length(event) <= 72 | ||
reason | text | null | CHECK char_length(reason) <= 2000 | ||
triggered_by | text | null | CHECK char_length(triggered_by) <= 255 | ||
principal | text | null | CHECK principal IS NOT NULL | ||
on_behalf_of | text | null | CHECK char_length(on_behalf_of) <= 255 |
Table checks: char_length(on_behalf_of) <= 255; char_length(reason) <= 2000; char_length(triggered_by) <= 255; principal IS NOT NULL; char_length(principal) <= 255; char_length(event) <= 72
application_flow_definition_id — cd_rollouts.application_flow_definition_id may be NULL, so the relationship to cd_application_flow_definitions is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id — cd_rollouts.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — cd_rollouts.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.application_id — cd_rollouts.application_id may be NULL, so the relationship to cd_applications is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.application_flow_definition_id, organization_id, group_id, application_id — cd_rollouts has several nullable FKs (application_flow_definition_id, organization_id, group_id, application_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
version_set_id | bigint | NOT NULL | cd_version_sets.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
started_at | timestamptz | null | |||
finished_at | timestamptz | null | |||
state | smallint | NOT NULL | 0 | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE | CHECK organization_id IS NOT NULL | |
application_id | bigint | null | cd_applications.id 1:N · CASCADE | CHECK application_id IS NOT NULL | |
application_flow_definition_id | bigint | null | cd_application_flow_definitions.id 1:N · CASCADE | ||
workflow_ref | text | null | CHECK char_length(workflow_ref) <= 255 | ||
iid | integer | null | CHECK iid IS NOT NULL |
Table checks: iid IS NOT NULL; char_length(workflow_ref) <= 255; application_id IS NOT NULL; organization_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
service_id | bigint | NOT NULL | cd_services.id 1:N · CASCADE | ||
environment_id | bigint | NOT NULL | cd_environments.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
observed_at | timestamptz | NOT NULL | |||
health | smallint | NOT NULL | 0 |
group_id — cd_services.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id — cd_services.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id, organization_id — cd_services has several nullable FKs (group_id, organization_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
application_id | bigint | NOT NULL | cd_applications.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
description | text | null | CHECK char_length(description) <= 2000 | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE | CHECK organization_id IS NOT NULL |
Table checks: organization_id IS NOT NULL; char_length(name) <= 255; char_length(description) <= 2000
group_id — cd_version_set_entries.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id — cd_version_set_entries.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.artifact_source_id — cd_version_set_entries.artifact_source_id may be NULL, so the relationship to cd_artifact_sources is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id, organization_id, artifact_source_id — cd_version_set_entries has several nullable FKs (group_id, organization_id, artifact_source_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
version_set_id | bigint | NOT NULL | cd_version_sets.id 1:N · CASCADE | ||
version_id | bigint | NOT NULL | cd_versions.id 1:N · CASCADE | ||
service_id | bigint | NOT NULL | cd_services.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
organization_id | bigint | null | organizations.id 1:N · CASCADE | CHECK organization_id IS NOT NULL | |
artifact_source_id | bigint | null | cd_artifact_sources.id 1:N · CASCADE | CHECK artifact_source_id IS NOT NULL |
Table checks: organization_id IS NOT NULL; artifact_source_id IS NOT NULL
created_by_id — cd_version_sets.created_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — cd_version_sets.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id — cd_version_sets.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.created_by_id, group_id, organization_id — cd_version_sets has several nullable FKs (created_by_id, group_id, organization_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
application_id | bigint | NOT NULL | cd_applications.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
entries_digest | text | null | CHECK char_length(entries_digest) <= 64 | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE | CHECK organization_id IS NOT NULL | |
description | text | null | CHECK char_length(description) <= 2000 | ||
created_by_id | bigint | null | users.id 1:N · SET NULL |
Table checks: char_length(entries_digest) <= 64; char_length(name) <= 255; organization_id IS NOT NULL; char_length(description) <= 2000
organization_id — cd_versions.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — cd_versions.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id, group_id — cd_versions has several nullable FKs (organization_id, group_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
artifact_source_id | bigint | NOT NULL | cd_artifact_sources.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
digest | text | null | CHECK char_length(digest) <= 255 | ||
reference | text | null | CHECK char_length(reference) <= 1024 | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE | CHECK organization_id IS NOT NULL |
Table checks: char_length(reference) <= 1024; organization_id IS NOT NULL; char_length(name) <= 255; char_length(digest) <= 255
Tables whose names start with chat_.
chat_teams.namespace_id → namespaces.id — one namespaces, at most one chat_teams · required · ON DELETE CASCADE · indexedlast_used_at — last_used_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.chat_names references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | |||
team_id | varchar | NOT NULL | |||
team_domain | varchar | null | |||
chat_id | varchar | NOT NULL | |||
chat_name | varchar | null | |||
last_used_at | timestamp | null | |||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL | |||
encrypted_token | bytea | null | |||
encrypted_token_iv | bytea | null | |||
duo_privacy_notice_acknowledged_at | timestamptz | null |
created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
team_id | varchar | null | |||
name | varchar | null | |||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL |
Tables whose names start with ci_.
ci_build_pending_states.partition_id, build_id → p_ci_builds.partition_id, id — one p_ci_builds, at most one ci_build_pending_states · required · ON DELETE CASCADE · indexedci_build_report_results.partition_id, build_id → p_ci_builds.partition_id, id — one p_ci_builds, at most one ci_build_report_results · required · ON DELETE CASCADE · indexedci_build_trace_chunks.partition_id, build_id → p_ci_builds.partition_id, id — one p_ci_builds, many ci_build_trace_chunks · required · ON DELETE CASCADE · not indexedci_builds_runner_session.partition_id, build_id → p_ci_builds.partition_id, id — one p_ci_builds, at most one ci_builds_runner_session · required · ON DELETE CASCADE · indexedci_cost_settings.runner_id → instance_type_ci_runners.id — one instance_type_ci_runners, at most one ci_cost_settings · required · ON DELETE CASCADE · indexedci_daily_build_group_report_results.partition_id, last_pipeline_id → p_ci_pipelines.partition_id, id — one p_ci_pipelines, many ci_daily_build_group_report_results · required · ON DELETE CASCADE · not indexedci_hosted_runners.runner_id → instance_type_ci_runners.id — one instance_type_ci_runners, at most one ci_hosted_runners · required · ON DELETE CASCADE · indexedci_job_artifact_states.partition_id, job_artifact_id → p_ci_job_artifacts.partition_id, id — one p_ci_job_artifacts, at most one ci_job_artifact_states · required · ON DELETE CASCADE · indexedci_job_variables.partition_id, job_id → p_ci_builds.partition_id, id — one p_ci_builds, many ci_job_variables · required · ON DELETE CASCADE · not indexedci_pending_builds.partition_id, build_id → p_ci_builds.partition_id, id — one p_ci_builds, at most one ci_pending_builds · required · ON DELETE CASCADE · indexedci_pipeline_artifacts.partition_id, pipeline_id → p_ci_pipelines.partition_id, id — one p_ci_pipelines, many ci_pipeline_artifacts · required · ON DELETE CASCADE · not indexedci_pipeline_chat_data.partition_id, pipeline_id → p_ci_pipelines.partition_id, id — one p_ci_pipelines, many ci_pipeline_chat_data · required · ON DELETE CASCADE · not indexedci_pipeline_messages.partition_id, pipeline_id → p_ci_pipelines.partition_id, id — one p_ci_pipelines, many ci_pipeline_messages · required · ON DELETE CASCADE · not indexedci_pipeline_metadata.partition_id, pipeline_id → p_ci_pipelines.partition_id, id — one p_ci_pipelines, many ci_pipeline_metadata · required · ON DELETE CASCADE · not indexedci_pipeline_schedule_inputs.pipeline_schedule_id → ci_pipeline_schedules.id — one ci_pipeline_schedules, many ci_pipeline_schedule_inputs · required · ON DELETE CASCADE · indexedci_pipeline_schedule_variables.pipeline_schedule_id → ci_pipeline_schedules.id — one ci_pipeline_schedules, many ci_pipeline_schedule_variables · required · ON DELETE CASCADE · indexedci_resources.partition_id, build_id → p_ci_builds.partition_id, id — one p_ci_builds, many ci_resources · optional · ON DELETE SET NULL · not indexedci_resources.resource_group_id → ci_resource_groups.id — one ci_resource_groups, many ci_resources · required · ON DELETE CASCADE · indexedci_runner_controller_instance_level_scopings.runner_controller_id → ci_runner_controllers.id — one ci_runner_controllers, at most one ci_runner_controller_instance_level_scopings · required · ON DELETE CASCADE · indexedci_runner_controller_runner_level_scopings.runner_id, runner_type → ci_runners.id, runner_type — one ci_runners, many ci_runner_controller_runner_level_scopings · required · ON DELETE CASCADE · indexedci_runner_controller_runner_level_scopings.runner_controller_id → ci_runner_controllers.id — one ci_runner_controllers, many ci_runner_controller_runner_level_scopings · required · ON DELETE CASCADE · indexedci_runner_controller_tokens.runner_controller_id → ci_runner_controllers.id — one ci_runner_controllers, many ci_runner_controller_tokens · required · ON DELETE CASCADE · indexedci_runner_namespaces.runner_id → group_type_ci_runners.id — one group_type_ci_runners, many ci_runner_namespaces · optional · ON DELETE CASCADE · indexedci_runner_projects.runner_id → project_type_ci_runners.id — one project_type_ci_runners, many ci_runner_projects · required · ON DELETE CASCADE · indexedci_runner_taggings_group_type.runner_id, runner_type → group_type_ci_runners.id, runner_type — one group_type_ci_runners, many ci_runner_taggings_group_type · required · ON DELETE CASCADE · indexedci_runner_taggings_instance_type.runner_id, runner_type → instance_type_ci_runners.id, runner_type — one instance_type_ci_runners, many ci_runner_taggings_instance_type · required · ON DELETE CASCADE · indexedci_runner_taggings_project_type.runner_id, runner_type → project_type_ci_runners.id, runner_type — one project_type_ci_runners, many ci_runner_taggings_project_type · required · ON DELETE CASCADE · indexedci_running_builds.partition_id, build_id → p_ci_builds.partition_id, id — one p_ci_builds, at most one ci_running_builds · required · ON DELETE CASCADE · indexedci_sources_pipelines.source_partition_id, source_job_id → p_ci_builds.partition_id, id — one p_ci_builds, many ci_sources_pipelines · optional · ON DELETE CASCADE · not indexedci_sources_pipelines.source_partition_id, source_pipeline_id → p_ci_pipelines.partition_id, id — one p_ci_pipelines, many ci_sources_pipelines · optional · ON DELETE CASCADE · not indexedci_sources_pipelines.partition_id, pipeline_id → p_ci_pipelines.partition_id, id — one p_ci_pipelines, many ci_sources_pipelines · optional · ON DELETE CASCADE · not indexedci_sources_projects.partition_id, pipeline_id → p_ci_pipelines.partition_id, id — one p_ci_pipelines, many ci_sources_projects · required · ON DELETE CASCADE · not indexedci_unit_test_failures.partition_id, build_id → p_ci_builds.partition_id, id — one p_ci_builds, many ci_unit_test_failures · required · ON DELETE CASCADE · not indexedci_unit_test_failures.unit_test_id → ci_unit_tests.id — one ci_unit_tests, many ci_unit_test_failures · required · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
build_id | bigint | NOT NULL | p_ci_builds.partition_id 1:1 · CASCADE | ||
state | smallint | null | |||
failure_reason | smallint | null | |||
trace_checksum | bytea | null | |||
trace_bytesize | bigint | null | |||
partition_id | bigint | NOT NULL | p_ci_builds.partition_id 1:1 · CASCADE | ||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
build_id PK | bigint | NOT NULL | p_ci_builds.partition_id 1:1 · CASCADE | ||
project_id | bigint | NOT NULL | |||
data | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
partition_id PK | bigint | NOT NULL | p_ci_builds.partition_id 1:1 · CASCADE |
ci_build_runtime_environments references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
build_id PK | bigint | NOT NULL | |||
partition_id PK | bigint | NOT NULL | |||
runtime_environment_id | bigint | null | |||
runner_machine_id | bigint | null | |||
project_id | bigint | NOT NULL | |||
suspend_on_success | boolean | NOT NULL | FALSE | ||
suspend_on_failure | boolean | NOT NULL | FALSE |
partition_id, build_id — PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_builds must scan ci_build_trace_chunks to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to ci_build_trace_chunks maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_builds rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
chunk_index | integer | NOT NULL | |||
data_store | integer | NOT NULL | |||
raw_data | bytea | null | |||
checksum | bytea | null | |||
lock_version | integer | NOT NULL | 0 | ||
build_id | bigint | NOT NULL | p_ci_builds.partition_id 1:N · CASCADE | ||
partition_id | bigint | NOT NULL | p_ci_builds.partition_id 1:N · CASCADE | ||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
url | varchar | NOT NULL | |||
certificate | varchar | null | |||
authorization | varchar | null | |||
build_id | bigint | NOT NULL | p_ci_builds.partition_id 1:1 · CASCADE | ||
partition_id | bigint | NOT NULL | p_ci_builds.partition_id 1:1 · CASCADE | ||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
runner_id PK | bigint | NOT NULL | instance_type_ci_runners.id 1:1 · CASCADE | ||
standard_factor | double precision | NOT NULL | 1.0 | ||
os_contribution_factor | double precision | NOT NULL | 0.008 | ||
os_plan_factor | double precision | NOT NULL | 0.5 |
partition_id, last_pipeline_id — PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_pipelines must scan ci_daily_build_group_report_results to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to ci_daily_build_group_report_results maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_pipelines rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
date | date | NOT NULL | |||
project_id | bigint | NOT NULL | |||
last_pipeline_id | bigint | NOT NULL | p_ci_pipelines.partition_id 1:N · CASCADE | ||
ref_path | text | NOT NULL | |||
group_name | text | NOT NULL | |||
data | jsonb | NOT NULL | |||
default_branch | boolean | NOT NULL | FALSE | ||
group_id | bigint | null | |||
partition_id | bigint | NOT NULL | p_ci_pipelines.partition_id 1:N · CASCADE |
ci_deleted_objects references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
file_store | smallint | NOT NULL | 1 | ||
pick_up_at | timestamptz | NOT NULL | now() | ||
store_dir | text | NOT NULL | CHECK char_length(store_dir) <= 1024 | ||
file | text | NOT NULL | |||
project_id | bigint | null | CHECK project_id IS NOT NULL | ||
created_at | timestamptz | NOT NULL | now() |
Table checks: char_length(store_dir) <= 1024; project_id IS NOT NULL
ci_freeze_periods references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
freeze_start | varchar(998) | NOT NULL | |||
freeze_end | varchar(998) | NOT NULL | |||
cron_timezone | varchar(255) | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
ci_gitlab_hosted_runner_monthly_usages references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
runner_id | bigint | NOT NULL | |||
runner_duration_seconds | bigint | NOT NULL | 0 | ||
project_id | bigint | NOT NULL | |||
root_namespace_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
billing_month | date | NOT NULL | CHECK billing_month = date_trunc(CAST('month' AS text), CAST(billing_month AS timestamptz)) | ||
notification_level | integer | NOT NULL | 100 | ||
compute_minutes_used | numeric(18, 4) | NOT NULL | 0.0 |
Table checks: billing_month = date_trunc(CAST('month' AS text), CAST(billing_month AS timestamptz))
ci_group_variables references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
key | varchar | NOT NULL | |||
encrypted_value | text | null | |||
encrypted_value_salt | varchar | null | |||
encrypted_value_iv | varchar | null | |||
group_id | bigint | NOT NULL | |||
protected | boolean | NOT NULL | FALSE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
masked | boolean | NOT NULL | FALSE | ||
variable_type | smallint | NOT NULL | 1 | ||
environment_scope | text | NOT NULL | CAST('*' AS text) | CHECK char_length(environment_scope) <= 255 | |
raw | boolean | NOT NULL | FALSE | ||
description | text | null | CHECK char_length(description) <= 255 | ||
hidden | boolean | NOT NULL | FALSE |
Table checks: char_length(environment_scope) <= 255; char_length(description) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
runner_id PK | bigint | NOT NULL | instance_type_ci_runners.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
ci_instance_runner_monthly_usages references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
runner_id | bigint | null | |||
runner_duration_seconds | bigint | NOT NULL | 0 | ||
project_id | bigint | null | |||
root_namespace_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
billing_month | date | NOT NULL | CHECK billing_month = date_trunc(CAST('month' AS text), CAST(billing_month AS timestamptz)) | ||
notification_level | integer | NOT NULL | 100 | ||
compute_minutes_used | numeric(18, 4) | NOT NULL | 0.0 |
Table checks: billing_month = date_trunc(CAST('month' AS text), CAST(billing_month AS timestamptz))
ci_instance_variables references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
variable_type | smallint | NOT NULL | 1 | ||
masked | boolean | null | FALSE | ||
protected | boolean | null | FALSE | ||
key | text | NOT NULL | CHECK char_length(key) <= 255 | ||
encrypted_value | text | null | |||
encrypted_value_iv | text | null | CHECK char_length(encrypted_value_iv) <= 255 | ||
raw | boolean | NOT NULL | FALSE | ||
description | text | null | CHECK char_length(description) <= 255 | ||
hidden | boolean | NOT NULL | FALSE |
Table checks: char_length(encrypted_value_iv) <= 255; char_length(key) <= 255; (variable_type = 2 AND char_length(encrypted_value) <= 67800) OR char_length(encrypted_value) <= 13579; char_length(description) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
job_artifact_id PK | bigint | NOT NULL | p_ci_job_artifacts.partition_id 1:1 · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | null | |||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 | ||
partition_id PK | bigint | NOT NULL | p_ci_job_artifacts.partition_id 1:1 · CASCADE | ||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: char_length(verification_failure) <= 255; project_id IS NOT NULL
ci_job_token_authorizations references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
accessed_project_id | bigint | NOT NULL | |||
origin_project_id | bigint | NOT NULL | |||
last_authorized_at | timestamptz | NOT NULL | |||
job_token_policies | jsonb | NOT NULL | CAST('{}' AS jsonb) |
ci_job_token_group_scope_links references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
source_project_id | bigint | NOT NULL | |||
target_group_id | bigint | NOT NULL | |||
added_by_id | bigint | null | |||
created_at | timestamptz | NOT NULL | |||
job_token_policies | jsonb | null | CAST('[]' AS jsonb) | ||
default_permissions | boolean | NOT NULL | TRUE | ||
autopopulated | boolean | NOT NULL | FALSE |
ci_job_token_project_scope_links references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
source_project_id | bigint | NOT NULL | |||
target_project_id | bigint | NOT NULL | |||
added_by_id | bigint | null | |||
created_at | timestamptz | NOT NULL | |||
direction | smallint | NOT NULL | 0 | ||
job_token_policies | jsonb | null | CAST('[]' AS jsonb) | ||
default_permissions | boolean | NOT NULL | TRUE | ||
autopopulated | boolean | NOT NULL | FALSE |
partition_id, job_id — PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_builds must scan ci_job_variables to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to ci_job_variables maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_builds rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
key | varchar | NOT NULL | |||
encrypted_value | text | null | |||
encrypted_value_iv | varchar | null | |||
job_id | bigint | NOT NULL | p_ci_builds.partition_id 1:N · CASCADE | ||
variable_type | smallint | NOT NULL | 1 | ||
source | smallint | NOT NULL | 0 | ||
raw | boolean | NOT NULL | FALSE | ||
partition_id | bigint | NOT NULL | p_ci_builds.partition_id 1:N · CASCADE | ||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
ci_minutes_additional_packs references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
namespace_id | bigint | NOT NULL | |||
expires_at | date | null | |||
number_of_minutes | integer | NOT NULL | |||
purchase_xid | text | null | CHECK char_length(purchase_xid) <= 50 |
Table checks: char_length(purchase_xid) <= 50
ci_namespace_mirrors references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | |||
traversal_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) |
ci_namespace_monthly_usages references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | |||
date | date | NOT NULL | CHECK date = date_trunc(CAST('month' AS text), CAST(date AS timestamptz)) | ||
notification_level | smallint | NOT NULL | 100 | ||
created_at | timestamptz | null | |||
amount_used | numeric(18, 4) | NOT NULL | 0.0 | ||
shared_runners_duration | bigint | NOT NULL | 0 | ||
shard_number | integer | NOT NULL | 1 |
Table checks: date = date_trunc(CAST('month' AS text), CAST(date AS timestamptz))
ci_partitions references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
current_from | timestamptz | null | |||
current_until | timestamptz | null | |||
builds_id_range | int8range | null | |||
pipelines_id_range | int8range | null |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
build_id | bigint | NOT NULL | p_ci_builds.partition_id 1:1 · CASCADE | ||
project_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
protected | boolean | NOT NULL | FALSE | ||
instance_runners_enabled | boolean | NOT NULL | FALSE | ||
namespace_id | bigint | null | |||
minutes_exceeded | boolean | NOT NULL | FALSE | ||
tag_ids | bigint[] | null | CAST('{}' AS bigint[]) | ||
namespace_traversal_ids | bigint[] | null | CAST('{}' AS bigint[]) | ||
partition_id | bigint | NOT NULL | p_ci_builds.partition_id 1:1 · CASCADE | ||
plan_id | bigint | null | |||
plan_name_uid | smallint | null | |||
runner_machine_id | bigint | null |
partition_id, pipeline_id — PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_pipelines must scan ci_pipeline_artifacts to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to ci_pipeline_artifacts maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_pipelines rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
pipeline_id | bigint | NOT NULL | p_ci_pipelines.partition_id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | |||
size | integer | NOT NULL | |||
file_store | smallint | NOT NULL | 1 | ||
file_type | smallint | NOT NULL | |||
file_format | smallint | NOT NULL | |||
file | text | null | CHECK char_length(file) <= 255 | ||
expire_at | timestamptz | null | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | null | |||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 | ||
locked | smallint | null | 2 | ||
partition_id | bigint | NOT NULL | p_ci_pipelines.partition_id 1:N · CASCADE |
Table checks: char_length(file) <= 255; file IS NOT NULL; char_length(verification_failure) <= 255
partition_id, pipeline_id — PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_pipelines must scan ci_pipeline_chat_data to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to ci_pipeline_chat_data maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_pipelines rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
chat_name_id | bigint | NOT NULL | |||
response_url | text | NOT NULL | |||
pipeline_id | bigint | NOT NULL | p_ci_pipelines.partition_id 1:N · CASCADE | ||
partition_id | bigint | NOT NULL | p_ci_pipelines.partition_id 1:N · CASCADE | ||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
partition_id, pipeline_id — PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_pipelines must scan ci_pipeline_messages to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to ci_pipeline_messages maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_pipelines rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
severity | smallint | NOT NULL | 0 | ||
content | text | NOT NULL | CHECK char_length(content) <= 10000 | ||
pipeline_id | bigint | NOT NULL | p_ci_pipelines.partition_id 1:N · CASCADE | ||
partition_id | bigint | NOT NULL | p_ci_pipelines.partition_id 1:N · CASCADE | ||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: char_length(content) <= 10000; project_id IS NOT NULL
partition_id, pipeline_id — PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_pipelines must scan ci_pipeline_metadata to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to ci_pipeline_metadata maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_pipelines rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id | bigint | NOT NULL | |||
pipeline_id PK | bigint | NOT NULL | p_ci_pipelines.partition_id 1:N · CASCADE | ||
name | text | null | CHECK char_length(name) <= 255 | ||
auto_cancel_on_new_commit | smallint | NOT NULL | 0 | ||
auto_cancel_on_job_failure | smallint | NOT NULL | 0 | ||
partition_id | bigint | NOT NULL | p_ci_pipelines.partition_id 1:N · CASCADE | ||
security_policy_protected_branch_bypassed | boolean | NOT NULL | FALSE |
Table checks: char_length(name) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
pipeline_schedule_id | bigint | NOT NULL | ci_pipeline_schedules.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
value | jsonb | null |
Table checks: char_length(name) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
key | varchar | NOT NULL | |||
encrypted_value | text | null | |||
encrypted_value_salt | varchar | null | |||
encrypted_value_iv | varchar | null | |||
pipeline_schedule_id | bigint | NOT NULL | ci_pipeline_schedules.id 1:N · CASCADE | ||
created_at | timestamptz | null | |||
updated_at | timestamptz | null | |||
variable_type | smallint | NOT NULL | 1 | ||
raw | boolean | NOT NULL | FALSE | ||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
next_run_at — next_run_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
description | varchar | null | |||
ref | varchar | null | |||
cron | varchar | null | |||
cron_timezone | varchar | null | |||
next_run_at | timestamp | null | |||
project_id | bigint | null | CHECK project_id IS NOT NULL | ||
owner_id | bigint | null | |||
active | boolean | null | TRUE | ||
created_at | timestamp | null | |||
updated_at | timestamp | null |
Table checks: project_id IS NOT NULL
ci_project_metrics references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | |||
first_pipeline_succeeded_at | timestamptz | null | |||
ci_config_generated_by | text | null | CHECK char_length(ci_config_generated_by) <= 255 | ||
ci_config_first_generated_at | timestamptz | null | |||
first_ai_pipeline_results_viewed_at | timestamptz | null |
Table checks: char_length(ci_config_generated_by) <= 255
ci_project_mirrors references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL |
ci_project_monthly_usages references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
date | date | NOT NULL | CHECK date = date_trunc(CAST('month' AS text), CAST(date AS timestamptz)) | ||
created_at | timestamptz | null | |||
amount_used | numeric(18, 4) | NOT NULL | 0.0 | ||
shared_runners_duration | bigint | NOT NULL | 0 |
Table checks: date = date_trunc(CAST('month' AS text), CAST(date AS timestamptz))
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
lock_version | integer | NOT NULL | 0 | ||
status | smallint | NOT NULL | 0 | ||
ref_path | text | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | |||
key | varchar(255) | NOT NULL | |||
process_mode | smallint | NOT NULL | 0 |
partition_id, build_id — PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_builds must scan ci_resources to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to ci_resources maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_builds rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.partition_id, build_id — ci_resources.partition_id.build_id may be NULL, so the relationship to p_ci_builds is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
resource_group_id | bigint | NOT NULL | ci_resource_groups.id 1:N · CASCADE | ||
build_id | bigint | null | p_ci_builds.partition_id 1:N · SET NULL | ||
partition_id | bigint | null | p_ci_builds.partition_id 1:N · SET NULL | ||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
runner_controller_id | bigint | NOT NULL | ci_runner_controllers.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
runner_controller_id | bigint | NOT NULL | ci_runner_controllers.id 1:N · CASCADE | ||
runner_id | bigint | NOT NULL | ci_runners.id 1:N · CASCADE | ||
runner_type PK | smallint | NOT NULL | ci_runners.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
runner_type, runner_id — runner_id points at different tables depending on runner_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.ci_runner_controller_runner_level_scopings_instance_type references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
runner_controller_id | bigint | NOT NULL | |||
runner_id | bigint | NOT NULL | |||
runner_type PK | smallint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
description | text | null | CHECK char_length(description) <= 1024 | ||
token_digest | text | NOT NULL | CHECK char_length(token_digest) <= 255 | ||
runner_controller_id | bigint | NOT NULL | ci_runner_controllers.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
last_used_at | timestamptz | null |
Table checks: char_length(description) <= 1024; char_length(token_digest) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
description | text | null | CHECK char_length(description) <= 1024 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
enabled | boolean | NOT NULL | FALSE | ||
state | smallint | NOT NULL | 0 |
Table checks: char_length(description) <= 1024
runner_type, runner_id — runner_id points at different tables depending on runner_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.ci_runner_machines references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
runner_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
contacted_at | timestamptz | null | |||
creation_state | smallint | NOT NULL | 0 | ||
executor_type | smallint | null | |||
runner_type PK | smallint | NOT NULL | |||
config | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
system_xid | text | NOT NULL | CHECK char_length(system_xid) <= 64 | ||
platform | text | null | CHECK char_length(platform) <= 255 | ||
architecture | text | null | CHECK char_length(architecture) <= 255 | ||
revision | text | null | CHECK char_length(revision) <= 255 | ||
ip_address | text | null | CHECK char_length(ip_address) <= 1024 | ||
version | text | null | CHECK char_length(version) <= 2048 | ||
runtime_features | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
organization_id | bigint | null | |||
labels | jsonb | NOT NULL | CAST('{}' AS jsonb) |
Table checks: char_length(system_xid) <= 64; char_length(revision) <= 255; char_length(version) <= 2048; char_length(ip_address) <= 1024; char_length(platform) <= 255; char_length(architecture) <= 255
runner_id — ci_runner_namespaces.runner_id may be NULL, so the relationship to group_type_ci_runners is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
runner_id | bigint | null | group_type_ci_runners.id 1:N · CASCADE | ||
namespace_id | bigint | null | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL
created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
runner_id | bigint | NOT NULL | project_type_ci_runners.id 1:N · CASCADE | ||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
runner_type, runner_id — runner_id points at different tables depending on runner_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.ci_runner_taggings references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
tag_id | bigint | NOT NULL | |||
runner_id | bigint | NOT NULL | |||
runner_type PK | smallint | NOT NULL | |||
organization_id | bigint | null | |||
tag_name | text | null | CHECK tag_name IS NOT NULL |
Table checks: tag_name IS NOT NULL; char_length(tag_name) <= 1024
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
tag_id | bigint | NOT NULL | |||
runner_id | bigint | NOT NULL | group_type_ci_runners.id 1:N · CASCADE | ||
runner_type PK | smallint | NOT NULL | group_type_ci_runners.id 1:N · CASCADE | ||
organization_id | bigint | null | CHECK organization_id IS NOT NULL | ||
tag_name | text | null | CHECK tag_name IS NOT NULL |
Table checks: organization_id IS NOT NULL; tag_name IS NOT NULL; char_length(tag_name) <= 1024
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
tag_id | bigint | NOT NULL | |||
runner_id | bigint | NOT NULL | instance_type_ci_runners.id 1:N · CASCADE | ||
runner_type PK | smallint | NOT NULL | instance_type_ci_runners.id 1:N · CASCADE | ||
organization_id | bigint | null | CHECK organization_id IS NULL | ||
tag_name | text | null | CHECK tag_name IS NOT NULL |
Table checks: organization_id IS NULL; tag_name IS NOT NULL; char_length(tag_name) <= 1024
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
tag_id | bigint | NOT NULL | |||
runner_id | bigint | NOT NULL | project_type_ci_runners.id 1:N · CASCADE | ||
runner_type PK | smallint | NOT NULL | project_type_ci_runners.id 1:N · CASCADE | ||
organization_id | bigint | null | CHECK organization_id IS NOT NULL | ||
tag_name | text | null | CHECK tag_name IS NOT NULL |
Table checks: organization_id IS NOT NULL; tag_name IS NOT NULL; char_length(tag_name) <= 1024
ci_runner_versions references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
version PK | text | NOT NULL | CHECK char_length(version) <= 2048 | ||
status | smallint | null |
Table checks: char_length(version) <= 2048
public_projects_minutes_cost_factor — public_projects_minutes_cost_factor is binary floating point. 0.1 + 0.2 ≠ 0.3; sums drift; reconciliation against the ledger will be off by cents.private_projects_minutes_cost_factor — private_projects_minutes_cost_factor is binary floating point. 0.1 + 0.2 ≠ 0.3; sums drift; reconciliation against the ledger will be off by cents.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
creator_id | bigint | null | |||
created_at | timestamptz | null | |||
updated_at | timestamptz | null | |||
contacted_at | timestamptz | null | |||
token_expires_at | timestamptz | null | |||
public_projects_minutes_cost_factor | double precision | NOT NULL | 1.0 | ||
private_projects_minutes_cost_factor | double precision | NOT NULL | 1.0 | ||
access_level | integer | NOT NULL | 0 | ||
maximum_timeout | integer | null | |||
runner_type PK | smallint | NOT NULL | |||
registration_type | smallint | NOT NULL | 0 | ||
creation_state | smallint | NOT NULL | 0 | ||
active | boolean | NOT NULL | TRUE | ||
run_untagged | boolean | NOT NULL | TRUE | ||
locked | boolean | NOT NULL | FALSE | ||
name | text | null | CHECK char_length(name) <= 256 | ||
token_encrypted | text | null | CHECK char_length(token_encrypted) <= 512 | ||
description | text | null | CHECK char_length(description) <= 1024 | ||
maintainer_note | text | null | CHECK char_length(maintainer_note) <= 1024 | ||
allowed_plans | text[] | NOT NULL | CAST('{}' AS text[]) | ||
allowed_plan_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
organization_id | bigint | null | |||
allowed_plan_name_uids | smallint[] | NOT NULL | CAST('{}' AS smallint[]) | ||
token_rotation_deadline | timestamptz | null | |||
uuid | uuid | null |
Table checks: char_length(token_encrypted) <= 512; char_length(name) <= 256; char_length(maintainer_note) <= 1024; char_length(description) <= 1024
runner_type, runner_id — runner_id points at different tables depending on runner_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
build_id | bigint | NOT NULL | p_ci_builds.partition_id 1:1 · CASCADE | ||
project_id | bigint | NOT NULL | |||
runner_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
runner_type | smallint | NOT NULL | |||
partition_id | bigint | NOT NULL | p_ci_builds.partition_id 1:1 · CASCADE | ||
runner_owner_namespace_xid | bigint | null |
ci_secure_file_states references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
ci_secure_file_id PK | bigint | NOT NULL | |||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | null | |||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 | ||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL; char_length(verification_failure) <= 255
ci_secure_files references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | smallint | NOT NULL | 1 | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
file | text | NOT NULL | CHECK char_length(file) <= 255 | ||
checksum | bytea | NOT NULL | |||
key_data | text | null | CHECK char_length(key_data) <= 128 | ||
metadata | jsonb | null | |||
expires_at | timestamptz | null |
Table checks: char_length(file) <= 255; char_length(name) <= 255; char_length(key_data) <= 128
source_partition_id, source_job_id — PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_builds must scan ci_sources_pipelines to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to ci_sources_pipelines maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_builds rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.source_partition_id, source_pipeline_id — PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_pipelines must scan ci_sources_pipelines to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to ci_sources_pipelines maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_pipelines rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.partition_id, pipeline_id — PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_pipelines must scan ci_sources_pipelines to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to ci_sources_pipelines maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_pipelines rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.source_partition_id, source_job_id — ci_sources_pipelines.source_partition_id.source_job_id may be NULL, so the relationship to p_ci_builds is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.source_partition_id, source_pipeline_id — ci_sources_pipelines.source_partition_id.source_pipeline_id may be NULL, so the relationship to p_ci_pipelines is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.partition_id, pipeline_id — ci_sources_pipelines.partition_id.pipeline_id may be NULL, so the relationship to p_ci_pipelines is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | null | CHECK project_id IS NOT NULL | ||
source_project_id | bigint | null | |||
source_job_id | bigint | null | p_ci_builds.partition_id 1:N · CASCADE | ||
partition_id | bigint | NOT NULL | p_ci_pipelines.partition_id 1:N · CASCADE | ||
source_partition_id | bigint | NOT NULL | p_ci_pipelines.partition_id 1:N · CASCADE | ||
pipeline_id | bigint | null | p_ci_pipelines.partition_id 1:N · CASCADE | ||
source_pipeline_id | bigint | null | p_ci_pipelines.partition_id 1:N · CASCADE |
Table checks: project_id IS NOT NULL
partition_id, pipeline_id — PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_pipelines must scan ci_sources_projects to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to ci_sources_projects maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_pipelines rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
pipeline_id | bigint | NOT NULL | p_ci_pipelines.partition_id 1:N · CASCADE | ||
source_project_id | bigint | NOT NULL | |||
partition_id | bigint | NOT NULL | p_ci_pipelines.partition_id 1:N · CASCADE |
ci_subscriptions_projects references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
downstream_project_id | bigint | NOT NULL | |||
upstream_project_id | bigint | NOT NULL | |||
author_id | bigint | null |
created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.ci_triggers references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
project_id | bigint | null | CHECK project_id IS NOT NULL | ||
owner_id | bigint | NOT NULL | |||
description | varchar | null | |||
expires_at | timestamptz | null | |||
token_encrypted | text | null | CHECK char_length(token_encrypted) <= 255 |
Table checks: char_length(token_encrypted) <= 255; project_id IS NOT NULL
partition_id, build_id — PostgreSQL does not index FK columns automatically. Every DELETE/UPDATE on p_ci_builds must scan ci_unit_test_failures to check the constraint, and every join from the parent side is a sequential scan. The scan cost grows with the child table forever, so this gets worse on its own.
An index is not free. Every write to ci_unit_test_failures maintains it — insert, update and delete alike — and it takes disk. Weigh that against how often p_ci_builds rows are actually deleted or re-keyed — if the answer is never, the scan never happens and the index only costs.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
failed_at | timestamptz | NOT NULL | |||
unit_test_id | bigint | NOT NULL | ci_unit_tests.id 1:N · CASCADE | ||
build_id | bigint | NOT NULL | p_ci_builds.partition_id 1:N · CASCADE | ||
partition_id | bigint | NOT NULL | p_ci_builds.partition_id 1:N · CASCADE | ||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
key_hash | text | NOT NULL | CHECK char_length(key_hash) <= 64 | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
suite_name | text | NOT NULL | CHECK char_length(suite_name) <= 255 |
Table checks: char_length(name) <= 255; char_length(key_hash) <= 64; char_length(suite_name) <= 255
environment_scope — environment_scope is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.ci_variables references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
key | varchar | NOT NULL | |||
encrypted_value | text | null | |||
encrypted_value_salt | varchar | null | |||
encrypted_value_iv | varchar | null | |||
project_id | bigint | NOT NULL | |||
protected | boolean | NOT NULL | FALSE | ||
environment_scope | varchar | NOT NULL | CAST('*' AS varchar) | ||
masked | boolean | NOT NULL | FALSE | ||
variable_type | smallint | NOT NULL | 1 | ||
raw | boolean | NOT NULL | FALSE | ||
description | text | null | CHECK char_length(description) <= 255 | ||
hidden | boolean | NOT NULL | FALSE |
Table checks: char_length(description) <= 255
Tables whose names start with cloud_.
No foreign keys in this domain.
cloud_connector_access references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
data | jsonb | null | |||
catalog | jsonb | null |
cloud_connector_keys references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
secret_key | jsonb | null |
Tables whose names start with cluster_.
cluster_agent_migrations.agent_id → cluster_agents.id — one cluster_agents, many cluster_agent_migrations · required · ON DELETE CASCADE · indexedcluster_agent_migrations.cluster_id → clusters.id — one clusters, at most one cluster_agent_migrations · required · ON DELETE CASCADE · indexedcluster_agent_migrations.issue_id → issues.id — one issues, many cluster_agent_migrations · optional · ON DELETE SET NULL · indexedcluster_agent_migrations.project_id → projects.id — one projects, many cluster_agent_migrations · required · ON DELETE CASCADE · indexedcluster_agent_tokens.project_id → projects.id — one projects, many cluster_agent_tokens · optional · ON DELETE CASCADE · indexedcluster_agent_tokens.created_by_user_id → users.id — one users, many cluster_agent_tokens · optional · ON DELETE SET NULL · indexedcluster_agent_tokens.agent_id → cluster_agents.id — one cluster_agents, many cluster_agent_tokens · required · ON DELETE CASCADE · indexedcluster_agent_url_configurations.agent_id → cluster_agents.id — one cluster_agents, many cluster_agent_url_configurations · required · ON DELETE CASCADE · indexedcluster_agent_url_configurations.created_by_user_id → users.id — one users, many cluster_agent_url_configurations · optional · ON DELETE SET NULL · indexedcluster_agent_url_configurations.project_id → projects.id — one projects, many cluster_agent_url_configurations · required · ON DELETE CASCADE · indexedcluster_agents.created_by_user_id → users.id — one users, many cluster_agents · optional · ON DELETE SET NULL · indexedcluster_agents.project_id → projects.id — one projects, many cluster_agents · required · ON DELETE CASCADE · indexedcluster_enabled_grants.namespace_id → namespaces.id — one namespaces, at most one cluster_enabled_grants · required · ON DELETE CASCADE · indexedcluster_groups.group_id → namespaces.id — one namespaces, many cluster_groups · required · ON DELETE CASCADE · indexedcluster_groups.cluster_id → clusters.id — one clusters, many cluster_groups · required · ON DELETE CASCADE · indexedcluster_platforms_kubernetes.organization_id → organizations.id — one organizations, many cluster_platforms_kubernetes · optional · ON DELETE CASCADE · indexedcluster_platforms_kubernetes.group_id → namespaces.id — one namespaces, many cluster_platforms_kubernetes · optional · ON DELETE CASCADE · indexedcluster_platforms_kubernetes.project_id → projects.id — one projects, many cluster_platforms_kubernetes · optional · ON DELETE CASCADE · indexedcluster_platforms_kubernetes.cluster_id → clusters.id — one clusters, at most one cluster_platforms_kubernetes · required · ON DELETE CASCADE · indexedcluster_projects.project_id → projects.id — one projects, many cluster_projects · required · ON DELETE CASCADE · indexedcluster_projects.cluster_id → clusters.id — one clusters, many cluster_projects · required · ON DELETE CASCADE · indexedcluster_providers_aws.project_id → projects.id — one projects, many cluster_providers_aws · optional · ON DELETE CASCADE · indexedcluster_providers_aws.group_id → namespaces.id — one namespaces, many cluster_providers_aws · optional · ON DELETE CASCADE · indexedcluster_providers_aws.organization_id → organizations.id — one organizations, many cluster_providers_aws · optional · ON DELETE CASCADE · indexedcluster_providers_aws.cluster_id → clusters.id — one clusters, at most one cluster_providers_aws · required · ON DELETE CASCADE · indexedcluster_providers_gcp.group_id → namespaces.id — one namespaces, many cluster_providers_gcp · optional · ON DELETE CASCADE · indexedcluster_providers_gcp.project_id → projects.id — one projects, many cluster_providers_gcp · optional · ON DELETE CASCADE · indexedcluster_providers_gcp.organization_id → organizations.id — one organizations, many cluster_providers_gcp · optional · ON DELETE CASCADE · indexedcluster_providers_gcp.cluster_id → clusters.id — one clusters, at most one cluster_providers_gcp · required · ON DELETE CASCADE · indexedissue_id — cluster_agent_migrations.issue_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
cluster_id | bigint | NOT NULL | clusters.id 1:1 · CASCADE | ||
agent_id | bigint | NOT NULL | cluster_agents.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
issue_id | bigint | null | issues.id 1:N · SET NULL | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
agent_install_status | smallint | NOT NULL | |||
agent_install_message | text | null | CHECK char_length(agent_install_message) <= 255 |
Table checks: char_length(agent_install_message) <= 255
project_id — cluster_agent_tokens.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.created_by_user_id — cluster_agent_tokens.created_by_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id, created_by_user_id — cluster_agent_tokens has several nullable FKs (project_id, created_by_user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
agent_id | bigint | NOT NULL | cluster_agents.id 1:N · CASCADE | ||
token_encrypted | text | NOT NULL | CHECK char_length(token_encrypted) <= 255 | ||
created_by_user_id | bigint | null | users.id 1:N · SET NULL | ||
description | text | null | CHECK char_length(description) <= 1024 | ||
name | text | null | CHECK name IS NOT NULL | ||
last_used_at | timestamptz | null | |||
status | smallint | NOT NULL | 0 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: name IS NOT NULL; char_length(name) <= 255; char_length(description) <= 1024; project_id IS NOT NULL; char_length(token_encrypted) <= 255
created_by_user_id — cluster_agent_url_configurations.created_by_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
agent_id | bigint | NOT NULL | cluster_agents.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_by_user_id | bigint | null | users.id 1:N · SET NULL | ||
status | smallint | NOT NULL | 0 | ||
url | text | NOT NULL | CHECK char_length(url) <= 2048 | ||
ca_cert | text | null | CHECK char_length(ca_cert) <= 16384 | ||
client_key | text | null | CHECK char_length(client_key) <= 16384 | ||
client_cert | text | null | CHECK char_length(client_cert) <= 16384 | ||
tls_host | text | null | CHECK char_length(tls_host) <= 2048 | ||
public_key | bytea | null | |||
encrypted_private_key | bytea | null | |||
encrypted_private_key_iv | bytea | null |
Table checks: char_length(tls_host) <= 2048; char_length(ca_cert) <= 16384; char_length(client_cert) <= 16384; char_length(client_key) <= 16384; char_length(url) <= 2048
created_by_user_id — cluster_agents.created_by_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
created_by_user_id | bigint | null | users.id 1:N · SET NULL | ||
has_vulnerabilities | boolean | NOT NULL | FALSE | ||
connection_mode | smallint | NOT NULL | 0 | ||
is_receptive | boolean | NOT NULL | FALSE |
Table checks: char_length(name) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
cluster_id | bigint | NOT NULL | clusters.id 1:N · CASCADE | ||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE |
organization_id — cluster_platforms_kubernetes.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — cluster_platforms_kubernetes.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — cluster_platforms_kubernetes.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
cluster_id | bigint | NOT NULL | clusters.id 1:1 · CASCADE | ||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL | |||
api_url | text | null | |||
ca_cert | text | null | |||
namespace | varchar | null | |||
username | varchar | null | |||
encrypted_password | text | null | |||
encrypted_password_iv | varchar | null | |||
encrypted_token | text | null | |||
encrypted_token_iv | varchar | null | |||
authorization_type | smallint | null | |||
organization_id | bigint | null | organizations.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE |
Table checks: num_nonnulls(group_id, organization_id, project_id) = 1
cluster_id, project_id — cluster_projects looks like a many-to-many link table but has no unique constraint across (cluster_id, project_id). The same pair can be inserted twice; every join through it will double-count.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
cluster_id | bigint | NOT NULL | clusters.id 1:N · CASCADE | ||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL |
project_id — cluster_providers_aws.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — cluster_providers_aws.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id — cluster_providers_aws.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
cluster_id | bigint | NOT NULL | clusters.id 1:1 · CASCADE | ||
num_nodes | integer | NOT NULL | |||
status | integer | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
key_name | varchar(255) | NOT NULL | |||
role_arn | varchar(2048) | NOT NULL | |||
region | varchar(255) | NOT NULL | |||
vpc_id | varchar(255) | NOT NULL | |||
subnet_ids | varchar(255)[] | NOT NULL | CAST('{}' AS varchar[]) | ||
security_group_id | varchar(255) | NOT NULL | |||
instance_type | varchar(255) | NOT NULL | |||
access_key_id | varchar(255) | null | |||
encrypted_secret_access_key_iv | varchar(255) | null | |||
encrypted_secret_access_key | text | null | |||
session_token | text | null | |||
status_reason | text | null | |||
kubernetes_version | text | NOT NULL | CAST('1.14' AS text) | CHECK char_length(kubernetes_version) <= 30 | |
organization_id | bigint | null | organizations.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE |
Table checks: num_nonnulls(group_id, organization_id, project_id) = 1; char_length(kubernetes_version) <= 30
group_id — cluster_providers_gcp.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — cluster_providers_gcp.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id — cluster_providers_gcp.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.machine_type — machine_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
cluster_id | bigint | NOT NULL | clusters.id 1:1 · CASCADE | ||
status | integer | null | |||
num_nodes | integer | NOT NULL | |||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL | |||
status_reason | text | null | |||
gcp_project_id | varchar | NOT NULL | |||
zone | varchar | NOT NULL | |||
machine_type | varchar | null | |||
operation_id | varchar | null | |||
endpoint | varchar | null | |||
encrypted_access_token | text | null | |||
encrypted_access_token_iv | varchar | null | |||
legacy_abac | boolean | NOT NULL | FALSE | ||
cloud_run | boolean | NOT NULL | FALSE | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE |
Table checks: num_nonnulls(group_id, organization_id, project_id) = 1
Tables whose names start with clusters_.
clusters.organization_id → organizations.id — one organizations, many clusters · optional · ON DELETE CASCADE · indexedclusters.group_id → namespaces.id — one namespaces, many clusters · optional · ON DELETE CASCADE · indexedclusters.management_project_id → projects.id — one projects, many clusters · optional · ON DELETE SET NULL · indexedclusters.project_id → projects.id — one projects, many clusters · optional · ON DELETE CASCADE · indexedclusters.user_id → users.id — one users, many clusters · optional · ON DELETE SET NULL · indexedclusters_kubernetes_namespaces.organization_id → organizations.id — one organizations, many clusters_kubernetes_namespaces · optional · ON DELETE CASCADE · indexedclusters_kubernetes_namespaces.group_id → namespaces.id — one namespaces, many clusters_kubernetes_namespaces · optional · ON DELETE CASCADE · indexedclusters_kubernetes_namespaces.sharding_project_id → projects.id — one projects, many clusters_kubernetes_namespaces · optional · ON DELETE CASCADE · indexedclusters_kubernetes_namespaces.cluster_project_id → cluster_projects.id — one cluster_projects, many clusters_kubernetes_namespaces · optional · ON DELETE SET NULL · indexedclusters_kubernetes_namespaces.cluster_id → clusters.id — one clusters, many clusters_kubernetes_namespaces · required · ON DELETE CASCADE · indexedclusters_kubernetes_namespaces.environment_id → environments.id — one environments, many clusters_kubernetes_namespaces · optional · ON DELETE SET NULL · indexedclusters_kubernetes_namespaces.project_id → projects.id — one projects, many clusters_kubernetes_namespaces · optional · ON DELETE SET NULL · indexedclusters_managed_resources.cluster_agent_id → cluster_agents.id — one cluster_agents, many clusters_managed_resources · required · ON DELETE CASCADE · indexedclusters_managed_resources.project_id → projects.id — one projects, many clusters_managed_resources · required · ON DELETE CASCADE · indexedclusters_managed_resources.environment_id → environments.id — one environments, many clusters_managed_resources · required · ON DELETE CASCADE · indexedorganization_id — clusters.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — clusters.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.management_project_id — clusters.management_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — clusters.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id — clusters.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.environment_scope — environment_scope is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.organization_id, group_id, management_project_id, project_id, user_id — clusters has several nullable FKs (organization_id, group_id, management_project_id, project_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | null | users.id 1:N · SET NULL | ||
provider_type | integer | null | |||
platform_type | integer | null | |||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL | |||
enabled | boolean | null | TRUE | ||
name | varchar | NOT NULL | |||
environment_scope | varchar | NOT NULL | CAST('*' AS varchar) | ||
cluster_type | smallint | NOT NULL | 3 | ||
domain | varchar | null | |||
managed | boolean | NOT NULL | TRUE | ||
namespace_per_environment | boolean | NOT NULL | TRUE | ||
management_project_id | bigint | null | projects.id 1:N · SET NULL | ||
cleanup_status | smallint | NOT NULL | 1 | ||
cleanup_status_reason | text | null | |||
helm_major_version | integer | NOT NULL | 3 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE |
Table checks: num_nonnulls(group_id, organization_id, project_id) = 1
organization_id — clusters_kubernetes_namespaces.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — clusters_kubernetes_namespaces.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.sharding_project_id — clusters_kubernetes_namespaces.sharding_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.cluster_project_id — clusters_kubernetes_namespaces.cluster_project_id may be NULL, so the relationship to cluster_projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.environment_id — clusters_kubernetes_namespaces.environment_id may be NULL, so the relationship to environments is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — clusters_kubernetes_namespaces.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id, group_id, sharding_project_id, cluster_project_id, environment_id, project_id — clusters_kubernetes_namespaces has several nullable FKs (organization_id, group_id, sharding_project_id, cluster_project_id, environment_id, project_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
cluster_id | bigint | NOT NULL | clusters.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · SET NULL | ||
cluster_project_id | bigint | null | cluster_projects.id 1:N · SET NULL | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
encrypted_service_account_token | text | null | |||
encrypted_service_account_token_iv | varchar | null | |||
namespace | varchar | NOT NULL | |||
service_account_name | varchar | null | |||
environment_id | bigint | null | environments.id 1:N · SET NULL | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
sharding_project_id | bigint | null | projects.id 1:N · CASCADE |
Table checks: num_nonnulls(group_id, organization_id, sharding_project_id) = 1
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
build_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
environment_id | bigint | NOT NULL | environments.id 1:N · CASCADE | ||
cluster_agent_id | bigint | NOT NULL | cluster_agents.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
template_name | text | null | CHECK char_length(template_name) <= 1024 | ||
tracked_objects | jsonb | NOT NULL | CAST('[]' AS jsonb) | ||
deletion_strategy | smallint | NOT NULL | 0 |
Table checks: char_length(template_name) <= 1024
Tables whose names start with compliance_.
