:hedgehog: PostHog is the leading platform for building self-driving products. Our developer tools – AI observability, analytics, session replay, flags, experiments, error tracking, logs, and more – capture all the context agents need to diagnose problems, uncover opportunities, and ship fixes. Steer it all from Slack, web, desktop, or the MCP.
Python ab-testingai-analyticsanalyticscdpdata-warehouseexperimentsfeature-flagsjavascriptproduct-analyticspythonreactsession-replaysurveystypescriptweb-analytics
fix(data-warehouse): skip enrichment for a deleted schema (#121347)
Tom Owersf0d604b67adc
diff --git a/products/warehouse_sources/backend/temporal/data_imports/workflow_activities/enrich_table_semantics.py b/products/warehouse_sources/backend/temporal/data_imports/workflow_activities/enrich_table_semantics.pyindex 5709b1a202ec..b67b731a366c 100644--- a/products/warehouse_sources/backend/temporal/data_imports/workflow_activities/enrich_table_semantics.py+++ b/products/warehouse_sources/backend/temporal/data_imports/workflow_activities/enrich_table_semantics.py@@ -297,13 +297,21 @@ def enrich_table_semantics_sync(team_id: int, schema_id: uuid.UUID) -> dict[str,# Retried individually, not via `with_internal_db_retries` on the whole function, because the# LLM call further down is expensive and must not be repeated for a blip reaching our own database.- team = retry_internal_db_operation(- lambda: (- Team.objects.select_related("organization")- .only("id", "uuid", "organization_id", "organization__is_ai_data_processing_approved")- .get(id=team_id)+ # The team can also be legitimately gone by the time this fire-and-forget child workflow runs+ # (deleted between the post-import gate check and now) — not a bug, so skip rather than let+ # DoesNotExist reach the activity's except block and error tracking. No analytics event here:+ # emit_completed needs the team this lookup just failed to find.+ try:+ team = retry_internal_db_operation(+ lambda: (+ Team.objects.select_related("organization")+ .only("id", "uuid", "organization_id", "organization__is_ai_data_processing_approved")+ .get(id=team_id)+ ))- )+ except Team.DoesNotExist:+ log.info("warehouse_enrichment.skipped", reason="team_deleted")+ return {"status": "skipped", "reason": "team_deleted"}# Context accumulated as we learn it; shared by the "started" and "completed" analytics events.event_props: dict[str, Any] = {"schema_id": str(schema_id)}@@ -318,13 +326,22 @@ def enrich_table_semantics_sync(team_id: int, schema_id: uuid.UUID) -> dict[str,emit_completed("skipped", reason="ai_data_processing_not_approved")return {"status": "skipped", "reason": "ai_data_processing_not_approved"}- schema = retry_internal_db_operation(- lambda: (- ExternalDataSchema.objects.select_related("source", "table")- .filter(team_id=team_id, deleted=False)- .get(id=schema_id)+ # The schema can be legitimately gone by the time this fire-and-forget child workflow runs+ # (deleted between the post-import gate check and now) — not a bug, so skip like the other+ # not-found cases below rather than let DoesNotExist reach the activity's except block and+ # error tracking.+ try:+ schema = retry_internal_db_operation(+ lambda: (+ ExternalDataSchema.objects.select_related("source", "table")+ .filter(team_id=team_id, deleted=False)+ .get(id=schema_id)+ ))- )+ except ExternalDataSchema.DoesNotExist:+ log.info("warehouse_enrichment.skipped", reason="schema_deleted")+ emit_completed("skipped", reason="schema_deleted")+ return {"status": "skipped", "reason": "schema_deleted"}table = schema.tableevent_props["source_type"] = schema.source.source_typeevent_props["schema_name"] = schema.namediff --git a/products/warehouse_sources/backend/temporal/data_imports/workflow_activities/tests/test_enrich_table_semantics.py b/products/warehouse_sources/backend/temporal/data_imports/workflow_activities/tests/test_enrich_table_semantics.pyindex 65eb238c3f94..bd6c31de08b4 100644--- a/products/warehouse_sources/backend/temporal/data_imports/workflow_activities/tests/test_enrich_table_semantics.py+++ b/products/warehouse_sources/backend/temporal/data_imports/workflow_activities/tests/test_enrich_table_semantics.py@@ -528,6 +528,28 @@ class TestEnrichTableSemanticsSync:assert result["reason"] == "ai_data_processing_not_approved"assert _annotations(team, table) == {}+ def test_skipped_when_team_was_deleted(self):+ # Same fire-and-forget race as the schema case below, but for the team itself.+ team = _team()+ team_id = team.pk+ team.delete()+ with patch.object(enrich, "_generate_descriptions") as mock_llm:+ result = enrich_table_semantics_sync(team_id, uuid.uuid4())+ mock_llm.assert_not_called()+ assert result == {"status": "skipped", "reason": "team_deleted"}++ def test_skipped_when_schema_was_deleted(self):+ # This activity is a fire-and-forget child workflow started after a sync completes, so the+ # schema can be deleted by the user before it actually runs. That's a benign race, not a bug.+ team = _team()+ schema, table = _make_schema(team, columns=[{"name": "amount", "data_type": "Int64", "is_nullable": False}])+ schema.soft_delete()+ with patch.object(enrich, "_generate_descriptions") as mock_llm:+ result = enrich_table_semantics_sync(team.pk, schema.id)+ mock_llm.assert_not_called()+ assert result == {"status": "skipped", "reason": "schema_deleted"}+ assert _annotations(team, table) == {}+def test_skipped_when_table_has_no_columns(self):team = _team()schema, table = _make_schema(team, columns=[])