To validate sample-dedup actually works, validate() runs
separated thread from sample-dedup and verifies
two following things.
1. check sample-dedup starts properly.
2. check references of all the chunk objects' in chunk tier
exists in designated base pool.
This routune repeats for max_valication_cnt times while
sample-dedup is running. If it doesn't raise any fail while the loop,
we can pretend sample-dedup works accurately.
If not, assert() will stop this test.
In case that a reference of chunk object doesn't exist in base pool,
validate() gives a second chance after repairing it (chunk-repair op)
to deal with false-positive reference inconsistency.
Signed-off-by: Sungmin Lee <sung_min.lee@samsung.com>
Internal directories: '_nogroup', '_index', '_legacy', '_deleting'
1. Internal directories should be filtered in 'subvolmegroup ls' command.
2. Internal directories should not be accepted as a group name.
Fixes: https://tracker.ceph.com/issues/55762
Signed-off-by: Nikhilkumar Shelke <nshelke@redhat.com>
docs: minor doc fixes of showing in progress clones for a snapshot
Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>
Reviewed-by: Ramana Raja <rraja@redhat.com>
Reviewed-by: Rishabh Dave <ridave@redhat.com>
Reviewed-by: Neeraj Pratap Singh <neesingh@redhat.com>
Reviewed-by: Kotresh HR <khiremat@redhat.com>
Reviewed-by: Anthony D'Atri <anthony.datri@gmail.com>
Reviewed-by: Dhairya Parmar <dparmar@redhat.com
Existing object-dedup command always creates temporary object and
perform set-chunk to make the input object manifest---this is because
current tier-flush, which is called by object-dedup, works only if
the target object is manifest, otherwise it just return 0.
So, object-dedup may not work if the target is already manifest.
Moreover, it causes unnecessary overhead if the target object is not manifest.
To solve these, this commit makes existing tier-flush to
change object's state to manifest implicitly.
With this change, the only required operation when calling object-dedup
is a tier-flush and object-dedup can work with the manifest object.
Signed-off-by: Myoungwon Oh <myoungwon.oh@samsung.com>
Missed this one, it seems the test was still passing until
recently but I guess the "master" builds are now gone
Signed-off-by: Adam King <adking@redhat.com>
Validate subvolume discover on upgrade from
legacy subvolume to v1. The handcrafted
`.meta' file on legacy subvolume root should
not be used for any subvolume apis like getpath,
authorize.
Signed-off-by: Kotresh HR <khiremat@redhat.com>
rbd: don't default empty pool name unless namespace is specified
Reviewed-by: Christopher Hoffman <choffman@redhat.com>
Reviewed-by: Mykola Golub <mgolub@suse.com>
1. When `test_pool_min_size` hit the case where
`not all PGs are active or peered` we dump
each PG state that doesn't have active or
peered state
2. Improve logs message in `inject_pause()`.
3. Add logs for the `test_map_discontinuity()`.
4. In the function, `choose_action()`,
added more logs regarding `chance_down`.
5. Added more loggings to
`primary_affinity()`,
`thrash_pg_upmap_items()`,
`thrash_pg_upmap()`.
6. Make self.is_clean() dump the pgs that
are not active+clean.
Signed-off-by: Kamoltat <ksirivad@redhat.com>
The I/O workload in this test is xfstests (qa/run_xfstests_qemu.sh)
which isn't subjected to any timeout other than global max_job_time
limit in any other subsuite (e.g. qemu/workloads/qemu_xfstests.yaml).
But here, there is a parallel "op" workload defined as a workunit.
The workunit task has a default timeout of 3 hours which is effectively
imposed on the entire job. In the "rbd cache = false" configuration,
it's sometimes exceeded.
Fixes: https://tracker.ceph.com/issues/48038
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Commit 96f05a7956 ("rbd: delay determination of default pool name")
broke "rbd perf image iostat" and "rbd perf image iotop" GLOBAL_POOL_KEY
support (the ability to blend all rbd pools together into a single
view).
Fixes: https://tracker.ceph.com/issues/56561
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
It doesn't really thrash anything, just repeatedly restarts the
workload on top of a dirty cache file. rbd_pwl_cache_recovery is
more on point and gets covered by existing CODEOWNERS.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
add thrash test for persistent write log cache. run rbd bench
on persistent write log cache, thrashes rbd bench, test the
recovery function of persistent write log cache.
Signed-off-by: Yin Congmin <congmin.yin@intel.com>
