init-ceph: check if /lib/lsb/init-functions exists

On OSX/FreeBSD, /lib/lsb/init-functions does not exist

Signed-off-by: Yan, Zheng <zyan@redhat.com>
(cherry picked from commit bb1fa7f3993dfcd1243b2a9a0bb432c269df215f)

Merge pull request #7656 from dzafman/wip-14769-hammer

ceph.in: Notify user that 'tell' can't be used in interactive mode

Reviewed-by: Loic Dachary <ldachary@redhat.com>
Reviewed-by: Ken Dreyer <kdreyer@redhat.com>

Merge pull request #7648 from tchaikov/wip-13783-hammer

mon/LogMonitor: use the configured facility if log to syslog

Reviewed-by: Loic Dachary <ldachary@redhat.com>

Merge pull request #7801 from dachary/wip-14863-hammer

hammer: unittest_crypto: benchmark 100,000 CryptoKey::encrypt() calls

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

unittest_crypto: benchmark 100,000 CryptoKey::encrypt() calls

Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit fb4b6c52d8e31e8dedfa8aecc16c389f0e7d86cf)

Conflicts:
src/test/crypto.cc : complements the incorrect cherry-pick
           df3f971eafda9c54881c13dcf47f996f18e17028 see
           http://tracker.ceph.com/issues/14863 for more information

Merge pull request #7185 from dachary/wip-14325-hammer

hammer: rgw: radosgw-admin bucket check --fix not work

Reviewed-by: Yehuda Sadeh <ysadehwe@redhat.com>

Merge pull request #7188 from dachary/wip-13799-hammer

hammer: rgw: default quota params

Reviewed-by: Yehuda Sadeh <ysadehwe@redhat.com>

Merge pull request #7414 from dachary/wip-14493-hammer

hammer: make rgw_fronends more forgiving of whitespace

Reviewed-by: Yehuda Sadeh <ysadehwe@redhat.com>

Merge pull request #7415 from dachary/wip-14554-hammer

hammer: Unable to bring up OSD's after dealing with FULL cluster (OSD assert with /include/interval_set.h: 386: FAILED assert(_size >= 0))

Merge pull request #7576 from dachary/wip-14579-hammer

hammer: mon sync does not copy config-key

Merge pull request #7577 from dachary/wip-14689-hammer

hammer: consume_maps clearing of waiting_for_pg needs to check the spg_t shard for acting set membership

Merge pull request #7590 from SUSE/wip-14720-hammer

hammer: OSDMap: reset osd_primary_affinity shared_ptr when deepish_copy_from

Merge pull request #7645 from dachary/wip-14760-hammer

hammer: OSD coredumps with leveldb compact on mount = true

Merge pull request #7485 from dillaman/wip-14611-hammer

librbd: flattening an rbd image with active IO can lead to hang

Reviewed-by: Loic Dachary <ldachary@redhat.com>

Merge pull request #7488 from ceph/wip-auth-hammer

hammer: auth: use libnss more safely

Reviewed-by: Loic Dachary <ldachary@redhat.com>

Merge pull request #7475 from dachary/wip-14602-hammer

hammer: ceph-disk: use blkid instead of sgdisk -i

Reviewed-by: Loic Dachary <ldachary@redhat.com>

Merge pull request #7456 from dachary/wip-14587-hammer

hammer: qa/workunits/post-file.sh: sudo

Reviewed-by: Loic Dachary <ldachary@redhat.com>

Merge pull request #6946 from dachary/wip-14087-shard-size

osd: log inconsistent shard sizes

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

Merge pull request #7110 from david-z/wip-hammer-fix-logrotate-ceph-fuse

[ceph-fuse] fix ceph-fuse writing to stale log file after log rotation

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

0.94.6

ceph.in: Notify user that 'tell' can't be used in interactive mode

Signed-off-by: David Zafman <dzafman@redhat.com>
(cherry picked from commit a341d977d42a3010267b64e15cebae3085507297)

Conflicts:
src/ceph.in (trivial resolution)

mon/LogMonitor: use the configured facility if log to syslog

Fixes: #13748
Signed-off-by: Kefu Chai <kchai@redhat.com>
(cherry picked from commit 4d2ca37d3b8f067d0cf0f7cfea24964ac181115d)

os/LevelDBStore:fix bug when compact_on_mount

Initialize the performance counter(l_rocksdb_compact)
before use it in compact().

