What signal is sent to the container when it is stopped/replaced

What process signal is sent to the running container before a new version is deployed? Our application currently listens for SIGTERM & SIGKILL in order to shutdown gracefully, but these never seem to be triggered.


Hello @david111,

This part is handled by Kubernetes and to my knowledge it’s supposed to be a SIGTERM (docs).
Can you give an example how you handle the signal and / or what make you think those are not sent to your containers?


Many thanks for the link @bchastanier - useful info!

One of our services runs jobs that need to be shutdown (somewhat) gracefully. The job state is tracked in the database and I would expect the state to be failed or retry, but it still appears as active after the pod has been replaced.

I guess it must be the app that is not handling the signal correctly. So I will continue to debug. Thanks !