[SRU] ceph 10.2.3

[James Page]

This point release fixes several important bugs in RBD mirroring, RGW multi-site, CephFS, and RADOS.

We recommend that all v10.2.x users upgrade.

Notable changes in this release include:
* build/ops: 60-ceph-partuuid-workaround-rules still needed by debian jessie (udev 215-17) (#16351, runsisi, Loic Dachary)
* build/ops: ceph Resource Agent does not work with systemd (#14828, Nathan Cutler)
* build/ops: ceph-base requires parted (#16095, Ken Dreyer)
* build/ops: ceph-osd-prestart.sh contains Upstart-specific code (#15984, Nathan Cutler)
* build/ops: mount.ceph: move from ceph-base to ceph-common and add symlink in /sbin for SUSE (#16598, #16645, Nathan Cutler, Dan Horák, Ricardo Dias, Kefu Chai)
* build/ops: need rocksdb commit 7ca731b12ce for ppc64le build (#17092, Nathan Cutler)
* build/ops: rpm: OBS needs ExclusiveArch (#16936, Michel Normand)
* cli: ceph command line tool chokes on ceph –w (the dash is unicode 'en dash' &ndash, copy-paste to reproduce) (#12287, Oleh Prypin, Kefu Chai)
* common: expose buffer const_iterator symbols (#16899, Noah Watkins)
* common: global-init: fixup chown of the run directory along with log and asok files (#15607, Karol Mroz)
* fs: ceph-fuse: link to libtcmalloc or jemalloc (#16655, Yan, Zheng)
* fs: client: crash in unmount when fuse_use_invalidate_cb is enabled (#16137, Yan, Zheng)
* fs: client: fstat cap release (#15723, Yan, Zheng, Noah Watkins)
* fs: essential backports for OpenStack Manila (#15406, #15614, #15615, John Spray, Ramana Raja, Xiaoxi Chen)
* fs: fix double-unlock on shutdown (#17126, Greg Farnum)
* fs: fix mdsmap print_summary with standby replays (#15705, John Spray)
* fs: fuse mounted file systems fails SAMBA CTDB ping_pong rw test with v9.0.2 (#12653, #15634, Yan, Zheng)
* librados: Add cleanup message with time to rados bench output (#15704, Vikhyat Umrao)
* librados: Missing export for rados_aio_get_version in src/include/rados/librados.h (#15535, Jim Wright)
* librados: osd: bad flags can crash the osd (#16012, Sage Weil)
* librbd: Close journal and object map before flagging exclusive lock as released (#16450, Jason Dillaman)
* librbd: Crash when utilizing advisory locking API functions (#16364, Jason Dillaman)
* librbd: ExclusiveLock object leaked when switching to snapshot (#16446, Jason Dillaman)
* librbd: FAILED assert(object_no < m_object_map.size()) (#16561, Jason Dillaman)
* librbd: Image removal doesn't necessarily clean up all rbd_mirroring entries (#16471, Jason Dillaman)
* librbd: Object map/fast-diff invalidated if journal replays the same snap remove event (#16350, Jason Dillaman)
* librbd: Timeout sending mirroring notification shouldn't result in failure (#16470, Jason Dillaman)
* librbd: Whitelist EBUSY error from snap unprotect for journal replay (#16445, Jason Dillaman)
* librbd: cancel all tasks should wait until finisher is done (#16517, Haomai Wang)
* librbd: delay acquiring lock if image watch has failed (#16923, Jason Dillaman)
* librbd: fix missing return statement if failed to get mirror image state (#16600, runsisi)
* librbd: flag image as updated after proxying maintenance op (#16404, Jason Dillaman)
* librbd: mkfs.xfs slow performance with discards and object map (#16707, #16689, Jason Dillaman)
* librbd: potential use after free on refresh error (#16519, Mykola Golub)
* librbd: rbd-nbd does not properly handle resize notifications (#15715, Mykola Golub)
* librbd: the option 'rbd_cache_writethrough_until_flush=true' dosn't work (#16740, #16386, #16708, #16654, #16478, Mykola Golub, xinxin shu, Xiaowei Chen, Jason Dillaman)
* mds: tell command blocks forever with async messenger (TestVolumeClient.test_evict_client failure) (#16288, Douglas Fuller)
* mds: Confusing MDS log message when shut down with stalled journaler reads (#15689, John Spray)
* mds: Deadlock on shutdown active rank while busy with metadata IO (#16042, Patrick Donnelly)
* mds: Failing file operations on kernel based cephfs mount point leaves unaccessible file behind on hammer 0.94.7 (#16013, Yan, Zheng)
* mds: Fix shutting down mds timed-out due to deadlock (#16396, Zhi Zhang)
* mds: MDSMonitor fixes (#16136, xie xingguo)
* mds: MDSMonitor::check_subs() is very buggy (#16022, Yan, Zheng)