Signed-off-by: Xiaoxi Chen <xiaoxi.chen@intel.com>
(cherry picked from commit 932d59d084090297c89cfa3ff7df0a952f807bab)

OSDMap: reset osd_primary_affinity shared_ptr when deepish_copy_from

Base on current logic, OSDMonitor may call create_pending and
encode_pending twice for the some epoch.

In encode_pending:

tmp.deepish_copy_from(osdmap);
tmp.apply_incremental(pending_inc);

This Op would change the tmp osd_primary_affinity, but the osd_primary_affinity
is declared as ceph::shared_ptr, so this would change the osdmap too. When this
round encode_pending is proposed fail. We may call encode_pending again, but the
osdmap is changed last round, so the pending_inc would be wrong.

Fixes: #14686
Signed-off-by: Xinze Chi <xinze@xsky.com>
(cherry picked from commit 990b437f4e616a87f4f7438e51945d531170ca83)

OSD::consume_map: correctly remove pg shards which are no longer acting

A particular shard may be non-active even if the osd is in the acting
set.

Fixes: 14278
Backport: hammer
Signed-off-by: Samuel Just <sjust@redhat.com>
(cherry picked from commit bd2fdf8c56c1aeed6cea3eb59046dbb6af470111)

mon: add mon_config_key prefix when sync full

Signed-off-by: Xiaowei Chen <chen.xiaowei@h3c.com>
(cherry picked from commit b77cd31b161e78467bbb6476776c0f1d43e615cf)

Merge pull request #7501 from SUSE/wip-14624-hammer

hammer: fsx failed to compile

Reviewed-by: Loic Dachary <ldachary@redhat.com>

Merge pull request #7524 from ktdreyer/wip-14637-hammer-man-radosgw-admin-orphans

hammer: doc: regenerate man pages, add orphans commands to radosgw-admin(8)

Merge pull request #7526 from ceph/wip-14516-hammer

hammer: rgw-admin: document orphans commands in usage

Merge pull request #7441 from odivlad/backport-pr-14569

[backport] hammer: rgw: Make RGW_MAX_PUT_SIZE configurable

Reviewed-by: Yehuda Sadeh <ysadehwe@redhat.com>

Merge pull request #7442 from odivlad/backport-pr-14570

[backport] rgw: fix wrong etag calculation during POST on S3 bucket.

Reviewed-by: Yehuda Sadeh <ysadehwe@redhat.com>

rgw-admin: document orphans commands in usage

Fixes: #14516
Signed-off-by: Yehuda Sadeh <yehuda@redhat.com>
(cherry picked from commit 105a76bf542e05b739d5a03ca8ae55432350f107)

Conflicts:
src/rgw/rgw_admin.cc (trivial resolution)

[backport] rgw: fix wrong etag calculation during POST on S3 bucket.

Closes: #14570
(cherry picked from commit 742906a)

Signed-off-by: Radoslaw Zarzynski <rzarzynski@mirantis.com>
Signed-off-by: Vladislav Odintsov <odivlad@gmail.com>
Tested-by: Vladislav Odintsov <odivlad@gmail.com>

[backport] rgw: Make RGW_MAX_PUT_SIZE configurable

The 5GB limit of a single operation uploading was part of S3 spec.
However some private setups may have some special requirements
on this limit. It's more convinent to have a configurable value.

