Description of problem:
In RHCS 3.0, all stats are reported to the ceph-mgr daemon. However, in order for this to occur, the same ports that are open for the Ceph OSDs (6800-7100) must also be open. If this does not happen, problems arise.
The cluster shortly devolves from HEALTH_OK to HEALTH_WARN, with the PGs stuck and stale. ceph -s or ceph -w will show that 100% of placement groups are "unknown."
User will expect to see HEALTH_OK, but will see HEALTH_WARN.
We probably need to cover what happens if the ceph-mgr daemon doesn't get installed, if it is not running, and if it is running but the correct ports aren't open.
Please share the actual doc link
(In reply to Harish NV Rao from comment #7)
> Please share the actual doc link
A gentle reminder.