cmake: only display the result of checking nasm once

if nasm is able to emit AVX512 instructions, we can assume that it's
able to generate AVX2 instructions as well. so no need to print "Found
nasm" multiple times.

and by chaining mutual exclusive branches with "elseif", we can have
better readability.

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

cmake: drop duplicated variable

not_arch_x32 replicates ${_support_x64}. and a variable starts with
"not" is just difficult to parse. so drop it.

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

Merge pull request #30027 from majianpeng/optimize-mutex-contention

os/bluestore: optimize mutex contention

Reviewed-by: Igor Fedotov <ifedotov@suse.com>

Merge pull request #36088 from rhcs-dashboard/fix-45822-master

mgr/dashboard:minimize console log traces of Ceph backend API test

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

Merge pull request #38651 from ifed01/wip-ifed-new-omap-naming

os/bluestore: introduce new "per-pg" naming scheme for omap

Reviewed-by: Josh Durgin <jdurgin@redhat.com>
Reviewed-by: Adam Kupczyk <akupczyk@redhat.com>

Merge pull request #38680 from ifed01/wip-ifed-rocksdb-options-annex

os/bluestore: introduce bluestore_rocksdb_options_annex config parame…

Reviewed-by: Josh Durgin <jdurgin@redhat.com>
Reviewed-by: Adam Kupczyk <akupczyk@redhat.com>

Merge pull request #38738 from ronen-fr/wip-ronenf-rawformat

librbd/migration: removing an unused capture

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

Merge pull request #38709 from wuhongsong/aios_be_truncated

blk: add upper bound of bluestore_deferred_batch_ops* options

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

Merge pull request #38655 from dillaman/wip-47828

librbd: ensure stripe unit is at least 512 bytes

Reviewed-by: David Disseldorp <ddiss@suse.de>
Reviewed-by: Mykola Golub <mgolub@suse.com>

librbd/migration: removing an unused capture

...silencing a compiler warning.

Signed-off-by: Ronen Friedman <rfriedma@redhat.com>

Merge pull request #38733 from ukernel/wip-yanzheng-email

mailmap: update affiliation for Yan, Zheng

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

Merge pull request #38237 from myoungwon/wip-fix-clone-size-mismatch

osd: fix clone size mismatch when deduped object is evicted

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

mailmap: update affiliation for Yan, Zheng

Signed-off-by: "Yan, Zheng" <ukernel@gmail.com>

Merge pull request #37704 from Huber-ming/log_error

radosgw-admin: fix error message when datalog list

Reviewed-by: Daniel Gryniewicz <dang@redhat.com>

blk: add upper bound of bluestore_deferred_batch_ops* options

so the number of pending io does not overflow when being passed to submit_batch().

Signed-off-by: hzwuhongsong <hzwuhongsong@corp.netease.com>

Merge pull request #38728 from xxhdx1985126/wip-crimson-osd_op-no-splice

crimson/osd: do not splice osd_op indata

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

crimson/osd: make sure PGBackend's modification operations not change osd_op state

Signed-off-by: Xuehan Xu <xxhdx1985126@gmail.com>

crimson/osd: do not splice osd_op indata

As each osd_op's indata holds only its own data, there's no need to drop their it.
On the other hand, indata is still needed when osd_op needs to be redone in case of
acting set change.

Signed-off-by: Xuehan Xu <xxhdx1985126@gmail.com>

Merge pull request #38633 from rhcs-dashboard/fix-logs-copy-button

mgr/dashboard: Fix for copy2Clipboard failing to copy on the first time

Reviewed-by: Alfonso Martínez <almartin@redhat.com>
Reviewed-by: Ernesto Puerta <epuertat@redhat.com>
Reviewed-by: Nizamudeen A <nia@redhat.com>
Reviewed-by: Aashish Sharma <aasharma@redhat.com>

mgr/dashboard: Fix for copy2Clipboard failing to copy on the first time.

Fixes: https://tracker.ceph.com/issues/48601
Signed-off-by: Avan Thakkar <athakkar@redhat.com>
Fix for the copy2Clipboard button to work on the first click itself.

