Worker SIGTERM behavior

We’re running octopus workers in kubernetes, and are noticing that workers may terminate in the middle of a deployment if the underlying host is being emptied or similar.

My question is: What is the expected behavior for workers receiving a stop signal while performing a dpeloy? are they supposed to complete the deploy before shutting down or will they kill themselves immediately regardless of any ongoing deploy?

Hi @trond!

Thanks for reaching out, and for the great question. Sadly, the tentacle agent does not have a “drain” mode built-in, so a SIGTERM would result in the immediate termination of the pod/deployment process…

One option, just off the top of my head, could be doing a pgrep -P 1 loop in your PreStop Lifecycle hook to see if there are any child processes running off the tentacle process, and if there are results returned, sleep for x seconds - this would simulate a graceful shutdown and let the current deployment processes finish.

I hope this helps, and please let me know if you have any further questions.

This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.