Closes: http://tracker.ceph.com/issues/14569
(cherry picked from commit df97f28)

Signed-off-by: Yuan Zhou <yuan.zhou@intel.com>
Signed-off-by: Vladislav Odintsov <odivlad@gmail.com>
Tested-by: Vladislav Odintsov <odivlad@gmail.com>

doc: add orphans commands to radosgw-admin(8)

The --help text was added in 105a76bf542e05b739d5a03ca8ae55432350f107.
Add the corresponding entries to the man page.

Fixes: #14637
Signed-off-by: Ken Dreyer <kdreyer@redhat.com>
(cherry picked from commit ec162f068b40f594c321df5caa9fe2541551b89e)
  Cherry-pick to hammer includes nroff source change (in master the
  nroff sources are no longer present in Git.)

man: rebuild manpages

following the procedure in admin/manpage-howto.txt.

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

fsx: checkout old version until it compiles properly on miras

I sent a patch to xfstests upstream at
http://article.gmane.org/gmane.comp.file-systems.fstests/1665, but
until that's fixed we need a version that works in our test lab.

Signed-off-by: Greg Farnum <gfarnum@redhat.com>
(cherry picked from commit 7d52372ae74878ebd001036ff0a7aad525eb15b6)

Merge pull request #7454 from dachary/wip-14584-hammer

hammer: fsstress.sh fails

Reviewed-by: Loic Dachary <ldachary@redhat.com>

Merge pull request #6918 from asheplyakov/hammer-bug-12449

osd: check for full before changing the cached obc (hammer)

Reviewed-by: Loic Dachary <ldachary@redhat.com>

qa/workunits/post-file.sh: sudo

Only root can read the key in /usr/share/...

Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit 82b0243ec318457e376288a6b32487fbea059705)

qa/workunits/post-file: pick a dir that's readable by world

Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit 04b00029e16971abf1937facbbff8da00d6a3452)

qa/workunits/post-file.sh: use /etc/default

base-files doesn't existing on rhel.  /etc/default is everywhere and it's
small.

Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit 52235e32561ef3e91f07c59c8fd89c4df244f461)

librbd: ensure librados callbacks are flushed prior to destroying image

Fixes: #14092
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit 98157ab3274bd960e4487e34f5a83e9c921a6ac8)

librbd: simplify IO flush handling

Add a new convenience method to ImageCtx for handling flush
requests and cleanup flush handling with dealing with the cache.

Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit e8d8096babaf15fe8af717297d512f580aff6c18)

WorkQueue: PointerWQ drain no longer waits for other queues

If another (independent) queue was processing, drain could
block waiting.  Instead, allow drain to exit quickly if
no items are being processed and the queue is empty for
the current WQ.

Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit b118d7df1e34387b6e5649a5b205cf061598d0d4)

test: new librbd flatten test case

AIO operations after a flatten operation were previously
hanging during the close of the parent image.

Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit b59158d9f0c3dbe4b47be1dfd27aaa5ad92e6063)

ceph-disk: use blkid instead of sgdisk -i

sgdisk -i 1 /dev/vdb opens /dev/vdb in write mode which indirectly
triggers a BLKRRPART ioctl from udev (starting version 214 and up) when
the device is closed (see below for the udev release note). The
implementation of this ioctl by the kernel (even old kernels) removes
all partitions and adds them again (similar to what partprobe does
explicitly).

The side effects of partitions disappearing while ceph-disk is running
are devastating.

sgdisk is replaced by blkid which only opens the device in read mode and
will not trigger this unexpected behavior.