Merge pull request #37267 from TheJJ/fix-up-acting-docs

doc: fix up and acting set explanation

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

doc: fix and improve the explainations of up and acting osd sets

acting = where clients requests are processed
up = where pgs were planned to be by crush+upmap

Fixes: https://tracker.ceph.com/issues/48718
Signed-off-by: Jonas Jelten <jj@sft.lol>

mailmap: update mail for Jonas Jelten

Signed-off-by: Jonas Jelten <jj@sft.lol>

Merge pull request #38719 from fengchunsong/master

test/perf_local: Add div32,prefetch,memory barrier bench for arm64

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

Merge pull request #38568 from batrick/i48594

pybind/mgr/cephadm: limit rgw osd caps

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

Merge pull request #38644 from pcuzner/fix-missing-py3-error

mgr/cephadm: correct connection failure message

Reviewed-by: Juan Miguel Olmo Martínez <jolmomar@redhat.com>
Reviewed-by: Sebastian Wagner <sebastian.wagner@suse.com>

Merge pull request #38654 from Daniel-Pivonka/osddryrunbug

mgr/cephadm:fix bug in orch apply osd --dry-run where empty table printed

Reviewed-by: Juan Miguel Olmo Martínez <jolmomar@redhat.com>
Reviewed-by: Sebastian Wagner <sebastian.wagner@suse.com>

test/perf_local: Add div32,prefetch,memory barrier bench for arm64

arm64 support div32,prefetch,memory barrier, so add them to benchmark.

Signed-off-by: Chunsong Feng <fengchunsong@huawei.com>
Signed-off-by: luo rixin <luorixin@huawei.com>

Merge pull request #38717 from swinds24/fix_compile_error

kv/LevelDBStore: fix incorrect ceph_abort_msg() usage

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

kv/LevelDBStore: fix incorrect ceph_abort_msg() usage

Compile error is encountered when HAVE_LEVELDB_FILTER_POLICY is not
defined.

/ceph/src/kv/LevelDBStore.cc: In member function 'int LevelDBStore::load_leveldb_options(bool, leveldb::Options&)':
/ceph/src/kv/LevelDBStore.cc:96:22: error: invalid initialization of reference of type 'const string& {aka const std::__cxx11::basic_string<char>&}' from expression of type 'bool'ceph_abort_msg(0 == "bloom size set but installed leveldb doesn't support bloom filters");
                    ~~^~~~
/ceph/src/include/ceph_assert.h:90:69: note: in definition of macro 'ceph_abort_msg'
   ::ceph::__ceph_abort( __FILE__, __LINE__, __CEPH_ASSERT_FUNCTION, msg)
                                                                     ^~~
/ceph/src/include/ceph_assert.h:62:19: note: in passing argument 4 of 'void ceph::__ceph_abort(const char*, int, const char*, const string&)'
 [[noreturn]] void __ceph_abort(const char *file, int line, const char *func,
                   ^~~~~~~~~~~~
src/kv/CMakeFiles/kv.dir/build.make:130: recipe for target 'src/kv/CMakeFiles/kv.dir/LevelDBStore.cc.o' failed

Signed-off-by: Jerry Lee <leisurelysw24@gmail.com>

Merge pull request #38702 from mkogan1/wip-fix-vstart-o

vstart: fix -o ceph.conf options processing

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

Merge pull request #38711 from Rethan/fix-osdmap-message-max-bytes-comments

option: update comments of option osd_map_message_max_bytes

option: update comments of option osd_map_message_max_bytes

Option osd_map_message_max_bytes indicates the max bytes of MOSDMap message, not maps count
as osd_map_message_max.

Signed-off-by: haoyixing <haoyixing@kuaishou.com>

Merge pull request #38700 from tchaikov/wip-seastar-msgr-test-timeout

test/crimson: add abort timer to messenger test

Reviewed-by: Yingxin Cheng <yingxin.cheng@intel.com>

Merge pull request #38576 from myoungwon/wip-fix-seg-fault-manifest-count

osd : prevent accessing deleted reference

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

Merge pull request #38662 from Songweibin/wip-121918

osd/ReplicatedBackend: handle tailing zeros of recovering objects nicer

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

Merge pull request #38665 from tchaikov/wip-arm32-cleanups

erasure-code, test, pybind: various arm32 related cleanups.

