Merge pull request #14540 from smithfarm/wip-18910-kraken

kraken: rbd-nbd: check /sys/block/nbdX/size to ensure kernel mapped correctly

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

Merge pull request #16140 from smithfarm/wip-20271-kraken

kraken: tests: LibRadosMiscConnectFailure.ConnectFailure hang

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

Merge pull request #16112 from smithfarm/wip-20499-kraken

kraken: tests: ObjectStore/StoreTest.OnodeSizeTracking/2 fails on bluestore

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

Merge pull request #14960 from yehudasa/wip-19704

kraken: civetweb: move to post 1.8 version

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

Merge pull request #16143 from smithfarm/wip-20365-kraken

kraken: mon: osd crush set crushmap need sanity check

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

Merge pull request #16138 from smithfarm/wip-20034-kraken

kraken: ceph-disk: Racing between partition creation & device node creation

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

Merge pull request #16135 from smithfarm/wip-20010-kraken

kraken: ceph-disk: separate ceph-osd --check-needs-* logs

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

Merge pull request #16215 from liewegas/wip-17743

kraken: CMakeLists.txt: disable memstore make check test

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

CMakeLists.txt: disable memstore make check test

This fails frequently on kraken, but is probably a wontfix.

See http://tracker.ceph.com/issues/17743

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

Merge pull request #14067 from asheplyakov/19322-bp-kraken

kraken: rgw: fix break inside of yield in RGWFetchAllMetaCR

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

Merge pull request #14509 from smithfarm/wip-18499-kraken

kraken: rgw: Realm set does not create a new period

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

Merge pull request #14511 from smithfarm/wip-18772-kraken

kraken: rgw: crash when updating period with placement group

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

Merge pull request #14513 from smithfarm/wip-18843-kraken

kraken: rgw: usage stats and quota are not operational for multi-tenant users

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

Merge pull request #14515 from smithfarm/wip-18904-kraken

kraken: rgw: first write also tries to read object

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

Merge pull request #14516 from smithfarm/wip-18909-kraken

kraken: rgw: the swift container acl does not support field .ref

Reviewed-by: Orit Wasserman <owasserm@redhat.com>
Reviewed-by: Radoslaw Zarzynski <rzarzynski@mirantis.com>

Merge pull request #14517 from smithfarm/wip-19047-kraken

kraken: rgw: fix use of marker in List::list_objects()

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

Merge pull request #14524 from smithfarm/wip-19477-kraken

kraken: rgw: S3 v4 authentication issue with X-Amz-Expires

Reviewed-by: Orit Wasserman <owasserm@redhat.com>
Reviewed-by: Radoslaw Zarzynski <rzarzynski@mirantis.com>

Merge pull request #14525 from smithfarm/wip-19479-kraken

kraken: rgw: zonegroupmap set does not work

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

Merge pull request #14528 from smithfarm/wip-19534-kraken

kraken: rgw: Error parsing xml when get bucket lifecycle

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

Merge pull request #14606 from asheplyakov/19608-kraken

kraken: rgw: multisite: fetch_remote_obj() gets wrong version when copying from remote

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

Merge pull request #15384 from asheplyakov/20015-bp-kraken

kraken: cls/rgw: list_plain_entries() stops before bi_log entries

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

Merge pull request #14522 from smithfarm/wip-19472-kraken

kraken: rgw: add the remove-x-delete feature to cancel swift object expiration

Reviewed-by: Radoslaw Zarzynski <rzarzynski@mirantis.com>

Merge pull request #14523 from smithfarm/wip-19475-kraken

kraken: rgw: multisite: EPERM when trying to read SLO objects as system/admin user

Reviewed-by: Radoslaw Zarzynski <rzarzynski@mirantis.com>

Merge pull request #14519 from smithfarm/wip-19175-kraken

kraken: swift API: cannot disable object versioning with empty X-Versions-Location

Reviewed-by: Radoslaw Zarzynski <rzarzynski@mirantis.com>

Merge pull request #14526 from smithfarm/wip-19524-kraken

kraken: rgw: 'radosgw-admin zone create' command with specified zone-id creates a zone with different id

Reviewed-by: Radoslaw Zarzynski <rzarzynski@mirantis.com>

Merge pull request #14529 from smithfarm/wip-19573-kraken

kraken: rgw: Response header of swift API returned by radosgw does not contain x-openstack-request-id. But Swift returns it.

