v0.94.10 Hammer

This Hammer point release fixes several bugs and adds two new features.

We recommend that all hammer v0.94.x users upgrade.

For more detailed information, see the complete changelog.

New Features

ceph-objectstore-tool and ceph-monstore-tool now enable user torebuild the monitor database from OSDs. (This feature is especially useful whenall monitors fail to boot due to leveldb corruption.)

In RADOS Gateway, it is now possible to reshard an existing bucket’s indexusing an off-line tool.

Usage:

$ radosgw-admin bucket reshard –bucket=<bucket_name> –num_shards=<num_shards>

This will create a new linked bucket instance that points to the newly createdindex objects. The old bucket instance still exists and currently it’s up tothe user to manually remove the old bucket index objects. (Note that bucketresharding currently requires that all IO (especially writes) to the specificbucket is quiesced.)

Other Notable Changes

• build/ops: ceph-create-keys loops forever (issue#17753, pr#12805, Alfredo Deza)

• build/ops: improve ceph.in error message (issue#11101, pr#10905, Kefu Chai)

• build/ops: make stop.sh more portable (issue#16918, pr#10569, Mykola Golub)

• build/ops: remove SYSTEMD_RUN from initscript (issue#16440, issue#7627, pr#9873, Vladislav Odintsov)

• cephx: Fix multiple segfaults due to attempts to encrypt or decrypt (issue#16266, pr#11930, Brad Hubbard)

• common: SIGABRT in TrackedOp::dump() via dump_ops_in_flight() (issue#8885, pr#12121, Jianpeng Ma, Zhiqiang Wang, David Zafman)

• common: os/ObjectStore: fix _update_op for split dest_cid (issue#15345, pr#12071, Sage Weil)

• crush: reset bucket->h.items[i] when removing tree item (issue#16525, pr#10724, Kefu Chai)

• doc: add “Upgrading to Hammer” section (issue#17386, pr#11372, Kefu Chai)

• doc: add orphan options to radosgw-admin –help and man page (issue#17281, issue#17280, pr#11140, Abhishek Lekshmanan, Casey Bodley, Ken Dreyer, Thomas Serlin)

• doc: clarify that RGW bucket object versioning is supported (issue#16574, pr#10437, Yuan Zhou, shawn chen)

• librados: bad flags can crash the osd (issue#16012, pr#11936, Jianpeng Ma, Sage Weil)

• librbd: ceph 10.2.2 rbd status on image format 2 returns “(2) No such file or directory” (issue#16887, pr#10987, Jason Dillaman)

• librbd: diffs to clone’s first snapshot should include parent diffs (issue#18068, pr#12446, Jason Dillaman)

• librbd: image.stat() call in librbdpy fails sometimes (issue#17310, pr#11949, Jason Dillaman)

• librbd: request exclusive lock if current owner cannot execute op (issue#16171, pr#12018, Mykola Golub)

• mds: fix cephfs-java ftruncate unit test failure (issue#11258, pr#11939, Yan, Zheng)

• mon: %USED of ceph df is wrong (issue#16933, pr#11934, Kefu Chai)

• mon: MonmapMonitor should return success when MON will be removed (issue#17725, pr#12006, Joao Eduardo Luis)

• mon: OSDMonitor: Missing nearfull flag set (issue#17390, pr#11273, Igor Podoski)

• mon: OSDs marked OUT wrongly after monitor failover (issue#17719, pr#11946, Dong Wu)

• mon: fix memory leak in prepare_beacon (issue#17285, pr#10238, Igor Podoski)

• mon: osd flag health message is misleading (issue#18175, pr#12687, Sage Weil)

• mon: prepare_pgtemp needs to only update up_thru if newer than the existing one (issue#16185, pr#11937, Samuel Just)

• mon: return size_t from MonitorDBStore::Transaction::size() (issue#14217, pr#10904, Kefu Chai)

• mon: send updated monmap to its subscribers (issue#17558, pr#11457, Kefu Chai)

• msgr: OpTracker needs to release the message throttle in _unregistered (issue#14248, pr#11938, Samuel Just)

• msgr: simple/Pipe: error decoding addr (issue#18072, pr#12266, Sage Weil)

• osd: PG::_update_calc_stats wrong for CRUSH_ITEM_NONE up set items (issue#16998, pr#11933, Samuel Just)

• osd: PG::choose_acting valgrind error or ./common/hobject.h: 182: FAILED assert(!max || (*this == hobject_t(hobject_t::get_max()))) (issue#13967, pr#11932, Tao Chang)

• osd: ReplicatedBackend::build_push_op: add a second config to limit omap entries/chunk independently of object data (issue#16128, pr#12417, Wanlong Gao)

