moomindani commented on issue #64609: URL: https://github.com/apache/airflow/issues/64609#issuecomment-6740609897
Two things left, in my view: 1. A docs fix. The deferrable-mode section added by #68017 (`run_now.rst` and `submit_run.rst`) tells users to set "the dedicated operator parameters `retry_limit` and `retry_delay`", but the operator arguments are `databricks_retry_limit` and `databricks_retry_delay`. Followed literally, `retry_delay=` sets the task-level retry delay from `BaseOperator` and never reaches the hook, and `retry_limit=` is not an operator argument at all. I missed this when approving #68017. Fixed in #76744. 2. A decision on Option 1 (pass an import path, import it in the trigger), which nobody has started. I would not pursue it: the strategies in the report are configured instances (`wait_incrementing(start=30, increment=30, max=300)`), not importable functions, so a path alone cannot express them, and supporting kwargs would mean inventing a serialization format for tenacity objects. With `databricks_retry_args` unset, the hook already builds `stop_after_attempt(databricks_retry_limit)` plus `wait_exponential(min=databricks_retry_delay, max=2**databricks_retry_limit)` in the triggerer too, which covers the reported use case closely enough. So I'd suggest closing this once the docs fix merges. If someone does need custom tenacity strategies in deferrable mode, a new feature request scoped to that would be cleaner than keeping this `priority:high` bug open. --- Drafted-by: Claude Code (Opus 5.5); reviewed by @moomindani before posting -- 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]
