rpm: Fix typo in summary of prometheus-alerts package

Signed-off-by: Matthias Bach <marix@marix.org>

Merge pull request #36915 from yuriw/wip-yuriw-46529-master

qa/tests: use bionic only for old clients in rados/thrash-old-clients

Reviewed-by: Sebastian Wagner <sebastian.wagner@suse.com>
Reviewed-by: Neha Ojha <nojha@redhat.com>

Merge pull request #36700 from leseb/retry-on-lock

ceph-volume: retry when acquiring lock fails

Merge pull request #36889 from idryomov/wip-relax-preauth-asserts

msg/async/ProtocolV2: allow rxbuf/txbuf get bigger in testing

Reviewed-by: Radoslaw Zarzynski <rzarzyns@redhat.com>

Merge pull request #36241 from jan--f/c-v-mpath-support

ceph-volume: support for mpath devices

Merge pull request #36917 from zdover23/wip-doc-rados-init-mems-to-mon-host

doc/rados: s/initial members/mon host

Reviewed-by: Brad Hubbard <bhubbard@redhat.com>

Merge pull request #36800 from adk3798/cephadm-46654

cephadm: Change service type to forking via systemd

Reviewed-by: Sebastian Wagner <sebastian.wagner@suse.com>

doc/rados: s/initial members/mon host

This changes an old string, "initial members"
to a current string, "mon host".

Fixes: https://tracker.ceph.com/issues/47116
Signed-off-by: Zac Dover <zac.dover@gmail.com>

Merge pull request #36879 from tchaikov/wip-doc-stable-release

doc: bump up sphinx to 3.2.1 and define "stable-release" global substitution

Reviewed-by: Willem Jan Withagen <wjw@digiware.nl>
Reviewed-by: Neha Ojha <nojha@redhat.com>

Merge pull request #36897 from tchaikov/wip-cmake-asan

cmake: expose Sanitizers_COMPILE_OPTIONS as a list

Reviewed-by: Brad Hubbard <bhubbard@redhat.com>

doc: fix broken hyper link

and always reference multisite.rst using tha label not the file.

Signed-off-by: Kefu Chai <kchai@redhat.com>

doc: fix broken hyper link

Signed-off-by: Kefu Chai <kchai@redhat.com>

doc: define label before section title

so we can see the title of section when following the link

Signed-off-by: Kefu Chai <kchai@redhat.com>

doc: add sphinx.ext.mathjax for math-to-MathML rendering

Signed-off-by: Kefu Chai <kchai@redhat.com>

doc: fix broken hyper link

this link was broken in 1427905c473e352e7cac1a9ac209cddb82544b57

Signed-off-by: Kefu Chai <kchai@redhat.com>

doc: always refernce terms in case sensitive mode

to silence warning from sphinx like:

ceph/doc/install/ceph-deploy/quick-start-preflight.rst:5: WARNING: term node not found in case sensitive match.made a reference to Node instead.

Signed-off-by: Kefu Chai <kchai@redhat.com>

include/rados/librados.h: avoid redefinition of rados_object_list_item

doxygen complains at seeing rados_object_list_item gets defined twice.
so let's fix it.

Signed-off-by: Kefu Chai <kchai@redhat.com>

doc: replace luminous with |stable-release|

"luminous" is a magic number in these contexts, and we should use a
constant for representing it.

the "sphinx_substitution_extensions" sphinx extension is introduced for
performing the global subsitution.

Signed-off-by: Kefu Chai <kchai@redhat.com>

Merge pull request #36907 from yuvalif/fix-issue-47209

cmake: make crimson libraries static to prevent cmake error

Reviewed-by: Kefu Chai <kchai@redhat.com>

qa/tests: use bionic only for old clients in rados/thrash-old-clients

Fixes: https://tracker.ceph.com/issues/46529
Signed-off-by: Yuri Weinstein <yweinste@redhat.com>

Merge PR #36089 into master

* refs/pull/36089/head:
mds/OpenFileTable: add option that disables dirfrags prefetch
mds: use parent/d_name in cap_reconnect_t to speedup MDCache::open_ino()
client: encode inode's parent/d_name in cap_reconnect_t
mds: explicitly track files with client writeable ranges
mds: properly set CLIENTWRITEABLE flag for imported caps
mds: in-place update projected inode's client ranges
mds: only add inodes that clients want Frw or excl caps to open file table
mds/OpenFileTable: reduce anchor map lookup during prefetch
mds/OpenFileTable: merge dirfrag set into anchor map