• osd: crash on EIO during deep-scrubbing (issue#16034, pr#11935, Nathan Cutler)

• osd: filestore: FALLOC_FL_PUNCH_HOLE must be used with FALLOC_FL_KEEP_SIZE (issue#18446, pr#13041, xinxin shu)

• osd: fix cached_removed_snaps bug in PGPool::update after map gap (issue#18628, issue#15943, pr#12906, Samuel Just)

• osd: fix collection_list shadow return value (issue#17713, pr#11927, Haomai Wang)

• osd: fix fiemap issue in xfs when #extents > 1364 (issue#17610, pr#11615, Kefu Chai, Ning Yao)

• osd: update PGPool to detect map gaps and reset cached_removed_snaps (issue#15943, pr#11676, Samuel Just)

• rbd: export diff should open image as read-only (issue#17671, pr#11948, liyankun)

• rbd: fix parameter check (issue#18237, pr#12312, Yankun Li)

• rbd: fix possible rbd data corruption (issue#16002, pr#11618, Yan, Zheng, Greg Farnum)

• rgw: Anonymous user is able to read bucket with authenticated read ACL (issue#13207, pr#11045, rahul.1aggarwal@gmail.com)

• rgw: COPY broke multipart files uploaded under dumpling (issue#16435, pr#11950, Yehuda Sadeh)

• rgw: TempURL in radosgw behaves now like its Swift’s counterpart. (issue#18316, pr#12619, Radoslaw Zarzynski)

• rgw: default quota fixes (issue#16410, pr#10839, Pavan Rallabhandi, Daniel Gryniewicz)

• rgw: do not abort when accept a CORS request with short origin (issue#18187, pr#12398, LiuYang)

• rgw: do not omap_getvals with (u64)-1 max (issue#17985, pr#12418, Yehuda Sadeh, Sage Weil)

• rgw: fix crash when client posts object with null condition (issue#17635, pr#11809, Yehuda Sadeh)

• rgw: fix inconsistent uid/email handling in radosgw-admin (issue#13598, pr#11952, Matt Benjamin)

• rgw: implement offline resharding command (issue#17745, pr#12227, Yehuda Sadeh, Orit Wasserman, weiqiaomiao)

• rgw: swift: ranged request on a DLO provides wrong values in Content-Range HTTP header (issue#13452, pr#11951, Radoslaw Zarzynski)

• rgw: the value of total_time is wrong in the result of ‘radosgw-admin log show’ opt (issue#17598, pr#11899, weiqiaomiao)

• tests: Cannot clone ceph/s3-tests.git (missing branch) (issue#18384, pr#12744, Orit Wasserman)

• tests: Cannot reserve CentOS 7.2 smithi machines (issue#18401, pr#12762, Nathan Cutler)

• tests: OSDs commit suicide in rbd suite when testing on btrfs (issue#18397, pr#12758, Nathan Cutler)

• tests: Workunits needlessly wget from git.ceph.com (issue#18336, issue#18271, issue#18388, pr#12685, Sage Weil, Nathan Cutler)

• tests: cephfs test failures (ceph.com/qa is broken, should be download.ceph.com/qa) (issue#18574, pr#13022, John Spray)

• tests: merge ceph-qa-suite (pr#12455, Sage Weil)

• tests: objecter_requests workunit fails on wip branches (issue#18393, pr#12759, Sage Weil)

• tests: populate mnt_point in qa/tasks/ceph.py (issue#18383, pr#12743, Nathan Cutler)

• tests: qemu/tests/qemu-iotests/077 fails in dumpling, hammer, and jewel (issue#10773, pr#12423, Jason Dillaman)

• tests: run fs/thrash on xfs instead of btrfs (issue#17151, pr#13039, Nathan Cutler)

• tests: update Ubuntu image url after ceph.com refactor (issue#18542, pr#12957, Jason Dillaman)

• tests: update rbd/singleton/all/formatted-output.yaml to support ceph-ci (issue#18440, [pr#12824 ](http://github.com/ceph/ceph/pull/12824), Venky Shankar, Nathan Cutler)

• tools: add a tool to rebuild mon store from OSD (issue#17179, issue#17400, pr#11125, Kefu Chai, xie xingguo)

• tools: ceph-objectstore-tool crashes if –journal-path (issue#17307, pr#11929, Kefu Chai)

• tools: ceph-objectstore-tool: add a way to split filestore directories offline (issue#17220, pr#11253, Josh Durgin)

• tools: crushtool –compile generates output despite missing item (issue#17306, pr#11931, Kefu Chai)