Currently there is no way to see which daemons were involved in a slow
op after the op has cleared. This change allows us to record which
daemons were implicated in the logs.
Partially fixes: http://tracker.ceph.com/issues/23205
Signed-off-by: Brad Hubbard <bhubbard@redhat.com>
(cherry picked from commit b5263176de)
CZ mirror is no longer running because company sponsoring this mirror
isn't able to provide it anymore.
I'm removing this mirror from list of official mirrors.
Signed-off-by: Tomáš Kukrál <tom@6shore.net>
Change pg state names according to osd/osd_types.cc as this is what
ceph -s and the prometheus exporter will present to users.
Signed-off-by: Jan Fajerski <jfajerski@suse.com>
* refs/pull/21805/head:
qa: move snap-hierarchy out of snaps workunits
qa: try snap format upgrade with multimds cluster
qa: add mds deactivation procedure for upgrades
Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>
The snapshot hierarchy it leaves behind can't be cleaned up by `rm -rf` which
breaks workunit cleanup. So, don't run this as part of normal snaps test.
Signed-off-by: Patrick Donnelly <pdonnell@redhat.com>
When run on PRs that have non-ASCII characters in their titles, the script
fails like this when run in a non UTF-8 environment:
UnicodeEncodeError: 'latin-1' codec can't encode character u'\u2026' in
position 651: ordinal not in range(256)
Since failing like that is not useful, avoid the error by forcing ASCII
encoding and filtering out any non-ASCII characters.
Fixes: http://tracker.ceph.com/issues/24003
Signed-off-by: Nathan Cutler <ncutler@suse.com>
* refs/pull/21374/head:
qa: add test for snap format upgrade
mds: initialize SnapServer::snaprealm_v2_since after journal replay
mds: properly distinguish cap update from snap flush
mds: update dev document of cephfs snapshot
doc: add release notes for cephfs snapshot
mds: allow snapshot by default for new filesystem
mds: close past parents after snaprealm format gets converted
mds: automaticly allow multi-active MDS after scrubbing all inodes
mds: don't mark primary dentry damaged if inode has been repaired
mds: upgrade snaprealm format during scrub
mds: allow scrubbing mdsdir
mds: cleanup scrub code
mds: show health warning if multimds with old format snapshots
mds: automaticly allow multi-active MDS after removing all old snapshots
mds: disallow multi-active MDS if snapshot was ever created by pre-mimic mds
mds: validate SnapInfo::long_name before using it
mds: don't bump snaptable last_snap when renaming snapshot
mds: properly save snaptable after upgrading version
Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>
Service names are not always osd.X or mon.X, they might be
rgw.radosgw.rgw1
This would lead to:
Unhandled exception from module 'influx' while running on mgr.mon01: too many values to unpack
Only split on the first dot as the rest is the service name
Fixes: http://tracker.ceph.com/issues/23996
Signed-off-by: Wido den Hollander <wido@42on.com>
If a PG is incomplete when the pool is deleted we'll dereference invalid
iterators here.
Fixes: http://tracker.ceph.com/issues/23980
Signed-off-by: Sage Weil <sage@redhat.com>
Currently there is no way to see which daemons were involved in a slow
op after the op has cleared. This change allows us to record which
daemons were implicated in the logs.
Partially fixes: http://tracker.ceph.com/issues/23205
Signed-off-by: Brad Hubbard <bhubbard@redhat.com>
If we don't note that we don't reply then we don't close out the routed
mon request and the op will appear as slow on the forwarding mon.
Fixes: http://tracker.ceph.com/issues/23769
Signed-off-by: Sage Weil <sage@redhat.com>
* refs/pull/21065/head:
qa/cephfs: test if evicted client unmounts without hanging
qa/tasks: allow custom timeout for umount_wait()
client: don't hang when MDS sessions are evicted
Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>
