The GitHub Actions job "Tests" on 
airflow.git/fix/celery-worker-health-check-active-queues has failed.
Run started by GitHub user antonio-mello-ai (triggered by antonio-mello-ai).

Head commit for run:
71c51d4b5a7030a4e8833e6586fd979a83f5c420 / Antonio <[email protected]>
feat(celery): add worker-health-check command to detect catatonic state after 
broker restart

After a Redis broker restart, a Celery worker can enter a catatonic state
where it reconnects at transport level (celery inspect ping returns OK) but
silently loses its queue consumer registrations (active_queues returns None).
Tasks pile up in the queue and are never picked up.

The existing health check in docker-compose and Kubernetes only uses
`celery inspect ping`, which does not detect this state.

This change adds a new `airflow celery worker-health-check` CLI command that
performs a two-stage check:
1. Ping check: verifies the worker is alive and reachable.
2. Active queues check: verifies the worker has registered queue consumers.

If either check fails, the command exits with a non-zero status so that
container orchestrators (Docker, Kubernetes) can restart the worker.

The docker-compose.yaml health check for airflow-worker is updated to use
the new command instead of the raw `celery inspect ping`.

Closes #63580

Co-Authored-By: Claude Opus 4.6 <[email protected]>

Report URL: https://github.com/apache/airflow/actions/runs/23085150038

With regards,
GitHub Actions via GitBox


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to