Merge pull request #22786 from pdvian/wip-24535-luminous

luminous: client: _ll_drop_pins may access invalid iterator

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

Merge pull request #22811 from smithfarm/wip-24694-luminous

luminous: PurgeQueue sometimes ignores Journaler errors

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

Merge pull request #23329 from ceph/backport-luminous-23234

luminous: ceph-volume: PVolumes.get() should return one PV when using name or uuid

Reviewed-by: Alfredo Deza <adeza@redhat.com>

Merge pull request #22814 from smithfarm/wip-24748-luminous

luminous: change default filestore_merge_threshold to -10

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

Merge pull request #22839 from vumrao/wip-vumrao-backport-d12ba37e0d0-to-luminous

luminous: osd: change log level when withholding pg creation

Reviewed-by: Josh Durgin <jdurgin@redhat.com>
Reviewed-by: Neha Ojha <nojha@redhat.com>
Reviewed-by: Brad Hubbard <bhubbard@redhat.com>

Merge pull request #22979 from VictorDenisov/backport_24033

luminous: rados.pyx: make all exceptions accept keyword arguments

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

Merge pull request #23064 from pdvian/wip-24886-luminous

luminous: os/bluestore: fix races on SharedBlob::coll in ~SharedBlob.

Reviewed-by: Sage Weil <sage@redhat.com>

Merge pull request #23331 from alfredodeza/wip-rm25170

luminous: ceph-volume add a __release__ string, to help version-conditional calls

Reviewed-by: Andrew Schoen <aschoen@redhat.com>

Merge pull request #23293 from smithfarm/wip-upgrade-cleanup-luminous

luminous: qa: drop upgrade/luminous-x

Reviewed-by: Yuri Weinstein <yweins@redhat.com>

ceph-volume add a __release__ string, to help version-conditional calls

Signed-off-by: Alfredo Deza <adeza@redhat.com>

ceph-volume: PVolumes.get() should return one PV when using name or uuid

It is possible to get duplicated pv entires from the 'pvs' lvm command.
If we're using PVolumes.get() with either pv_name or pv_uuid we can
safely return a single PVolume object. However, if we use pv_tags with
PVolumes.get() we must still raise a MultiplePVsError if many pvs
are found that have the tags, because they are not guaranteed to be the
same pv, as would be the case with pv_name or pv_uuid.

Fixes: http://tracker.ceph.com/issues/24784
Signed-off-by: Andrew Schoen <aschoen@redhat.com>
(cherry picked from commit e5e866e8ecaf0522f3ebac83fa6fbc5a666acd58)

Merge pull request #23238 from alfredodeza/luminous-rm24993

luminous ceph-volume process.call with stdin in Python 3 fix

Reviewed-by: Andrew Schoen <aschoen@redhat.com>

Merge pull request #23296 from b-ranto/wip-volume-selinux-luminous

[luminous] ceph-volume: Restore SELinux context

Reviewed-by: Alfredo Deza <adeza@redhat.com>

ceph-volume: Restore SELinux context

We need to reset the root context of the file system after mounting it.
Otherwise, the SELinux policy rules will not be preserved.

Fixes: https://tracker.ceph.com/issues/24785
Signed-off-by: Boris Ranto <branto@redhat.com>
(cherry picked from commit 61dbf4ac2af7d1e9da927bbc4006c9fbccb55478)

ceph-volume tests.util verify as_bytes behavior

Signed-off-by: Alfredo Deza <adeza@redhat.com>
(cherry picked from commit 1802c386da3bf1e77f20d8a5456edd810a631c6f)

ceph-volume tests catch utf-8 encoded stdin errors

Signed-off-by: Alfredo Deza <adeza@redhat.com>
(cherry picked from commit 31f565e90bb346c2be982f7daba31431b5368677)

ceph-volume process use the as_bytes util for stdin

Signed-off-by: Alfredo Deza <adeza@redhat.com>
(cherry picked from commit 6a9368c1db468e05944607f9814f19053de80b6b)

ceph-volume util create a helper for converting to bytes

Signed-off-by: Alfredo Deza <adeza@redhat.com>
(cherry picked from commit fb518165445479cf08eefc8a2755ad78dcafd07f)