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

Merge PR #36887 into master

* refs/pull/36887/head:
qa: ignore expected mds failover message

Reviewed-by: Jeff Layton <jlayton@redhat.com>

Merge PR #36846 into master

* refs/pull/36846/head:
client: support getting ceph.dir.rsnaps vxattr
messages: fix decode version for snap_btime

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

Merge PR #36777 into master

* refs/pull/36777/head:
osdc/Journaler: do not call onsafe->complete() if onsafe is 0

Reviewed-by: Zheng Yan <zyan@redhat.com>
Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

Merge PR #36672 into master

* refs/pull/36672/head:
client: handle readdir reply without Fs cap

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

Merge PR #36631 into master

* refs/pull/36631/head:
client: dump frag's repmap if dir_replicated is true

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

Merge PR #36885 into master

* refs/pull/36885/head:
mon: ignore statfs requests on invalid poolid
qa: test for statfs on deleted file system
qa: move run_shell args to kwargs

Reviewed-by: Josh Durgin <jdurgin@redhat.com>
Reviewed-by: Kefu Chai <kchai@redhat.com>

Merge pull request #36881 from mhackett/wip-mhackett-47189

doc: document tuning of object gateway garbage collection

ceph-volume: retry when acquiring lock fails

When preaparing the osd device with --mkfs, the ceph-osd binary tries to
acquire an exclusive lock on the device (soon to become an OSD).
Unfortunately, when running in containers, we have seen cases where
there is a race between ceph-osd and systemd-udevd to acquire a lock on
the device. Sometimes systemd-udevd gets the lock and releases it soon
so that the ceph-osd gets sometimes the lock is still held and because
ceph-osd uses LOCK_NB the command fails.

This commit retries if the lock cannot be acquired, up to 5 times for 5
seconds, this should be more than enough to acquire the lock and
proceed with the OSD mkfs.

Unfortunately, this is so transient that we cannot lock earlier from c-v,
this won't do anything.

Fixes: https://tracker.ceph.com/issues/47010
Signed-off-by: Sébastien Han <seb@redhat.com>

crimson/cmake: fix copy&paste error

Signed-off-by: Yuval Lifshitz <ylifshit@redhat.com>

crimson/cmake: make libraries static to prevent cmake error

Fixes: https://tracker.ceph.com/issues/47209
Signed-off-by: Yuval Lifshitz <ylifshit@redhat.com>

Merge pull request #36826 from agayev/fix-hmsmr-compilation

os/bluestore: Restore HM-SMR functionality on BlueStore

Reviewed-by: Igor Fedotov <ifedotov@suse.com>
Reviewed-by: Kefu Chai <kchai@redhat.com>

Merge pull request #34653 from clwluvw/bucket-stats

rgw: Add bucket name to bucket stats error logging

Merge pull request #36857 from tchaikov/wip-crimson-conf

crimson/osd: support "dump_pgstate_history" command and update legacy conf vars

Reviewed-by: Radoslaw Zarzynski <rzarzyns@redhat.com>

crimson/osd: improve the const correctness

it should suffice to pass a `const OSD&` to the ctor of `OsdStatusHook`

Signed-off-by: Kefu Chai <kchai@redhat.com>

crimson/osd: support "dump_pgstate_history" command

it is required by the "ceph_test_admin_socket_output" test.

Signed-off-by: Kefu Chai <kchai@redhat.com>

crimson/admin: make injected_args optional

this option for `injectargs` command is optional, so we should not
crash if it does not exist.

Signed-off-by: Kefu Chai <kchai@redhat.com>

crimson/common: do not check for experimental/filesystem anymore

seastar is only compatible with C++17 and C++20, so no need to check for
<experimental/filesystem> anymore. the C++ standard library should be
C++17 compatible.

Signed-off-by: Kefu Chai <kchai@redhat.com>

crimson/common: call config::update_legacy_vals()

so the legacy vals are populated with the changes received from conffile.

Signed-off-by: Kefu Chai <kchai@redhat.com>

common/config: make update_legacy_vals() public

so crimson can call it in `ConfigProxy::parse_config_files()` after calling
`get_config().parse_buffer()`. so the legacy vals are populated with the
changes received from conffile.