compliance_framework_security_policies.security_policy_id → security_policies.id — one security_policies, many compliance_framework_security_policies · optional · ON DELETE CASCADE · indexedcompliance_framework_security_policies.namespace_id → namespaces.id — one namespaces, many compliance_framework_security_policies · optional · ON DELETE CASCADE · indexedcompliance_framework_security_policies.project_id → projects.id — one projects, many compliance_framework_security_policies · optional · ON DELETE CASCADE · indexedcompliance_framework_security_policies.framework_id → compliance_management_frameworks.id — one compliance_management_frameworks, many compliance_framework_security_policies · required · ON DELETE CASCADE · indexedcompliance_framework_security_policies.policy_configuration_id → security_orchestration_policy_configurations.id — one security_orchestration_policy_configurations, many compliance_framework_security_policies · required · ON DELETE CASCADE · indexedcompliance_management_frameworks.namespace_id → namespaces.id — one namespaces, many compliance_management_frameworks · required · ON DELETE CASCADE · indexedcompliance_requirements.namespace_id → namespaces.id — one namespaces, many compliance_requirements · required · ON DELETE CASCADE · indexedcompliance_requirements.framework_id → compliance_management_frameworks.id — one compliance_management_frameworks, many compliance_requirements · required · ON DELETE CASCADE · indexedcompliance_requirements_controls.namespace_id → namespaces.id — one namespaces, many compliance_requirements_controls · required · ON DELETE CASCADE · indexedcompliance_requirements_controls.compliance_requirement_id → compliance_requirements.id — one compliance_requirements, many compliance_requirements_controls · required · ON DELETE CASCADE · indexedsecurity_policy_id — compliance_framework_security_policies.security_policy_id may be NULL, so the relationship to security_policies is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id — compliance_framework_security_policies.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — compliance_framework_security_policies.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.security_policy_id, namespace_id, project_id — compliance_framework_security_policies has several nullable FKs (security_policy_id, namespace_id, project_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
framework_id | bigint | NOT NULL | compliance_management_frameworks.id 1:N · CASCADE | ||
policy_configuration_id | bigint | NOT NULL | security_orchestration_policy_configurations.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
policy_index | smallint | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
security_policy_id | bigint | null | security_policies.id 1:N · CASCADE |
Table checks: num_nonnulls(namespace_id, project_id) = 1
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
description | text | NOT NULL | CHECK char_length(description) <= 255 | ||
color | text | NOT NULL | CHECK char_length(color) <= 10 | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
pipeline_configuration_full_path | text | null | CHECK char_length(pipeline_configuration_full_path) <= 255 | ||
created_at | timestamptz | null | |||
updated_at | timestamptz | null | |||
source_id | bigint | null | |||
template_id | text | null | CHECK char_length(template_id) <= 255 | ||
template_version | integer | null |
Table checks: char_length(color) <= 10; char_length(description) <= 255; char_length(name) <= 255; char_length(pipeline_configuration_full_path) <= 255; char_length(template_id) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
framework_id | bigint | NOT NULL | compliance_management_frameworks.id 1:N · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
description | text | NOT NULL | CHECK char_length(description) <= 500 |
Table checks: char_length(description) <= 500; char_length(name) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
compliance_requirement_id | bigint | NOT NULL | compliance_requirements.id 1:N · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
name | smallint | NOT NULL | |||
control_type | smallint | NOT NULL | |||
expression | text | null | CHECK char_length(expression) <= 255 | ||
encrypted_secret_token | bytea | null | CHECK octet_length(encrypted_secret_token) <= 256 | ||
encrypted_secret_token_iv | bytea | null | |||
external_url | text | null | CHECK char_length(external_url) <= 1024 | ||
external_control_name | text | null | CHECK char_length(external_control_name) <= 255 | ||
ping_enabled | boolean | NOT NULL | TRUE |
Table checks: char_length(expression) <= 255; char_length(external_url) <= 1024; octet_length(encrypted_secret_token) <= 256; char_length(external_control_name) <= 255
Tables whose names start with container_.
container_expiration_policies.project_id → projects.id — one projects, at most one container_expiration_policies · required · ON DELETE CASCADE · indexedcontainer_registry_protection_rules.project_id → projects.id — one projects, many container_registry_protection_rules · required · ON DELETE CASCADE · indexedcontainer_registry_protection_tag_rules.project_id → projects.id — one projects, many container_registry_protection_tag_rules · required · ON DELETE CASCADE · indexedcontainer_repositories.project_id → projects.id — one projects, many container_repositories · required · ON DELETE NO ACTION · indexedcontainer_repository_states.project_id → projects.id — one projects, many container_repository_states · optional · ON DELETE CASCADE · indexedcontainer_repository_states.container_repository_id → container_repositories.id — one container_repositories, at most one container_repository_states · required · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | projects.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
next_run_at | timestamptz | null | |||
name_regex | varchar(255) | null | CAST('.*' AS varchar) | ||
cadence | varchar(12) | NOT NULL | CAST('1d' AS varchar) | ||
older_than | varchar(12) | null | CAST('90d' AS varchar) | ||
keep_n | integer | null | 10 | ||
enabled | boolean | NOT NULL | FALSE | ||
name_regex_keep | text | null | CHECK char_length(name_regex_keep) <= 255 |
Table checks: char_length(name_regex_keep) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
repository_path_pattern | text | null | CHECK repository_path_pattern IS NOT NULL | ||
minimum_access_level_for_push | smallint | null | |||
minimum_access_level_for_delete | smallint | null |
Table checks: num_nonnulls(minimum_access_level_for_delete, minimum_access_level_for_push) > 0; repository_path_pattern IS NOT NULL; char_length(repository_path_pattern) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
minimum_access_level_for_push | smallint | null | |||
minimum_access_level_for_delete | smallint | null | |||
tag_name_pattern | text | NOT NULL | CHECK char_length(tag_name_pattern) <= 255 |
Table checks: char_length(tag_name_pattern) <= 255; num_nonnulls(minimum_access_level_for_delete, minimum_access_level_for_push) <> 1
created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N | ||
name | varchar | NOT NULL | |||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL | |||
status | smallint | null | |||
expiration_policy_started_at | timestamptz | null | |||
expiration_policy_cleanup_status | smallint | NOT NULL | 0 | ||
expiration_policy_completed_at | timestamptz | null | |||
last_cleanup_deleted_tags_count | integer | null | |||
delete_started_at | timestamptz | null | |||
status_updated_at | timestamptz | null | |||
failed_deletion_count | integer | NOT NULL | 0 | CHECK failed_deletion_count >= 0 | |
next_delete_attempt_at | timestamptz | null |
Table checks: failed_deletion_count >= 0
project_id — container_repository_states.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
container_repository_id PK | bigint | NOT NULL | container_repositories.id 1:1 · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: char_length(verification_failure) <= 255; project_id IS NOT NULL
Tables whose names start with custom_.
custom_dashboard_search_data.custom_dashboard_id → custom_dashboards.id — one custom_dashboards, at most one custom_dashboard_search_data · required · ON DELETE CASCADE · indexedcustom_dashboard_search_data.organization_id → organizations.id — one organizations, many custom_dashboard_search_data · required · ON DELETE CASCADE · indexedcustom_dashboard_versions.organization_id → organizations.id — one organizations, many custom_dashboard_versions · required · ON DELETE CASCADE · indexedcustom_dashboard_versions.updated_by_id → users.id — one users, many custom_dashboard_versions · optional · ON DELETE SET NULL · indexedcustom_dashboard_versions.custom_dashboard_id → custom_dashboards.id — one custom_dashboards, many custom_dashboard_versions · required · ON DELETE NO ACTION · indexedcustom_dashboards.created_by_id → users.id — one users, many custom_dashboards · optional · ON DELETE SET NULL · indexedcustom_dashboards.namespace_id → namespaces.id — one namespaces, many custom_dashboards · optional · ON DELETE SET NULL · indexedcustom_dashboards.organization_id → organizations.id — one organizations, many custom_dashboards · required · ON DELETE CASCADE · indexedcustom_dashboards.updated_by_id → users.id — one users, many custom_dashboards · optional · ON DELETE SET NULL · indexedcustom_emoji.creator_id → users.id — one users, many custom_emoji · required · ON DELETE CASCADE · indexedcustom_emoji.namespace_id → namespaces.id — one namespaces, many custom_emoji · required · ON DELETE CASCADE · indexedcustom_field_select_options.custom_field_id → custom_fields.id — one custom_fields, many custom_field_select_options · required · ON DELETE CASCADE · indexedcustom_field_select_options.namespace_id → namespaces.id — one namespaces, many custom_field_select_options · required · ON DELETE CASCADE · indexedcustom_fields.namespace_id → namespaces.id — one namespaces, many custom_fields · required · ON DELETE CASCADE · indexedcustom_fields.created_by_id → users.id — one users, many custom_fields · optional · ON DELETE SET NULL · indexedcustom_fields.updated_by_id → users.id — one users, many custom_fields · optional · ON DELETE SET NULL · indexedcustom_software_licenses.project_id → projects.id — one projects, many custom_software_licenses · required · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
custom_dashboard_id | bigint | NOT NULL | custom_dashboards.id 1:1 · CASCADE | ||
name | text | NOT NULL | CAST('' AS text) | CHECK char_length(name) <= 255 | |
description | text | NOT NULL | CAST('' AS text) | CHECK char_length(description) <= 2048 | |
search_vector | tsvector | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE |
Table checks: char_length(name) <= 255; char_length(description) <= 2048
updated_by_id — custom_dashboard_versions.updated_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
custom_dashboard_id | bigint | NOT NULL | custom_dashboards.id 1:N | ||
version_number | integer | NOT NULL | |||
config | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(config) = CAST('object' AS text) | |
updated_by_id | bigint | null | users.id 1:N · SET NULL | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE |
Table checks: jsonb_typeof(config) = CAST('object' AS text)
created_by_id — custom_dashboards.created_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id — custom_dashboards.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.updated_by_id — custom_dashboards.updated_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.created_by_id, namespace_id, updated_by_id — custom_dashboards has several nullable FKs (created_by_id, namespace_id, updated_by_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | null | namespaces.id 1:N · SET NULL | ||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
created_by_id | bigint | null | users.id 1:N · SET NULL | ||
updated_by_id | bigint | null | users.id 1:N · SET NULL | ||
lock_version | integer | NOT NULL | 0 | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
description | text | NOT NULL | CAST('' AS text) | CHECK char_length(description) <= 2048 | |
config | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(config) = CAST('object' AS text) | |
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
Table checks: char_length(name) <= 255; char_length(description) <= 2048; jsonb_typeof(config) = CAST('object' AS text)
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 36 | ||
file | text | NOT NULL | CHECK char_length(file) <= 255 | ||
external | boolean | NOT NULL | TRUE | ||
creator_id | bigint | NOT NULL | users.id 1:N · CASCADE |
Table checks: char_length(name) <= 36; char_length(file) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
custom_field_id | bigint | NOT NULL | custom_fields.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
position | smallint | NOT NULL | |||
value | text | NOT NULL | CHECK char_length(value) <= 255 |
Table checks: char_length(value) <= 255
created_by_id — custom_fields.created_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.updated_by_id — custom_fields.updated_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.created_by_id, updated_by_id — custom_fields has several nullable FKs (created_by_id, updated_by_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
archived_at | timestamptz | null | |||
field_type | smallint | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
created_by_id | bigint | null | users.id 1:N · SET NULL | ||
updated_by_id | bigint | null | users.id 1:N · SET NULL |
Table checks: char_length(name) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 |
Table checks: char_length(name) <= 255
Tables whose names start with customer_.
customer_relations_contacts.group_id → namespaces.id — one namespaces, many customer_relations_contacts · required · ON DELETE CASCADE · indexedcustomer_relations_contacts.organization_id → customer_relations_organizations.id — one customer_relations_organizations, many customer_relations_contacts · optional · ON DELETE CASCADE · indexedcustomer_relations_organizations.group_id → namespaces.id — one namespaces, many customer_relations_organizations · required · ON DELETE CASCADE · indexedorganization_id — customer_relations_contacts.organization_id may be NULL, so the relationship to customer_relations_organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
organization_id | bigint | null | customer_relations_organizations.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
state | smallint | NOT NULL | 1 | ||
phone | text | null | CHECK char_length(phone) <= 32 | ||
first_name | text | NOT NULL | CHECK char_length(first_name) <= 255 | ||
last_name | text | NOT NULL | CHECK char_length(last_name) <= 255 | ||
email | text | null | CHECK char_length(email) <= 255 | ||
description | text | null | CHECK char_length(description) <= 1024 |
Table checks: char_length(first_name) <= 255; char_length(description) <= 1024; char_length(last_name) <= 255; char_length(phone) <= 32; char_length(email) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
state | smallint | NOT NULL | 1 | ||
default_rate | numeric(18, 2) | null | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
description | text | null | CHECK char_length(description) <= 1024 |
Table checks: char_length(name) <= 255; char_length(description) <= 1024
Tables whose names start with dast_.
dast_pre_scan_verification_steps.dast_pre_scan_verification_id → dast_pre_scan_verifications.id — one dast_pre_scan_verifications, many dast_pre_scan_verification_steps · required · ON DELETE CASCADE · indexeddast_pre_scan_verifications.dast_profile_id → dast_profiles.id — one dast_profiles, many dast_pre_scan_verifications · required · ON DELETE CASCADE · indexeddast_profile_schedules.dast_profile_id → dast_profiles.id — one dast_profiles, at most one dast_profile_schedules · required · ON DELETE CASCADE · indexeddast_profiles.dast_scanner_profile_id → dast_scanner_profiles.id — one dast_scanner_profiles, many dast_profiles · required · ON DELETE CASCADE · indexeddast_profiles.dast_site_profile_id → dast_site_profiles.id — one dast_site_profiles, many dast_profiles · required · ON DELETE CASCADE · indexeddast_profiles_pipelines.dast_profile_id → dast_profiles.id — one dast_profiles, many dast_profiles_pipelines · required · ON DELETE CASCADE · indexeddast_profiles_tags.dast_profile_id → dast_profiles.id — one dast_profiles, many dast_profiles_tags · required · ON DELETE CASCADE · indexeddast_scanner_profiles_builds.dast_scanner_profile_id → dast_scanner_profiles.id — one dast_scanner_profiles, many dast_scanner_profiles_builds · required · ON DELETE CASCADE · indexeddast_site_profile_secret_variables.dast_site_profile_id → dast_site_profiles.id — one dast_site_profiles, many dast_site_profile_secret_variables · required · ON DELETE CASCADE · indexeddast_site_profiles.dast_site_id → dast_sites.id — one dast_sites, many dast_site_profiles · required · ON DELETE CASCADE · indexeddast_site_profiles_builds.dast_site_profile_id → dast_site_profiles.id — one dast_site_profiles, many dast_site_profiles_builds · required · ON DELETE CASCADE · indexeddast_site_validations.dast_site_token_id → dast_site_tokens.id — one dast_site_tokens, many dast_site_validations · required · ON DELETE CASCADE · indexeddast_sites.dast_site_validation_id → dast_site_validations.id — one dast_site_validations, many dast_sites · optional · ON DELETE SET NULL · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
dast_pre_scan_verification_id | bigint | NOT NULL | dast_pre_scan_verifications.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
name | text | null | CHECK char_length(name) <= 255 | ||
verification_errors | text[] | NOT NULL | CAST('{}' AS text[]) | ||
check_type | smallint | NOT NULL | 0 | ||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL; char_length(name) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
dast_profile_id | bigint | NOT NULL | dast_profiles.id 1:N · CASCADE | ||
ci_pipeline_id | bigint | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
{"owner":"group::dynamic analysis","description":"Scheduling for scans using DAST Profiles"}
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
dast_profile_id | bigint | NOT NULL | dast_profiles.id 1:1 · CASCADE | ||
user_id | bigint | null | |||
next_run_at | timestamptz | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
active | boolean | NOT NULL | TRUE | ||
cron | text | NOT NULL | CHECK char_length(cron) <= 255 | ||
cadence | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
timezone | text | NOT NULL | CHECK char_length(timezone) <= 255 | ||
starts_at | timestamptz | NOT NULL | now() |
Table checks: char_length(cron) <= 255; char_length(timezone) <= 255
{"owner":"group::dynamic analysis","description":"Profile used to run a DAST on-demand scan"}
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
dast_site_profile_id | bigint | NOT NULL | dast_site_profiles.id 1:N · CASCADE | ||
dast_scanner_profile_id | bigint | NOT NULL | dast_scanner_profiles.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
description | text | NOT NULL | CHECK char_length(description) <= 255 | ||
branch_name | text | null | CHECK char_length(branch_name) <= 255 |
Table checks: char_length(name) <= 255; char_length(branch_name) <= 255; char_length(description) <= 255
{"owner":"group::dynamic analysis","description":"Join table between DAST Profiles and CI Pipelines"}
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
dast_profile_id PK | bigint | NOT NULL | dast_profiles.id 1:N · CASCADE | ||
ci_pipeline_id PK | bigint | NOT NULL | |||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
dast_profile_id | bigint | NOT NULL | dast_profiles.id 1:N · CASCADE | ||
tag_id | bigint | NOT NULL | |||
project_id | bigint | null | CHECK project_id IS NOT NULL | ||
tag_name | text | null | CHECK tag_name IS NOT NULL |
Table checks: project_id IS NOT NULL; tag_name IS NOT NULL; char_length(tag_name) <= 1024
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | |||
spider_timeout | smallint | null | |||
target_timeout | smallint | null | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
scan_type | smallint | NOT NULL | 1 | ||
use_ajax_spider | boolean | NOT NULL | FALSE | ||
show_debug_messages | boolean | NOT NULL | FALSE |
Table checks: char_length(name) <= 255
{"owner":"group::dynamic analysis","description":"Join table between DAST Scanner Profiles and CI Builds"}
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
dast_scanner_profile_id PK | bigint | NOT NULL | dast_scanner_profiles.id 1:N · CASCADE | ||
ci_build_id PK | bigint | NOT NULL | |||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
{"owner":"group::dynamic analysis","description":"Secret variables used in DAST on-demand scans"}
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
dast_site_profile_id | bigint | NOT NULL | dast_site_profiles.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
variable_type | smallint | NOT NULL | 1 | ||
key | text | NOT NULL | CHECK char_length(key) <= 255 | ||
encrypted_value | bytea | NOT NULL | CHECK length(encrypted_value) <= 13352 | ||
encrypted_value_iv | bytea | NOT NULL | CHECK length(encrypted_value_iv) <= 17 | ||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: length(encrypted_value) <= 13352; char_length(key) <= 255; length(encrypted_value_iv) <= 17; project_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
dast_site_id | bigint | NOT NULL | dast_sites.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
excluded_urls | text[] | NOT NULL | CAST('{}' AS text[]) | ||
auth_enabled | boolean | NOT NULL | FALSE | ||
auth_url | text | null | CHECK char_length(auth_url) <= 1024 | ||
auth_username_field | text | null | CHECK char_length(auth_username_field) <= 255 | ||
auth_password_field | text | null | CHECK char_length(auth_password_field) <= 255 | ||
auth_username | text | null | CHECK char_length(auth_username) <= 255 | ||
target_type | smallint | NOT NULL | 0 | ||
scan_method | smallint | NOT NULL | 0 | ||
auth_submit_field | text | null | CHECK char_length(auth_submit_field) <= 255 | ||
scan_file_path | text | null | CHECK char_length(scan_file_path) <= 1024 | ||
optional_variables | jsonb | NOT NULL | CAST('[]' AS jsonb) |
Table checks: char_length(auth_username_field) <= 255; char_length(name) <= 255; char_length(scan_file_path) <= 1024; char_length(auth_submit_field) <= 255; char_length(auth_password_field) <= 255; char_length(auth_url) <= 1024; char_length(auth_username) <= 255
{"owner":"group::dynamic analysis","description":"Join table between DAST Site Profiles and CI Builds"}
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
dast_site_profile_id PK | bigint | NOT NULL | dast_site_profiles.id 1:N · CASCADE | ||
ci_build_id PK | bigint | NOT NULL | |||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
expired_at | timestamptz | null | |||
token | text | NOT NULL | CHECK char_length(token) <= 255 | ||
url | text | NOT NULL | CHECK char_length(url) <= 255 |
Table checks: char_length(token) <= 255; char_length(url) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
dast_site_token_id | bigint | NOT NULL | dast_site_tokens.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
validation_started_at | timestamptz | null | |||
validation_passed_at | timestamptz | null | |||
validation_failed_at | timestamptz | null | |||
validation_last_retried_at | timestamptz | null | |||
validation_strategy | smallint | NOT NULL | |||
url_base | text | NOT NULL | CHECK char_length(url_base) <= 255 | ||
url_path | text | NOT NULL | CHECK char_length(url_path) <= 255 | ||
state | text | NOT NULL | CAST('pending' AS text) | CHECK char_length(state) <= 255 | |
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: char_length(url_path) <= 255; char_length(state) <= 255; project_id IS NOT NULL; char_length(url_base) <= 255
dast_site_validation_id — dast_sites.dast_site_validation_id may be NULL, so the relationship to dast_site_validations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
url | text | NOT NULL | CHECK char_length(url) <= 255 | ||
dast_site_validation_id | bigint | null | dast_site_validations.id 1:N · SET NULL |
Table checks: char_length(url) <= 255
Tables whose names start with dependency_.
dependency_firewall_activity_stats.dependency_firewall_policy_rule_id → dependency_firewall_policy_rules.id — one dependency_firewall_policy_rules, many dependency_firewall_activity_stats · optional · ON DELETE CASCADE · indexeddependency_firewall_activity_stats.project_id → projects.id — one projects, many dependency_firewall_activity_stats · required · ON DELETE CASCADE · indexeddependency_firewall_policy_rules.security_policy_management_project_id → projects.id — one projects, many dependency_firewall_policy_rules · required · ON DELETE CASCADE · indexeddependency_firewall_policy_rules.security_policy_id → security_policies.id — one security_policies, many dependency_firewall_policy_rules · required · ON DELETE CASCADE · indexeddependency_list_export_part_upload_states.dependency_list_export_part_upload_id → dependency_list_export_part_uploads.id — one dependency_list_export_part_uploads, at most one dependency_list_export_part_upload_states · required · ON DELETE CASCADE · indexeddependency_list_export_part_upload_states.organization_id → organizations.id — one organizations, many dependency_list_export_part_upload_states · required · ON DELETE CASCADE · indexeddependency_list_export_parts.dependency_list_export_id → dependency_list_exports.id — one dependency_list_exports, many dependency_list_export_parts · required · ON DELETE CASCADE · indexeddependency_list_export_upload_states.project_id → projects.id — one projects, many dependency_list_export_upload_states · optional · ON DELETE CASCADE · indexeddependency_list_export_upload_states.organization_id → organizations.id — one organizations, many dependency_list_export_upload_states · optional · ON DELETE CASCADE · indexeddependency_list_export_upload_states.dependency_list_export_upload_id → dependency_list_export_uploads.id — one dependency_list_export_uploads, at most one dependency_list_export_upload_states · required · ON DELETE CASCADE · indexeddependency_list_export_upload_states.namespace_id → namespaces.id — one namespaces, many dependency_list_export_upload_states · optional · ON DELETE CASCADE · indexeddependency_proxy_blobs.group_id → namespaces.id — one namespaces, many dependency_proxy_blobs · required · ON DELETE CASCADE · indexeddependency_proxy_group_settings.group_id → namespaces.id — one namespaces, many dependency_proxy_group_settings · required · ON DELETE CASCADE · indexeddependency_proxy_image_ttl_group_policies.group_id → namespaces.id — one namespaces, at most one dependency_proxy_image_ttl_group_policies · required · ON DELETE CASCADE · indexeddependency_proxy_manifest_states.group_id → namespaces.id — one namespaces, many dependency_proxy_manifest_states · optional · ON DELETE CASCADE · indexeddependency_proxy_manifest_states.dependency_proxy_manifest_id → dependency_proxy_manifests.id — one dependency_proxy_manifests, at most one dependency_proxy_manifest_states · required · ON DELETE CASCADE · indexeddependency_proxy_manifests.group_id → namespaces.id — one namespaces, many dependency_proxy_manifests · required · ON DELETE CASCADE · indexeddependency_proxy_packages_settings.project_id → projects.id — one projects, at most one dependency_proxy_packages_settings · required · ON DELETE CASCADE · indexeddependency_firewall_policy_rule_id — dependency_firewall_activity_stats.dependency_firewall_policy_rule_id may be NULL, so the relationship to dependency_firewall_policy_rules is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
dependency_firewall_policy_rule_id | bigint | null | dependency_firewall_policy_rules.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
count | bigint | NOT NULL | 0 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
stat_time | timestamptz | NOT NULL | |||
outcome | smallint | NOT NULL |
Table checks: outcome <> 2 OR dependency_firewall_policy_rule_id IS NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
security_policy_id | bigint | NOT NULL | security_policies.id 1:N · CASCADE | ||
security_policy_management_project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
rule_index | integer | NOT NULL | 0 | ||
type | smallint | NOT NULL | 0 | ||
content | jsonb | NOT NULL | CAST('{}' AS jsonb) |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
dependency_list_export_part_upload_id | bigint | NOT NULL | dependency_list_export_part_uploads.id 1:1 · CASCADE | ||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 |
Table checks: char_length(verification_failure) <= 255
model_type — model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.model_type, model_id — model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | nextval(CAST('uploads_id_seq' AS regclass)) | ||
size | bigint | NOT NULL | |||
model_id | bigint | NOT NULL | |||
uploaded_by_user_id | bigint | null | |||
organization_id | bigint | null | CHECK organization_id IS NOT NULL | ||
namespace_id | bigint | null | |||
project_id | bigint | null | |||
created_at | timestamp | null | |||
store | integer | NOT NULL | 1 | ||
version | integer | null | 1 | ||
path | text | NOT NULL | CHECK char_length(path) <= 511 | ||
checksum | text | null | CHECK char_length(checksum) <= 64 | ||
model_type PK | text | NOT NULL | |||
uploader | text | NOT NULL | |||
mount_point | text | null | |||
secret | text | null |
Table checks: char_length(path) <= 511; organization_id IS NOT NULL; char_length(checksum) <= 64
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
dependency_list_export_id | bigint | NOT NULL | dependency_list_exports.id 1:N · CASCADE | ||
start_id | bigint | NOT NULL | |||
end_id | bigint | NOT NULL | |||
organization_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | integer | null | |||
file | text | null | CHECK char_length(file) <= 255 |
Table checks: char_length(file) <= 255
project_id — dependency_list_export_upload_states.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id — dependency_list_export_upload_states.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id — dependency_list_export_upload_states.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
dependency_list_export_upload_id | bigint | NOT NULL | dependency_list_export_uploads.id 1:1 · CASCADE | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 |
Table checks: char_length(verification_failure) <= 255; num_nonnulls(namespace_id, organization_id, project_id) = 1
model_type — model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.model_type, model_id — model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | nextval(CAST('uploads_id_seq' AS regclass)) | ||
size | bigint | NOT NULL | |||
model_id | bigint | NOT NULL | |||
uploaded_by_user_id | bigint | null | |||
organization_id | bigint | null | |||
namespace_id | bigint | null | |||
project_id | bigint | null | |||
created_at | timestamp | null | |||
store | integer | NOT NULL | 1 | ||
version | integer | null | 1 | ||
path | text | NOT NULL | CHECK char_length(path) <= 511 | ||
checksum | text | null | CHECK char_length(checksum) <= 64 | ||
model_type PK | text | NOT NULL | |||
uploader | text | NOT NULL | |||
mount_point | text | null | |||
secret | text | null |
Table checks: char_length(path) <= 511; char_length(checksum) <= 64; num_nonnulls(namespace_id, organization_id, project_id) = 1
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | null | |||
user_id | bigint | null | |||
file_store | integer | null | |||
status | smallint | NOT NULL | 0 | ||
file | text | null | CHECK char_length(file) <= 255 | ||
group_id | bigint | null | |||
pipeline_id | bigint | null | |||
export_type | smallint | NOT NULL | 0 | ||
organization_id | bigint | null | |||
expires_at | timestamptz | null | |||
send_email | boolean | NOT NULL | FALSE |
Table checks: num_nonnulls(group_id, organization_id, project_id) = 1; char_length(file) <= 255
{"owner":"group::geo","description":"Geo-specific table to store the verification state of DependencyProxy::Blob objects"}
dependency_proxy_blob_states references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
dependency_proxy_blob_id PK | bigint | NOT NULL | |||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 | ||
group_id | bigint | null | CHECK group_id IS NOT NULL |
Table checks: char_length(verification_failure) <= 255; group_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
size | bigint | null | |||
file_store | integer | null | |||
file_name | varchar | NOT NULL | |||
file | text | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
read_at | timestamptz | NOT NULL | now() |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
enabled | boolean | NOT NULL | TRUE | ||
identity | jsonb | null | |||
secret | jsonb | null |
Table checks: num_nonnulls(identity, secret) = 2 OR num_nulls(identity, secret) = 2
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
group_id PK | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
ttl | integer | null | 90 | ||
enabled | boolean | NOT NULL | FALSE |
group_id — dependency_proxy_manifest_states.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
dependency_proxy_manifest_id PK | bigint | NOT NULL | dependency_proxy_manifests.id 1:1 · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK group_id IS NOT NULL |
Table checks: group_id IS NOT NULL; char_length(verification_failure) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
size | bigint | null | |||
file_store | smallint | null | |||
file_name | text | NOT NULL | CHECK char_length(file_name) <= 255 | ||
file | text | NOT NULL | CHECK char_length(file) <= 255 | ||
digest | text | NOT NULL | CHECK char_length(digest) <= 255 | ||
content_type | text | null | CHECK char_length(content_type) <= 255 | ||
status | smallint | NOT NULL | 0 | ||
read_at | timestamptz | NOT NULL | now() |
Table checks: char_length(file) <= 255; char_length(content_type) <= 255; char_length(file_name) <= 255; char_length(digest) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id PK | bigint | NOT NULL | projects.id 1:1 · CASCADE | ||
enabled | boolean | null | FALSE | ||
maven_external_registry_url | text | null | CHECK char_length(maven_external_registry_url) <= 255 | ||
encrypted_maven_external_registry_username | bytea | null | CHECK octet_length(encrypted_maven_external_registry_username) <= 1020 | ||
encrypted_maven_external_registry_username_iv | bytea | null | CHECK octet_length(encrypted_maven_external_registry_username_iv) <= 1020 | ||
encrypted_maven_external_registry_password | bytea | null | CHECK octet_length(encrypted_maven_external_registry_password) <= 1020 | ||
encrypted_maven_external_registry_password_iv | bytea | null | CHECK octet_length(encrypted_maven_external_registry_password_iv) <= 1020 | ||
npm_external_registry_url | text | null | CHECK char_length(npm_external_registry_url) <= 255 | ||
encrypted_npm_external_registry_basic_auth | bytea | null | CHECK octet_length(encrypted_npm_external_registry_basic_auth) <= 1020 | ||
encrypted_npm_external_registry_basic_auth_iv | bytea | null | CHECK octet_length(encrypted_npm_external_registry_basic_auth_iv) <= 1020 | ||
encrypted_npm_external_registry_auth_token | bytea | null | CHECK octet_length(encrypted_npm_external_registry_auth_token) <= 1020 | ||
encrypted_npm_external_registry_auth_token_iv | bytea | null | CHECK octet_length(encrypted_npm_external_registry_auth_token_iv) <= 1020 |
Table checks: char_length(npm_external_registry_url) <= 255; num_nulls(encrypted_maven_external_registry_username, encrypted_maven_external_registry_password) = 0 OR num_nulls(encrypted_maven_external_registry_username, encrypted_maven_external_registry_password) = 2; octet_length(encrypted_maven_external_registry_username) <= 1020; octet_length(encrypted_npm_external_registry_auth_token) <= 1020; octet_length(encrypted_npm_external_registry_basic_auth) <= 1020; octet_length(encrypted_npm_external_registry_basic_auth_iv) <= 1020; num_nulls(encrypted_npm_external_registry_basic_auth, encrypted_npm_external_registry_auth_token) > 0; char_length(maven_external_registry_url) <= 255; octet_length(encrypted_maven_external_registry_password) <= 1020; octet_length(encrypted_npm_external_registry_auth_token_iv) <= 1020; octet_length(encrypted_maven_external_registry_password_iv) <= 1020; octet_length(encrypted_maven_external_registry_username_iv) <= 1020
Tables whose names start with deploy_.
deploy_keys_projects.project_id → projects.id — one projects, many deploy_keys_projects · required · ON DELETE CASCADE · indexeddeploy_tokens.project_id → projects.id — one projects, many deploy_tokens · optional · ON DELETE CASCADE · indexeddeploy_tokens.creator_id → users.id — one users, many deploy_tokens · optional · ON DELETE SET NULL · indexeddeploy_tokens.group_id → namespaces.id — one namespaces, many deploy_tokens · optional · ON DELETE CASCADE · indexedcreated_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
deploy_key_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
can_push | boolean | NOT NULL | FALSE |
project_id — deploy_tokens.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.creator_id — deploy_tokens.creator_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — deploy_tokens.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id, creator_id, group_id — deploy_tokens has several nullable FKs (project_id, creator_id, group_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
revoked | boolean | null | FALSE | ||
read_repository | boolean | NOT NULL | FALSE | ||
read_registry | boolean | NOT NULL | FALSE | ||
expires_at | timestamptz | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
name | varchar | NOT NULL | |||
username | varchar | null | |||
token_encrypted | varchar(255) | null | |||
deploy_token_type | smallint | NOT NULL | 2 | ||
write_registry | boolean | NOT NULL | FALSE | ||
read_package_registry | boolean | NOT NULL | FALSE | ||
write_package_registry | boolean | NOT NULL | FALSE | ||
creator_id | bigint | null | users.id 1:N · SET NULL | ||
read_virtual_registry | boolean | NOT NULL | FALSE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
write_virtual_registry | boolean | NOT NULL | FALSE | ||
seven_days_notification_sent_at | timestamptz | null | |||
thirty_days_notification_sent_at | timestamptz | null | |||
sixty_days_notification_sent_at | timestamptz | null |
Table checks: num_nonnulls(group_id, project_id) = 1
Tables whose names start with deployment_.
deployment_approvals.user_id → users.id — one users, many deployment_approvals · required · ON DELETE CASCADE · indexeddeployment_approvals.deployment_id → deployments.id — one deployments, many deployment_approvals · required · ON DELETE CASCADE · indexeddeployment_approvals.approval_rule_id → protected_environment_approval_rules.id — one protected_environment_approval_rules, many deployment_approvals · optional · ON DELETE SET NULL · indexeddeployment_approvals.project_id → projects.id — one projects, many deployment_approvals · optional · ON DELETE CASCADE · indexeddeployment_clusters.project_id → projects.id — one projects, many deployment_clusters · optional · ON DELETE CASCADE · indexeddeployment_clusters.deployment_id → deployments.id — one deployments, at most one deployment_clusters · required · ON DELETE CASCADE · indexeddeployment_merge_requests.project_id → projects.id — one projects, many deployment_merge_requests · optional · ON DELETE CASCADE · indexeddeployment_merge_requests.environment_id → environments.id — one environments, many deployment_merge_requests · optional · ON DELETE CASCADE · indexeddeployment_merge_requests.merge_request_id → merge_requests.id — one merge_requests, many deployment_merge_requests · required · ON DELETE CASCADE · indexeddeployment_merge_requests.deployment_id → deployments.id — one deployments, many deployment_merge_requests · required · ON DELETE CASCADE · indexedapproval_rule_id — deployment_approvals.approval_rule_id may be NULL, so the relationship to protected_environment_approval_rules is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — deployment_approvals.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.approval_rule_id, project_id — deployment_approvals has several nullable FKs (approval_rule_id, project_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
deployment_id | bigint | NOT NULL | deployments.id 1:N · CASCADE | ||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
status | smallint | NOT NULL | |||
comment | text | null | CHECK char_length(comment) <= 255 | ||
approval_rule_id | bigint | null | protected_environment_approval_rules.id 1:N · SET NULL | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL | |
ci_build_id | bigint | null |
Table checks: project_id IS NOT NULL; char_length(comment) <= 255
project_id — deployment_clusters.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
deployment_id PK | bigint | NOT NULL | deployments.id 1:1 · CASCADE | ||
cluster_id | bigint | NOT NULL | |||
kubernetes_namespace | varchar(255) | null | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
project_id — deployment_merge_requests.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.environment_id — deployment_merge_requests.environment_id may be NULL, so the relationship to environments is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id, environment_id — deployment_merge_requests has several nullable FKs (project_id, environment_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
deployment_id PK | bigint | NOT NULL | deployments.id 1:N · CASCADE | ||
merge_request_id PK | bigint | NOT NULL | merge_requests.id 1:N · CASCADE | ||
environment_id | bigint | null | environments.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
Tables whose names start with design_.
design_management_action_upload_states.namespace_id → namespaces.id — one namespaces, many design_management_action_upload_states · required · ON DELETE CASCADE · indexeddesign_management_action_upload_states.design_management_action_upload_id → design_management_action_uploads.id — one design_management_action_uploads, at most one design_management_action_upload_states · required · ON DELETE CASCADE · indexeddesign_management_designs.namespace_id → namespaces.id — one namespaces, many design_management_designs · optional · ON DELETE CASCADE · indexeddesign_management_designs.project_id → projects.id — one projects, many design_management_designs · required · ON DELETE CASCADE · indexeddesign_management_designs.issue_id → issues.id — one issues, many design_management_designs · optional · ON DELETE CASCADE · indexeddesign_management_designs_versions.namespace_id → namespaces.id — one namespaces, many design_management_designs_versions · optional · ON DELETE CASCADE · indexeddesign_management_designs_versions.design_id → design_management_designs.id — one design_management_designs, many design_management_designs_versions · required · ON DELETE CASCADE · indexeddesign_management_designs_versions.version_id → design_management_versions.id — one design_management_versions, many design_management_designs_versions · required · ON DELETE CASCADE · indexeddesign_management_repositories.namespace_id → namespaces.id — one namespaces, many design_management_repositories · optional · ON DELETE CASCADE · indexeddesign_management_repositories.project_id → projects.id — one projects, at most one design_management_repositories · required · ON DELETE CASCADE · indexeddesign_management_repository_states.namespace_id → namespaces.id — one namespaces, many design_management_repository_states · optional · ON DELETE CASCADE · indexeddesign_management_repository_states.design_management_repository_id → design_management_repositories.id — one design_management_repositories, at most one design_management_repository_states · required · ON DELETE CASCADE · indexeddesign_management_versions.namespace_id → namespaces.id — one namespaces, many design_management_versions · optional · ON DELETE CASCADE · indexeddesign_management_versions.author_id → users.id — one users, many design_management_versions · optional · ON DELETE SET NULL · indexeddesign_management_versions.issue_id → issues.id — one issues, many design_management_versions · optional · ON DELETE CASCADE · indexeddesign_user_mentions.namespace_id → namespaces.id — one namespaces, many design_user_mentions · optional · ON DELETE CASCADE · indexeddesign_user_mentions.note_id → notes.id — one notes, at most one design_user_mentions · required · ON DELETE CASCADE · indexeddesign_user_mentions.design_id → design_management_designs.id — one design_management_designs, many design_user_mentions · required · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
design_management_action_upload_id | bigint | NOT NULL | design_management_action_uploads.id 1:1 · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 |
Table checks: char_length(verification_failure) <= 255
model_type — model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.model_type, model_id — model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | nextval(CAST('uploads_id_seq' AS regclass)) | ||
size | bigint | NOT NULL | |||
model_id | bigint | NOT NULL | |||
uploaded_by_user_id | bigint | null | |||
organization_id | bigint | null | |||
namespace_id | bigint | null | CHECK namespace_id IS NOT NULL | ||
project_id | bigint | null | |||
created_at | timestamp | null | |||
store | integer | NOT NULL | 1 | ||
version | integer | null | 1 | ||
path | text | NOT NULL | CHECK char_length(path) <= 511 | ||
checksum | text | null | CHECK char_length(checksum) <= 64 | ||
model_type PK | text | NOT NULL | |||
uploader | text | NOT NULL | |||
mount_point | text | null | |||
secret | text | null |
Table checks: char_length(path) <= 511; namespace_id IS NOT NULL; char_length(checksum) <= 64
namespace_id — design_management_designs.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.issue_id — design_management_designs.issue_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id, issue_id — design_management_designs has several nullable FKs (namespace_id, issue_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
issue_id | bigint | null | issues.id 1:N · CASCADE | ||
filename | varchar | NOT NULL | CHECK char_length(CAST(filename AS text)) <= 255 | ||
relative_position | integer | null | |||
iid | integer | null | CHECK iid IS NOT NULL | ||
cached_markdown_version | integer | null | |||
description | text | null | CHECK char_length(description) <= 1000000 | ||
description_html | text | null | |||
imported_from | smallint | NOT NULL | 0 | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: char_length(CAST(filename AS text)) <= 255; char_length(description) <= 1000000; iid IS NOT NULL; namespace_id IS NOT NULL
namespace_id — design_management_designs_versions.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
design_id | bigint | NOT NULL | design_management_designs.id 1:N · CASCADE | ||
version_id | bigint | NOT NULL | design_management_versions.id 1:N · CASCADE | ||
event | smallint | NOT NULL | 0 | ||
image_v432x230 | varchar(255) | null | |||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL
namespace_id — design_management_repositories.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL
namespace_id — design_management_repository_states.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
design_management_repository_id PK | bigint | NOT NULL | design_management_repositories.id 1:1 · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL; char_length(verification_failure) <= 255
namespace_id — design_management_versions.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.author_id — design_management_versions.author_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.issue_id — design_management_versions.issue_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id, author_id, issue_id — design_management_versions has several nullable FKs (namespace_id, author_id, issue_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
sha | bytea | NOT NULL | |||
issue_id | bigint | null | issues.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
author_id | bigint | null | users.id 1:N · SET NULL | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL
namespace_id — design_user_mentions.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
design_id | bigint | NOT NULL | design_management_designs.id 1:N · CASCADE | ||
mentioned_users_ids | bigint[] | null | |||
mentioned_projects_ids | bigint[] | null | |||
mentioned_groups_ids | bigint[] | null | |||
note_id | bigint | NOT NULL | notes.id 1:1 · CASCADE | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL
Tables whose names start with dora_.
dora_configurations.project_id → projects.id — one projects, at most one dora_configurations · required · ON DELETE CASCADE · indexeddora_daily_metrics.project_id → projects.id — one projects, many dora_daily_metrics · optional · ON DELETE CASCADE · indexeddora_daily_metrics.environment_id → environments.id — one environments, many dora_daily_metrics · required · ON DELETE CASCADE · indexeddora_performance_scores.project_id → projects.id — one projects, many dora_performance_scores · required · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:1 · CASCADE | ||
branches_for_lead_time_for_changes | text[] | NOT NULL | CAST('{}' AS text[]) |
project_id — dora_daily_metrics.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
environment_id | bigint | NOT NULL | environments.id 1:N · CASCADE | ||
date | date | NOT NULL | |||
deployment_frequency | integer | null | CHECK deployment_frequency >= 0 | ||
lead_time_for_changes_in_seconds | integer | null | CHECK lead_time_for_changes_in_seconds >= 0 | ||
time_to_restore_service_in_seconds | integer | null | |||
incidents_count | integer | null | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL; deployment_frequency >= 0; lead_time_for_changes_in_seconds >= 0
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
date | date | NOT NULL | |||
deployment_frequency | smallint | null | |||
lead_time_for_changes | smallint | null | |||
time_to_restore_service | smallint | null | |||
change_failure_rate | smallint | null |
Tables whose names start with duo_.
duo_workflow_session_artifacts.workflow_id → duo_workflows_workflows.id — one duo_workflows_workflows, at most one duo_workflow_session_artifacts · required · ON DELETE CASCADE · indexedduo_workflow_session_artifacts.project_id → projects.id — one projects, many duo_workflow_session_artifacts · optional · ON DELETE CASCADE · indexedduo_workflow_session_artifacts.namespace_id → namespaces.id — one namespaces, many duo_workflow_session_artifacts · optional · ON DELETE CASCADE · indexedduo_workflow_session_artifacts.user_id → users.id — one users, many duo_workflow_session_artifacts · required · ON DELETE CASCADE · indexedduo_workflows_checkpoint_writes.namespace_id → namespaces.id — one namespaces, many duo_workflows_checkpoint_writes · optional · ON DELETE CASCADE · indexedduo_workflows_checkpoint_writes.project_id → projects.id — one projects, many duo_workflows_checkpoint_writes · optional · ON DELETE CASCADE · indexedduo_workflows_checkpoint_writes.workflow_id → duo_workflows_workflows.id — one duo_workflows_workflows, many duo_workflows_checkpoint_writes · required · ON DELETE CASCADE · indexedduo_workflows_events.namespace_id → namespaces.id — one namespaces, many duo_workflows_events · optional · ON DELETE CASCADE · indexedduo_workflows_events.workflow_id → duo_workflows_workflows.id — one duo_workflows_workflows, many duo_workflows_events · required · ON DELETE CASCADE · indexedduo_workflows_events.project_id → projects.id — one projects, many duo_workflows_events · optional · ON DELETE CASCADE · indexedduo_workflows_workflow_merge_requests.project_id → projects.id — one projects, many duo_workflows_workflow_merge_requests · optional · ON DELETE CASCADE · indexedduo_workflows_workflow_merge_requests.merge_request_id → merge_requests.id — one merge_requests, many duo_workflows_workflow_merge_requests · required · ON DELETE CASCADE · indexedduo_workflows_workflow_merge_requests.workflow_id → duo_workflows_workflows.id — one duo_workflows_workflows, many duo_workflows_workflow_merge_requests · required · ON DELETE CASCADE · indexedduo_workflows_workflow_merge_requests.namespace_id → namespaces.id — one namespaces, many duo_workflows_workflow_merge_requests · optional · ON DELETE CASCADE · indexedduo_workflows_workflow_notes.namespace_id → namespaces.id — one namespaces, many duo_workflows_workflow_notes · optional · ON DELETE CASCADE · indexedduo_workflows_workflow_notes.project_id → projects.id — one projects, many duo_workflows_workflow_notes · optional · ON DELETE CASCADE · indexedduo_workflows_workflow_notes.workflow_id → duo_workflows_workflows.id — one duo_workflows_workflows, many duo_workflows_workflow_notes · required · ON DELETE CASCADE · indexedduo_workflows_workflow_notes.note_id → notes.id — one notes, many duo_workflows_workflow_notes · required · ON DELETE CASCADE · indexedduo_workflows_workflow_pipelines.namespace_id → namespaces.id — one namespaces, many duo_workflows_workflow_pipelines · optional · ON DELETE CASCADE · indexedduo_workflows_workflow_pipelines.project_id → projects.id — one projects, many duo_workflows_workflow_pipelines · optional · ON DELETE CASCADE · indexedduo_workflows_workflow_pipelines.workflow_id → duo_workflows_workflows.id — one duo_workflows_workflows, many duo_workflows_workflow_pipelines · required · ON DELETE CASCADE · indexedduo_workflows_workflow_work_items.project_id → projects.id — one projects, many duo_workflows_workflow_work_items · optional · ON DELETE CASCADE · indexedduo_workflows_workflow_work_items.namespace_id → namespaces.id — one namespaces, many duo_workflows_workflow_work_items · optional · ON DELETE CASCADE · indexedduo_workflows_workflow_work_items.work_item_id → issues.id — one issues, many duo_workflows_workflow_work_items · required · ON DELETE CASCADE · indexedduo_workflows_workflow_work_items.workflow_id → duo_workflows_workflows.id — one duo_workflows_workflows, many duo_workflows_workflow_work_items · required · ON DELETE CASCADE · indexedduo_workflows_workflows.project_id → projects.id — one projects, many duo_workflows_workflows · optional · ON DELETE CASCADE · indexedduo_workflows_workflows.ai_catalog_item_version_id → ai_catalog_item_versions.id — one ai_catalog_item_versions, many duo_workflows_workflows · optional · ON DELETE SET NULL · indexedduo_workflows_workflows.namespace_id → namespaces.id — one namespaces, many duo_workflows_workflows · optional · ON DELETE CASCADE · indexedduo_workflows_workflows.user_id → users.id — one users, many duo_workflows_workflows · required · ON DELETE CASCADE · indexedduo_workflows_workflows.issue_id → issues.id — one issues, many duo_workflows_workflows · optional · ON DELETE CASCADE · indexedduo_workflows_workflows.ai_catalog_item_id → ai_catalog_items.id — one ai_catalog_items, many duo_workflows_workflows · optional · ON DELETE SET NULL · indexedduo_workflows_workflows.service_account_id → users.id — one users, many duo_workflows_workflows · optional · ON DELETE SET NULL · indexedduo_workflows_workflows.agent_identity_id → ai_agent_identities.id — one ai_agent_identities, many duo_workflows_workflows · optional · ON DELETE SET NULL · indexedduo_workflows_workflows.merge_request_id → merge_requests.id — one merge_requests, many duo_workflows_workflows · optional · ON DELETE CASCADE · indexedduo_workflows_workloads.workflow_id → duo_workflows_workflows.id — one duo_workflows_workflows, many duo_workflows_workloads · required · ON DELETE CASCADE · indexedduo_workflows_workloads.project_id → projects.id — one projects, many duo_workflows_workloads · required · ON DELETE CASCADE · indexedproject_id — duo_workflow_session_artifacts.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id — duo_workflow_session_artifacts.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id, namespace_id — duo_workflow_session_artifacts has several nullable FKs (project_id, namespace_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
workflow_id | bigint | NOT NULL | duo_workflows_workflows.id 1:1 · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL | |
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
status | smallint | NOT NULL | 0 | ||
workflow_definition | text | NOT NULL | CAST('software_development' AS text) | CHECK char_length(workflow_definition) <= 255 | |
credits_used | double precision | NOT NULL | 0.0 | ||
model_used | text | NOT NULL | CAST('' AS text) | CHECK char_length(model_used) <= 255 | |
workflow_created_at | timestamptz | NOT NULL | |||
workflow_updated_at | timestamptz | NOT NULL |
Table checks: char_length(workflow_definition) <= 255; char_length(model_used) <= 255; namespace_id IS NOT NULL
namespace_id — duo_workflows_checkpoint_writes.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — duo_workflows_checkpoint_writes.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
workflow_id | bigint | NOT NULL | duo_workflows_workflows.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
idx | integer | NOT NULL | |||
thread_ts | text | NOT NULL | CHECK char_length(thread_ts) <= 255 | ||
task | text | NOT NULL | CHECK char_length(task) <= 255 | ||
channel | text | NOT NULL | CHECK char_length(channel) <= 255 | ||
write_type | text | NOT NULL | CHECK char_length(write_type) <= 255 | ||
data | text | NOT NULL | CHECK char_length(data) <= 10000 | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE |
Table checks: char_length(data) <= 10000; num_nonnulls(namespace_id, project_id) = 1; char_length(write_type) <= 255; char_length(task) <= 255; char_length(channel) <= 255; char_length(thread_ts) <= 255
namespace_id — duo_workflows_events.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — duo_workflows_events.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
workflow_id | bigint | NOT NULL | duo_workflows_workflows.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
event_type | smallint | NOT NULL | |||
event_status | smallint | NOT NULL | |||
message | text | null | CHECK char_length(message) <= 16384 | ||
correlation_id_value | text | null | CHECK char_length(correlation_id_value) <= 128 | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE |
Table checks: char_length(message) <= 16384; char_length(correlation_id_value) <= 128; num_nonnulls(namespace_id, project_id) = 1
project_id — duo_workflows_workflow_merge_requests.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id — duo_workflows_workflow_merge_requests.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
workflow_id | bigint | NOT NULL | duo_workflows_workflows.id 1:N · CASCADE | ||
merge_request_id | bigint | NOT NULL | merge_requests.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
link_type | smallint | NOT NULL |
Table checks: num_nonnulls(namespace_id, project_id) = 1
namespace_id — duo_workflows_workflow_notes.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — duo_workflows_workflow_notes.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
workflow_id | bigint | NOT NULL | duo_workflows_workflows.id 1:N · CASCADE | ||
note_id | bigint | NOT NULL | notes.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
link_type | smallint | NOT NULL |
Table checks: num_nonnulls(namespace_id, project_id) = 1
namespace_id — duo_workflows_workflow_pipelines.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — duo_workflows_workflow_pipelines.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
workflow_id | bigint | NOT NULL | duo_workflows_workflows.id 1:N · CASCADE | ||
pipeline_id | bigint | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
link_type | smallint | NOT NULL |
Table checks: num_nonnulls(namespace_id, project_id) = 1
project_id — duo_workflows_workflow_work_items.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id — duo_workflows_workflow_work_items.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
workflow_id | bigint | NOT NULL | duo_workflows_workflows.id 1:N · CASCADE | ||
work_item_id | bigint | NOT NULL | issues.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
link_type | smallint | NOT NULL |
Table checks: num_nonnulls(namespace_id, project_id) = 1
project_id — duo_workflows_workflows.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.ai_catalog_item_version_id — duo_workflows_workflows.ai_catalog_item_version_id may be NULL, so the relationship to ai_catalog_item_versions is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id — duo_workflows_workflows.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.issue_id — duo_workflows_workflows.issue_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.ai_catalog_item_id — duo_workflows_workflows.ai_catalog_item_id may be NULL, so the relationship to ai_catalog_items is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.service_account_id — duo_workflows_workflows.service_account_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.agent_identity_id — duo_workflows_workflows.agent_identity_id may be NULL, so the relationship to ai_agent_identities is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.merge_request_id — duo_workflows_workflows.merge_request_id may be NULL, so the relationship to merge_requests is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id, ai_catalog_item_version_id, namespace_id, issue_id, ai_catalog_item_id, service_account_id, agent_identity_id, merge_request_id — duo_workflows_workflows has several nullable FKs (project_id, ai_catalog_item_version_id, namespace_id, issue_id, ai_catalog_item_id, service_account_id, agent_identity_id, merge_request_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
goal | text | null | CHECK char_length(goal) <= 16384 | ||
agent_privileges | smallint[] | NOT NULL | CAST('{1,2}' AS smallint[]) | ||
workflow_definition | text | NOT NULL | CAST('software_development' AS text) | CHECK char_length(workflow_definition) <= 255 | |
allow_agent_to_request_user | boolean | NOT NULL | TRUE | ||
pre_approved_agent_privileges | smallint[] | NOT NULL | CAST('{1,2}' AS smallint[]) | ||
image | text | null | CHECK char_length(image) <= 2048 | ||
environment | smallint | null | |||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
ai_catalog_item_version_id | bigint | null | ai_catalog_item_versions.id 1:N · SET NULL | ||
issue_id | bigint | null | issues.id 1:N · CASCADE | ||
merge_request_id | bigint | null | merge_requests.id 1:N · CASCADE | ||
service_account_id | bigint | null | users.id 1:N · SET NULL | ||
tool_call_approvals | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
ai_catalog_item_id | bigint | null | ai_catalog_items.id 1:N · SET NULL | ||
summary | text | null | CHECK char_length(summary) <= 1024 | ||
messaging_callback_context | jsonb | null | |||
title | text | null | CHECK char_length(title) <= 40 | ||
model_metadata_json | text | null | CHECK char_length(model_metadata_json) <= 1024 | ||
incremental_checkpoints_enabled | boolean | NOT NULL | FALSE | ||
agent_type | text | null | CHECK char_length(agent_type) <= 50 | ||
jsonl_sha256 | text | null | CHECK char_length(jsonl_sha256) <= 64 | ||
idempotency_key | text | null | CHECK char_length(idempotency_key) <= 255 | ||
sync_type | smallint | null | |||
agent_identity_id | bigint | null | ai_agent_identities.id 1:N · SET NULL | ||
flow_metadata_json | text | null | CHECK char_length(flow_metadata_json) <= 1024 | ||
web_search_enabled | boolean | NOT NULL | FALSE | ||
trigger_source | smallint | NOT NULL | 0 | ||
trigger_flow_trigger_id | bigint | null |
Table checks: char_length(title) <= 40; char_length(summary) <= 1024; char_length(goal) <= 16384; char_length(image) <= 2048; char_length(flow_metadata_json) <= 1024; char_length(agent_type) <= 50; num_nonnulls(namespace_id, project_id) = 1; char_length(idempotency_key) <= 255; char_length(jsonl_sha256) <= 64; char_length(model_metadata_json) <= 1024; char_length(workflow_definition) <= 255; num_nonnulls(issue_id, merge_request_id) <= 1
project_id, workflow_id — duo_workflows_workloads looks like a many-to-many link table but has no unique constraint across (project_id, workflow_id). The same pair can be inserted twice; every join through it will double-count.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
workflow_id | bigint | NOT NULL | duo_workflows_workflows.id 1:N · CASCADE | ||
workload_id | bigint | NOT NULL |
Tables whose names start with elastic_.