The problem does not show on Ubuntu 14.04 because it is running udev <
214 but shows on CentOS 7 which is running udev > 214.

git clone git://anonscm.debian.org/pkg-systemd/systemd.git
systemd/NEWS:
CHANGES WITH 214:

        * As an experimental feature, udev now tries to lock the
          disk device node (flock(LOCK_SH|LOCK_NB)) while it
          executes events for the disk or any of its partitions.
          Applications like partitioning programs can lock the
          disk device node (flock(LOCK_EX)) and claim temporary
          device ownership that way; udev will entirely skip all event
          handling for this disk and its partitions. If the disk
          was opened for writing, the close will trigger a partition
          table rescan in udev's "watch" facility, and if needed
          synthesize "change" events for the disk and all its partitions.
          This is now unconditionally enabled, and if it turns out to
          cause major problems, we might turn it on only for specific
          devices, or might need to disable it entirely. Device Mapper
          devices are excluded from this logic.

http://tracker.ceph.com/issues/14094 Fixes: #14094

Signed-off-by: Ilya Dryomov <idryomov@redhat.com>
Signed-off-by: Loic Dachary <loic@dachary.org>
(cherry picked from commit 9dce05a8cdfc564c5162885bbb67a04ad7b95c5a)

Conflicts:
src/ceph-disk: keep get_partition_type as it is because
        some hammer users may rely on the fact that it is able
        to fallback to sgdisk if blkid is old. Chances are an
        old blkid also means an old udev that does not have the
        problem this fix is adressing. The get_partition_uuid
        is modified to try blkid first, with the same rationale.

qa/fsstress.sh: fix 'cp not writing through dangling symlink'

On some test machines, /usr/lib/ltp/testcases/bin/fsstress is
dangling symlink. 'cp -f' is impotent in this case.

Fixes: #12710
Signed-off-by: Yan, Zheng <zyan@redhat.com>
(cherry picked from commit 479f2a760baf6af4714d3391a366a6b3acf1bcaf)

[ceph-fuse] fix ceph-fuse writing to stale log file after log rotation

This fix should be applied to hammer branch. It can't be directly applied to master branch, because logrotate.conf is changed on matser since ceph-osd, ceph-mon, etc, is controlled by systemd with user/group as 'ceph' by default, while ceph-fuse might be started as root privilege by external users.

Signed-off-by: Zhi Zhang <zhangz.david@outlook.com>

Merge pull request #7236 from athanatos/wip-14376

config_opts: increase suicide timeout to 300 to match recovery

http://pulpito.ceph.com/sage-2016-01-30_09:58:32-rados-wip-sage-testing-hammer---basic-mira/

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

Merge pull request #6450 from dachary/wip-13672-hammer

tests: testprofile must be removed before it is re-created

http://pulpito.ceph.com/sage-2016-01-30_09:58:32-rados-wip-sage-testing-hammer---basic-mira/

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

Merge pull request #6680 from SUSE/wip-13859-hammer

hammer: ceph.spec.in License line does not reflect COPYING

http://pulpito.ceph.com/sage-2016-01-30_09:58:32-rados-wip-sage-testing-hammer---basic-mira/

Merge pull request #6791 from branch-predictor/bp-5812-backport

Race condition in rados bench

http://pulpito.ceph.com/sage-2016-01-30_09:58:32-rados-wip-sage-testing-hammer---basic-mira/

Merge pull request #6973 from dreamhost/wip-configure-hammer

configure.ac: no use to add "+" before ac_ext=c

http://pulpito.ceph.com/sage-2016-01-30_09:58:32-rados-wip-sage-testing-hammer---basic-mira/

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

Merge pull request #7206 from dzafman/wip-14292

osd/PG.cc: 3837: FAILED assert(0 == "Running incompatible OSD")

http://pulpito.ceph.com/sage-2016-01-30_09:58:32-rados-wip-sage-testing-hammer---basic-mira/

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

Merge pull request #7207 from rldleblanc/recency_fix_for_hammer

hammer: osd/ReplicatedPG: Recency fix for Hammer

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

Merge pull request #7347 from tchaikov/wip-hammer-10093

tools: ceph-monstore-tool must do out_store.close()

http://pulpito.ceph.com/sage-2016-01-30_09:58:32-rados-wip-sage-testing-hammer---basic-mira/