Reviewed-by: Radoslaw Zarzynski <rzarzynski@mirantis.com>

Merge pull request #16113 from smithfarm/wip-19807-kraken

kraken: tests: remove hard-coded image name from TestLibRBD.Mirror

Reviewed-by: Jason Dillaman <dillaman@redhat.com>
Reviewed-by: Mykola Golub <mgolub@mirantis.com>

Merge pull request #16132 from smithfarm/wip-20522-kraken

kraken: FAILED assert(object_contexts.empty()) (live on master only from Jan-Feb 2017, all other instances are different)

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

Merge pull request #15962 from asheplyakov/20443-bp-kraken

kraken: osd: unlock sdata_op_ordering_lock with sdata_lock hold to avoid missing wakeup signal

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

Merge pull request #15066 from asheplyakov/19916-bp-kraken

kraken: osd: osd/PrimaryLogPG: do not call on_shutdown() if (pg.deleting)

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

Merge pull request #14942 from shinobu-x/wip-18293-kraken

kraken: osd: leaked MOSDMap

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

Merge pull request #14736 from smithfarm/wip-19326-kraken

kraken: bluestore bdev: flush no-op optimization is racy

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

Merge pull request #14732 from smithfarm/wip-19560-kraken

kraken: objecter: full_try behavior not consistent with osd

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

Merge pull request #13542 from shinobu-x/wip-18929-kraken

 kraken: osd: bogus assert when checking acting set on recovery completion in rados/upgrade

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

Merge pull request #14939 from rzarzynski/wip-rgw-19754-kraken

kraken: rgw: fix RadosGW hang during multi-chunk upload of AWSv4

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

Merge pull request #14530 from smithfarm/wip-19574-kraken

kraken: rgw: unsafe access in RGWListBucket_ObjStore_SWIFT::send_response()

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

Merge pull request #13868 from smithfarm/wip-19160-kraken

kraken: rgw: multisite: RGWMetaSyncShardControlCR gives up on EIO

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

mon: osd crush set crushmap need sanity check

The sanity check verifying the new crushmap does not remove crush rules
that are in use is not exclusive to ceph setcrushmap.

Fixes: http://tracker.ceph.com/issues/19302
Signed-off-by: Loic Dachary <loic@dachary.org>
(cherry picked from commit ed760457bf154c10adf75c6df046eecab7eb8e4b)

ceph_test_rados_api_misc: fix LibRadosMiscConnectFailure.ConnectFailure retry

Fixes: http://tracker.ceph.com/issues/19901
Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit 087dff80cac707ee9bcb5bcfc98cb3ec047bd49f)

ceph-disk: Adding retry loop in get_partition_dev()

There is very rare cases where get_partition_dev() is called before the actual partition is available in /sys/block/<device>.

It appear that waiting a very short is usually enough to get the partition beein populated.

Analysis:
update_partition() is supposed to be enough to avoid any racing between events sent by parted/sgdisk/partprobe and
the actual creation on the /sys/block/<device>/* entrypoint.
On our CI that race occurs pretty often but trying to reproduce it locally never been possible.

This patch is almost a workaround rather than a fix to the real problem.
It offer retrying after a very short to be make a chance the device to appear.
This approach have been succesful on the CI.

Note his patch is not changing the timing when the device is perfectly created on time and just differ by a 1/5th up to 2 seconds when the bug occurs.

