Ilya Dryomov [Mon, 14 Aug 2023 11:16:59 +0000 (13:16 +0200)]
qa/suites/upgrade/octopus-x: skip TestClsRbd.mirror_snapshot test
The behavior of the class method changed in reef; the change was
backported to pacific and quincy. An octopus test binary used against
pacific OSDs produces an expected failure:
[ RUN ] TestClsRbd.mirror_snapshot
.../ceph-15.2.17/src/test/cls_rbd/test_cls_rbd.cc:2279: Failure
Expected equality of these values:
-85
mirror_image_snapshot_unlink_peer(&ioctx, oid, 1, "peer2")
Which is: 0
[ FAILED ] TestClsRbd.mirror_snapshot (6 ms)
Conflicts:
src/pybind/mgr/rbd_support/mirror_snapshot_schedule.py
- Above conflict was due to commit e4a16e2
("mgr/rbd_support: add type annotation") not in pacific
Conflicts:
src/pybind/mgr/rbd_support/module.py
- Above conflict was due to commit dcb51b0
("mgr/rbd_support: define commands using CLICommand") not in pacific
Ramana Raja [Wed, 10 May 2023 18:37:44 +0000 (14:37 -0400)]
rbd_support: recover from "double blocklisting"
Recover from being blocklisted while recovering from blocklisting.
When the rbd_support module is being set up to recover from client
blocklisting, the module's new rados client connection can also get
blocklisted. Currently, this will cause the recovery to fail and
the module will remain inoperable. Instead, retry module recovery
when the new client gets blocklisted during the module setup in the
recovery thread.
Conflicts:
src/pybind/mgr/rbd_support/mirror_snapshot_schedule.py
src/pybind/mgr/rbd_support/module.py
src/pybind/mgr/rbd_support/perf.py
src/pybind/mgr/rbd_support/task.py
src/pybind/mgr/rbd_support/trash_purge_schedule.py
- Above conflicts were due to commit e4a16e2
("mgr/rbd_support: add type annotation") not in pacific
- Above conflicts were due to commit dcb51b0
("mgr/rbd_support: define commands using CLICommand") not in pacific
Ramana Raja [Wed, 15 Feb 2023 15:12:54 +0000 (10:12 -0500)]
mgr/rbd_support: recover from rados client blocklisting
In certain scenarios the OSDs were slow to process RBD requests.
This lead to the rbd_support module's RBD client not being able to
gracefully handover a RBD exclusive lock to another RBD client.
After the condition persisted for some time, the other RBD client
forcefully acquired the lock by blocklisting the rbd_support module's
RBD client, and consequently blocklisted the module's RADOS client. The
rbd_support module stopped working. To recover the module, the entire
mgr service had to be restarted which reloaded other mgr modules.
Instead of recovering the rbd_support module from client blocklisting
by being disruptive to other mgr modules, recover the module
automatically without restarting the mgr serivce. On client getting
blocklisted, shutdown the module's handlers and blocklisted client,
create a new rados client for the module, and start the new handlers.
Conflicts:
src/pybind/mgr/rbd_support/mirror_snapshot_schedule.py
src/pybind/mgr/rbd_support/module.py
src/pybind/mgr/rbd_support/perf.py
src/pybind/mgr/rbd_support/task.py
src/pybind/mgr/rbd_support/trash_purge_schedule.py
- Above conflicts were due to commit e4a16e2
("mgr/rbd_support: add type annotation") not in pacific
- Above conflicts were due to commit dcb51b0
("mgr/rbd_support: define commands using CLICommand") not in pacific
Ramana Raja [Mon, 30 Jan 2023 07:21:54 +0000 (02:21 -0500)]
mgr: store names of modules that register RADOS clients in the MgrMap
The MgrMap stores a list of RADOS clients' addresses registered by the
mgr modules. During failover of ceph-mgr, the list is used to blocklist
clients belonging to the failed ceph-mgr.
Store the names of the mgr modules that registered the RADOS clients
along with the clients' addresses in the MgrMap. During debugging, this
allows easy identification of the mgr module that registered a
particular RADOS client by just dumping the MgrMap (`ceph mgr dump`).
Following is the MgrMap output with a module's client name displayed
along with its client addrvec,
$ ceph mgr dump | jq '.active_clients[0]'
{
"name": "devicehealth",
"addrvec": [
{
"type": "v2",
"addr": "10.0.0.148:0",
"nonce": 612376578
}
]
}
Igor Fedotov [Fri, 11 Nov 2022 14:31:19 +0000 (17:31 +0300)]
os/bluestore: introduce a cooldown period for failed BlueFS allocations.
When using bluefs_shared_alloc_size one might get a long-lasting state when
that large chunks are not available any more and fallback to shared
device min alloc size occurs. The introduced cooldown is intended to
prevent repetitive allocation attempts with bluefs_shared_alloc_size for
a while. The rationale is to eliminate performance penalty these failing
attempts might cause.
Adam Emerson [Thu, 20 Jul 2023 01:03:44 +0000 (21:03 -0400)]
build: install-deps.sh installs system boost on Jammy
Since on Jammy system boost is new enough for Pacific and we don't have
Jammy packages for older boost (we only have those for Bionic), just
install the system packages rather than fetching ceph-libboost.
No analogous commit exists in main as while main's Jammy case installs
ceph-libboost, we just need a system package here.
Fixes: https://tracker.ceph.com/issues/62103 Signed-off-by: Adam Emerson <aemerson@redhat.com>
Adam Emerson [Wed, 19 Jul 2023 21:12:08 +0000 (17:12 -0400)]
build: Remove old ceph-libboost* packages in install-deps
Here, we extract `clean_boost_on_ubuntu()` and call it before other
installs on Debian distributions so that if we install a system boost,
a potentially newer `ceph-libboost` won't get in the way.