elastic_group_index_statuses.namespace_id → namespaces.id — one namespaces, at most one elastic_group_index_statuses · required · ON DELETE CASCADE · indexedelastic_reindexing_slices.elastic_reindexing_subtask_id → elastic_reindexing_subtasks.id — one elastic_reindexing_subtasks, many elastic_reindexing_slices · required · ON DELETE CASCADE · indexedelastic_reindexing_subtasks.elastic_reindexing_task_id → elastic_reindexing_tasks.id — one elastic_reindexing_tasks, many elastic_reindexing_subtasks · required · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
namespace_id PK | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
wiki_indexed_at | timestamptz | null | |||
last_wiki_commit | bytea | null |
elastic_index_settings references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
number_of_replicas | smallint | NOT NULL | 1 | ||
number_of_shards | smallint | NOT NULL | 5 | ||
alias_name | text | NOT NULL | CHECK char_length(alias_name) <= 255 |
Table checks: char_length(alias_name) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
elastic_reindexing_subtask_id | bigint | NOT NULL | elastic_reindexing_subtasks.id 1:N · CASCADE | ||
elastic_slice | smallint | NOT NULL | 0 | ||
elastic_max_slice | smallint | NOT NULL | 0 | ||
retry_attempt | smallint | NOT NULL | 0 | ||
elastic_task | text | null | CHECK char_length(elastic_task) <= 255 |
Table checks: char_length(elastic_task) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
elastic_reindexing_task_id | bigint | NOT NULL | elastic_reindexing_tasks.id 1:N · CASCADE | ||
alias_name | text | NOT NULL | CHECK char_length(alias_name) <= 255 | ||
index_name_from | text | NOT NULL | CHECK char_length(index_name_from) <= 255 | ||
index_name_to | text | NOT NULL | CHECK char_length(index_name_to) <= 255 | ||
elastic_task | text | null | CHECK char_length(elastic_task) <= 255 | ||
documents_count_target | integer | null | |||
documents_count | integer | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
Table checks: char_length(elastic_task) <= 255; char_length(alias_name) <= 255; char_length(index_name_from) <= 255; char_length(index_name_to) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
state | smallint | NOT NULL | 0 | ||
in_progress | boolean | NOT NULL | TRUE | ||
error_message | text | null | CHECK char_length(error_message) <= 255 | ||
delete_original_index_at | timestamptz | null | |||
max_slices_running | smallint | NOT NULL | 60 | ||
slice_multiplier | smallint | NOT NULL | 2 | ||
targets | text[] | null | |||
options | jsonb | NOT NULL | CAST('{}' AS jsonb) |
Table checks: char_length(error_message) <= 255
Tables whose names start with elasticsearch_.
elasticsearch_indexed_namespaces.namespace_id → namespaces.id — one namespaces, at most one elasticsearch_indexed_namespaces · required · ON DELETE CASCADE · indexedelasticsearch_indexed_projects.project_id → projects.id — one projects, at most one elasticsearch_indexed_projects · required · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
namespace_id PK | bigint | NOT NULL | namespaces.id 1:1 · CASCADE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id PK | bigint | NOT NULL | projects.id 1:1 · CASCADE |
Tables whose names start with enabled_.
enabled_foundational_flow_check_results.enabled_foundational_flow_id → enabled_foundational_flows.id — one enabled_foundational_flows, many enabled_foundational_flow_check_results · required · ON DELETE CASCADE · indexedenabled_foundational_flow_check_results.organization_id → organizations.id — one organizations, many enabled_foundational_flow_check_results · required · ON DELETE CASCADE · indexedenabled_foundational_flows.project_id → projects.id — one projects, many enabled_foundational_flows · optional · ON DELETE CASCADE · indexedenabled_foundational_flows.catalog_item_id → ai_catalog_items.id — one ai_catalog_items, many enabled_foundational_flows · required · ON DELETE CASCADE · indexedenabled_foundational_flows.namespace_id → namespaces.id — one namespaces, many enabled_foundational_flows · optional · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
enabled_foundational_flow_id | bigint | NOT NULL | enabled_foundational_flows.id 1:N · CASCADE | ||
check_id | smallint | NOT NULL | |||
status | smallint | NOT NULL | |||
message | text | null | CHECK char_length(message) <= 4096 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
Table checks: char_length(message) <= 4096
project_id — enabled_foundational_flows.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id — enabled_foundational_flows.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
catalog_item_id | bigint | NOT NULL | ai_catalog_items.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
Table checks: num_nonnulls(namespace_id, project_id) = 1
Tables whose names start with epic_.
epic_issues.work_item_parent_link_id → work_item_parent_links.id — one work_item_parent_links, at most one epic_issues · optional · ON DELETE CASCADE · indexedepic_issues.namespace_id → namespaces.id — one namespaces, many epic_issues · optional · ON DELETE CASCADE · indexedepic_issues.issue_id → issues.id — one issues, at most one epic_issues · required · ON DELETE CASCADE · indexedepic_issues.epic_id → epics.id — one epics, many epic_issues · required · ON DELETE CASCADE · indexedepic_user_mentions.group_id → namespaces.id — one namespaces, many epic_user_mentions · optional · ON DELETE CASCADE · indexedepic_user_mentions.note_id → notes.id — one notes, many epic_user_mentions · optional · ON DELETE CASCADE · indexedepic_user_mentions.epic_id → epics.id — one epics, many epic_user_mentions · required · ON DELETE CASCADE · indexedwork_item_parent_link_id — epic_issues.work_item_parent_link_id may be NULL, so the relationship to work_item_parent_links is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id — epic_issues.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.work_item_parent_link_id, namespace_id — epic_issues has several nullable FKs (work_item_parent_link_id, namespace_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
epic_id | bigint | NOT NULL | epics.id 1:N · CASCADE | ||
issue_id | bigint | NOT NULL | issues.id 1:1 · CASCADE | ||
relative_position | integer | null | |||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL | |
work_item_parent_link_id | bigint | null | work_item_parent_links.id 1:1 · CASCADE | CHECK work_item_parent_link_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL; work_item_parent_link_id IS NOT NULL
group_id — epic_user_mentions.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.note_id — epic_user_mentions.note_id may be NULL, so the relationship to notes is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id, note_id — epic_user_mentions has several nullable FKs (group_id, note_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
epic_id | bigint | NOT NULL | epics.id 1:N · CASCADE | ||
mentioned_users_ids | bigint[] | null | |||
mentioned_projects_ids | bigint[] | null | |||
mentioned_groups_ids | bigint[] | null | |||
note_id | bigint | null | notes.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK group_id IS NOT NULL |
Table checks: group_id IS NOT NULL
Tables whose names start with error_.
error_tracking_client_keys.project_id → projects.id — one projects, many error_tracking_client_keys · required · ON DELETE CASCADE · indexederror_tracking_error_events.project_id → projects.id — one projects, many error_tracking_error_events · optional · ON DELETE CASCADE · indexederror_tracking_error_events.error_id → error_tracking_errors.id — one error_tracking_errors, many error_tracking_error_events · required · ON DELETE CASCADE · indexederror_tracking_errors.project_id → projects.id — one projects, many error_tracking_errors · required · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
active | boolean | NOT NULL | TRUE | ||
public_key | text | NOT NULL | CHECK char_length(public_key) <= 255 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
Table checks: char_length(public_key) <= 255
project_id — error_tracking_error_events.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
error_id | bigint | NOT NULL | error_tracking_errors.id 1:N · CASCADE | ||
description | text | NOT NULL | CHECK char_length(description) <= 1024 | ||
environment | text | null | CHECK char_length(environment) <= 255 | ||
level | text | null | CHECK char_length(level) <= 255 | ||
occurred_at | timestamptz | NOT NULL | |||
payload | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL; char_length(description) <= 1024; char_length(level) <= 255; char_length(environment) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
description | text | NOT NULL | CHECK char_length(description) <= 1024 | ||
actor | text | NOT NULL | CHECK char_length(actor) <= 255 | ||
first_seen_at | timestamptz | NOT NULL | now() | ||
last_seen_at | timestamptz | NOT NULL | now() | ||
platform | text | null | CHECK char_length(platform) <= 255 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
events_count | bigint | NOT NULL | 0 | ||
status | smallint | NOT NULL | 0 |
Table checks: char_length(name) <= 255; char_length(actor) <= 255; char_length(description) <= 1024; char_length(platform) <= 255
Tables whose names start with external_.
external_approval_rules.project_id → projects.id — one projects, many external_approval_rules · required · ON DELETE CASCADE · indexedexternal_status_checks.project_id → projects.id — one projects, many external_status_checks · required · ON DELETE CASCADE · indexedexternal_status_checks_protected_branches.project_id → projects.id — one projects, many external_status_checks_protected_branches · optional · ON DELETE CASCADE · indexedexternal_status_checks_protected_branches.protected_branch_id → protected_branches.id — one protected_branches, many external_status_checks_protected_branches · required · ON DELETE CASCADE · indexedexternal_status_checks_protected_branches.external_status_check_id → external_status_checks.id — one external_status_checks, many external_status_checks_protected_branches · required · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
external_url | text | NOT NULL | CHECK char_length(external_url) <= 255 | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 |
Table checks: char_length(name) <= 255; char_length(external_url) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | |||
pull_request_iid | integer | NOT NULL | |||
status | smallint | NOT NULL | |||
source_branch | varchar(255) | NOT NULL | |||
target_branch | varchar(255) | NOT NULL | |||
source_repository | varchar(255) | NOT NULL | |||
target_repository | varchar(255) | NOT NULL | |||
source_sha | bytea | NOT NULL | |||
target_sha | bytea | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
external_url | text | NOT NULL | CHECK char_length(external_url) <= 255 | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
encrypted_shared_secret | bytea | null | |||
encrypted_shared_secret_iv | bytea | null |
Table checks: char_length(name) <= 255; char_length(external_url) <= 255
project_id — external_status_checks_protected_branches.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
external_status_check_id | bigint | NOT NULL | external_status_checks.id 1:N · CASCADE | ||
protected_branch_id | bigint | NOT NULL | protected_branches.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
Tables whose names start with fork_.
fork_network_members.forked_from_project_id → projects.id — one projects, many fork_network_members · optional · ON DELETE SET NULL · indexedfork_network_members.fork_network_id → fork_networks.id — one fork_networks, many fork_network_members · required · ON DELETE CASCADE · indexedfork_network_members.project_id → projects.id — one projects, at most one fork_network_members · required · ON DELETE CASCADE · indexedfork_networks.root_project_id → projects.id — one projects, at most one fork_networks · optional · ON DELETE SET NULL · indexedforked_from_project_id — fork_network_members.forked_from_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
fork_network_id | bigint | NOT NULL | fork_networks.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:1 · CASCADE | ||
forked_from_project_id | bigint | null | projects.id 1:N · SET NULL |
root_project_id — fork_networks.root_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
root_project_id | bigint | null | projects.id 1:1 · SET NULL | ||
deleted_root_project_name | varchar | null | |||
organization_id | bigint | NOT NULL |
Tables whose names start with geo_.
geo_event_log.cache_invalidation_event_id → geo_cache_invalidation_events.id — one geo_cache_invalidation_events, many geo_event_log · optional · ON DELETE CASCADE · indexedgeo_event_log.geo_event_id → geo_events.id — one geo_events, many geo_event_log · optional · ON DELETE CASCADE · indexedgeo_node_namespace_links.geo_node_id → geo_nodes.id — one geo_nodes, many geo_node_namespace_links · required · ON DELETE CASCADE · indexedgeo_node_organization_links.geo_node_id → geo_nodes.id — one geo_nodes, many geo_node_organization_links · required · ON DELETE CASCADE · indexedgeo_node_statuses.geo_node_id → geo_nodes.id — one geo_nodes, at most one geo_node_statuses · required · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
key | varchar | NOT NULL |
geo_ci_job_artifact_verification_summaries references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
bucket_number | integer | NOT NULL | |||
total_count | integer | NOT NULL | 0 | ||
verified_count | integer | NOT NULL | 0 | ||
failed_count | integer | NOT NULL | 0 | ||
state | smallint | NOT NULL | 0 | ||
last_calculated_at | timestamptz | null | |||
state_changed_at | timestamptz | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
cache_invalidation_event_id — geo_event_log.cache_invalidation_event_id may be NULL, so the relationship to geo_cache_invalidation_events is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.geo_event_id — geo_event_log.geo_event_id may be NULL, so the relationship to geo_events is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.cache_invalidation_event_id, geo_event_id — geo_event_log has several nullable FKs (cache_invalidation_event_id, geo_event_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamp | NOT NULL | |||
cache_invalidation_event_id | bigint | null | geo_cache_invalidation_events.id 1:N · CASCADE | ||
geo_event_id | bigint | null | geo_events.id 1:N · CASCADE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
replicable_name | varchar(255) | NOT NULL | |||
event_name | varchar(255) | NOT NULL | |||
payload | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
created_at | timestamptz | NOT NULL |
created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
geo_node_id | bigint | NOT NULL | geo_nodes.id 1:N · CASCADE | ||
namespace_id | bigint | NOT NULL | |||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
geo_node_id | bigint | NOT NULL | geo_nodes.id 1:N · CASCADE | ||
organization_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
last_event_date — last_event_date stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.cursor_last_event_date — cursor_last_event_date stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.last_successful_status_check_at — last_successful_status_check_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
geo_node_id | bigint | NOT NULL | geo_nodes.id 1:1 · CASCADE | ||
db_replication_lag_seconds | integer | null | |||
last_event_id | bigint | null | |||
last_event_date | timestamp | null | |||
cursor_last_event_id | bigint | null | |||
cursor_last_event_date | timestamp | null | |||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL | |||
last_successful_status_check_at | timestamp | null | |||
status_message | varchar | null | |||
replication_slots_count | integer | null | |||
replication_slots_used_count | integer | null | |||
replication_slots_max_retained_wal_bytes | bigint | null | |||
version | varchar | null | |||
revision | varchar | null | |||
storage_configuration_digest | bytea | null | |||
projects_count | integer | null | |||
status | jsonb | NOT NULL | CAST('{}' AS jsonb) |
selective_sync_type — selective_sync_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
primary | boolean | NOT NULL | FALSE | ||
oauth_application_id | bigint | null | |||
enabled | boolean | NOT NULL | TRUE | ||
access_key | varchar | null | |||
encrypted_secret_access_key | varchar | null | |||
encrypted_secret_access_key_iv | varchar | null | |||
clone_url_prefix | varchar | null | |||
files_max_capacity | integer | NOT NULL | 10 | ||
repos_max_capacity | integer | NOT NULL | 10 | ||
url | varchar | NOT NULL | |||
selective_sync_type | varchar | null | |||
selective_sync_shards | text | null | |||
verification_max_capacity | integer | NOT NULL | 10 | ||
minimum_reverification_interval | integer | NOT NULL | 90 | ||
internal_url | varchar | null | |||
name | varchar | NOT NULL | |||
container_repositories_max_capacity | integer | NOT NULL | 2 | ||
created_at | timestamptz | null | |||
updated_at | timestamptz | null | |||
sync_object_storage | boolean | NOT NULL | FALSE | ||
blob_download_timeout | integer | NOT NULL | 28800 | ||
checksum_mismatch_report_threshold | integer | NOT NULL | 3 | ||
checksum_mismatch_self_heal_cooldown_minutes | integer | NOT NULL | 60 |
Tables whose names start with gitlab_.
gitlab_subscriptions.hosted_plan_id → plans.id — one plans, many gitlab_subscriptions · optional · ON DELETE CASCADE · indexedgitlab_subscriptions.namespace_id → namespaces.id — one namespaces, at most one gitlab_subscriptions · optional · ON DELETE CASCADE · indexedgitlab_subscription_histories references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
gitlab_subscription_created_at | timestamptz | null | |||
gitlab_subscription_updated_at | timestamptz | null | |||
start_date | date | null | |||
end_date | date | null | |||
trial_ends_on | date | null | |||
namespace_id | bigint | null | CHECK namespace_id IS NOT NULL | ||
hosted_plan_id | bigint | null | |||
max_seats_used | integer | null | |||
seats | integer | null | |||
trial | boolean | null | |||
change_type | smallint | null | |||
gitlab_subscription_id | bigint | NOT NULL | |||
created_at | timestamptz | null | |||
trial_starts_on | date | null | |||
auto_renew | boolean | null | |||
trial_extension_type | smallint | null | |||
seats_in_use | integer | null | |||
hosted_plan_name_uid | smallint | null | |||
contract_overages_allowed | boolean | null |
Table checks: namespace_id IS NOT NULL
hosted_plan_id — gitlab_subscriptions.hosted_plan_id may be NULL, so the relationship to plans is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id — gitlab_subscriptions.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.hosted_plan_id, namespace_id — gitlab_subscriptions has several nullable FKs (hosted_plan_id, namespace_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
start_date | date | null | |||
end_date | date | null | |||
trial_ends_on | date | null | |||
namespace_id | bigint | null | namespaces.id 1:1 · CASCADE | CHECK namespace_id IS NOT NULL | |
hosted_plan_id | bigint | null | plans.id 1:N · CASCADE | ||
max_seats_used | integer | null | 0 | ||
seats | integer | null | 0 | ||
trial | boolean | null | FALSE | ||
trial_starts_on | date | null | |||
auto_renew | boolean | null | |||
seats_in_use | integer | NOT NULL | 0 | ||
seats_owed | integer | NOT NULL | 0 | ||
trial_extension_type | smallint | null | |||
max_seats_used_changed_at | timestamptz | null | |||
last_seat_refresh_at | timestamptz | null | |||
hosted_plan_name_uid | smallint | null | |||
contract_overages_allowed | boolean | NOT NULL | TRUE |
Table checks: namespace_id IS NOT NULL
Tables whose names start with govern_.
govern_policy_enforcements.govern_policy_id → govern_policies.id — one govern_policies, many govern_policy_enforcements · required · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
organization_id | bigint | NOT NULL | |||
namespace_id | bigint | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
version | integer | NOT NULL | 1 | ||
trigger_type | smallint | NOT NULL | |||
mode | smallint | NOT NULL | 2 | ||
lifecycle_state | smallint | NOT NULL | 0 | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
description | text | null | CHECK char_length(description) <= 4096 | ||
scope_rego | text | null | CHECK char_length(scope_rego) <= 4096 | ||
policy_scope | jsonb | null | |||
rules | jsonb | NOT NULL | CAST('[]' AS jsonb) | ||
actions | jsonb | NOT NULL | CAST('[]' AS jsonb) |
Table checks: char_length(description) <= 4096; char_length(scope_rego) <= 4096; char_length(name) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
organization_id | bigint | NOT NULL | |||
govern_policy_id | bigint | NOT NULL | govern_policies.id 1:N · CASCADE | ||
project_id | bigint | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
last_evaluated_at | timestamptz | null | |||
state | smallint | NOT NULL | 0 |
Tables whose names start with gpg_.
gpg_key_subkeys.user_id → users.id — one users, many gpg_key_subkeys · optional · ON DELETE CASCADE · indexedgpg_key_subkeys.gpg_key_id → gpg_keys.id — one gpg_keys, many gpg_key_subkeys · required · ON DELETE CASCADE · indexedgpg_keys.user_id → users.id — one users, many gpg_keys · optional · ON DELETE CASCADE · indexedgpg_signatures.project_id → projects.id — one projects, many gpg_signatures · optional · ON DELETE CASCADE · indexedgpg_signatures.gpg_key_subkey_id → gpg_key_subkeys.id — one gpg_key_subkeys, many gpg_signatures · optional · ON DELETE SET NULL · indexedgpg_signatures.gpg_key_id → gpg_keys.id — one gpg_keys, many gpg_signatures · optional · ON DELETE SET NULL · indexeduser_id — gpg_key_subkeys.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
gpg_key_id | bigint | NOT NULL | gpg_keys.id 1:N · CASCADE | ||
keyid | bytea | null | |||
fingerprint | bytea | null | |||
user_id | bigint | null | users.id 1:N · CASCADE | CHECK user_id IS NOT NULL |
Table checks: user_id IS NOT NULL
user_id — gpg_keys.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
user_id | bigint | null | users.id 1:N · CASCADE | CHECK user_id IS NOT NULL | |
primary_keyid | bytea | null | |||
fingerprint | bytea | null | |||
key | text | null | |||
externally_verified | boolean | NOT NULL | FALSE | ||
externally_verified_at | timestamptz | null |
Table checks: user_id IS NOT NULL
project_id — gpg_signatures.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.gpg_key_subkey_id — gpg_signatures.gpg_key_subkey_id may be NULL, so the relationship to gpg_key_subkeys is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.gpg_key_id — gpg_signatures.gpg_key_id may be NULL, so the relationship to gpg_keys is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id, gpg_key_subkey_id, gpg_key_id — gpg_signatures has several nullable FKs (project_id, gpg_key_subkey_id, gpg_key_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL | |
gpg_key_id | bigint | null | gpg_keys.id 1:N · SET NULL | ||
commit_sha | bytea | null | |||
gpg_key_primary_keyid | bytea | null | |||
gpg_key_user_name | text | null | |||
gpg_key_user_email | text | null | |||
verification_status | smallint | NOT NULL | 0 | ||
gpg_key_subkey_id | bigint | null | gpg_key_subkeys.id 1:N · SET NULL | ||
committer_email | text | null | CHECK char_length(committer_email) <= 255 |
Table checks: char_length(committer_email) <= 255; project_id IS NOT NULL
Tables whose names start with import_.
import_export_upload_upload_states.project_id → projects.id — one projects, many import_export_upload_upload_states · optional · ON DELETE CASCADE · indexedimport_export_upload_upload_states.import_export_upload_upload_id → import_export_upload_uploads.id — one import_export_upload_uploads, at most one import_export_upload_upload_states · required · ON DELETE CASCADE · indexedimport_export_upload_upload_states.namespace_id → namespaces.id — one namespaces, many import_export_upload_upload_states · optional · ON DELETE CASCADE · indexedimport_export_uploads.user_id → users.id — one users, many import_export_uploads · optional · ON DELETE SET NULL · indexedimport_export_uploads.group_id → namespaces.id — one namespaces, many import_export_uploads · optional · ON DELETE CASCADE · indexedimport_export_uploads.project_id → projects.id — one projects, many import_export_uploads · optional · ON DELETE CASCADE · indexedimport_failures.group_id → namespaces.id — one namespaces, many import_failures · optional · ON DELETE CASCADE · indexedimport_failures.organization_id → organizations.id — one organizations, many import_failures · optional · ON DELETE CASCADE · indexedimport_failures.user_id → users.id — one users, many import_failures · optional · ON DELETE CASCADE · indexedimport_failures.project_id → projects.id — one projects, many import_failures · optional · ON DELETE CASCADE · indexedimport_offline_configurations.organization_id → organizations.id — one organizations, many import_offline_configurations · required · ON DELETE CASCADE · indexedimport_offline_configurations.bulk_import_id → bulk_imports.id — one bulk_imports, many import_offline_configurations · optional · ON DELETE CASCADE · indexedimport_offline_configurations.offline_export_id → import_offline_exports.id — one import_offline_exports, many import_offline_configurations · optional · ON DELETE CASCADE · indexedimport_offline_exports.organization_id → organizations.id — one organizations, many import_offline_exports · required · ON DELETE CASCADE · indexedimport_offline_exports.user_id → users.id — one users, many import_offline_exports · required · ON DELETE CASCADE · indexedimport_placeholder_memberships.namespace_id → namespaces.id — one namespaces, many import_placeholder_memberships · required · ON DELETE CASCADE · indexedimport_placeholder_memberships.group_id → namespaces.id — one namespaces, many import_placeholder_memberships · optional · ON DELETE CASCADE · indexedimport_placeholder_memberships.project_id → projects.id — one projects, many import_placeholder_memberships · optional · ON DELETE CASCADE · indexedimport_placeholder_memberships.source_user_id → import_source_users.id — one import_source_users, many import_placeholder_memberships · required · ON DELETE CASCADE · indexedimport_placeholder_user_details.placeholder_user_id → users.id — one users, many import_placeholder_user_details · required · ON DELETE CASCADE · indexedimport_placeholder_user_details.organization_id → organizations.id — one organizations, many import_placeholder_user_details · required · ON DELETE CASCADE · indexedimport_placeholder_user_details.namespace_id → namespaces.id — one namespaces, many import_placeholder_user_details · optional · ON DELETE SET NULL · indexedimport_source_user_placeholder_references.namespace_id → namespaces.id — one namespaces, many import_source_user_placeholder_references · required · ON DELETE CASCADE · indexedimport_source_user_placeholder_references.source_user_id → import_source_users.id — one import_source_users, many import_source_user_placeholder_references · required · ON DELETE CASCADE · indexedimport_source_users.reassigned_by_user_id → users.id — one users, many import_source_users · optional · ON DELETE SET NULL · indexedimport_source_users.reassign_to_user_id → users.id — one users, many import_source_users · optional · ON DELETE SET NULL · indexedimport_source_users.placeholder_user_id → users.id — one users, many import_source_users · optional · ON DELETE SET NULL · indexedimport_source_users.namespace_id → namespaces.id — one namespaces, many import_source_users · required · ON DELETE CASCADE · indexedproject_id — import_export_upload_upload_states.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id — import_export_upload_upload_states.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
import_export_upload_upload_id | bigint | NOT NULL | import_export_upload_uploads.id 1:1 · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 |
Table checks: num_nonnulls(namespace_id, project_id) = 1; char_length(verification_failure) <= 255
model_type — model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.model_type, model_id — model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | nextval(CAST('uploads_id_seq' AS regclass)) | ||
size | bigint | NOT NULL | |||
model_id | bigint | NOT NULL | |||
uploaded_by_user_id | bigint | null | |||
organization_id | bigint | null | |||
namespace_id | bigint | null | |||
project_id | bigint | null | |||
created_at | timestamp | null | |||
store | integer | NOT NULL | 1 | ||
version | integer | null | 1 | ||
path | text | NOT NULL | CHECK char_length(path) <= 511 | ||
checksum | text | null | CHECK char_length(checksum) <= 64 | ||
model_type PK | text | NOT NULL | |||
uploader | text | NOT NULL | |||
mount_point | text | null | |||
secret | text | null |
Table checks: char_length(path) <= 511; num_nonnulls(namespace_id, project_id) = 1; char_length(checksum) <= 64
user_id — import_export_uploads.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — import_export_uploads.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — import_export_uploads.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id, group_id, project_id — import_export_uploads has several nullable FKs (user_id, group_id, project_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
import_file | text | null | |||
export_file | text | null | |||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
remote_import_url | text | null | CHECK char_length(remote_import_url) <= 2048 | ||
user_id | bigint | null | users.id 1:N · SET NULL |
Table checks: char_length(remote_import_url) <= 2048; num_nonnulls(group_id, project_id) = 1
group_id — import_failures.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id — import_failures.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id — import_failures.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — import_failures.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id, organization_id, user_id, project_id — import_failures has several nullable FKs (group_id, organization_id, user_id, project_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
relation_index | integer | null | |||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
relation_key | varchar(64) | null | |||
exception_class | varchar(128) | null | |||
correlation_id_value | varchar(128) | null | |||
exception_message | varchar(255) | null | |||
retry_count | integer | null | |||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
source | varchar(128) | null | |||
external_identifiers | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
user_id | bigint | null | users.id 1:N · CASCADE | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE |
Table checks: num_nonnulls(group_id, organization_id, project_id) = 1
bulk_import_id — import_offline_configurations.bulk_import_id may be NULL, so the relationship to bulk_imports is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.offline_export_id — import_offline_configurations.offline_export_id may be NULL, so the relationship to import_offline_exports is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
offline_export_id | bigint | null | import_offline_exports.id 1:N · CASCADE | ||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
provider | smallint | NOT NULL | |||
bucket | text | NOT NULL | CHECK char_length(bucket) <= 256 | ||
export_prefix | text | NOT NULL | CHECK char_length(export_prefix) <= 255 | ||
object_storage_credentials | jsonb | NOT NULL | |||
bulk_import_id | bigint | null | bulk_imports.id 1:N · CASCADE | ||
entity_prefix_mapping | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
source_hostname | text | null | CHECK char_length(source_hostname) <= 255 |
Table checks: char_length(source_hostname) <= 255; char_length(bucket) <= 256; num_nonnulls(bulk_import_id, offline_export_id) = 1; char_length(export_prefix) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
has_failures | boolean | NOT NULL | FALSE | ||
source_hostname | text | null | CHECK char_length(source_hostname) <= 255 |
Table checks: char_length(source_hostname) <= 255
group_id — import_placeholder_memberships.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — import_placeholder_memberships.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id, project_id — import_placeholder_memberships has several nullable FKs (group_id, project_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
source_user_id | bigint | NOT NULL | import_source_users.id 1:N · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
expires_at | date | null | |||
access_level | smallint | NOT NULL |
namespace_id — import_placeholder_user_details.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
placeholder_user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
namespace_id | bigint | null | namespaces.id 1:N · SET NULL | ||
deletion_attempts | integer | NOT NULL | 0 | ||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
last_deletion_attempt_at | timestamptz | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
source_user_id | bigint | NOT NULL | import_source_users.id 1:N · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
numeric_key | bigint | null | |||
created_at | timestamptz | NOT NULL | |||
model | text | NOT NULL | CHECK char_length(model) <= 150 | ||
user_reference_column | text | NOT NULL | CHECK char_length(user_reference_column) <= 50 | ||
composite_key | jsonb | null | |||
alias_version | smallint | NOT NULL |
Table checks: char_length(user_reference_column) <= 50; char_length(model) <= 150
reassigned_by_user_id — import_source_users.reassigned_by_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.reassign_to_user_id — import_source_users.reassign_to_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.placeholder_user_id — import_source_users.placeholder_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.reassigned_by_user_id, reassign_to_user_id, placeholder_user_id — import_source_users has several nullable FKs (reassigned_by_user_id, reassign_to_user_id, placeholder_user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
placeholder_user_id | bigint | null | users.id 1:N · SET NULL | ||
reassign_to_user_id | bigint | null | users.id 1:N · SET NULL | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
source_username | text | null | CHECK char_length(source_username) <= 255 | ||
source_name | text | null | CHECK char_length(source_name) <= 255 | ||
source_user_identifier | text | NOT NULL | CHECK char_length(source_user_identifier) <= 255 | ||
source_hostname | text | NOT NULL | CHECK char_length(source_hostname) <= 255 | ||
import_type | text | NOT NULL | CHECK char_length(import_type) <= 255 | ||
reassigned_by_user_id | bigint | null | users.id 1:N · SET NULL | ||
reassignment_error | text | null | CHECK char_length(reassignment_error) <= 255 | ||
reassignment_token | text | null | CHECK char_length(reassignment_token) <= 32 |
Table checks: char_length(reassignment_error) <= 255; char_length(import_type) <= 255; char_length(source_username) <= 255; char_length(source_name) <= 255; char_length(source_user_identifier) <= 255; char_length(reassignment_token) <= 32; char_length(source_hostname) <= 255
Tables whose names start with issuable_.
issuable_metric_image_upload_states.namespace_id → namespaces.id — one namespaces, many issuable_metric_image_upload_states · required · ON DELETE CASCADE · indexedissuable_metric_image_upload_states.issuable_metric_image_upload_id → issuable_metric_image_uploads.id — one issuable_metric_image_uploads, at most one issuable_metric_image_upload_states · required · ON DELETE CASCADE · indexedissuable_metric_images.namespace_id → namespaces.id — one namespaces, many issuable_metric_images · optional · ON DELETE CASCADE · indexedissuable_metric_images.issue_id → issues.id — one issues, many issuable_metric_images · required · ON DELETE CASCADE · indexedissuable_resource_links.namespace_id → namespaces.id — one namespaces, many issuable_resource_links · optional · ON DELETE CASCADE · indexedissuable_resource_links.issue_id → issues.id — one issues, many issuable_resource_links · required · ON DELETE CASCADE · indexedissuable_severities.namespace_id → namespaces.id — one namespaces, many issuable_severities · optional · ON DELETE CASCADE · indexedissuable_severities.issue_id → issues.id — one issues, at most one issuable_severities · required · ON DELETE CASCADE · indexedissuable_slas.namespace_id → namespaces.id — one namespaces, many issuable_slas · optional · ON DELETE CASCADE · indexedissuable_slas.issue_id → issues.id — one issues, at most one issuable_slas · required · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
issuable_metric_image_upload_id | bigint | NOT NULL | issuable_metric_image_uploads.id 1:1 · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 |
Table checks: char_length(verification_failure) <= 255
model_type — model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.model_type, model_id — model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | nextval(CAST('uploads_id_seq' AS regclass)) | ||
size | bigint | NOT NULL | |||
model_id | bigint | NOT NULL | |||
uploaded_by_user_id | bigint | null | |||
organization_id | bigint | null | |||
namespace_id | bigint | null | CHECK namespace_id IS NOT NULL | ||
project_id | bigint | null | |||
created_at | timestamp | null | |||
store | integer | NOT NULL | 1 | ||
version | integer | null | 1 | ||
path | text | NOT NULL | CHECK char_length(path) <= 511 | ||
checksum | text | null | CHECK char_length(checksum) <= 64 | ||
model_type PK | text | NOT NULL | |||
uploader | text | NOT NULL | |||
mount_point | text | null | |||
secret | text | null |
Table checks: char_length(path) <= 511; namespace_id IS NOT NULL; char_length(checksum) <= 64
namespace_id — issuable_metric_images.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
issue_id | bigint | NOT NULL | issues.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | smallint | null | |||
file | text | NOT NULL | CHECK char_length(file) <= 255 | ||
url | text | null | CHECK char_length(url) <= 255 | ||
url_text | text | null | CHECK char_length(url_text) <= 128 | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL; char_length(url_text) <= 128; char_length(url) <= 255; char_length(file) <= 255
namespace_id — issuable_resource_links.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
issue_id | bigint | NOT NULL | issues.id 1:N · CASCADE | ||
link_text | text | null | CHECK char_length(link_text) <= 255 | ||
link | text | NOT NULL | CHECK char_length(link) <= 2200 | ||
link_type | smallint | NOT NULL | 0 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
is_unique | boolean | null | |||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: char_length(link) <= 2200; namespace_id IS NOT NULL; char_length(link_text) <= 255
namespace_id — issuable_severities.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
issue_id | bigint | NOT NULL | issues.id 1:1 · CASCADE | ||
severity | smallint | NOT NULL | 0 | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL
namespace_id — issuable_slas.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
issue_id | bigint | NOT NULL | issues.id 1:1 · CASCADE | ||
due_at | timestamptz | NOT NULL | |||
label_applied | boolean | NOT NULL | FALSE | ||
issuable_closed | boolean | NOT NULL | FALSE | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL
Tables whose names start with jira_.
jira_connect_installations.organization_id → organizations.id — one organizations, many jira_connect_installations · optional · ON DELETE CASCADE · indexedjira_connect_subscriptions.namespace_id → namespaces.id — one namespaces, many jira_connect_subscriptions · required · ON DELETE CASCADE · indexedjira_connect_subscriptions.jira_connect_installation_id → jira_connect_installations.id — one jira_connect_installations, many jira_connect_subscriptions · required · ON DELETE CASCADE · indexedjira_imports.project_id → projects.id — one projects, many jira_imports · required · ON DELETE CASCADE · indexedjira_imports.label_id → labels.id — one labels, many jira_imports · optional · ON DELETE SET NULL · indexedjira_imports.user_id → users.id — one users, many jira_imports · optional · ON DELETE SET NULL · indexedjira_tracker_data.organization_id → organizations.id — one organizations, many jira_tracker_data · optional · ON DELETE CASCADE · indexedjira_tracker_data.group_id → namespaces.id — one namespaces, many jira_tracker_data · optional · ON DELETE CASCADE · indexedjira_tracker_data.integration_id → integrations.id — one integrations, many jira_tracker_data · optional · ON DELETE CASCADE · indexedjira_tracker_data.project_id → projects.id — one projects, many jira_tracker_data · optional · ON DELETE CASCADE · indexedorganization_id — jira_connect_installations.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
client_key | varchar | null | |||
encrypted_shared_secret | varchar | null | |||
encrypted_shared_secret_iv | varchar | null | |||
base_url | varchar | null | |||
instance_url | text | null | CHECK char_length(instance_url) <= 255 | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE | CHECK organization_id IS NOT NULL | |
display_url | text | null | CHECK char_length(display_url) <= 255 | ||
cloud_id | text | null | CHECK char_length(cloud_id) <= 255 | ||
jira_api_base_url | text | null | CHECK char_length(jira_api_base_url) <= 512 | ||
encrypted_forge_system_token | text | null | CHECK char_length(encrypted_forge_system_token) <= 8192 | ||
encrypted_forge_system_token_iv | text | null | CHECK char_length(encrypted_forge_system_token_iv) <= 255 |
Table checks: char_length(cloud_id) <= 255; char_length(jira_api_base_url) <= 512; char_length(instance_url) <= 255; char_length(encrypted_forge_system_token) <= 8192; organization_id IS NOT NULL; char_length(encrypted_forge_system_token_iv) <= 255; char_length(display_url) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
jira_connect_installation_id | bigint | NOT NULL | jira_connect_installations.id 1:N · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
label_id — jira_imports.label_id may be NULL, so the relationship to labels is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id — jira_imports.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.label_id, user_id — jira_imports has several nullable FKs (label_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
user_id | bigint | null | users.id 1:N · SET NULL | ||
label_id | bigint | null | labels.id 1:N · SET NULL | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
finished_at | timestamptz | null | |||
jira_project_xid | bigint | NOT NULL | |||
total_issue_count | integer | NOT NULL | 0 | ||
imported_issues_count | integer | NOT NULL | 0 | ||
failed_to_import_count | integer | NOT NULL | 0 | ||
status | smallint | NOT NULL | 0 | ||
jid | varchar(255) | null | |||
jira_project_key | varchar(255) | NOT NULL | |||
jira_project_name | varchar(255) | NOT NULL | |||
scheduled_at | timestamptz | null | |||
error_message | text | null | CHECK char_length(error_message) <= 1000 |
Table checks: char_length(error_message) <= 1000
organization_id — jira_tracker_data.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — jira_tracker_data.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.integration_id — jira_tracker_data.integration_id may be NULL, so the relationship to integrations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — jira_tracker_data.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id, group_id, integration_id, project_id — jira_tracker_data has several nullable FKs (organization_id, group_id, integration_id, project_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
encrypted_url | varchar | null | |||
encrypted_url_iv | varchar | null | |||
encrypted_api_url | varchar | null | |||
encrypted_api_url_iv | varchar | null | |||
encrypted_username | varchar | null | |||
encrypted_username_iv | varchar | null | |||
encrypted_password | varchar | null | |||
encrypted_password_iv | varchar | null | |||
jira_issue_transition_id | varchar | null | |||
project_key | text | null | CHECK char_length(project_key) <= 255 | ||
issues_enabled | boolean | NOT NULL | FALSE | ||
deployment_type | smallint | NOT NULL | 0 | ||
vulnerabilities_issuetype | text | null | CHECK char_length(vulnerabilities_issuetype) <= 255 | ||
vulnerabilities_enabled | boolean | NOT NULL | FALSE | ||
jira_issue_transition_automatic | boolean | NOT NULL | FALSE | ||
integration_id | bigint | null | integrations.id 1:N · CASCADE | ||
jira_issue_prefix | text | null | CHECK char_length(jira_issue_prefix) <= 255 | ||
jira_issue_regex | text | null | CHECK char_length(jira_issue_regex) <= 255 | ||
jira_auth_type | smallint | NOT NULL | 0 | ||
project_keys | text[] | NOT NULL | CAST('{}' AS text[]) | ||
customize_jira_issue_enabled | boolean | null | FALSE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE | ||
jira_check_enabled | boolean | NOT NULL | FALSE | ||
jira_assignee_check_enabled | boolean | NOT NULL | FALSE | ||
jira_status_check_enabled | boolean | NOT NULL | FALSE | ||
jira_exists_check_enabled | boolean | NOT NULL | FALSE | ||
jira_allowed_statuses_string | text | null | CHECK char_length(jira_allowed_statuses_string) <= 1024 |
Table checks: char_length(vulnerabilities_issuetype) <= 255; char_length(project_key) <= 255; char_length(jira_issue_prefix) <= 255; char_length(jira_issue_regex) <= 255; char_length(jira_allowed_statuses_string) <= 1024; num_nonnulls(group_id, organization_id, project_id) = 1
Tables whose names start with label_.
label_links.namespace_id → namespaces.id — one namespaces, many label_links · optional · ON DELETE CASCADE · indexedlabel_links.label_id → labels.id — one labels, many label_links · optional · ON DELETE CASCADE · indexedlabel_priorities.label_id → labels.id — one labels, many label_priorities · required · ON DELETE CASCADE · indexedlabel_priorities.project_id → projects.id — one projects, many label_priorities · required · ON DELETE CASCADE · indexednamespace_id — label_links.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.label_id — label_links.label_id may be NULL, so the relationship to labels is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.target_type — target_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.target_type, target_id — target_id points at different tables depending on target_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.namespace_id, label_id — label_links has several nullable FKs (namespace_id, label_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
label_id | bigint | null | labels.id 1:N · CASCADE | ||
target_id | bigint | null | |||
target_type | varchar | null | |||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL
created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
label_id | bigint | NOT NULL | labels.id 1:N · CASCADE | ||
priority | integer | NOT NULL | |||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL |
Tables whose names start with ldap_.
ldap_admin_role_links.organization_id → organizations.id — one organizations, many ldap_admin_role_links · optional · ON DELETE CASCADE · indexedldap_admin_role_links.member_role_id → member_roles.id — one member_roles, many ldap_admin_role_links · required · ON DELETE CASCADE · indexedldap_group_links.member_role_id → member_roles.id — one member_roles, many ldap_group_links · optional · ON DELETE SET NULL · indexedorganization_id — ldap_admin_role_links.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
member_role_id | bigint | NOT NULL | member_roles.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
provider | text | NOT NULL | CHECK char_length(provider) <= 255 | ||
cn | text | null | CHECK char_length(cn) <= 255 | ||
filter | text | null | CHECK char_length(filter) <= 255 | ||
sync_status | smallint | NOT NULL | 0 | ||
sync_started_at | timestamptz | null | |||
sync_ended_at | timestamptz | null | |||
last_successful_sync_at | timestamptz | null | |||
sync_error | text | null | CHECK char_length(sync_error) <= 255 | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE | CHECK organization_id IS NOT NULL |
Table checks: char_length(sync_error) <= 255; char_length(filter) <= 255; char_length(cn) <= 255; organization_id IS NOT NULL; char_length(provider) <= 255
member_role_id — ldap_group_links.member_role_id may be NULL, so the relationship to member_roles is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
cn | varchar | null | |||
group_access | integer | NOT NULL | |||
group_id | bigint | NOT NULL | |||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
provider | varchar | null | |||
filter | varchar | null | |||
member_role_id | bigint | null | member_roles.id 1:N · SET NULL |
Tables whose names start with lfs_.
lfs_file_locks.user_id → users.id — one users, many lfs_file_locks · required · ON DELETE CASCADE · indexedlfs_file_locks.project_id → projects.id — one projects, many lfs_file_locks · required · ON DELETE CASCADE · indexedlfs_objects_projects.project_id → projects.id — one projects, many lfs_objects_projects · required · ON DELETE CASCADE · indexedproject_id, user_id — lfs_file_locks looks like a many-to-many link table but has no unique constraint across (project_id, user_id). The same pair can be inserted twice; every join through it will double-count.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
created_at | timestamp | NOT NULL | |||
path | varchar(511) | null |
lfs_object_states references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
lfs_object_id PK | bigint | NOT NULL | |||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | null | |||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 |
Table checks: char_length(verification_failure) <= 255
created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.lfs_objects references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
oid | varchar | NOT NULL | |||
size | bigint | NOT NULL | |||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
file | varchar | null | |||
file_store | integer | null | 1 | CHECK file_store IS NOT NULL |
Table checks: file_store IS NOT NULL
created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
lfs_object_id | bigint | null | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
repository_type | smallint | null | |||
oid | text | null | CHECK char_length(oid) <= 255 |
Table checks: char_length(oid) <= 255
Tables whose names start with member_.
member_approvals.member_namespace_id → namespaces.id — one namespaces, many member_approvals · required · ON DELETE CASCADE · indexedmember_approvals.member_role_id → member_roles.id — one member_roles, many member_approvals · optional · ON DELETE SET NULL · indexedmember_approvals.member_id → members.id — one members, many member_approvals · optional · ON DELETE CASCADE · indexedmember_roles.organization_id → organizations.id — one organizations, many member_roles · optional · ON DELETE CASCADE · indexedmember_roles.namespace_id → namespaces.id — one namespaces, many member_roles · optional · ON DELETE CASCADE · indexedmember_role_id — member_approvals.member_role_id may be NULL, so the relationship to member_roles is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.member_id — member_approvals.member_id may be NULL, so the relationship to members is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.member_role_id, member_id — member_approvals has several nullable FKs (member_role_id, member_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
reviewed_at | timestamptz | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
member_id | bigint | null | members.id 1:N · CASCADE | ||
member_namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
requested_by_id | bigint | null | |||
reviewed_by_id | bigint | null | |||
new_access_level | integer | NOT NULL | |||
old_access_level | integer | null | |||
status | smallint | NOT NULL | 0 | ||
user_id | bigint | NOT NULL | |||
member_role_id | bigint | null | member_roles.id 1:N · SET NULL | ||
metadata | jsonb | NOT NULL | CAST('{}' AS jsonb) |
organization_id — member_roles.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id — member_roles.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
base_access_level | integer | null | |||
name | text | NOT NULL | CAST('Custom' AS text) | CHECK char_length(name) <= 255 | |
description | text | null | CHECK char_length(description) <= 255 | ||
occupies_seat | boolean | NOT NULL | FALSE | ||
permissions | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE |
Table checks: char_length(description) <= 255; char_length(name) <= 255; num_nonnulls(namespace_id, organization_id) = 1
Tables whose names start with members_.
members.member_namespace_id → namespaces.id — one namespaces, many members · optional · ON DELETE CASCADE · indexedmembers.member_role_id → member_roles.id — one member_roles, many members · optional · ON DELETE SET NULL · indexedmembers_deletion_schedules.scheduled_by_id → users.id — one users, many members_deletion_schedules · required · ON DELETE CASCADE · indexedmembers_deletion_schedules.user_id → users.id — one users, many members_deletion_schedules · required · ON DELETE CASCADE · indexedmembers_deletion_schedules.namespace_id → namespaces.id — one namespaces, many members_deletion_schedules · required · ON DELETE CASCADE · indexedmember_namespace_id — members.member_namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.member_role_id — members.member_role_id may be NULL, so the relationship to member_roles is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.source_type — source_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.type — type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.source_type, source_id — source_id points at different tables depending on source_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.member_namespace_id, member_role_id — members has several nullable FKs (member_namespace_id, member_role_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.invite_accepted_at — invite_accepted_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.requested_at — requested_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
access_level | integer | NOT NULL | |||
source_id | bigint | NOT NULL | |||
source_type | varchar | NOT NULL | |||
user_id | bigint | null | |||
notification_level | integer | NOT NULL | |||
type | varchar | null | |||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
created_by_id | bigint | null | |||
invite_email | varchar | null | |||
invite_token | varchar | null | |||
invite_accepted_at | timestamp | null | |||
requested_at | timestamp | null | |||
expires_at | date | null | |||
ldap | boolean | NOT NULL | FALSE | ||
override | boolean | NOT NULL | FALSE | ||
state | smallint | null | 0 | ||
invite_email_success | boolean | NOT NULL | TRUE | ||
member_namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK member_namespace_id IS NOT NULL | |
member_role_id | bigint | null | member_roles.id 1:N · SET NULL | ||
expiry_notified_at | timestamptz | null | |||
request_accepted_at | timestamptz | null |
Table checks: member_namespace_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
scheduled_by_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
ip_address | inet | null |
Tables whose names start with ml_.
ml_candidate_metadata.project_id → projects.id — one projects, many ml_candidate_metadata · optional · ON DELETE CASCADE · indexedml_candidate_metadata.candidate_id → ml_candidates.id — one ml_candidates, many ml_candidate_metadata · required · ON DELETE CASCADE · indexedml_candidate_metrics.project_id → projects.id — one projects, many ml_candidate_metrics · optional · ON DELETE CASCADE · indexedml_candidate_metrics.candidate_id → ml_candidates.id — one ml_candidates, many ml_candidate_metrics · optional · ON DELETE CASCADE · indexedml_candidate_params.project_id → projects.id — one projects, many ml_candidate_params · optional · ON DELETE CASCADE · indexedml_candidate_params.candidate_id → ml_candidates.id — one ml_candidates, many ml_candidate_params · optional · ON DELETE CASCADE · indexedml_candidates.project_id → projects.id — one projects, many ml_candidates · optional · ON DELETE CASCADE · indexedml_candidates.experiment_id → ml_experiments.id — one ml_experiments, many ml_candidates · required · ON DELETE CASCADE · indexedml_candidates.package_id → packages_packages.id — one packages_packages, many ml_candidates · optional · ON DELETE SET NULL · indexedml_candidates.model_version_id → ml_model_versions.id — one ml_model_versions, at most one ml_candidates · optional · ON DELETE CASCADE · indexedml_candidates.user_id → users.id — one users, many ml_candidates · optional · ON DELETE SET NULL · indexedml_experiment_metadata.project_id → projects.id — one projects, many ml_experiment_metadata · optional · ON DELETE CASCADE · indexedml_experiment_metadata.experiment_id → ml_experiments.id — one ml_experiments, many ml_experiment_metadata · required · ON DELETE CASCADE · indexedml_experiments.project_id → projects.id — one projects, many ml_experiments · required · ON DELETE CASCADE · indexedml_experiments.user_id → users.id — one users, many ml_experiments · optional · ON DELETE SET NULL · indexedml_experiments.model_id → ml_models.id — one ml_models, many ml_experiments · optional · ON DELETE CASCADE · indexedml_model_metadata.project_id → projects.id — one projects, many ml_model_metadata · optional · ON DELETE CASCADE · indexedml_model_metadata.model_id → ml_models.id — one ml_models, many ml_model_metadata · required · ON DELETE CASCADE · indexedml_model_version_metadata.model_version_id → ml_model_versions.id — one ml_model_versions, many ml_model_version_metadata · required · ON DELETE CASCADE · indexedml_model_version_metadata.project_id → projects.id — one projects, many ml_model_version_metadata · required · ON DELETE CASCADE · indexedml_model_versions.package_id → packages_packages.id — one packages_packages, many ml_model_versions · optional · ON DELETE SET NULL · indexedml_model_versions.model_id → ml_models.id — one ml_models, many ml_model_versions · required · ON DELETE CASCADE · indexedml_model_versions.project_id → projects.id — one projects, many ml_model_versions · required · ON DELETE CASCADE · indexedml_models.project_id → projects.id — one projects, many ml_models · required · ON DELETE CASCADE · indexedml_models.user_id → users.id — one users, many ml_models · optional · ON DELETE SET NULL · indexedproject_id — ml_candidate_metadata.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
candidate_id | bigint | NOT NULL | ml_candidates.id 1:N · CASCADE | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
value | text | NOT NULL | CHECK char_length(value) <= 5000 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: char_length(name) <= 255; char_length(value) <= 5000; project_id IS NOT NULL
project_id — ml_candidate_metrics.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.candidate_id — ml_candidate_metrics.candidate_id may be NULL, so the relationship to ml_candidates is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id, candidate_id — ml_candidate_metrics has several nullable FKs (project_id, candidate_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
candidate_id | bigint | null | ml_candidates.id 1:N · CASCADE | CHECK candidate_id IS NOT NULL | |
value | double precision | null | |||
step | integer | null | |||
is_nan | bytea | null | |||
name | text | NOT NULL | CHECK char_length(name) <= 250 | ||
tracked_at | bigint | null | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: char_length(name) <= 250; candidate_id IS NOT NULL; project_id IS NOT NULL
project_id — ml_candidate_params.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.candidate_id — ml_candidate_params.candidate_id may be NULL, so the relationship to ml_candidates is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id, candidate_id — ml_candidate_params has several nullable FKs (project_id, candidate_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
candidate_id | bigint | null | ml_candidates.id 1:N · CASCADE | CHECK candidate_id IS NOT NULL | |
name | text | NOT NULL | CHECK char_length(name) <= 250 | ||
value | text | NOT NULL | CHECK char_length(value) <= 250 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: char_length(name) <= 250; char_length(value) <= 250; candidate_id IS NOT NULL; project_id IS NOT NULL
project_id — ml_candidates.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.package_id — ml_candidates.package_id may be NULL, so the relationship to packages_packages is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.model_version_id — ml_candidates.model_version_id may be NULL, so the relationship to ml_model_versions is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id — ml_candidates.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id, package_id, model_version_id, user_id — ml_candidates has several nullable FKs (project_id, package_id, model_version_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
experiment_id | bigint | NOT NULL | ml_experiments.id 1:N · CASCADE | ||
user_id | bigint | null | users.id 1:N · SET NULL | ||
start_time | bigint | null | |||
end_time | bigint | null | |||
status | smallint | NOT NULL | 0 | ||
name | text | null | CHECK char_length(name) <= 255 | ||
package_id | bigint | null | packages_packages.id 1:N · SET NULL | ||
eid | uuid | null | CHECK eid IS NOT NULL | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL | |
internal_id | bigint | null | |||
ci_build_id | bigint | null | |||
model_version_id | bigint | null | ml_model_versions.id 1:1 · CASCADE |
Table checks: project_id IS NOT NULL; char_length(name) <= 255; eid IS NOT NULL
project_id — ml_experiment_metadata.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
experiment_id | bigint | NOT NULL | ml_experiments.id 1:N · CASCADE | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
value | text | NOT NULL | CHECK char_length(value) <= 5000 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: char_length(name) <= 255; char_length(value) <= 5000; project_id IS NOT NULL
user_id — ml_experiments.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.model_id — ml_experiments.model_id may be NULL, so the relationship to ml_models is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id, model_id — ml_experiments has several nullable FKs (user_id, model_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
iid | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
user_id | bigint | null | users.id 1:N · SET NULL | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
deleted_on | timestamptz | null | |||
model_id | bigint | null | ml_models.id 1:N · CASCADE |
Table checks: char_length(name) <= 255
project_id — ml_model_metadata.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
model_id | bigint | NOT NULL | ml_models.id 1:N · CASCADE | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
value | text | NOT NULL | CHECK char_length(value) <= 5000 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: char_length(value) <= 5000; char_length(name) <= 255; project_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
model_version_id | bigint | NOT NULL | ml_model_versions.id 1:N · CASCADE | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
value | text | NOT NULL | CHECK char_length(value) <= 5000 |
Table checks: char_length(name) <= 255; char_length(value) <= 5000
package_id — ml_model_versions.package_id may be NULL, so the relationship to packages_packages is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
model_id | bigint | NOT NULL | ml_models.id 1:N · CASCADE | ||
package_id | bigint | null | packages_packages.id 1:N · SET NULL | ||
version | text | NOT NULL | CHECK char_length(version) <= 255 | ||
description | text | null | CHECK char_length(description) <= 10000 | ||
semver_major | integer | null | |||
semver_minor | integer | null | |||
semver_patch | integer | null | |||
semver_prerelease | text | null | CHECK char_length(semver_prerelease) <= 255 | ||
cached_markdown_version | integer | null | |||
description_html | text | null | CHECK char_length(description_html) <= 50000 |
Table checks: char_length(semver_prerelease) <= 255; char_length(version) <= 255; char_length(description_html) <= 50000; char_length(description) <= 10000
user_id — ml_models.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
description | text | null | CHECK char_length(description) <= 10000 | ||
user_id | bigint | null | users.id 1:N · SET NULL | ||
cached_markdown_version | integer | null | |||
description_html | text | null | CHECK char_length(description_html) <= 50000 |
Table checks: char_length(name) <= 255; char_length(description_html) <= 50000; char_length(description) <= 10000
Tables whose names start with note_.