ceph-volume tests validate the new utils for prompts

Signed-off-by: Alfredo Deza <adeza@redhat.com>
(cherry picked from commit 2d988337a96fe6cc8fffb37e4641c86cd47e1139)

ceph-volume tests verify str_to_int utility

Signed-off-by: Alfredo Deza <adeza@redhat.com>
(cherry picked from commit 56b1bf633d3d16cf335043450df95f191e2907a4)

ceph-volume util create prompt helpers

Signed-off-by: Alfredo Deza <adeza@redhat.com>
(cherry picked from commit 0283fcc38b13c2d090b1d25be732e6f5f43b834c)

ceph-volume util move the str_to_int utility out of lvm

Signed-off-by: Alfredo Deza <adeza@redhat.com>
(cherry picked from commit cb5b30ab43f37adc45fb6114452b37722c65501e)

ceph-volume tests run to log_descriptors, verifying utf-8

Signed-off-by: Alfredo Deza <adeza@redhat.com>

ceph-volume process decode to utf-8 when logging descriptors

Signed-off-by: Alfredo Deza <adeza@redhat.com>

ceph-volume tests.process ensure that stdin works on py3 and py2.7

Signed-off-by: Alfredo Deza <adeza@redhat.com>

ceph-volume process encode stdin with utf-8 to prevent bytes TypeError

Signed-off-by: Alfredo Deza <adeza@redhat.com>

luminous: qa: drop upgrade/luminous-x

This suite would typically be run from the mimic or master branches, not
luminous.

This commit is intentionally not a cherry-pick, since it is relevant to
luminous only.

Signed-off-by: Nathan Cutler <ncutler@suse.com>

Merge pull request #23274 from yuriw/wip-yuriw-25123-luminous

qa/upgrade/luminous-p2p: cleaned up "#branch: luminous" lines

Reviewed-by: Nathan Cutler <ncutler@suse.com>

Merge pull request #23179 from smithfarm/wip-25066-luminous

luminous: build/ops: install-deps.sh fails on newest openSUSE Leap

Reviewed-by: Tim Serong <tserong@suse.com>

qa/tests: cleaned up "#branch: luminous" lines

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

Merge pull request #23265 from yuriw/wip-yuriw-25123-luminous

qa/tests: added 12.2.7 to the mix

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

Merge pull request #23150 from alfredodeza/luminous-rm24952

luminous: ceph-volume lvm.listing only include devices if they exist

Reviewed-by: Andrew Schoen <aschoen@redhat.com>

qa/tests: added 12.2.7 to the mix
also added 'overall HEALTH_' to the whitelist

Fixes: https://tracker.ceph.com/issues/25123
Signed-off-by: Yuri Weinstein <yweinste@redhat.com>

Merge pull request #23177 from smithfarm/wip-25063-luminous

luminous: doc: ceph-bluestore-tool manpage not getting rendered correctly

Reviewed-by: Sage Weil <sage@redhat.com>
Reviewed-by: Igor Fedotov <ifedotov@suse.com>

ceph-volume lvm.listing only include devices if they exist

Signed-off-by: Alfredo Deza <adeza@redhat.com>
(cherry picked from commit 37565f81634f47c389926d300ed5f92b223bddea)

ceph-volume: do not pin the testinfra version for the simple tests

For testinfra to work with a 2.6.x version of ansible it must use the
latest version.

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
(cherry picked from commit 70b86012b56fd4b66fc0c7ce88a6cfbe895a0702)

Merge pull request #23264 from ceph/backport-luminous-23263

luminous: ceph-volume: update version of ansible to 2.6.x for simple tests

Reviewed-by: Alfredo Deza <adeza@redhat.com>

ceph-volume: update version of ansible to 2.6.x for simple tests

ceph-ansible now requires a 2.5.x or 2.6.x version of ansible if you're
using the master branch. This updates our functional tests for the
simple subcommand to use a 2.6.x version of ansible.

