mds: consider max age of dispatch queue when finding stale client

session renew messages from clients can be in the dispatch queue,
waiting for getting dispatched.

Signed-off-by: "Yan, Zheng" <zyan@redhat.com>

mds: reset heartbeat map at potential time-consuming places

Signed-off-by: Yan, Zheng <zyan@redhat.com>
Fixes: http://tracker.ceph.com/issues/26858

mds: change MDSRank::finished_queue to deque

Signed-off-by: "Yan, Zheng" <zyan@redhat.com>

Merge pull request #23372 from cbodley/wip-25214

qa/rgw: override valgrind --max-threads for radosgw

Reviewed-by: Matt Benjamin <mbenjamin@redhat.com>

qa/rgw: override valgrind --max-threads for radosgw

radosgw now uses 512 frontend threads by default, and valgrind won't
start with its default --max-threads=500

Fixes: http://tracker.ceph.com/issues/25214
Signed-off-by: Casey Bodley <cbodley@redhat.com>

Merge pull request #23145 from cbodley/wip-qa-rgw-admin-rest

qa/rgw: add radosgw-admin-rest task to singleton suite

Reviewed-by: Orit Wasserman <owasserm@redhat.com>

Merge PR #23195 into master

* refs/pull/23195/head:
mds: use vector for context lists
common: make C_ContextsBase container agnostic
common: make finish_contexts container agnostic
mds: add vector def for MDS contexts
mds: use compact map to manage waiting list

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

Merge PR #23353 into master

* refs/pull/23353/head:
doc: explain how to mkdir under .snap

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

doc: explain how to mkdir under .snap

Prior to this change, the CephFS snapshot instructions were not clear
whether users should "mkdir" the ".snap" directory or not.

Update the documentation to clarify that users should make a directory
under .snap with a name of their choice.

Thanks to John Spray <jspray@redhat.com> on ceph-users for this
information.

Signed-off-by: Ken Dreyer <kdreyer@redhat.com>

Merge pull request #23355 from alfredodeza/wip-rm25173

ceph-volume do not use stdin in luminous

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

Merge pull request #23366 from tchaikov/wip-cmake-no-common-crc-aarch64

cmake: do not link against common_crc_aarch64

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

cmake: do not link against common_crc_aarch64

it's included by libcrc32 already

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

Merge PR #23340 into master

* refs/pull/23340/head:
osd/PGLog.cc: use lgeneric_subdout instead of generic_dout

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

Merge pull request #23342 from ZVampirEM77/wip-sync-error-trim-param

rgw-admin: add "--trim-delay-ms" introduction for 'sync error trim'

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

Merge pull request #23360 from badone/wip-release-schedule-update-august-2018

doc/releases: Update releases to August '18

Reviewed-by: Greg Farnum <gfarnum@redhat.com>
Reviewed-by: Abhishek Lekshmanan <abhishek@suse.com>

ceph-volume tests verify that luminous doesn't use stdin

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

ceph-volume util do not use stdin for luminous

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

Merge pull request #23224 from votdev/rest_client_timeout

mgr/dashboard: Set timeout in RestClient calls

Reviewed-by: Lenz Grimmer <lgrimmer@suse.com>
Reviewed-by: Sebastian Wagner <swagner@suse.com>
Reviewed-by: Tatjana Dehler <tdehler@suse.com>

Merge pull request #21881 from sebastian-philipp/dashboard-pool-patch

mgr/dashboard: Add Pool update endpoint

Reviewed-by: Stephan Müller <smueller@suse.com>
Reviewed-by: Tatjana Dehler <tdehler@suse.com>

Merge pull request #23284 from tchaikov/wip-seastar-config

crimson/common: write configs synchronously on shard.0

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

doc/releases: Update releases to August '18

Mimic    13.2.1
Luminous 12.2.5, 12.2.6, 12.2.7
Jewel    10.2.11

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

rgw-admin: add "--trim-delay-ms" introduction for 'sync error trim'

Signed-off-by: Enming.Zhang <enming.zhang@umcloud.com>

Merge PR #22825 into master

* refs/pull/22825/head:
common: FreeBSD does not have /etc/os-release

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

Merge PR #22998 into master

* refs/pull/22998/head:
filestore: add pgid in filestore pg dir split log message

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

Merge PR #23134 into master

* refs/pull/23134/head:
common: check completion condition before waiting

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

Merge PR #23223 into master

* refs/pull/23223/head:
osd/PG: kill dead functions and related options
iosd/osd_type: kill unused input ec_pool for iterate_mayberw_back_to
common: kill dead options
osd/PG: do not initialize up/acting twice
osd/PG: clear missing_loc properly if last location is gone