As the sources.list.d being removed in the original cleanup code isn't
the one we're currently installing in the install code, add a removal
for the currently used source, then do apt-update so packages from the
removed source are no longer included as available.
Two subsidiary dev packages from conflicting boost libraries can be
installed, but it leaves apt in an inconsistent state. To clean this
up, add `--fix-missing` to the removal line and call
`clean_boost_on_ubuntu()` before other uses of apt.
Fixes: https://tracker.ceph.com/issues/62097 Signed-off-by: Adam Emerson <aemerson@redhat.com>
(cherry picked from commit 0c3f511e14af639b6509e69b889258b2f718f8fd)
Conflicts:
install-deps.sh
- Different boost version for Pacific than Squid.
- ci_debug does not exist in Pacific
- whitespace
- No INSTALL_EXTRA
Fixes: https://tracker.ceph.com/issues/62103 Signed-off-by: Adam Emerson <aemerson@redhat.com>
Kotresh HR [Tue, 6 Jun 2023 07:39:00 +0000 (13:09 +0530)]
mds: Fix the linkmerge assert check
Let's say there is a hardlink created as below.
touch file1
ln file1 hl_file1
In this case 'file1' holds the primary inode and 'hl_file' holds
the remote inode (holds primary inode number and other required info).
Now, if the 'file1' is deleted first, it's moved to a stray directory
and can't be deleted because the hardlink 'hl_file1' still exists
which requires primary inode. So on straydn eval, this primary
inode is linked to 'hl_file1 and remote inode is removed. This is called
the linkmerge/stray reintegration. So in the linkmerge case, the srcdnl
is primary straydn and the destdnl is the remote.
Jos Collin [Mon, 24 Jul 2023 08:46:52 +0000 (14:16 +0530)]
qa: fix cephfs-mirror unwinding and 'fs volume create/rm' order
* Fixes the 'fs volume create' happens before the cephfs-mirror daemon start.
* Fixes the 'fs volume rm' happen only after the cephfs-mirror daemon unwinding.
- This prevents the issue of mirror-daemon not returning from a libcephfs call, as
the volumes were deleted during the cephfs_mirror_thrash ing.
Fixes: https://tracker.ceph.com/issues/61182 Signed-off-by: Jos Collin <jcollin@redhat.com>
(cherry picked from commit b9a1a3cdf9770bcb27d6e08ddbc059f01674f4b8)
Jos Collin [Fri, 23 Jun 2023 06:16:26 +0000 (11:46 +0530)]
mds: MDLog::_recovery_thread: handle the errors gracefully
A write fails if the MDS is already blocklisted due to the 'fs fail' issued by the qa tests.
Handle those write failures gracefully, even when the MDS is stopping.
Fixes: https://tracker.ceph.com/issues/61201 Signed-off-by: Jos Collin <jcollin@redhat.com>
(cherry picked from commit d562905dcfb5b8a45ce7042c543720ef8b0fa05b)
N Balachandran [Fri, 14 Jul 2023 12:56:15 +0000 (18:26 +0530)]
rbd-mirror: fix race preventing local image deletion
On primary image deletion, a race between InstanceReplayer::release_image()
(which calls ImageReplayer::stop())and ImageReplayer::handle_bootstrap()
may prevent the non-primary image from being deleted. Because
ImageReplayer::handle_bootstrap() checks if the start has been canceled
before processing -ENOLINK, m_delete_requested is not set to true and the
local image is not deleted.
This commit sets m_delete_requested to true before checking if the
start has been canceled to ensure that the image is deleted during shut_down,
and updates ImageReplayer::bootstrap() to cancel the BootstrapRequest
and prevent a potentially long time spent in an image sync if ImageReplayer
stop is requested.
Fixes: https://tracker.ceph.com/issues/61672 Signed-off-by: N Balachandran <nibalach@redhat.com>
(cherry picked from commit 1df7921b916579c273cb597e234c9c6b721674d7)
Dhairya Parmar [Mon, 22 May 2023 10:37:34 +0000 (16:07 +0530)]
mds: remove code to bypass dumping empty header scrub info
Previously when ~mdsdir was scrubbed at CephFS root, it's header
was kept empty, thus it became necessary to not dump it's values
for 'scrub status'. Now since both the scrubs(~mdsdir and root)
run under the same header, this code is no more needed.
Dhairya Parmar [Mon, 22 May 2023 10:36:24 +0000 (16:06 +0530)]
mds: dump_values no more needed
Previouly, two individual scrubs were initiated to scrub ~mdsdir
at root where the ~mdsdir scrub wasn't provided any tag thus, it
was necessary to not dump it's values for output of 'scrub start'.
Now since mdsdir and root scrub run under single header, there is
no need for this anymore, thus removing this redundant code.
Dhairya Parmar [Mon, 22 May 2023 07:04:51 +0000 (12:34 +0530)]
mds: enqueue ~mdsdir at the time of enqueing root
This would avoid the need to run individual scrubs for
~mdsdir and root, i.e. run both the scrubs under the
same header, this also helps to avoid edge case where
in case ~mdsdir is huge and it's taking time to scrub it,
the scrub status would report something like this until
root inodes kick in:
{
"status": "scrub active (757 inodes in the stack)",
"scrubs": {}
}
Xiubo Li [Wed, 28 Jun 2023 13:59:53 +0000 (21:59 +0800)]
client: force sending cap revoke ack always
If just before the revoke request, which will increase the 'seq', is
sent out the clients released the corresponding caps and sent out
the cap update request with old 'seq', the mds will miss checking
the seqs and calculating the caps.
projects / ceph.git / log