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
(cherry picked from commit 654578928dd840801967c5409b5b19ad6cf90317)

Merge pull request #23245 from alfredodeza/luminous-ansible-26

luminous ceph-volume tests/functional use Ansible 2.6

Reviewed-by: Andrew Schoen <aschoen@redhat.com

ceph-volume tests/functional use testinfra latest for ansible 2.6

Signed-off-by: Alfredo Deza <adeza@redhat.com>
(cherry picked from commit 4f48ceab975539c9ce989aade7a1f3ae7d5b51d5)

ceph-volume tests/functional use Ansible 2.6

The ceph-ansible project now refuses to use 2.4 even though it works
with that version

Signed-off-by: Alfredo Deza <adeza@redhat.com>
(cherry picked from commit 9570864dd614d2442278e2ba3501788fefd753e3)

Merge pull request #23235 from rhcs-dashboard/wip-25102-luminous

pybind/mgr/mgr_module: make rados handle available to all modules

pybind/mgr/mgr_module: make rados handle available to all modules

Fixes: http://tracker.ceph.com/issues/25102
Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit 366905fb5779ecb110ec4216afe7cde14fbbb2d8)
Signed-off-by: Ernesto Puerta <epuertat@redhat.com>

Merge pull request #23227 from liewegas/wip-25057-luminous

luminous: osd/OSDMap: CRUSH_TUNABLES5 added in jewel, not kraken

osd/OSDMap: CRUSH_TUNABLES5 added in jewel, not kraken

In 0ceb5c0f3694725c7043ea99b077816da2ca44a7 I screwed up the version that
CRUSH_TUANBLES5 (chooseleaf_stable) was added: it comes from
043a7378ec4e828f2568ea3906d9076848161ccd, v10.0.3, which leads up to
jewel.

This, in turn, means that when jewel tunables are set we do not actually
encode the jewel tunables.  At least, luminous doesn't; jewel uses a
different code path to decode how to encode the OSDMap.

This, in turn, leads to crc errors:

- luminous mons running
- require_osd_release is jewel
- jewel tunables are set.  mon in-memory crushmap gets chooseleaf_stable=1
- osds are luminous, but don't see the new field in the crushmap
- upgrade finishes, and require_osd_Release becomes luminous
- the full osdmap encodes the chooseleaf_stable field, finally, and it
  is 1
- the osds try to encode the same thing, but their crushmap has 0.
-> crc mismatch

This is a regression from 0ceb5c0f3694725c7043ea99b077816da2ca44a7,
backported to luminous in 686b054f48c50859bd5e8bad93007bf9d7d7899c in
12.2.6.

Fixes: 686b054f48c50859bd5e8bad93007bf9d7d7899c
Fixes: http://tracker.ceph.com/issues/25057
Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit 7e9d24d0602abefa4cca245f5f86dcd4b412b97a)

Merge pull request #23219 from liewegas/wip-slow-requests-upgrade-luminous

luminous: qa/suites/upgrade/jewel-x: whitelist 'slow requests'

qa/suites/upgrade/jewel-x: whitelist 'slow requests'

This message is generated by jewel but not luminous.

Signed-off-by: Sage Weil <sage@redhat.com>

Add new OpenSUSE Leap id for install-deps.sh

Signed-off-by: Kyr Shatskyy <kyrylo.shatskyy@gmail.com>
(cherry picked from commit bd0448c0b5a09a9314fddf1757502529801366e3)

Conflicts:
install-deps.sh (trivial resolution)

doc: use :command: for subcommands in ceph-bluestore-tool manpage

Older versions of Sphinx, such as the one in CentOS 7, do not render "..
option::" lines correctly if the option contains a hyphen but does not start
with a hyphen. And ceph-bluestore-tool appears to be the only Ceph manpage
affected by this bug.

Fixes: http://tracker.ceph.com/issues/24800
Signed-off-by: Nathan Cutler <ncutler@suse.com>
(cherry picked from commit 7c9229e7876dfc1bf6e415ed7cce924c3b50925d)

Merge pull request #23147 from alfredodeza/luminous-rm24961

ceph-volume tests/functional run lvm list after OSD provisioning

Reviewed-by: Andrew Schoen <aschoen@redhat.com>

ceph-volume tests/functional run lvm list after OSD provisioning

Signed-off-by: Alfredo Deza <adeza@redhat.com>
(cherry picked from commit e01fb206d67f26724f3b03e4b22f5ba6eea22207)

Merge pull request #23102 from ceph/backport-luminous-23093

luminous: ceph-volume: an OSD ID must exist and be destroyed before reuse

Reviewed-by: Alfredo Deza <adeza@redhat.com>

Merge pull request #23128 from ceph/backport-luminous-23117

luminous: ceph-volume: unmount lvs correctly before zapping

Reviewed-by: Alfredo Deza <adeza@redhat.com>

ceph-volume: unmount lvs correctly before zapping

If the lv found does not include the meatadata to construct an osd path,
then we should try to unmount it's path instead.