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

Merge pull request #22990 from tchaikov/wip-cmake-link-static-libstdc++

cmake: fix "WITH_STATIC_LIBSTDCXX"

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

Merge PR #22692 into master

* refs/pull/22692/head:
doc/mgr/devicehealth: document devicehealth module
doc/rados/operations/health-checks: document DEVICE_HEALTH* messages
mgr/devicehealth: fix style for returns
mgr/devicehealth: use constants for health warnings
mgr/devicehealth: deal with as many daemons as we can until limit
mgr/devicehealth: warn if too many daemons are expected to fail soon
mgr/devicehealth: set primary-affinity 0 for failing devices
msg/devicehealth: fix config options
mgr/devicehealth: only fetch osdmap once from check_health
mgr/devicehealth: revise health messages
mgr/devicehealth: add 'device check-health' command and run periodically
mgr/devicehealth: fix new options
mgr/devicehealth: add helpers to life_expectancy_response()
mgr/devicehealth: simplify setting defaults
common/blkdev remove debug statements

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

Merge PR #23157 into master

* refs/pull/23157/head:
Provided API to change umask

Merge PR #23354 into master

* refs/pull/23354/head:
src/osd/PG.cc: remove redundant call to trim_log()

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

doc/mgr/devicehealth: document devicehealth module

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

doc/rados/operations/health-checks: document DEVICE_HEALTH* messages

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

mgr/devicehealth: fix style for returns

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

mgr/devicehealth: use constants for health warnings

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

mgr/devicehealth: deal with as many daemons as we can until limit

Process as many OSDs as we can until we hit the min_in_ratio.

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

mgr/devicehealth: warn if too many daemons are expected to fail soon

Refuse to mark out *all* OSDs.

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

mgr/devicehealth: set primary-affinity 0 for failing devices

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

msg/devicehealth: fix config options

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

mgr/devicehealth: only fetch osdmap once from check_health

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

mgr/devicehealth: revise health messages

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

mgr/devicehealth: add 'device check-health' command and run periodically

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

mgr/devicehealth: fix new options

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

mgr/devicehealth: add helpers to life_expectancy_response()

- if mark_out_threshold is met we write to log.warn instead of raising a
  health warning.
- check that OSD is 'in' before calling mark_out().
- raise a health warning in case OSD is marked 'out' but still has PGs
  attached to it.
- cast thresholds default values to string.
- add SCSI multipath support to health warning message.
- change health warning message.

Signed-off-by: Yaarit Hatuka <yaarithatuka@gmail.com>

mgr/devicehealth: simplify setting defaults

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

common/blkdev remove debug statements

Signed-off-by: Yaarit Hatuka yaarithatuka@gmail.com

Merge PR #23334 into master

* refs/pull/23334/head:
pybind/rados/rados: do not pass prval from stack

Reviewed-by: John Spray <john.spray@redhat.com>
Reviewed-by: Josh Durgin <jdurgin@redhat.com>

Merge pull request #23217 from dzafman/wip-25085

osd: Allow repair of an object with a bad data_digest in object_info on all replicas

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

src/osd/PG.cc: remove redundant call to trim_log()

This change is motived by the failure tracked in
https://tracker.ceph.com/issues/25198. The failure highlights a case, when a
call to trim_log() after the PG has recovered, races with the previous op,
on a replica OSD. Since the previous operation has not completed, the
last_complete value for that OSD is not valid, when we try to trim the
log. It is also worth noting that the race is due to MOSDPGTrim going through
the strict queue as a peering message vs regular ops going through the
non-strict queue.

During the investigation of this bug, we noticed that, with
https://tracker.ceph.com/issues/23979, we allow pg log trimming to
happen on the primary and replicas, whenever we cross the upper bound of
the pg log. This also ensures that pg log trimming happens while processing
any new op.

Therefore, the function trim_log(), which earlier served the purpose of
trimming logs on the primary and replicas, just before the PG went into
the Recovered state, is no more required. This acted like a last line of
defense to trim logs, when we did not need the logs any more. But, this call
seems redundant now, because, we are limiting the pg log length at all times.

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

pybind/rados/rados: do not pass prval from stack

The prval is a pointer to an int to write the final completion code of
the rados op.  This can't be on the stack since we immediately leave the
current scope after preparing the op (looong before we do the rados op).

We keep the tuple return value to avoid breaking users of this API
(devicehealth module, gnocchi at a minimum).

Fixes: http://tracker.ceph.com/issues/25175
Signed-off-by: Sage Weil <sage@redhat.com>

Merge pull request #23348 from ceph/wip-rm24957

ceph-volume: adds test for `ceph-volume lvm list /dev/sda`

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

ceph-volume: adds a unit test for `lvm list /dev/sda`

This test is to prove that the issue from
http://tracker.ceph.com/issues/24957 was fixed
by http://tracker.ceph.com/issues/24784

When running lvm list against a raw device it should handle
gracefully the situation where there are multiple PVs with the
name of the given device.

