admin/build-doc: depend on zlib1g-dev and graphviz

The docs currently require zlib1g-dev (on Ubuntu) and graphviz (on RHEL
and Ubuntu) in order to build.

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

Merge pull request #7328 from ceph/wip-mtime-bugfix

Bug fix for mtime anomalies in RadosGW and other places

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

Merge pull request #7453 from ceph/wip-14566

msg/async: fix potential race condition

Merge pull request #7505 from xiexingguo/xxg-wip-kstorelru

os/kstore: insert new onode to the front position of onode LRU

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

Merge pull request #7492 from jjhuo/master

os/bluestore: insert new onode to the front position of onode LRU

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

Merge branch 'xxg-wip-14474' of git://github.com/xiexingguo/ceph

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

Merge pull request #7380 from yuyuyu101/async-nip

msg/async: reduce extra tcp packet for message ack

Merge pull request #7398 from theanalyst/fix/12744

mon: PG Monitor should report waiting for backfill

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

Merge pull request #7447 from liewegas/wip-parallel-rados-test

test: run rados api tests in parallel

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

Merge pull request #7455 from yuyuyu101/bufferlist-move

buffer: use move construct to append/push_back/push_front

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

Merge pull request #7459 from dwj192/mon-created

mon: fix monmap creation stamp

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

Merge pull request #7477 from xiexingguo/wip-fix-obj

osdc/objecter: fix race condition and result code overflow

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

Merge pull request #7482 from branch-predictor/bp-faster-health

mon: reduce CPU and memory manager pressure of pg health check

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

Merge pull request #7518 from ceph/wip-14637-man-radosgw-admin-orphans

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

Merge pull request #6268 from dachary/wip-13453-recovery-counters

recovery counters

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

Merge pull request #7507 from tchaikov/wip-cmake

cmake: add KernelDevice.cc to libos_srcs

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

cmake: add KernelDevice.cc to libos_srcs

which fixes the build

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

Merge remote-tracking branch 'gh/jewel'

Merge branch 'wip-rados-bench'

rados: Add units to rados bench output

Fixes: #12248
Signed-off-by: Gu Zhongyan <guzhongyan@360.com>
Reviewed-by: Sage Weil <sage@redhat.com>

Merge pull request #7386 from liewegas/wip-features

osd: skip redirect encoding if we can

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

os/kstore: insert new onode to the front position of onode LRU

See https://github.com/jjhuo/ceph/commit/caed88264d77f860376d4213cc98309a5daf3af2 for detail information.

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

Merge branch 'weighted-priority-queue' of git://github.com/rldleblanc/ceph

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

Merge branch 'gelf-logging-backend' of git://github.com/irq0/ceph

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

scripts/run-coverity: fix upload process

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

Merge pull request #7357 from zhengbinh3c/master

osd: fix memory leak in ReplicatedPG.cc error path

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

Merge pull request #7362 from xiexingguo/xxg-wip-fixbs

osd: BlueStore: fix fsck and blockdevice read-relevant issue

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

Merge pull request #7468 from liewegas/wip-test-fuse

qa/workunits/objectstore/test_fuse.sh: fix root check

Merge pull request #7476 from tchaikov/wip-include-expat-in-source

rgw: do not include "expat.h" in header file

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

Merge pull request #7479 from roidayan/fix_xio_compilation

msg/xio: fix compilation

test_bufferlist: add move tests for bufferlist

Signed-off-by: Haomai Wang <haomai@xsky.com>

os/bluestore: insert new onode to the front position of onode LRU

When new added onode comes, it should go to the front of the LRU,
since it's least recently used.

Signed-off-by: Jianjian Huo <samuel.huo@gmail.com>

bluestore/bluefs_types: fix imcomplete output message

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

BlueFS: add a guard for log space allocation during mkfs

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

BlueFS: fix race condition

Fixes: #14480
Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>

BlueFS: fix typo

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

BlueFS: remove unused local variable

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

BlueFS: remove redundant intermediate buffer pad

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

BlueFS: stop alloc if unable to replay

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

BlueFS: fix unprecise calculation of sync interval

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

BlueFS: add log after updating prefer_bdev field.

Fixes: #14474
Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>

qa/workunits/rados/test.sh: bash

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

buffer: add operator= for ptr/bufferlist rvalue

Signed-off-by: Haomai Wang <haomai@xsky.com>

misc: using move construct to avoid extra atomic inc/dec

Signed-off-by: Haomai Wang <haomai@xsky.com>

buffer: add ptr move construct

Signed-off-by: Haomai Wang <haomai@xsky.com>

ceph_test_objectstore: fix warning

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

os/bluestore/KernelDevice: explicit on ctor

This was lost in the wip-spdk merge.

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

doc: update PG doc with the backfill change

Since the state reported is `backfilling` and `backfill_wait`

Signed-off-by: Abhishek Lekshmanan <abhishek@suse.com>

mon: PG Monitor should report backfill_wait

Instead of backfill as it is actually waiting for backfill instead of
actually backfilling

Fixes: #12744
Reported-by: Dan Mick <dmick@redhat.com>
Signed-off-by: Abhishek Lekshmanan <abhishek@suse.com>

Merge pull request #7467 from liewegas/wip-scrub-start

osd: fix scrub start hobject

Merge branch 'wip-spdk' of git://github.com/yuyuyu101/ceph

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

Merge remote-tracking branch 'me/wip-lockdep'

Merge pull request #7438 from ifed01/wip-pgbackend-cleanup

osd: cleanup OSD/PGBackend interface a bit

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

Merge pull request #7480 from trociny/fix-journal-commit_position_task

journal: reset commit_position_task_ctx pointer after task complete

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

mon: reduce CPU and memory manager pressure of pg health check

When we don't need details about which PGs are stuck, we can just iterate
once over pg_stat map and increase appropriate counters, no need to
repeatedly spam memory manager with stuck_pgs map population (we'll only
use its size). Even if we need details, first check if we actually have any
PGs in any of requested states, so in most cases we won't iterate needlessly
five times over all PG stat map.