* mds: Session::check_access() is buggy (#16358, Yan, Zheng)
* mds: StrayManager.cc: 520: FAILED assert(dnl->is_primary()) (#15920, Yan, Zheng)
* mds: enforce a dirfrag limit on entries (#16164, Patrick Donnelly)
* mds: fix SnapRealm::have_past_parents_open() (#16299, Yan, Zheng)
* mds: fix getattr starve setattr (#16154, Yan, Zheng)
* mds: wrongly treat symlink inode as normal file/dir when symlink inode is stale on kcephfs (#15702, Zhi Zhang)
* mon: "mon metadata" fails when only one monitor exists (#15866, John Spray, Kefu Chai)
* mon: Monitor: validate prefix on handle_command() (#16297, You Ji)
* mon: OSDMonitor: drop pg temps from not the current primary (#16127, Samuel Just)
* mon: prepare_pgtemp needs to only update up_thru if newer than the existing one (#16185, Samuel Just)
* msgr: AsyncConnection::lockmsg/async lockdep cycle: AsyncMessenger::lock, MDSDaemon::mds_lock, AsyncConnection::lock (#16237, Haomai Wang)
* msgr: async messenger mon crash (#16378, #16418, Haomai Wang)
* msgr: backports of all asyncmsgr fixes to jewel (#15503, #15372, Yan Jun, Haomai Wang, Piotr Dałek)
* msgr: msg/async: connection race hang (#15849, Haomai Wang)
* osd: FileStore: umount hang because sync thread doesn't exit (#15695, Kefu Chai)
* osd: Fixes for list-inconsistent-* (#15766, #16192, #15719, David Zafman)
* osd: New pools have bogus stuck inactive/unclean HEALTH_ERR messages until they are first active and clean (#14952, Sage Weil)
* osd: OSD crash with Hammer to Jewel Upgrade: void FileStore::init_temp_collections() (#16672, David Zafman)
* osd: OSD failed to subscribe skipped osdmaps after ceph osd pause (#17023, Kefu Chai)
* osd: ObjectCacher split BufferHead read fix (#16002, Greg Farnum)
* osd: ReplicatedBackend doesn't increment stats on pull, only push (#16277, Kefu Chai)
* osd: Scrub error: 0/1 pinned (#15952, Samuel Just)
* osd: crash adding snap to purged_snaps in ReplicatedPG::WaitingOnReplicas (#15943, Samuel Just)
* osd: partprobe intermittent issues during ceph-disk prepare (#15176, Marius Vollmer, Loic Dachary)
* osd: saw valgrind issues in ReplicatedPG::new_repop (#16801, Kefu Chai)
* osd: sparse_read on ec pool should return extends with correct offset (#16138, kofiliu)
* osd:sched_time not actually randomized (#15890, xie xingguo)
* rbd: ImageReplayer::is_replaying does not include flush state (#16970, Jason Dillaman)
* rbd: Journal duplicate op detection can cause lockdep error (#16363, Jason Dillaman)
* rbd: Journal needs to handle duplicate maintenance op tids (#16362, Jason Dillaman)
* rbd: Unable to disable journaling feature if in unexpected mirror state (#16348, Jason Dillaman)
* rbd: bashism in src/rbdmap (#16608, Jason Dillaman)
* rbd: doc: format 2 now is the default image format (#17026, Chengwei Yang)
* rbd: hen journaling is enabled, a flush request shouldn't flush the cache (#15761, Yuan Zhou)
* rbd: possible race condition during journal transition from replay to ready (#16198, Jason Dillaman)
* rbd: qa/workunits/rbd: respect RBD_CREATE_ARGS environment variable (#16289, Mykola Golub)
* rbd: rbd-mirror should disable proxied maintenance ops for non-primary image (#16411, Jason Dillaman)
* rbd: rbd-mirror: FAILED assert(m_local_image_ctx->object_map != nullptr) (#16558, Jason Dillaman)
* rbd: rbd-mirror: FAILED assert(m_on_update_status_finish == nullptr) (#16956, Jason Dillaman)
* rbd: rbd-mirror: FAILED assert(m_state == STATE_STOPPING) (#16980, Jason Dillaman)
* rbd: rbd-mirror: ensure replay status formatter has completed before stopping replay (#16352, Jason Dillaman)
* rbd: rbd-mirror: include local pool id in resync throttle unique key (#16536, #15239, #16488, #16491, #16329, #15108, #15670, Ricardo Dias, Jason Dillaman)
* rbd: rbd-mirror: potential race condition accessing local image journal (#16230, Jason Dillaman)
* rbd: rbd-mirror: reduce memory footprint during journal replay (#16321, #16489, #16622, #16539, #16223, #16349, Mykola Golub, Jason Dillaman)
* rgw: A query on a static large object fails with 404 error (#16015, Radoslaw Zarzynski)
* rgw: Add zone rename to radosgw_admin (#16934, Shilpa Jagannath)
* rgw: Bucket index shards orphaned after bucket delete (#16412, Orit Wasserman)
* rgw: Bug when using port 443s in rgw. (#16548, Pritha Srivastava)