A typical output from the build running on a CI with that code.
command_check_call: Running command: /usr/bin/udevadm settle --timeout=600
get_dm_uuid: get_dm_uuid /dev/sda uuid path is /sys/dev/block/8:0/dm/uuid
get_partition_dev: Try 1/10 : partition 2 for /dev/sda does not in /sys/block/sda
get_partition_dev: Found partition 2 for /dev/sda after 1 tries
        get_dm_uuid: get_dm_uuid /dev/sda uuid path is /sys/dev/block/8:0/dm/uuid
get_dm_uuid: get_dm_uuid /dev/sda2 uuid path is /sys/dev/block/8:2/dm/uuid

fixes: #19428

Signed-off-by: Erwan Velu <erwan@redhat.com>
(cherry picked from commit 93e7b95ed8b4c78daebf7866bb1f0826d7199075)

ceph-disk: Reporting /sys directory in get_partition_dev()

When get_partition_dev() fails, it reports the following message :
    ceph_disk.main.Error: Error: partition 2 for /dev/sdb does not appear to exist
The code search for a directory inside the /sys/block/get_dev_name(os.path.realpath(dev)).

The issue here is the error message doesn't report that path when failing while it might be involved in.

This patch is about reporting where the code was looking at when trying to estimate if the partition was available.

Signed-off-by: Erwan Velu <erwan@redhat.com>
(cherry picked from commit 413c9fcfbe8e6ab33d73b8428090ccacc33c5d15)

ceph-disk: separate ceph-osd --check-needs-* logs

It is using the OSD id zero but have nothing to do with OSD zero and
this is confusing to the user. The log themselves do not need to be kept
around and are stored in the run directory so that they can be disposed
of after reboot.

Fixes: http://tracker.ceph.com/issues/19888
Signed-off-by: Loic Dachary <loic@dachary.org>
(cherry picked from commit c7b3c46bd63b78475868e405bf20d9c142f0336a)

ReplicatedBackend: don't queue Context outside of ObjectStore with obc

We only flush the ObjectStore callbacks, not everything else.  Thus,
there isn't a guarrantee that the obc held by pull_complete_info will
be cleaned up before the Flush callback is triggered.  Instead, just
defer clearing the pull state until the callback (it'll be cleaned up
during the interval change) and remove the ObjectContext from
pull_complete_info.

Introduced: 68defc2b0561414711d4dd0a76bc5d0f46f8a3f8
Fixes: http://tracker.ceph.com/issues/18927
Signed-off-by: Samuel Just <sjust@redhat.com>
(cherry picked from commit 51eee55c475a3d931844831e040ed3d66ee59af4)

osd/: don't leak context for Blessed*Context or RecoveryQueueAsync

This has always been a bug, but until
68defc2b0561414711d4dd0a76bc5d0f46f8a3f8, nothing deleted those contexts
without calling complete().

Fixes: http://tracker.ceph.com/issues/18809
Bug shadowed until: 68defc2b0561414711d4dd0a76bc5d0f46f8a3f8
Signed-off-by: Samuel Just <sjust@redhat.com>
(cherry picked from commit 91b74235027c8a4872dcab6b37767b12c3267061)

