Description of problem:
Customer has a cluster where their NTP access was accidentally blocked by a firewall rule. This caused the cluster to think it was having intermittent latency issues with etcd and failures with ODF resulting in the loss of access to PVCs even though there were no actual latency issues, just a clock skew.

Another odd side effect of this was that whenever this would happen, all of their operators would reinstall themselves. This last issue is concerning because it would cause additional outages and is currently unexplained.

Version-Release number of selected component (if applicable): OCP 4.10.18

How reproducible:  It happened during a clock skew everything time etcd was reporting latency issues. I don't know if it's easy to reproduce in a test cluster.  The one thing that seems to cause it is a clock skew that results in etcd producing 

Actual results: All OLM installed operators reinstall themselves

Expected results: There should be no change in operators