Sending a manifest update on an erroring or crashing pod should force a deleting of the pod

[pcfreak30]

If you have a pod in an error or crashloopbackoff state, and send a manifest update, say to update env vars, it should detect the state and delete the pod. this will force the deployment/statefulset to update and reflect the new config and re-spawn new containers.

Currently if this happens the only route is to close the deployment or make a custom image that never allows the container to crash.