Refactor: Adjust bulk delete batch size for EventAttachment

Co-authored-by: tillman.elser <[email protected]>

Author: cursoragent

src/sentry/db/deletion.py

@@ -79,7 +79,7 @@ def _continuous_query(self, query):
             cursor.execute(query)
             results = cursor.rowcount > 0
 
-    def iterator(self, chunk_size=100, batch_size=10000) -> Generator[tuple[int, ...]]:
+    def iterator(self, chunk_size=100, batch_size=1000) -> Generator[tuple[int, ...]]:
         assert self.days is not None
         assert self.dtfield is not None
 

src/sentry/runner/commands/cleanup.py

@@ -584,6 +584,7 @@ def run_bulk_deletes_in_deletes(
     models_attempted: set[str],
 ) -> None:
     from sentry.db.deletion import BulkDeleteQuery
+    from sentry.models.eventattachment import EventAttachment
 
     debug_output("Running bulk deletes in DELETES")
     for model_tp, dtfield, order_by in deletes:
@@ -603,7 +604,9 @@ def run_bulk_deletes_in_deletes(
                 order_by=order_by,
             )
 
-            for chunk in q.iterator(chunk_size=100):
+            # Use smaller batch size for EventAttachment to avoid query timeouts on massive tables
+            batch_size = 100 if model_tp == EventAttachment else 1000
+            for chunk in q.iterator(chunk_size=100, batch_size=batch_size):
                 task_queue.put((imp, chunk))
 
     # Ensure all tasks are completed before exiting

tests/sentry/db/test_deletion.py

@@ -86,3 +86,26 @@ def test_iteration_descending(self) -> None:
             results.update(chunk)
 
         assert results == expected_group_ids
+
+    def test_iteration_with_custom_batch_size(self) -> None:
+        target_project = self.project
+        expected_group_ids = {self.create_group().id for i in range(2)}
+
+        other_project = self.create_project()
+        self.create_group(other_project)
+        self.create_group(other_project)
+
+        # Test with custom batch_size parameter
+        iterator = BulkDeleteQuery(
+            model=Group,
+            project_id=target_project.id,
+            dtfield="last_seen",
+            order_by="last_seen",
+            days=0,
+        ).iterator(chunk_size=1, batch_size=100)
+
+        results: set[int] = set()
+        for chunk in iterator:
+            results.update(chunk)
+
+        assert results == expected_group_ids