commit 45107e21c568dd033c2f0a3107dec8f0b0e58374 (tag: refs/tags/v10.2.2) Author: Jenkins Build Slave User Date: Tue Jun 14 11:43:24 2016 +0000 10.2.2 commit 008c5af59b045f9b6b8046a4c6d5e915cbe18bb3 Merge: b733244 dd1ea65 Author: Jason Dillaman Date: Tue Jun 14 07:32:44 2016 -0400 Merge pull request #9692 from dillaman/wip-16292 jewel: librbd: remove should ignore mirror errors from older OSDs Reviewed-by: Jason Dillaman commit dd1ea65c2c8ae966c1f07ce78f0d1f16c9a315d9 Author: Jason Dillaman Date: Mon Jun 13 15:29:43 2016 -0400 librbd: remove should ignore mirror errors from older OSDs Fixes: http://tracker.ceph.com/issues/16268 Signed-off-by: Jason Dillaman (cherry picked from commit 5a97a9b7ef238c4ceb16225a0b937f9a0cdbb2b6) commit b733244b3dc46e11689ddfb46187877c01347d3a Merge: 1feddef 2656881 Author: Jason Dillaman Date: Tue Jun 14 07:24:28 2016 -0400 Merge pull request #9691 from dillaman/wip-16260-jewel jewel: librbd: do not shut down exclusive lock while acquiring' Reviewed-by: Jason Dillaman commit 1feddef07c182d5a50b3aa3c4a7ff87f21dbd3e1 Merge: ffb64cd a38caf9 Author: Jason Dillaman Date: Tue Jun 14 06:57:38 2016 -0400 Merge pull request #9690 from dillaman/wip-16290 jewel: rbd-mirror: FAILED assert(!m_status_watcher) Reviewed-by: Jason Dillaman commit 26568819e331ef3adec6667a801f07058d94845b Author: Jason Dillaman Date: Mon Jun 13 20:20:33 2016 -0400 librbd: track in-flight flush events for journaling The tracking for flush events was accidentally dropped. This can result in the journal being closed too early since it might miss in-flight flush ops. Signed-off-by: Jason Dillaman (cherry picked from commit a1ea1edce9332fa1730625629e3b4e16f02caaf4) commit a85fbb4e9a9a409738e9f031c20a8e2beb1b514f Author: Jason Dillaman Date: Mon Jun 13 12:00:28 2016 -0400 librbd: do not shut down exclusive lock while acquiring Fixes: http://tracker.ceph.com/issues/16260 Signed-off-by: Jason Dillaman (cherry picked from commit c5694fc6766fb8e213c4b65d2cd7b9d066b07ff7) commit a38caf90c98a04d57b81519803271f9d85fb5d56 Author: Mykola Golub Date: Sun Jun 12 20:32:45 2016 +0300 rbd-mirror: FAILED assert(!m_status_watcher) Fixes: http://tracker.ceph.com/issues/16245 Signed-off-by: Mykola Golub (cherry picked from commit b8bfc0f2efb44d3eff7a5fd8bca24516c8815d62) commit ffb64cd0538cee05b4f360fa8334829da268d1b9 Merge: 271406c edef63d Author: Jason Dillaman Date: Tue Jun 14 06:54:51 2016 -0400 Merge pull request #9654 from dillaman/wip-16246 jewel: librbd: recursive lock possible when disabling journaling Reviewed-by: Jason Dillaman commit 271406c77a6176057ba73b55a110a193d6872a48 Merge: b5e344e 0579a48 Author: Jason Dillaman Date: Tue Jun 14 06:54:31 2016 -0400 Merge pull request #9673 from dillaman/wip-16265 jewel: rbd-mirror: do not re-use image id from mirror directory if creating image Reviewed-by: Jason Dillaman commit b5e344e1e3670214da803ff6f7eb6e50bf52e46f Merge: 1720664 739f343 Author: Jason Dillaman Date: Mon Jun 13 23:37:06 2016 -0400 Merge pull request #9674 from dillaman/wip-16267 jewel: qa/workunits: ensure replay has started before checking position Reviewed-by: Jason Dillaman commit 739f343ec880eacd13b935d493b5dbfcc5efa387 Author: Jason Dillaman Date: Mon Jun 13 00:19:37 2016 -0400 qa/workunits: ensure replay has started before checking position Fixes: http://tracker.ceph.com/issues/16248 Signed-off-by: Jason Dillaman (cherry picked