You could add (debug) logs for starters ;-) There was a thread [1] describing something quite similar, pointing to this bug report [2]. In recent versions it's supposed to be fixed although I don't see the tracker or PR number in the release notes of both pacific and quincy. Can you verify if you're hitting the same bug? Thanks, Eugen [1] https://lists.ceph.io/hyperkitty/list/ceph-users@ceph.io/thread/KPVJSA5WHO7U... [2] https://tracker.ceph.com/issues/49132 Zitat von Nathan Harper <nathharper@gmail.com>:

Hi,

We're having sporadic problems with a CephFS filesystem where MDSs end up on the OSD blocklist. We're still digging around looking for a cause (Ceph related or other infrastructure cause).

The cluster isn't massive (68 OSDs spread over 34 hosts), each host is a VM, with MGR/MON/MDS on non-OSD hosts.

Running Ceph 16.2.10

Any suggestions for debugging this further? _______________________________________________ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-leave@ceph.io