Describe the issue:
There is a too strong statement about monitoring stack persistent storage.
> In OpenShift Container Platform 4.1 deployed on bare metal, Prometheus and Alertmanager cannot have persistent storage and are ephemeral only. For use cases other than bare metal, you must use dynamic provisioning to use persistent storage. In particular, do not use NFS for the monitoring stack.
The only important thing is "Use block storage. Network storage like NFS is not supported as it's known to lead data corruption."
The others are not needed, like:
- need dynamic provisioning
- bare metal cannot have persistent storage for monitoring stack
They are false. For example, bare metal with static provisioned iSCSI storage class PVs should be fine.
Suggestions for improvement:
Seems like the following commit fixes this doc issue. Closing.