- What I did
I changed the cleanup code of TestEventsOOMDisableTrue, so that if the container is already removed by the time the defer call is triggered, it does not fail the test - How I did it
I used dockerCmdWithResult (which let client code choose what to do with the potential command line error) instead of dockerCmd (which fails the test if the command line does not succeed) - How to verify it
Run the CI :)
@AkihiroSuda the container spin on doing x=$x$x$x$x, consuming more and more memory.
his test checks that the oom killer does not take action for 20s, but at some point the process in the container fails to allocate more memory and crashes.