note_diff_files.namespace_id → namespaces.id — one namespaces, many note_diff_files · optional · ON DELETE CASCADE · indexednote_diff_files.diff_note_id → notes.id — one notes, at most one note_diff_files · required · ON DELETE CASCADE · indexednote_duo_metadata.workflow_id → duo_workflows_workflows.id — one duo_workflows_workflows, many note_duo_metadata · required · ON DELETE CASCADE · indexednote_duo_metadata.namespace_id → namespaces.id — one namespaces, many note_duo_metadata · required · ON DELETE CASCADE · indexednote_duo_metadata.note_id → notes.id — one notes, at most one note_duo_metadata · required · ON DELETE CASCADE · indexednote_metadata.namespace_id → namespaces.id — one namespaces, many note_metadata · optional · ON DELETE CASCADE · indexednote_metadata.note_id → notes.id — one notes, at most one note_metadata · required · ON DELETE CASCADE · indexednamespace_id — note_diff_files.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.a_mode — a_mode is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.b_mode — b_mode is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
diff | text | NOT NULL | |||
new_file | boolean | NOT NULL | |||
renamed_file | boolean | NOT NULL | |||
deleted_file | boolean | NOT NULL | |||
a_mode | varchar | NOT NULL | |||
b_mode | varchar | NOT NULL | |||
new_path | text | NOT NULL | |||
old_path | text | NOT NULL | |||
diff_note_id | bigint | NOT NULL | notes.id 1:1 · CASCADE | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
note_id PK | bigint | NOT NULL | notes.id 1:1 · CASCADE | ||
workflow_id | bigint | NOT NULL | duo_workflows_workflows.id 1:N · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | null | |||
updated_at | timestamptz | null |
namespace_id — note_metadata.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
note_id PK | bigint | NOT NULL | notes.id 1:1 · CASCADE | ||
email_participant | text | null | CHECK char_length(email_participant) <= 255 | ||
created_at | timestamptz | null | |||
updated_at | timestamptz | null | |||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: char_length(email_participant) <= 255; namespace_id IS NOT NULL
Tables whose names start with oauth_.
oauth_access_grants.organization_id → organizations.id — one organizations, many oauth_access_grants · required · ON DELETE CASCADE · indexedoauth_access_tokens.organization_id → organizations.id — one organizations, many oauth_access_tokens · required · ON DELETE CASCADE · indexedoauth_applications.organization_id → organizations.id — one organizations, many oauth_applications · optional · ON DELETE CASCADE · indexedoauth_consents.user_id → users.id — one users, many oauth_consents · required · ON DELETE CASCADE · indexedoauth_consents.client_id → oauth_applications.uid — one oauth_applications, many oauth_consents · required · ON DELETE CASCADE · indexedoauth_device_grants.application_id → oauth_applications.id — one oauth_applications, many oauth_device_grants · required · ON DELETE CASCADE · indexedoauth_device_grants.organization_id → organizations.id — one organizations, many oauth_device_grants · required · ON DELETE CASCADE · indexedoauth_openid_requests.organization_id → organizations.id — one organizations, many oauth_openid_requests · required · ON DELETE CASCADE · indexedoauth_openid_requests.access_grant_id → oauth_access_grants.id — one oauth_access_grants, many oauth_openid_requests · required · ON DELETE CASCADE · indexedcreated_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.revoked_at — revoked_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
resource_owner_id | bigint | NOT NULL | |||
application_id | bigint | NOT NULL | |||
token | varchar | NOT NULL | |||
expires_in | integer | NOT NULL | |||
redirect_uri | text | NOT NULL | |||
created_at | timestamp | NOT NULL | |||
revoked_at | timestamp | null | |||
scopes | varchar | null | |||
code_challenge | text | null | CHECK char_length(code_challenge) <= 128 | ||
code_challenge_method | text | null | CHECK char_length(code_challenge_method) <= 5 | ||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE |
Table checks: char_length(code_challenge) <= 128; char_length(code_challenge_method) <= 5
revoked_at — revoked_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
resource_owner_id | bigint | null | |||
application_id | bigint | null | |||
token | varchar | NOT NULL | |||
refresh_token | varchar | null | |||
expires_in | integer | null | 7200 | CHECK expires_in IS NOT NULL | |
revoked_at | timestamp | null | |||
created_at | timestamp | NOT NULL | |||
scopes | varchar | null | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE |
Table checks: expires_in IS NOT NULL
organization_id — oauth_applications.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.owner_type — owner_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.owner_type, owner_id — owner_id points at different tables depending on owner_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
name | varchar | NOT NULL | |||
uid | varchar | NOT NULL | |||
secret | varchar | NOT NULL | |||
redirect_uri | text | NOT NULL | |||
scopes | text | NOT NULL | CAST('' AS text) | CHECK char_length(scopes) <= 2048 | |
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
owner_id | bigint | null | |||
owner_type | varchar | null | |||
trusted | boolean | NOT NULL | FALSE | ||
confidential | boolean | NOT NULL | TRUE | ||
expire_access_tokens | boolean | NOT NULL | FALSE | ||
ropc_enabled | boolean | NOT NULL | TRUE | ||
dynamic | boolean | NOT NULL | FALSE | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE | CHECK organization_id IS NOT NULL | |
device_code_enabled | boolean | NOT NULL | TRUE |
Table checks: char_length(scopes) <= 2048; organization_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
status | smallint | NOT NULL | |||
client_id | text | NOT NULL | oauth_applications.uid 1:N · CASCADE | ||
consent_challenge | text | NOT NULL | CHECK char_length(consent_challenge) <= 100 | ||
requested_scopes | text[] | null | CHECK cardinality(requested_scopes) <= 50 | ||
granted_scopes | text[] | null | CHECK cardinality(granted_scopes) <= 50 |
Table checks: char_length(consent_challenge) <= 100; cardinality(granted_scopes) <= 50; cardinality(requested_scopes) <= 50
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
resource_owner_id | bigint | null | |||
application_id | bigint | NOT NULL | oauth_applications.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
last_polling_at | timestamptz | null | |||
expires_in | integer | NOT NULL | |||
scopes | text | NOT NULL | CAST('' AS text) | CHECK char_length(scopes) <= 255 | |
device_code | text | NOT NULL | CHECK char_length(device_code) <= 255 | ||
user_code | text | null | CHECK char_length(user_code) <= 255 | ||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE |
Table checks: char_length(device_code) <= 255; char_length(scopes) <= 255; char_length(user_code) <= 255
access_grant_id, organization_id — oauth_openid_requests looks like a many-to-many link table but has no unique constraint across (access_grant_id, organization_id). The same pair can be inserted twice; every join through it will double-count.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
access_grant_id | bigint | NOT NULL | oauth_access_grants.id 1:N · CASCADE | ||
nonce | varchar | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE |
Tables whose names start with observability_.
observability_group_o11y_settings.group_id → namespaces.id — one namespaces, many observability_group_o11y_settings · required · ON DELETE CASCADE · indexedobservability_logs_issues_connections.project_id → projects.id — one projects, many observability_logs_issues_connections · required · ON DELETE CASCADE · indexedobservability_logs_issues_connections.issue_id → issues.id — one issues, many observability_logs_issues_connections · required · ON DELETE CASCADE · indexedobservability_metrics_issues_connections.project_id → projects.id — one projects, many observability_metrics_issues_connections · optional · ON DELETE CASCADE · indexedobservability_metrics_issues_connections.namespace_id → namespaces.id — one namespaces, many observability_metrics_issues_connections · required · ON DELETE CASCADE · indexedobservability_metrics_issues_connections.issue_id → issues.id — one issues, many observability_metrics_issues_connections · required · ON DELETE CASCADE · indexedobservability_traces_issues_connections.project_id → projects.id — one projects, many observability_traces_issues_connections · required · ON DELETE CASCADE · indexedobservability_traces_issues_connections.issue_id → issues.id — one issues, many observability_traces_issues_connections · required · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
o11y_service_url | text | NOT NULL | CHECK char_length(o11y_service_url) <= 255 | ||
o11y_service_user_email | text | NOT NULL | CHECK char_length(o11y_service_user_email) <= 255 | ||
o11y_service_password | jsonb | NOT NULL | |||
o11y_service_post_message_encryption_key | jsonb | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
o11y_otel_url | text | null | CHECK char_length(o11y_otel_url) <= 255 | ||
o11y_mcp_url | text | null | CHECK char_length(o11y_mcp_url) <= 255 |
Table checks: char_length(o11y_service_user_email) <= 255; char_length(o11y_mcp_url) <= 255; char_length(o11y_otel_url) <= 255; char_length(o11y_service_url) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
issue_id | bigint | NOT NULL | issues.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
log_timestamp | timestamptz | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
severity_number | smallint | NOT NULL | |||
service_name | text | NOT NULL | CHECK char_length(service_name) <= 500 | ||
trace_identifier | text | NOT NULL | CHECK char_length(trace_identifier) <= 128 | ||
log_fingerprint | text | NOT NULL | CHECK char_length(log_fingerprint) <= 128 |
Table checks: char_length(trace_identifier) <= 128; char_length(log_fingerprint) <= 128; char_length(service_name) <= 500
project_id — observability_metrics_issues_connections.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
issue_id | bigint | NOT NULL | issues.id 1:N · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
metric_type | smallint | NOT NULL | |||
metric_name | text | NOT NULL | CHECK char_length(metric_name) <= 500 | ||
project_id | bigint | null | projects.id 1:N · CASCADE |
Table checks: char_length(metric_name) <= 500
issue_id, project_id — observability_traces_issues_connections looks like a many-to-many link table but has no unique constraint across (issue_id, project_id). The same pair can be inserted twice; every join through it will double-count.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
issue_id | bigint | NOT NULL | issues.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
trace_identifier | text | NOT NULL | CHECK char_length(trace_identifier) <= 128 |
Table checks: char_length(trace_identifier) <= 128
Tables whose names start with operations_.
operations_feature_flags.project_id → projects.id — one projects, many operations_feature_flags · required · ON DELETE CASCADE · indexedoperations_feature_flags_clients.project_id → projects.id — one projects, many operations_feature_flags_clients · required · ON DELETE CASCADE · indexedoperations_feature_flags_issues.project_id → projects.id — one projects, many operations_feature_flags_issues · optional · ON DELETE CASCADE · indexedoperations_feature_flags_issues.feature_flag_id → operations_feature_flags.id — one operations_feature_flags, many operations_feature_flags_issues · required · ON DELETE CASCADE · indexedoperations_feature_flags_issues.issue_id → issues.id — one issues, many operations_feature_flags_issues · required · ON DELETE CASCADE · indexedoperations_scopes.project_id → projects.id — one projects, many operations_scopes · optional · ON DELETE CASCADE · indexedoperations_scopes.strategy_id → operations_strategies.id — one operations_strategies, many operations_scopes · required · ON DELETE CASCADE · indexedoperations_strategies.project_id → projects.id — one projects, many operations_strategies · optional · ON DELETE CASCADE · indexedoperations_strategies.feature_flag_id → operations_feature_flags.id — one operations_feature_flags, many operations_strategies · required · ON DELETE CASCADE · indexedoperations_strategies_user_lists.project_id → projects.id — one projects, many operations_strategies_user_lists · optional · ON DELETE CASCADE · indexedoperations_strategies_user_lists.strategy_id → operations_strategies.id — one operations_strategies, many operations_strategies_user_lists · required · ON DELETE CASCADE · indexedoperations_strategies_user_lists.user_list_id → operations_user_lists.id — one operations_user_lists, many operations_strategies_user_lists · required · ON DELETE CASCADE · indexedoperations_user_lists.project_id → projects.id — one projects, many operations_user_lists · required · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
active | boolean | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
name | varchar | NOT NULL | |||
description | text | null | |||
iid | integer | NOT NULL | |||
version | smallint | NOT NULL | 1 |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
token_encrypted | varchar | null | |||
last_feature_flag_updated_at | timestamptz | null |
project_id — operations_feature_flags_issues.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
feature_flag_id | bigint | NOT NULL | operations_feature_flags.id 1:N · CASCADE | ||
issue_id | bigint | NOT NULL | issues.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
project_id — operations_scopes.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
strategy_id | bigint | NOT NULL | operations_strategies.id 1:N · CASCADE | ||
environment_scope | varchar(255) | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
project_id — operations_strategies.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
feature_flag_id | bigint | NOT NULL | operations_feature_flags.id 1:N · CASCADE | ||
name | varchar(255) | NOT NULL | |||
parameters | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
project_id — operations_strategies_user_lists.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
strategy_id | bigint | NOT NULL | operations_strategies.id 1:N · CASCADE | ||
user_list_id | bigint | NOT NULL | operations_user_lists.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
iid | integer | NOT NULL | |||
name | varchar(255) | NOT NULL | |||
user_xids | text | NOT NULL | CAST('' AS text) |
Tables whose names start with organization_.
organization_cluster_agent_mappings.cluster_agent_id → cluster_agents.id — one cluster_agents, at most one organization_cluster_agent_mappings · required · ON DELETE CASCADE · indexedorganization_cluster_agent_mappings.organization_id → organizations.id — one organizations, many organization_cluster_agent_mappings · required · ON DELETE CASCADE · indexedorganization_cluster_agent_mappings.creator_id → users.id — one users, many organization_cluster_agent_mappings · optional · ON DELETE SET NULL · indexedorganization_detail_upload_states.organization_detail_upload_id → organization_detail_uploads.id — one organization_detail_uploads, at most one organization_detail_upload_states · required · ON DELETE CASCADE · indexedorganization_detail_upload_states.organization_id → organizations.id — one organizations, many organization_detail_upload_states · required · ON DELETE CASCADE · indexedorganization_details.organization_id → organizations.id — one organizations, at most one organization_details · required · ON DELETE CASCADE · indexedorganization_foundational_agent_statuses.organization_id → organizations.id — one organizations, many organization_foundational_agent_statuses · required · ON DELETE CASCADE · indexedorganization_isolations.organization_id → organizations.id — one organizations, at most one organization_isolations · required · ON DELETE CASCADE · indexedorganization_push_rules.organization_id → organizations.id — one organizations, at most one organization_push_rules · required · ON DELETE CASCADE · indexedorganization_settings.organization_id → organizations.id — one organizations, at most one organization_settings · required · ON DELETE CASCADE · indexedorganization_user_details.user_id → users.id — one users, many organization_user_details · required · ON DELETE CASCADE · indexedorganization_user_details.organization_id → organizations.id — one organizations, many organization_user_details · required · ON DELETE CASCADE · indexedorganization_users.organization_id → organizations.id — one organizations, many organization_users · required · ON DELETE RESTRICT · indexedorganization_users.user_id → users.id — one users, many organization_users · required · ON DELETE CASCADE · indexedcreator_id — organization_cluster_agent_mappings.creator_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
cluster_agent_id | bigint | NOT NULL | cluster_agents.id 1:1 · CASCADE | ||
creator_id | bigint | null | users.id 1:N · SET NULL | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
organization_detail_upload_id | bigint | NOT NULL | organization_detail_uploads.id 1:1 · CASCADE | ||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 |
Table checks: char_length(verification_failure) <= 255
model_type — model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.model_type, model_id — model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | nextval(CAST('uploads_id_seq' AS regclass)) | ||
size | bigint | NOT NULL | |||
model_id | bigint | NOT NULL | |||
uploaded_by_user_id | bigint | null | |||
organization_id | bigint | null | CHECK organization_id IS NOT NULL | ||
namespace_id | bigint | null | |||
project_id | bigint | null | |||
created_at | timestamp | null | |||
store | integer | NOT NULL | 1 | ||
version | integer | null | 1 | ||
path | text | NOT NULL | CHECK char_length(path) <= 511 | ||
checksum | text | null | CHECK char_length(checksum) <= 64 | ||
model_type PK | text | NOT NULL | |||
uploader | text | NOT NULL | |||
mount_point | text | null | |||
secret | text | null |
Table checks: char_length(path) <= 511; organization_id IS NOT NULL; char_length(checksum) <= 64
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
organization_id PK | bigint | NOT NULL | organizations.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
cached_markdown_version | integer | null | |||
description | text | null | CHECK char_length(description) <= 1024 | ||
description_html | text | null | |||
avatar | text | null | CHECK char_length(avatar) <= 255 | ||
state_metadata | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
soft_deleted_at | timestamptz | null |
Table checks: char_length(description) <= 1024; char_length(avatar) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
reference | text | NOT NULL | CHECK char_length(reference) <= 255 | ||
enabled | boolean | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
Table checks: char_length(reference) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
isolated | boolean | NOT NULL | FALSE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:1 · CASCADE | ||
max_file_size | integer | NOT NULL | 0 | ||
member_check | boolean | NOT NULL | FALSE | ||
prevent_secrets | boolean | NOT NULL | FALSE | ||
commit_committer_name_check | boolean | NOT NULL | FALSE | ||
deny_delete_tag | boolean | null | |||
reject_unsigned_commits | boolean | null | |||
commit_committer_check | boolean | null | |||
reject_non_dco_commits | boolean | null | |||
commit_message_regex | text | null | CHECK char_length(commit_message_regex) <= 511 | ||
branch_name_regex | text | null | CHECK char_length(branch_name_regex) <= 511 | ||
commit_message_negative_regex | text | null | CHECK char_length(commit_message_negative_regex) <= 2047 | ||
author_email_regex | text | null | CHECK char_length(author_email_regex) <= 511 | ||
file_name_regex | text | null | CHECK char_length(file_name_regex) <= 511 |
Table checks: char_length(author_email_regex) <= 511; char_length(branch_name_regex) <= 511; char_length(commit_message_negative_regex) <= 2047; char_length(commit_message_regex) <= 511; char_length(file_name_regex) <= 511
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
organization_id PK | bigint | NOT NULL | organizations.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
settings | jsonb | NOT NULL | CAST('{}' AS jsonb) |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
username | text | NOT NULL | CHECK char_length(username) <= 510 | ||
display_name | text | NOT NULL | CHECK char_length(display_name) <= 510 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
Table checks: char_length(display_name) <= 510; char_length(username) <= 510
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · RESTRICT | ||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
access_level | smallint | NOT NULL | 10 |
Tables whose names start with packages_.
packages_build_infos.project_id → projects.id — one projects, many packages_build_infos · optional · ON DELETE CASCADE · indexedpackages_build_infos.package_id → packages_packages.id — one packages_packages, many packages_build_infos · required · ON DELETE CASCADE · indexedpackages_cargo_metadata.project_id → projects.id — one projects, many packages_cargo_metadata · required · ON DELETE CASCADE · indexedpackages_cargo_metadata.package_id → packages_packages.id — one packages_packages, at most one packages_cargo_metadata · required · ON DELETE CASCADE · indexedpackages_cleanup_policies.project_id → projects.id — one projects, at most one packages_cleanup_policies · required · ON DELETE CASCADE · indexedpackages_composer_metadata.project_id → projects.id — one projects, many packages_composer_metadata · optional · ON DELETE CASCADE · indexedpackages_composer_metadata.package_id → packages_packages.id — one packages_packages, at most one packages_composer_metadata · required · ON DELETE CASCADE · indexedpackages_packages.creator_id → users.id — one users, many packages_packages · optional · ON DELETE SET NULL · indexedpackages_packages.project_id → projects.id — one projects, many packages_packages · required · ON DELETE CASCADE · indexedpackages_composer_packages.creator_id → users.id — one users, many packages_composer_packages · optional · ON DELETE SET NULL · indexedpackages_composer_packages.project_id → projects.id — one projects, many packages_composer_packages · required · ON DELETE CASCADE · indexedpackages_conan_file_metadata.project_id → projects.id — one projects, many packages_conan_file_metadata · optional · ON DELETE CASCADE · indexedpackages_conan_file_metadata.package_revision_id → packages_conan_package_revisions.id — one packages_conan_package_revisions, many packages_conan_file_metadata · optional · ON DELETE CASCADE · indexedpackages_conan_file_metadata.package_reference_id → packages_conan_package_references.id — one packages_conan_package_references, many packages_conan_file_metadata · optional · ON DELETE CASCADE · indexedpackages_conan_file_metadata.recipe_revision_id → packages_conan_recipe_revisions.id — one packages_conan_recipe_revisions, many packages_conan_file_metadata · optional · ON DELETE CASCADE · indexedpackages_conan_file_metadata.package_file_id → packages_package_files.id — one packages_package_files, at most one packages_conan_file_metadata · required · ON DELETE CASCADE · indexedpackages_conan_metadata.project_id → projects.id — one projects, many packages_conan_metadata · optional · ON DELETE CASCADE · indexedpackages_conan_metadata.package_id → packages_packages.id — one packages_packages, many packages_conan_metadata · required · ON DELETE CASCADE · indexedpackages_conan_package_references.package_id → packages_packages.id — one packages_packages, many packages_conan_package_references · required · ON DELETE CASCADE · indexedpackages_conan_package_references.recipe_revision_id → packages_conan_recipe_revisions.id — one packages_conan_recipe_revisions, many packages_conan_package_references · optional · ON DELETE CASCADE · indexedpackages_conan_package_references.project_id → projects.id — one projects, many packages_conan_package_references · required · ON DELETE CASCADE · indexedpackages_conan_package_revisions.project_id → projects.id — one projects, many packages_conan_package_revisions · required · ON DELETE CASCADE · indexedpackages_conan_package_revisions.package_reference_id → packages_conan_package_references.id — one packages_conan_package_references, many packages_conan_package_revisions · required · ON DELETE CASCADE · indexedpackages_conan_package_revisions.package_id → packages_packages.id — one packages_packages, many packages_conan_package_revisions · required · ON DELETE CASCADE · indexedpackages_conan_recipe_revisions.project_id → projects.id — one projects, many packages_conan_recipe_revisions · required · ON DELETE CASCADE · indexedpackages_conan_recipe_revisions.package_id → packages_packages.id — one packages_packages, many packages_conan_recipe_revisions · required · ON DELETE CASCADE · indexedpackages_debian_file_metadata.project_id → projects.id — one projects, many packages_debian_file_metadata · optional · ON DELETE CASCADE · indexedpackages_debian_file_metadata.package_file_id → packages_package_files.id — one packages_package_files, at most one packages_debian_file_metadata · required · ON DELETE CASCADE · indexedpackages_debian_group_architectures.group_id → namespaces.id — one namespaces, many packages_debian_group_architectures · optional · ON DELETE CASCADE · indexedpackages_debian_group_architectures.distribution_id → packages_debian_group_distributions.id — one packages_debian_group_distributions, many packages_debian_group_architectures · required · ON DELETE CASCADE · indexedpackages_debian_group_component_files.group_id → namespaces.id — one namespaces, many packages_debian_group_component_files · optional · ON DELETE CASCADE · indexedpackages_debian_group_component_files.architecture_id → packages_debian_group_architectures.id — one packages_debian_group_architectures, many packages_debian_group_component_files · optional · ON DELETE RESTRICT · indexedpackages_debian_group_component_files.component_id → packages_debian_group_components.id — one packages_debian_group_components, many packages_debian_group_component_files · required · ON DELETE RESTRICT · indexedpackages_debian_group_components.group_id → namespaces.id — one namespaces, many packages_debian_group_components · optional · ON DELETE CASCADE · indexedpackages_debian_group_components.distribution_id → packages_debian_group_distributions.id — one packages_debian_group_distributions, many packages_debian_group_components · required · ON DELETE CASCADE · indexedpackages_debian_group_distribution_keys.group_id → namespaces.id — one namespaces, many packages_debian_group_distribution_keys · optional · ON DELETE CASCADE · indexedpackages_debian_group_distribution_keys.distribution_id → packages_debian_group_distributions.id — one packages_debian_group_distributions, many packages_debian_group_distribution_keys · required · ON DELETE CASCADE · indexedpackages_debian_group_distributions.group_id → namespaces.id — one namespaces, many packages_debian_group_distributions · required · ON DELETE RESTRICT · indexedpackages_debian_group_distributions.creator_id → users.id — one users, many packages_debian_group_distributions · optional · ON DELETE SET NULL · indexedpackages_debian_project_architectures.project_id → projects.id — one projects, many packages_debian_project_architectures · optional · ON DELETE CASCADE · indexedpackages_debian_project_architectures.distribution_id → packages_debian_project_distributions.id — one packages_debian_project_distributions, many packages_debian_project_architectures · required · ON DELETE CASCADE · indexedpackages_debian_project_component_file_states.project_id → projects.id — one projects, many packages_debian_project_component_file_states · required · ON DELETE CASCADE · indexedpackages_debian_project_component_files.project_id → projects.id — one projects, many packages_debian_project_component_files · optional · ON DELETE CASCADE · indexedpackages_debian_project_component_files.architecture_id → packages_debian_project_architectures.id — one packages_debian_project_architectures, many packages_debian_project_component_files · optional · ON DELETE RESTRICT · indexedpackages_debian_project_component_files.component_id → packages_debian_project_components.id — one packages_debian_project_components, many packages_debian_project_component_files · required · ON DELETE RESTRICT · indexedpackages_debian_project_components.project_id → projects.id — one projects, many packages_debian_project_components · optional · ON DELETE CASCADE · indexedpackages_debian_project_components.distribution_id → packages_debian_project_distributions.id — one packages_debian_project_distributions, many packages_debian_project_components · required · ON DELETE CASCADE · indexedpackages_debian_project_distribution_keys.project_id → projects.id — one projects, many packages_debian_project_distribution_keys · optional · ON DELETE CASCADE · indexedpackages_debian_project_distribution_keys.distribution_id → packages_debian_project_distributions.id — one packages_debian_project_distributions, many packages_debian_project_distribution_keys · required · ON DELETE CASCADE · indexedpackages_debian_project_distributions.creator_id → users.id — one users, many packages_debian_project_distributions · optional · ON DELETE SET NULL · indexedpackages_debian_project_distributions.project_id → projects.id — one projects, many packages_debian_project_distributions · required · ON DELETE RESTRICT · indexedpackages_debian_publications.project_id → projects.id — one projects, many packages_debian_publications · optional · ON DELETE CASCADE · indexedpackages_debian_publications.distribution_id → packages_debian_project_distributions.id — one packages_debian_project_distributions, many packages_debian_publications · required · ON DELETE CASCADE · indexedpackages_debian_publications.package_id → packages_packages.id — one packages_packages, at most one packages_debian_publications · required · ON DELETE CASCADE · indexedpackages_dependencies.project_id → projects.id — one projects, many packages_dependencies · optional · ON DELETE CASCADE · indexedpackages_dependency_links.project_id → projects.id — one projects, many packages_dependency_links · optional · ON DELETE CASCADE · indexedpackages_dependency_links.dependency_id → packages_dependencies.id — one packages_dependencies, many packages_dependency_links · required · ON DELETE CASCADE · indexedpackages_dependency_links.package_id → packages_packages.id — one packages_packages, many packages_dependency_links · required · ON DELETE CASCADE · indexedpackages_helm_file_metadata.project_id → projects.id — one projects, many packages_helm_file_metadata · optional · ON DELETE CASCADE · indexedpackages_helm_file_metadata.package_file_id → packages_package_files.id — one packages_package_files, at most one packages_helm_file_metadata · required · ON DELETE CASCADE · indexedpackages_helm_metadata_cache_states.project_id → projects.id — one projects, many packages_helm_metadata_cache_states · optional · ON DELETE CASCADE · indexedpackages_helm_metadata_cache_states.packages_helm_metadata_cache_id → packages_helm_metadata_caches.id — one packages_helm_metadata_caches, at most one packages_helm_metadata_cache_states · required · ON DELETE CASCADE · indexedpackages_maven_metadata.project_id → projects.id — one projects, many packages_maven_metadata · optional · ON DELETE CASCADE · indexedpackages_maven_metadata.package_id → packages_packages.id — one packages_packages, many packages_maven_metadata · required · ON DELETE CASCADE · indexedpackages_npm_metadata.project_id → projects.id — one projects, many packages_npm_metadata · optional · ON DELETE CASCADE · indexedpackages_npm_metadata.package_id → packages_packages.id — one packages_packages, at most one packages_npm_metadata · required · ON DELETE CASCADE · indexedpackages_nuget_dependency_link_metadata.project_id → projects.id — one projects, many packages_nuget_dependency_link_metadata · optional · ON DELETE CASCADE · indexedpackages_nuget_dependency_link_metadata.dependency_link_id → packages_dependency_links.id — one packages_dependency_links, at most one packages_nuget_dependency_link_metadata · required · ON DELETE CASCADE · indexedpackages_nuget_metadata.project_id → projects.id — one projects, many packages_nuget_metadata · optional · ON DELETE CASCADE · indexedpackages_nuget_metadata.package_id → packages_packages.id — one packages_packages, at most one packages_nuget_metadata · required · ON DELETE CASCADE · indexedpackages_nuget_symbols.package_id → packages_packages.id — one packages_packages, many packages_nuget_symbols · optional · ON DELETE SET NULL · indexedpackages_package_file_build_infos.project_id → projects.id — one projects, many packages_package_file_build_infos · optional · ON DELETE CASCADE · indexedpackages_package_file_build_infos.package_file_id → packages_package_files.id — one packages_package_files, many packages_package_file_build_infos · required · ON DELETE CASCADE · indexedpackages_package_files.package_id → packages_packages.id — one packages_packages, many packages_package_files · required · ON DELETE CASCADE · indexedpackages_protection_rules.project_id → projects.id — one projects, many packages_protection_rules · required · ON DELETE CASCADE · indexedpackages_pypi_file_metadata.package_file_id → packages_package_files.id — one packages_package_files, at most one packages_pypi_file_metadata · required · ON DELETE CASCADE · indexedpackages_pypi_file_metadata.project_id → projects.id — one projects, many packages_pypi_file_metadata · required · ON DELETE CASCADE · indexedpackages_pypi_metadata.project_id → projects.id — one projects, many packages_pypi_metadata · optional · ON DELETE CASCADE · indexedpackages_pypi_metadata.package_id → packages_packages.id — one packages_packages, at most one packages_pypi_metadata · required · ON DELETE CASCADE · indexedpackages_rpm_metadata.project_id → projects.id — one projects, many packages_rpm_metadata · optional · ON DELETE CASCADE · indexedpackages_rpm_metadata.package_id → packages_packages.id — one packages_packages, at most one packages_rpm_metadata · required · ON DELETE CASCADE · indexedpackages_rpm_repository_files.project_id → projects.id — one projects, many packages_rpm_repository_files · required · ON DELETE CASCADE · indexedpackages_rubygems_metadata.project_id → projects.id — one projects, many packages_rubygems_metadata · optional · ON DELETE CASCADE · indexedpackages_rubygems_metadata.package_id → packages_packages.id — one packages_packages, at most one packages_rubygems_metadata · required · ON DELETE CASCADE · indexedpackages_tags.project_id → projects.id — one projects, many packages_tags · optional · ON DELETE CASCADE · indexedpackages_tags.package_id → packages_packages.id — one packages_packages, many packages_tags · required · ON DELETE CASCADE · indexedpackages_terraform_module_metadata.package_id → packages_packages.id — one packages_packages, at most one packages_terraform_module_metadata · required · ON DELETE CASCADE · indexedpackages_terraform_module_metadata.project_id → projects.id — one projects, many packages_terraform_module_metadata · required · ON DELETE CASCADE · indexedproject_id — packages_build_infos.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
package_id | bigint | NOT NULL | packages_packages.id 1:N · CASCADE | ||
pipeline_id | bigint | null | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
package_id PK | bigint | NOT NULL | packages_packages.id 1:1 · CASCADE | ||
index_content | jsonb | null | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
normalized_name | text | null | CHECK char_length(normalized_name) <= 64 | ||
normalized_version | text | null | CHECK char_length(normalized_version) <= 255 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
Table checks: char_length(normalized_name) <= 64; char_length(normalized_version) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id PK | bigint | NOT NULL | projects.id 1:1 · CASCADE | ||
next_run_at | timestamptz | null | |||
keep_n_duplicated_package_files | text | NOT NULL | CAST('all' AS text) | CHECK char_length(keep_n_duplicated_package_files) <= 255 |
Table checks: char_length(keep_n_duplicated_package_files) <= 255
project_id — packages_composer_metadata.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
package_id PK | bigint | NOT NULL | packages_packages.id 1:1 · CASCADE | ||
target_sha | bytea | NOT NULL | CHECK octet_length(target_sha) <= 64 | ||
composer_json | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
version_cache_sha | bytea | null | CHECK octet_length(version_cache_sha) <= 255 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL; octet_length(target_sha) <= 64; octet_length(version_cache_sha) <= 255
creator_id — packages_packages.creator_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
name | varchar | NOT NULL | |||
version | varchar | null | |||
package_type | smallint | NOT NULL | |||
creator_id | bigint | null | users.id 1:N · SET NULL | ||
status | smallint | NOT NULL | 0 | ||
last_downloaded_at | timestamptz | null | |||
status_message | text | null | CHECK char_length(status_message) <= 255 |
Table checks: char_length(status_message) <= 255
creator_id — packages_composer_packages.creator_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | nextval(CAST('packages_packages_id_seq' AS regclass)) | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
creator_id | bigint | null | users.id 1:N · SET NULL | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
last_downloaded_at | timestamptz | null | |||
status | smallint | NOT NULL | 0 | ||
name | text | NOT NULL | |||
version | text | null | |||
target_sha | bytea | null | CHECK octet_length(target_sha) <= 64 | ||
version_cache_sha | bytea | null | CHECK octet_length(version_cache_sha) <= 255 | ||
status_message | text | null | CHECK char_length(status_message) <= 255 | ||
composer_json | jsonb | NOT NULL | CAST('{}' AS jsonb) |
Table checks: char_length(status_message) <= 255; octet_length(target_sha) <= 64; octet_length(version_cache_sha) <= 255
project_id — packages_conan_file_metadata.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.package_revision_id — packages_conan_file_metadata.package_revision_id may be NULL, so the relationship to packages_conan_package_revisions is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.package_reference_id — packages_conan_file_metadata.package_reference_id may be NULL, so the relationship to packages_conan_package_references is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.recipe_revision_id — packages_conan_file_metadata.recipe_revision_id may be NULL, so the relationship to packages_conan_recipe_revisions is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id, package_revision_id, package_reference_id, recipe_revision_id — packages_conan_file_metadata has several nullable FKs (project_id, package_revision_id, package_reference_id, recipe_revision_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | packages_package_files.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
conan_file_type | smallint | NOT NULL | |||
recipe_revision_id | bigint | null | packages_conan_recipe_revisions.id 1:N · CASCADE | ||
package_revision_id | bigint | null | packages_conan_package_revisions.id 1:N · CASCADE | ||
package_reference_id | bigint | null | packages_conan_package_references.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL; NOT(conan_file_type = 1 AND package_reference_id IS NOT NULL)
packages_conan_jwt_signing_keys references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
secret_key | jsonb | NOT NULL |
project_id — packages_conan_metadata.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
package_id | bigint | NOT NULL | packages_packages.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
package_username | varchar(255) | NOT NULL | |||
package_channel | varchar(255) | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
recipe_revision_id — packages_conan_package_references.recipe_revision_id may be NULL, so the relationship to packages_conan_recipe_revisions is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
package_id | bigint | NOT NULL | packages_packages.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
recipe_revision_id | bigint | null | packages_conan_recipe_revisions.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
reference | bytea | NOT NULL | CHECK octet_length(reference) <= 20 | ||
info | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK char_length(CAST(info AS text)) <= 20000 |
Table checks: char_length(CAST(info AS text)) <= 20000; octet_length(reference) <= 20
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
package_id | bigint | NOT NULL | packages_packages.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
package_reference_id | bigint | NOT NULL | packages_conan_package_references.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
revision | bytea | NOT NULL | CHECK octet_length(revision) <= 20 | ||
status | smallint | NOT NULL | 0 |
Table checks: octet_length(revision) <= 20
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
package_id | bigint | NOT NULL | packages_packages.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
revision | bytea | NOT NULL | CHECK octet_length(revision) <= 20 | ||
status | smallint | NOT NULL | 0 |
Table checks: octet_length(revision) <= 20
project_id — packages_debian_file_metadata.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
package_file_id PK | bigint | NOT NULL | packages_package_files.id 1:1 · CASCADE | ||
file_type | smallint | NOT NULL | |||
component | text | null | CHECK char_length(component) <= 255 | ||
architecture | text | null | CHECK char_length(architecture) <= 255 | ||
fields | jsonb | null | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: char_length(component) <= 255; project_id IS NOT NULL; char_length(architecture) <= 255
group_id — packages_debian_group_architectures.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
distribution_id | bigint | NOT NULL | packages_debian_group_distributions.id 1:N · CASCADE | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK group_id IS NOT NULL |
Table checks: group_id IS NOT NULL; char_length(name) <= 255
group_id — packages_debian_group_component_files.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.architecture_id — packages_debian_group_component_files.architecture_id may be NULL, so the relationship to packages_debian_group_architectures is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id, architecture_id — packages_debian_group_component_files has several nullable FKs (group_id, architecture_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
component_id | bigint | NOT NULL | packages_debian_group_components.id 1:N · RESTRICT | ||
architecture_id | bigint | null | packages_debian_group_architectures.id 1:N · RESTRICT | ||
size | integer | NOT NULL | |||
file_type | smallint | NOT NULL | |||
compression_type | smallint | null | |||
file_store | smallint | NOT NULL | 1 | ||
file | text | NOT NULL | CHECK char_length(file) <= 255 | ||
file_sha256 | bytea | NOT NULL | CHECK octet_length(file_sha256) <= 64 | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK group_id IS NOT NULL |
Table checks: group_id IS NOT NULL; char_length(file) <= 255; octet_length(file_sha256) <= 64
group_id — packages_debian_group_components.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
distribution_id | bigint | NOT NULL | packages_debian_group_distributions.id 1:N · CASCADE | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK group_id IS NOT NULL |
Table checks: char_length(name) <= 255; group_id IS NOT NULL
group_id — packages_debian_group_distribution_keys.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
distribution_id | bigint | NOT NULL | packages_debian_group_distributions.id 1:N · CASCADE | ||
encrypted_private_key | text | NOT NULL | |||
encrypted_private_key_iv | text | NOT NULL | |||
encrypted_passphrase | text | NOT NULL | |||
encrypted_passphrase_iv | text | NOT NULL | |||
public_key | text | NOT NULL | CHECK char_length(public_key) <= 524288 | ||
fingerprint | text | NOT NULL | CHECK char_length(fingerprint) <= 255 | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK group_id IS NOT NULL |
Table checks: group_id IS NOT NULL; char_length(fingerprint) <= 255; char_length(public_key) <= 524288
creator_id — packages_debian_group_distributions.creator_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · RESTRICT | ||
creator_id | bigint | null | users.id 1:N · SET NULL | ||
valid_time_duration_seconds | integer | null | |||
file_store | smallint | NOT NULL | 1 | ||
automatic | boolean | NOT NULL | TRUE | ||
automatic_upgrades | boolean | NOT NULL | FALSE | ||
codename | text | NOT NULL | CHECK char_length(codename) <= 255 | ||
suite | text | null | CHECK char_length(suite) <= 255 | ||
origin | text | null | CHECK char_length(origin) <= 255 | ||
label | text | null | CHECK char_length(label) <= 255 | ||
version | text | null | CHECK char_length(version) <= 255 | ||
description | text | null | CHECK char_length(description) <= 255 | ||
file | text | null | CHECK char_length(file) <= 255 | ||
file_signature | text | null | CHECK char_length(file_signature) <= 4096 | ||
signed_file | text | null | CHECK char_length(signed_file) <= 255 | ||
signed_file_store | smallint | NOT NULL | 1 |
Table checks: char_length(signed_file) <= 255; char_length(description) <= 255; char_length(file_signature) <= 4096; char_length(version) <= 255; char_length(codename) <= 255; char_length(origin) <= 255; char_length(file) <= 255; char_length(label) <= 255; char_length(suite) <= 255
project_id — packages_debian_project_architectures.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
distribution_id | bigint | NOT NULL | packages_debian_project_distributions.id 1:N · CASCADE | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: char_length(name) <= 255; project_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
packages_debian_project_component_file_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 |
Table checks: char_length(verification_failure) <= 255
project_id — packages_debian_project_component_files.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.architecture_id — packages_debian_project_component_files.architecture_id may be NULL, so the relationship to packages_debian_project_architectures is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id, architecture_id — packages_debian_project_component_files has several nullable FKs (project_id, architecture_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
component_id | bigint | NOT NULL | packages_debian_project_components.id 1:N · RESTRICT | ||
architecture_id | bigint | null | packages_debian_project_architectures.id 1:N · RESTRICT | ||
size | integer | NOT NULL | |||
file_type | smallint | NOT NULL | |||
compression_type | smallint | null | |||
file_store | smallint | NOT NULL | 1 | ||
file | text | NOT NULL | CHECK char_length(file) <= 255 | ||
file_sha256 | bytea | NOT NULL | CHECK octet_length(file_sha256) <= 64 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL; char_length(file) <= 255; octet_length(file_sha256) <= 64
project_id — packages_debian_project_components.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
distribution_id | bigint | NOT NULL | packages_debian_project_distributions.id 1:N · CASCADE | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: char_length(name) <= 255; project_id IS NOT NULL
project_id — packages_debian_project_distribution_keys.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
distribution_id | bigint | NOT NULL | packages_debian_project_distributions.id 1:N · CASCADE | ||
encrypted_private_key | text | NOT NULL | |||
encrypted_private_key_iv | text | NOT NULL | |||
encrypted_passphrase | text | NOT NULL | |||
encrypted_passphrase_iv | text | NOT NULL | |||
public_key | text | NOT NULL | CHECK char_length(public_key) <= 524288 | ||
fingerprint | text | NOT NULL | CHECK char_length(fingerprint) <= 255 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: char_length(fingerprint) <= 255; project_id IS NOT NULL; char_length(public_key) <= 524288
creator_id — packages_debian_project_distributions.creator_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · RESTRICT | ||
creator_id | bigint | null | users.id 1:N · SET NULL | ||
valid_time_duration_seconds | integer | null | |||
file_store | smallint | NOT NULL | 1 | ||
automatic | boolean | NOT NULL | TRUE | ||
automatic_upgrades | boolean | NOT NULL | FALSE | ||
codename | text | NOT NULL | CHECK char_length(codename) <= 255 | ||
suite | text | null | CHECK char_length(suite) <= 255 | ||
origin | text | null | CHECK char_length(origin) <= 255 | ||
label | text | null | CHECK char_length(label) <= 255 | ||
version | text | null | CHECK char_length(version) <= 255 | ||
description | text | null | CHECK char_length(description) <= 255 | ||
file | text | null | CHECK char_length(file) <= 255 | ||
file_signature | text | null | CHECK char_length(file_signature) <= 4096 | ||
signed_file | text | null | CHECK char_length(signed_file) <= 255 | ||
signed_file_store | smallint | NOT NULL | 1 |
Table checks: char_length(origin) <= 255; char_length(label) <= 255; char_length(codename) <= 255; char_length(version) <= 255; char_length(signed_file) <= 255; char_length(suite) <= 255; char_length(file_signature) <= 4096; char_length(description) <= 255; char_length(file) <= 255
project_id — packages_debian_publications.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
package_id | bigint | NOT NULL | packages_packages.id 1:1 · CASCADE | ||
distribution_id | bigint | NOT NULL | packages_debian_project_distributions.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
project_id — packages_dependencies.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
name | varchar(255) | NOT NULL | |||
version_pattern | varchar(255) | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
project_id — packages_dependency_links.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
package_id | bigint | NOT NULL | packages_packages.id 1:N · CASCADE | ||
dependency_id | bigint | NOT NULL | packages_dependencies.id 1:N · CASCADE | ||
dependency_type | smallint | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
project_id — packages_helm_file_metadata.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
package_file_id PK | bigint | NOT NULL | packages_package_files.id 1:1 · CASCADE | ||
channel | text | NOT NULL | CHECK char_length(channel) <= 255 | ||
metadata | jsonb | null | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: char_length(channel) <= 255; project_id IS NOT NULL
project_id — packages_helm_metadata_cache_states.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
packages_helm_metadata_cache_id | bigint | NOT NULL | packages_helm_metadata_caches.id 1:1 · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL; char_length(verification_failure) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
last_downloaded_at | timestamptz | null | |||
project_id | bigint | NOT NULL | |||
size | integer | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
file_store | integer | null | 1 | ||
channel | text | NOT NULL | CHECK char_length(channel) <= 255 | ||
file | text | NOT NULL | CHECK char_length(file) <= 255 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 255 |
Table checks: char_length(object_storage_key) <= 255; char_length(file) <= 255; char_length(channel) <= 255
project_id — packages_maven_metadata.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
package_id | bigint | NOT NULL | packages_packages.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
app_group | varchar | NOT NULL | |||
app_name | varchar | NOT NULL | |||
app_version | varchar | null | |||
path | varchar(512) | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
project_id — packages_npm_metadata.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
package_id PK | bigint | NOT NULL | packages_packages.id 1:1 · CASCADE | ||
package_json | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK char_length(CAST(package_json AS text)) < 20000 | |
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL; char_length(CAST(package_json AS text)) < 20000
packages_npm_metadata_caches references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
last_downloaded_at | timestamptz | null | |||
project_id | bigint | null | CHECK project_id IS NOT NULL | ||
file_store | integer | null | 1 | ||
size | integer | NOT NULL | |||
file | text | NOT NULL | CHECK char_length(file) <= 255 | ||
package_name | text | NOT NULL | |||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 255 | ||
status | smallint | NOT NULL | 0 |
Table checks: char_length(file) <= 255; project_id IS NOT NULL; char_length(object_storage_key) <= 255
project_id — packages_nuget_dependency_link_metadata.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
dependency_link_id PK | bigint | NOT NULL | packages_dependency_links.id 1:1 · CASCADE | ||
target_framework | text | NOT NULL | CHECK char_length(target_framework) <= 255 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL; char_length(target_framework) <= 255
project_id — packages_nuget_metadata.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
package_id PK | bigint | NOT NULL | packages_packages.id 1:1 · CASCADE | ||
license_url | text | null | CHECK char_length(license_url) <= 255 | ||
project_url | text | null | CHECK char_length(project_url) <= 255 | ||
icon_url | text | null | CHECK char_length(icon_url) <= 255 | ||
authors | text | null | CHECK char_length(authors) <= 255 | ||
description | text | null | CHECK char_length(description) <= 4000 | ||
normalized_version | text | null | CHECK char_length(normalized_version) <= 255 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL; char_length(normalized_version) <= 255; char_length(description) <= 4000; char_length(authors) <= 255; char_length(icon_url) <= 255; char_length(license_url) <= 255; char_length(project_url) <= 255
packages_nuget_symbol_states references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
packages_nuget_symbol_id | bigint | NOT NULL | |||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 | ||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL; char_length(verification_failure) <= 255
package_id — packages_nuget_symbols.package_id may be NULL, so the relationship to packages_packages is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
package_id | bigint | null | packages_packages.id 1:N · SET NULL | ||
size | integer | NOT NULL | |||
file_store | smallint | null | 1 | ||
file | text | NOT NULL | CHECK char_length(file) <= 255 | ||
file_path | text | NOT NULL | CHECK char_length(file_path) <= 255 | ||
signature | text | NOT NULL | CHECK char_length(signature) <= 255 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 255 | ||
file_sha256 | bytea | null | CHECK octet_length(file_sha256) <= 64 | ||
status | smallint | NOT NULL | 0 | ||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: char_length(file) <= 255; char_length(object_storage_key) <= 255; char_length(file_path) <= 255; project_id IS NOT NULL; char_length(signature) <= 255; octet_length(file_sha256) <= 64
project_id — packages_package_file_build_infos.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
package_file_id | bigint | NOT NULL | packages_package_files.id 1:N · CASCADE | ||
pipeline_id | bigint | null | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
packages_package_file_states references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
package_file_id | bigint | NOT NULL | |||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | null | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 | ||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL; char_length(verification_failure) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
package_id | bigint | NOT NULL | packages_packages.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
size | bigint | null | |||
file_store | integer | null | 1 | CHECK file_store IS NOT NULL | |
file_md5 | bytea | null | CHECK octet_length(file_md5) <= 32 | ||
file_sha1 | bytea | null | CHECK octet_length(file_sha1) <= 40 | ||
file_name | varchar | NOT NULL | |||
file | text | NOT NULL | |||
file_sha256 | bytea | null | CHECK octet_length(file_sha256) <= 64 | ||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
verification_failure | varchar(255) | null | |||
verification_retry_count | integer | null | |||
verification_checksum | bytea | null | |||
verification_state | smallint | NOT NULL | 0 | ||
verification_started_at | timestamptz | null | |||
status | smallint | NOT NULL | 0 | ||
file_final_path | text | null | CHECK char_length(file_final_path) <= 1024 | ||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: char_length(file_final_path) <= 1024; project_id IS NOT NULL; file_store IS NOT NULL; octet_length(file_md5) <= 32; octet_length(file_sha1) <= 40; octet_length(file_sha256) <= 64
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
package_type | smallint | NOT NULL | |||
package_name_pattern | text | NOT NULL | CHECK char_length(package_name_pattern) <= 255 | ||
minimum_access_level_for_push | smallint | null | |||
minimum_access_level_for_delete | smallint | null | |||
pattern | text | null | CHECK char_length(pattern) <= 255 | ||
pattern_type | smallint | NOT NULL | 0 | ||
target_field | smallint | NOT NULL | 0 |
Table checks: num_nonnulls(minimum_access_level_for_delete, minimum_access_level_for_push) > 0; char_length(pattern) <= 255; char_length(package_name_pattern) <= 255
package_file_id, project_id — packages_pypi_file_metadata looks like a many-to-many link table but has no unique constraint across (package_file_id, project_id). The same pair can be inserted twice; every join through it will double-count.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
package_file_id PK | bigint | NOT NULL | packages_package_files.id 1:1 · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
required_python | text | NOT NULL | CAST('' AS text) | CHECK char_length(required_python) <= 255 |
Table checks: char_length(required_python) <= 255
project_id — packages_pypi_metadata.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
package_id PK | bigint | NOT NULL | packages_packages.id 1:1 · CASCADE | ||
required_python | text | null | CAST('' AS text) | CHECK required_python IS NOT NULL | |
metadata_version | text | null | CHECK char_length(metadata_version) <= 16 | ||
summary | text | null | CHECK char_length(summary) <= 255 | ||
keywords | text | null | CHECK char_length(keywords) <= 1024 | ||
author_email | text | null | CHECK char_length(author_email) <= 2048 | ||
description | text | null | CHECK char_length(description) <= 4000 | ||
description_content_type | text | null | CHECK char_length(description_content_type) <= 128 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: required_python IS NOT NULL; char_length(keywords) <= 1024; char_length(metadata_version) <= 16; char_length(required_python) <= 255; char_length(author_email) <= 2048; char_length(summary) <= 255; project_id IS NOT NULL; char_length(description) <= 4000; char_length(description_content_type) <= 128
project_id — packages_rpm_metadata.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
package_id PK | bigint | NOT NULL | packages_packages.id 1:1 · CASCADE | ||
release | text | NOT NULL | CAST('1' AS text) | CHECK char_length(release) <= 128 | |
summary | text | NOT NULL | CAST('' AS text) | CHECK char_length(summary) <= 1000 | |
description | text | NOT NULL | CAST('' AS text) | CHECK char_length(description) <= 5000 | |
arch | text | NOT NULL | CAST('' AS text) | CHECK char_length(arch) <= 255 | |
license | text | null | CHECK char_length(license) <= 1000 | ||
url | text | null | CHECK char_length(url) <= 1000 | ||
epoch | integer | NOT NULL | 0 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: char_length(arch) <= 255; project_id IS NOT NULL; char_length(description) <= 5000; char_length(url) <= 1000; char_length(license) <= 1000; char_length(summary) <= 1000; char_length(release) <= 128
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
file_store | integer | null | 1 | ||
status | smallint | NOT NULL | 0 | ||
size | integer | null | |||
file_md5 | bytea | null | CHECK octet_length(file_md5) <= 32 | ||
file_sha1 | bytea | null | CHECK octet_length(file_sha1) <= 40 | ||
file_sha256 | bytea | null | CHECK octet_length(file_sha256) <= 64 | ||
file | text | NOT NULL | CHECK char_length(file) <= 255 | ||
file_name | text | NOT NULL | CHECK char_length(file_name) <= 255 |
Table checks: char_length(file) <= 255; char_length(file_name) <= 255; octet_length(file_md5) <= 32; octet_length(file_sha1) <= 40; octet_length(file_sha256) <= 64
project_id — packages_rubygems_metadata.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
package_id PK | bigint | NOT NULL | packages_packages.id 1:1 · CASCADE | ||
authors | text | null | CHECK char_length(authors) <= 255 | ||
files | text | null | CHECK char_length(files) <= 30000 | ||
summary | text | null | CHECK char_length(summary) <= 1024 | ||
description | text | null | CHECK char_length(description) <= 1024 | ||
email | text | null | CHECK char_length(email) <= 255 | ||
homepage | text | null | CHECK char_length(homepage) <= 255 | ||
licenses | text | null | CHECK char_length(licenses) <= 30000 | ||
metadata | text | null | CHECK char_length(metadata) <= 30000 | ||
author | text | null | CHECK char_length(author) <= 255 | ||
bindir | text | null | CHECK char_length(bindir) <= 255 | ||
cert_chain | text | null | CHECK char_length(cert_chain) <= 255 | ||
executables | text | null | CHECK char_length(executables) <= 30000 | ||
extensions | text | null | CHECK char_length(extensions) <= 30000 | ||
extra_rdoc_files | text | null | CHECK char_length(extra_rdoc_files) <= 30000 | ||
platform | text | null | CHECK char_length(platform) <= 255 | ||
post_install_message | text | null | CHECK char_length(post_install_message) <= 255 | ||
rdoc_options | text | null | CHECK char_length(rdoc_options) <= 30000 | ||
require_paths | text | null | CHECK char_length(require_paths) <= 30000 | ||
required_ruby_version | text | null | CHECK char_length(required_ruby_version) <= 255 | ||
required_rubygems_version | text | null | CHECK char_length(required_rubygems_version) <= 255 | ||
requirements | text | null | CHECK char_length(requirements) <= 30000 | ||
rubygems_version | text | null | CHECK char_length(rubygems_version) <= 255 | ||
signing_key | text | null | CHECK char_length(signing_key) <= 255 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: char_length(description) <= 1024; char_length(email) <= 255; char_length(rubygems_version) <= 255; char_length(post_install_message) <= 255; char_length(required_rubygems_version) <= 255; char_length(platform) <= 255; char_length(cert_chain) <= 255; char_length(extensions) <= 30000; char_length(licenses) <= 30000; char_length(summary) <= 1024; char_length(executables) <= 30000; char_length(homepage) <= 255; char_length(bindir) <= 255; char_length(authors) <= 255; char_length(signing_key) <= 255; char_length(extra_rdoc_files) <= 30000; char_length(author) <= 255; char_length(require_paths) <= 30000; char_length(rdoc_options) <= 30000; char_length(required_ruby_version) <= 255; project_id IS NOT NULL; char_length(metadata) <= 30000; char_length(files) <= 30000; char_length(requirements) <= 30000
packages_rubygems_spec_files references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | |||
size | integer | NOT NULL | |||
file_store | integer | null | 1 | ||
status | smallint | NOT NULL | 0 | ||
file_name | text | NOT NULL | CHECK char_length(file_name) <= 255 | ||
file | text | NOT NULL | CHECK char_length(file) <= 255 | ||
object_storage_key | text | NOT NULL | CHECK char_length(object_storage_key) <= 255 |
Table checks: char_length(file_name) <= 255; char_length(object_storage_key) <= 255; char_length(file) <= 255
project_id — packages_tags.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
package_id | bigint | NOT NULL | packages_packages.id 1:N · CASCADE | ||
name | varchar(255) | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
package_id PK | bigint | NOT NULL | packages_packages.id 1:1 · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
fields | jsonb | NOT NULL | CHECK char_length(CAST(fields AS text)) <= 10485760 | ||
semver_major | integer | null | |||
semver_minor | integer | null | |||
semver_prerelease | text | null | CHECK char_length(semver_prerelease) <= 255 | ||
semver_patch | bigint | null |
Table checks: char_length(semver_prerelease) <= 255; char_length(CAST(fields AS text)) <= 10485760
Tables whose names start with pages_.