Merge pull request #15792 from liewegas/wip-bluestore-leaks-kraken

kraken: os/bluestore: deep decode onode value

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

Merge pull request #15729 from asheplyakov/20315-bp-kraken

kraken: mon: fail to form large quorum; msg/async busy loop

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

Merge pull request #15084 from asheplyakov/19928-bp-kraken

kraken: mon crash on shutdown, lease_ack_timeout event

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

Merge pull request #14735 from smithfarm/wip-19685-kraken

kraken: osd: Give requested scrubs a higher priority

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

Merge pull request #14733 from smithfarm/wip-19561-kraken

kraken: api_misc: [  FAILED  ] LibRadosMiscConnectFailure.ConnectFailure

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

Merge pull request #13883 from shinobu-x/wip-19119-kraken

kraken: pre-jewel "osd rm" incrementals are misinterpreted

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

Merge pull request #14646 from gregsfortytwo/wip-17916-kraken

kraken: osd: pglog: with config, don't assert in the presence of stale diverg…

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

Merge pull request #15237 from asheplyakov/20035-bp-kraken

kraken: mon/PGMap: factor mon_osd_full_ratio into MAX AVAIL calc

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

Merge pull request #14331 from shinobu-x/wip-19396-kraken

kraken: Objecter::epoch_barrier isn't respected in _op_submit()

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

Merge pull request #16083 from smithfarm/wip-18677-kraken

kraken: osd: metadata reports filestore when using bluestore

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

Merge pull request #16098 from smithfarm/wip-19621-kraken

kraken: rbd-nbd: add signal handler

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

Merge pull request #15612 from smithfarm/wip-19611-kraken

kraken: rbd: Issues with C API image metadata retrieval functions

Reviewed-by: Jason Dillaman <dillaman@redhat.com>
Reviewed-by: Mykola Golub <mgolub@mirantis.com>

Merge pull request #14833 from smithfarm/wip-19794-kraken

kraken: tests: test_notify.py: assert(not image.is_exclusive_lock_owner()) on line 147

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

Merge pull request #14543 from smithfarm/wip-19173-kraken

kraken: rbd: rbd_clone_copy_on_read ineffective with exclusive-lock

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

Merge pull request #14974 from dillaman/wip-19872-kraken

kraken: rbd-mirror: failover and failback of unmodified image results in split-brain

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

Merge pull request #14934 from dillaman/wip-19833-kraken

kraken: cls_rbd: default initialize snapshot namespace for legacy clients

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

Merge pull request #14539 from smithfarm/wip-18771-kraken

kraken: rbd: Improve compatibility between librbd + krbd for the data pool

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

test: remove hard-coded image name from RBD metadata test

Fixes: http://tracker.ceph.com/issues/19798
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit 8f72e745e157cc12d76ca6babe956c5698ee297f)

librbd: relax "is parent mirrored" check when enabling mirroring for pool

If the parent is in the same pool and has the journaling feature enabled
we can assume the mirroring will eventually be enabled for it.

Fixes: http://tracker.ceph.com/issues/19130
Signed-off-by: Mykola Golub <mgolub@mirantis.com>
(cherry picked from commit fe31bca22f90ce02f461d6421a4f66539db888d3)

rbd-mirror: check image mirroring state when bootstrapping

Fixes: http://tracker.ceph.com/issues/18447
Signed-off-by: Mykola Golub <mgolub@mirantis.com>
(cherry picked from commit 5fc5a8ac895524f05eed6e7db20b0dda3a8cb60f)

rbd-mirror: async request to test if image is primary

Signed-off-by: Mykola Golub <mgolub@mirantis.com>
(cherry picked from commit 0a1cb35caacdf85029f31a0364dc07a5d7462f5f)

rbd-mirror: hold owner lock when testing if lock owner

Signed-off-by: Mykola Golub <mgolub@mirantis.com>
(cherry picked from commit 82aa89668d666f434cd19ff444223017b5512c6a)

os/bluestore: fix OnodeSizeTracking testing

[  FAILED  ] 1 test, listed below:
[  FAILED  ] ObjectStore/StoreTest.OnodeSizeTracking/2, where GetParam() = "bluestore"

 1 FAILED TEST

The above test failure happens as the bluestore mount() process
will try to load all collections and put them into the coll_map,
which will be also tracked as mempool::bluestore_meta_other.
So total_bytes from mempool won't be equal to zero.

Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
(cherry picked from commit 1d97862f716ce24bfe227b192ecc5e1eb3f76757)

os/bluestore/KernelDevice: fix uninit value

Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit 42db0c70bc7ef595f0925657c043ce081799b2b9)

Conflicts:
    src/os/bluestore/KernelDevice.h
      - flush_lock mutex is still used in kraken, so keep it

os/bluestore/BlueFS: fix flush_bdev placement

We need to flush any new writes on any fsync().  Notably, this includes
the rocksdb log. However, previously _fsync was only doing a bdev flush if
we also had a dirty bluefs journal and called into _sync_and_flush_journal.
If we didn't, we weren't doing a flush() at all, which could lead to
corrupted data.

Fix this by moving the first flush_bdev *out* of _sync_and_flush_log.  (The
second one is there to flush the bluefs journal; the first one was to
ensure prior writes are stable.)  Instead, flush prior writes in all of the
callers prior to calling _sync_and_flush_log.  This includes _fsync (and
fixes the bug by covering the non-journal-flush path) as well as several
other callers.

Fixes: http://tracker.ceph.com/issues/19250
Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit 2924a96493d8570317e55854a25fc64911ecf151)

os/bluestore/KernelDevice: make flush() thread safe

flush() may be called from multiple racing threads (notably, rocksdb can call fsync via
bluefs at any time), and we need to make sure that if one thread sees the io_since_flush
command and does an actual flush, that other racing threads also wait until that flush is
complete.  This is accomplished with a simple mutex!

Also, set the flag on IO *completion*, since flush is only a promise about
completed IOs, not submitted IOs.

Document.

Fixes: http://tracker.ceph.com/issues/19251
Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit 6b3c52643c8e5fa820c53d96608862b7649c3fd0)

Merge pull request #14065 from asheplyakov/19212-bp-kraken

kraken: rgw: "cluster [WRN] bad locator @X on object @X...." in cluster log

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

Merge pull request #16092 from smithfarm/wip-20150-kraken

kraken: ceph-disk: fails if OSD udev rule triggers prior to mount of /var

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

Merge pull request #14604 from asheplyakov/19647-kraken

kraken: ceph-disk: enable directory backed OSD at boot time

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

rbd-nbd: remove debug message from do_unmap

Global context is not initialized when do_unmap is called.

Signed-off-by: Pan Liu <liupan1111@gmail.com>
(cherry picked from commit cd748f9b260a3c388b45091ff0bfc041257acd0c)

rbd-nbd: s/cpp_error/cpp_strerror/ to fix FTBFS

the build failure was introduced by ff4dcf0

Signed-off-by: Kefu Chai <kchai@redhat.com>
(cherry picked from commit c12ecd984df0f9ac0d7de81b9d689f6182a3646a)

rbd-nbd: polish the output info before and after ioctl NBD_DISCONNECT.

Signed-off-by: Pan Liu <liupan1111@gmail.com>
(cherry picked from commit ff4dcf029028e8a3636ac71a6c5ac5380bf274e4)

rbd-nbd: support signal handle for SIGHUP, SIGINT, and SIGTERM.

Fixes: http://tracker.ceph.com/issues/19349
Signed-off-by: Pan Liu <liupan1111@gmail.com>
(cherry picked from commit 3ba01aa6ce052d1afa42132feffc2353d73caae6)

Merge pull request #14616 from smithfarm/wip-18723-kraken

kraken: osd: calc_clone_subsets misuses try_read_lock vs missing

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

ceph-disk: do not setup_statedir on trigger

trigger may run when statedir is unavailable and does not use it.

Fixes: http://tracker.ceph.com/issues/19941
Signed-off-by: Loic Dachary <loic@dachary.org>
(cherry picked from commit 16bfbdd3d9988523bba31aace516c303057daa58)

Merge pull request #14852 from yehudasa/wip-rgw-support-ragweed-kraken

kraken: rgw: add apis to support ragweed suite

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

Merge pull request #13514 from rzarzynski/wip-rgw-controllable-204-kraken

kraken: rgw: make sending Content-Length in 204 and 304 controllable

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

Merge pull request #14403 from shinobu-x/wip-17331-kraken

kraken: ceph-disk list reports mount error for OSD having mount options with SELinux context

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

osd: Return correct osd_objectstore in OSD metadata

Do not simply read the configuration value as it might have changed
during OSD startup by reading the type from disk.