from commit 6a68b68e0d2741e6694fd28d31766808ba881815) commit 0579a483bf7b0f1cc590fb1dbc6c740c17679dac Author: Jason Dillaman Date: Mon Jun 13 02:09:00 2016 -0400 rbd-mirror: reset local image id when creating new image Fixes: http://tracker.ceph.com/issues/16253 Signed-off-by: Jason Dillaman (cherry picked from commit 1948047acbc129de2982d53e130da1f577af224b) commit ec809870cc2902d2bdf6543d911e23e71567303c Author: Jason Dillaman Date: Mon Jun 13 02:07:54 2016 -0400 rbd-mirror: handle an attempt to delete a non-mirrored image Signed-off-by: Jason Dillaman (cherry picked from commit 45dda67b3ef3d5dc5abb327fc1f242140f004d90) commit 6c303007af20a317f02b5caf7d3cb9f2214d2ee5 Author: Jason Dillaman Date: Mon Jun 13 02:03:29 2016 -0400 cls_rbd: disallow setting an empty mirror image entry to disabling Signed-off-by: Jason Dillaman (cherry picked from commit 7b2f99a97ff64451f3ae72081186dff41c507271) commit edef63d56bc329638f79ce8017e72810dcd768f1 Author: Jason Dillaman Date: Sun Jun 12 21:42:59 2016 -0400 librbd: avoid holding owner_lock waiting for exclusive lock Fixes: http://tracker.ceph.com/issues/16235 Signed-off-by: Jason Dillaman (cherry picked from commit b96ab805eed49d4bdb494c8919d2e1ee7ad0a7e6) commit 1720664ddcb32dd300b2bd046ff15ed4916676a4 Merge: f902309 2eaaf2e Author: Gregory Farnum Date: Mon Jun 13 01:26:53 2016 -0700 Merge pull request #9656 from gregsfortytwo/wip-jewel-16024 Jewel client: fix bugs accidentally disabling readahead commit f902309bc9c159c0d6a798c0398a95f64bb682ed Merge: 89d6545 d61e3dd Author: Gregory Farnum Date: Mon Jun 13 01:22:22 2016 -0700 Merge pull request #9655 from gregsfortytwo/wip-jewel-15508 Jewel mds: order directories by hash and fix simultaneous readdir races commit d61e3ddf6d99a4b9b72fd769d635bf9b2bffaf0e Author: Yan, Zheng Date: Mon May 9 22:12:48 2016 +0800 client: fix simultaneous readdirs race Current readdir code uses list to track the order of the dentries in readdir replies. When handling a readdir reply, it pushes the resulting dentries to the back of directory's dentry_list. After readdir finishes, the dentry_list reflects how MDS sorts dentries. This method is racy when there are simultaneous readdirs. The fix is use vector instead of list to trace how dentries are sorted in its parent directory. As long as shared_gen doesn't change, each dentry is at fixed position of the vector. So cocurrent readdirs do not affect each other. Fixes: http://tracker.ceph.com/issues/15508 Signed-off-by: Yan, Zheng (cherry picked from commit 9d297c5e98f814b282dadc379ab70dfa678db73e) Signed-off-by: Greg Farnum Date: Mon Apr 25 23:24:46 2016 +0800 client: move dir_{release,ordered}_count into class Inode We close Inode::dir when it's empty. Once closing the dir, we lose track of {release,ordered}_count. This causes direcotry to be wrongly marked as complete. (dir is trimmed to empty in the middle of readdir) Signed-off-by: Yan, Zheng (cherry picked from commit 235fcf68f5d96a4d9d6cc260d12da912fa7ea4a8) Signed-off-by: Greg Farnum Date: Mon Apr 25 17:59:35 2016 +0800 ceph_test_libcephfs: check order of entries in readdir result Signed-off-by: Yan, Zheng (cherry picked from commit 9b17d145c15a469caa29853312b8dcb0b19f55f0) Signed-off-by: Greg Farnum Date: Mon Apr 25 15:52:32 2016 +0800 mds: don't reset readdir offset if client supports hash order dentry Now the ordering of dentries is stable across