[at88mph]

No description provided.

[pdowler]

Were you able to provoke an OOM and see the jvm and container exit?

Did the heap dump actually work? iirc the attempt to cat the result to stdout so it ends up in logs did not work last time I wanted to see it

[at88mph]

Correct. I ran two Kubernetes deployments; one with the new switch and one without:

One with:
-Xmx32m -XX:+ExitOnOutOfMemoryError

and another with
-Xmx32m

below is a view of the two Pods. The oom-exit one actually died (Error), and the other is still running as it's not registered as an error. To get appropriate restarts, the end application (e.g. cavern) will need to have a proper livenessProbe in place.