* rgw: Fallback to Host header for bucket name. (#15975, Robin H. Johnson)
* rgw: Fix civetweb IPv6 (#16928, Robin H. Johnson)
* rgw: Increase log level for messages occuring while running rgw admin command (#16935, Shilpa Jagannath)
* rgw: No Last-Modified, Content-Size and X-Object-Manifest headers if no segments in DLO manifest (#15812, Radoslaw Zarzynski)
* rgw: RGWPeriodPuller tries to pull from itself (#16939, Casey Bodley)
* rgw: Set Access-Control-Allow-Origin to a Asterisk if allowed in a rule (#15348, Wido den Hollander)
* rgw: Swift API returns double space usage and objects of account metadata (#16188, Albert Tu)
* rgw: account/container metadata not actually present in a request are deleted during POST through Swift API (#15977, #15779, Radoslaw Zarzynski)
* rgw: add socket backlog setting for via ceph.conf (#16406, Feng Guo)
* rgw: add tenant support to multisite sync (#16469, #16121, #16665, Yehuda Sadeh, Josh Durgin, Casey Bodley, Pritha Srivastava)
* rgw: add_zone only clears master_zone if --master=false (#15901, Casey Bodley)
* rgw: aws4 parsing issue (#15940, #15939, Yehuda Sadeh)
* rgw: aws4: add STREAMING-AWS4-HMAC-SHA256-PAYLOAD support (#16146, Radoslaw Zarzynski, Javier M. Mellid)
* rgw: backport merge of static sites fixes (#15555, #15532, #15531, Robin H. Johnson)
* rgw: can set negative max_buckets on RGWUserInfo (#14534, Yehuda Sadeh)
* rgw: cleanup radosgw-admin temp command as it was deprecated (#16023, Vikhyat Umrao)
* rgw: comparing return code to ERR_NOT_MODIFIED in rgw_rest_s3.cc (needs minus sign) (#16327, Nathan Cutler)
* rgw: custom metadata aren't camelcased in Swift's responses (#15902, Radoslaw Zarzynski)
* rgw: data sync stops after getting error in all data log sync shards (#16530, Yehuda Sadeh)
* rgw: default zone and zonegroup cannot be added to a realm (#16839, Casey Bodley)
* rgw: document multi tenancy (#16635, Pete Zaitcev)
* rgw: don't unregister request if request is not connected to manager (#15911, Yehuda Sadeh)
* rgw: failed to create bucket after upgrade from hammer to jewel (#16627, Orit Wasserman)
* rgw: fix ldap bindpw parsing (#16286, Matt Benjamin)
* rgw: fix multi-delete query param parsing. (#16618, Robin H. Johnson)
* rgw: improve support for Swift's object versioning. (#15925, Radoslaw Zarzynski)
* rgw: initial slashes are not properly handled in Swift's BulkDelete (#15948, Radoslaw Zarzynski)
* rgw: master: build failures with boost > 1.58 (#16392, #16391, Abhishek Lekshmanan)
* rgw: multisite segfault on ~RGWRealmWatcher if realm was deleted (#16817, Casey Bodley)
* rgw: multisite sync races with deletes (#16222, #16464, #16220, #16143, Yehuda Sadeh, Casey Bodley)
* rgw: multisite: preserve zone's extra pool (#16712, Abhishek Lekshmanan)
* rgw: object expirer's hints might be trimmed without processing in some circumstances (#16705, #16684, Radoslaw Zarzynski)
* rgw: radosgw-admin failure for user create after upgrade from hammer to jewel (#15937, Orit Wasserman, Abhishek Lekshmanan)
* rgw: radosgw-admin: EEXIST messages for create operations (#15720, Abhishek Lekshmanan)
* rgw: radosgw-admin: inconsistency in uid/email handling (#13598, Matt Benjamin)
* rgw: realm pull fails when using apache frontend (#15846, Orit Wasserman)
* rgw: retry on bucket sync errors (#16108, Yehuda Sadeh)
* rgw: s3website: x-amz-website-redirect-location header returns malformed HTTP response (#15531, Robin H. Johnson)
* rgw: segfault in RGWOp_MDLog_Notify (#16666, Casey Bodley)
* rgw: segmentation fault on error_repo in data sync (#16603, Casey Bodley)
* rgw: selinux denials in RGW (#16126, Boris Ranto)
* rgw: support size suffixes for --max-size in radosgw-admin command (#16004, Vikhyat Umrao)
* rgw: updating CORS/ACLs might not work in some circumstances (#15976, Radoslaw Zarzynski)
* rgw: use zone endpoints instead of zonegroup endpoints (#16834, Casey Bodley)
* tests: improve rbd-mirror test case coverage (#16197, Mykola Golub, Jason Dillaman)
* tests: rados/test.sh workunit timesout on OpenStack (#15403, Loic Dachary)
* tools: ceph-disk: Accept bcache devices as data disks (#13278, Peter Sabaini)
* tools: src/script/subman fails with KeyError: 'nband' (#16961, Loic Dachary, Ali Maredia)