Fixes: http://tracker.ceph.com/issues/18638
Signed-off-by: Wido den Hollander <wido@42on.com>
(cherry picked from commit 8fe6a0303b02ac1033f5bfced9f94350fe3e33de)

Conflicts:
    src/osd/OSD.cc
      - g_conf->osd_objectstore was changed to cct->_conf->osd_objectstore by
        1d5e967a05ddbcceb10efe3b57e242b3b6b7eb8c which is not in kraken

rgw: fix use of marker in List::list_objects()

Fixes: http://tracker.ceph.com/issues/18331
List marker is an index key, so treat it as such. This
fixes infinite loop in orphans find command.

Signed-off-by: Yehuda Sadeh <yehuda@redhat.com>
(cherry picked from commit a5d1fa0587184f43c69d8e03114b58d43f320781)

rgw: fix RadosGW hang during multi-chunk upload of AWSv4.

Fixes: http://tracker.ceph.com/issues/19754
Signed-off-by: Radoslaw Zarzynski <rzarzynski@mirantis.com>
(cherry picked from commit 72c1e2e351d984d0425a20f2c772951cbc36f13e)

rgw/lifecycle: do not send lifecycle rules when GetLifeCycle failed

Now, RGW will send two HTTP responses when GetLifeCycle failed. The first one is
Error Respnse like 404, and the second is lifecycle rules. It will breaks s3 sdk
and s3 utilities.

Fixes: http://tracker.ceph.com/issues/19363
Signed-off-by: liuchang0812 <liuchang0812@gmail.com>
(cherry picked from commit c3c0c828da5a64ca896475c1b0c369fde1bbd76a)

  rgw: fix response header of Swift API

Response header of Swift API returned by radosgw does not contain
"x-openstack-request-id", but Swift returns it. Enhance the
compatibility of radosgw.

Fixes: http://tracker.ceph.com/issues/19443
Signed-off-by: tone-zhang <tone.zhang@linaro.org>
(cherry picked from commit e96db213079ab5e026156ab4b38418d1d4c23d27)

rgw: fix crash when listing objects via swift

Fixes: http://tracker.ceph.com/issues/19249
Signed-off-by: Yehuda Sadeh <yehuda@redhat.com>
(cherry picked from commit a9ec5e8ce184e19c009863db4d3519f9d8af91bd)

Conflicts:
        src/rgw/rgw_rest_swift.cc ("key" element of RGWObjEnt struct
           is not a reference; fix)

librbd: introduce new constants for tracking max block name prefix

Fixes: http://tracker.ceph.com/issues/18653
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit 2c08629c99d90aa7676b59263c055c9f1f577039)

rbd-nbd: no need create asok file for unmap and list-mapped commands.

Fixes: http://tracker.ceph.com/issues/17951
Signed-off-by: Pan Liu <liupan1111@gmail.com>
(cherry picked from commit 72352653d585ef89043a4ece371b5c0cb3f6f32a)

rbd-nbd: check /sys/block/nbdX/size to ensure kernel mapped correctly

Fixes: http://tracker.ceph.com/issues/18335
Signed-off-by: Mykola Golub <mgolub@mirantis.com>
(cherry picked from commit 596e5ea8a5df72002672eef0a6d20572ca6f60f0)

librbd: acquire exclusive-lock during copy on read

Fixes: http://tracker.ceph.com/issues/18888
Signed-off-by: Venky Shankar <vshankar@redhat.com>
(cherry picked from commit 7dba5311b12011a4a6e8564e68150e54c5af5ddd)