directory fragmentation. There is no need to reset readdir offset if directory get fragmented in the middle of readdir. Signed-off-by: Yan, Zheng (cherry picked from commit 98a01af2cd8e01f14ab7be32c31a43571ef8bb87) Signed-off-by: Greg Farnum Date: Mon Apr 25 15:31:27 2016 +0800 client: using hash value to compose dentry offset If MDS sorts dentries in dirfrag in hash order, we use hash value to compose dentry offset. dentry offset is: (0xff << 52) | ((24 bits hash) << 28) | (the nth entry hash hash collision) This offset is stable across directory fragmentation. Signed-off-by: Yan, Zheng (cherry picked from commit 680766ec131b95271e320f54dfe6d69ea8d4fbb3) Signed-off-by: Greg Farnum Date: Mon Apr 25 10:50:17 2016 +0800 client: record 'offset' for each entry of dir_result_t::buffer This is preparation for using hash value as dentry 'offset' Signed-off-by: Yan, Zheng (cherry picked from commit bd6546e5e8bfb4ef8927807492d63d5285534618) Signed-off-by: Greg Farnum Date: Mon May 9 11:51:36 2016 +0800 client: fix cached readdir after seekdir Client::seekdir doesn't reset dirp->at_cache_name for a forward seek within same frag. So the dentry with name == at_cache_name may not be the one prior to the readdir postion. Signed-off-by: Yan, Zheng (cherry picked from commit 0e32115bae2f1ac2c59b57852976b0de5587abac) Signed-off-by: Greg Farnum Date: Thu Apr 21 17:31:10 2016 +0800 mds: define end/complete in readdir reply as single u16 flags so that we can introduce new flags for readdir reply. Signed-off-by: Yan, Zheng (cherry picked from commit 92cfbdf526d1030891da366299b635e589dd5f8e) Signed-off-by: Greg Farnum Date: Sat Apr 23 16:03:48 2016 +0800 mds: sort dentries in CDir in hash order This gives us stable ordering of dentries. (Previously ordering of dentries changes after directory gets fragmented) Signed-off-by: Yan, Zheng (cherry picked from commit f483224a0bc70b9eb5f62cd6cbb8b97df5b65953) Signed-off-by: Greg Farnum Date: Fri Apr 22 09:43:54 2016 +0800 client: save readdir result into dir_result_t directly Current code saves the readdir result into MedaRequest, then updates dir_result_t according to MetaRequest. I can't see any reason why we need to do this. Signed-off-by: Yan, Zheng (cherry picked from commit db5d60d1b8d01afd43da09a063f00aa6ab312903) Signed-off-by: Greg Farnum Date: Fri Apr 22 09:31:00 2016 +0800 client: don't allocate dir_result_t::buffer dynamically Signed-off-by: Yan, Zheng (cherry picked from commit c41ceb9f1425432295a52965de8d4a23b15fe5e5) Signed-off-by: Greg Farnum Date: Thu Apr 21 20:53:55 2016 +0800 client: simplify 'offset in frag' don't distinguish leftmost frag from other frags. always use 2 as first entry's offset. Signed-off-by: Yan, Zheng (cherry picked from commit 6572c2a24ea2abec4c7eecef92a292c2cc72c762) Signed-off-by: Greg Farnum Date: Fri Jun 10 17:33:04 2016 -0400 client: move readahead ref get/put to obj constr The readahead context was incrementing the file handle reference count but only decreasing in finish which is not called if readahead is a no-op. This fixes a bug caught in testing where readahead was causing inode to become disconnected: 2016-06-10 19:46:48.953018 7f2a4351be80 1 client.4125 dump_inode: DISCONNECTED inode 10000000502 #10000000502 ref 110000000502.head(faked_ino=2307 ref=1 ll_ref=0 cap_refs={1024=0,2048=0,4096=0,8192=0} open={1=0,2=0} mode=100666 size=4194304/0 mtime=2016-06-10 