1- a decompiled and recompiled was parsing the class bucket ids but it
wasn't actually using them.
2- rebuild_roots_with_classes() was throwing out the old ids and assigning
new ids when the tree was rebuilt.
Fix by passing in a (potentially partial) class_bucket map into
populate_classes(). Take care to allocate new bucket ids that don't
collide with previously used ids.
Signed-off-by: Sage Weil <sage@redhat.com>
* refs/remotes/upstream/pull/16766/head:
cephfs: Clarify health warning
cephfs: Always print filesystem name
Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>
Previously, the FSMap summary omitted the filesystem
name in clusters with only one filesystem. Print the
file system name no matter what.
Signed-off-by: Douglas Fuller <dfuller@redhat.com>
we cannot overwrite existing dev class, and "osd_class_update_on_start"
is true by default (see 0c885d6). so we should remove all device classes before
setting them.
Signed-off-by: Kefu Chai <kchai@redhat.com>
116cf759c8
will now hide all shadow trees(roots), so this is not applicable anymore
(actually it is misleading).
Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
We might use 'crush link' to link same host into
different roots, which as a result can cause different
shadow trees reference same hosts too.
This means we may need to destory the same buckets(hosts, racks, etc.)
multiple times during rebuilding all shadow trees and hence 'remove_root'
shall be idempotent.
Fixes: http://tracker.ceph.com/issues/20845
Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
* refs/remotes/upstream/pull/16631/head:
cephfs: Resend discover when joining from starting state
cephfs: Wait for MDCache to be ready before starting log segment
Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>
* refs/remotes/upstream/pull/16363/head:
mds: fix health checks for multiple filesystems
mon: fix log message on mds rank stop
mon: uniform way of referring to MDS daemons
mds: revise health checks for _FAILED/_DEGRADED
mon/MDSMonitor: log messages on MDS rank changes
Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>
rgw: match wildcards in StringLike policy conditions
Reviewed-by: Pritha Srivastava <prsrivas@redhat.com>
Reviewed-by: Adam C. Emerson <aemerson@redhat.com>
If ceph_fsid is not found, which is the case for legacy dmcrypted OSD,
the cluster must default to ceph, as it was before support for non
standard cluster names was introduced.
Fixes: http://tracker.ceph.com/issues/20893
Signed-off-by: Loic Dachary <loic@dachary.org>
We test EC profiles with m=1 here, and mapgap can lead to incomplete pgs
because it takes an osd down and waits for healthy.
Fixes: http://tracker.ceph.com/issues/20844
Signed-off-by: Sage Weil <sage@redhat.com>
remote/smithi025/log/ceph.log.gz:2017-08-03 07:02:15.049074 mon.b mon.0 172.21.15.25:6789/0 197 : cluster [INF] Manager daemon x is unresponsive, replacing it with standby daemon y
remote/smithi025/log/ceph.log.gz:2017-08-03 07:03:10.078032 mon.b mon.0 172.21.15.25:6789/0 226 : cluster [WRN] Manager daemon x is unresponsive. No standby daemons available.
x and y may be swapped, so whitelist the rest of the string.
Signed-off-by: Sage Weil <sage@redhat.com>
ceph-osd: fix auto detect which objectstore is currently running
Reviewed-by: Kefu Chai <kchai@redhat.com>
Reviewed-by: Chunmei Liu <chunmei.liu@intel.com>
