Description of problem: Add a recommendation bullet point under "CHAPTER 14. ADD OSDS" which suggests using the mkfs.xfs defaults employed by ceph-deploy. Possibly include a note which states that using non-defaults could result in poor behavior. Additional info: This is to combat customers changing away from the defaults used to format OSDs and encountering bugs in their environments at a later date. For example, a customer encountered https://bugzilla.redhat.com/show_bug.cgi?id=1278992 when they set '-n 65536' on their OSDs, which is a bug that is entirely due to them using such a high directory block size that is non-default. While the high directory size issue is a bug in XFS, it is a complicated fix which may take a long time before making it into the RHEL code base. We can try to prevent further customers from hitting this issue by mentioning this in the Ceph documentation.
May HPE have access to this BZ? Thank You Joe Kachuck