antonio-mello-ai commented on PR #63583: URL: https://github.com/apache/airflow/pull/63583#issuecomment-4062945433
Rebased on latest `main` and addressed all feedback. Here's the summary: ### Addressing @eladkal's question: is this still an open bug? The referenced issues are closed, but the bug persists in recent Celery versions: - **[Discussion #9095](https://github.com/celery/celery/discussions/9095)**: Multiple users reporting the same behavior with **Celery 5.5.3** through **Nov 2025** — workers reconnect but stop consuming tasks. Maintainer @auvipy acknowledged in **May 2025**: *"help us to reproduce this on celery 5.5.x+ please"* - **[celery/celery#9631](https://github.com/celery/celery/issues/9631)** (open, Mar 2025): Celery consumers losing connection behind Redis HAProxy — same root mechanic - No Celery 5.6.x changelog mentions a fix for this reconnection behavior ### Addressing @Vamsi-klu's feedback All five points addressed in this push: | # | Feedback | Resolution | |---|----------|------------| | 1 | `os.EX_UNAVAILABLE` is Unix-only | Replaced with `raise SystemExit("descriptive message")`, consistent with the rest of this file | | 2 | No timeout on inspect calls | Added `timeout=5.0` to `inspect()` constructor | | 3 | No exception handling for broker errors | Added `try/except` around both `ping()` and `active_queues()` with `log.exception()` and clean exit | | 4 | Missing test for broker connection errors | Added two new tests: `test_health_check_fails_when_broker_unreachable` and `test_health_check_fails_when_active_queues_raises` | | 5 | `import socket` can be module-level | Moved to top-level imports | All 9 tests passing, ruff + pre-commit hooks clean. -- 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]