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

Merge pull request #38677 from ilanddev/mgr-deadlock-fix

mgr/ActivePyModules.cc: always release GIL before attempting to acquire a lock

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

Merge pull request #38678 from tchaikov/wip-denc

include/denc: use pair<const K,V> in range-based for loop

Reviewed-by: Adam C. Emerson <aemerson@redhat.com>

Merge PR #37428 into master

* refs/pull/37428/head:
mds: move duplicated inner loop out
mds: try to replicate hot dir to restarted MDS

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

Merge PR #36597 into master

* refs/pull/36597/head:
client: make the send_to_auth display of mds_requests more accurate

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

Merge PR #38037 into master

* refs/pull/38037/head:
qa: test_readahead add kernel client support

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

Merge PR #38631 into master

* refs/pull/38631/head:
MDSLogContext: fix typo in MDSLogContextBase and add more comments

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

Merge PR #38672 into master

* refs/pull/38672/head:
mds: increase the l_mds_openino_backtrace_fetch when fetching the backtrace

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

Merge PR #38692 into master

* refs/pull/38692/head:
pybind/cephfs: pass command as size-1 array of strings

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

Merge PR #38668 into master

* refs/pull/38668/head:
osdc: Fix return type of map waiting functions
monc: Use post rather than defer
common/async: Hold lock in constructor of blocked_result
osdc: Don't keep wrapping completion in wait_for_latest_osdmap
neorados: Replace unsafe uses of dispatch with post

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

os/bluestore: upgrade omap naming scheme to be per-pg

Signed-off-by: Igor Fedotov <ifedotov@suse.com>

Merge pull request #38707 from ceph/wip-yuriw-octopus-x-4-master

qa/tests: initial check in for octopus-x/parallel tests (try 2)

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

qa/tasks/cephadm: include first mgr host when applying mgrs

This prevents the first mgr from being shut down due to lack of
appropriate placements.

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

vstart: fix -o ceph.conf options processing

options passed to vstart.sh via the -o param and bunched on the same
line without line termination between them
hence not being parsed.

ex:
```
cat ./ceph.conf
...
debug_ms=0 debug_objecter=0 bluestore_debug_enforce_settings=ssd
bluestore_block_size=214748364800
...
```

Fixes: https://tracker.ceph.com/issues/48708
Signed-off-by: Mark Kogan <mkogan@redhat.com>

test/crimson: add abort timer to messenger test

Use a 120s timer to force abort the messenger test which should normally
succeeds in 60s, in order to:
* Fail the make check to print logs about where it is blocking;
* Free the listening IP address for the next make check;

Signed-off-by: Yingxin Cheng <yingxin.cheng@intel.com>
Signed-off-by: Kefu Chai <kchai@redhat.com>

mgr/dashboard:minimize console log traces of Dashboard backend API tests

Changed log levels in some files to reduce console log traces

Fixes: https://tracker.ceph.com/issues/45822
Signed-off-by: Aashish Sharma <aasharma@redhat.com>

qa/tests: add light octopus-x/copy (cephadm based)

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

Merge PR #37721 into master

* refs/pull/37721/head:
test: add tests for new snapshot APIs
pybind/cephfs: python bindings for new snapshot APIs
client: Snapshot cephfs APIs
mds: include snapshot metadata in trace reply to client
mds: include snap metadata on-disk on snapshot creation
mds: include metadata in SnapInfo structure

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

Merge pull request #38468 from IlsooByun/fix_lc_mp_iter

rgw: lc: fix multipart iteration problem

Merge PR #38611 into master

* refs/pull/38611/head:
qa/cephfs/nfs: Add test for readonly export

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

pybind/cephfs: pass command as size-1 array of strings

Fixes: https://tracker.ceph.com/issues/48701
Signed-off-by: Patrick Donnelly <pdonnell@redhat.com>

os/bluestore: introduce bluestore_rocksdb_options_annex config parameter.

This simplifies modifying a subset of rocksdb settings since it
eliminates the need to re-provide bluestore's rocksdb settings.

Signed-off-by: Igor Fedotov <ifedotov@suse.com>

Merge pull request #38187 from zdover23/wip-doc-dev-unit-tests-5-of-7

doc/dev: update unit-tests tox section

Reviewed-by: Ernesto Puerta <epuertat@redhat.com>

Merge pull request #38414 from pcuzner/fix-drivegroup-example

doc: Update drivegroup spec examples

Reviewed-by: Juan Miguel Olmo Martínez <jolmomar@redhat.com>

doc/dev: update unit-tests tox section

This commit
- alters the structure of the
  unit-test section of the Developer Guide
  that discusses the tox tool, which is
  used to run certain unit tests
- corrects some grammar
- updates the list of locations of tox.ini
  files
- This also incorporates N. Cutler's suggestions.

