Celery worker stops consuming tasks after Redis connection reset and fails to recover automatically
criticalWhen Redis broker connection is reset (network interruption or Redis closing idle connections), Celery logs 'Connection to broker lost. Trying to re-establish the connection' but the worker stops consuming tasks from Redis and does not recover automatically. Tasks remain queued until worker process is manually restarted. The connection reset error occurs during heartbeat checks or message restoration attempts.
Monitor for 'Connection to broker lost' warnings followed by ConnectionResetError from Redis. If task consumption stops (celery.task.received stops incrementing while celery.queue.length grows), restart the worker process/container. Verify network stability between worker and Redis. Check Redis timeout and idle connection settings. Consider implementing automated worker health checks that restart workers when task consumption stalls despite non-zero queue length.