Signed-off-by: Kefu Chai <kchai@redhat.com>

Merge pull request #36891 from tchaikov/wip-less-warnings

crimson/mon,common: replace #warning with // TODO

Reviewed-by: Radoslaw Zarzynski <rzarzyns@redhat.com>

Merge pull request #36892 from tchaikov/wip-crimson-cleanups

crimson/osd: return EAGAIN if !can_serve_replica_read() and cleanups

Reviewed-by: Radoslaw Zarzynski <rzarzyns@redhat.com>

Merge pull request #36905 from sebastian-philipp/cephadm-pep8-test-migration

mgr/cephadm: PEP8tify test_migration.py

Reviewed-by: Adam King <adking@redhat.com>

Merge pull request #36676 from matthewoliver/cephadm_orch_ipv6

cephadm: auto wrap and unwrap ipv6 addresses

Reviewed-by: Laura Paduano <lpaduano@suse.com>
Reviewed-by: Sebastian Wagner <sebastian.wagner@suse.com>

Merge pull request #36896 from xxhdx1985126/wip-47133

crimson/pg: check acting set change before appending pglog

Reviewed-by: Kefu Chai <kchai@redhat.com>

mgr/cephadm: PEP8tify test_migration.py

Signed-off-by: Sebastian Wagner <sebastian.wagner@suse.com>

Merge pull request #36877 from sebastian-philipp/mgr-module-fix-kwargs-cli-command

mgr/mgr_module.py: CLICommand: Fix parsing of kwargs arguments

Reviewed-by: Jan Fajerski <jfajerski@suse.com>
Reviewed-by: Kefu Chai <kchai@redhat.com>

Merge pull request #36859 from tspmelo/wip-fix-onselect

mgr/dashboard: Fix stale table actions

Reviewed-by: Volker Theile <vtheile@suse.com>
Reviewed-by: Stephan Müller <smueller@suse.com>
Reviewed-by: Laura Paduano <lpaduano@suse.com>

cmake: expose Sanitizers_COMPILE_OPTIONS as a list

* cmake/modules/FindSanitizers.cmake: do not pollute CMAKE_REQUIRED_FLAGS
* cmake/modules/FindSanitizers.cmake: expose Sanitizers_COMPILE_OPTIONS
  as a list
* CMakeLists.txt: append Sanitizers_COMPILE_OPTIONS to
  *_LINKER_FLAGS after replacing ";" with " " in it.

Signed-off-by: Kefu Chai <kchai@redhat.com>

Merge pull request #36839 from dillaman/wip-46875

librbd: flush all incomplete in-flight IOs upon image close

Reviewed-by: Mykola Golub <mgolub@suse.com>

Merge pull request #36819 from vumrao/wip-vumrao-46663

mon/PGMap: add pg count for pools in the ceph df command

Reviewed-by: Neha Ojha <nojha@redhat.com>
Reviewed-by: Kefu Chai <kchai@redhat.com>

crimson/pg: check acting set change before appending pglog

Fixes: https://tracker.ceph.com/issues/47133
Signed-off-by: Xuehan Xu <xxhdx1985126@gmail.com>

Merge pull request #33852 from rosinL/wip-fix-enqueue-dead-qp-deadlock

msg/async/rdma: avoid enqueue_dead_qp deadlock

Reviewed-by: Peng Liu <liupeng37@baidu.com>

Merge pull request #36767 from tchaikov/wip-47061

test/osd/RadosModel: check retval of read op

Reviewed-by: Samuel Just <sjust@redhat.com>

crimson/osd: return EAGAIN if !can_serve_replica_read()

it's necessary for serving balanced read.

Signed-off-by: Kefu Chai <kchai@redhat.com>

crimson/osd: check for primary out side of is_misdirected()

so we can check can_serve_replica_read() without checking primary again.

Signed-off-by: Kefu Chai <kchai@redhat.com>

crimson/osd: encode OI with CEPH_FEATURES_ALL

Signed-off-by: Kefu Chai <kchai@redhat.com>

crimson/osd: use logger() instead of crimson::get_logger(...)

for better readability

Signed-off-by: Kefu Chai <kchai@redhat.com>

crimson/osd: call PeeringState::get_peer_missing(peer)

as PeeringState::get_peer_missing(const pg_shard_t &peer) already checks
for the non-existent shard, and because `get_replica_recovery_order()`
never returns primary osd, it's safe to use
`PeeringState::get_peer_missing(peer)` instead of
`pg->get_peering_state().get_peer_missing().find(peer)`.