Fixes: http://tracker.ceph.com/issues/24796
Signed-off-by: Andrew Schoen <aschoen@redhat.com>
(cherry picked from commit 5b175a5bb68264b80c00cb4f04026a4d19ca20ad)

ceph-volume: an osd ID must exist and be destroyed before reuse

Fixes: https://tracker.ceph.com/issues/24044
Signed-off-by: Andrew Schoen <aschoen@redhat.com>
(cherry picked from commit 02a13f2de0783ff3c680a6042388a4692250edfb)

ceph-volume util/prepare.py

Replaces PR#21887
Rename function check_id to osd_id_exists in util/prepare.py
Fix boolean analysis of osd_id_exists when --osd-id is provided on command line

Signed-off-by: Ron Allred <rallred@itrefined.com>
(cherry picked from commit 4641ebb075adc987e83d7440f9756104fc04f83a)

12.2.7

Merge PR #23055 into luminous

* refs/pull/23055/head:
common/options: disable osd_skip_data_digest
common/options: make osd_skip_data_digest a DEV option
common: add osd_distrust_data_digest option
osd: apply osd_skip_data_digest to read operations
osd: do not skip_data_digests during scrub
osd/PrimaryLogPG: use legacy style for osd_skip_data_digest

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

os/bluestore: fix races on SharedBlob::coll in ~SharedBlob.

Example scenario:
  ```
  A: BlueStore::Collection::split_cache(src, dest)
  A: std::lock(src->cache->lock, dest->cache->lock)

  B: SharedBlob::~SharedBlob
  B: waits coll->cache->lock // coll == src

  A: sb->coll := dest
  A: unlocks both src and dest's cache mutexes

  C: // any locked operation on dest's cache
  C: acquires dest->cache->lock
  C: begins the op

  B: // with the src's cache mutex acquired
  B: BufferSpace::_clear(coll->cache) // coll == dest
  B: // oops, B operates on cache already locked by C
  ```

Fixes: http://tracker.ceph.com/issues/24859
Signed-off-by: Radoslaw Zarzynski <rzarzyns@redhat.com>
(cherry picked from commit d1b16749c7d32eead6e326c492ef9edf0d03a2b8)

common/options: disable osd_skip_data_digest

This is flawed in that it stops doing digests based on the primary's
backend store, not the replicas.

Signed-off-by: Sage Weil <sage@redhat.com>

common/options: make osd_skip_data_digest a DEV option

And document/warn to never turn it off after it's been on.

Signed-off-by: Sage Weil <sage@redhat.com>

common: add osd_distrust_data_digest option

If we have reason to distrust the stored full-object oi data digests,
this option will ignore them (and opportunistically clear them).  It
basically affects the same behaviors that osd_skip_data_digest except
the previous option only takes effect if bluestore is in use, while this
option is unconditional.

This serves a workaround for clusters that suffered the 12.2.6 bug that
produced bad full-object digests and *also* had PGs' primaries move from
bluestore to filestore nodes.  In those situations, this option can be
set to true.

After all OSDs are upgrade and all PGs come back with a clean deep
scrub this option can be disabled again.

Signed-off-by: Sage Weil <sage@redhat.com>

osd: apply osd_skip_data_digest to read operations

If we would skip updating data digests, apply that same decision to read
operations as well.

Signed-off-by: Sage Weil <sage@redhat.com>

osd: do not skip_data_digests during scrub

The goal was to skip storing per-object digests.  Scrub still needs to
calculate these digests in order to do a deep scrub, however!

This is not a cherry-pick of master because the code has changed a bit
and because the master fix is not merged yet.

Fixes: http://tracker.ceph.com/issues/24922
Signed-off-by: Sage Weil <sage@redhat.com>

osd/PrimaryLogPG: use legacy style for osd_skip_data_digest

This is more efficient.