pages_deployment_states.project_id → projects.id — one projects, many pages_deployment_states · optional · ON DELETE CASCADE · indexedpages_deployment_states.pages_deployment_id → pages_deployments.id — one pages_deployments, at most one pages_deployment_states · required · ON DELETE CASCADE · indexedpages_deployments.project_id → projects.id — one projects, many pages_deployments · required · ON DELETE CASCADE · indexedpages_domain_acme_orders.project_id → projects.id — one projects, many pages_domain_acme_orders · optional · ON DELETE CASCADE · indexedpages_domain_acme_orders.pages_domain_id → pages_domains.id — one pages_domains, many pages_domain_acme_orders · required · ON DELETE CASCADE · indexedpages_domains.project_id → projects.id — one projects, many pages_domains · optional · ON DELETE CASCADE · indexedproject_id — pages_deployment_states.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
pages_deployment_id PK | bigint | NOT NULL | pages_deployments.id 1:1 · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
verification_retry_count | smallint | null | |||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL; char_length(verification_failure) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
ci_build_id | bigint | null | |||
file_store | smallint | NOT NULL | |||
file | text | NOT NULL | CHECK char_length(file) <= 255 | ||
file_count | integer | NOT NULL | |||
file_sha256 | bytea | NOT NULL | |||
size | bigint | null | CHECK size IS NOT NULL | ||
root_directory | text | null | CAST('public' AS text) | CHECK char_length(root_directory) <= 255 | |
path_prefix | text | null | CHECK char_length(path_prefix) <= 128 | ||
build_ref | text | null | CHECK char_length(build_ref) <= 512 | ||
deleted_at | timestamptz | null | |||
upload_ready | boolean | null | FALSE | ||
expires_at | timestamptz | null |
Table checks: char_length(path_prefix) <= 128; size IS NOT NULL; char_length(root_directory) <= 255; char_length(build_ref) <= 512; char_length(file) <= 255
project_id — pages_domain_acme_orders.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
pages_domain_id | bigint | NOT NULL | pages_domains.id 1:N · CASCADE | ||
expires_at | timestamptz | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
url | varchar | NOT NULL | |||
challenge_token | varchar | NOT NULL | |||
challenge_file_content | text | NOT NULL | |||
encrypted_private_key | text | NOT NULL | |||
encrypted_private_key_iv | text | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
project_id — pages_domains.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL | |
certificate | text | null | |||
encrypted_key | text | null | |||
encrypted_key_iv | varchar | null | |||
encrypted_key_salt | varchar | null | |||
domain | varchar | null | |||
verified_at | timestamptz | null | |||
verification_code | varchar | NOT NULL | |||
enabled_until | timestamptz | null | |||
remove_at | timestamptz | null | |||
auto_ssl_enabled | boolean | NOT NULL | FALSE | ||
certificate_valid_not_before | timestamptz | null | |||
certificate_valid_not_after | timestamptz | null | |||
certificate_source | smallint | NOT NULL | 0 | ||
wildcard | boolean | NOT NULL | FALSE | ||
usage | smallint | NOT NULL | 0 | ||
scope | smallint | NOT NULL | 2 | ||
auto_ssl_failed | boolean | NOT NULL | FALSE | ||
updated_at | timestamptz | null |
Table checks: project_id IS NOT NULL
Tables whose names start with personal_.
personal_access_token_granular_scopes.personal_access_token_id → personal_access_tokens.id — one personal_access_tokens, many personal_access_token_granular_scopes · required · ON DELETE CASCADE · indexedpersonal_access_token_granular_scopes.granular_scope_id → granular_scopes.id — one granular_scopes, many personal_access_token_granular_scopes · required · ON DELETE CASCADE · indexedpersonal_access_token_granular_scopes.organization_id → organizations.id — one organizations, many personal_access_token_granular_scopes · required · ON DELETE CASCADE · indexedpersonal_access_token_last_used_ips.personal_access_token_id → personal_access_tokens.id — one personal_access_tokens, many personal_access_token_last_used_ips · required · ON DELETE CASCADE · indexedpersonal_access_token_last_used_ips.organization_id → organizations.id — one organizations, many personal_access_token_last_used_ips · required · ON DELETE CASCADE · indexedpersonal_access_tokens.group_id → namespaces.id — one namespaces, many personal_access_tokens · optional · ON DELETE SET NULL · indexedpersonal_access_tokens.previous_personal_access_token_id → personal_access_tokens.id — one personal_access_tokens, many personal_access_tokens · optional · ON DELETE SET NULL · indexedpersonal_access_tokens.organization_id → organizations.id — one organizations, many personal_access_tokens · required · ON DELETE CASCADE · indexedpersonal_access_tokens.user_id → users.id — one users, many personal_access_tokens · required · ON DELETE CASCADE · indexedpersonal_snippet_upload_states.organization_id → organizations.id — one organizations, many personal_snippet_upload_states · required · ON DELETE CASCADE · indexedpersonal_snippet_upload_states.personal_snippet_upload_id → snippet_uploads.id — one snippet_uploads, at most one personal_snippet_upload_states · required · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
personal_access_token_id | bigint | NOT NULL | personal_access_tokens.id 1:N · CASCADE | ||
granular_scope_id | bigint | NOT NULL | granular_scopes.id 1:N · CASCADE |
organization_id, personal_access_token_id — personal_access_token_last_used_ips looks like a many-to-many link table but has no unique constraint across (organization_id, personal_access_token_id). The same pair can be inserted twice; every join through it will double-count.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
personal_access_token_id | bigint | NOT NULL | personal_access_tokens.id 1:N · CASCADE | ||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
ip_address | inet | null |
group_id — personal_access_tokens.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.previous_personal_access_token_id — personal_access_tokens.previous_personal_access_token_id may be NULL, so the relationship to personal_access_tokens is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id, previous_personal_access_token_id — personal_access_tokens has several nullable FKs (group_id, previous_personal_access_token_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
name | varchar | NOT NULL | |||
revoked | boolean | null | FALSE | ||
expires_at | date | null | |||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL | |||
scopes | varchar | NOT NULL | CAST('--- [] ' AS varchar) | - [] | |
impersonation | boolean | NOT NULL | FALSE | ||
token_digest | varchar | null | |||
expire_notification_delivered | boolean | NOT NULL | FALSE | ||
last_used_at | timestamptz | null | |||
after_expiry_notification_delivered | boolean | NOT NULL | FALSE | ||
previous_personal_access_token_id | bigint | null | personal_access_tokens.id 1:N · SET NULL | ||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
seven_days_notification_sent_at | timestamptz | null | |||
thirty_days_notification_sent_at | timestamptz | null | |||
sixty_days_notification_sent_at | timestamptz | null | |||
description | text | null | CHECK char_length(description) <= 255 | ||
group_id | bigint | null | namespaces.id 1:N · SET NULL | ||
user_type | smallint | null | |||
granular | boolean | NOT NULL | FALSE | ||
sudo | boolean | NOT NULL | FALSE |
Table checks: char_length(description) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
personal_snippet_upload_id | bigint | NOT NULL | snippet_uploads.id 1:1 · CASCADE | ||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 |
Table checks: char_length(verification_failure) <= 255
Tables whose names start with snippet_.
snippet_repositories.snippet_organization_id → organizations.id — one organizations, many snippet_repositories · optional · ON DELETE CASCADE · indexedsnippet_repositories.snippet_project_id → projects.id — one projects, many snippet_repositories · optional · ON DELETE CASCADE · indexedsnippet_repositories.snippet_id → snippets.id — one snippets, at most one snippet_repositories · required · ON DELETE CASCADE · indexedsnippet_repository_states.snippet_repository_id → snippet_repositories.snippet_id — one snippet_repositories, at most one snippet_repository_states · required · ON DELETE CASCADE · indexedsnippet_repository_states.snippet_organization_id → organizations.id — one organizations, many snippet_repository_states · optional · ON DELETE CASCADE · indexedsnippet_repository_states.snippet_project_id → projects.id — one projects, many snippet_repository_states · optional · ON DELETE CASCADE · indexedsnippet_repository_storage_moves.snippet_project_id → projects.id — one projects, many snippet_repository_storage_moves · optional · ON DELETE CASCADE · indexedsnippet_repository_storage_moves.snippet_organization_id → organizations.id — one organizations, many snippet_repository_storage_moves · optional · ON DELETE CASCADE · indexedsnippet_repository_storage_moves.snippet_id → snippets.id — one snippets, many snippet_repository_storage_moves · required · ON DELETE CASCADE · indexedsnippet_statistics.snippet_organization_id → organizations.id — one organizations, many snippet_statistics · optional · ON DELETE CASCADE · indexedsnippet_statistics.snippet_project_id → projects.id — one projects, many snippet_statistics · optional · ON DELETE CASCADE · indexedsnippet_statistics.snippet_id → snippets.id — one snippets, at most one snippet_statistics · required · ON DELETE CASCADE · indexedsnippet_user_mentions.snippet_project_id → projects.id — one projects, many snippet_user_mentions · optional · ON DELETE CASCADE · indexedsnippet_user_mentions.snippet_organization_id → organizations.id — one organizations, many snippet_user_mentions · optional · ON DELETE CASCADE · indexedsnippet_user_mentions.snippet_id → snippets.id — one snippets, many snippet_user_mentions · required · ON DELETE CASCADE · indexedsnippet_user_mentions.note_id → notes.id — one notes, many snippet_user_mentions · optional · ON DELETE CASCADE · indexedsnippet_organization_id — snippet_repositories.snippet_organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.snippet_project_id — snippet_repositories.snippet_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
snippet_id PK | bigint | NOT NULL | snippets.id 1:1 · CASCADE | ||
shard_id | bigint | null | |||
disk_path | varchar(80) | NOT NULL | |||
verification_retry_count | smallint | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_started_at | timestamptz | null | |||
snippet_project_id | bigint | null | projects.id 1:N · CASCADE | ||
snippet_organization_id | bigint | null | organizations.id 1:N · CASCADE |
Table checks: num_nonnulls(snippet_organization_id, snippet_project_id) = 1; char_length(verification_failure) <= 255
snippet_organization_id — snippet_repository_states.snippet_organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.snippet_project_id — snippet_repository_states.snippet_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
snippet_repository_id | bigint | NOT NULL | snippet_repositories.snippet_id 1:1 · CASCADE | ||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | null | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 | ||
snippet_project_id | bigint | null | projects.id 1:N · CASCADE | ||
snippet_organization_id | bigint | null | organizations.id 1:N · CASCADE |
Table checks: char_length(verification_failure) <= 255; num_nonnulls(snippet_organization_id, snippet_project_id) = 1
snippet_project_id — snippet_repository_storage_moves.snippet_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.snippet_organization_id — snippet_repository_storage_moves.snippet_organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
snippet_id | bigint | NOT NULL | snippets.id 1:N · CASCADE | ||
state | smallint | NOT NULL | 1 | ||
source_storage_name | text | NOT NULL | CHECK char_length(source_storage_name) <= 255 | ||
destination_storage_name | text | NOT NULL | CHECK char_length(destination_storage_name) <= 255 | ||
error_message | text | null | CHECK char_length(error_message) <= 256 | ||
snippet_project_id | bigint | null | projects.id 1:N · CASCADE | ||
snippet_organization_id | bigint | null | organizations.id 1:N · CASCADE |
Table checks: num_nonnulls(snippet_organization_id, snippet_project_id) = 1; char_length(error_message) <= 256; char_length(destination_storage_name) <= 255; char_length(source_storage_name) <= 255
snippet_organization_id — snippet_statistics.snippet_organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.snippet_project_id — snippet_statistics.snippet_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
snippet_id PK | bigint | NOT NULL | snippets.id 1:1 · CASCADE | ||
repository_size | bigint | NOT NULL | 0 | ||
file_count | bigint | NOT NULL | 0 | ||
commit_count | bigint | NOT NULL | 0 | ||
snippet_project_id | bigint | null | projects.id 1:N · CASCADE | ||
snippet_organization_id | bigint | null | organizations.id 1:N · CASCADE |
Table checks: num_nonnulls(snippet_organization_id, snippet_project_id) = 1
model_type — model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.model_type, model_id — model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | nextval(CAST('uploads_id_seq' AS regclass)) | ||
size | bigint | NOT NULL | |||
model_id | bigint | NOT NULL | |||
uploaded_by_user_id | bigint | null | |||
organization_id | bigint | null | CHECK organization_id IS NOT NULL | ||
namespace_id | bigint | null | |||
project_id | bigint | null | |||
created_at | timestamp | null | |||
store | integer | NOT NULL | 1 | ||
version | integer | null | 1 | ||
path | text | NOT NULL | CHECK char_length(path) <= 511 | ||
checksum | text | null | CHECK char_length(checksum) <= 64 | ||
model_type PK | text | NOT NULL | |||
uploader | text | NOT NULL | |||
mount_point | text | null | |||
secret | text | null |
Table checks: char_length(path) <= 511; char_length(checksum) <= 64; organization_id IS NOT NULL
snippet_project_id — snippet_user_mentions.snippet_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.snippet_organization_id — snippet_user_mentions.snippet_organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.note_id — snippet_user_mentions.note_id may be NULL, so the relationship to notes is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.snippet_project_id, snippet_organization_id, note_id — snippet_user_mentions has several nullable FKs (snippet_project_id, snippet_organization_id, note_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
snippet_id | bigint | NOT NULL | snippets.id 1:N · CASCADE | ||
mentioned_users_ids | bigint[] | null | |||
mentioned_projects_ids | bigint[] | null | |||
mentioned_groups_ids | bigint[] | null | |||
note_id | bigint | null | notes.id 1:N · CASCADE | ||
snippet_project_id | bigint | null | projects.id 1:N · CASCADE | ||
snippet_organization_id | bigint | null | organizations.id 1:N · CASCADE |
Table checks: num_nonnulls(snippet_organization_id, snippet_project_id) = 1
Tables whose names start with pm_.
pm_affected_packages.pm_advisory_id → pm_advisories.id — one pm_advisories, many pm_affected_packages · required · ON DELETE CASCADE · indexedpm_malware_affected_packages.pm_malware_advisory_id → pm_malware_advisories.id — one pm_malware_advisories, many pm_malware_affected_packages · required · ON DELETE CASCADE · indexedpm_package_version_licenses.pm_package_version_id → pm_package_versions.id — one pm_package_versions, many pm_package_version_licenses · required · ON DELETE CASCADE · indexedpm_package_version_licenses.pm_license_id → pm_licenses.id — one pm_licenses, many pm_package_version_licenses · required · ON DELETE CASCADE · indexedpm_package_versions.pm_package_id → pm_packages.id — one pm_packages, many pm_package_versions · required · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
advisory_xid | text | NOT NULL | CHECK char_length(advisory_xid) <= 36 | ||
published_date | date | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
source_xid | smallint | NOT NULL | |||
title | text | null | CHECK char_length(title) <= 256 | ||
description | text | null | CHECK char_length(description) <= 8192 | ||
cvss_v2 | text | null | CHECK char_length(cvss_v2) <= 128 | ||
cvss_v3 | text | null | CHECK char_length(cvss_v3) <= 128 | ||
urls | text[] | null | CAST('{}' AS text[]) | CHECK cardinality(urls) <= 20 | |
identifiers | jsonb | NOT NULL | |||
cve | text | null | CHECK char_length(cve) <= 24 | ||
cvss_v4 | text | null | CHECK char_length(cvss_v4) <= 180 |
Table checks: char_length(cvss_v2) <= 128; char_length(advisory_xid) <= 36; char_length(cve) <= 24; char_length(cvss_v3) <= 128; char_length(title) <= 256; char_length(description) <= 8192; cardinality(urls) <= 20; char_length(cvss_v4) <= 180
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
pm_advisory_id | bigint | NOT NULL | pm_advisories.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
purl_type | smallint | NOT NULL | |||
package_name | text | NOT NULL | CHECK char_length(package_name) <= 256 | ||
distro_version | text | NOT NULL | CAST('' AS text) | CHECK char_length(distro_version) <= 256 | |
solution | text | null | CHECK char_length(solution) <= 2048 | ||
affected_range | text | NOT NULL | CHECK char_length(affected_range) <= 512 | ||
fixed_versions | text[] | null | CAST('{}' AS text[]) | CHECK cardinality(fixed_versions) <= 10 | |
overridden_advisory_fields | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
versions | jsonb | NOT NULL | CAST('[]' AS jsonb) |
Table checks: char_length(package_name) <= 256; char_length(affected_range) <= 512; char_length(solution) <= 2048; char_length(distro_version) <= 256; cardinality(fixed_versions) <= 10
pm_checkpoints references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
sequence | integer | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
purl_type | smallint | NOT NULL | |||
chunk | smallint | NOT NULL | |||
data_type | smallint | NOT NULL | 1 | ||
version_format | smallint | NOT NULL | 1 | ||
id PK | bigint | NOT NULL | |||
full_sync_target_sequence | integer | null |
pm_cve_enrichment references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
epss_score | double precision | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
cve | text | NOT NULL | CHECK char_length(cve) <= 24 | ||
is_known_exploit | boolean | NOT NULL | FALSE |
Table checks: char_length(cve) <= 24
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
spdx_identifier | text | NOT NULL | CHECK char_length(spdx_identifier) <= 50 | ||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() |
Table checks: char_length(spdx_identifier) <= 50
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
published_date | date | NOT NULL | |||
withdrawn_date | date | null | |||
source_xid | smallint | NOT NULL | |||
advisory_xid | text | NOT NULL | CHECK char_length(advisory_xid) <= 36 | ||
title | text | null | CHECK char_length(title) <= 2048 | ||
description | text | null | CHECK char_length(description) <= 10000 | ||
identifiers | jsonb | NOT NULL | CAST('[]' AS jsonb) | ||
urls | text[] | NOT NULL | CAST('{}' AS text[]) |
Table checks: char_length(title) <= 2048; char_length(description) <= 10000; char_length(advisory_xid) <= 36
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
pm_malware_advisory_id | bigint | NOT NULL | pm_malware_advisories.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
purl_type | smallint | NOT NULL | |||
package_name | text | NOT NULL | CHECK char_length(package_name) <= 512 | ||
affected_range | text | NOT NULL | CHECK char_length(affected_range) <= 10000 |
Table checks: char_length(package_name) <= 512; char_length(affected_range) <= 10000
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
pm_package_version_id | bigint | NOT NULL | pm_package_versions.id 1:N · CASCADE | ||
pm_license_id | bigint | NOT NULL | pm_licenses.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
id PK | bigint | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
pm_package_id | bigint | NOT NULL | pm_packages.id 1:N · CASCADE | ||
version | text | NOT NULL | CHECK char_length(version) <= 255 | ||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() |
Table checks: char_length(version) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
purl_type | smallint | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
licenses | jsonb | null |
Table checks: char_length(name) <= 255
Tables whose names start with postgres_.
No foreign keys in this domain.
postgres_async_foreign_key_validations references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 63 | ||
table_name | text | NOT NULL | CHECK char_length(table_name) <= 127 | ||
last_error | text | null | CHECK char_length(last_error) <= 10000 | ||
attempts | integer | NOT NULL | 0 | ||
constraint_type | smallint | NOT NULL | 0 |
Table checks: char_length(last_error) <= 10000; char_length(name) <= 63; char_length(table_name) <= 127
postgres_async_indexes references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 63 | ||
definition | text | NOT NULL | CHECK char_length(definition) <= 2048 | ||
table_name | text | NOT NULL | CHECK char_length(table_name) <= 127 | ||
attempts | integer | NOT NULL | 0 | ||
last_error | text | null | CHECK char_length(last_error) <= 10000 |
Table checks: char_length(definition) <= 2048; char_length(last_error) <= 10000; char_length(name) <= 63; char_length(table_name) <= 127
postgres_reindex_actions references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
action_start | timestamptz | NOT NULL | |||
action_end | timestamptz | null | |||
ondisk_size_bytes_start | bigint | NOT NULL | |||
ondisk_size_bytes_end | bigint | null | |||
state | smallint | NOT NULL | 0 | ||
index_identifier | text | NOT NULL | CHECK char_length(index_identifier) <= 255 | ||
bloat_estimate_bytes_start | bigint | null |
Table checks: char_length(index_identifier) <= 255
postgres_reindex_queued_actions references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
index_identifier | text | NOT NULL | CHECK char_length(index_identifier) <= 255 | ||
state | smallint | NOT NULL | 0 | ||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() |
Table checks: char_length(index_identifier) <= 255
Tables whose names start with protected_.
protected_branch_merge_access_levels.protected_branch_project_id → projects.id — one projects, many protected_branch_merge_access_levels · optional · ON DELETE CASCADE · indexedprotected_branch_merge_access_levels.protected_branch_id → protected_branches.id — one protected_branches, many protected_branch_merge_access_levels · required · ON DELETE CASCADE · indexedprotected_branch_merge_access_levels.group_id → namespaces.id — one namespaces, many protected_branch_merge_access_levels · optional · ON DELETE CASCADE · indexedprotected_branch_merge_access_levels.member_role_id → member_roles.id — one member_roles, many protected_branch_merge_access_levels · optional · ON DELETE RESTRICT · indexedprotected_branch_merge_access_levels.protected_branch_namespace_id → namespaces.id — one namespaces, many protected_branch_merge_access_levels · optional · ON DELETE CASCADE · indexedprotected_branch_merge_access_levels.user_id → users.id — one users, many protected_branch_merge_access_levels · optional · ON DELETE CASCADE · indexedprotected_branch_push_access_levels.deploy_key_id → keys.id — one keys, many protected_branch_push_access_levels · optional · ON DELETE CASCADE · indexedprotected_branch_push_access_levels.protected_branch_namespace_id → namespaces.id — one namespaces, many protected_branch_push_access_levels · optional · ON DELETE CASCADE · indexedprotected_branch_push_access_levels.group_id → namespaces.id — one namespaces, many protected_branch_push_access_levels · optional · ON DELETE CASCADE · indexedprotected_branch_push_access_levels.member_role_id → member_roles.id — one member_roles, many protected_branch_push_access_levels · optional · ON DELETE RESTRICT · indexedprotected_branch_push_access_levels.protected_branch_id → protected_branches.id — one protected_branches, many protected_branch_push_access_levels · required · ON DELETE CASCADE · indexedprotected_branch_push_access_levels.protected_branch_project_id → projects.id — one projects, many protected_branch_push_access_levels · optional · ON DELETE CASCADE · indexedprotected_branch_push_access_levels.user_id → users.id — one users, many protected_branch_push_access_levels · optional · ON DELETE CASCADE · indexedprotected_branch_unprotect_access_levels.protected_branch_project_id → projects.id — one projects, many protected_branch_unprotect_access_levels · optional · ON DELETE CASCADE · indexedprotected_branch_unprotect_access_levels.protected_branch_namespace_id → namespaces.id — one namespaces, many protected_branch_unprotect_access_levels · optional · ON DELETE CASCADE · indexedprotected_branch_unprotect_access_levels.member_role_id → member_roles.id — one member_roles, many protected_branch_unprotect_access_levels · optional · ON DELETE RESTRICT · indexedprotected_branch_unprotect_access_levels.user_id → users.id — one users, many protected_branch_unprotect_access_levels · optional · ON DELETE CASCADE · indexedprotected_branch_unprotect_access_levels.group_id → namespaces.id — one namespaces, many protected_branch_unprotect_access_levels · optional · ON DELETE CASCADE · indexedprotected_branch_unprotect_access_levels.protected_branch_id → protected_branches.id — one protected_branches, many protected_branch_unprotect_access_levels · required · ON DELETE CASCADE · indexedprotected_branches.project_id → projects.id — one projects, many protected_branches · optional · ON DELETE CASCADE · indexedprotected_branches.namespace_id → namespaces.id — one namespaces, many protected_branches · optional · ON DELETE CASCADE · indexedprotected_environment_approval_rules.protected_environment_group_id → namespaces.id — one namespaces, many protected_environment_approval_rules · optional · ON DELETE CASCADE · indexedprotected_environment_approval_rules.group_id → namespaces.id — one namespaces, many protected_environment_approval_rules · optional · ON DELETE CASCADE · indexedprotected_environment_approval_rules.user_id → users.id — one users, many protected_environment_approval_rules · optional · ON DELETE CASCADE · indexedprotected_environment_approval_rules.protected_environment_project_id → projects.id — one projects, many protected_environment_approval_rules · optional · ON DELETE CASCADE · indexedprotected_environment_approval_rules.protected_environment_id → protected_environments.id — one protected_environments, many protected_environment_approval_rules · required · ON DELETE CASCADE · indexedprotected_environment_deploy_access_levels.protected_environment_group_id → namespaces.id — one namespaces, many protected_environment_deploy_access_levels · optional · ON DELETE CASCADE · indexedprotected_environment_deploy_access_levels.protected_environment_project_id → projects.id — one projects, many protected_environment_deploy_access_levels · optional · ON DELETE CASCADE · indexedprotected_environment_deploy_access_levels.group_id → namespaces.id — one namespaces, many protected_environment_deploy_access_levels · optional · ON DELETE CASCADE · indexedprotected_environment_deploy_access_levels.user_id → users.id — one users, many protected_environment_deploy_access_levels · optional · ON DELETE CASCADE · indexedprotected_environment_deploy_access_levels.protected_environment_id → protected_environments.id — one protected_environments, many protected_environment_deploy_access_levels · required · ON DELETE CASCADE · indexedprotected_environments.group_id → namespaces.id — one namespaces, many protected_environments · optional · ON DELETE CASCADE · indexedprotected_environments.project_id → projects.id — one projects, many protected_environments · optional · ON DELETE CASCADE · indexedprotected_tag_create_access_levels.deploy_key_id → keys.id — one keys, many protected_tag_create_access_levels · optional · ON DELETE CASCADE · indexedprotected_tag_create_access_levels.project_id → projects.id — one projects, many protected_tag_create_access_levels · optional · ON DELETE CASCADE · indexedprotected_tag_create_access_levels.group_id → namespaces.id — one namespaces, many protected_tag_create_access_levels · optional · ON DELETE CASCADE · indexedprotected_tag_create_access_levels.protected_tag_id → protected_tags.id — one protected_tags, many protected_tag_create_access_levels · required · ON DELETE CASCADE · indexedprotected_tag_create_access_levels.user_id → users.id — one users, many protected_tag_create_access_levels · optional · ON DELETE CASCADE · indexedprotected_tags.project_id → projects.id — one projects, many protected_tags · required · ON DELETE CASCADE · indexedprotected_branch_project_id — protected_branch_merge_access_levels.protected_branch_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — protected_branch_merge_access_levels.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.member_role_id — protected_branch_merge_access_levels.member_role_id may be NULL, so the relationship to member_roles is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.protected_branch_namespace_id — protected_branch_merge_access_levels.protected_branch_namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id — protected_branch_merge_access_levels.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.protected_branch_project_id, group_id, member_role_id, protected_branch_namespace_id, user_id — protected_branch_merge_access_levels has several nullable FKs (protected_branch_project_id, group_id, member_role_id, protected_branch_namespace_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
protected_branch_id | bigint | NOT NULL | protected_branches.id 1:N · CASCADE | ||
access_level | integer | null | 40 | ||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL | |||
user_id | bigint | null | users.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
protected_branch_project_id | bigint | null | projects.id 1:N · CASCADE | ||
protected_branch_namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
member_role_id | bigint | null | member_roles.id 1:N · RESTRICT |
Table checks: num_nonnulls(protected_branch_namespace_id, protected_branch_project_id) = 1
deploy_key_id — protected_branch_push_access_levels.deploy_key_id may be NULL, so the relationship to keys is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.protected_branch_namespace_id — protected_branch_push_access_levels.protected_branch_namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — protected_branch_push_access_levels.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.member_role_id — protected_branch_push_access_levels.member_role_id may be NULL, so the relationship to member_roles is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.protected_branch_project_id — protected_branch_push_access_levels.protected_branch_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id — protected_branch_push_access_levels.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.deploy_key_id, protected_branch_namespace_id, group_id, member_role_id, protected_branch_project_id, user_id — protected_branch_push_access_levels has several nullable FKs (deploy_key_id, protected_branch_namespace_id, group_id, member_role_id, protected_branch_project_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
protected_branch_id | bigint | NOT NULL | protected_branches.id 1:N · CASCADE | ||
access_level | integer | null | 40 | ||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL | |||
user_id | bigint | null | users.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
deploy_key_id | bigint | null | keys.id 1:N · CASCADE | ||
protected_branch_project_id | bigint | null | projects.id 1:N · CASCADE | ||
protected_branch_namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
member_role_id | bigint | null | member_roles.id 1:N · RESTRICT |
Table checks: num_nonnulls(protected_branch_namespace_id, protected_branch_project_id) = 1
protected_branch_project_id — protected_branch_unprotect_access_levels.protected_branch_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.protected_branch_namespace_id — protected_branch_unprotect_access_levels.protected_branch_namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.member_role_id — protected_branch_unprotect_access_levels.member_role_id may be NULL, so the relationship to member_roles is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id — protected_branch_unprotect_access_levels.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — protected_branch_unprotect_access_levels.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.protected_branch_project_id, protected_branch_namespace_id, member_role_id, user_id, group_id — protected_branch_unprotect_access_levels has several nullable FKs (protected_branch_project_id, protected_branch_namespace_id, member_role_id, user_id, group_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
protected_branch_id | bigint | NOT NULL | protected_branches.id 1:N · CASCADE | ||
access_level | integer | null | 40 | ||
user_id | bigint | null | users.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
protected_branch_project_id | bigint | null | projects.id 1:N · CASCADE | ||
protected_branch_namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
member_role_id | bigint | null | member_roles.id 1:N · RESTRICT |
Table checks: num_nonnulls(protected_branch_namespace_id, protected_branch_project_id) = 1
project_id — protected_branches.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id — protected_branches.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
name | varchar | NOT NULL | |||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
code_owner_approval_required | boolean | NOT NULL | FALSE | ||
allow_force_push | boolean | NOT NULL | FALSE | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE |
Table checks: project_id IS NULL <> namespace_id IS NULL
protected_environment_group_id — protected_environment_approval_rules.protected_environment_group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — protected_environment_approval_rules.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id — protected_environment_approval_rules.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.protected_environment_project_id — protected_environment_approval_rules.protected_environment_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.protected_environment_group_id, group_id, user_id, protected_environment_project_id — protected_environment_approval_rules has several nullable FKs (protected_environment_group_id, group_id, user_id, protected_environment_project_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
protected_environment_id | bigint | NOT NULL | protected_environments.id 1:N · CASCADE | ||
user_id | bigint | null | users.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
access_level | smallint | null | |||
required_approvals | smallint | NOT NULL | CHECK required_approvals > 0 | ||
group_inheritance_type | smallint | NOT NULL | 0 | ||
protected_environment_project_id | bigint | null | projects.id 1:N · CASCADE | ||
protected_environment_group_id | bigint | null | namespaces.id 1:N · CASCADE |
Table checks: num_nonnulls(protected_environment_group_id, protected_environment_project_id) = 1; (access_level IS NOT NULL AND group_id IS NULL AND user_id IS NULL) OR (user_id IS NOT NULL AND access_level IS NULL AND group_id IS NULL) OR (group_id IS NOT NULL AND user_id IS NULL AND access_level IS NULL); required_approvals > 0
protected_environment_group_id — protected_environment_deploy_access_levels.protected_environment_group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.protected_environment_project_id — protected_environment_deploy_access_levels.protected_environment_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — protected_environment_deploy_access_levels.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id — protected_environment_deploy_access_levels.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.protected_environment_group_id, protected_environment_project_id, group_id, user_id — protected_environment_deploy_access_levels has several nullable FKs (protected_environment_group_id, protected_environment_project_id, group_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
access_level | integer | null | |||
protected_environment_id | bigint | NOT NULL | protected_environments.id 1:N · CASCADE | ||
user_id | bigint | null | users.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
group_inheritance_type | smallint | NOT NULL | 0 | ||
protected_environment_project_id | bigint | null | projects.id 1:N · CASCADE | ||
protected_environment_group_id | bigint | null | namespaces.id 1:N · CASCADE |
Table checks: num_nonnulls(protected_environment_group_id, protected_environment_project_id) = 1; num_nonnulls(user_id, group_id, access_level) = 1
group_id — protected_environments.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — protected_environments.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
name | varchar | NOT NULL | |||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
required_approval_count | integer | NOT NULL | 0 | CHECK required_approval_count >= 0 |
Table checks: project_id IS NULL <> group_id IS NULL; required_approval_count >= 0
deploy_key_id — protected_tag_create_access_levels.deploy_key_id may be NULL, so the relationship to keys is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — protected_tag_create_access_levels.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — protected_tag_create_access_levels.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id — protected_tag_create_access_levels.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.deploy_key_id, project_id, group_id, user_id — protected_tag_create_access_levels has several nullable FKs (deploy_key_id, project_id, group_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
protected_tag_id | bigint | NOT NULL | protected_tags.id 1:N · CASCADE | ||
access_level | integer | null | 40 | ||
user_id | bigint | null | users.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL | |||
deploy_key_id | bigint | null | keys.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
name | varchar | NOT NULL | |||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL |
Tables whose names start with push_.
push_event_payloads.project_id → projects.id — one projects, many push_event_payloads · optional · ON DELETE CASCADE · indexedpush_event_payloads.event_id → events.id — one events, at most one push_event_payloads · required · ON DELETE CASCADE · indexedpush_rules.project_id → projects.id — one projects, many push_rules · optional · ON DELETE CASCADE · indexedproject_id — push_event_payloads.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
commit_count | bigint | NOT NULL | |||
action | smallint | NOT NULL | |||
ref_type | smallint | NOT NULL | |||
commit_from | bytea | null | |||
commit_to | bytea | null | |||
ref | text | null | |||
commit_title | varchar(70) | null | |||
ref_count | integer | null | |||
event_id PK | bigint | NOT NULL | events.id 1:1 · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
project_id — push_rules.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
commit_message_regex | varchar | null | CHECK char_length(CAST(commit_message_regex AS text)) <= 511 | ||
deny_delete_tag | boolean | null | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL | |
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
author_email_regex | varchar | null | CHECK char_length(CAST(author_email_regex AS text)) <= 511 | ||
member_check | boolean | NOT NULL | FALSE | ||
file_name_regex | varchar | null | CHECK char_length(CAST(file_name_regex AS text)) <= 511 | ||
is_sample | boolean | null | FALSE | ||
max_file_size | integer | NOT NULL | 0 | ||
prevent_secrets | boolean | NOT NULL | FALSE | ||
branch_name_regex | varchar | null | CHECK char_length(CAST(branch_name_regex AS text)) <= 511 | ||
reject_unsigned_commits | boolean | null | |||
commit_committer_check | boolean | null | |||
regexp_uses_re2 | boolean | null | TRUE | ||
commit_message_negative_regex | varchar | null | CHECK char_length(CAST(commit_message_negative_regex AS text)) <= 2047 | ||
reject_non_dco_commits | boolean | null | |||
commit_committer_name_check | boolean | NOT NULL | FALSE | ||
organization_id | bigint | null |
Table checks: char_length(CAST(author_email_regex AS text)) <= 511; char_length(CAST(branch_name_regex AS text)) <= 511; char_length(CAST(commit_message_negative_regex AS text)) <= 2047; char_length(CAST(commit_message_regex AS text)) <= 511; char_length(CAST(file_name_regex AS text)) <= 511; project_id IS NOT NULL
Tables whose names start with requirements_.
requirements.issue_id → issues.id — one issues, at most one requirements · optional · ON DELETE CASCADE · indexedrequirements.project_id → projects.id — one projects, many requirements · required · ON DELETE CASCADE · indexedrequirements_management_test_reports.project_id → projects.id — one projects, many requirements_management_test_reports · optional · ON DELETE CASCADE · indexedrequirements_management_test_reports.issue_id → issues.id — one issues, many requirements_management_test_reports · optional · ON DELETE CASCADE · indexedrequirements_management_test_reports.author_id → users.id — one users, many requirements_management_test_reports · optional · ON DELETE SET NULL · indexedissue_id — requirements.issue_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
iid | integer | NOT NULL | |||
issue_id | bigint | null | issues.id 1:1 · CASCADE | CHECK issue_id IS NOT NULL |
Table checks: issue_id IS NOT NULL
project_id — requirements_management_test_reports.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.issue_id — requirements_management_test_reports.issue_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.author_id — requirements_management_test_reports.author_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id, issue_id, author_id — requirements_management_test_reports has several nullable FKs (project_id, issue_id, author_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
author_id | bigint | null | users.id 1:N · SET NULL | ||
state | smallint | NOT NULL | |||
build_id | bigint | null | |||
issue_id | bigint | null | issues.id 1:N · CASCADE | ||
uses_legacy_iid | boolean | NOT NULL | TRUE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
Tables whose names start with resource_.
resource_iteration_events.triggered_by_id → issues.id — one issues, many resource_iteration_events · optional · ON DELETE SET NULL · indexedresource_iteration_events.namespace_id → namespaces.id — one namespaces, many resource_iteration_events · required · ON DELETE CASCADE · indexedresource_iteration_events.user_id → users.id — one users, many resource_iteration_events · required · ON DELETE SET NULL · indexedresource_iteration_events.merge_request_id → merge_requests.id — one merge_requests, many resource_iteration_events · optional · ON DELETE CASCADE · indexedresource_iteration_events.issue_id → issues.id — one issues, many resource_iteration_events · optional · ON DELETE CASCADE · indexedresource_iteration_events.iteration_id → sprints.id — one sprints, many resource_iteration_events · optional · ON DELETE CASCADE · indexedresource_label_events.namespace_id → namespaces.id — one namespaces, many resource_label_events · required · ON DELETE CASCADE · indexedresource_label_events.issue_id → issues.id — one issues, many resource_label_events · optional · ON DELETE CASCADE · indexedresource_label_events.epic_id → epics.id — one epics, many resource_label_events · optional · ON DELETE CASCADE · indexedresource_label_events.merge_request_id → merge_requests.id — one merge_requests, many resource_label_events · optional · ON DELETE CASCADE · indexedresource_label_events.label_id → labels.id — one labels, many resource_label_events · optional · ON DELETE SET NULL · indexedresource_label_events.user_id → users.id — one users, many resource_label_events · optional · ON DELETE SET NULL · indexedresource_link_events.system_note_metadata_id → system_note_metadata.id — one system_note_metadata, at most one resource_link_events · optional · ON DELETE CASCADE · indexedresource_link_events.namespace_id → namespaces.id — one namespaces, many resource_link_events · optional · ON DELETE CASCADE · indexedresource_link_events.user_id → users.id — one users, many resource_link_events · required · ON DELETE SET NULL · indexedresource_link_events.child_work_item_id → issues.id — one issues, many resource_link_events · required · ON DELETE CASCADE · indexedresource_link_events.issue_id → issues.id — one issues, many resource_link_events · required · ON DELETE CASCADE · indexedresource_milestone_events.namespace_id → namespaces.id — one namespaces, many resource_milestone_events · required · ON DELETE CASCADE · indexedresource_milestone_events.merge_request_id → merge_requests.id — one merge_requests, many resource_milestone_events · optional · ON DELETE CASCADE · indexedresource_milestone_events.issue_id → issues.id — one issues, many resource_milestone_events · optional · ON DELETE CASCADE · indexedresource_milestone_events.milestone_id → milestones.id — one milestones, many resource_milestone_events · optional · ON DELETE CASCADE · indexedresource_milestone_events.user_id → users.id — one users, many resource_milestone_events · optional · ON DELETE SET NULL · indexedresource_state_events.namespace_id → namespaces.id — one namespaces, many resource_state_events · required · ON DELETE CASCADE · indexedresource_state_events.issue_id → issues.id — one issues, many resource_state_events · optional · ON DELETE CASCADE · indexedresource_state_events.merge_request_id → merge_requests.id — one merge_requests, many resource_state_events · optional · ON DELETE CASCADE · indexedresource_state_events.source_merge_request_id → merge_requests.id — one merge_requests, many resource_state_events · optional · ON DELETE SET NULL · indexedresource_state_events.epic_id → epics.id — one epics, many resource_state_events · optional · ON DELETE CASCADE · indexedresource_state_events.user_id → users.id — one users, many resource_state_events · optional · ON DELETE SET NULL · indexedresource_weight_events.namespace_id → namespaces.id — one namespaces, many resource_weight_events · optional · ON DELETE CASCADE · indexedresource_weight_events.issue_id → issues.id — one issues, many resource_weight_events · required · ON DELETE CASCADE · indexedresource_weight_events.user_id → users.id — one users, many resource_weight_events · optional · ON DELETE SET NULL · indexedtriggered_by_id — resource_iteration_events.triggered_by_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.merge_request_id — resource_iteration_events.merge_request_id may be NULL, so the relationship to merge_requests is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.issue_id — resource_iteration_events.issue_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.iteration_id — resource_iteration_events.iteration_id may be NULL, so the relationship to sprints is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.triggered_by_id, merge_request_id, issue_id, iteration_id — resource_iteration_events has several nullable FKs (triggered_by_id, merge_request_id, issue_id, iteration_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · SET NULL | ||
issue_id | bigint | null | issues.id 1:N · CASCADE | ||
merge_request_id | bigint | null | merge_requests.id 1:N · CASCADE | ||
iteration_id | bigint | null | sprints.id 1:N · CASCADE | CHECK iteration_id IS NOT NULL | |
created_at | timestamptz | NOT NULL | |||
action | smallint | NOT NULL | |||
automated | boolean | NOT NULL | FALSE | ||
triggered_by_id | bigint | null | issues.id 1:N · SET NULL | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE |
Table checks: iteration_id IS NOT NULL
issue_id — resource_label_events.issue_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.epic_id — resource_label_events.epic_id may be NULL, so the relationship to epics is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.merge_request_id — resource_label_events.merge_request_id may be NULL, so the relationship to merge_requests is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.label_id — resource_label_events.label_id may be NULL, so the relationship to labels is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id — resource_label_events.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.issue_id, epic_id, merge_request_id, label_id, user_id — resource_label_events has several nullable FKs (issue_id, epic_id, merge_request_id, label_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
action | integer | NOT NULL | |||
issue_id | bigint | null | issues.id 1:N · CASCADE | ||
merge_request_id | bigint | null | merge_requests.id 1:N · CASCADE | ||
epic_id | bigint | null | epics.id 1:N · CASCADE | ||
label_id | bigint | null | labels.id 1:N · SET NULL | ||
user_id | bigint | null | users.id 1:N · SET NULL | ||
created_at | timestamptz | NOT NULL | |||
reference | text | null | |||
imported_from | smallint | NOT NULL | 0 | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE |
Table checks: num_nonnulls(epic_id, issue_id, merge_request_id) = 1
system_note_metadata_id — resource_link_events.system_note_metadata_id may be NULL, so the relationship to system_note_metadata is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id — resource_link_events.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.system_note_metadata_id, namespace_id — resource_link_events has several nullable FKs (system_note_metadata_id, namespace_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
action | smallint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · SET NULL | ||
issue_id | bigint | NOT NULL | issues.id 1:N · CASCADE | ||
child_work_item_id | bigint | NOT NULL | issues.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
system_note_metadata_id | bigint | null | system_note_metadata.id 1:1 · CASCADE | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL
merge_request_id — resource_milestone_events.merge_request_id may be NULL, so the relationship to merge_requests is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.issue_id — resource_milestone_events.issue_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.milestone_id — resource_milestone_events.milestone_id may be NULL, so the relationship to milestones is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id — resource_milestone_events.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.merge_request_id, issue_id, milestone_id, user_id — resource_milestone_events has several nullable FKs (merge_request_id, issue_id, milestone_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | null | users.id 1:N · SET NULL | ||
issue_id | bigint | null | issues.id 1:N · CASCADE | ||
merge_request_id | bigint | null | merge_requests.id 1:N · CASCADE | ||
milestone_id | bigint | null | milestones.id 1:N · CASCADE | ||
action | smallint | NOT NULL | |||
state | smallint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
imported_from | smallint | NOT NULL | 0 | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE |
Table checks: num_nonnulls(issue_id, merge_request_id) = 1
issue_id — resource_state_events.issue_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.merge_request_id — resource_state_events.merge_request_id may be NULL, so the relationship to merge_requests is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.source_merge_request_id — resource_state_events.source_merge_request_id may be NULL, so the relationship to merge_requests is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.epic_id — resource_state_events.epic_id may be NULL, so the relationship to epics is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id — resource_state_events.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.issue_id, merge_request_id, source_merge_request_id, epic_id, user_id — resource_state_events has several nullable FKs (issue_id, merge_request_id, source_merge_request_id, epic_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | null | users.id 1:N · SET NULL | ||
issue_id | bigint | null | issues.id 1:N · CASCADE | ||
merge_request_id | bigint | null | merge_requests.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
state | smallint | NOT NULL | |||
epic_id | bigint | null | epics.id 1:N · CASCADE | ||
source_commit | text | null | CHECK char_length(source_commit) <= 40 | ||
close_after_error_tracking_resolve | boolean | NOT NULL | FALSE | ||
close_auto_resolve_prometheus_alert | boolean | NOT NULL | FALSE | ||
source_merge_request_id | bigint | null | merge_requests.id 1:N · SET NULL | ||
imported_from | smallint | NOT NULL | 0 | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE |
Table checks: num_nonnulls(epic_id, issue_id, merge_request_id) = 1; char_length(source_commit) <= 40
namespace_id — resource_weight_events.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id — resource_weight_events.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id, user_id — resource_weight_events has several nullable FKs (namespace_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | null | users.id 1:N · SET NULL | ||
issue_id | bigint | NOT NULL | issues.id 1:N · CASCADE | ||
weight | integer | null | |||
created_at | timestamptz | NOT NULL | |||
previous_weight | integer | null | |||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL
Tables whose names start with saml_.
saml_group_links.member_role_id → member_roles.id — one member_roles, many saml_group_links · optional · ON DELETE SET NULL · indexedsaml_group_links.group_id → namespaces.id — one namespaces, many saml_group_links · required · ON DELETE CASCADE · indexedsaml_providers.member_role_id → member_roles.id — one member_roles, many saml_providers · optional · ON DELETE SET NULL · indexedsaml_providers.group_id → namespaces.id — one namespaces, many saml_providers · required · ON DELETE CASCADE · indexedmember_role_id — saml_group_links.member_role_id may be NULL, so the relationship to member_roles is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
access_level | smallint | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
saml_group_name | text | NOT NULL | CHECK char_length(saml_group_name) <= 255 | ||
member_role_id | bigint | null | member_roles.id 1:N · SET NULL | ||
assign_duo_seats | boolean | NOT NULL | FALSE | ||
scim_group_uid | uuid | null | |||
provider | text | null | CHECK char_length(provider) <= 255 |
Table checks: char_length(saml_group_name) <= 255; char_length(provider) <= 255
member_role_id — saml_providers.member_role_id may be NULL, so the relationship to member_roles is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
enabled | boolean | NOT NULL | |||
certificate_fingerprint | varchar | NOT NULL | |||
sso_url | varchar | NOT NULL | |||
enforced_sso | boolean | NOT NULL | FALSE | ||
enforced_group_managed_accounts | boolean | NOT NULL | FALSE | ||
prohibited_outer_forks | boolean | NOT NULL | TRUE | ||
default_membership_role | smallint | NOT NULL | 10 | ||
git_check_enforced | boolean | NOT NULL | FALSE | ||
member_role_id | bigint | null | member_roles.id 1:N · SET NULL | ||
disable_password_authentication_for_enterprise_users | boolean | null | FALSE |
Tables whose names start with saved_.
