This commit drops two workarounds in drive groups for the old batch
behaviour. OSDs on LVs no longer need to be created individually by
create (related to functional testing) and ceph-volume batch now filters
unavailable drives itself.
Signed-off-by: Jan Fajerski <jfajerski@suse.com>
This restores legacy batch behavior and also adds some initial test and
adjusts existing tests to changes.
Signed-off-by: Jan Fajerski <jfajerski@suse.com>
This completely refactors the batch code in order to make use of the
create/prepare code path for creating OSDs instead of having a second
code path doing this. This not only eases the maintenance burden but
also adds various features and fixes bugs. This subcommand can now
handle LVs, replace OSDs, reuse VGs and has a better notion of
idempotency.
Signed-off-by: Jan Fajerski <jfajerski@suse.com>
This new property returns the free space in any VGs present. If no VGs
are on the device we project how much space a VG will have.
Signed-off-by: Jan Fajerski <jfajerski@suse.com>
This add the disk.Size class as all size related argument types. We
often create this class form args like this anyway and it enables users
to pass not only bytes but also strings like 50G.
Signed-off-by: Jan Fajerski <jfajerski@suse.com>
This makes it easier to create valid Namespace objects/arg lists when
programmatically calling create/prepare.
Signed-off-by: Jan Fajerski <jfajerski@suse.com>
It's no need to hold the lock when copying the data, which may
take a long time.
Fixes: https://tracker.ceph.com/issues/47047
Signed-off-by: Xiubo Li <xiubli@redhat.com>
* refs/pull/37009/head:
client: remove the unecessary client_lock for objector->read/write()
client: fix the comment for RWRef
Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>
there is no need to probe the device to generate the argparse help
message.
also removing the test on the function as it's not there anymore.
Signed-off-by: Marc Gariepy <gariepy.marc@gmail.com>
Fixes: https://tracker.ceph.com/issues/47502
The prevailing consensus is that rationales of direct fixes need
not be included in the commit messages. The matter must be
made clear to the reviewers, but it's not important how this is
accomplished.
(Perusers of the git history interested in finding out this information
are free to search for it by linking the commit in question to the PR it
originated from, and possibly from there to the related tracker issues.)
Signed-off-by: Nathan Cutler <ncutler@suse.com>
src/librbd/crypto/BlockCrypto.cc:5:10: fatal error: 'alloca.h' file not found
^~~~~~~~~~
1 error generated.
Including <stdlib.h> is enough
fixes: https://tracker.ceph.com/issues/47614
Signed-off-by: Willem Jan Withagen <wjw@digiware.nl>
This resolves the potential for an assertion failure in the cache
constructors that occurs when the data IoCtx is not valid. Errors
are deferred until actual IOs are issued against the invalid data
pool.
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
to silence the warning like
follow the suggestion from pip
ERROR: After October 2020 you may experience errors when installing or updating packages. This is because pip will change the way that it resolves dependency conflicts.
We recommend you use --use-feature=2020-resolver to test your packages with the new resolver before it becomes the default.
Fixes: https://tracker.ceph.com/issues/47636
Signed-off-by: Kefu Chai <kchai@redhat.com>
The file mgr/volumes/fs/operations/pin_util.py imports distutils.util for using
strtobool and thus the python package is required.
Signed-off-by: Johannes M. Scheuermann <joh.scheuer@gmail.com>
When encoding the inode backtrace, it may take a bit longer time.
If we hold the mds_lock or other locks, it may cause other threads
to idle wait.
This will queue the encoding and storing work in the mds finisher
to get rid of the locks holded.
Fixes: https://tracker.ceph.com/issues/47148
Signed-off-by: Xiubo Li <xiubli@redhat.com>
When encoding the dentries infomation to remove or update, it may
take a bit longer time. If we hold the mds_lock or other locks, it
may cause other threads to idle wait.
This will queue the encoding and storing work in the mds finisher
to get rid of the locks holded.
Fixes: https://tracker.ceph.com/issues/47148
Signed-off-by: Xiubo Li <xiubli@redhat.com>
ceph-csi has added support for passing custom map and unmap options via
mapOptions and unmapOptions storage class parameters. However, it also
uses --read-only for implementing ROX (ReadOnlyMany) PVs. If the user
supplies "mapOptions: rw", they will get around the intended read-only
restriction (at least on the block device).
ceph-csi could be patched to use "-o ro", but it actually makes sense
for common options to win over device type-specific equivalents.
Fixes: https://tracker.ceph.com/issues/47625
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
mgr/dashboard: display devices' health information within a tabset
Reviewed-by: Patrick Seidensal <pnawracay@suse.com>
Reviewed-by: Volker Theile <vtheile@suse.com>
