feat(worker): use lightweight liveness & add Celery readiness probe

[honghainguyen777]

This update prevents worker pods from restart loops caused by Celery ping timeouts.

• Liveness probe: replaced the Celery ping with a lightweight command (rm -f /tmp/health.txt) to only check container health.

• Readiness probe: added an optional Celery app.control.ping() check with configurable timeout (celeryTimeoutSeconds).

The liveness probe now ensures the container is alive, while readiness reflects broker connectivity - avoiding restarts when Celery or the broker is slow (increase celeryTimeoutSeconds to allow the process to complete)

[patsevanton]

Could you update you fork?

[honghainguyen777]

Could you update you fork?

Hi @patsevanton done! :)

[patsevanton]

@honghainguyen777 how can I reproduce an error or a situation in a laboratory setting?

[honghainguyen777]

@honghainguyen777 how can I reproduce an error or a situation in a laboratory setting?

Great question! Are you using RabbitMQ as the broker? You can try to make your broker to respond slower than the timeout, for example, you try to slow down your pod network so the communication between the Worker pod and RabbitMQ server gets timeout.