This command has not worked since hammer and makes no sense in an
environment where we do not allow empty caps
Fixes: http://tracker.ceph.com/issues/37663
Signed-off-by: Brad Hubbard <bhubbard@redhat.com>
rgw: fix stats for versioned buckets after reshard.
Reviewed-by: Matt Benjamin <mbenjami@redhat.com>
Reviewed-By: Abhishek Lekshmanan <abhishek@suse.com>
Otherwise, ``ceph-volume lvm activate --all`` will skip OSDs that are
already running. So the test that verifies that permissions are
corrected will fail because ``activate`` will simply skip them
Signed-off-by: Alfredo Deza <adeza@redhat.com>
mgr/dashboard: Validation for duplicate RGW user email
Reviewed-by: Alfonso Martínez <almartin@redhat.com>
Reviewed-by: Laura Paduano <lpaduano@suse.com>
librados2 and librbd1 are installed as a dependency of qemu-kvm.
qemu-kvm is installed by ceph-cm-ansible, see [1].
in thrash-old-clients, jewel packages are installed, but yum does
not allow downgrade unless it's required explicitly. in this change,
we downgrade librbd1 and librados2 to address this issue.
currently, the ceph packages shipped by CentOS/RHEL 7 are still an old
version of jewel. so this issue only kicks in when we try to install
hammer.
this change should address failures like
Command failed on smithi136 with status 1: '\n sudo yum -y install
rbd-fuse\n '
found in rados/thrash-old-clients tests.
---
[1]
3db1cbdc22 (diff-f2b05d775fedff6c5c6689f564b32f1c)
Fixes: http://tracker.ceph.com/issues/37618
Signed-off-by: Kefu Chai <kchai@redhat.com>
* refs/pull/25222/head:
mds: do not call Journaler::_trim twice
Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>
Reviewed-by: Zheng Yan <zyan@redhat.com>
* refs/pull/25229/head:
mds: fix incorrect l_pq_executing_ops statistics when meet an invalid item in purge queue
Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>
Reviewed-by: Zheng Yan <zyan@redhat.com>
* refs/pull/25109/head:
client: do not move f->pos untill success write
Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>
Reviewed-by: Zheng Yan <zyan@redhat.com>
* refs/pull/25220/head:
mds: queue inodes into MDCache::export_pin_queue after setattr is journaled
Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>
The ceph-ansible stable-3.2 branch does not have the ceph-facts role
but master does. We need to make sure to import this role only when
it's supported.
Signed-off-by: Andrew Schoen <aschoen@redhat.com>
This change adds a single log entry at level 1 to indicate when a
bucket resharding successfully completes.
Signed-off-by: J. Eric Ivancich <ivancich@redhat.com>
It's now possible to send an array of notification configs to the notification
service. These array is cached for, by default 500ms, before this array
is unified and than converted to full notifications that are seen by the
user.
Useful if different service can come to the same conclusions regarding
the notification. This is the case for the prometheus notification service
and the prometheus alert service.
Fixes: https://tracker.ceph.com/issues/37469
Signed-off-by: Stephan Müller <smueller@suse.com>
This model will now be taken as single argument in the 'show' method
of the notification service. Through using this model it's easy to
define notifications before sending them out. Which can be helpful
if you have to process an array of possible notification and the
possibility that those notifications would look a like.
This scenario doesn't exist in the current code base, but it's the
case for new feature to show alerts from Prometheus.
Fixes: https://tracker.ceph.com/issues/37469
Signed-off-by: Stephan Müller <smueller@suse.com>