Signed-off-by: Kefu Chai <kchai@redhat.com>

crimson/osd: eliminate a local variable

for better readability

Signed-off-by: Kefu Chai <kchai@redhat.com>

test/osd/RadosModel: assure that tid seq is contiguous

before this change:

4:  finishing write tid 1 to gen81094472-1
4:  finishing write tid 2 to gen81094472-1
4:  finishing write tid 3 to gen81094472-1
4:  finishing write tid 5 to gen81094472-1
4:  finishing write tid 6 to gen81094472-1

after this change:

1:  finishing write tid 1 to gen81117124-1
1:  finishing write tid 2 to gen81117124-1
1:  finishing write tid 3 to gen81117124-1
1:  finishing write tid 4 to gen81117124-1
1:  finishing write tid 5 to gen81117124-1

more pleasant when reading the logging messages.

Signed-off-by: Kefu Chai <kchai@redhat.com>

test/osd/RadosModel: check retval of read op

* check racing read's return value
* check if racing read completes

Fixes: https://tracker.ceph.com/issues/47061
Signed-off-by: Kefu Chai <kchai@redhat.com>

test/osd/RadosModel: do not set done

we don't reuse TestOp, and `done` should have been reset in the
constructor.

Signed-off-by: Kefu Chai <kchai@redhat.com>

test/osd/RadosModel: use range-based for loop

Signed-off-by: Kefu Chai <kchai@redhat.com>

test/osd/RadosModel: use in-class initiailization

Signed-off-by: Kefu Chai <kchai@redhat.com>

test/osd: use stack for ObjectDesc::iterator::stack

better readability this way

Signed-off-by: Kefu Chai <kchai@redhat.com>

crimson/mon: replace #warning with TODO

less distracting when fixing compiling error / warnings this way

Signed-off-by: Kefu Chai <kchai@redhat.com>

common/admin_socket: replace #warning with TODO

less distracting when fixing compiling error / warnings this way

Signed-off-by: Kefu Chai <kchai@redhat.com>

Merge pull request #36815 from sebastian-philipp/cephadm-pep8-module

mgr/cephadm: PEP8tify module.py

Reviewed-by: Adam King <adking@redhat.com>
Reviewed-by: Michael Fritch <mfritch@suse.com>

mgr/cephadm: PEP8tify module.py

Signed-off-by: Sebastian Wagner <sebastian.wagner@suse.com>

Merge pull request #36858 from adk3798/cephadm-47171

mgr/cephadm: Verify non-empty list in get_active_daemon functions

Reviewed-by: Sebastian Wagner <sebastian.wagner@suse.com>

Merge pull request #36849 from sebastian-philipp/cephadm-kick-when-monmap

mgr/cephadm: kick serve loop to trigger ceph.conf deployment

Reviewed-by: Michael Fritch <mfritch@suse.com>

Merge pull request #36753 from sebastian-philipp/mgr-cephadm-schedule-redeploy

mgr/cephadm: Make daemon actions asynchronous

Reviewed-by: Adam King <adking@redhat.com>
Reviewed-by: Michael Fritch <mfritch@suse.com>

mon: ignore statfs requests on invalid poolid

Otherwise, we'll see a SIGSEGV when trying to get the size in
get_pool_free_space.

Fixes: https://tracker.ceph.com/issues/47182
Signed-off-by: Patrick Donnelly <pdonnell@redhat.com>

qa: test for statfs on deleted file system

Signed-off-by: Patrick Donnelly <pdonnell@redhat.com>

qa: move run_shell args to kwargs

To avoid listing all run args.

Signed-off-by: Patrick Donnelly <pdonnell@redhat.com>

Merge pull request #35891 from changchengx/multi_stress_watch

test/multi_stress_watch: use internel API to connect cluster

Reviewed-by: Kefu Chai <kchai@redhat.com>

Merge pull request #36868 from zhoufeng2021/removeunsed

msg/async: cleanup unused function local_deliver()

Reviewed-by: Kefu Chai <kchai@redhat.com>

test/osd/RadosModel: use insert_or_assign()

for better performance, and less repeatings.