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

Merge pull request #7411 from dachary/wip-14467-hammer

hammer: disable filestore_xfs_extsize by default

http://pulpito.ceph.com/sage-2016-01-30_09:58:32-rados-wip-sage-testing-hammer---basic-mira/

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

Merge pull request #7412 from dachary/wip-14470-hammer

tools: tool for artificially inflate the leveldb of the mon store for testing purposes

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

Merge pull request #7446 from liewegas/wip-14537-hammer

mon: compact full epochs also

http://pulpito.ceph.com/sage-2016-01-30_09:58:32-rados-wip-sage-testing-hammer---basic-mira/

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

Merge pull request #7182 from dachary/wip-14143-hammer

hammer: Verify self-managed snapshot functionality on image create

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

Merge pull request #7183 from dachary/wip-14283-hammer

hammer: rbd: fix bench-write

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

Merge pull request #7416 from dachary/wip-14466-hammer

hammer: rbd-replay does not check for EOF and goes to endless loop

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

Merge pull request #7417 from dachary/wip-14553-hammer

hammer: rbd: TaskFinisher::cancel should remove event from SafeTimer

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

Merge pull request #7407 from dillaman/wip-14543-hammer

librbd: ImageWatcher shouldn't block the notification thread

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

Merge pull request #6980 from dillaman/wip-14063-hammer

librbd: fix merge-diff for >2GB diff-files

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

mon: compact full epochs also

by compacting the ${prefix}.${start}..${prefix}..${end} does not
necessary compact the range of ${prefix}."full_"${start}..
${prefix}."full_"${end}. so when more and more epochs get trimmed
with out a full range compaction, the size of monitor store could
be very large.

Fixes: #14537
Signed-off-by: Kefu Chai <kchai@redhat.com>
(cherry picked from commit 93d633a25ad8c36c972bb766c38187e2612041e1)

Check for full before changing the cached obc

ReplicatedPG::prepare_transaction(): check if the pool is full before
updating the cached ObjectContext to avoid the discrepancy between
the cached and the actual object size (and other metadata).
While at it improve the check itself: consider cluster full flag,
not just the pool full flag, also consider object count changes too,
not just bytes.

Based on commit a1eb380c3d5254f9f1fe34b4629e51d77fe010c1

Fixes: #13335
Signed-off-by: Alexey Sheplyakov <asheplyakov@mirantis.com>

Merge pull request #6353 from theanalyst/wip-13513-hammer

rgw: value of Swift API's X-Object-Manifest header is not url_decoded during segment look up

Reviewed-by: Yehuda Sadeh <ysadehwe@redhat.com>

Merge pull request #6620 from SUSE/wip-13820-hammer

hammer: rgw: Setting ACL on Object removes ETag

Reviewed-by: Yehuda Sadeh <ysadehwe@redhat.com>

Merge pull request #7186 from dachary/wip-13888-hammer

hammer: rgw: orphans finish segfaults

Reviewed-by: Yehuda Sadeh <ysadehwe@redhat.com>

osd: recency should look at newest (not oldest) hitsets

Reported-by: xinxin shu <xinxin.shu@intel.com>
Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit 528eae92d010ea34ae8cf0e8b2290aaa5e058d24)

Conflicts:
src/osd/ReplicatedPG.cc
                    code section was moved to ReplicatedPG::maybe_promote
                    in master.
Signed-off-by: Robert LeBlanc <robert.leblanc@endurance.com>

osd/ReplicatedPG: fix promotion recency logic

Recency is defined as how many of the last N hitsets an object
must appear in in order to be promoted.  The previous logic did
nothing of the sort... it checked for the object in any one of
the last N hitsets, which led to way to many promotions and killed
any chance of the cache performing properly.

While we are here, we can simplify the code to drop the max_in_*
fields (no longer necessary).

Note that we may still want a notion of 'temperature' that does
tolerate the object missing in one of the recent hitsets.. but
that would be different than recency, and should probably be
modeled after the eviction temperature model.