For more detailed information refer to the complete changelog[1] and the
release notes[2]

Getting Ceph
------------

* Git at git://github.com/ceph/ceph.git
* Tarball at http://download.ceph.com/tarballs/ceph-10.2.3.tar.gz
* For packages, see http://ceph.com/docs/master/install/get-packages
* For ceph-deploy, see http://ceph.com/docs/master/install/install-ceph-deploy

[1]: http://docs.ceph.com/docs/master/_downloads/v10.2.3.txt
[2]: http://docs.ceph.com/docs/master/release-notes/#v10-2-3-jewel

Regards,
Abhishek

[David Medberry (med)]

Does this include the changes/fixes that James Troup reported in bug #1628750? Or do we expect an urgent update to this?

[James Page (james-page)]

Plan is to pick that fix at the same time.

[James Page (james-page)]

Dumping packages for all targets here:

https://launchpad.net/~openstack-ubuntu-testing/+archive/ubuntu/ceph-sru

Once they build and test OK I'll upload for SRU; note that yakkety is broken in other ways atm (bug 1629102).

[James Page (james-page)]

Uploaded to xenial-proposed; SRU testing is covered as part of the OpenStack Team Stable Release Update process:

  https://wiki.ubuntu.com/OpenStack/StableReleaseUpdates

[Launchpad Janitor (janitor)]

This bug was fixed in the package ceph - 10.2.3-0ubuntu1

---------------
ceph (10.2.3-0ubuntu1) yakkety; urgency=medium

  * New upstream point release (LP: #1628809):
    - d/p/rocksdb-flags.patch: Dropped, included upstream.
    - d/p/*: Refreshed.
    - d/p/32bit-ftbfs.patch: Cherry pick fix for 32bit arch compat.
    - d/ceph-{fs-common,fuse}.install: Fix install locations
      for mount{.fuse}.ceph.
  * Limit the amount of data per chunk in omap push operations to 64k,
    ensuring that OSD threads don't hit timeouts during recovery
    operations (LP: #1628750):
    - d/p/osd-limit-omap-data-in-push-op.patch: Cherry pick fix from
      upstream master branch.

 -- James Page <email address hidden> Thu, 29 Sep 2016 21:44:33 +0100

[Brian Murray (brian-murray)]

Hello James, or anyone else affected,

Accepted ceph into xenial-proposed. The package will build now and be available at https://launchpad.net/ubuntu/+source/ceph/10.2.3-0ubuntu0.16.04.1 in a few hours, and then in the -proposed repository.