saved_replies.user_id → users.id — one users, many saved_replies · required · ON DELETE CASCADE · indexedsaved_views.created_by_id → users.id — one users, many saved_views · optional · ON DELETE SET NULL · indexedsaved_views.updated_by_id → users.id — one users, many saved_views · optional · ON DELETE SET NULL · indexedsaved_views.namespace_id → namespaces.id — one namespaces, many saved_views · required · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
content | text | NOT NULL | CHECK char_length(content) <= 10000 |
Table checks: char_length(content) <= 10000; char_length(name) <= 255
created_by_id — saved_views.created_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.updated_by_id — saved_views.updated_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.created_by_id, updated_by_id — saved_views has several nullable FKs (created_by_id, updated_by_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_by_id | bigint | null | users.id 1:N · SET NULL | ||
version | integer | NOT NULL | |||
private | boolean | NOT NULL | TRUE | ||
name | text | NOT NULL | CHECK char_length(name) <= 140 | ||
description | text | null | CHECK char_length(description) <= 140 | ||
sort | smallint | null | |||
filter_data | jsonb | null | |||
display_settings | jsonb | null | |||
lock_version | integer | NOT NULL | 0 | ||
updated_by_id | bigint | null | users.id 1:N · SET NULL |
Table checks: char_length(name) <= 140; char_length(description) <= 140
Tables whose names start with sbom_.
sbom_component_versions.component_id → sbom_components.id — one sbom_components, many sbom_component_versions · required · ON DELETE CASCADE · indexedsbom_graph_paths.ancestor_id → sbom_occurrences.id — one sbom_occurrences, many sbom_graph_paths · required · ON DELETE CASCADE · indexedsbom_graph_paths.descendant_id → sbom_occurrences.id — one sbom_occurrences, many sbom_graph_paths · required · ON DELETE CASCADE · indexedsbom_occurrence_refs.sbom_occurrence_id → sbom_occurrences.id — one sbom_occurrences, many sbom_occurrence_refs · required · ON DELETE CASCADE · indexedsbom_occurrence_refs.security_project_tracked_context_id → security_project_tracked_contexts.id — one security_project_tracked_contexts, many sbom_occurrence_refs · required · ON DELETE CASCADE · indexedsbom_occurrences.component_version_id → sbom_component_versions.id — one sbom_component_versions, many sbom_occurrences · optional · ON DELETE CASCADE · indexedsbom_occurrences.source_package_id → sbom_source_packages.id — one sbom_source_packages, many sbom_occurrences · optional · ON DELETE CASCADE · indexedsbom_occurrences.source_id → sbom_sources.id — one sbom_sources, many sbom_occurrences · optional · ON DELETE CASCADE · indexedsbom_occurrences.component_id → sbom_components.id — one sbom_components, many sbom_occurrences · required · ON DELETE CASCADE · indexedsbom_occurrences_vulnerabilities.vulnerability_id → vulnerabilities.id — one vulnerabilities, many sbom_occurrences_vulnerabilities · required · ON DELETE CASCADE · indexedsbom_occurrences_vulnerabilities.sbom_occurrence_ref_id → sbom_occurrence_refs.id — one sbom_occurrence_refs, many sbom_occurrences_vulnerabilities · optional · ON DELETE CASCADE · indexedsbom_occurrences_vulnerabilities.sbom_occurrence_id → sbom_occurrences.id — one sbom_occurrences, many sbom_occurrences_vulnerabilities · required · ON DELETE CASCADE · indexedsbom_vulnerability_scans.sbom_vulnerability_scan_result_id → sbom_vulnerability_scan_results.id — one sbom_vulnerability_scan_results, many sbom_vulnerability_scans · optional · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
component_id | bigint | NOT NULL | sbom_components.id 1:N · CASCADE | ||
version | text | NOT NULL | CHECK char_length(version) <= 255 | ||
source_package_name | text | null | CHECK char_length(source_package_name) <= 255 | ||
organization_id | bigint | NOT NULL |
Table checks: char_length(source_package_name) <= 255; char_length(version) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
component_type | smallint | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
purl_type | smallint | null | |||
organization_id | bigint | NOT NULL |
Table checks: char_length(name) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
ancestor_id | bigint | NOT NULL | sbom_occurrences.id 1:N · CASCADE | ||
descendant_id | bigint | NOT NULL | sbom_occurrences.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | |||
path_length | integer | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
top_level_ancestor | boolean | NOT NULL | FALSE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
sbom_occurrence_id | bigint | NOT NULL | sbom_occurrences.id 1:N · CASCADE | ||
security_project_tracked_context_id | bigint | NOT NULL | security_project_tracked_contexts.id 1:N · CASCADE | ||
commit_sha | bytea | NOT NULL | |||
pipeline_id | bigint | null |
component_version_id — sbom_occurrences.component_version_id may be NULL, so the relationship to sbom_component_versions is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.source_package_id — sbom_occurrences.source_package_id may be NULL, so the relationship to sbom_source_packages is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.source_id — sbom_occurrences.source_id may be NULL, so the relationship to sbom_sources is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.component_version_id, source_package_id, source_id — sbom_occurrences has several nullable FKs (component_version_id, source_package_id, source_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
component_version_id | bigint | null | sbom_component_versions.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | |||
pipeline_id | bigint | null | |||
source_id | bigint | null | sbom_sources.id 1:N · CASCADE | ||
commit_sha | bytea | NOT NULL | |||
component_id | bigint | NOT NULL | sbom_components.id 1:N · CASCADE | ||
uuid | uuid | NOT NULL | |||
package_manager | text | null | CHECK char_length(package_manager) <= 255 | ||
component_name | text | null | CHECK char_length(component_name) <= 255 | ||
input_file_path | text | null | CHECK char_length(input_file_path) <= 1024 | ||
licenses | jsonb | null | CAST('[]' AS jsonb) | ||
highest_severity | smallint | null | |||
vulnerability_count | integer | NOT NULL | 0 | ||
source_package_id | bigint | null | sbom_source_packages.id 1:N · CASCADE | ||
archived | boolean | NOT NULL | FALSE | ||
traversal_ids | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | ||
ancestors | jsonb | NOT NULL | CAST('[]' AS jsonb) | ||
reachability | smallint | null | 0 | ||
partition_id | bigint | null | 1 |
Table checks: char_length(package_manager) <= 255; char_length(component_name) <= 255; char_length(input_file_path) <= 1024
sbom_occurrence_ref_id — sbom_occurrences_vulnerabilities.sbom_occurrence_ref_id may be NULL, so the relationship to sbom_occurrence_refs is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
sbom_occurrence_id | bigint | NOT NULL | sbom_occurrences.id 1:N · CASCADE | ||
vulnerability_id | bigint | NOT NULL | vulnerabilities.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | null | CHECK project_id IS NOT NULL | ||
vulnerability_occurrence_id | bigint | null | |||
sbom_occurrence_ref_id | bigint | null | sbom_occurrence_refs.id 1:N · CASCADE |
Table checks: project_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
purl_type | smallint | NOT NULL | |||
created_at | timestamptz | NOT NULL | now() | ||
updated_at | timestamptz | NOT NULL | now() | ||
organization_id | bigint | NOT NULL |
Table checks: char_length(name) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
source_type | smallint | NOT NULL | |||
source | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
organization_id | bigint | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | |||
result_file_store | smallint | null | 1 | ||
result_file | text | null | CHECK char_length(result_file) <= 255 |
Table checks: char_length(result_file) <= 255
sbom_vulnerability_scan_result_id — sbom_vulnerability_scans.sbom_vulnerability_scan_result_id may be NULL, so the relationship to sbom_vulnerability_scan_results is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | |||
build_id | bigint | NOT NULL | |||
status | smallint | null | 0 | ||
sbom_file_store | smallint | null | 1 | ||
result_file_store | smallint | null | 1 | ||
sbom_file | text | null | CHECK char_length(sbom_file) <= 255 | ||
sbom_file_final_path | text | null | CHECK char_length(sbom_file_final_path) <= 1024 | ||
result_file | text | null | CHECK char_length(result_file) <= 255 | ||
error_message | text | null | CHECK char_length(error_message) <= 1024 | ||
sbom_vulnerability_scan_result_id | bigint | null | sbom_vulnerability_scan_results.id 1:N · CASCADE | ||
sbom_digest | text | null | CHECK char_length(sbom_digest) <= 1024 |
Table checks: char_length(error_message) <= 1024; char_length(sbom_file_final_path) <= 1024; char_length(sbom_digest) <= 1024; char_length(result_file) <= 255; char_length(sbom_file) <= 255
Tables whose names start with scan_.
scan_execution_policy_rules.security_policy_management_project_id → projects.id — one projects, many scan_execution_policy_rules · required · ON DELETE CASCADE · indexedscan_execution_policy_rules.security_policy_id → security_policies.id — one security_policies, many scan_execution_policy_rules · required · ON DELETE CASCADE · indexedscan_result_policies.approval_policy_rule_id → approval_policy_rules.id — one approval_policy_rules, many scan_result_policies · optional · ON DELETE CASCADE · indexedscan_result_policies.project_id → projects.id — one projects, many scan_result_policies · optional · ON DELETE CASCADE · indexedscan_result_policies.namespace_id → namespaces.id — one namespaces, many scan_result_policies · optional · ON DELETE CASCADE · indexedscan_result_policies.security_orchestration_policy_configuration_id → security_orchestration_policy_configurations.id — one security_orchestration_policy_configurations, many scan_result_policies · required · ON DELETE CASCADE · indexedscan_result_policy_violation_details.project_id → projects.id — one projects, many scan_result_policy_violation_details · required · ON DELETE CASCADE · indexedscan_result_policy_violation_details.scan_result_policy_violation_id → scan_result_policy_violations.id — one scan_result_policy_violations, many scan_result_policy_violation_details · required · ON DELETE CASCADE · indexedscan_result_policy_violations.merge_request_id → merge_requests.id — one merge_requests, many scan_result_policy_violations · required · ON DELETE CASCADE · indexedscan_result_policy_violations.approval_policy_rule_id → approval_policy_rules.id — one approval_policy_rules, many scan_result_policy_violations · optional · ON DELETE CASCADE · indexedscan_result_policy_violations.project_id → projects.id — one projects, many scan_result_policy_violations · required · ON DELETE CASCADE · indexedscan_result_policy_violations.scan_result_policy_id → scan_result_policies.id — one scan_result_policies, many scan_result_policy_violations · optional · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
security_policy_id | bigint | NOT NULL | security_policies.id 1:N · CASCADE | ||
security_policy_management_project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
rule_index | smallint | NOT NULL | |||
type | smallint | NOT NULL | |||
content | jsonb | NOT NULL | CAST('{}' AS jsonb) |
approval_policy_rule_id — scan_result_policies.approval_policy_rule_id may be NULL, so the relationship to approval_policy_rules is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — scan_result_policies.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id — scan_result_policies.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.approval_policy_rule_id, project_id, namespace_id — scan_result_policies has several nullable FKs (approval_policy_rule_id, project_id, namespace_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
security_orchestration_policy_configuration_id | bigint | NOT NULL | security_orchestration_policy_configurations.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
orchestration_policy_idx | smallint | NOT NULL | |||
license_states | text[] | null | CAST('{}' AS text[]) | ||
match_on_inclusion | boolean | null | |||
role_approvers | integer[] | null | CAST('{}' AS integer[]) | ||
age_value | integer | null | CHECK age_value IS NULL OR age_value >= 0 | ||
age_operator | smallint | null | |||
age_interval | smallint | null | |||
vulnerability_attributes | jsonb | null | CAST('{}' AS jsonb) | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
rule_idx | smallint | null | CHECK rule_idx IS NULL OR rule_idx >= 0 | ||
project_approval_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
commits | smallint | null | |||
send_bot_message | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
fallback_behavior | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
policy_tuning | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
action_idx | smallint | NOT NULL | 0 | ||
custom_roles | bigint[] | NOT NULL | CAST('{}' AS bigint[]) | CHECK array_position(custom_roles, CAST(NULL AS bigint)) IS NULL | |
licenses | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
approval_policy_rule_id | bigint | null | approval_policy_rules.id 1:N · CASCADE | ||
malicious_rule | boolean | null |
Table checks: age_value IS NULL OR age_value >= 0; num_nonnulls(namespace_id, project_id) >= 1; rule_idx IS NULL OR rule_idx >= 0; array_position(custom_roles, CAST(NULL AS bigint)) IS NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
scan_result_policy_violation_id | bigint | NOT NULL | scan_result_policy_violations.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
policy_rule_type | smallint | NOT NULL | |||
finding_state | smallint | null | |||
finding_uuid | text | null | CHECK char_length(finding_uuid) <= 50 | ||
license_name | text | null | CHECK char_length(license_name) <= 255 | ||
dependencies | text[] | NOT NULL | CAST('{}' AS text[]) | ||
commit_shas | text[] | NOT NULL | CAST('{}' AS text[]) | ||
metadata | jsonb | NOT NULL | CAST('{}' AS jsonb) |
Table checks: char_length(license_name) <= 255; char_length(finding_uuid) <= 50
approval_policy_rule_id — scan_result_policy_violations.approval_policy_rule_id may be NULL, so the relationship to approval_policy_rules is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.scan_result_policy_id — scan_result_policy_violations.scan_result_policy_id may be NULL, so the relationship to scan_result_policies is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
scan_result_policy_id | bigint | null | scan_result_policies.id 1:N · CASCADE | ||
merge_request_id | bigint | NOT NULL | merge_requests.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
violation_data | jsonb | null | |||
approval_policy_rule_id | bigint | null | approval_policy_rules.id 1:N · CASCADE | ||
status | smallint | NOT NULL | 1 |
Table checks: approval_policy_rule_id IS NOT NULL OR scan_result_policy_id IS NOT NULL
Tables whose names start with scim_.
scim_group_memberships.user_id → users.id — one users, many scim_group_memberships · required · ON DELETE CASCADE · indexedscim_identities.group_id → namespaces.id — one namespaces, many scim_identities · optional · ON DELETE CASCADE · indexedscim_identities.user_id → users.id — one users, many scim_identities · required · ON DELETE CASCADE · indexedscim_oauth_access_tokens.organization_id → organizations.id — one organizations, many scim_oauth_access_tokens · optional · ON DELETE CASCADE · indexedscim_oauth_access_tokens.group_id → namespaces.id — one namespaces, many scim_oauth_access_tokens · optional · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
scim_group_uid | uuid | NOT NULL |
group_id — scim_identities.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
active | boolean | null | FALSE | ||
extern_uid | varchar(255) | NOT NULL |
organization_id — scim_oauth_access_tokens.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — scim_oauth_access_tokens.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id, group_id — scim_oauth_access_tokens has several nullable FKs (organization_id, group_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
token_encrypted | varchar | NOT NULL | |||
organization_id | bigint | null | organizations.id 1:N · CASCADE | CHECK organization_id IS NOT NULL |
Table checks: organization_id IS NOT NULL
Tables whose names start with secret_.
secret_detection_token_statuses.vulnerability_occurrence_id → vulnerability_occurrences.id — one vulnerability_occurrences, at most one secret_detection_token_statuses · required · ON DELETE CASCADE · indexedsecret_rotation_infos.project_id → projects.id — one projects, many secret_rotation_infos · required · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
vulnerability_occurrence_id PK | bigint | NOT NULL | vulnerability_occurrences.id 1:1 · CASCADE | ||
project_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
status | smallint | NOT NULL | 0 | ||
last_verified_at | timestamptz | null |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
secret_name | text | NOT NULL | CHECK char_length(secret_name) <= 255 | ||
secret_metadata_version | integer | NOT NULL | |||
rotation_interval_days | integer | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
next_reminder_at | timestamptz | NOT NULL | |||
last_reminder_at | timestamptz | null |
Table checks: char_length(secret_name) <= 255
Tables whose names start with secrets_.
secrets_manager_namespace_enrollments.namespace_id → namespaces.id — one namespaces, at most one secrets_manager_namespace_enrollments · required · ON DELETE CASCADE · indexedsecrets_management_recovery_keys references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
active | boolean | NOT NULL | FALSE | ||
key | jsonb | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
beta | boolean | NOT NULL | TRUE |
Tables whose names start with service_.
service_desk_custom_email_credentials.project_id → projects.id — one projects, at most one service_desk_custom_email_credentials · required · ON DELETE CASCADE · indexedservice_desk_custom_email_verifications.triggerer_id → users.id — one users, many service_desk_custom_email_verifications · optional · ON DELETE SET NULL · indexedservice_desk_custom_email_verifications.project_id → projects.id — one projects, at most one service_desk_custom_email_verifications · required · ON DELETE CASCADE · indexedservice_desk_settings.file_template_project_id → projects.id — one projects, many service_desk_settings · optional · ON DELETE SET NULL · indexedservice_desk_settings.project_id → projects.id — one projects, at most one service_desk_settings · required · ON DELETE CASCADE · indexedservice_access_tokens references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
encrypted_token | bytea | NOT NULL | |||
encrypted_token_iv | bytea | NOT NULL | |||
expires_at | timestamptz | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | projects.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
smtp_port | integer | null | |||
smtp_address | text | null | CHECK char_length(smtp_address) <= 255 | ||
encrypted_smtp_username | bytea | null | |||
encrypted_smtp_username_iv | bytea | null | |||
encrypted_smtp_password | bytea | null | |||
encrypted_smtp_password_iv | bytea | null | |||
smtp_authentication | smallint | null |
Table checks: char_length(smtp_address) <= 255
triggerer_id — service_desk_custom_email_verifications.triggerer_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | projects.id 1:1 · CASCADE | ||
triggerer_id | bigint | null | users.id 1:N · SET NULL | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
triggered_at | timestamptz | null | |||
state | smallint | NOT NULL | 0 | ||
error | smallint | null | |||
encrypted_token | bytea | null | |||
encrypted_token_iv | bytea | null |
file_template_project_id — service_desk_settings.file_template_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | projects.id 1:1 · CASCADE | ||
issue_template_key | varchar(255) | null | |||
outgoing_name | varchar(255) | null | |||
project_key | varchar(255) | null | |||
file_template_project_id | bigint | null | projects.id 1:N · SET NULL | ||
custom_email_enabled | boolean | NOT NULL | FALSE | ||
custom_email | text | null | CHECK char_length(custom_email) <= 255 | ||
service_desk_enabled | boolean | NOT NULL | TRUE | ||
add_external_participants_from_cc | boolean | NOT NULL | FALSE | ||
reopen_issue_on_external_participant_note | boolean | NOT NULL | FALSE | ||
tickets_confidential_by_default | boolean | NOT NULL | TRUE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_key_address_slug | text | null | CHECK char_length(project_key_address_slug) <= 500 |
Table checks: char_length(project_key_address_slug) <= 500; char_length(custom_email) <= 255
Tables whose names start with slack_.
slack_api_scopes.organization_id → organizations.id — one organizations, many slack_api_scopes · optional · ON DELETE CASCADE · indexedslack_integrations.project_id → projects.id — one projects, many slack_integrations · optional · ON DELETE CASCADE · indexedslack_integrations.group_id → namespaces.id — one namespaces, many slack_integrations · optional · ON DELETE CASCADE · indexedslack_integrations.integration_id → integrations.id — one integrations, many slack_integrations · optional · ON DELETE CASCADE · indexedslack_integrations.organization_id → organizations.id — one organizations, many slack_integrations · optional · ON DELETE CASCADE · indexedslack_integrations_scopes.organization_id → organizations.id — one organizations, many slack_integrations_scopes · optional · ON DELETE CASCADE · indexedslack_integrations_scopes.group_id → namespaces.id — one namespaces, many slack_integrations_scopes · optional · ON DELETE CASCADE · indexedslack_integrations_scopes.project_id → projects.id — one projects, many slack_integrations_scopes · optional · ON DELETE CASCADE · indexedslack_integrations_scopes.slack_api_scope_id → slack_api_scopes.id — one slack_api_scopes, many slack_integrations_scopes · required · ON DELETE CASCADE · indexedslack_integrations_scopes.slack_integration_id → slack_integrations.id — one slack_integrations, many slack_integrations_scopes · required · ON DELETE CASCADE · indexedorganization_id — slack_api_scopes.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 100 | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE | CHECK organization_id IS NOT NULL |
Table checks: char_length(name) <= 100; organization_id IS NOT NULL
project_id — slack_integrations.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — slack_integrations.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.integration_id — slack_integrations.integration_id may be NULL, so the relationship to integrations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id — slack_integrations.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id, group_id, integration_id, organization_id — slack_integrations has several nullable FKs (project_id, group_id, integration_id, organization_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
team_id | varchar | NOT NULL | |||
team_name | varchar | NOT NULL | |||
alias | varchar | NOT NULL | |||
user_id | varchar | NOT NULL | |||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL | |||
bot_user_id | text | null | CHECK char_length(bot_user_id) <= 255 | ||
encrypted_bot_access_token | bytea | null | |||
encrypted_bot_access_token_iv | bytea | null | |||
integration_id | bigint | null | integrations.id 1:N · CASCADE | CHECK integration_id IS NOT NULL | |
organization_id | bigint | null | organizations.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE |
Table checks: num_nonnulls(group_id, organization_id, project_id) = 1; char_length(bot_user_id) <= 255; integration_id IS NOT NULL
organization_id — slack_integrations_scopes.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — slack_integrations_scopes.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — slack_integrations_scopes.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
slack_api_scope_id | bigint | NOT NULL | slack_api_scopes.id 1:N · CASCADE | ||
slack_integration_id | bigint | NOT NULL | slack_integrations.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE |
Table checks: num_nonnulls(group_id, organization_id, project_id) = 1
Tables whose names start with status_.
status_check_responses.external_status_check_id → external_status_checks.id — one external_status_checks, many status_check_responses · required · ON DELETE CASCADE · indexedstatus_check_responses.project_id → projects.id — one projects, many status_check_responses · optional · ON DELETE CASCADE · indexedstatus_check_responses.merge_request_id → merge_requests.id — one merge_requests, many status_check_responses · required · ON DELETE CASCADE · indexedstatus_page_published_incidents.namespace_id → namespaces.id — one namespaces, many status_page_published_incidents · optional · ON DELETE CASCADE · indexedstatus_page_published_incidents.issue_id → issues.id — one issues, at most one status_page_published_incidents · required · ON DELETE CASCADE · indexedstatus_page_settings.project_id → projects.id — one projects, at most one status_page_settings · required · ON DELETE CASCADE · indexedproject_id — status_check_responses.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
merge_request_id | bigint | NOT NULL | merge_requests.id 1:N · CASCADE | ||
external_approval_rule_id | bigint | null | |||
sha | bytea | NOT NULL | |||
external_status_check_id | bigint | NOT NULL | external_status_checks.id 1:N · CASCADE | ||
status | smallint | NOT NULL | 0 | ||
retried_at | timestamptz | null | |||
created_at | timestamptz | NOT NULL | now() | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
namespace_id — status_page_published_incidents.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
issue_id | bigint | NOT NULL | issues.id 1:1 · CASCADE | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | projects.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
enabled | boolean | NOT NULL | FALSE | ||
aws_s3_bucket_name | varchar(63) | NOT NULL | |||
aws_region | varchar(255) | NOT NULL | |||
aws_access_key | varchar(255) | NOT NULL | |||
encrypted_aws_secret_key | varchar(255) | NOT NULL | |||
encrypted_aws_secret_key_iv | varchar(255) | NOT NULL | |||
status_page_url | text | null | CHECK char_length(status_page_url) <= 1024 |
Table checks: char_length(status_page_url) <= 1024
Tables whose names start with subscription_.
subscription_add_on_purchases.subscription_add_on_id → subscription_add_ons.id — one subscription_add_ons, many subscription_add_on_purchases · required · ON DELETE CASCADE · indexedsubscription_add_on_purchases.namespace_id → namespaces.id — one namespaces, many subscription_add_on_purchases · optional · ON DELETE CASCADE · indexedsubscription_add_on_purchases.organization_id → organizations.id — one organizations, many subscription_add_on_purchases · required · ON DELETE CASCADE · indexedsubscription_seat_assignments.namespace_id → namespaces.id — one namespaces, many subscription_seat_assignments · optional · ON DELETE CASCADE · indexedsubscription_seat_assignments.user_id → users.id — one users, many subscription_seat_assignments · required · ON DELETE CASCADE · indexedsubscription_seat_assignments.organization_id → organizations.id — one organizations, many subscription_seat_assignments · required · ON DELETE CASCADE · indexedsubscription_user_add_on_assignment_versions.organization_id → organizations.id — one organizations, many subscription_user_add_on_assignment_versions · required · ON DELETE CASCADE · indexedsubscription_user_add_on_assignments.add_on_purchase_id → subscription_add_on_purchases.id — one subscription_add_on_purchases, many subscription_user_add_on_assignments · required · ON DELETE CASCADE · indexedsubscription_user_add_on_assignments.user_id → users.id — one users, many subscription_user_add_on_assignments · required · ON DELETE CASCADE · indexedsubscription_user_add_on_assignments.organization_id → organizations.id — one organizations, many subscription_user_add_on_assignments · optional · ON DELETE CASCADE · indexednamespace_id — subscription_add_on_purchases.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
subscription_add_on_id | bigint | NOT NULL | subscription_add_ons.id 1:N · CASCADE | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
quantity | integer | NOT NULL | |||
expires_on | date | NOT NULL | |||
purchase_xid | text | NOT NULL | CHECK char_length(purchase_xid) <= 255 | ||
last_assigned_users_refreshed_at | timestamptz | null | |||
trial | boolean | NOT NULL | FALSE | ||
started_at | date | null | CHECK started_at IS NOT NULL | ||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
subscription_add_on_uid | smallint | null | CHECK subscription_add_on_uid IS NOT NULL |
Table checks: char_length(purchase_xid) <= 255; subscription_add_on_uid IS NOT NULL; started_at IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
name | smallint | NOT NULL | |||
description | text | NOT NULL | CHECK char_length(description) <= 512 |
Table checks: char_length(description) <= 512
namespace_id — subscription_seat_assignments.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
last_activity_on | timestamptz | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
seat_type | smallint | null |
item_type, item_id — item_id points at different tables depending on item_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
item_id | bigint | null | |||
purchase_id | bigint | null | |||
user_id | bigint | null | |||
created_at | timestamptz | null | |||
item_type | text | NOT NULL | CHECK char_length(item_type) <= 255 | ||
event | text | NOT NULL | CHECK char_length(event) <= 255 | ||
namespace_path | text | null | CHECK char_length(namespace_path) <= 255 | ||
add_on_name | text | null | CHECK char_length(add_on_name) <= 255 | ||
whodunnit | text | null | CHECK char_length(whodunnit) <= 255 | ||
object | jsonb | null |
Table checks: char_length(item_type) <= 255; char_length(event) <= 255; char_length(namespace_path) <= 255; char_length(add_on_name) <= 255; char_length(whodunnit) <= 255
organization_id — subscription_user_add_on_assignments.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
add_on_purchase_id | bigint | NOT NULL | subscription_add_on_purchases.id 1:N · CASCADE | ||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
organization_id | bigint | null | organizations.id 1:N · CASCADE | CHECK organization_id IS NOT NULL |
Table checks: organization_id IS NOT NULL
Tables whose names start with system_.
system_access_group_microsoft_applications.group_id → namespaces.id — one namespaces, at most one system_access_group_microsoft_applications · required · ON DELETE CASCADE · indexedsystem_access_group_microsoft_graph_access_tokens.system_access_group_microsoft_application_id → system_access_group_microsoft_applications.id — one system_access_group_microsoft_applications, at most one system_access_group_microsoft_graph_access_tokens · optional · ON DELETE CASCADE · indexedsystem_access_group_microsoft_graph_access_tokens.group_id → namespaces.id — one namespaces, many system_access_group_microsoft_graph_access_tokens · required · ON DELETE CASCADE · indexedsystem_access_microsoft_applications.namespace_id → namespaces.id — one namespaces, at most one system_access_microsoft_applications · optional · ON DELETE CASCADE · indexedsystem_access_microsoft_graph_access_tokens.system_access_microsoft_application_id → system_access_microsoft_applications.id — one system_access_microsoft_applications, at most one system_access_microsoft_graph_access_tokens · optional · ON DELETE CASCADE · indexedsystem_note_metadata.namespace_id → namespaces.id — one namespaces, many system_note_metadata · optional · ON DELETE CASCADE · indexedsystem_note_metadata.note_id → notes.id — one notes, at most one system_note_metadata · required · ON DELETE CASCADE · indexedsystem_note_metadata.description_version_id → description_versions.id — one description_versions, many system_note_metadata · optional · ON DELETE SET NULL · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
enabled | boolean | NOT NULL | FALSE | ||
tenant_xid | text | NOT NULL | CHECK char_length(tenant_xid) <= 255 | ||
client_xid | text | NOT NULL | CHECK char_length(client_xid) <= 255 | ||
login_endpoint | text | NOT NULL | CAST('https://login.microsoftonline.com' AS text) | CHECK char_length(login_endpoint) <= 255 | |
graph_endpoint | text | NOT NULL | CAST('https://graph.microsoft.com' AS text) | CHECK char_length(graph_endpoint) <= 255 | |
encrypted_client_secret | bytea | NOT NULL | |||
encrypted_client_secret_iv | bytea | NOT NULL |
Table checks: char_length(graph_endpoint) <= 255; char_length(login_endpoint) <= 255; char_length(tenant_xid) <= 255; char_length(client_xid) <= 255
system_access_group_microsoft_application_id — system_access_group_microsoft_graph_access_tokens.system_access_group_microsoft_application_id may be NULL, so the relationship to system_access_group_microsoft_applications is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
system_access_group_microsoft_application_id | bigint | null | system_access_group_microsoft_applications.id 1:1 · CASCADE | ||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
expires_in | integer | NOT NULL | |||
encrypted_token | bytea | NOT NULL | |||
encrypted_token_iv | bytea | NOT NULL |
namespace_id — system_access_microsoft_applications.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
namespace_id | bigint | null | namespaces.id 1:1 · CASCADE | ||
enabled | boolean | NOT NULL | FALSE | ||
tenant_xid | text | NOT NULL | CHECK char_length(tenant_xid) <= 255 | ||
client_xid | text | NOT NULL | CHECK char_length(client_xid) <= 255 | ||
login_endpoint | text | NOT NULL | CAST('https://login.microsoftonline.com' AS text) | CHECK char_length(login_endpoint) <= 255 | |
graph_endpoint | text | NOT NULL | CAST('https://graph.microsoft.com' AS text) | CHECK char_length(graph_endpoint) <= 255 | |
encrypted_client_secret | bytea | NOT NULL | |||
encrypted_client_secret_iv | bytea | NOT NULL |
Table checks: char_length(login_endpoint) <= 255; char_length(tenant_xid) <= 255; char_length(graph_endpoint) <= 255; char_length(client_xid) <= 255
system_access_microsoft_application_id — system_access_microsoft_graph_access_tokens.system_access_microsoft_application_id may be NULL, so the relationship to system_access_microsoft_applications is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
system_access_microsoft_application_id | bigint | null | system_access_microsoft_applications.id 1:1 · CASCADE | ||
expires_in | integer | NOT NULL | |||
encrypted_token | bytea | NOT NULL | |||
encrypted_token_iv | bytea | NOT NULL |
namespace_id — system_note_metadata.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.description_version_id — system_note_metadata.description_version_id may be NULL, so the relationship to description_versions is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id, description_version_id — system_note_metadata has several nullable FKs (namespace_id, description_version_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
commit_count | integer | null | |||
action | varchar | null | |||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL | |||
description_version_id | bigint | null | description_versions.id 1:N · SET NULL | ||
note_id | bigint | NOT NULL | notes.id 1:1 · CASCADE | ||
id PK | bigint | NOT NULL | |||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL
Tables whose names start with tag_.
tag_gpg_signatures.gpg_key_subkey_id → gpg_key_subkeys.id — one gpg_key_subkeys, many tag_gpg_signatures · optional · ON DELETE CASCADE · indexedtag_gpg_signatures.gpg_key_id → gpg_keys.id — one gpg_keys, many tag_gpg_signatures · optional · ON DELETE SET NULL · indexedtag_gpg_signatures.project_id → projects.id — one projects, many tag_gpg_signatures · required · ON DELETE CASCADE · indexedtag_ssh_signatures.project_id → projects.id — one projects, many tag_ssh_signatures · required · ON DELETE CASCADE · indexedtag_ssh_signatures.key_id → keys.id — one keys, many tag_ssh_signatures · optional · ON DELETE SET NULL · indexedtag_x509_signatures.x509_certificate_id → x509_certificates.id — one x509_certificates, many tag_x509_signatures · required · ON DELETE CASCADE · indexedtag_x509_signatures.project_id → projects.id — one projects, many tag_x509_signatures · required · ON DELETE CASCADE · indexedgpg_key_subkey_id — tag_gpg_signatures.gpg_key_subkey_id may be NULL, so the relationship to gpg_key_subkeys is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.gpg_key_id — tag_gpg_signatures.gpg_key_id may be NULL, so the relationship to gpg_keys is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.gpg_key_subkey_id, gpg_key_id — tag_gpg_signatures has several nullable FKs (gpg_key_subkey_id, gpg_key_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
gpg_key_id | bigint | null | gpg_keys.id 1:N · SET NULL | ||
gpg_key_subkey_id | bigint | null | gpg_key_subkeys.id 1:N · CASCADE | ||
verification_status | smallint | NOT NULL | 0 | ||
object_name | bytea | NOT NULL | |||
gpg_key_primary_keyid | bytea | NOT NULL | |||
gpg_key_user_name | text | null | CHECK char_length(gpg_key_user_name) <= 255 | ||
gpg_key_user_email | text | null | CHECK char_length(gpg_key_user_email) <= 255 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
Table checks: char_length(gpg_key_user_name) <= 255; char_length(gpg_key_user_email) <= 255
key_id — tag_ssh_signatures.key_id may be NULL, so the relationship to keys is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
key_id | bigint | null | keys.id 1:N · SET NULL | ||
verification_status | smallint | NOT NULL | 0 | ||
object_name | bytea | NOT NULL | |||
key_fingerprint_sha256 | bytea | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
x509_certificate_id | bigint | NOT NULL | x509_certificates.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
verification_status | smallint | NOT NULL | 0 | ||
object_name | bytea | NOT NULL |
Tables whose names start with targeted_.
targeted_message_dismissals.namespace_id → namespaces.id — one namespaces, many targeted_message_dismissals · required · ON DELETE CASCADE · indexedtargeted_message_dismissals.user_id → users.id — one users, many targeted_message_dismissals · required · ON DELETE CASCADE · indexedtargeted_message_dismissals.targeted_message_id → targeted_messages.id — one targeted_messages, many targeted_message_dismissals · required · ON DELETE CASCADE · indexedtargeted_message_namespaces.namespace_id → namespaces.id — one namespaces, many targeted_message_namespaces · required · ON DELETE CASCADE · indexedtargeted_message_namespaces.targeted_message_id → targeted_messages.id — one targeted_messages, many targeted_message_namespaces · required · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
targeted_message_id | bigint | NOT NULL | targeted_messages.id 1:N · CASCADE | ||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
targeted_message_id | bigint | NOT NULL | targeted_messages.id 1:N · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
target_type | smallint | NOT NULL | 0 | ||
starts_at | timestamptz | NOT NULL | |||
ends_at | timestamptz | NOT NULL | |||
roles | smallint[] | NOT NULL | CAST('{}' AS smallint[]) |
Table checks: starts_at < ends_at
Tables whose names start with terraform_.
terraform_state_protection_rules.project_id → projects.id — one projects, many terraform_state_protection_rules · required · ON DELETE CASCADE · indexedterraform_state_versions.project_id → projects.id — one projects, many terraform_state_versions · optional · ON DELETE CASCADE · indexedterraform_state_versions.created_by_user_id → users.id — one users, many terraform_state_versions · optional · ON DELETE SET NULL · indexedterraform_state_versions.terraform_state_id → terraform_states.id — one terraform_states, many terraform_state_versions · required · ON DELETE CASCADE · indexedterraform_states.locked_by_user_id → users.id — one users, many terraform_states · optional · ON DELETE SET NULL · indexedterraform_states.project_id → projects.id — one projects, many terraform_states · required · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
minimum_access_level_for_write | smallint | NOT NULL | |||
allowed_from | smallint | NOT NULL | 0 | ||
state_name | text | NOT NULL | CHECK char_length(state_name) <= 255 |
Table checks: char_length(state_name) <= 255
terraform_state_version_states references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
terraform_state_version_id | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 |
Table checks: char_length(verification_failure) <= 255
project_id — terraform_state_versions.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.created_by_user_id — terraform_state_versions.created_by_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id, created_by_user_id — terraform_state_versions has several nullable FKs (project_id, created_by_user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
terraform_state_id | bigint | NOT NULL | terraform_states.id 1:N · CASCADE | ||
created_by_user_id | bigint | null | users.id 1:N · SET NULL | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
version | integer | NOT NULL | |||
file_store | smallint | NOT NULL | |||
file | text | NOT NULL | CHECK char_length(file) <= 255 | ||
verification_retry_count | smallint | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 | ||
ci_build_id | bigint | null | |||
verification_started_at | timestamptz | null | |||
verification_state | smallint | NOT NULL | 0 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL | |
is_encrypted | boolean | NOT NULL | TRUE |
Table checks: char_length(file) <= 255; project_id IS NOT NULL; char_length(verification_failure) <= 255
locked_by_user_id — terraform_states.locked_by_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id, name — terraform_states soft-deletes via deleted_at, but (project_id, name) is unique across live AND deleted rows. A user who deletes their account can never sign up again with the same value, and ON CONFLICT upserts will resurrect ghosts.uuid — terraform_states soft-deletes via deleted_at, but (uuid) is unique across live AND deleted rows. A user who deletes their account can never sign up again with the same value, and ON CONFLICT upserts will resurrect ghosts.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
file_store | smallint | null | |||
file | varchar(255) | null | |||
lock_xid | varchar(255) | null | |||
locked_at | timestamptz | null | |||
locked_by_user_id | bigint | null | users.id 1:N · SET NULL | ||
uuid | varchar(32) | NOT NULL | |||
name | varchar(255) | NOT NULL | |||
versioning_enabled | boolean | NOT NULL | TRUE | ||
deleted_at | timestamptz | null | |||
activerecord_lock_version | integer | NOT NULL | 0 |
Tables whose names start with wiki_.
wiki_page_meta.namespace_id → namespaces.id — one namespaces, many wiki_page_meta · optional · ON DELETE CASCADE · indexedwiki_page_meta.project_id → projects.id — one projects, many wiki_page_meta · optional · ON DELETE CASCADE · indexedwiki_page_meta_user_mentions.namespace_id → namespaces.id — one namespaces, many wiki_page_meta_user_mentions · required · ON DELETE CASCADE · indexedwiki_page_meta_user_mentions.note_id → notes.id — one notes, many wiki_page_meta_user_mentions · required · ON DELETE CASCADE · indexedwiki_page_meta_user_mentions.wiki_page_meta_id → wiki_page_meta.id — one wiki_page_meta, many wiki_page_meta_user_mentions · required · ON DELETE CASCADE · indexedwiki_page_slugs.project_id → projects.id — one projects, many wiki_page_slugs · optional · ON DELETE CASCADE · indexedwiki_page_slugs.namespace_id → namespaces.id — one namespaces, many wiki_page_slugs · optional · ON DELETE CASCADE · indexedwiki_page_slugs.wiki_page_meta_id → wiki_page_meta.id — one wiki_page_meta, many wiki_page_slugs · required · ON DELETE CASCADE · indexednamespace_id — wiki_page_meta.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — wiki_page_meta.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
title | varchar(255) | NOT NULL | |||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
deleted_at | timestamptz | null |
Table checks: num_nonnulls(namespace_id, project_id) = 1
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
wiki_page_meta_id | bigint | NOT NULL | wiki_page_meta.id 1:N · CASCADE | ||
note_id | bigint | NOT NULL | notes.id 1:N · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
mentioned_users_ids | bigint[] | null | |||
mentioned_projects_ids | bigint[] | null | |||
mentioned_groups_ids | bigint[] | null |
project_id — wiki_page_slugs.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id — wiki_page_slugs.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
canonical | boolean | NOT NULL | FALSE | ||
wiki_page_meta_id | bigint | NOT NULL | wiki_page_meta.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
slug | varchar(2048) | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE |
Table checks: num_nonnulls(namespace_id, project_id) = 1
wiki_repository_states references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
project_wiki_repository_id | bigint | NOT NULL | |||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | null | |||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 | ||
project_id | bigint | null | CHECK project_id IS NOT NULL |
Table checks: char_length(verification_failure) <= 255; project_id IS NOT NULL
Tables whose names start with work_.
work_item_agent_plans.namespace_id → namespaces.id — one namespaces, many work_item_agent_plans · required · ON DELETE CASCADE · indexedwork_item_agent_plans.work_item_id → issues.id — one issues, at most one work_item_agent_plans · required · ON DELETE CASCADE · indexedwork_item_colors.issue_id → issues.id — one issues, at most one work_item_colors · required · ON DELETE CASCADE · indexedwork_item_colors.namespace_id → namespaces.id — one namespaces, many work_item_colors · required · ON DELETE CASCADE · indexedwork_item_current_statuses.custom_status_id → work_item_custom_statuses.id — one work_item_custom_statuses, many work_item_current_statuses · optional · ON DELETE SET NULL · indexedwork_item_current_statuses.work_item_id → issues.id — one issues, at most one work_item_current_statuses · required · ON DELETE CASCADE · indexedwork_item_current_statuses.namespace_id → namespaces.id — one namespaces, many work_item_current_statuses · required · ON DELETE CASCADE · indexedwork_item_custom_lifecycle_statuses.namespace_id → namespaces.id — one namespaces, many work_item_custom_lifecycle_statuses · required · ON DELETE CASCADE · indexedwork_item_custom_lifecycle_statuses.status_id → work_item_custom_statuses.id — one work_item_custom_statuses, many work_item_custom_lifecycle_statuses · required · ON DELETE CASCADE · indexedwork_item_custom_lifecycle_statuses.lifecycle_id → work_item_custom_lifecycles.id — one work_item_custom_lifecycles, many work_item_custom_lifecycle_statuses · required · ON DELETE CASCADE · indexedwork_item_custom_lifecycles.namespace_id → namespaces.id — one namespaces, many work_item_custom_lifecycles · required · ON DELETE CASCADE · indexedwork_item_custom_lifecycles.default_open_status_id → work_item_custom_statuses.id — one work_item_custom_statuses, many work_item_custom_lifecycles · required · ON DELETE CASCADE · indexedwork_item_custom_lifecycles.updated_by_id → users.id — one users, many work_item_custom_lifecycles · optional · ON DELETE SET NULL · indexedwork_item_custom_lifecycles.created_by_id → users.id — one users, many work_item_custom_lifecycles · optional · ON DELETE SET NULL · indexedwork_item_custom_lifecycles.default_duplicate_status_id → work_item_custom_statuses.id — one work_item_custom_statuses, many work_item_custom_lifecycles · required · ON DELETE CASCADE · indexedwork_item_custom_lifecycles.default_closed_status_id → work_item_custom_statuses.id — one work_item_custom_statuses, many work_item_custom_lifecycles · required · ON DELETE CASCADE · indexedwork_item_custom_status_mappings.namespace_id → namespaces.id — one namespaces, many work_item_custom_status_mappings · required · ON DELETE CASCADE · indexedwork_item_custom_status_mappings.new_status_id → work_item_custom_statuses.id — one work_item_custom_statuses, many work_item_custom_status_mappings · required · ON DELETE CASCADE · indexedwork_item_custom_status_mappings.old_status_id → work_item_custom_statuses.id — one work_item_custom_statuses, many work_item_custom_status_mappings · required · ON DELETE RESTRICT · indexedwork_item_custom_statuses.namespace_id → namespaces.id — one namespaces, many work_item_custom_statuses · required · ON DELETE CASCADE · indexedwork_item_custom_statuses.updated_by_id → users.id — one users, many work_item_custom_statuses · optional · ON DELETE SET NULL · indexedwork_item_custom_statuses.created_by_id → users.id — one users, many work_item_custom_statuses · optional · ON DELETE SET NULL · indexedwork_item_custom_types.organization_id → organizations.id — one organizations, many work_item_custom_types · optional · ON DELETE CASCADE · indexedwork_item_custom_types.namespace_id → namespaces.id — one namespaces, many work_item_custom_types · optional · ON DELETE CASCADE · indexedwork_item_date_field_values.namespace_id → namespaces.id — one namespaces, many work_item_date_field_values · required · ON DELETE CASCADE · indexedwork_item_date_field_values.work_item_id → issues.id — one issues, many work_item_date_field_values · required · ON DELETE CASCADE · indexedwork_item_date_field_values.custom_field_id → custom_fields.id — one custom_fields, many work_item_date_field_values · required · ON DELETE CASCADE · indexedwork_item_dates_sources.start_date_sourcing_milestone_id → milestones.id — one milestones, many work_item_dates_sources · optional · ON DELETE SET NULL · indexedwork_item_dates_sources.start_date_sourcing_work_item_id → issues.id — one issues, many work_item_dates_sources · optional · ON DELETE SET NULL · indexedwork_item_dates_sources.due_date_sourcing_work_item_id → issues.id — one issues, many work_item_dates_sources · optional · ON DELETE SET NULL · indexedwork_item_dates_sources.due_date_sourcing_milestone_id → milestones.id — one milestones, many work_item_dates_sources · optional · ON DELETE SET NULL · indexedwork_item_dates_sources.issue_id → issues.id — one issues, at most one work_item_dates_sources · required · ON DELETE CASCADE · indexedwork_item_dates_sources.namespace_id → namespaces.id — one namespaces, many work_item_dates_sources · required · ON DELETE CASCADE · indexedwork_item_number_field_values.namespace_id → namespaces.id — one namespaces, many work_item_number_field_values · required · ON DELETE CASCADE · indexedwork_item_number_field_values.work_item_id → issues.id — one issues, many work_item_number_field_values · required · ON DELETE CASCADE · indexedwork_item_number_field_values.custom_field_id → custom_fields.id — one custom_fields, many work_item_number_field_values · required · ON DELETE CASCADE · indexedwork_item_parent_links.namespace_id → namespaces.id — one namespaces, many work_item_parent_links · optional · ON DELETE CASCADE · indexedwork_item_parent_links.work_item_parent_id → issues.id — one issues, many work_item_parent_links · required · ON DELETE CASCADE · indexedwork_item_parent_links.work_item_id → issues.id — one issues, at most one work_item_parent_links · required · ON DELETE CASCADE · indexedwork_item_positions.namespace_id → namespaces.id — one namespaces, many work_item_positions · required · ON DELETE CASCADE · indexedwork_item_positions.work_item_id → issues.id — one issues, at most one work_item_positions · required · ON DELETE CASCADE · indexedwork_item_progresses.namespace_id → namespaces.id — one namespaces, many work_item_progresses · optional · ON DELETE CASCADE · indexedwork_item_progresses.issue_id → issues.id — one issues, at most one work_item_progresses · required · ON DELETE CASCADE · indexedwork_item_select_field_values.namespace_id → namespaces.id — one namespaces, many work_item_select_field_values · required · ON DELETE CASCADE · indexedwork_item_select_field_values.work_item_id → issues.id — one issues, many work_item_select_field_values · required · ON DELETE CASCADE · indexedwork_item_select_field_values.custom_field_select_option_id → custom_field_select_options.id — one custom_field_select_options, many work_item_select_field_values · required · ON DELETE CASCADE · indexedwork_item_select_field_values.custom_field_id → custom_fields.id — one custom_fields, many work_item_select_field_values · required · ON DELETE CASCADE · indexedwork_item_settings.namespace_id → namespaces.id — one namespaces, many work_item_settings · optional · ON DELETE CASCADE · indexedwork_item_settings.organization_id → organizations.id — one organizations, many work_item_settings · optional · ON DELETE CASCADE · indexedwork_item_text_field_values.namespace_id → namespaces.id — one namespaces, many work_item_text_field_values · required · ON DELETE CASCADE · indexedwork_item_text_field_values.work_item_id → issues.id — one issues, many work_item_text_field_values · required · ON DELETE CASCADE · indexedwork_item_text_field_values.custom_field_id → custom_fields.id — one custom_fields, many work_item_text_field_values · required · ON DELETE CASCADE · indexedwork_item_transitions.promoted_to_epic_id → epics.id — one epics, many work_item_transitions · optional · ON DELETE SET NULL · indexedwork_item_transitions.work_item_id → issues.id — one issues, at most one work_item_transitions · required · ON DELETE CASCADE · indexedwork_item_transitions.duplicated_to_id → issues.id — one issues, many work_item_transitions · optional · ON DELETE SET NULL · indexedwork_item_transitions.moved_to_id → issues.id — one issues, many work_item_transitions · optional · ON DELETE SET NULL · indexedwork_item_transitions.namespace_id → namespaces.id — one namespaces, many work_item_transitions · required · ON DELETE CASCADE · indexedwork_item_type_custom_fields.custom_field_id → custom_fields.id — one custom_fields, many work_item_type_custom_fields · required · ON DELETE CASCADE · indexedwork_item_type_custom_fields.namespace_id → namespaces.id — one namespaces, many work_item_type_custom_fields · required · ON DELETE CASCADE · indexedwork_item_type_custom_lifecycles.lifecycle_id → work_item_custom_lifecycles.id — one work_item_custom_lifecycles, many work_item_type_custom_lifecycles · required · ON DELETE CASCADE · indexedwork_item_type_custom_lifecycles.namespace_id → namespaces.id — one namespaces, many work_item_type_custom_lifecycles · required · ON DELETE CASCADE · indexedwork_item_type_user_preferences.user_id → users.id — one users, many work_item_type_user_preferences · required · ON DELETE CASCADE · indexedwork_item_type_user_preferences.namespace_id → namespaces.id — one namespaces, many work_item_type_user_preferences · required · ON DELETE CASCADE · indexedwork_item_type_visibilities.namespace_id → namespaces.id — one namespaces, many work_item_type_visibilities · required · ON DELETE CASCADE · indexedwork_item_type_visibility_defaults.namespace_id → namespaces.id — one namespaces, many work_item_type_visibility_defaults · optional · ON DELETE CASCADE · indexedwork_item_type_visibility_defaults.organization_id → organizations.id — one organizations, many work_item_type_visibility_defaults · optional · ON DELETE CASCADE · indexedwork_item_weights_sources.namespace_id → namespaces.id — one namespaces, many work_item_weights_sources · required · ON DELETE CASCADE · indexedwork_item_weights_sources.work_item_id → issues.id — one issues, at most one work_item_weights_sources · required · ON DELETE CASCADE · indexed| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
work_item_id PK | bigint | NOT NULL | issues.id 1:1 · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
cached_markdown_version | integer | null | |||
file_store | smallint | NOT NULL | 1 | ||
readiness_score | smallint | null | |||
ai_planning_enabled | boolean | NOT NULL | FALSE |
issue_id, namespace_id — work_item_colors looks like a many-to-many link table but has no unique constraint across (issue_id, namespace_id). The same pair can be inserted twice; every join through it will double-count.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
issue_id PK | bigint | NOT NULL | issues.id 1:1 · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
color | text | NOT NULL | CHECK char_length(color) <= 7 |
Table checks: char_length(color) <= 7
custom_status_id — work_item_current_statuses.custom_status_id may be NULL, so the relationship to work_item_custom_statuses is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
work_item_id | bigint | NOT NULL | issues.id 1:1 · CASCADE | ||
system_defined_status_id | bigint | null | |||
custom_status_id | bigint | null | work_item_custom_statuses.id 1:N · SET NULL | ||
updated_at | timestamptz | NOT NULL |
Table checks: num_nonnulls(custom_status_id, system_defined_status_id) > 0
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
lifecycle_id | bigint | NOT NULL | work_item_custom_lifecycles.id 1:N · CASCADE | ||
status_id | bigint | NOT NULL | work_item_custom_statuses.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
position | integer | NOT NULL | 0 | CHECK position >= 0 |
Table checks: position >= 0
updated_by_id — work_item_custom_lifecycles.updated_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.created_by_id — work_item_custom_lifecycles.created_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.updated_by_id, created_by_id — work_item_custom_lifecycles has several nullable FKs (updated_by_id, created_by_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
default_open_status_id | bigint | NOT NULL | work_item_custom_statuses.id 1:N · CASCADE | ||
default_closed_status_id | bigint | NOT NULL | work_item_custom_statuses.id 1:N · CASCADE | ||
default_duplicate_status_id | bigint | NOT NULL | work_item_custom_statuses.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 64 | ||
created_by_id | bigint | null | users.id 1:N · SET NULL | ||
updated_by_id | bigint | null | users.id 1:N · SET NULL |
Table checks: char_length(name) <= 64
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
old_status_id | bigint | NOT NULL | work_item_custom_statuses.id 1:N · RESTRICT | ||
new_status_id | bigint | NOT NULL | work_item_custom_statuses.id 1:N · CASCADE | ||
work_item_type_id | bigint | NOT NULL | |||
valid_from | timestamptz | null | |||
valid_until | timestamptz | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
old_status_role | smallint | null |
Table checks: valid_from IS NULL OR valid_until IS NULL OR valid_from < valid_until; old_status_id <> new_status_id
updated_by_id — work_item_custom_statuses.updated_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.created_by_id — work_item_custom_statuses.created_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.updated_by_id, created_by_id — work_item_custom_statuses has several nullable FKs (updated_by_id, created_by_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
category | smallint | NOT NULL | 1 | CHECK category > 0 | |
name | text | NOT NULL | CHECK char_length(name) <= 32 | ||
description | text | null | CHECK char_length(description) <= 128 | ||
color | text | NOT NULL | CHECK char_length(color) <= 7 | ||
created_by_id | bigint | null | users.id 1:N · SET NULL | ||
updated_by_id | bigint | null | users.id 1:N · SET NULL | ||
converted_from_system_defined_status_identifier | smallint | null |
Table checks: char_length(color) <= 7; char_length(name) <= 32; char_length(description) <= 128; category > 0; name !~ CAST('^["''`]|["''`]$|[\x00-\x1F\x7F]' AS text)
organization_id — work_item_custom_types.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id — work_item_custom_types.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | CHECK id >= 1001 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
organization_id | bigint | null | organizations.id 1:N · CASCADE | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
icon_name | smallint | NOT NULL | 0 | ||
converted_from_system_defined_type_identifier | smallint | null | |||
name | text | NOT NULL | CHECK char_length(name) <= 48 | ||
archived | boolean | NOT NULL | FALSE |
Table checks: id >= 1001; char_length(name) <= 48; num_nonnulls(namespace_id, organization_id) = 1
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
work_item_id | bigint | NOT NULL | issues.id 1:N · CASCADE | ||
custom_field_id | bigint | NOT NULL | custom_fields.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
value | date | NOT NULL |
start_date_sourcing_milestone_id — work_item_dates_sources.start_date_sourcing_milestone_id may be NULL, so the relationship to milestones is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.start_date_sourcing_work_item_id — work_item_dates_sources.start_date_sourcing_work_item_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.due_date_sourcing_work_item_id — work_item_dates_sources.due_date_sourcing_work_item_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.due_date_sourcing_milestone_id — work_item_dates_sources.due_date_sourcing_milestone_id may be NULL, so the relationship to milestones is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.start_date_sourcing_milestone_id, start_date_sourcing_work_item_id, due_date_sourcing_work_item_id, due_date_sourcing_milestone_id — work_item_dates_sources has several nullable FKs (start_date_sourcing_milestone_id, start_date_sourcing_work_item_id, due_date_sourcing_work_item_id, due_date_sourcing_milestone_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
issue_id PK | bigint | NOT NULL | issues.id 1:1 · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
start_date_is_fixed | boolean | NOT NULL | FALSE | ||
due_date_is_fixed | boolean | NOT NULL | FALSE | ||
start_date | date | null | |||
due_date | date | null | |||
start_date_sourcing_work_item_id | bigint | null | issues.id 1:N · SET NULL | ||
start_date_sourcing_milestone_id | bigint | null | milestones.id 1:N · SET NULL | ||
due_date_sourcing_work_item_id | bigint | null | issues.id 1:N · SET NULL | ||
due_date_sourcing_milestone_id | bigint | null | milestones.id 1:N · SET NULL | ||
start_date_fixed | date | null | |||
due_date_fixed | date | null |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
work_item_id | bigint | NOT NULL | issues.id 1:N · CASCADE | ||
custom_field_id | bigint | NOT NULL | custom_fields.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
value | numeric | NOT NULL |
namespace_id — work_item_parent_links.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
work_item_id | bigint | NOT NULL | issues.id 1:1 · CASCADE | ||
work_item_parent_id | bigint | NOT NULL | issues.id 1:N · CASCADE | ||
relative_position | integer | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
work_item_id PK | bigint | NOT NULL | issues.id 1:1 · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
relative_position | bigint | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
relative_positioning_namespace_id | bigint | null |
namespace_id — work_item_progresses.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
issue_id PK | bigint | NOT NULL | issues.id 1:1 · CASCADE | ||
progress | smallint | NOT NULL | 0 | ||
start_value | double precision | NOT NULL | 0.0 | ||
end_value | double precision | NOT NULL | 100.0 | ||
current_value | double precision | NOT NULL | 0.0 | ||
rollup_progress | boolean | NOT NULL | TRUE | ||
reminder_frequency | smallint | NOT NULL | 0 | ||
last_reminder_sent_at | timestamptz | null | |||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
work_item_id | bigint | NOT NULL | issues.id 1:N · CASCADE | ||
custom_field_id | bigint | NOT NULL | custom_fields.id 1:N · CASCADE | ||
custom_field_select_option_id | bigint | NOT NULL | custom_field_select_options.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
namespace_id — work_item_settings.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id — work_item_settings.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
organization_id | bigint | null | organizations.id 1:N · CASCADE | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
customizable_type_visibility | boolean | NOT NULL | FALSE |
Table checks: num_nonnulls(namespace_id, organization_id) = 1
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
work_item_id | bigint | NOT NULL | issues.id 1:N · CASCADE | ||
custom_field_id | bigint | NOT NULL | custom_fields.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
value | text | NOT NULL | CHECK char_length(value) <= 1024 |
Table checks: char_length(value) <= 1024
promoted_to_epic_id — work_item_transitions.promoted_to_epic_id may be NULL, so the relationship to epics is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.duplicated_to_id — work_item_transitions.duplicated_to_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.moved_to_id — work_item_transitions.moved_to_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.promoted_to_epic_id, duplicated_to_id, moved_to_id — work_item_transitions has several nullable FKs (promoted_to_epic_id, duplicated_to_id, moved_to_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
work_item_id PK | bigint | NOT NULL | issues.id 1:1 · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
moved_to_id | bigint | null | issues.id 1:N · SET NULL | ||
duplicated_to_id | bigint | null | issues.id 1:N · SET NULL | ||
promoted_to_epic_id | bigint | null | epics.id 1:N · SET NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
custom_field_id | bigint | NOT NULL | custom_fields.id 1:N · CASCADE | ||
work_item_type_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
lifecycle_id, namespace_id — work_item_type_custom_lifecycles looks like a many-to-many link table but has no unique constraint across (lifecycle_id, namespace_id). The same pair can be inserted twice; every join through it will double-count.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
work_item_type_id | bigint | NOT NULL | |||
lifecycle_id | bigint | NOT NULL | work_item_custom_lifecycles.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
work_item_type_id | bigint | null | |||
sort | text | null | CHECK char_length(sort) <= 255 | ||
display_settings | jsonb | NOT NULL | CAST('{}' AS jsonb) | CHECK jsonb_typeof(display_settings) = CAST('object' AS text) |
Table checks: char_length(sort) <= 255; jsonb_typeof(display_settings) = CAST('object' AS text)
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
work_item_type_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
enabled | boolean | NOT NULL | TRUE | ||
propagate | boolean | NOT NULL | FALSE |
namespace_id — work_item_type_visibility_defaults.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id — work_item_type_visibility_defaults.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
work_item_type_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
enabled | boolean | NOT NULL | TRUE | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE |
Table checks: num_nonnulls(namespace_id, organization_id) = 1
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
work_item_id PK | bigint | NOT NULL | issues.id 1:1 · CASCADE | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
rolled_up_weight | bigint | null | |||
rolled_up_completed_weight | bigint | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
Tables whose names start with workspace_.
