When execution of an event thread fails, cleanup is not executed

[mewitte]

This includes docker container and networks, as well as the temporary yamls in the pwd

[lpirl]

Maybe some kind of job queue, holding jobs that need to be done before the script exists, could help here? Of course, the main procedure would have to catch all exceptions, run the pending cleanup jobs and finally re-raise the caught exception and exit.
I made good experiences with this kind of pattern (e.g., see class Cleaner and def main in https://github.com/lpirl/admintools/blob/master/cli-application-template.py – no brain surgery).