fix(data-imports): stop blaming credentials for a cancelled warehouse query

products/warehouse_sources/backend/models/table.py

@@ -30,7 +30,12 @@
 
 from posthog.clickhouse.client import sync_execute
 from posthog.clickhouse.query_tagging import Feature, Product, tag_queries
-from posthog.errors import CORRUPTED_PARQUET_METADATA_MESSAGE, wrap_clickhouse_query_error
+from posthog.errors import (
+    CORRUPTED_PARQUET_METADATA_MESSAGE,
+    QueryErrorCategory,
+    classify_query_error,
+    wrap_clickhouse_query_error,
+)
 from posthog.exceptions_capture import capture_exception
 from posthog.models.utils import CreatedMetaFields, DeletedMetaFields, UpdatedMetaFields, UUIDTModel, sane_repr
 from posthog.schema_enums import DatabaseSerializedFieldType
@@ -932,6 +937,13 @@ def _safe_expose_ch_error(self, err):
         if not hasattr(err, "message"):
             raise err
 
+        # A cancelled query here means our own client timed out reading, not a bad file or bucket.
+        if classify_query_error(err) == QueryErrorCategory.CANCELLED:
+            raise Exception(
+                "Reading the files from your storage bucket took too long and the query was cancelled. "
+                "This is usually temporary - try again, or narrow the URL pattern if the dataset is very large."
+            )
+
         for key, value in ExtractErrors.items():
             if key in raw_message:
                 raise Exception(value)

products/warehouse_sources/backend/tests/test_table.py

@@ -140,6 +140,11 @@ def test_transient_errors_without_message_are_reraised_untouched(self, err: Exce
             DataWarehouseTable()._safe_expose_ch_error(err)
         assert exc_info.value is err
 
+    def test_cancelled_query_gets_a_timeout_message_instead_of_storage_bucket_blame(self) -> None:
+        # code 394 QUERY_WAS_CANCELLED here means our own client timed out reading, not bad files.
+        with pytest.raises(Exception, match="took too long"):
+            DataWarehouseTable()._safe_expose_ch_error(ServerException("DB::Exception: Query was cancelled.", code=394))
+
     def test_delta_kernel_permission_error_gets_actionable_message(self) -> None:
         # Delta-format tables (the default for every warehouse_sources synced table) read via
         # ClickHouse's DeltaLake kernel, whose object_store errors use different wording than