Signed-off-by: Sage Weil <sage@redhat.com>

Merge pull request #23039 from liewegas/wip-24903-workaround-luminous

rpm,deb: fix chown to use -h

rpm,deb: fix chown to use -h

Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit 92207c76b91c26c0558688a358ade7c0d94d1b59)

Merge PR #23028 into luminous

* refs/pull/23028/head:
osd/PrimaryLogPG: clear data digest on WRITEFULL if skip_data_digest

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

Merge PR #23035 into luminous

* refs/pull/23035/head:
rpm,deb: chown bluestore block* links to ceph.ceph on install/upgrade

rpm,deb: chown bluestore block* links to ceph.ceph on install/upgrade

This works around https://tracker.ceph.com/issues/24903

Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit 58cde3fd793c037beca6645cc30b2ae1c30e2af1)

Merge pull request #22369 from linuxbox2/wip-luminous-rgw-cache-time-added

Wip luminous rgw cache time added

Merge PR #22962 into luminous

* refs/pull/22962/head:
osd/PrimaryLogPG: rebuild attrs from clients

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

Merge PR #22976 into luminous

* refs/pull/22976/head:
qa/standalone/osd/ec-error-rollforward: reproduce bug 24597
qa/standalone/osd/repro_long_log.sh: fix test
osd/PG: do not blindly roll forward to log.head

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

Merge PR #23007 into luminous

* refs/pull/23007/head:
qa/suites/rados/singleton/all/osd-recovery*: whitelist REQUEST_SLOW

osd/PrimaryLogPG: clear data digest on WRITEFULL if skip_data_digest

If we detect we are on bluestore and should skip the data digest, we
need to clear any old digest on WRITEFULL.  This aligns us with the
WRITE behavior, which is also either an update or a clear (but never
neither!).

Fixes: https://tracker.ceph.com/issues/23871
Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit fe5038c7f9577327f82913b4565712c53903ee48)

Merge PR #22872 into luminous

* refs/pull/22872/head:
doc: outline steps to take CephFS fs down

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>
Reviewed-by: Nathan Cutler <ncutler@suse.com>

Merge PR #22963 into luminous

* refs/pull/22963/head:
mon/AuthMonitor: improve error message

Reviewed-by: Douglas Fuller <dfuller@redhat.com>

qa/suites/rados/singleton/all/osd-recovery*: whitelist REQUEST_SLOW

Signed-off-by: Sage Weil <sage@redhat.com>

qa/standalone/osd/ec-error-rollforward: reproduce bug 24597

This reproduces http://tracker.ceph.com/issues/24597

Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit 4108ebc0abae83777ebc69bda610167f39e22109)

qa/standalone/osd/repro_long_log.sh: fix test

The log trimming case wasn't quite right.  Before HEAD^ we were
rolling forward too aggressively and miscalculating the can_rollforward_to,
which affected the trim_to calculation.

Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit 4f9fdd98e2dde3979ec88a3f761d76f7b297415c)

osd/PG: do not blindly roll forward to log.head

If we are told we can roll forward by the primary, we should only roll
forward as far as the primary says we can.

This probably came out of the similar case in append_log(), but notably
that roll_forward() only happens if !transaction_applied (i.e., backfill
target), and that condition is not checked here.

Fixes: http://tracker.ceph.com/issues/24597
Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit 858ff647ff7bd1f6277823b5185ead5ffebc05b0)

rados.pyx: make all exceptions accept keyword arguments

Moving the code that makes exceptions accept keyword arguments from the
exception OSError to Error prevents a crash when exceptions not
inheriting OSError are received.

Fixes: http://tracker.ceph.com/issues/24033
Signed-off-by: Rishabh Dave <ridave@redhat.com>
(cherry picked from commit 16458025da1c5854f1ef3840ac7d58fb5a715c64)

mon/AuthMonitor: improve error message

Improve the error message for 'ceph fs authorize' where the
given entity exists but has different auth caps.

Fixes: http://tracker.ceph.com/issues/21765
Signed-off-by: Douglas Fuller <dfuller@redhat.com>
(cherry picked from commit 447b3d4852acd2db656c973cc224fb77d3fff590)

osd/PrimaryLogPG: rebuild attrs from clients

Ensure that buffers coming in via client ops are rebuilt before populating
the PGTransaction.  This ensures that we don't pin the raw buffers for
the entire incoming message in memory.

In the past we've addressed this issue at the ObjectStore layer, but we
did not consider the attr_cache in ObjectContext.  Rebuilding the buffers
at this point will sanitize any incoming attribute before it reaches
either PGBackend or ObjectContext object_cache or the ObjectStore
implementation.