Please help us by testing this new package. See https://wiki.ubuntu.com/Testing/EnableProposed for documentation how to enable and use -proposed. Your feedback will aid us getting this update out to other Ubuntu users.

If this package fixes the bug for you, please add a comment to this bug, mentioning the version of the package you tested, and change the tag from verification-needed to verification-done. If it does not fix the bug for you, please add a comment stating that, and change the tag to verification-failed. In either case, details of your testing will help us make a better decision.

Further information regarding the verification process can be found at https://wiki.ubuntu.com/QATeam/PerformingSRUVerification . Thank you in advance!

[James Page (james-page)]

Hello James, or anyone else affected,

Accepted ceph into mitaka-proposed. The package will build now and be available in the Ubuntu Cloud Archive in a few hours, and then in the -proposed repository.

Please help us by testing this new package. To enable the -proposed repository:

  sudo add-apt-repository cloud-archive:mitaka-proposed
  sudo apt-get update

Your feedback will aid us getting this update out to other Ubuntu users.

If this package fixes the bug for you, please add a comment to this bug, mentioning the version of the package you tested, and change the tag from verification-mitaka-needed to verification-mitaka-done. If it does not fix the bug for you, please add a comment stating that, and change the tag to verification-mitaka-failed. In either case, details of your testing will help us make a better decision.

Further information regarding the verification process can be found at https://wiki.ubuntu.com/QATeam/PerformingSRUVerification . Thank you in advance!

[Brian Murray (brian-murray)]

The lack of verification on this bug and bug 1628750 are preventing the acceptance of another ceph upload into -proposed.

[James Page (james-page)]

Brian

I'd like to stack the fix for bug 1628750 ontop of this SRU, rather than deliver two sets of updates to end users if possible.

[James Page (james-page)]

Oh wait - not bug 1628750 - I'd like to stack bug 1587261

[Brian Murray (brian-murray)]

Hello James, or anyone else affected,

Accepted ceph into xenial-proposed. The package will build now and be available at https://launchpad.net/ubuntu/+source/ceph/10.2.3-0ubuntu0.16.04.2 in a few hours, and then in the -proposed repository.

Please help us by testing this new package. See https://wiki.ubuntu.com/Testing/EnableProposed for documentation on how to enable and use -proposed.Your feedback will aid us getting this update out to other Ubuntu users.

If this package fixes the bug for you, please add a comment to this bug, mentioning the version of the package you tested, and change the tag from verification-needed to verification-done. If it does not fix the bug for you, please add a comment stating that, and change the tag to verification-failed. In either case, details of your testing will help us make a better decision.

Further information regarding the verification process can be found at https://wiki.ubuntu.com/QATeam/PerformingSRUVerification . Thank you in advance!

[Robie Basak (racb)]

The verification of the Stable Release Update for ceph has completed successfully and the package has now been released to -updates. Subsequently, the Ubuntu Stable Release Updates Team is being unsubscribed and will not receive messages about this bug report. In the event that you encounter a regression using the package from -updates please report a new bug using ubuntu-bug and tag the bug report regression-update so we can easily find any regressions.

[Launchpad Janitor (janitor)]

This bug was fixed in the package ceph - 10.2.3-0ubuntu0.16.04.2

---------------
ceph (10.2.3-0ubuntu0.16.04.2) xenial; urgency=medium

  * rgw: Fixes for creation times for buckets (LP: #1587261):
    - d/p/rgw_rados-creation_time.patch: Backport fix from upstream master.
      Fix logic error that leads to creation time being 0 instead of current
      time when creating buckets.

ceph (10.2.3-0ubuntu0.16.04.1) xenial; urgency=medium

  * New upstream stable release (LP: #1628809).
    - d/p/*: Refresh.
    - d/p/rocksdb-flags.patch: Dropped, accepted upstream.
    - d/p/32bit-ftbfs.patch: Cherry pick fix for 32bit arch compat.
    - d/ceph-{fs-common,fuse}.install: Fix install locations
      for mount{.fuse}.ceph.
  * Limit the amount of data per chunk in omap push operations to 64k,
    ensuring that OSD threads don't hit timeouts during recovery
    operations (LP: #1628750):
    - d/p/osd-limit-omap-data-in-push-op.patch: Cherry pick fix from
      upstream master branch.

 -- Frode Nordahl <email address hidden> Fri, 28 Oct 2016 13:50:40 +0200