We noticed that our failed prefect jobs that use kubernetes still keep running, is there a configuration in the prefect deployment that we have missed?

Have you got a finished_job_ttl in play?

No didn’t, thanks for the hint!