When we deploy two kafka clusters into one namespace, it looks that CO doesn't reconcile Kafka pods when JVM options change. ZK rolling update is performed, but Kafka stays in same state for at least 1 hour (after that time it's rolled). Cannot attach any CO logs, because CO log is stuck during this time and no logs are available at all.
Steps to reproduce:
- run LogSettingST#testGcLoggingSetDisabled system test