qa/standalone/osd/divergent-prior.sh: Divergent test 3 with pg_autoscale_mode on pick divergent osd
Reviewed-by: Kamoltat (Junior) Sirivadhna <ksirivad@redhat.com>
qa/tasks: set defer_client_eviction_on_laggy_osds=false in api tests
Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>
Reviewed-by: Dhairya Parmar <dparmar@redhat.com>
Reviewed-by: Nizamudeen A <nia@redhat.com>
We expect laggy OSDs in this testing environment,
so it makes sense to disable this warning.
Fixes: https://tracker.ceph.com/issues/61907
Signed-off-by: Laura Flores <lflores@redhat.com>
In rocksdb 7.0, all envlibrados files were moved to a separate repository (ref: https://github.com/facebook/rocksdb/pull/9206).
The new repo is temporary and serves as an example before it is finalized where and who to host RADOS support.
Since this new repo is outside of the rocksdb repo and in an unceratin state, we should remove support for it in main
and Reef test suites. Quincy and below still use rocksdb 6.0, so the same does not apply.
Fixes: https://tracker.ceph.com/issues/59057
Signed-off-by: Laura Flores <lflores@redhat.com>
The rgw spec migration code, intended to formalize
the rgw_frontend_type spec option, doesn't work with
simple specs i.e.
service_type: rgw
service_id: rgw.1
service_name: rgw.rgw.1
placement:
label: rgw
because the migration code assumes there will always be
a "spec" section inside the spec. This is the case for
more involved rgw specs such as
service_type: rgw
service_id: foo
placement:
label: rgw
count_per_host: 2
spec:
rgw_realm: myrealm
rgw_zone: myzone
rgw_frontend_type: "beast"
rgw_frontend_port: 5000
which is what the migration is actually concerned about
(verification of the rgw_frontend_type in these specs).
In the case where the spec is more simple, we should
just leave the spec alone and move on. Unfortunately
the current code assumes the field will always be
there and hits an unhandled KeyError when trying to
migrate the more simple specs. This causes the
cephadm module to crash shortly after starting an
upgrade to a version that includes this migration
and it's very difficult to find the root cause. This
can be worked around by adding fields to the rgw
spec before upgrade so the "spec" field exists in
the spec and the migration works as intended.
This commit fixes the migration in the simple
case as well as adding testing for that case to
both the unit tests and orch/cephadm teuthology
upgrade tests
Fixes: https://tracker.ceph.com/issues/61889
Signed-off-by: Adam King <adking@redhat.com>
* refs/pull/49971/head:
doc/cephfs: document MDS_CLIENTS_LAGGY health warning
qa: ignore warnings
qa: add test cases to check client eviction if an OSD is laggy
mds,messages: enable beacon to report clients lagginess
mds: do not evict client on laggy osds
common: add new config option to defer client eviction
osd: add method to check for laggy osds
Reviewed-by: Venky Shankar <vshankar@redhat.com>
qa/tasks/s3tests.py was adding `not test_of_sts` and `not
webidentity_test` by default, and the rgw/sts suites were adding
those attrs back in `extra_attrs`
when `extra_attrs` was changed to be additive, this started causing
InvocationError failures
instead of using `extra_attrs` to control these filters,
qa/tasks/s3tests.py now uses the `sts_tests` and `webidentity_tests`
flags from the sts yaml files to control whether or not the `not
test_of_sts`/`not webidentity_test` attrs are added to the pytest
command line
Fixes: https://tracker.ceph.com/issues/61838
Signed-off-by: Casey Bodley <cbodley@redhat.com>
Otherwise, the MDS that just got replaced can transition to a rank
for another file system and the test cannot deterministically infer
which MDS needs to checked.
Fixes: http://tracker.ceph.com/issues/61764
Signed-off-by: Venky Shankar <vshankar@redhat.com>
the s3tests.py task is filtering out several attrs by default. but
when dbstore uses `extra_attrs` to add 'not fails_on_dbstore', it
overwrites those other filters
Signed-off-by: Casey Bodley <cbodley@redhat.com>
pybind/rados: keep byte representation if decode fails
Reviewed-by: Samuel Just <sjust@redhat.com>
Reviewed-by: John Mulligan <jmulligan@redhat.com>
Reviewed-by: Dhairya Parmar <dparmar@redhat.com>
cephadm _orch deploy will eventually be the only deploy method so
we need to update the test script.
Signed-off-by: John Mulligan <jmulligan@redhat.com>
The current version is pretty useless:
- "rbd bench" writes the same byte (0xff) over and over again, so
almost all checksumming is in vain
- snapshots are taken in a steady state (i.e. not under I/O), so no
race conditions can get exposed
- even with these caveats, it's not wired up into the suite
Redo this workunit to be a reliable reproducer for the issue fixed
in the previous commit and wire it up for both krbd and rbd-nbd.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
qa: wait for 100 seconds to make sure the quota to be enforced
Reviewed-by: Venky Shankar <vshankar@redhat.com>
Reviewed-by: Rishabh Dave <ridave@redhat.com>
Centos Stream 8 has removed the 'device-mapper-devel', 'libedit-devel'
and 'userspace-rcu-devel' packages from the mirrors and we need to
install it from powertools repo.
Fixes: https://tracker.ceph.com/issues/59683
Signed-off-by: Xiubo Li <xiubli@redhat.com>
When creating new pool, the current code pick the divergent osd by
the first pg out of pg dump pgs, that can be in "unknown" status
which means the up_primary = -1 and that will fail the test.
We need to wait unitl the first pg is active+clean
Fixes: https://tracker.ceph.com/issues/56034
Signed-off-by: Nitzan Mordechai <nmordech@redhat.com>