Backport: infernalis, hammer
Reported-by: Nick Fisk <nick@fisk.me.uk>
Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit 180c8743addc5ae2f1db9c58cd2996ca6e7ac18b)

Conflicts:
src/osd/ReplicatedPG.cc
                    code section was moved to ReplicatedPG::maybe_promote
                    in master.
Signed-off-by: Robert LeBlanc <robert.leblanc@endurance.com>

Merge pull request #5789 from SUSE/wip-12928-hammer

rpm: libcephfs_jni1 has no %post and %postun

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

Merge pull request #7434 from tchaikov/wip-14441-hammer

man: document listwatchers cmd in "rados" manpage

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

man: document listwatchers cmd in "rados" manpage

Signed-off-by: Kefu Chai <kchai@redhat.com>
(cherry picked from commit c2e391301efc43f0b431e89737246b2c43bf10a9)

man/rados.8: also added the rendered man.8 man page, as we don't
             put the generated man pages in master anymore. but
             they are still in the hammer's source repo.

rbd: remove canceled tasks from timer thread

When canceling scheduled tasks using the timer thread, TaskFinisher::cancel
does not call SafeTimer::cancel_event, so events fire anyway. Add this call.

Fixes: #14476
Signed-off-by: Douglas Fuller <dfuller@redhat.com>
(cherry picked from commit 2aa0f318c862dbe3027d74d345671506605778eb)

rbd-replay: handle EOF gracefully

Fixes: #14452
Signed-off-by: Mykola Golub <mgolub@mirantis.com>
(cherry picked from commit c59b84c3e2c9bbda68219e4d2288a889dd9ca6cb)

Conflicts:
src/rbd_replay/BufferReader.cc: trivial resolution
        caused by s/CEPH_PAGE_SIZE/CEPH_BUFFER_APPEND_SIZE/
        in the context of one hunk.

PG::activate(): handle unexpected cached_removed_snaps more gracefully

PGPool::update(): ditto

Fixes: #14428
Backport: infernalis, hammer, firefly

Signed-off-by: Alexey Sheplyakov <asheplyakov@mirantis.com>
(cherry picked from commit aba6746b850e9397ff25570f08d0af8847a7162c)

rgw: warn on suspicious civetweb frontend parameters

Warns about incorrect spacing near 'port' (e.g., 'port = 80').

Fixes issue #12038.

Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
(cherry picked from commit 93497e576a946ed7e59f4a4299e447b65aa1774b)

tools: monstore: add 'show-versions' command.

Using this tool, the first/last committed version of maps will be shown.

Signed-off-by: Cilang Zhao <zhao.cilang@h3c.com>
(cherry picked from commit 21e6ba0c18428caff45733e6b43d197be38af8bb)

tools: ceph_monstore_tool: add inflate-pgmap command

this command repeatly add the latest pgmap to the monstore in order
to inflate it. the command helps with the testing of some monstore
related performance issue of monitor

Fixes: #14217
Signed-off-by: Kefu Chai <kchai@redhat.com>
(cherry picked from commit b0f6af814a58dd374ca95e84a4e381f8ef1368de)

tools:support printing the crushmap in readable fashion.

Signed-off-by: Bo Cai <cai.bo@h3c.com>
(cherry picked from commit b7faf678ed0241abc7eb86b9baaa6db669a22d53)

tools:print the map infomation in human readable format.

Signed-off-by: Bo Cai <cai.bo@h3c.com>
(cherry picked from commit 5ed8cdc19150382c946a373ec940d76f98e6ecb7)

tools:remove the local file when get map failed.