Signed-off-by: Zac Dover <zac.dover@gmail.com>

Merge pull request #38621 from anthonyeleven/anthonyeleven/bluestore-doc-cleanup

doc/rados/configuration: clean up BlueStore config and enhance DB sizing

Reviewed-by: Igor Fedotov <ifedotov@suse.com>

qa/cephfs/nfs: Add test for readonly export

Fixes: https://tracker.ceph.com/issues/48622
Signed-off-by: Varsha Rao <varao@redhat.com>

Merge pull request #38499 from aclamk/wip-bs-onode-prune-deleted

os/bluestore: Prune deleted onodes

Reviewed-by: Igor Fedotov <ifedotov@suse.com>
Reviewed-by: Josh Durgin <jdurgin@redhat.com>

Merge pull request #38650 from dmick/wip-dmick-docker-mirror

qa/distros: configure podman to pull from docker-mirror

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

include/denc: use pair<const K,V> in range-based for loop

map<K,V>::value_type is pair<const K, V>, so if we use range-based for
loop when iterating through a map, we should use pair<const K,V> instead
of pair<K,V>, the latter also compiles, but it might create a temporary
object of pair<K,V> from pair<const K,V>. GCC-11 complains at seeing
this:

../src/include/denc.h:1002:21: warning: loop variable ‘e’ of type ‘const T&’ {aka ‘const std::pair<OSDPerfMetricQuery, OSDPerfMetricReport>&’} binds to a tem\
porary constructed from type ‘const std::pair<const OSDPerfMetricQuery, OSDPerfMetricReport>’ [-Wrange-loop-constru
ct]
 1002 |       for (const T& e : s) {
      |                     ^

this change

* use the value_type of container in `maplike_details<Container>`,
  so we can avoid the overhead of creating temporay objects when
  encoding a map
* define denc_traits for std::pair<const A, B> as well, so the elements
  of a map can be encoded using denc facility

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

Merge PR #38666 into master

* refs/pull/38666/head:
doc: Fix mds capabilities for manual deployment

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

Merge pull request #37156 from CloudFerro/wip-allow-dynamic-levels

os/bluestore: allow dynamic levels

Reviewed-by: Igor Fedotov <ifedotov@suse.com>
Reviewed-by: Adam Kupczyk <akupczyk@redhat.com>
Reviewed-by: Mark Nelson <mnelson@redhat.com>

Merge PR #38669 into master

* refs/pull/38669/head:
mds: fix typo in debug logs

Reviewed-by: Rishabh Dave <ridave@redhat.com>
Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>
Reviewed-by: Varsha Rao <varao@redhat.com>

Merge pull request #38674 from mkogan1/rgw-fix-t48688

rgw: fix 'this->ondisk_version' will always evaluate to 'true'

Reviewed-by: Ronen Friedman <rfriedma@redhat.com>
Reviewed-by: Casey Bodley <cbodley@redhat.com>

mgr/ActivePyModules.cc: always release GIL before attempting to acquire a lock

A thread that holds the GIL while attempting to acquire a mutex will cause a deadlock
if another thread owns the mutex and is waiting on the GIL. The GIL must not be treated
like an ordinary mutex since it may be preempted at any time or released when doing
blocking I/O. Such deadlocks are severe since they starve all threads from access to the
GIL and therefore prevent any Python code from running until the mgr process is restarted.

Fixes: https://tracker.ceph.com/issues/39264
Signed-off-by: Cory Snyder <csnyder@iland.com>

Merge pull request #38653 from dillaman/wip-48647

librbd: clear implicitly enabled feature bits when creating images

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

Merge pull request #38670 from Rethan/doc-fix-max_part-description

doc/rbd-nbd: fix description for command rbd-nbd's parameter --max_part

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

rgw: fix 'this->ondisk_version' will always evaluate to 'true'

fix:
```
src/rgw/CMakeFiles/radosgw.dir/rgw_rest_metadata.cc.o
../src/rgw/rgw_rest_metadata.cc:267:13: warning: address of
'this->ondisk_version' will always evaluate to 'true'
[-Wpointer-bool-conversion]
                                       &ondisk_version);
```
Fixes: https://tracker.ceph.com/issues/48688
Signed-off-by: Mark Kogan <mkogan@redhat.com>

Merge pull request #38259 from rhcs-dashboard/fix-jwt-token

mgr/dashboard: Use secure cookies to store JWT Token

Reviewed-by: Alfonso Martínez <almartin@redhat.com>
Reviewed-by: Ernesto Puerta <epuertat@redhat.com>
Reviewed-by: Nizamudeen A <nia@redhat.com>

mds: increase the l_mds_openino_backtrace_fetch when fetching the backtrace

Signed-off-by: Xiubo Li <xiubli@redhat.com>

osd/ReplicatedBackend: handle tailing zeros of recovering objects nicer

E.g.,
(1) recovery_info.copy_subset = [0, 4194304]
(2) copy_subset(fiemap) = [0, 8192]
(3) out_op->data_included.span_of(copy_subset, 0, 4194304) = [0, 8192]

We'll set new_progress.data_recovered_to = 8192 as a result, and
fail to complete the push in one go.

Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>

doc/rbd-nbd: fix description for command rbd-nbd's parameter --max_part

The parameter --max_part should actually override module param max_part,
not as nbds_max as it said.

Signed-off-by: haoyixing <haoyixing@kuaishou.com>

mds: fix typo in debug logs

Signed-off-by: Xiubo Li <xiubli@redhat.com>

osdc: Fix return type of map waiting functions

Fixes: https://tracker.ceph.com/issues/47294
Signed-off-by: Adam C. Emerson <aemerson@redhat.com>

Merge pull request #38573 from batrick/i48600

osd: remove invalid put on message

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

monc: Use post rather than defer

Signed-off-by: Adam C. Emerson <aemerson@redhat.com>

common/async: Hold lock in constructor of blocked_result

More correct, though not the cause of error.

Signed-off-by: Adam C. Emerson <aemerson@redhat.com>

osdc: Don't keep wrapping completion in wait_for_latest_osdmap

Also remove unsafe use of dispatch.

Signed-off-by: Adam C. Emerson <aemerson@redhat.com>

neorados: Replace unsafe uses of dispatch with post

Dispatch may block the calling thread, so don't use it directly in a
called function. (In handlers is okay, though.)

Signed-off-by: Adam C. Emerson <aemerson@redhat.com>

cmake: install python scripts into /usr/sbin even if DESTDIR is empty

it's reported that on Alphine, "make install" installs
`ceph-volume-systemd` into `/usr/bin` instead of `/usr/sbin`, and this
breaks `systemd/ceph-volume@.service`. so in this change, we always
install this script into `/usr/sbin`, even if `DESTDIR` is not defined.

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

pybind/ceph_argparse.py: use a safe value for timeout

we have reports that on arm32 machines, it timed out immediately, so
to prevent it from int overflow, use a safer value instead of
(1 << (32 - 1)) - 1.

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

pybind/ceph_argparse.py: drop unused variable

and other PEP8 compliance cleanups

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

erasure-code: bail out early if chunks to be decoded is empty

otherwise we will read from uninitialized memory for blocksize.

update tests accordingly, as the number of decoded chunks is 0 if

* the number of available chunks is 0 and
* the number of requested chunks is not 0

after this change, because, in that case, the decode fails before any
chunk is allocated. but otherwise, the output chunks are still
allocated. so instead checking the number of output chunks, the test is
removed.

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

doc: Fix mds capabilities for manual deployment

Now matches auth-config-ref.rst, and actually works.

Signed-off-by: Ivan Jager <aij+git@mrph.org>

test/erasure-code: fold variables with the same values

for better readability

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

test/erasure-code: define variables closer to where they are used

for better readability.

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

erasure-code: print warnings using dout(10)

stderr is closed once the process is daemonized, while `derr` is for
fatal error or logic error in program. dout(10) is more appropriate in
this case, as it is an expected error from callee's perspective.

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

Merge pull request #38646 from xxhdx1985126/wip-crimson-bug-fix

crimson/osd: don't get recovery read lock in PGRecvery::on_local_recover()

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

crimson/osd: make sure the read lock is acquired when trying to get recovery read

Signed-off-by: Xuehan Xu <xxhdx1985126@gmail.com>

crimson/osd: don't get recovery read lock in PGRecvery::on_local_recover()

Now we've changed drop_recovery_read to only care about the recovery_read flag, so
we shouldn't get read lock when acquiring recovery read, otherwise there would be
a chance in which the read lock can't get released

Signed-off-by: Xuehan Xu <xxhdx1985126@gmail.com>

Merge pull request #38291 from aserdean/windows_common_proccess_name

common: Add get_process_name implementation for Windows

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

Merge pull request #38630 from tchaikov/wip-crimson-cleanups

crimson/osd: cleanups to drop some do_with() calls

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

Merge pull request #38522 from tchaikov/wip-mon-health-cleanup

mon/HealthMonitor: split large method into smaller ones and cleanups

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

Merge PR #38579 into master

* refs/pull/38579/head:
mds: ensure dirfrags are fetched once
mds: refactor variable names

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