I am not sure this related to RBD, but in case it is, this would be an important bug to fix. Running LVM on top of RBD, XFS filesystem on top of that, consumed in RHEL 7.4. When running a large read operation and doing LVM snapshots during that operation, the block being read winds up all zeroes in pagecache. Dropping the caches syncs up the block with what's on "disk" and everything is fine. Working on steps to reproduce simply - ceph is Luminous 12.2.7, RHEL client is Jewel 10.2.10-17.el7cp -- Alex Gorbachev Storcium _______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com