Signed-off-by: Kefu Chai <kchai@redhat.com>

test/osd/RadosModel: add more comments

Signed-off-by: Kefu Chai <kchai@redhat.com>

test/osd/RadosModel: rename local variable

for better readability

Signed-off-by: Kefu Chai <kchai@redhat.com>

Merge pull request #36888 from batrick/release-notes-union

doc: use union merge strategy for resolving conflicts in PendingRelea…

Reviewed-by: Kefu Chai <kchai@redhat.com>

msg/async/ProtocolV2: allow rxbuf/txbuf get bigger in testing

We have a kernel client test case that constructs huge auth tickets
to exercise the three related code paths in the kernel.  One of the
tickets is bigger than 1000000 bytes, as required for triggering the
third code path.

We haven't bumped into this assert earlier because the kernel client
is still on msgr v1.  However, "rbd map" and "rbd unmap" commands
started connecting to the cluster in commit 96f05a7956b3 ("rbd: delay
determination of default pool name") and that happens via msgr v2.

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>

Merge pull request #36886 from batrick/librados-statfs-header

librados: remove unused header

Reviewed-by: Kefu Chai <kchai@redhat.com>

msg/async/rdma: avoid enqueue_dead_qp deadlock

In RDMADispatcher::handle_async_event, when case
IBV_EVENT_QP_LAST_WQE_REACHED matched, enqueue_dead_qp wants the lock,
result in deadlock here. Change enqueue_dead_qp to lockless.

Fixes: https://tracker.ceph.com/issues/44298
Signed-off-by: Chunsong Feng <fengchunsong@huawei.com>
Signed-off-by: luorixin <luorixin@huawei.com>

librados: remove unused headers

Signed-off-by: Patrick Donnelly <pdonnell@redhat.com>

doc: use union merge strategy for resolving conflicts in PendingReleaseNotes

Generally, the conflict resolution performed manually is always to do a
union: automate this. This avoids pointless rebases of PRs.

Signed-off-by: Patrick Donnelly <pdonnell@redhat.com>

Merge PR #36701 into master

* refs/pull/36701/head:
client: make Client::open() pass proper cap mask to path_walk

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

Merge PR #36641 into master

* refs/pull/36641/head:
mds : skip show_cache if level of log is lower than 7

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>
Reviewed-by: Zheng Yan <zyan@redhat.com>

qa: ignore expected mds failover message

Fixes: https://tracker.ceph.com/issues/47202
Signed-off-by: Patrick Donnelly <pdonnell@redhat.com>

Merge PR #36882 into master

* refs/pull/36882/head:
qa/tests: Added "--force-priority"

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

Merge pull request #36729 from changchengx/zero_malloc

crush: do not malloc for empty bucket and avoid member variables after memset() with 0

Reviewed-by: Kefu Chai <kchai@redhat.com>

Merge pull request #36820 from tchaikov/wip-test-multi_stress_watch

test/multi_stress_watch: use condition_variable for level triggering

Reviewed-By: Josh Durgin <jdurgin@redhat.com>
Reviewed-by: Neha Ojha <nojha@redhat.com>
Reviewed-by: Samuel Just <sjust@redhat.com>

Merge pull request #36832 from neha-ojha/wip-clean-temps

tools/osdmaptool.cc: add ability to clean_temps

Reviewed-By: Josh Durgin <jdurgin@redhat.com>

qa/tests: Added "--force-priority"

Signed-off-by: Yuri Weinstein <yweinste@redhat.com>

doc: document tuning of object gateway garbage collection
Fixes: https://tracker.ceph.com/issues/47189
Signed-off-by: mhackett <mhackett@redhat.com>

admin/doc-requirements: bump up sphinx to 3.2.1

* use the latest stable sphinx
* unpin breathe, sphinx-autodoc-typehints, so pip can choose the
  compatible versions without worrying about the compile failures.

Signed-off-by: Kefu Chai <kchai@redhat.com>

doc/conf.py: define CEPH_RADOS_API for breathe

otherwise we could have following errors:

Invalid C declaration: Expected identifier in nested name, got keyword: int [error at 18]
      CEPH_RADOS_API int rados_aio_append (rados_ioctx_t io, const char *oid, rados_completion_t completion, const char *buf, size_t len)
      ------------------^

Signed-off-by: Kefu Chai <kchai@redhat.com>

script/gen_state_diagram.py: s/python/python3/

python does not necessarily on a modern distro which has only python3,
so use python3 instead.

Signed-off-by: Kefu Chai <kchai@redhat.com>

Merge pull request #36854 from idryomov/wip-krbd-default-pool-fix

rbd: fix default pool handling for krbd map/unmap

Reviewed-by: Jason Dillaman <dillaman@redhat.com>