arsenelupin14 commented on issue #67707: URL: https://github.com/apache/airflow/issues/67707#issuecomment-4581806209
Hello, I've been tracing through the codebase and can confirm your findings exactly. In Airflow 3, the patch_task_instance endpoint calls _patch_task_instance_state, which directly delegates to dag.set_task_instance_state(...). Inside dag.py, the set_task_instance_state method has hardcoded logic to always flush the session and clear downstream tasks that are in failed/upstream_failed state. My plan is: - Add an optional clear_downstream: bool parameter to dag.set_task_instance_state (defaulting to True to preserve existing internal usages). - Expose this flag via PatchTaskInstanceBody and BulkTaskInstanceBody in the FastAPI routes. But, Should hardcode clear_downstream=False in the Task Instances view API call to strictly restore Airflow 2 parity or add an explicit toggle in the Task Instances UI so users can choose whether to resume downstreams on the fly? -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
