This is an issue from upstream: http://tracker.ceph.com/issues/17466 https://www.mail-archive.com/ceph-users@lists.ceph.com/msg32925.html 10.2.3 (or RHCS 2.0) mons will crash (and continue to crash if restarted) if an older MDS was started with the "mds_standby_for_rank" setting set (due to a decode bug when handling older messages), or if a newer MDS was started with an mds_standby_for_fscid setting that did not correspond to an existing filesystem (due to a failure to validate this field).
Moving this bug to verified state. steps followed for verification 1. Configure cluster with 3 mon and 2 mds in which one MDS with "mds_standby_for_rank" flag(same rank) for 1 file system 2. Add one more MDS daemon with "mds_standby_for_fscid" which should not corresponds to particular file system. 3. stop active MDS service 4. MON should not crash.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://rhn.redhat.com/errata/RHSA-2016-2815.html