Question, before I get too far down this path. Does the actionrunner nodes bleed off the current task(s) before shutting down when a redeploy is requested? Also, fairly new to Kubernetes and wanna make sure we don’t create new processes that will cause workflow failures. Thanks.
By redeploying if you meant upgrade…
From what i gathered using this model in our production for little over 8 months, updates to st2actionrunner pods are
rollingupdate. So, in general in-flight actions will always be satisfied before the pod or underlying container goes down in the
organic upgrades. You can of course kill the pod which will disrupt that and fail the action.
Not sure about the
new processes part of the question. But
st2WorkflowEngine will also follow the same model and will satisfy the in-flight workflows during upgrades.
I am pretty sure you answered my question.