Signed-off-by: Bo Cai <cai.bo@h3c.com>
(cherry picked from commit 0b03b32d8ba76fe9f6f1158e68eb440e3670393a)

tools: ceph_monstore_tool: describe behavior of rewrite command

Signed-off-by: Joao Eduardo Luis <joao@suse.de>
(cherry picked from commit a881f9385feb0f5a61fa22357984d6f291c08177)

tools/ceph-monstore-tools: add rewrite command

"rewrite" command will
 - add a new osdmap version to update current osdmap held by OSDMonitor
 - add a new paxos version, as a proposal it will
   * rewrite all osdmap epochs from specified epoch to  the last_committed
     one with the specified crush map.
   * add the new osdmap which is added just now
so the leader monitor can trigger a recovery process to apply the transaction
to all monitors in quorum, and hence bring them back to normal after being
injected with a faulty crushmap.

Fixes: #11815
Signed-off-by: Kefu Chai <kchai@redhat.com>
(cherry picked from commit 306373427836ca0c2418dbe6caab26d74d94d12e)

tools: ceph-monstore-tool must do out_store.close()

this change fixes the "store-copy" command.
Like the bug reported in http://tracker.ceph.com/issues/10093.

Signed-off-by: huangjun <hjwsm1989@gmail.com>
(cherry picked from commit d85e0f8c50fce62be012506f7ffcb1cdeb0c819c)

osd: disable filestore_xfs_extsize by default

This option involves a tradeoff: When disabled, fragmentation is worse,
but large sequential writes are faster. When enabled, large sequential
writes are slower, but fragmentation is reduced.

Fixes: #14397
(cherry picked from commit aed85775bf53c273786ce4999320134822722af5)

Merge pull request #7316 from ceph/wip-deb-lttng-hammer

deb: strip tracepoint libraries from Wheezy/Precise builds

All other "modern" Debian-based OSes have a functional LTTng-UST. Since only hammer needs to build on these older distros, this fix only affects the deb building process for those two releases(since autoconf detects that LTTng is broken).

Reviewed-by: Loic Dachary <ldachary@redhat.com>

Merge pull request #7187 from dachary/wip-13831-hammer

hammer: init script reload doesn't work on EL7

Reviewed-by: Loic Dachary <ldachary@redhat.com>

librbd: ImageWatcher shouldn't block the notification thread

Blocking the notification thread will also result in librados async
callbacks becoming blocked (since they use the same thread).

Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit 7e2019a72733dff43e55c9b22df12939d584f87d)

 Conflicts:
src/librbd/ImageWatcher.[cc|h]: fewer RPC messages

librados_test_stub: watch/notify now behaves similar to librados

Notifications are executed via the same librados AIO callback
thread, so it's now possible to catch deadlock.

Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit 43722571838a2d78ca8583b5a1ea381cd988de0b)

tests: simulate writeback flush during snap create

Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit e9570fcf4be9cd5c06937769d074fa52ffb32e4b)

Merge pull request #7365 from liewegas/wip-tcmalloc-hammer

osd: pass tcmalloc env through to ceph-osd

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

upstart/ceph-osd.conf: pass TCMALLOC_MAX_TOTAL_THREAD_CACHE_BYTES through

Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit a95a3d34cccb9f9f0782a96ca05e5fe9c2b01772)

init-ceph: pass TCMALLOC_MAX_TOTAL_THREAD_CACHE_BYTES through

..when set in the sysconfig/default file.

Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit 03e01ae263ef207046933890821ae75d5e60d9b8)

[source default/sysconfig file too]

tools: ceph-monstore-tool must do out_store.close()

this change fixes the "store-copy" command.
Like the bug reported in http://tracker.ceph.com/issues/10093.

Signed-off-by: huangjun <hjwsm1989@gmail.com>
(cherry picked from commit d85e0f8c50fce62be012506f7ffcb1cdeb0c819c)

deb: strip tracepoint libraries from Wheezy/Precise builds

These releases do not (by default) have a sane LTTng-UST environment,
which results in autoconf disabling support for tracing.

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

Merge pull request #7252 from ceph/wip-13483-hammer

qa: remove legacy OS support from rbd/qemu-iotests

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