This at least halves the time needed by pg health gathering (~3ms,
down from ~6-7ms) on cluster with 31832 PGs and Intel Xeon E5-2640
CPU.

Signed-off-by: Piotr Dałek <piotr.dalek@ts.fujitsu.com>

script/run-coverity: fix user

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

journal: reset commit_position_task_ctx pointer after task complete

Otherwise, schedule_commit_task() would never schedule the next task
and the commit position would not be committed until shutdown.

Signed-off-by: Mykola Golub <mgolub@mirantis.com>

Merge remote-tracking branch 'gh/wip-da-SCA-20150910'

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

xio: fix compilation

Signed-off-by: Roi Dayan <roid@mellanox.com>

osdc/objecter: update crush location in a safer way

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

test/librados: fix hash position overflow issue

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

rados: fix result code overflow

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

Merge pull request #7473 from yehudasa/wip-14516

rgw-admin: document orphans commands in usage

osdc/objecter: fix race condition

Fixes: #14603
Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>

Merge pull request #7451 from wjin/fix

msg/async: fix array boundary

Reviewed-by: Haomai Wang <haomai@xsky.com>
Reviewed-by: Kefu Chai <kchai@redhat.com>

rgw: do not include "expat.h" in heade file

Building Ceph without the Expat development library currently results in
the following build error:

~/ceph> ./configure --without-fuse --without-tcmalloc --without-radosgw
...
~/ceph> make
...
CXX ceph_dencoder-rgw_dencoder.o
In file included from rgw/rgw_dencoder.cc:6:0:
rgw/rgw_acl_s3.h:9:19: fatal error: expat.h: No such file or directory

The ceph-dencoder binary is built with rgw_dencoder.cc, which includes
rgw_acl_s3.h, which in turn includes expat.h. However, configure.ac
currently only checks for Expat as part of radosgw.

but we don't need expat.h for testing the encoding of rgw structs, so
move "#include <expat.h>" into the source file to avoid this dependency.
and as a side effect, this also speed up the compilation.

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

BlueStore: fix wrong action scope of enode

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

Merge pull request #7472 from badone/wip-ceph-disk-typo

ceph-disk: Fix trivial typo

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

BlueStore: add error handling for clone

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

BlueStore: add error handling for clone_range

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

BlueStore: fix wrong decode logic of nid_max

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

BlueStore: fix unhandled return code from blockdevice APIs

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

BlueStore: try to scan as many errors as possible

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

BlueStore: fix duplicated counting of errors

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

BlueStore: fix enode verification logic

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

BlueStore: add a guard for collection loading

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

BlueStore: add a guard for overlay loading

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

Merge pull request #6813 from manens/pybind_rados_snapshot_read

pybind: Implementation of Ioctx.set_read to allow read from snapshots

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

rgw-admin: document orphans commands in usage

Fixes: #14516
Signed-off-by: Yehuda Sadeh <yehuda@redhat.com>

ceph-disk: Fix trivial typo

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

msg: ignore request_redirect_t encode/decode when we not need

Signed-off-by: Xinze Chi <xinze@xsky.com>

Merge pull request #7445 from xiexingguo/xxg-wip-mem

osd: memstore: fix blocksize for statfs

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

qa/workunits/objectstore/test_fuse.sh: fix root check

$EUID not defined everywhere.

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

os/memstore: disable lockdep on Collection::lock

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

os/kstore: disable lockdep on Collection::lock

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

os/bluestore: disable lockdep on Collection::lock

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

os/filestore/CollectionIndex: disable lockdep on access_lock

This overflows the lockdep table when there are too many PGs. This
code is static so I'm not worried about new locking bugs.

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

common/RWLock: alloc lockdep to be disabled

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

lockdep: dump lock names if we run out of IDs

..just in case we are doing something silly, like misnaming locks.

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

os/bluestore: fix typo (again)

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

os/bluestore: fix typo

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

Merge pull request #7400 from yuyuyu101/wip-13938

librbd: Avoid create two threads per image

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

Merge pull request #7409 from dillaman/wip-librbd-watch-notify

tests: improve simulation of watch/notify handling

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

Merge pull request #6562 from Aran85/authedpermcheck

rgw: fix the condition of  authenticated user's permit check .

osd/PG: fix scrub start object

This was changed in 05d79faa512210b0f0a91640d18db33b887a6e73 but it needs
to be more precise to avoid breaking BlueStore's assertions (that the
start position is within the collection range).

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

qa/workunits/rados/test.sh: run tests in parallel by default

Faster!  8 min vs the original 21 min on my vstart cluster.

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

Merge pull request #7291 from majianpeng/vstart.sh

vstart.sh: If use '-n' delete related source rather than consider '-k'.

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

osd/ECTransaction: Removing unused local

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

osd/PGBackend: PGBackend interface cleanup - make some interface methods pure virtual instead of assertion usage.
Signed-off-by: Igor Fedotov <ifedotov@mirantis.com>