Sorry there is bug in 13.2.2 that breaks compatibility of purge queue disk format. Please downgrading mds to 13.2.1, then run 'ceph mds repaired cephfs_name:0'. Regards Yan, Zheng On Mon, Oct 8, 2018 at 9:20 AM Alfredo Daniel Rezinovsky <alfrenovsky@xxxxxxxxx> wrote: > > Cluster with 4 nodes > > node 1: 2 HDDs > node 2: 3 HDDs > node 3: 3 HDDs > node 4: 2 HDDs > > After a problem with upgrade from 13.2.1 to 13.2.2 (I restarted the > nodes 1 at a time, think that was the problem) > > I upgraded with ubuntu apt-get upgrade. I had 1 active mds at a time > when did the upgrade. > > All MDSs stopped working > > Status shows 1 crashed and no one in standby. > > If I restart an MDS status shows replay then crash with this log output: > > ceph version 13.2.2 (02899bfda814146b021136e9d8e80eba494e1126) mimic > (stable) > 1: (()+0x3f5480) [0x555de8a51480] > 2: (()+0x12890) [0x7f6e4cb41890] > 3: (gsignal()+0xc7) [0x7f6e4bc39e97] > 4: (abort()+0x141) [0x7f6e4bc3b801] > 5: (ceph::__ceph_assert_fail(char const*, char const*, int, char > const*)+0x250) [0x7f6e4d22a710] > 6: (()+0x26c787) [0x7f6e4d22a787] > 7: (EMetaBlob::replay(MDSRank*, LogSegment*, MDSlaveUpdate*)+0x5f4b) > [0x555de8a3c83b] > 8: (EUpdate::replay(MDSRank*)+0x39) [0x555de8a3dd79] > 9: (MDLog::_replay_thread()+0x864) [0x555de89e6e04] > 10: (MDLog::ReplayThread::entry()+0xd) [0x555de8784ebd] > 11: (()+0x76db) [0x7f6e4cb366db] > 12: (clone()+0x3f) [0x7f6e4bd1c88f] > NOTE: a copy of the executable, or `objdump -rdS <executable>` is needed > to interpret this > > journal reports OK > > Now im trying: > > cephfs-data-scan scan_extents cephfs_data > > > _______________________________________________ > ceph-users mailing list > ceph-users@xxxxxxxxxxxxxx > http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com _______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com