Fixes: http://tracker.ceph.com/issues/24768
Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit 0a0f8d5e6fa8276f06a9b7a7b18ea6f15fa8cb86)

12.2.6

Merge pull request #22911 from dillaman/wip-24740-luminous

luminous: qa/workunits/rbd: adapt import_export test to handle multiple units

Reviewed-by: Nathan Cutler <ncutler@suse.com>

Merge remote-tracking branch 'private/wip-mon-snap-caps-luminous' into luminous

Merge remote-tracking branch 'private/wip-cx-luminous' into luminous

qa/workunits/rbd: adapt import_export test to handle multiple units

Fixes: http://tracker.ceph.com/issues/24740
Signed-off-by: Jason Dillaman <dillaman@redhat.com>

Merge pull request #22855 from jcsp/wip-luminous-cephfs-doc

[luminous] doc/cephfs: remove lingering "experimental" note about multimds

Reviewed-by: Nathan Cutler <ncutler@suse.com>

doc: outline steps to take CephFS fs down

Fixes: https://tracker.ceph.com/issues/23427
Signed-off-by: Patrick Donnelly <pdonnell@redhat.com>
(cherry picked from commit 9aa13bf0b0bf20b4fcf4949af60ebfff1e8fea0a)

doc/cephfs: remove lingering "experimental" note about multimds

Signed-off-by: John Spray <john.spray@redhat.com>

luminous: osd: change log level when withholding pg creation

Signed-off-by: Dan van der Ster <daniel.vanderster@cern.ch>
(cherry picked from commit d12ba37e0d08593f37a2e162d4ac3b83ada7f0b6)
Signed-off-by: Vikhyat Umrao <vumrao@redhat.com>

Merge pull request #22831 from ceph/backport-luminous-22828

luminous: ceph-volume: provide a nice errror message when missing ceph.conf

Reviewed-by: Alfredo Deza <adeza@redhat.com>

ceph-volume: always ignore a missing ceph conf in main.py

Now that we have a nice error message when a ceph.conf is missing
and we try to use values from it, maintaining a list of commands that
don't need ceph.conf isn't as helpful. We had actually missed 'simple
trigger' when we first implemented this causing all our luminous tests
for simple to fail when we backported.

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
(cherry picked from commit 0487a9ac60c5ba289b7dd7c730b014a298c5998f)

ceph-volume: show a nice error message when ceph.conf is not loaded

If the ceph configuration file is not loaded correctly and then values
from it are used then an undescript error message is shown, e.g.

AttributeError: 'property' object has no attribute 'get'

With this change that same error condition shows the message:

RuntimeError: No valid ceph configuration file was loaded.

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
(cherry picked from commit e42ecb831914b07c7c8559d4556a8aabd2accd0c)

osd/filestore: Change default filestore_merge_threshold to -1

Performance evaluations of medium to large size Ceph clusters have
demonstrated negligible performance impact from unnecessarily deep
directory hierarchies but significant performance impact from filestore
split and merge activity. Disable merges by default.

Fixes: http://tracker.ceph.com/issues/24686
Signed-off-by: Douglas Fuller <dfuller@redhat.com>
(cherry picked from commit 75f55f2dfc92a89ec9d821e00effad75839008c9)

Conflicts:
src/common/options.cc (retain LEVEL_ADVANCED in luminous because
            65bec679f250ecc8406ac777f373785a3e110031 is not being backported)

mds: improve error handling in PurgeQueue

Fixes: http://tracker.ceph.com/issues/24533
Signed-off-by: John Spray <john.spray@redhat.com>
(cherry picked from commit b0b7b84da84d69951977045e8812419c19268756)

client: fix bug #24491 _ll_drop_pins may access invalid iterator
Fixes: https://tracker.ceph.com/issues/24491
Signed-off-by: Liu Yangkuan <liuyangkuan@gmail.com>
(cherry picked from commit 97188fef8596f098aa43aa745a217b40a6c29ea4)

Conflicts:
src/client/Client.cc: Resolved in _ll_drop_pins for ldout

Merge pull request #22746 from ceph/backport-luminous-22724

luminous: ceph-volume: error on commands that need ceph.conf to operate

Reviewed-by: Alfredo Deza <adeza@redhat.com>