Conflicts:
    src/librbd/AioImageRequestWQ.h:
      - in master this file has morphed into src/librbd/io/ImageRequestWQ.h
      - kraken has AioImageRequest<ImageCtx> instead of ImageRequest<ImageCtx>
    src/librbd/image/RefreshRequest.cc:
      - rename image context element to "aio_work_queue" (from "io_work_queue")
        because kraken doesn't have de95d862f57b56738e04d77f2351622f83f17f4a
    src/test/librbd/image/test_mock_RefreshRequest.cc:
      - rename image context element to "aio_work_queue" (from "io_work_queue")
        because kraken doesn't have de95d862f57b56738e04d77f2351622f83f17f4a

PrimaryLogPG::try_lock_for_read: give up if missing

The only users calc_*_subsets might try to read_lock an object which is
missing on the primary.  Returning false in those cases is perfectly
reasonable and avoids the problem.

Fixes: http://tracker.ceph.com/issues/18583
Signed-off-by: Samuel Just <sjust@redhat.com>
(cherry picked from commit 3833440adea6f8bcb0093603c3a9d16360ed57ec)

ReplicatedBackend: take read locks for clone sources during recovery

Otherwise, we run the risk of a clone source which hasn't actually
come into existence yet being used if we grab a clone which *just*
got added the the ssc, but has not yet actually had time to be
created (can't rely on message ordering here since recovery messages
don't necessarily order with client IO!).

Fixes: http://tracker.ceph.com/issues/17831
Signed-off-by: Samuel Just <sjust@redhat.com>
(cherry picked from commit 68defc2b0561414711d4dd0a76bc5d0f46f8a3f8)

osd/PrimaryLogPG: do not expect FULL_TRY ops to get resent

The objecter will not resend FULL_TRY requests that were sent prior to
becoming full, so we should not discard them.

Fixes: http://tracker.ceph.com/issues/19430
Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit 3f7acdbc9a942fd18937dbcf07fbc7b752c50ba3)

ceph_test_librados_api_misc: fix stupid LibRadosMiscConnectFailure.ConnectFailure test

Sometimes the cond doesn't time out and it wakes up instead.  Just repeat
the test many times to ensure that at least once it times out (usually
it doesn't; it's pretty infrequent that it doesn't).

Fixes: http://tracker.ceph.com/issues/15368
Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit 8bc197400d94ee2716d3f2fa454247379a676cf9)

osd: Give requested scrub work a higher priority

Once started we now queue scrub work at higher priority than
scheduled scrubs.

Fixes: http://tracker.ceph.com/issues/15789
Signed-off-by: David Zafman <dzafman@redhat.com>
(cherry picked from commit ebab8b1f4f67fbdec1e147c580329c1e2b5cf7cd)

Conflicts:
src/osd/OSD.h - in kraken, the PGScrub() call is enclosed within
                   op_wq.queue(make_pair(...)) instead of enqueue_back()

Merge pull request #16069 from smithfarm/wip-20345-kraken

kraken: make check fails with Error EIO: load dlopen(build/lib/libec_FAKE.so): build/lib/libec_FAKE.so: cannot open shared object file: No such file or directory

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

ceph.spec.in, debian/control: Add bc to build dependencies

The bc is missing for ceph-helpers.sh

Fixes: http://tracker.ceph.com/issues/18876
Signed-off-by: Kyr Shatskyy <kyrylo.shatskyy@gmail.com>
(cherry picked from commit 3ff4be6d6896346b7b9ec7f158fcde8866faeb38)

qa/workunits/ceph-helpers: do not error out if is_clean

it would be a race otherwise, because we cannot be sure that the cluster
pgs are not all clean or not when run_osd() returns, but we can be sure
that they are expected to active+clean after a while. that's what
wait_for_clean() does.

Signed-off-by: Kefu Chai <kchai@redhat.com>
(cherry picked from commit 6cb4503a40ae4ebee5690fda024cae8d1a506bce)