workspace_agentk_states.workspace_id → workspaces.id — one workspaces, at most one workspace_agentk_states · required · ON DELETE CASCADE · indexedworkspace_agentk_states.project_id → projects.id — one projects, many workspace_agentk_states · required · ON DELETE CASCADE · indexedworkspace_tokens.project_id → projects.id — one projects, many workspace_tokens · required · ON DELETE CASCADE · indexedworkspace_tokens.workspace_id → workspaces.id — one workspaces, at most one workspace_tokens · required · ON DELETE CASCADE · indexedworkspace_variables.project_id → projects.id — one projects, many workspace_variables · optional · ON DELETE CASCADE · indexedworkspace_variables.workspace_id → workspaces.id — one workspaces, many workspace_variables · required · ON DELETE CASCADE · indexedproject_id, workspace_id — workspace_agentk_states looks like a many-to-many link table but has no unique constraint across (project_id, workspace_id). The same pair can be inserted twice; every join through it will double-count.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
workspace_id | bigint | NOT NULL | workspaces.id 1:1 · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
desired_config | jsonb | NOT NULL |
project_id, workspace_id — workspace_tokens looks like a many-to-many link table but has no unique constraint across (project_id, workspace_id). The same pair can be inserted twice; every join through it will double-count.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
workspace_id | bigint | NOT NULL | workspaces.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
token_encrypted | text | NOT NULL | CHECK char_length(token_encrypted) <= 512 |
Table checks: char_length(token_encrypted) <= 512; char_length(token_encrypted) <= 512
project_id — workspace_variables.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
workspace_id | bigint | NOT NULL | workspaces.id 1:N · CASCADE | ||
variable_type | smallint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
key | text | NOT NULL | CHECK char_length(key) <= 255 | ||
encrypted_value | bytea | NOT NULL | |||
encrypted_value_iv | bytea | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL | |
user_provided | boolean | NOT NULL | FALSE |
Table checks: char_length(key) <= 255; project_id IS NOT NULL
Tables whose names start with workspaces_.
workspaces.user_id → users.id — one users, many workspaces · required · ON DELETE CASCADE · indexedworkspaces.project_id → projects.id — one projects, many workspaces · required · ON DELETE CASCADE · indexedworkspaces.personal_access_token_id → personal_access_tokens.id — one personal_access_tokens, many workspaces · optional · ON DELETE RESTRICT · indexedworkspaces.cluster_agent_id → cluster_agents.id — one cluster_agents, many workspaces · required · ON DELETE CASCADE · indexedworkspaces_agent_config_versions.project_id → projects.id — one projects, many workspaces_agent_config_versions · required · ON DELETE CASCADE · indexedworkspaces_agent_configs.cluster_agent_id → cluster_agents.id — one cluster_agents, at most one workspaces_agent_configs · required · ON DELETE CASCADE · indexedworkspaces_agent_configs.project_id → projects.id — one projects, many workspaces_agent_configs · required · ON DELETE CASCADE · indexedpersonal_access_token_id — workspaces.personal_access_token_id may be NULL, so the relationship to personal_access_tokens is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
cluster_agent_id | bigint | NOT NULL | cluster_agents.id 1:N · CASCADE | ||
desired_state_updated_at | timestamptz | NOT NULL | |||
responded_to_agent_at | timestamptz | null | |||
name | text | NOT NULL | CHECK char_length(name) <= 64 | ||
namespace | text | NOT NULL | CHECK char_length(namespace) <= 64 | ||
desired_state | text | NOT NULL | CHECK char_length(desired_state) <= 32 | ||
actual_state | text | NOT NULL | CHECK char_length(actual_state) <= 32 | ||
devfile_path | text | null | CHECK char_length(devfile_path) <= 2048 | ||
devfile | text | null | CHECK char_length(devfile) <= 65535 | ||
processed_devfile | text | null | CHECK char_length(processed_devfile) <= 65535 | ||
url | text | null | CHECK char_length(url) <= 1024 | ||
deployment_resource_version | text | null | CHECK char_length(deployment_resource_version) <= 64 | ||
personal_access_token_id | bigint | null | personal_access_tokens.id 1:N · RESTRICT | CHECK personal_access_token_id IS NOT NULL | |
force_include_all_resources | boolean | NOT NULL | TRUE | ||
url_prefix | text | null | CHECK char_length(url_prefix) <= 256 | ||
url_query_string | text | null | CHECK char_length(url_query_string) <= 256 | ||
workspaces_agent_config_version | integer | NOT NULL | |||
project_ref | text | null | CHECK char_length(project_ref) <= 256 | ||
actual_state_updated_at | timestamptz | NOT NULL |
Table checks: char_length(name) <= 64; char_length(namespace) <= 64; char_length(deployment_resource_version) <= 64; char_length(project_ref) <= 256; char_length(processed_devfile) <= 65535; char_length(url_query_string) <= 256; char_length(actual_state) <= 32; char_length(url) <= 1024; project_ref IS NOT NULL; char_length(desired_state) <= 32; char_length(devfile_path) <= 2048; char_length(devfile) <= 65535; char_length(url_prefix) <= 256; personal_access_token_id IS NOT NULL
item_type, item_id — item_id points at different tables depending on item_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | null | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
item_id | bigint | NOT NULL | |||
event | text | NOT NULL | CHECK char_length(event) <= 20 | ||
whodunnit | text | null | CHECK char_length(whodunnit) <= 255 | ||
item_type | text | NOT NULL | CHECK char_length(item_type) <= 255 | ||
object | jsonb | null | |||
object_changes | jsonb | null |
Table checks: char_length(whodunnit) <= 255; char_length(event) <= 20; char_length(item_type) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
cluster_agent_id | bigint | NOT NULL | cluster_agents.id 1:1 · CASCADE | ||
workspaces_quota | bigint | NOT NULL | CAST('-1' AS integer) | ||
workspaces_per_user_quota | bigint | NOT NULL | CAST('-1' AS integer) | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
enabled | boolean | NOT NULL | |||
network_policy_enabled | boolean | NOT NULL | TRUE | ||
dns_zone | text | NOT NULL | CAST('' AS text) | CHECK char_length(dns_zone) <= 256 | |
gitlab_workspaces_proxy_namespace | text | NOT NULL | CAST('gitlab-workspaces' AS text) | CHECK char_length(gitlab_workspaces_proxy_namespace) <= 63 | |
network_policy_egress | jsonb | NOT NULL | CAST('[{"allow": "0.0.0.0/0", "except": ["10.0.0.0/8", "172.16.0.0/12", "192.168.0.0/16"]}]' AS jsonb) | ||
default_resources_per_workspace_container | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
max_resources_per_workspace | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
allow_privilege_escalation | boolean | NOT NULL | FALSE | ||
use_kubernetes_user_namespaces | boolean | NOT NULL | FALSE | ||
default_runtime_class | text | NOT NULL | CAST('' AS text) | CHECK char_length(default_runtime_class) <= 253 | |
annotations | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
labels | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
image_pull_secrets | jsonb | NOT NULL | CAST('[]' AS jsonb) | ||
max_active_hours_before_stop | smallint | NOT NULL | 36 | CHECK max_active_hours_before_stop > 0 | |
max_stopped_hours_before_termination | smallint | NOT NULL | 744 | CHECK max_stopped_hours_before_termination > 0 | |
shared_namespace | text | NOT NULL | CAST('' AS text) | CHECK char_length(shared_namespace) <= 63 | |
gitlab_workspaces_proxy_http_enabled | boolean | NOT NULL | TRUE | ||
gitlab_workspaces_proxy_ssh_enabled | boolean | NOT NULL | TRUE |
Table checks: char_length(shared_namespace) <= 63; max_stopped_hours_before_termination > 0; char_length(dns_zone) <= 256; (max_active_hours_before_stop + max_stopped_hours_before_termination) <= 8760; char_length(default_runtime_class) <= 253; max_active_hours_before_stop > 0; char_length(gitlab_workspaces_proxy_namespace) <= 63
Tables whose names start with x509_.
x509_certificates.project_id → projects.id — one projects, many x509_certificates · optional · ON DELETE CASCADE · indexedx509_certificates.x509_issuer_id → x509_issuers.id — one x509_issuers, many x509_certificates · required · ON DELETE CASCADE · indexedx509_commit_signatures.x509_certificate_id → x509_certificates.id — one x509_certificates, many x509_commit_signatures · required · ON DELETE CASCADE · indexedx509_commit_signatures.project_id → projects.id — one projects, many x509_commit_signatures · required · ON DELETE CASCADE · indexedx509_issuers.project_id → projects.id — one projects, many x509_issuers · optional · ON DELETE CASCADE · indexedproject_id — x509_certificates.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
subject_key_identifier | varchar(255) | NOT NULL | |||
subject | varchar(512) | null | |||
email | varchar(255) | NOT NULL | |||
serial_number | bytea | NOT NULL | CHECK octet_length(serial_number) <= 25 | ||
certificate_status | smallint | NOT NULL | 0 | ||
x509_issuer_id | bigint | NOT NULL | x509_issuers.id 1:N · CASCADE | ||
emails | varchar[] | NOT NULL | CAST('{}' AS varchar[]) | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL; octet_length(serial_number) <= 25
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
x509_certificate_id | bigint | NOT NULL | x509_certificates.id 1:N · CASCADE | ||
commit_sha | bytea | NOT NULL | |||
verification_status | smallint | NOT NULL | 0 | ||
committer_email | text | null | CHECK char_length(committer_email) <= 255 |
Table checks: char_length(committer_email) <= 255
project_id — x509_issuers.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
subject_key_identifier | varchar(255) | NOT NULL | |||
subject | varchar(255) | null | |||
crl_url | varchar(255) | null | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
Tables whose name prefix is theirs alone.
uploads.uploaded_by_user_id → users.id — one users, many uploads · optional · ON DELETE SET NULL · indexeduploads.project_id → projects.id — one projects, many uploads · optional · ON DELETE CASCADE · indexeduploads.namespace_id → namespaces.id — one namespaces, many uploads · optional · ON DELETE CASCADE · indexeduploads.organization_id → organizations.id — one organizations, many uploads · optional · ON DELETE CASCADE · indexedachievements.namespace_id → namespaces.id — one namespaces, many achievements · required · ON DELETE CASCADE · indexedactivation_metrics.user_id → users.id — one users, many activation_metrics · required · ON DELETE CASCADE · indexedactivation_metrics.namespace_id → namespaces.id — one namespaces, many activation_metrics · optional · ON DELETE CASCADE · indexedactivity_pub_releases_subscriptions.project_id → projects.id — one projects, many activity_pub_releases_subscriptions · required · ON DELETE CASCADE · indexedadmin_roles.organization_id → organizations.id — one organizations, many admin_roles · optional · ON DELETE CASCADE · indexedallowed_email_domains.group_id → namespaces.id — one namespaces, many allowed_email_domains · required · ON DELETE CASCADE · indexedapprovals.merge_request_id → merge_requests.id — one merge_requests, many approvals · required · ON DELETE CASCADE · indexedapprovals.project_id → projects.id — one projects, many approvals · optional · ON DELETE CASCADE · indexedarkose_sessions.user_id → users.id — one users, many arkose_sessions · required · ON DELETE CASCADE · indexedartifact_registry_namespace_mappings.organization_id → organizations.id — one organizations, at most one artifact_registry_namespace_mappings · required · ON DELETE CASCADE · indexedatlassian_identities.user_id → users.id — one users, at most one atlassian_identities · required · ON DELETE CASCADE · indexedauthentication_events.organization_id → organizations.id — one organizations, many authentication_events · required · ON DELETE CASCADE · indexedauthentication_events.user_id → users.id — one users, many authentication_events · optional · ON DELETE SET NULL · indexedautomation_rules.namespace_id → namespaces.id — one namespaces, many automation_rules · required · ON DELETE CASCADE · indexedaward_emoji.organization_id → organizations.id — one organizations, many award_emoji · optional · ON DELETE CASCADE · indexedaward_emoji.namespace_id → namespaces.id — one namespaces, many award_emoji · optional · ON DELETE CASCADE · indexedaws_roles.user_id → users.id — one users, at most one aws_roles · required · ON DELETE CASCADE · indexedbadges.project_id → projects.id — one projects, many badges · optional · ON DELETE CASCADE · indexedbadges.group_id → namespaces.id — one namespaces, many badges · optional · ON DELETE CASCADE · indexedbanned_users.user_id → users.id — one users, at most one banned_users · required · ON DELETE CASCADE · indexedburned_project_routes.organization_id → organizations.id — one organizations, many burned_project_routes · required · ON DELETE CASCADE · indexedcommit_user_mentions.namespace_id → namespaces.id — one namespaces, many commit_user_mentions · optional · ON DELETE CASCADE · indexedcommit_user_mentions.note_id → notes.id — one notes, at most one commit_user_mentions · required · ON DELETE CASCADE · indexedcompromised_password_detections.user_id → users.id — one users, many compromised_password_detections · required · ON DELETE CASCADE · indexedcoverage_fuzzing_corpuses.project_id → projects.id — one projects, many coverage_fuzzing_corpuses · required · ON DELETE CASCADE · indexedcoverage_fuzzing_corpuses.user_id → users.id — one users, many coverage_fuzzing_corpuses · optional · ON DELETE CASCADE · indexedcoverage_fuzzing_corpuses.package_id → packages_packages.id — one packages_packages, at most one coverage_fuzzing_corpuses · required · ON DELETE CASCADE · indexedcsv_issue_imports.user_id → users.id — one users, many csv_issue_imports · required · ON DELETE CASCADE · indexedcsv_issue_imports.project_id → projects.id — one projects, many csv_issue_imports · required · ON DELETE CASCADE · indexeddeployments.environment_id → environments.id — one environments, many deployments · required · ON DELETE CASCADE · indexeddeployments.project_id → projects.id — one projects, many deployments · required · ON DELETE CASCADE · indexeddescription_versions.namespace_id → namespaces.id — one namespaces, many description_versions · required · ON DELETE CASCADE · indexeddescription_versions.merge_request_id → merge_requests.id — one merge_requests, many description_versions · optional · ON DELETE CASCADE · indexeddescription_versions.issue_id → issues.id — one issues, many description_versions · optional · ON DELETE CASCADE · indexeddescription_versions.epic_id → epics.id — one epics, many description_versions · optional · ON DELETE CASCADE · indexeddesignated_beneficiaries.user_id → users.id — one users, many designated_beneficiaries · required · ON DELETE CASCADE · indexeddiff_note_positions.namespace_id → namespaces.id — one namespaces, many diff_note_positions · optional · ON DELETE CASCADE · indexeddiff_note_positions.note_id → notes.id — one notes, many diff_note_positions · required · ON DELETE CASCADE · indexeddingtalk_tracker_data.integration_id → integrations.id — one integrations, many dingtalk_tracker_data · required · ON DELETE CASCADE · indexeddraft_notes.project_id → projects.id — one projects, many draft_notes · optional · ON DELETE CASCADE · indexeddraft_notes.author_id → users.id — one users, many draft_notes · required · ON DELETE CASCADE · indexeddraft_notes.merge_request_id → merge_requests.id — one merge_requests, many draft_notes · required · ON DELETE CASCADE · indexedearly_access_program_tracking_events.user_id → users.id — one users, many early_access_program_tracking_events · required · ON DELETE CASCADE · indexedemails.user_id → users.id — one users, many emails · required · ON DELETE CASCADE · indexedenvironments.merge_request_id → merge_requests.id — one merge_requests, many environments · optional · ON DELETE SET NULL · indexedenvironments.cluster_agent_id → cluster_agents.id — one cluster_agents, many environments · optional · ON DELETE SET NULL · indexedenvironments.project_id → projects.id — one projects, many environments · required · ON DELETE CASCADE · indexedepics.due_date_sourcing_epic_id → epics.id — one epics, many epics · optional · ON DELETE SET NULL · indexedepics.start_date_sourcing_milestone_id → milestones.id — one milestones, many epics · optional · ON DELETE SET NULL · indexedepics.author_id → users.id — one users, many epics · required · ON DELETE CASCADE · indexedepics.due_date_sourcing_milestone_id → milestones.id — one milestones, many epics · optional · ON DELETE SET NULL · indexedepics.work_item_parent_link_id → work_item_parent_links.id — one work_item_parent_links, at most one epics · optional · ON DELETE SET NULL · indexedepics.start_date_sourcing_epic_id → epics.id — one epics, many epics · optional · ON DELETE SET NULL · indexedepics.closed_by_id → users.id — one users, many epics · optional · ON DELETE SET NULL · indexedepics.assignee_id → users.id — one users, many epics · optional · ON DELETE SET NULL · indexedepics.issue_id → issues.id — one issues, at most one epics · optional · ON DELETE CASCADE · indexedepics.parent_id → epics.id — one epics, many epics · optional · ON DELETE SET NULL · indexedepics.group_id → namespaces.id — one namespaces, many epics · required · ON DELETE CASCADE · indexedevents.personal_namespace_id → namespaces.id — one namespaces, many events · optional · ON DELETE CASCADE · indexedevents.project_id → projects.id — one projects, many events · optional · ON DELETE CASCADE · indexedevents.group_id → namespaces.id — one namespaces, many events · optional · ON DELETE CASCADE · indexedevidences.project_id → projects.id — one projects, many evidences · optional · ON DELETE CASCADE · indexedevidences.release_id → releases.id — one releases, many evidences · required · ON DELETE CASCADE · indexedexcluded_merge_requests.merge_request_id → merge_requests.id — one merge_requests, many excluded_merge_requests · required · ON DELETE CASCADE · indexedghost_user_migrations.user_id → users.id — one users, at most one ghost_user_migrations · required · ON DELETE CASCADE · indexedgrafana_integrations.project_id → projects.id — one projects, many grafana_integrations · required · ON DELETE CASCADE · indexedgranular_scopes.namespace_id → namespaces.id — one namespaces, many granular_scopes · optional · ON DELETE CASCADE · indexedgranular_scopes.organization_id → organizations.id — one organizations, many granular_scopes · required · ON DELETE CASCADE · indexedroutes.namespace_id → namespaces.id — one namespaces, at most one routes · optional · ON DELETE CASCADE · indexediam_outbox.organization_id → organizations.id — one organizations, many iam_outbox · required · ON DELETE CASCADE · indexedidentities.user_id → users.id — one users, many identities · optional · ON DELETE CASCADE · indexedidentities.saml_provider_id → saml_providers.id — one saml_providers, many identities · optional · ON DELETE CASCADE · indexedindex_statuses.project_id → projects.id — one projects, at most one index_statuses · required · ON DELETE CASCADE · indexedinsights.namespace_id → namespaces.id — one namespaces, many insights · required · ON DELETE CASCADE · indexedinsights.project_id → projects.id — one projects, many insights · required · ON DELETE CASCADE · indexedintegrations.project_id → projects.id — one projects, many integrations · optional · ON DELETE CASCADE · indexedintegrations.organization_id → organizations.id — one organizations, many integrations · optional · ON DELETE CASCADE · indexedintegrations.group_id → namespaces.id — one namespaces, many integrations · optional · ON DELETE CASCADE · indexedintegrations.inherit_from_id → integrations.id — one integrations, many integrations · optional · ON DELETE CASCADE · indexedinternal_ids.namespace_id → namespaces.id — one namespaces, many internal_ids · optional · ON DELETE CASCADE · indexedinternal_ids.project_id → projects.id — one projects, many internal_ids · optional · ON DELETE CASCADE · indexedip_restrictions.group_id → namespaces.id — one namespaces, many ip_restrictions · required · ON DELETE CASCADE · indexedissues.author_id → users.id — one users, many issues · optional · ON DELETE SET NULL · indexedissues.sprint_id → sprints.id — one sprints, many issues · optional · ON DELETE SET NULL · indexedissues.project_id → projects.id — one projects, many issues · optional · ON DELETE CASCADE · indexedissues.milestone_id → milestones.id — one milestones, many issues · optional · ON DELETE SET NULL · indexedissues.duplicated_to_id → issues.id — one issues, many issues · optional · ON DELETE SET NULL · indexedissues.moved_to_id → issues.id — one issues, many issues · optional · ON DELETE SET NULL · indexedissues.closed_by_id → users.id — one users, many issues · optional · ON DELETE SET NULL · indexedissues.namespace_id → namespaces.id — one namespaces, many issues · optional · ON DELETE CASCADE · indexedissues.promoted_to_epic_id → epics.id — one epics, many issues · optional · ON DELETE SET NULL · indexedissues.updated_by_id → users.id — one users, many issues · optional · ON DELETE SET NULL · indexediterations_cadences.group_id → namespaces.id — one namespaces, many iterations_cadences · required · ON DELETE CASCADE · indexedjob_environments.environment_id → environments.id — one environments, many job_environments · required · ON DELETE CASCADE · indexedjob_environments.project_id → projects.id — one projects, many job_environments · required · ON DELETE CASCADE · indexedjob_environments.deployment_id → deployments.id — one deployments, many job_environments · optional · ON DELETE CASCADE · indexedkeys.organization_id → organizations.id — one organizations, many keys · optional · ON DELETE CASCADE · indexedknowledge_graph_enabled_namespaces.root_namespace_id → namespaces.id — one namespaces, at most one knowledge_graph_enabled_namespaces · required · ON DELETE CASCADE · indexedlabels.project_id → projects.id — one projects, many labels · optional · ON DELETE CASCADE · indexedlabels.organization_id → organizations.id — one organizations, many labels · optional · ON DELETE CASCADE · indexedlabels.group_id → namespaces.id — one namespaces, many labels · optional · ON DELETE CASCADE · indexedlist_user_preferences.list_id → lists.id — one lists, many list_user_preferences · required · ON DELETE CASCADE · indexedlist_user_preferences.user_id → users.id — one users, many list_user_preferences · required · ON DELETE CASCADE · indexedlists.board_id → boards.id — one boards, many lists · required · ON DELETE CASCADE · indexedlists.custom_status_id → work_item_custom_statuses.id — one work_item_custom_statuses, many lists · optional · ON DELETE CASCADE · indexedlists.iteration_id → sprints.id — one sprints, many lists · optional · ON DELETE CASCADE · indexedlists.project_id → projects.id — one projects, many lists · optional · ON DELETE CASCADE · indexedlists.label_id → labels.id — one labels, many lists · optional · ON DELETE CASCADE · indexedlists.user_id → users.id — one users, many lists · optional · ON DELETE CASCADE · indexedlists.group_id → namespaces.id — one namespaces, many lists · optional · ON DELETE CASCADE · indexedlists.milestone_id → milestones.id — one milestones, many lists · optional · ON DELETE CASCADE · indexedmilestone_releases.project_id → projects.id — one projects, many milestone_releases · optional · ON DELETE CASCADE · indexedmilestone_releases.release_id → releases.id — one releases, many milestone_releases · required · ON DELETE CASCADE · indexedmilestone_releases.milestone_id → milestones.id — one milestones, many milestone_releases · required · ON DELETE CASCADE · indexedmilestones.group_id → namespaces.id — one namespaces, many milestones · optional · ON DELETE CASCADE · indexedmilestones.project_id → projects.id — one projects, many milestones · optional · ON DELETE CASCADE · indexedmobile_device_push_subscriptions.user_id → users.id — one users, many mobile_device_push_subscriptions · required · ON DELETE CASCADE · indexednon_sql_service_pings.organization_id → organizations.id — one organizations, many non_sql_service_pings · required · ON DELETE CASCADE · indexednotes.review_id → reviews.id — one reviews, many notes · optional · ON DELETE SET NULL · indexednotes.namespace_id → namespaces.id — one namespaces, many notes · optional · ON DELETE CASCADE · indexednotes.project_id → projects.id — one projects, many notes · optional · ON DELETE CASCADE · indexednotes.organization_id → organizations.id — one organizations, many notes · optional · ON DELETE CASCADE · indexedonboarding_progresses.namespace_id → namespaces.id — one namespaces, at most one onboarding_progresses · required · ON DELETE CASCADE · indexedpath_locks.project_id → projects.id — one projects, many path_locks · optional · ON DELETE CASCADE · indexedpath_locks.user_id → users.id — one users, many path_locks · optional · ON DELETE CASCADE · indexedsnippets.project_id → projects.id — one projects, many snippets · optional · ON DELETE CASCADE · indexedpipl_users.user_id → users.id — one users, at most one pipl_users · required · ON DELETE CASCADE · indexedplan_limits.plan_id → plans.id — one plans, at most one plan_limits · required · ON DELETE CASCADE · indexedpool_repositories.organization_id → organizations.id — one organizations, many pool_repositories · optional · ON DELETE CASCADE · indexedpool_repositories.shard_id → shards.id — one shards, many pool_repositories · required · ON DELETE RESTRICT · indexedqueries_service_pings.organization_id → organizations.id — one organizations, many queries_service_pings · required · ON DELETE CASCADE · indexedraw_usage_data.organization_id → organizations.id — one organizations, many raw_usage_data · required · ON DELETE CASCADE · indexedredirect_routes.namespace_id → namespaces.id — one namespaces, many redirect_routes · optional · ON DELETE CASCADE · indexedrelated_epic_links.group_id → namespaces.id — one namespaces, many related_epic_links · optional · ON DELETE CASCADE · indexedrelated_epic_links.issue_link_id → issue_links.id — one issue_links, at most one related_epic_links · optional · ON DELETE CASCADE · indexedrelated_epic_links.target_id → epics.id — one epics, many related_epic_links · required · ON DELETE CASCADE · indexedrelated_epic_links.source_id → epics.id — one epics, many related_epic_links · required · ON DELETE CASCADE · indexedrelation_import_trackers.project_id → projects.id — one projects, many relation_import_trackers · required · ON DELETE CASCADE · indexedrelease_links.project_id → projects.id — one projects, many release_links · optional · ON DELETE CASCADE · indexedrelease_links.release_id → releases.id — one releases, many release_links · required · ON DELETE CASCADE · indexedreleases.project_id → projects.id — one projects, many releases · optional · ON DELETE CASCADE · indexedreleases.author_id → users.id — one users, many releases · optional · ON DELETE SET NULL · indexedremote_mirrors.project_id → projects.id — one projects, many remote_mirrors · optional · ON DELETE CASCADE · indexedrepository_languages.project_id → projects.id — one projects, many repository_languages · required · ON DELETE CASCADE · indexedrequired_code_owners_sections.protected_branch_project_id → projects.id — one projects, many required_code_owners_sections · optional · ON DELETE CASCADE · indexedrequired_code_owners_sections.protected_branch_namespace_id → namespaces.id — one namespaces, many required_code_owners_sections · optional · ON DELETE CASCADE · indexedrequired_code_owners_sections.protected_branch_id → protected_branches.id — one protected_branches, many required_code_owners_sections · required · ON DELETE CASCADE · indexedreviews.author_id → users.id — one users, many reviews · optional · ON DELETE SET NULL · indexedreviews.merge_request_id → merge_requests.id — one merge_requests, many reviews · required · ON DELETE CASCADE · indexedreviews.project_id → projects.id — one projects, many reviews · required · ON DELETE CASCADE · indexedsentry_issues.namespace_id → namespaces.id — one namespaces, many sentry_issues · optional · ON DELETE CASCADE · indexedsentry_issues.issue_id → issues.id — one issues, at most one sentry_issues · required · ON DELETE CASCADE · indexedslsa_attestations.project_id → projects.id — one projects, many slsa_attestations · required · ON DELETE CASCADE · indexedsmartcard_identities.user_id → users.id — one users, many smartcard_identities · required · ON DELETE CASCADE · indexedsoftware_license_policies.custom_software_license_id → custom_software_licenses.id — one custom_software_licenses, many software_license_policies · optional · ON DELETE CASCADE · indexedsoftware_license_policies.approval_policy_rule_id → approval_policy_rules.id — one approval_policy_rules, many software_license_policies · optional · ON DELETE CASCADE · indexedsoftware_license_policies.project_id → projects.id — one projects, many software_license_policies · required · ON DELETE CASCADE · indexedsoftware_license_policies.scan_result_policy_id → scan_result_policies.id — one scan_result_policies, many software_license_policies · optional · ON DELETE CASCADE · indexedspam_logs.user_id → users.id — one users, many spam_logs · optional · ON DELETE CASCADE · indexedspam_logs.organization_id → organizations.id — one organizations, many spam_logs · optional · ON DELETE CASCADE · indexedsprints.iterations_cadence_id → iterations_cadences.id — one iterations_cadences, many sprints · optional · ON DELETE CASCADE · indexedsprints.group_id → namespaces.id — one namespaces, many sprints · optional · ON DELETE CASCADE · indexedssh_signatures.user_id → users.id — one users, many ssh_signatures · optional · ON DELETE SET NULL · indexedssh_signatures.project_id → projects.id — one projects, many ssh_signatures · required · ON DELETE CASCADE · indexedssh_signatures.key_id → keys.id — one keys, many ssh_signatures · optional · ON DELETE SET NULL · indexedsubscriptions.user_id → users.id — one users, many subscriptions · optional · ON DELETE CASCADE · indexedsubscriptions.project_id → projects.id — one projects, many subscriptions · optional · ON DELETE CASCADE · indexedsuggestions.namespace_id → namespaces.id — one namespaces, many suggestions · optional · ON DELETE CASCADE · indexedsuggestions.note_id → notes.id — one notes, many suggestions · required · ON DELETE CASCADE · indexedsupply_chain_attestation_states.project_id → projects.id — one projects, many supply_chain_attestation_states · optional · ON DELETE CASCADE · indexedtarget_branch_rules.project_id → projects.id — one projects, many target_branch_rules · required · ON DELETE CASCADE · indexedterm_agreements.user_id → users.id — one users, many term_agreements · required · ON DELETE CASCADE · indexedterm_agreements.term_id → application_setting_terms.id — one application_setting_terms, many term_agreements · required · ON DELETE NO ACTION · indexedtimelog_categories.namespace_id → namespaces.id — one namespaces, many timelog_categories · required · ON DELETE CASCADE · indexedtimelogs.project_id → projects.id — one projects, many timelogs · optional · ON DELETE CASCADE · indexedtimelogs.namespace_id → namespaces.id — one namespaces, many timelogs · required · ON DELETE CASCADE · indexedtimelogs.timelog_category_id → timelog_categories.id — one timelog_categories, many timelogs · optional · ON DELETE SET NULL · indexedtimelogs.issue_id → issues.id — one issues, many timelogs · optional · ON DELETE CASCADE · indexedtimelogs.merge_request_id → merge_requests.id — one merge_requests, many timelogs · optional · ON DELETE CASCADE · indexedtimelogs.note_id → notes.id — one notes, many timelogs · optional · ON DELETE SET NULL · indexedtodos.project_id → projects.id — one projects, many todos · optional · ON DELETE CASCADE · indexedtodos.organization_id → organizations.id — one organizations, many todos · optional · ON DELETE CASCADE · indexedtodos.note_id → notes.id — one notes, many todos · optional · ON DELETE CASCADE · indexedtodos.author_id → users.id — one users, many todos · required · ON DELETE CASCADE · indexedtodos.user_id → users.id — one users, many todos · required · ON DELETE CASCADE · indexedtodos.group_id → namespaces.id — one namespaces, many todos · optional · ON DELETE CASCADE · indexedtopics.organization_id → organizations.id — one organizations, many topics · required · ON DELETE CASCADE · indexedupcoming_reconciliations.organization_id → organizations.id — one organizations, many upcoming_reconciliations · required · ON DELETE CASCADE · indexedupcoming_reconciliations.namespace_id → namespaces.id — one namespaces, at most one upcoming_reconciliations · optional · ON DELETE CASCADE · indexedvs_code_settings.user_id → users.id — one users, many vs_code_settings · required · ON DELETE CASCADE · indexedvulnerabilities.finding_id → vulnerability_occurrences.id — one vulnerability_occurrences, many vulnerabilities · optional · ON DELETE CASCADE · indexedwebauthn_registrations.user_id → users.id — one users, many webauthn_registrations · required · ON DELETE CASCADE · indexedxray_reports.project_id → projects.id — one projects, many xray_reports · required · ON DELETE CASCADE · indexedzentao_tracker_data.project_id → projects.id — one projects, many zentao_tracker_data · optional · ON DELETE CASCADE · indexedzentao_tracker_data.group_id → namespaces.id — one namespaces, many zentao_tracker_data · optional · ON DELETE CASCADE · indexedzentao_tracker_data.organization_id → organizations.id — one organizations, many zentao_tracker_data · optional · ON DELETE CASCADE · indexedzentao_tracker_data.integration_id → integrations.id — one integrations, many zentao_tracker_data · required · ON DELETE CASCADE · indexedzoom_meetings.project_id → projects.id — one projects, many zoom_meetings · required · ON DELETE CASCADE · indexedzoom_meetings.issue_id → issues.id — one issues, many zoom_meetings · required · ON DELETE CASCADE · indexedgroups_visits references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
entity_id | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | |||
visited_at PK | timestamptz | NOT NULL |
JiHu-specific table
verification_codes references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at PK | timestamptz | NOT NULL | now() | ||
visitor_id_code PK | text | NOT NULL | CHECK char_length(visitor_id_code) <= 64 | ||
code PK | text | NOT NULL | CHECK char_length(code) <= 8 | ||
phone PK | text | NOT NULL | CHECK char_length(phone) <= 50 |
Table checks: char_length(code) <= 8; char_length(visitor_id_code) <= 64; char_length(phone) <= 50
uploaded_by_user_id — uploads.uploaded_by_user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — uploads.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id — uploads.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id — uploads.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.model_type — model_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.model_type, model_id — model_id points at different tables depending on model_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.uploaded_by_user_id, project_id, namespace_id, organization_id — uploads has several nullable FKs (uploaded_by_user_id, project_id, namespace_id, organization_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
size | bigint | NOT NULL | |||
model_id | bigint | NOT NULL | |||
uploaded_by_user_id | bigint | null | users.id 1:N · SET NULL | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
created_at | timestamp | null | |||
store | integer | NOT NULL | 1 | ||
version | integer | null | 1 | ||
path | text | NOT NULL | CHECK char_length(path) <= 511 | ||
checksum | text | null | CHECK char_length(checksum) <= 64 | ||
model_type PK | text | NOT NULL | |||
uploader | text | NOT NULL | |||
mount_point | text | null | |||
secret | text | null |
Table checks: char_length(path) <= 511; char_length(checksum) <= 64
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
avatar | text | null | CHECK char_length(avatar) <= 255 | ||
description | text | null | CHECK char_length(description) <= 1024 |
Table checks: char_length(name) <= 255; char_length(description) <= 1024; char_length(avatar) <= 255
namespace_id — activation_metrics.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
metric | smallint | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
status | smallint | NOT NULL | 1 | ||
shared_inbox_url | text | null | CHECK char_length(shared_inbox_url) <= 1024 | ||
subscriber_inbox_url | text | null | CHECK char_length(subscriber_inbox_url) <= 1024 | ||
subscriber_url | text | NOT NULL | CHECK char_length(subscriber_url) <= 1024 | ||
payload | jsonb | null |
Table checks: char_length(subscriber_inbox_url) <= 1024; char_length(subscriber_url) <= 1024; char_length(shared_inbox_url) <= 1024
organization_id — admin_roles.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
description | text | null | CHECK char_length(description) <= 255 | ||
permissions | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
organization_id | bigint | null | organizations.id 1:N · CASCADE | CHECK organization_id IS NOT NULL |
Table checks: organization_id IS NOT NULL; char_length(name) <= 255; char_length(description) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
domain | varchar(255) | NOT NULL |
appearances references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
title | varchar | NOT NULL | |||
description | text | NOT NULL | |||
logo | varchar | null | |||
updated_by | integer | null | |||
header_logo | varchar | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
description_html | text | null | |||
cached_markdown_version | integer | null | |||
new_project_guidelines | text | null | |||
new_project_guidelines_html | text | null | |||
header_message | text | null | |||
header_message_html | text | null | |||
footer_message | text | null | |||
footer_message_html | text | null | |||
message_background_color | text | null | |||
message_font_color | text | null | |||
favicon | varchar | null | |||
email_header_and_footer_enabled | boolean | NOT NULL | FALSE | ||
profile_image_guidelines | text | null | CHECK char_length(profile_image_guidelines) <= 4096 | ||
profile_image_guidelines_html | text | null | |||
pwa_short_name | text | null | CHECK char_length(pwa_short_name) <= 255 | ||
pwa_icon | text | null | CHECK char_length(pwa_icon) <= 1024 | ||
pwa_name | text | null | CHECK char_length(pwa_name) <= 255 | ||
pwa_description | text | null | CHECK char_length(pwa_description) <= 2048 | ||
member_guidelines | text | null | CHECK char_length(member_guidelines) <= 4096 | ||
member_guidelines_html | text | null | |||
site_name | text | null | CHECK char_length(site_name) <= 255 |
Table checks: char_length(profile_image_guidelines) <= 4096; char_length(pwa_name) <= 255; char_length(member_guidelines) <= 4096; char_length(pwa_short_name) <= 255; char_length(pwa_description) <= 2048; char_length(pwa_icon) <= 1024; char_length(site_name) <= 255
project_id — approvals.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
merge_request_id | bigint | NOT NULL | merge_requests.id 1:N · CASCADE | ||
user_id | bigint | NOT NULL | |||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
patch_id_sha | bytea | null | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.ar_internal_metadata references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
key PK | varchar | NOT NULL | |||
value | varchar | null | |||
created_at | timestamp(6) | NOT NULL | |||
updated_at | timestamp(6) | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
session_created_at | timestamptz | null | |||
checked_answer_at | timestamptz | null | |||
verified_at | timestamptz | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
global_score | integer | null | |||
custom_score | integer | null | |||
challenge_shown | boolean | NOT NULL | FALSE | ||
challenge_solved | boolean | NOT NULL | FALSE | ||
session_is_legit | boolean | NOT NULL | TRUE | ||
is_tor | boolean | NOT NULL | FALSE | ||
is_vpn | boolean | NOT NULL | FALSE | ||
is_proxy | boolean | NOT NULL | FALSE | ||
is_bot | boolean | NOT NULL | FALSE | ||
session_xid | text | NOT NULL | CHECK char_length(session_xid) <= 64 | ||
telltale_user | text | null | CHECK char_length(telltale_user) <= 128 | ||
user_agent | text | null | CHECK char_length(user_agent) <= 255 | ||
user_language_shown | text | null | CHECK char_length(user_language_shown) <= 64 | ||
device_xid | text | null | CHECK char_length(device_xid) <= 64 | ||
telltale_list | text[] | NOT NULL | CAST('{}' AS text[]) | ||
user_ip | text | null | CHECK char_length(user_ip) <= 64 | ||
country | text | null | CHECK char_length(country) <= 64 | ||
region | text | null | CHECK char_length(region) <= 64 | ||
city | text | null | CHECK char_length(city) <= 64 | ||
isp | text | null | CHECK char_length(isp) <= 128 | ||
connection_type | text | null | CHECK char_length(connection_type) <= 64 | ||
risk_band | text | null | CHECK char_length(risk_band) <= 64 | ||
risk_category | text | null | CHECK char_length(risk_category) <= 64 | ||
ja4_hash | text | null | CHECK char_length(ja4_hash) <= 64 | ||
canvas_fingerprint | bigint | null |
Table checks: char_length(ja4_hash) <= 64; char_length(user_agent) <= 255; char_length(telltale_user) <= 128; char_length(risk_band) <= 64; char_length(session_xid) <= 64; char_length(device_xid) <= 64; char_length(user_ip) <= 64; char_length(risk_category) <= 64; char_length(city) <= 64; char_length(isp) <= 128; char_length(region) <= 64; char_length(country) <= 64; char_length(user_language_shown) <= 64; char_length(connection_type) <= 64
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:1 · CASCADE | ||
ar_namespace_id | uuid | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
user_id PK | bigint | NOT NULL | users.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
expires_at | timestamptz | null | |||
extern_uid | text | NOT NULL | CHECK char_length(extern_uid) <= 255 | ||
encrypted_token | bytea | null | CHECK octet_length(encrypted_token) <= 5120 | ||
encrypted_token_iv | bytea | null | CHECK octet_length(encrypted_token_iv) <= 12 | ||
encrypted_refresh_token | bytea | null | CHECK octet_length(encrypted_refresh_token) <= 5000 | ||
encrypted_refresh_token_iv | bytea | null | CHECK octet_length(encrypted_refresh_token_iv) <= 12 |
Table checks: octet_length(encrypted_refresh_token_iv) <= 12; octet_length(encrypted_refresh_token) <= 5000; octet_length(encrypted_token_iv) <= 12; octet_length(encrypted_token) <= 5120; char_length(extern_uid) <= 255
user_id — authentication_events.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
user_id | bigint | null | users.id 1:N · SET NULL | ||
result | smallint | NOT NULL | |||
ip_address | inet | null | |||
provider | text | NOT NULL | CHECK char_length(provider) <= 64 | ||
user_name | text | NOT NULL | CHECK char_length(user_name) <= 255 | ||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE |
Table checks: char_length(user_name) <= 255; char_length(provider) <= 64
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
issues_events | boolean | NOT NULL | FALSE | ||
merge_requests_events | boolean | NOT NULL | FALSE | ||
permanently_disabled | boolean | NOT NULL | FALSE | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
rule | text | NOT NULL | CHECK char_length(rule) <= 2048 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
Table checks: char_length(name) <= 255; char_length(rule) <= 2048
organization_id — award_emoji.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id — award_emoji.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.awardable_type — awardable_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.awardable_type, awardable_id — awardable_id points at different tables depending on awardable_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
name | varchar | null | |||
user_id | bigint | null | |||
awardable_type | varchar | null | |||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
awardable_id | bigint | null | |||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE |
Table checks: num_nonnulls(namespace_id, organization_id) = 1
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
user_id PK | bigint | NOT NULL | users.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
role_arn | varchar(2048) | null | |||
role_external_id | varchar(64) | NOT NULL | |||
region | text | null | CHECK char_length(region) <= 255 |
Table checks: char_length(region) <= 255
project_id — badges.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — badges.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.type — type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
link_url | varchar | NOT NULL | |||
image_url | varchar | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
type | varchar | NOT NULL | |||
name | varchar(255) | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
Table checks: num_nonnulls(group_id, project_id) = 1
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
user_id PK | bigint | NOT NULL | users.id 1:1 · CASCADE | ||
projects_deleted | boolean | NOT NULL | FALSE |
starts_at — starts_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.ends_at — ends_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
message | text | NOT NULL | |||
starts_at | timestamp | NOT NULL | |||
ends_at | timestamp | NOT NULL | |||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL | |||
color | varchar | null | |||
font | varchar | null | |||
message_html | text | NOT NULL | |||
cached_markdown_version | integer | null | |||
target_path | varchar(255) | null | |||
broadcast_type | smallint | NOT NULL | 1 | ||
dismissable | boolean | null | |||
target_access_levels | integer[] | NOT NULL | CAST('{}' AS integer[]) | ||
theme | smallint | NOT NULL | 0 | ||
show_in_cli | boolean | NOT NULL | TRUE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | |||
burned_at | timestamptz | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
path | text | NOT NULL |
cells_outstanding_leases references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
uuid PK | uuid | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
namespace_id — commit_user_mentions.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
mentioned_users_ids | bigint[] | null | |||
mentioned_projects_ids | bigint[] | null | |||
mentioned_groups_ids | bigint[] | null | |||
commit_id | varchar | NOT NULL | |||
note_id | bigint | NOT NULL | notes.id 1:1 · CASCADE | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
resolved_at | timestamptz | null | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE |
JiHu-specific table
content_blocked_states references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
commit_sha | bytea | NOT NULL | |||
blob_sha | bytea | NOT NULL | |||
path | text | NOT NULL | CHECK char_length(path) <= 2048 | ||
container_identifier | text | NOT NULL | CHECK char_length(container_identifier) <= 255 |
Table checks: char_length(container_identifier) <= 255; char_length(path) <= 2048
created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.conversational_development_index_metrics references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
leader_issues | double precision | NOT NULL | |||
instance_issues | double precision | NOT NULL | |||
leader_notes | double precision | NOT NULL | |||
instance_notes | double precision | NOT NULL | |||
leader_milestones | double precision | NOT NULL | |||
instance_milestones | double precision | NOT NULL | |||
leader_boards | double precision | NOT NULL | |||
instance_boards | double precision | NOT NULL | |||
leader_merge_requests | double precision | NOT NULL | |||
instance_merge_requests | double precision | NOT NULL | |||
leader_ci_pipelines | double precision | NOT NULL | |||
instance_ci_pipelines | double precision | NOT NULL | |||
leader_environments | double precision | NOT NULL | |||
instance_environments | double precision | NOT NULL | |||
leader_deployments | double precision | NOT NULL | |||
instance_deployments | double precision | NOT NULL | |||
leader_projects_prometheus_active | double precision | null | |||
instance_projects_prometheus_active | double precision | null | |||
leader_service_desk_issues | double precision | NOT NULL | |||
instance_service_desk_issues | double precision | NOT NULL | |||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL | |||
percentage_boards | double precision | NOT NULL | 0.0 | ||
percentage_ci_pipelines | double precision | NOT NULL | 0.0 | ||
percentage_deployments | double precision | NOT NULL | 0.0 | ||
percentage_environments | double precision | NOT NULL | 0.0 | ||
percentage_issues | double precision | NOT NULL | 0.0 | ||
percentage_merge_requests | double precision | NOT NULL | 0.0 | ||
percentage_milestones | double precision | NOT NULL | 0.0 | ||
percentage_notes | double precision | NOT NULL | 0.0 | ||
percentage_projects_prometheus_active | double precision | NOT NULL | 0.0 | ||
percentage_service_desk_issues | double precision | NOT NULL | 0.0 | ||
usage_data_id | bigint | null |
country_access_logs references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
access_count_reset_at | timestamptz | null | |||
first_access_at | timestamptz | null | |||
last_access_at | timestamptz | null | |||
user_id | bigint | NOT NULL | |||
access_count | integer | NOT NULL | 0 | ||
country_code | smallint | NOT NULL |
user_id — coverage_fuzzing_corpuses.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
user_id | bigint | null | users.id 1:N · CASCADE | ||
package_id | bigint | NOT NULL | packages_packages.id 1:1 · CASCADE | ||
file_updated_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
project_id, user_id — csv_issue_imports looks like a many-to-many link table but has no unique constraint across (project_id, user_id). The same pair can be inserted twice; every join through it will double-count.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
deployable_type — deployable_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.deployable_type, deployable_id — deployable_id points at different tables depending on deployable_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
iid | integer | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
environment_id | bigint | NOT NULL | environments.id 1:N · CASCADE | ||
ref | varchar | NOT NULL | |||
tag | boolean | NOT NULL | |||
sha | varchar | NOT NULL | |||
user_id | bigint | null | |||
deployable_type | varchar | null | |||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
on_stop | varchar | null | |||
status | smallint | NOT NULL | |||
finished_at | timestamptz | null | |||
deployable_id | bigint | null | |||
archived | boolean | NOT NULL | FALSE |
merge_request_id — description_versions.merge_request_id may be NULL, so the relationship to merge_requests is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.issue_id — description_versions.issue_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.epic_id — description_versions.epic_id may be NULL, so the relationship to epics is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
issue_id | bigint | null | issues.id 1:N · CASCADE | ||
merge_request_id | bigint | null | merge_requests.id 1:N · CASCADE | ||
epic_id | bigint | null | epics.id 1:N · CASCADE | ||
description | text | null | |||
deleted_at | timestamptz | null | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE |
Table checks: num_nonnulls(epic_id, issue_id, merge_request_id) = 1
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | CURRENT_TIMESTAMP | ||
updated_at | timestamptz | NOT NULL | CURRENT_TIMESTAMP | ||
type | smallint | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
relationship | text | null | CHECK char_length(relationship) <= 255 | ||
email | text | null | CHECK char_length(email) <= 255 |
Table checks: char_length(relationship) <= 255; char_length(name) <= 255; char_length(email) <= 255
detached_partitions references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
drop_after | timestamptz | NOT NULL | |||
table_name | text | NOT NULL | CHECK char_length(table_name) <= 63 |
Table checks: char_length(table_name) <= 63
namespace_id — diff_note_positions.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
note_id | bigint | NOT NULL | notes.id 1:N · CASCADE | ||
old_line | integer | null | |||
new_line | integer | null | |||
diff_content_type | smallint | NOT NULL | |||
diff_type | smallint | NOT NULL | |||
line_code | varchar(255) | NOT NULL | |||
base_sha | bytea | NOT NULL | |||
start_sha | bytea | NOT NULL | |||
head_sha | bytea | NOT NULL | |||
old_path | text | NOT NULL | |||
new_path | text | NOT NULL | |||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL
JiHu-specific table
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
integration_id | bigint | NOT NULL | integrations.id 1:N · CASCADE | JiHu-specific column | |
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
corpid | text | null | JiHu-specific columnCHECK char_length(corpid) <= 255 |
Table checks: char_length(corpid) <= 255
project_id — draft_notes.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
merge_request_id | bigint | NOT NULL | merge_requests.id 1:N · CASCADE | ||
author_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
resolve_discussion | boolean | NOT NULL | FALSE | ||
discussion_id | varchar | null | |||
note | text | NOT NULL | |||
position | text | null | |||
original_position | text | null | |||
change_position | text | null | |||
commit_id | bytea | null | |||
line_code | text | null | CHECK char_length(line_code) <= 255 | ||
internal | boolean | NOT NULL | FALSE | ||
note_type | smallint | null | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL; char_length(line_code) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
event_name | text | NOT NULL | CHECK char_length(event_name) <= 255 | ||
event_label | text | null | CHECK char_length(event_label) <= 255 | ||
category | text | null | CHECK char_length(category) <= 255 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
Table checks: char_length(event_label) <= 255; char_length(event_name) <= 255; char_length(category) <= 255
created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.confirmed_at — confirmed_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.confirmation_sent_at — confirmation_sent_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
email | varchar | NOT NULL | |||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
confirmation_token | varchar | null | |||
confirmed_at | timestamp | null | |||
confirmation_sent_at | timestamp | null | |||
detumbled_email | text | null | CHECK char_length(detumbled_email) <= 255 |
Table checks: char_length(detumbled_email) <= 255
merge_request_id — environments.merge_request_id may be NULL, so the relationship to merge_requests is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.cluster_agent_id — environments.cluster_agent_id may be NULL, so the relationship to cluster_agents is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.environment_type — environment_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.state — state is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.merge_request_id, cluster_agent_id — environments has several nullable FKs (merge_request_id, cluster_agent_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
name | varchar | NOT NULL | |||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
external_url | varchar | null | |||
environment_type | varchar | null | |||
state | varchar | NOT NULL | CAST('available' AS varchar) | ||
slug | varchar | NOT NULL | |||
auto_stop_at | timestamptz | null | |||
auto_delete_at | timestamptz | null | |||
tier | smallint | null | |||
merge_request_id | bigint | null | merge_requests.id 1:N · SET NULL | ||
cluster_agent_id | bigint | null | cluster_agents.id 1:N · SET NULL | ||
kubernetes_namespace | text | null | CHECK char_length(kubernetes_namespace) <= 63 | ||
flux_resource_path | text | null | CHECK char_length(flux_resource_path) <= 255 | ||
description | text | null | CHECK char_length(description) <= 10000 | ||
description_html | text | null | CHECK char_length(description_html) <= 50000 | ||
cached_markdown_version | integer | null | |||
auto_stop_setting | smallint | NOT NULL | 0 |
Table checks: char_length(flux_resource_path) <= 255; char_length(description) <= 10000; char_length(kubernetes_namespace) <= 63; char_length(description_html) <= 50000
due_date_sourcing_epic_id — epics.due_date_sourcing_epic_id may be NULL, so the relationship to epics is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.start_date_sourcing_milestone_id — epics.start_date_sourcing_milestone_id may be NULL, so the relationship to milestones is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.due_date_sourcing_milestone_id — epics.due_date_sourcing_milestone_id may be NULL, so the relationship to milestones is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.work_item_parent_link_id — epics.work_item_parent_link_id may be NULL, so the relationship to work_item_parent_links is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.start_date_sourcing_epic_id — epics.start_date_sourcing_epic_id may be NULL, so the relationship to epics is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.closed_by_id — epics.closed_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.assignee_id — epics.assignee_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.issue_id — epics.issue_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.parent_id — epics.parent_id may be NULL, so the relationship to epics is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.due_date_sourcing_epic_id, start_date_sourcing_milestone_id, due_date_sourcing_milestone_id, work_item_parent_link_id, start_date_sourcing_epic_id, closed_by_id, assignee_id, issue_id, parent_id — epics has several nullable FKs (due_date_sourcing_epic_id, start_date_sourcing_milestone_id, due_date_sourcing_milestone_id, work_item_parent_link_id, start_date_sourcing_epic_id, closed_by_id, assignee_id, issue_id, parent_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.last_edited_at — last_edited_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.closed_at — closed_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