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

ceph-volume: move pvolumes fixture into conftest.py

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

Merge pull request #23336 from noahdesu/vstart-dashboard-no-rbd

vstart: disable dashboard when rbd not built

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

osd/PGLog.cc: use lgeneric_subdout instead of generic_dout

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

Merge PR #23297 into master

* refs/pull/23297/head:
ceph_volume_client: add delay for MDSMap to be distributed

Reviewed-by: Ramana Raja <rraja@redhat.com>

Merge PR #23308 into master

* refs/pull/23308/head:
doc: s/Ceph FS/CephFS

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

vstart: disable dashboard when rbd not built

dashboard doesn't load correctly without the rbd module, which means
vstart commands that interact with dashboard fail and vstart exits.

Signed-off-by: Noah Watkins <nwatkins@redhat.com>

Merge pull request #23292 from yuriw/wip-yuriw-25140-master

qa/tests: added 1st draft of mimic-x suite

Merge pull request #23302 from yuriw/wip-yuriw-crontab-master

qa/tests: added mimic-x to the schedule

Merge pull request #23321 from cernceph/dvanders_enable

ceph-volume: enable the ceph-osd during lvm activation

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

Merge pull request #23332 from alfredodeza/wip-rm25171

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

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

Merge pull request #23305 from smithfarm/wip-cleanup-upgrade

qa/upgrade: cleanup for nautilus

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

qa/tests: added 1st draft of mimic-x suite

Fixes: https://tracker.ceph.com/issues/25140
Signed-off-by: Yuri Weinstein <yweinste@redhat.com>

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

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

Merge pull request #23276 from tchaikov/wip-config-diff-lock

common/config: fix the lock in ConfigProxy::diff()

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

Merge pull request #23251 from neha-ojha/wip-25112

osd,mon: increase mon_max_pg_per_osd to 250

Reviewed-by: Josh Durgin <jdurgin@redhat.com>
Reviewed-by: Nathan Cutler <ncutler@suse.com>

Merge pull request #23249 from liewegas/wip-mon-cx-nautilus

osd/OSDMap: fix CEPHX_V2 osd requirement to nautilus, not mimic

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

Merge pull request #23229 from rjfd/wip-dashboard-query-params-bug

mgr/dashboard: fix query parameters in task annotated endpoints

Reviewed-by: Tiago Melo <tmelo@suse.com>

Merge PR #22607 into master

* refs/pull/22607/head:
common/options: convert many TYPE_[U]INT -> TYPE_SIZE
common/options: remove journal_max_corrupt_search

Reviewed-by: John Spray <john.spray@redhat.com>
Reviewed-by: Kefu Chai <kchai@redhat.com>

ceph-volume: enable ceph-osd during lvm activation

Enable the ceph-osd@<id> unit during lvm activate to link these
units to the ceph-osd.target.

Signed-off-by: Dan van der Ster <daniel.vanderster@cern.ch>
Fixes: http://tracker.ceph.com/issues/24152

ceph-volume: optional systemd enable --runtime

Allow units to be enabled but not persisted across a reboot,
and use this when enabling osds.

Signed-off-by: Dan van der Ster <daniel.vanderster@cern.ch>

Merge pull request #23287 from Devp00l/wip-duplicate-error-messages

mgr/dashboard: Fix duplicate error messages

Reviewed-by: Laura Paduano <lpaduano@suse.com>
Reviewed-by: Tiago Melo <tmelo@suse.com>
Reviewed-by: Volker Theile <vtheile@suse.com>

Merge pull request #22669 from votdev/feature_24574

mgr/dashboard: Cleanup RGW config checks

Reviewed-by: Lenz Grimmer <lgrimmer@suse.com>
Reviewed-by: Sebastian Wagner <swagner@suse.com>
Reviewed-by: Tatjana Dehler <tdehler@suse.com>

qa/upgrade: cleanup for nautilus

Drop unused suites, which ATM means all of them except upgrade/luminous-x
which recently got a cleanup in https://github.com/ceph/ceph/pull/23162

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

Merge pull request #23162 from smithfarm/wip-upgrade-cleanup

tests: upgrade/luminous-x: fix order of final-workload directory

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

cmake: do not link libs against ${ALLOC_LIBS}

let libcommon or libceph-common take care of this.

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

cmake,make-dist: build gperftools if WITH_STATIC_LIBSTDCXX

we could create a mini project to build a shared library, and use
try_compile() to test if the found gperftools is compiled with -fPIC.
but as we are targeting mostly xenial when enabling
WITH_STATIC_LIBSTDCXX, and google-perftools on xenial by default
is built without -fPIC. so let's keep it simple.