author_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
assignee_id | bigint | null | users.id 1:N · SET NULL | ||
iid | integer | NOT NULL | |||
cached_markdown_version | integer | null | |||
updated_by_id | bigint | null | |||
last_edited_by_id | bigint | null | |||
lock_version | integer | null | 0 | CHECK lock_version IS NOT NULL | |
start_date | date | null | |||
end_date | date | null | |||
last_edited_at | timestamp | null | |||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL | |||
title | varchar | NOT NULL | |||
title_html | varchar | NOT NULL | |||
description | text | null | |||
description_html | text | null | |||
start_date_sourcing_milestone_id | bigint | null | milestones.id 1:N · SET NULL | ||
due_date_sourcing_milestone_id | bigint | null | milestones.id 1:N · SET NULL | ||
start_date_fixed | date | null | |||
due_date_fixed | date | null | |||
start_date_is_fixed | boolean | null | |||
due_date_is_fixed | boolean | null | |||
closed_by_id | bigint | null | users.id 1:N · SET NULL | ||
closed_at | timestamp | null | |||
parent_id | bigint | null | epics.id 1:N · SET NULL | ||
relative_position | integer | null | |||
state_id | smallint | NOT NULL | 1 | ||
start_date_sourcing_epic_id | bigint | null | epics.id 1:N · SET NULL | ||
due_date_sourcing_epic_id | bigint | null | epics.id 1:N · SET NULL | ||
confidential | boolean | NOT NULL | FALSE | ||
external_key | varchar(255) | null | |||
color | text | null | CAST('#1068bf' AS text) | CHECK char_length(color) <= 7 | |
total_opened_issue_weight | integer | NOT NULL | 0 | ||
total_closed_issue_weight | integer | NOT NULL | 0 | ||
total_opened_issue_count | integer | NOT NULL | 0 | ||
total_closed_issue_count | integer | NOT NULL | 0 | ||
issue_id | bigint | null | issues.id 1:1 · CASCADE | CHECK issue_id IS NOT NULL | |
imported | smallint | NOT NULL | 0 | ||
imported_from | smallint | NOT NULL | 0 | ||
work_item_parent_link_id | bigint | null | work_item_parent_links.id 1:1 · SET NULL |
Table checks: issue_id IS NOT NULL; char_length(color) <= 7; lock_version IS NOT NULL
personal_namespace_id — events.personal_namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — events.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — events.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.target_type — target_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.target_type, target_id — target_id points at different tables depending on target_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id | bigint | null | projects.id 1:N · CASCADE | ||
author_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
action | smallint | NOT NULL | |||
target_type | varchar | null | |||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
fingerprint | bytea | null | CHECK octet_length(fingerprint) <= 128 | ||
id PK | bigint | NOT NULL | |||
target_id | bigint | null | |||
imported_from | smallint | NOT NULL | 0 | ||
personal_namespace_id | bigint | null | namespaces.id 1:N · CASCADE |
Table checks: octet_length(fingerprint) <= 128; group_id IS NOT NULL OR project_id IS NOT NULL OR personal_namespace_id IS NOT NULL
project_id — evidences.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
release_id | bigint | NOT NULL | releases.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
summary_sha | bytea | null | |||
summary | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
merge_request_id | bigint | NOT NULL | merge_requests.id 1:N · CASCADE |
created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.feature_gates references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
feature_key | varchar | NOT NULL | |||
key | varchar | NOT NULL | |||
value | varchar | null | |||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL |
created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.features references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
key | varchar | NOT NULL | |||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:1 · CASCADE | ||
initiator_user_id | bigint | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
hard_delete | boolean | NOT NULL | FALSE | ||
consume_after | timestamptz | NOT NULL | now() | ||
user_type | smallint | null |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
encrypted_token | varchar(255) | NOT NULL | |||
encrypted_token_iv | varchar(255) | NOT NULL | |||
grafana_url | varchar(1024) | NOT NULL | |||
enabled | boolean | NOT NULL | FALSE |
namespace_id — granular_scopes.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
permissions | jsonb | NOT NULL | CAST('[]' AS jsonb) | CHECK jsonb_typeof(permissions) = CAST('array' AS text) | |
access | smallint | NOT NULL | 0 |
Table checks: jsonb_typeof(permissions) = CAST('array' AS text)
namespace_id — routes.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.source_type — source_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.source_type, source_id — source_id points at different tables depending on source_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
source_id | bigint | NOT NULL | |||
source_type | varchar | NOT NULL | |||
path | varchar | NOT NULL | |||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
name | varchar | null | |||
namespace_id | bigint | null | namespaces.id 1:1 · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
name | varchar | NOT NULL |
created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.historical_data references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
date | date | null | |||
active_user_count | integer | null | |||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
recorded_at | timestamptz | null | CHECK recorded_at IS NOT NULL |
Table checks: recorded_at IS NOT NULL
entity_type, entity_id — entity_id points at different tables depending on entity_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
entity_id | bigint | NOT NULL | |||
l0_delivered_at | timestamptz | null | |||
l2_delivered_at | timestamptz | null | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
event_type | smallint | NOT NULL | |||
l0_attempts | smallint | NOT NULL | 0 | ||
l2_attempts | smallint | NOT NULL | 0 | ||
entity_type | text | NOT NULL | CHECK char_length(entity_type) <= 255 | ||
l0_last_error | text | null | CHECK char_length(l0_last_error) <= 4096 | ||
l2_last_error | text | null | CHECK char_length(l2_last_error) <= 4096 | ||
payload | jsonb | NOT NULL | CAST('{}' AS jsonb) |
Table checks: char_length(l2_last_error) <= 4096; char_length(l0_last_error) <= 4096; char_length(entity_type) <= 255
user_id — identities.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.saml_provider_id — identities.saml_provider_id may be NULL, so the relationship to saml_providers is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id, saml_provider_id — identities has several nullable FKs (user_id, saml_provider_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
extern_uid | varchar | null | |||
provider | varchar | null | |||
user_id | bigint | null | users.id 1:N · CASCADE | CHECK user_id IS NOT NULL | |
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
secondary_extern_uid | varchar | null | |||
saml_provider_id | bigint | null | saml_providers.id 1:N · CASCADE | ||
trusted_extern_uid | boolean | null | TRUE |
Table checks: user_id IS NOT NULL
indexed_at — indexed_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:1 · CASCADE | ||
indexed_at | timestamp | null | |||
note | text | null | |||
last_commit | varchar | null | |||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL | |||
last_wiki_commit | bytea | null | |||
wiki_indexed_at | timestamptz | null |
namespace_id, project_id — insights looks like a many-to-many link table but has no unique constraint across (namespace_id, project_id). The same pair can be inserted twice; every join through it will double-count.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE |
project_id — integrations.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id — integrations.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — integrations.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.inherit_from_id — integrations.inherit_from_id may be NULL, so the relationship to integrations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.category — category is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.project_id, organization_id, group_id, inherit_from_id — integrations has several nullable FKs (project_id, organization_id, group_id, inherit_from_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
active | boolean | NOT NULL | FALSE | ||
push_events | boolean | null | TRUE | ||
issues_events | boolean | null | TRUE | ||
merge_requests_events | boolean | null | TRUE | ||
tag_push_events | boolean | null | TRUE | ||
note_events | boolean | NOT NULL | TRUE | ||
category | varchar | NOT NULL | CAST('common' AS varchar) | ||
wiki_page_events | boolean | null | TRUE | ||
pipeline_events | boolean | NOT NULL | FALSE | ||
confidential_issues_events | boolean | NOT NULL | TRUE | ||
commit_events | boolean | NOT NULL | TRUE | ||
job_events | boolean | NOT NULL | FALSE | ||
confidential_note_events | boolean | null | TRUE | ||
deployment_events | boolean | NOT NULL | FALSE | ||
comment_on_event_enabled | boolean | NOT NULL | TRUE | ||
instance | boolean | NOT NULL | FALSE | ||
comment_detail | smallint | null | |||
inherit_from_id | bigint | null | integrations.id 1:N · CASCADE | ||
alert_events | boolean | null | |||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
type_new | text | null | CHECK char_length(type_new) <= 255 | ||
vulnerability_events | boolean | NOT NULL | FALSE | ||
archive_trace_events | boolean | NOT NULL | FALSE | ||
encrypted_properties | bytea | null | |||
encrypted_properties_iv | bytea | null | |||
incident_events | boolean | NOT NULL | FALSE | ||
group_mention_events | boolean | NOT NULL | FALSE | ||
group_confidential_mention_events | boolean | NOT NULL | FALSE | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE | ||
event_filters | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
filter | jsonb | NOT NULL | CAST('{}' AS jsonb) |
Table checks: num_nonnulls(group_id, organization_id, project_id) = 1; char_length(type_new) <= 255
namespace_id — internal_ids.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — internal_ids.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
usage | integer | NOT NULL | |||
last_value | integer | NOT NULL | |||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE |
Table checks: num_nonnulls(namespace_id, project_id) = 1
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
range | varchar | NOT NULL |
author_id — issues.author_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.sprint_id — issues.sprint_id may be NULL, so the relationship to sprints is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — issues.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.milestone_id — issues.milestone_id may be NULL, so the relationship to milestones is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.duplicated_to_id — issues.duplicated_to_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.moved_to_id — issues.moved_to_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.closed_by_id — issues.closed_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id — issues.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.promoted_to_epic_id — issues.promoted_to_epic_id may be NULL, so the relationship to epics is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.updated_by_id — issues.updated_by_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.author_id, sprint_id, project_id, milestone_id, duplicated_to_id, moved_to_id, closed_by_id, namespace_id, promoted_to_epic_id, updated_by_id — issues has several nullable FKs (author_id, sprint_id, project_id, milestone_id, duplicated_to_id, moved_to_id, closed_by_id, namespace_id, promoted_to_epic_id, updated_by_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.last_edited_at — last_edited_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
title | varchar | null | |||
author_id | bigint | null | users.id 1:N · SET NULL | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
description | text | null | |||
milestone_id | bigint | null | milestones.id 1:N · SET NULL | ||
iid | integer | null | |||
updated_by_id | bigint | null | users.id 1:N · SET NULL | ||
weight | integer | null | |||
confidential | boolean | NOT NULL | FALSE | ||
due_date | date | null | |||
moved_to_id | bigint | null | issues.id 1:N · SET NULL | ||
lock_version | integer | null | 0 | CHECK lock_version IS NOT NULL | |
title_html | text | null | |||
description_html | text | null | |||
time_estimate | integer | null | 0 | ||
relative_position | integer | null | |||
service_desk_reply_to | varchar | null | |||
cached_markdown_version | integer | null | |||
last_edited_at | timestamp | null | |||
last_edited_by_id | bigint | null | |||
discussion_locked | boolean | null | |||
closed_at | timestamptz | null | |||
closed_by_id | bigint | null | users.id 1:N · SET NULL | ||
state_id | smallint | NOT NULL | 1 | ||
duplicated_to_id | bigint | null | issues.id 1:N · SET NULL | ||
promoted_to_epic_id | bigint | null | epics.id 1:N · SET NULL | ||
health_status | smallint | null | |||
sprint_id | bigint | null | sprints.id 1:N · SET NULL | ||
blocking_issues_count | integer | NOT NULL | 0 | ||
upvotes_count | integer | NOT NULL | 0 | ||
work_item_type_id | bigint | null | CHECK work_item_type_id IS NOT NULL | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL | |
start_date | date | null | |||
imported_from | smallint | NOT NULL | 0 | ||
namespace_traversal_ids | bigint[] | null | CAST('{}' AS bigint[]) |
Table checks: work_item_type_id IS NOT NULL; namespace_id IS NOT NULL; lock_version IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
group_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
start_date | date | null | |||
duration_in_weeks | integer | null | |||
iterations_in_advance | integer | null | |||
active | boolean | NOT NULL | TRUE | ||
automatic | boolean | NOT NULL | TRUE | ||
title | text | NOT NULL | CHECK char_length(title) <= 255 | ||
roll_over | boolean | NOT NULL | FALSE | ||
description | text | null | CHECK char_length(description) <= 5000 | ||
next_run_date | date | null |
Table checks: char_length(description) <= 5000; char_length(title) <= 255
deployment_id — job_environments.deployment_id may be NULL, so the relationship to deployments is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
environment_id | bigint | NOT NULL | environments.id 1:N · CASCADE | ||
ci_pipeline_id | bigint | NOT NULL | |||
ci_job_id | bigint | NOT NULL | |||
deployment_id | bigint | null | deployments.id 1:N · CASCADE | ||
expanded_environment_name | text | NOT NULL | CHECK char_length(expanded_environment_name) <= 255 | ||
options | jsonb | NOT NULL | CAST('{}' AS jsonb) |
Table checks: char_length(expanded_environment_name) <= 255
organization_id — keys.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.type — type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.last_used_at — last_used_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | null | |||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
key | text | null | |||
title | varchar | null | |||
type | varchar | null | |||
fingerprint | varchar | null | |||
public | boolean | NOT NULL | FALSE | ||
last_used_at | timestamp | null | |||
fingerprint_sha256 | bytea | null | |||
expires_at | timestamptz | null | |||
expiry_notification_delivered_at | timestamptz | null | |||
before_expiry_notification_delivered_at | timestamptz | null | |||
usage_type | smallint | NOT NULL | 0 | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE | CHECK organization_id IS NOT NULL |
Table checks: organization_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
root_namespace_id | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
project_id — labels.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id — labels.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — labels.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.type — type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
title | varchar | null | |||
color | varchar | null | |||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
template | boolean | null | FALSE | ||
description | varchar | null | |||
description_html | text | null | |||
type | varchar | null | |||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
cached_markdown_version | integer | null | |||
lock_on_merge | boolean | NOT NULL | FALSE | ||
archived | boolean | NOT NULL | FALSE | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE |
Table checks: num_nonnulls(group_id, organization_id, project_id) = 1
created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.licenses references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
data | text | NOT NULL | |||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
cloud | boolean | null | FALSE | ||
last_synced_at | timestamptz | null |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
list_id | bigint | NOT NULL | lists.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
collapsed | boolean | null |
custom_status_id — lists.custom_status_id may be NULL, so the relationship to work_item_custom_statuses is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.iteration_id — lists.iteration_id may be NULL, so the relationship to sprints is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — lists.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.label_id — lists.label_id may be NULL, so the relationship to labels is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id — lists.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — lists.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.milestone_id — lists.milestone_id may be NULL, so the relationship to milestones is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.custom_status_id, iteration_id, project_id, label_id, user_id, group_id, milestone_id — lists has several nullable FKs (custom_status_id, iteration_id, project_id, label_id, user_id, group_id, milestone_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
board_id | bigint | NOT NULL | boards.id 1:N · CASCADE | ||
label_id | bigint | null | labels.id 1:N · CASCADE | ||
list_type | integer | NOT NULL | 1 | ||
position | integer | null | |||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL | |||
user_id | bigint | null | users.id 1:N · CASCADE | ||
milestone_id | bigint | null | milestones.id 1:N · CASCADE | ||
max_issue_count | integer | NOT NULL | 0 | ||
max_issue_weight | integer | NOT NULL | 0 | ||
limit_metric | varchar(20) | null | |||
iteration_id | bigint | null | sprints.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
custom_status_id | bigint | null | work_item_custom_statuses.id 1:N · CASCADE | ||
system_defined_status_identifier | smallint | null |
Table checks: list_type <> 6 OR num_nonnulls(system_defined_status_identifier, custom_status_id) = 1; num_nonnulls(group_id, project_id) = 1
project_id — milestone_releases.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
milestone_id PK | bigint | NOT NULL | milestones.id 1:N · CASCADE | ||
release_id PK | bigint | NOT NULL | releases.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
group_id — milestones.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — milestones.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.state — state is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
title | varchar | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
description | text | null | |||
due_date | date | null | |||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
state | varchar | null | |||
iid | integer | null | |||
title_html | text | null | |||
description_html | text | null | |||
start_date | date | null | |||
cached_markdown_version | integer | null | |||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
lock_version | integer | NOT NULL | 0 |
Table checks: num_nonnulls(group_id, project_id) = 1
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
last_seen_at | timestamptz | NOT NULL | now() | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
platform | smallint | NOT NULL | 0 | ||
apns_environment | smallint | NOT NULL | 0 | ||
payload_mode | smallint | NOT NULL | 0 | ||
device_token | jsonb | NOT NULL | |||
bundle_identifier | text | null | CHECK char_length(bundle_identifier) <= 255 | ||
device_name | text | null | CHECK char_length(device_name) <= 255 | ||
app_version | text | null | CHECK char_length(app_version) <= 64 | ||
locale | text | null | CHECK char_length(locale) <= 32 |
Table checks: char_length(locale) <= 32; char_length(app_version) <= 64; char_length(bundle_identifier) <= 255; char_length(device_name) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
recorded_at | timestamptz | NOT NULL | |||
payload | jsonb | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
metadata | jsonb | null |
review_id — notes.review_id may be NULL, so the relationship to reviews is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.namespace_id — notes.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — notes.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id — notes.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.type — type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.noteable_type, noteable_id — noteable_id points at different tables depending on noteable_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.review_id, namespace_id, project_id, organization_id — notes has several nullable FKs (review_id, namespace_id, project_id, organization_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.resolved_at — resolved_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
note | text | null | |||
noteable_type | varchar | null | CHECK noteable_type IS NOT NULL | ||
author_id | bigint | null | |||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
line_code | varchar | null | |||
commit_id | varchar | null | |||
noteable_id | bigint | null | |||
system | boolean | NOT NULL | FALSE | ||
st_diff | text | null | |||
updated_by_id | bigint | null | |||
type | varchar | null | |||
position | text | null | |||
original_position | text | null | |||
resolved_at | timestamp | null | |||
resolved_by_id | bigint | null | |||
discussion_id | varchar | null | |||
note_html | text | null | |||
cached_markdown_version | integer | null | |||
change_position | text | null | |||
resolved_by_push | boolean | null | |||
review_id | bigint | null | reviews.id 1:N · SET NULL | ||
confidential | boolean | null | |||
last_edited_at | timestamptz | null | |||
internal | boolean | NOT NULL | FALSE | ||
id PK | bigint | NOT NULL | |||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | ||
imported_from | smallint | NOT NULL | 0 | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE |
Table checks: noteable_type IS NOT NULL; num_nonnulls(namespace_id, organization_id, project_id) >= 1
source_type — source_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.source_type, source_id — source_id points at different tables depending on source_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.notification_settings references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | |||
source_id | bigint | null | |||
source_type | varchar | null | |||
level | integer | NOT NULL | 0 | ||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL | |||
new_note | boolean | null | |||
new_issue | boolean | null | |||
reopen_issue | boolean | null | |||
close_issue | boolean | null | |||
reassign_issue | boolean | null | |||
new_merge_request | boolean | null | |||
reopen_merge_request | boolean | null | |||
close_merge_request | boolean | null | |||
reassign_merge_request | boolean | null | |||
merge_merge_request | boolean | null | |||
failed_pipeline | boolean | null | |||
success_pipeline | boolean | null | |||
push_to_merge_request | boolean | null | |||
issue_due | boolean | null | |||
new_epic | boolean | null | |||
notification_email | varchar | null | |||
fixed_pipeline | boolean | null | |||
new_release | boolean | null | |||
moved_project | boolean | NOT NULL | TRUE | ||
change_reviewer_merge_request | boolean | null | |||
merge_when_pipeline_succeeds | boolean | NOT NULL | FALSE | ||
approver | boolean | NOT NULL | FALSE | ||
service_account_failed_pipeline | boolean | NOT NULL | FALSE | ||
service_account_success_pipeline | boolean | NOT NULL | FALSE | ||
service_account_fixed_pipeline | boolean | NOT NULL | FALSE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
merge_request_created_at | timestamptz | null | |||
pipeline_created_at | timestamptz | null | |||
user_added_at | timestamptz | null | |||
trial_started_at | timestamptz | null | |||
required_mr_approvals_enabled_at | timestamptz | null | |||
code_owners_enabled_at | timestamptz | null | |||
issue_created_at | timestamptz | null | |||
secure_dependency_scanning_run_at | timestamptz | null | |||
secure_dast_run_at | timestamptz | null | |||
license_scanning_run_at | timestamptz | null | |||
code_added_at | timestamptz | null | |||
ended_at | timestamptz | null | |||
duo_seat_assigned_at | timestamptz | null |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
name | text | NOT NULL | CAST('' AS text) | CHECK char_length(name) <= 255 | |
path | text | NOT NULL | CHECK char_length(path) <= 255 | ||
visibility_level | smallint | NOT NULL | 0 | ||
state | smallint | NOT NULL | 0 | ||
uuid | uuid | NOT NULL | gen_random_uuid_v7() |
Table checks: char_length(path) <= 255; char_length(name) <= 255
project_id — path_locks.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id — path_locks.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id, user_id — path_locks has several nullable FKs (project_id, user_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
path | varchar | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL | |
user_id | bigint | null | users.id 1:N · CASCADE | ||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL |
Table checks: project_id IS NOT NULL
project_id — snippets.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.type — type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
title | varchar | null | |||
content | text | null | |||
author_id | bigint | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
file_name | varchar | null | |||
type | varchar | null | |||
visibility_level | integer | NOT NULL | 0 | ||
title_html | text | null | |||
content_html | text | null | |||
cached_markdown_version | integer | null | |||
description | text | null | |||
description_html | text | null | |||
encrypted_secret_token | varchar(255) | null | |||
encrypted_secret_token_iv | varchar(255) | null | |||
secret | boolean | NOT NULL | FALSE | ||
repository_read_only | boolean | NOT NULL | FALSE | ||
imported_from | smallint | NOT NULL | 0 | ||
organization_id | bigint | null |
Table checks: num_nonnulls(organization_id, project_id) = 1
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
user_id PK | bigint | NOT NULL | users.id 1:1 · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
initial_email_sent_at | timestamptz | null | |||
last_access_from_pipl_country_at | timestamptz | NOT NULL | |||
state | smallint | NOT NULL | 0 |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
plan_id | bigint | NOT NULL | plans.id 1:1 · CASCADE | ||
ci_pipeline_size | integer | NOT NULL | 0 | ||
ci_active_jobs | integer | NOT NULL | 0 | ||
project_hooks | integer | NOT NULL | 100 | ||
group_hooks | integer | NOT NULL | 50 | ||
ci_project_subscriptions | integer | NOT NULL | 2 | ||
ci_pipeline_schedules | integer | NOT NULL | 10 | ||
offset_pagination_limit | integer | NOT NULL | 50000 | ||
ci_instance_level_variables | integer | NOT NULL | 25 | ||
storage_size_limit | integer | NOT NULL | 0 | ||
ci_max_artifact_size_lsif | integer | NOT NULL | 200 | ||
ci_max_artifact_size_archive | integer | NOT NULL | 0 | ||
ci_max_artifact_size_metadata | integer | NOT NULL | 0 | ||
ci_max_artifact_size_trace | integer | NOT NULL | 0 | ||
ci_max_artifact_size_junit | integer | NOT NULL | 0 | ||
ci_max_artifact_size_sast | integer | NOT NULL | 0 | ||
ci_max_artifact_size_dependency_scanning | integer | NOT NULL | 350 | ||
ci_max_artifact_size_container_scanning | integer | NOT NULL | 150 | ||
ci_max_artifact_size_dast | integer | NOT NULL | 0 | ||
ci_max_artifact_size_codequality | integer | NOT NULL | 0 | ||
ci_max_artifact_size_license_management | integer | NOT NULL | 0 | ||
ci_max_artifact_size_license_scanning | integer | NOT NULL | 100 | ||
ci_max_artifact_size_performance | integer | NOT NULL | 0 | ||
ci_max_artifact_size_metrics | integer | NOT NULL | 0 | ||
ci_max_artifact_size_metrics_referee | integer | NOT NULL | 0 | ||
ci_max_artifact_size_network_referee | integer | NOT NULL | 0 | ||
ci_max_artifact_size_dotenv | integer | NOT NULL | 0 | ||
ci_max_artifact_size_cobertura | integer | NOT NULL | 0 | ||
ci_max_artifact_size_terraform | integer | NOT NULL | 5 | ||
ci_max_artifact_size_accessibility | integer | NOT NULL | 0 | ||
ci_max_artifact_size_cluster_applications | integer | NOT NULL | 0 | ||
ci_max_artifact_size_secret_detection | integer | NOT NULL | 0 | ||
ci_max_artifact_size_requirements | integer | NOT NULL | 0 | ||
ci_max_artifact_size_coverage_fuzzing | integer | NOT NULL | 0 | ||
ci_max_artifact_size_browser_performance | integer | NOT NULL | 0 | ||
ci_max_artifact_size_load_performance | integer | NOT NULL | 0 | ||
ci_needs_size_limit | integer | NOT NULL | 50 | ||
conan_max_file_size | bigint | NOT NULL | CAST('3221225472' AS bigint) | ||
maven_max_file_size | bigint | NOT NULL | CAST('3221225472' AS bigint) | ||
npm_max_file_size | bigint | NOT NULL | 524288000 | ||
nuget_max_file_size | bigint | NOT NULL | 524288000 | ||
pypi_max_file_size | bigint | NOT NULL | CAST('3221225472' AS bigint) | ||
generic_packages_max_file_size | bigint | NOT NULL | CAST('5368709120' AS bigint) | ||
golang_max_file_size | bigint | NOT NULL | 104857600 | ||
debian_max_file_size | bigint | NOT NULL | CAST('3221225472' AS bigint) | ||
project_feature_flags | integer | NOT NULL | 200 | ||
ci_max_artifact_size_api_fuzzing | integer | NOT NULL | 0 | ||
ci_pipeline_deployments | integer | NOT NULL | 500 | ||
pull_mirror_interval_seconds | integer | NOT NULL | 300 | ||
daily_invites | integer | NOT NULL | 0 | ||
rubygems_max_file_size | bigint | NOT NULL | CAST('3221225472' AS bigint) | ||
terraform_module_max_file_size | bigint | NOT NULL | 1073741824 | ||
helm_max_file_size | bigint | NOT NULL | 5242880 | ||
ci_registered_group_runners | integer | NOT NULL | 1000 | ||
ci_registered_project_runners | integer | NOT NULL | 1000 | ||
ci_daily_pipeline_schedule_triggers | integer | NOT NULL | 0 | ||
ci_max_artifact_size_running_container_scanning | integer | NOT NULL | 0 | ||
ci_max_artifact_size_cluster_image_scanning | integer | NOT NULL | 0 | ||
ci_jobs_trace_size_limit | integer | NOT NULL | 100 | ||
pages_file_entries | integer | NOT NULL | 200000 | ||
dast_profile_schedules | integer | NOT NULL | 1 | ||
external_audit_event_destinations | integer | NOT NULL | 5 | ||
dotenv_variables | integer | NOT NULL | 20 | ||
dotenv_size | integer | NOT NULL | 5120 | ||
pipeline_triggers | integer | NOT NULL | 25000 | ||
project_ci_secure_files | integer | NOT NULL | 100 | ||
repository_size | bigint | null | |||
security_policy_scan_execution_schedules | integer | NOT NULL | 0 | ||
web_hook_calls_mid | integer | NOT NULL | 0 | ||
web_hook_calls_low | integer | NOT NULL | 0 | ||
project_ci_variables | integer | NOT NULL | 8000 | ||
group_ci_variables | integer | NOT NULL | 30000 | ||
ci_max_artifact_size_cyclonedx | integer | NOT NULL | 1 | ||
rpm_max_file_size | bigint | NOT NULL | CAST('5368709120' AS bigint) | ||
ci_max_artifact_size_requirements_v2 | integer | NOT NULL | 0 | ||
pipeline_hierarchy_size | integer | NOT NULL | 1000 | ||
enforcement_limit | integer | NOT NULL | 0 | ||
notification_limit | integer | NOT NULL | 0 | ||
dashboard_limit_enabled_at | timestamptz | null | |||
web_hook_calls | integer | NOT NULL | 0 | ||
project_access_token_limit | integer | NOT NULL | 0 | ||
google_cloud_logging_configurations | integer | NOT NULL | 5 | ||
ml_model_max_file_size | bigint | NOT NULL | CAST('10737418240' AS bigint) | ||
limits_history | jsonb | NOT NULL | CAST('{}' AS jsonb) | ||
updated_at | timestamptz | NOT NULL | now() | ||
ci_max_artifact_size_annotations | integer | NOT NULL | 0 | ||
ci_job_annotations_size | integer | NOT NULL | 81920 | ||
ci_job_annotations_num | integer | NOT NULL | 20 | ||
file_size_limit_mb | double precision | NOT NULL | 100.0 | ||
audit_events_amazon_s3_configurations | integer | NOT NULL | 5 | ||
ci_max_artifact_size_repository_xray | bigint | NOT NULL | 1073741824 | ||
active_versioned_pages_deployments_limit_by_namespace | integer | NOT NULL | 1000 | ||
ci_max_artifact_size_jacoco | bigint | NOT NULL | 0 | ||
import_placeholder_user_limit_tier_1 | integer | NOT NULL | 0 | ||
import_placeholder_user_limit_tier_2 | integer | NOT NULL | 0 | ||
import_placeholder_user_limit_tier_3 | integer | NOT NULL | 0 | ||
import_placeholder_user_limit_tier_4 | integer | NOT NULL | 0 | ||
ci_max_artifact_size_slsa_provenance_statement | bigint | NOT NULL | 0 | ||
cargo_max_file_size | bigint | NOT NULL | CAST('5368709120' AS bigint) | ||
plan_name_uid | smallint | null | |||
ci_max_artifact_size_sarif | integer | NOT NULL | 10 | ||
max_pipelines_per_merge_train | smallint | NOT NULL | 20 |
created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL | |||
name | varchar | null | |||
title | varchar | null | |||
plan_name_uid | smallint | null | CHECK plan_name_uid IS NOT NULL |
Table checks: plan_name_uid IS NOT NULL
organization_id — pool_repositories.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.state — state is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
shard_id | bigint | NOT NULL | shards.id 1:N · RESTRICT | ||
disk_path | varchar | null | |||
state | varchar | null | |||
source_project_id | bigint | null | |||
organization_id | bigint | null | organizations.id 1:N · CASCADE | CHECK organization_id IS NOT NULL |
Table checks: organization_id IS NOT NULL
programming_languages references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
name | varchar | NOT NULL | |||
color | varchar | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
language_id | bigint | null |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
recorded_at | timestamptz | NOT NULL | |||
payload | jsonb | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
recorded_at | timestamptz | NOT NULL | |||
sent_at | timestamptz | null | |||
payload | jsonb | NOT NULL | |||
version_usage_data_id_value | bigint | null | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE |
namespace_id — redirect_routes.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.source_type — source_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.source_type, source_id — source_id points at different tables depending on source_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
source_id | bigint | NOT NULL | |||
source_type | varchar | NOT NULL | |||
path | varchar | NOT NULL | |||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL | |||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
relation | smallint | NOT NULL | |||
status | smallint | NOT NULL | 0 |
project_id — release_links.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
release_id | bigint | NOT NULL | releases.id 1:N · CASCADE | ||
url | varchar | NOT NULL | |||
name | varchar | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
filepath | varchar(128) | null | |||
link_type | smallint | null | 0 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL
project_id — releases.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.author_id — releases.author_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id, author_id — releases has several nullable FKs (project_id, author_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
tag | varchar | null | CHECK tag IS NOT NULL | ||
description | text | null | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL | |
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL | |||
description_html | text | null | |||
cached_markdown_version | integer | null | |||
author_id | bigint | null | users.id 1:N · SET NULL | ||
name | varchar | null | |||
sha | varchar | null | |||
released_at | timestamptz | NOT NULL | |||
release_published_at | timestamptz | null |
Table checks: project_id IS NOT NULL; tag IS NOT NULL
project_id — remote_mirrors.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.update_status — update_status is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.last_update_at — last_update_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.last_successful_update_at — last_successful_update_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.last_update_started_at — last_update_started_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL | |
url | varchar | null | |||
enabled | boolean | null | FALSE | ||
update_status | varchar | null | |||
last_update_at | timestamp | null | |||
last_successful_update_at | timestamp | null | |||
last_error | varchar | null | |||
encrypted_credentials | text | null | |||
encrypted_credentials_iv | varchar | null | |||
encrypted_credentials_salt | varchar | null | |||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL | |||
last_update_started_at | timestamp | null | |||
only_protected_branches | boolean | NOT NULL | FALSE | ||
remote_name | varchar | null | |||
error_notification_sent | boolean | null | |||
keep_divergent_refs | boolean | null | |||
mirror_branch_regex | text | null | CHECK char_length(mirror_branch_regex) <= 255 |
Table checks: project_id IS NOT NULL; char_length(mirror_branch_regex) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
project_id PK | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
programming_language_id PK | bigint | NOT NULL | |||
share | double precision | NOT NULL | |||
language_id | bigint | null |
protected_branch_project_id — required_code_owners_sections.protected_branch_project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.protected_branch_namespace_id — required_code_owners_sections.protected_branch_namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
protected_branch_id | bigint | NOT NULL | protected_branches.id 1:N · CASCADE | ||
name | text | NOT NULL | CHECK char_length(name) <= 1024 | ||
protected_branch_project_id | bigint | null | projects.id 1:N · CASCADE | ||
protected_branch_namespace_id | bigint | null | namespaces.id 1:N · CASCADE |
Table checks: char_length(name) <= 1024; num_nonnulls(protected_branch_namespace_id, protected_branch_project_id) = 1
author_id — reviews.author_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
author_id | bigint | null | users.id 1:N · SET NULL | ||
merge_request_id | bigint | NOT NULL | merge_requests.id 1:N · CASCADE | ||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL |
schema_migrations references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
version PK | varchar | NOT NULL | |||
finished_at | timestamptz | null | now() |
namespace_id — sentry_issues.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
issue_id | bigint | NOT NULL | issues.id 1:1 · CASCADE | ||
sentry_issue_identifier | bigint | NOT NULL | |||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
build_id | bigint | null | |||
status | smallint | NOT NULL | 0 | ||
expire_at | timestamptz | null | |||
predicate_kind | smallint | NOT NULL | 0 | ||
predicate_type | text | NOT NULL | CHECK char_length(predicate_type) <= 255 | ||
subject_digest | text | NOT NULL | CHECK char_length(subject_digest) <= 255 | ||
file | text | null | CHECK char_length(file) <= 255 | ||
file_store | smallint | null | 1 | ||
iid | integer | null | |||
predicate_file | text | null | CHECK char_length(predicate_file) <= 1024 | ||
predicate_file_store | smallint | NOT NULL | 1 |
Table checks: char_length(file) <= 255; char_length(subject_digest) <= 255; char_length(predicate_type) <= 255; char_length(predicate_file) <= 1024
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
subject | varchar | NOT NULL | |||
issuer | varchar | NOT NULL |
custom_software_license_id — software_license_policies.custom_software_license_id may be NULL, so the relationship to custom_software_licenses is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.approval_policy_rule_id — software_license_policies.approval_policy_rule_id may be NULL, so the relationship to approval_policy_rules is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.scan_result_policy_id — software_license_policies.scan_result_policy_id may be NULL, so the relationship to scan_result_policies is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.custom_software_license_id, approval_policy_rule_id, scan_result_policy_id — software_license_policies has several nullable FKs (custom_software_license_id, approval_policy_rule_id, scan_result_policy_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
classification | integer | NOT NULL | 0 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
scan_result_policy_id | bigint | null | scan_result_policies.id 1:N · CASCADE | ||
custom_software_license_id | bigint | null | custom_software_licenses.id 1:N · CASCADE | ||
approval_policy_rule_id | bigint | null | approval_policy_rules.id 1:N · CASCADE | ||
software_license_spdx_identifier | text | null | CHECK char_length(software_license_spdx_identifier) <= 255 |
Table checks: num_nonnulls(custom_software_license_id, software_license_spdx_identifier) = 1; char_length(software_license_spdx_identifier) <= 255
user_id — spam_logs.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id — spam_logs.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.noteable_type — noteable_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.user_id, organization_id — spam_logs has several nullable FKs (user_id, organization_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | null | users.id 1:N · CASCADE | CHECK user_id IS NOT NULL | |
source_ip | varchar | null | |||
user_agent | varchar | null | |||
via_api | boolean | null | |||
noteable_type | varchar | null | |||
title | varchar | null | |||
description | text | null | |||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL | |||
submitted_as_ham | boolean | NOT NULL | FALSE | ||
recaptcha_verified | boolean | NOT NULL | FALSE | ||
target_id | bigint | null | |||
organization_id | bigint | null | organizations.id 1:N · CASCADE | CHECK organization_id IS NOT NULL |
Table checks: organization_id IS NOT NULL; user_id IS NOT NULL
iterations_cadence_id — sprints.iterations_cadence_id may be NULL, so the relationship to iterations_cadences is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — sprints.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.iterations_cadence_id, group_id — sprints has several nullable FKs (iterations_cadence_id, group_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
start_date | date | null | CHECK start_date IS NOT NULL | ||
due_date | date | null | CHECK due_date IS NOT NULL | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK group_id IS NOT NULL | |
iid | integer | NOT NULL | |||
cached_markdown_version | integer | null | |||
title | text | null | CHECK char_length(title) <= 255 | ||
title_html | text | null | |||
description | text | null | |||
description_html | text | null | |||
state_enum | smallint | NOT NULL | 1 | ||
iterations_cadence_id | bigint | null | iterations_cadences.id 1:N · CASCADE | ||
sequence | integer | null |
Table checks: group_id IS NOT NULL; char_length(title) <= 255; start_date IS NOT NULL; due_date IS NOT NULL
user_id — ssh_signatures.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.key_id — ssh_signatures.key_id may be NULL, so the relationship to keys is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.user_id, key_id — ssh_signatures has several nullable FKs (user_id, key_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
key_id | bigint | null | keys.id 1:N · SET NULL | ||
verification_status | smallint | NOT NULL | 0 | ||
commit_sha | bytea | NOT NULL | |||
user_id | bigint | null | users.id 1:N · SET NULL | ||
key_fingerprint_sha256 | bytea | null | |||
committer_email | text | null | CHECK char_length(committer_email) <= 255 |
Table checks: char_length(committer_email) <= 255
user_id — subscriptions.user_id may be NULL, so the relationship to users is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id — subscriptions.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.subscribable_type — subscribable_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.subscribable_type, subscribable_id — subscribable_id points at different tables depending on subscribable_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.user_id, project_id — subscriptions has several nullable FKs (user_id, project_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | null | users.id 1:N · CASCADE | CHECK user_id IS NOT NULL | |
subscribable_id | bigint | null | |||
subscribable_type | varchar | null | |||
subscribed | boolean | null | |||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
project_id | bigint | null | projects.id 1:N · CASCADE |
Table checks: user_id IS NOT NULL
namespace_id — suggestions.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
relative_order | smallint | NOT NULL | |||
applied | boolean | NOT NULL | FALSE | ||
commit_id | varchar | null | |||
from_content | text | NOT NULL | |||
to_content | text | NOT NULL | |||
lines_above | integer | NOT NULL | 0 | ||
lines_below | integer | NOT NULL | 0 | ||
outdated | boolean | NOT NULL | FALSE | ||
note_id | bigint | NOT NULL | notes.id 1:N · CASCADE | ||
namespace_id | bigint | null | namespaces.id 1:N · CASCADE | CHECK namespace_id IS NOT NULL |
Table checks: namespace_id IS NOT NULL
project_id — supply_chain_attestation_states.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
supply_chain_attestation_id | bigint | NOT NULL | |||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | NOT NULL | 0 | ||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 | ||
project_id | bigint | null | projects.id 1:N · CASCADE | CHECK project_id IS NOT NULL |
Table checks: project_id IS NOT NULL; char_length(verification_failure) <= 255
tags references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
name | varchar | null | |||
taggings_count | integer | null | 0 |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
target_branch | text | NOT NULL |
Table checks: char_length(name) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
term_id | bigint | NOT NULL | application_setting_terms.id 1:N | ||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
accepted | boolean | NOT NULL | FALSE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL |
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
billing_rate | numeric(18, 4) | null | 0.0 | ||
billable | boolean | NOT NULL | FALSE | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
description | text | null | CHECK char_length(description) <= 1024 | ||
color | text | NOT NULL | CAST('#6699cc' AS text) | CHECK char_length(color) <= 7 |
Table checks: char_length(name) <= 255; char_length(color) <= 7; char_length(description) <= 1024
project_id — timelogs.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.timelog_category_id — timelogs.timelog_category_id may be NULL, so the relationship to timelog_categories is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.issue_id — timelogs.issue_id may be NULL, so the relationship to issues is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.merge_request_id — timelogs.merge_request_id may be NULL, so the relationship to merge_requests is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.note_id — timelogs.note_id may be NULL, so the relationship to notes is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.project_id, timelog_category_id, issue_id, merge_request_id, note_id — timelogs has several nullable FKs (project_id, timelog_category_id, issue_id, merge_request_id, note_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.spent_at — spent_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
time_spent | integer | NOT NULL | |||
user_id | bigint | null | |||
created_at | timestamp | NOT NULL | |||
updated_at | timestamp | NOT NULL | |||
issue_id | bigint | null | issues.id 1:N · CASCADE | ||
merge_request_id | bigint | null | merge_requests.id 1:N · CASCADE | ||
spent_at | timestamp | null | now() | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
summary | text | null | CHECK char_length(summary) <= 255 | ||
note_id | bigint | null | notes.id 1:N · SET NULL | ||
timelog_category_id | bigint | null | timelog_categories.id 1:N · SET NULL | ||
namespace_id | bigint | NOT NULL | namespaces.id 1:N · CASCADE |
Table checks: char_length(summary) <= 255; num_nonnulls(issue_id, merge_request_id) = 1
project_id — todos.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id — todos.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.note_id — todos.note_id may be NULL, so the relationship to notes is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — todos.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.target_type — target_type is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.state — state is a short string that clearly takes a fixed set of values, but the database accepts anything. Typos become new states, and nobody can list the legal values without reading application code.target_type, target_id — target_id points at different tables depending on target_type. No FK can express that, so orphans accumulate silently and every join needs a CASE. Acceptable for append-only audit data; painful anywhere the target must still exist.project_id, organization_id, note_id, group_id — todos has several nullable FKs (project_id, organization_id, note_id, group_id) that look like an exclusive arc — a row should point at exactly one of them. Nothing stops zero or both.created_at — created_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.updated_at — updated_at stores wall-clock time with no zone. It reads back differently depending on the session's TimeZone, and DST transitions produce ambiguous values.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
target_id | bigint | null | |||
target_type | varchar | NOT NULL | |||
author_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
action | integer | NOT NULL | |||
state | varchar | NOT NULL | |||
created_at | timestamp | null | |||
updated_at | timestamp | null | |||
commit_id | varchar | null | |||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
resolved_by_action | smallint | null | |||
note_id | bigint | null | notes.id 1:N · CASCADE | ||
snoozed_until | timestamptz | null | |||
organization_id | bigint | null | organizations.id 1:N · CASCADE |
Table checks: num_nonnulls(group_id, organization_id, project_id) = 1
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
avatar | text | null | CHECK char_length(avatar) <= 255 | ||
description | text | null | CHECK char_length(description) <= 1024 | ||
total_projects_count | bigint | NOT NULL | 0 | ||
non_private_projects_count | bigint | NOT NULL | 0 | ||
title | text | null | CHECK char_length(title) <= 255 | ||
slug | text | null | CHECK char_length(slug) <= 255 | ||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE | ||
cached_markdown_version | integer | NOT NULL | 0 | ||
description_html | text | null | CHECK char_length(description_html) <= 50000 |
Table checks: char_length(slug) <= 255; char_length(title) <= 255; char_length(avatar) <= 255; char_length(description) <= 1024; char_length(name) <= 255; char_length(description_html) <= 50000
namespace_id — upcoming_reconciliations.namespace_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
namespace_id | bigint | null | namespaces.id 1:1 · CASCADE | ||
next_reconciliation_date | date | NOT NULL | |||
display_alert_from | date | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
organization_id | bigint | NOT NULL | organizations.id 1:N · CASCADE |
upload_states references nothing and nothing references it. Either it is a staging/log table (fine, say so), or it is dead, or it is the seed of a second data model growing beside the first.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
verification_started_at | timestamptz | null | |||
verification_retry_at | timestamptz | null | |||
verified_at | timestamptz | null | |||
upload_id PK | bigint | NOT NULL | |||
verification_state | smallint | NOT NULL | 0 | ||
verification_retry_count | smallint | null | |||
verification_checksum | bytea | null | |||
verification_failure | text | null | CHECK char_length(verification_failure) <= 255 |
Table checks: char_length(verification_failure) <= 255
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
setting_type | text | NOT NULL | CHECK char_length(setting_type) <= 256 | ||
content | text | NOT NULL | CHECK char_length(content) <= 524288 | ||
uuid | uuid | null | CHECK uuid IS NOT NULL | ||
version | integer | null | 0 | CHECK version IS NOT NULL | |
settings_context_hash | text | null | CHECK char_length(settings_context_hash) <= 255 |
Table checks: version IS NOT NULL; uuid IS NOT NULL; char_length(content) <= 524288; char_length(settings_context_hash) <= 255; char_length(setting_type) <= 256; (setting_type = CAST('extensions' AS text) AND settings_context_hash IS NOT NULL) OR (setting_type <> CAST('extensions' AS text) AND settings_context_hash IS NULL)
finding_id — vulnerabilities.finding_id may be NULL, so the relationship to vulnerability_occurrences is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | |||
author_id | bigint | NOT NULL | |||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
title | varchar(255) | NOT NULL | |||
title_html | text | null | |||
description | text | null | |||
description_html | text | null | |||
state | smallint | NOT NULL | 1 | ||
severity | smallint | NOT NULL | |||
severity_overridden | boolean | null | FALSE | ||
resolved_by_id | bigint | null | |||
resolved_at | timestamptz | null | |||
report_type | smallint | NOT NULL | |||
cached_markdown_version | integer | null | |||
confirmed_by_id | bigint | null | |||
confirmed_at | timestamptz | null | |||
dismissed_at | timestamptz | null | |||
dismissed_by_id | bigint | null | |||
resolved_on_default_branch | boolean | NOT NULL | FALSE | ||
present_on_default_branch | boolean | NOT NULL | TRUE | ||
detected_at | timestamptz | null | now() | ||
finding_id | bigint | null | vulnerability_occurrences.id 1:N · CASCADE | CHECK finding_id IS NOT NULL | |
cvss | jsonb | null | CAST('[]' AS jsonb) | ||
auto_resolved | boolean | NOT NULL | FALSE | ||
uuid | uuid | null | |||
solution | text | null | CHECK char_length(solution) <= 7000 | ||
partition_id | bigint | null | 1 |
Table checks: finding_id IS NOT NULL; char_length(solution) <= 7000
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
user_id | bigint | NOT NULL | users.id 1:N · CASCADE | ||
counter | bigint | NOT NULL | 0 | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
credential_xid | text | NOT NULL | CHECK char_length(credential_xid) <= 1364 | ||
name | text | NOT NULL | CHECK char_length(name) <= 255 | ||
public_key | text | NOT NULL | |||
authentication_mode | smallint | NOT NULL | 2 | ||
passkey_eligible | boolean | NOT NULL | FALSE | ||
last_used_at | timestamptz | null |
Table checks: char_length(name) <= 255; char_length(credential_xid) <= 1364
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
lang | text | NOT NULL | CHECK char_length(lang) <= 255 | ||
payload | jsonb | NOT NULL |
Table checks: char_length(lang) <= 255
project_id — zentao_tracker_data.project_id may be NULL, so the relationship to projects is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.group_id — zentao_tracker_data.group_id may be NULL, so the relationship to namespaces is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.organization_id — zentao_tracker_data.organization_id may be NULL, so the relationship to organizations is optional. That is legitimate (e.g. approved_by before approval) but often a modelling shrug: a row with no owner, or two nullable FKs that are secretly an either/or.| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
integration_id | bigint | NOT NULL | integrations.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
encrypted_url | bytea | null | |||
encrypted_url_iv | bytea | null | |||
encrypted_api_url | bytea | null | |||
encrypted_api_url_iv | bytea | null | |||
encrypted_zentao_product_xid | bytea | null | |||
encrypted_zentao_product_xid_iv | bytea | null | |||
encrypted_api_token | bytea | null | |||
encrypted_api_token_iv | bytea | null | |||
project_id | bigint | null | projects.id 1:N · CASCADE | ||
group_id | bigint | null | namespaces.id 1:N · CASCADE | ||
organization_id | bigint | null | organizations.id 1:N · CASCADE |
Table checks: num_nonnulls(group_id, organization_id, project_id) = 1
| Column | Type | Null | Default | References | Notes |
|---|---|---|---|---|---|
id PK | bigint | NOT NULL | |||
project_id | bigint | NOT NULL | projects.id 1:N · CASCADE | ||
issue_id | bigint | NOT NULL | issues.id 1:N · CASCADE | ||
created_at | timestamptz | NOT NULL | |||
updated_at | timestamptz | NOT NULL | |||
issue_status | smallint | NOT NULL | 1 | ||
url | varchar(255) | null |