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

Provided API to change umask

Signed-off-by: Pavani Rajula <rpavani1998@gmail.com>

qa/tests: added mimic-x to the schedule

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

common/options: convert many TYPE_[U]INT -> TYPE_SIZE

Note that the _cost options are in fact in units of bytes.

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

cmake: link against libsnappy.a if WITH_STATIC_LIBSTDCXX

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

rocksdb: pickup change to link against libsnappy.a

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

cmake: fix "WITH_STATIC_LIBSTDCXX"

- do not link libkv with ALLOC_LIBS, it turns out that if we link
tcmalloc *before* -static-libstdc++ -static-libgcc, libstdc++ and gcc
libs will show up in `ldd` output
- add `-static-libstdc++ -static-libgcc` to CMAKE_SHARED_LINKER_FLAGS
and CMAKE_EXE_LINKER_FLAGS instead of adding them to all shared
libraries and executable. simpler this way.
- link against libtcmalloc statically, because libtcmalloc is a C++
library, linking against it dynamically and linking against C++ runtime
statically will pull in depdencies on two versions of C++ runtime, which
will bring down the app at run-time.
- do not pass '-pie' to linker when building executable if
`WITH_STATIC_LIBSTDCXX` and tcmalloc is used, because the static tcmalloc
is not compiled with PIC.
- only apply '-pie' if ENABLE_SHARED is enabled.

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

crimson/common: write configs synchronously on shard.0

to avoid potential racings on the same shard. before this change, we
apply the change in async. after this change, all changes happens on the
owner shard (i.e. shard.0), and the changes are applied synchronously.

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

doc: s/Ceph FS/CephFS

Fixes: https://github.com/ceph/ceph/pull/22784#discussion_r200755460
Signed-off-by: Jos Collin <jcollin@redhat.com>

crimson/common: throw if fails to change config

simpler this way, and this allows us to have more detailed error message
so we can present it to end-user. skipping the updating step if no
changes is made is nice to have, but changing settings is not in the
critical path. so let's keep it simple.

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

common/config: do not use magic number in set_mon_vals()

before this change, we compare the retcode of _set_val() with 1, and 0,
which are pratically magic numbers. after this change, we use
ConfigValues::set_value_result_t for non-error retcode.

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

Merge pull request #23043 from tchaikov/wip-python-cephfs-dependencies

deb,rpm: fix python-cephfs dependencies

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

ceph_volume_client: add delay for MDSMap to be distributed

Otherwise the setxattr will fail if the mds has not yet received the MDSMap
which adds the new data pool.

Fixes: https://tracker.ceph.com/issues/25141
Signed-off-by: Patrick Donnelly <pdonnell@redhat.com>

Merge pull request #23288 from theanalyst/doc/releases/13.2.1

doc: releases: mimic 13.2.1 release notes

Reviewed-By: Sage Weil <sweil@redhat.com>

Merge pull request #23278 from b-ranto/wip-volume-selinux

ceph-volume: Restore SELinux context

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

Merge pull request #23106 from noahdesu/always-on-mgr-modules2

mgr: create always on class of modules

Merge pull request #23143 from joke-lee/qa-admin-api-user-info-through-access-key

qa: add test for https://github.com/ceph/ceph/pull/22790

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

Merge pull request #23289 from alfredodeza/wip-rm24993-2

ceph-volume ensure encoded bytes are always used

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

qa/rgw: fix 'policy show' tests in radosgw_admin_rest task

'policy show' returns a json-encoded representation of
RGWAccessControlPolicy, while key.get_xml_acl() returns
RGWAccessControlPolicy_S3 encoded as xml. so even with '&format=xml',
the strings won't match

Signed-off-by: Casey Bodley <cbodley@redhat.com>

qa/rgw: add bucket-id param to 'bucket link' admin api test

Signed-off-by: Casey Bodley <cbodley@redhat.com>

qa/rgw: radosgw_admin_rest task tolerates empty responses

result.json() throws a 'JSONDecodeError: Expecting value: line 1 column 1'
for requests that return no body, such as 'user rm' 'key rm' 'subuser
rm', 'bucket unlink', etc

Signed-off-by: Casey Bodley <cbodley@redhat.com>

qa/rgw: radosgw_admin_rest task uses rgw.role_endpoints for connections

Signed-off-by: Casey Bodley <cbodley@redhat.com>

qa/rgw: radosgw_admin_rest task uses util.rgw.rgwadmin()

Signed-off-by: Casey Bodley <cbodley@redhat.com>

qa/rgw: add radosgw-admin-rest task to singleton suite

Signed-off-by: Casey Bodley <cbodley@redhat.com>