19:29:45.107417 caps=-(0=pAsLsXsFscr) objectset[10000000502 ts 2/4012653 objects 0 dirty_or_tx 0] 0x7f2a24300d00) 2016-06-10 19:46:48.953032 7f2a4351be80 2 client.4125 cache still has 0+155 items, waiting (for caps to release?) Signed-off-by: Patrick Donnelly (cherry picked from commit b5a0773259ce9f2ef44e8e6925ab11dc1a5e55e6) Signed-off-by: Greg Farnum Date: Thu May 26 16:32:58 2016 -0400 client: use fh min readahead to enable readahead The client was using the configured maximum readahead of 0 which was changed in 95ee69989129750fddce6a3b5644238c4b88ed74. This prevented readahead from ever running (without setting a different default from 0). Fixes: http://tracker.ceph.com/issues/16024 Signed-off-by: Patrick Donnelly (cherry picked from commit 01c179bcce2dc8a269d78eb1be8198b12bd161f7) Signed-off-by: Greg Farnum Date: Fri May 27 11:43:31 2016 -0400 client: update comment Default was changed to unlimited in 11e4680e7394e6e02612441c964e05207b97737b. Signed-off-by: Patrick Donnelly (cherry picked from commit bee02a073f4c551798bc490618a0e8d862893c25) Signed-off-by: Greg Farnum Date: Thu May 26 15:08:52 2016 -0400 client: use layout.get_period() helper Signed-off-by: Patrick Donnelly (cherry picked from commit b331d12a43a4177369a003c3ab9d525da6a89950) Signed-off-by: Greg Farnum Date: Fri Jun 10 15:44:26 2016 -0400 librbd: recursive lock possible when disabling journaling If pool-level mirroring is enabled, deleting the journal could cause a deadlock attempting to delete remote peer sync-point snapshots. Fixes: http://tracker.ceph.com/issues/16235 Signed-off-by: Jason Dillaman (cherry picked from commit fb255e6c3cd44c8d24c53e3cd70395a11a712574) commit 89d65456ed1303b09e9bc054bbb62c0808db4565 Merge: c526386 e15bfff Author: Jason Dillaman Date: Sun Jun 12 08:16:35 2016 -0400 Merge pull request #9553 from Abhishekvrshny/wip-16038-jewel jewel: pybind: rbd API should default features parameter to None Reviewed-by: Jason Dillaman commit c526386b4b087029e6fc00315b7726a2cf9caecb Merge: 6f9f830 6af7b22 Author: Jason Dillaman Date: Sun Jun 12 08:13:16 2016 -0400 Merge pull request #9552 from Abhishekvrshny/wip-15981-jewel jewel: qa: dynamic_features.sh races with image deletion Reviewed-by: Jason Dillaman commit 6f9f8307d7cff313bd2f476236591141d3c45c71 Merge: 7edf393 dc38a26 Author: Jason Dillaman Date: Sun Jun 12 08:08:39 2016 -0400 Merge pull request #9629 from dillaman/wip-16231 jewel: rbd-mirror: fix deletion propagation edge cases Reviewed-by: Jason Dillaman commit 7edf393743b2f7811d784e02a42fce13af19e8bc Merge: 39f530a bc658ed Author: Jason Dillaman Date: Sun Jun 12 08:08:23 2016 -0400 Merge pull request #9630 from dillaman/wip-16233 jewel: rbd-mirror: do not propagate deletions when pool unavailable Reviewed-by: Jason Dillaman commit 39f530a224d85cd12294ce125d8548cbe1e3e1c0 Merge: bd454ff cb50679 Author: Jason Dillaman Date: Sun Jun 12 08:07:42 2016 -0400 Merge pull request #9628 from dillaman/wip-16224 jewel: rbd-mirror: fix journal shut down ordering Reviewed-by: Jason Dillaman commit bd454ff3a95244d2be2177180814439667db73f1 Merge: 3fe2f1b 3a2cd16 Author: Jason Dillaman Date: Sun Jun 12 08:07:10 2016 -0400 Merge pull request #9627 from dillaman/wip-16217 jewel: rbd-mirror: refresh image after creating sync point Reviewed-by: Jason Dillaman commit 3fe2f1b652d7d6d80f9375174885b002db503f40 Merge: 2a9fee9 99ff1fc Author: Sage Weil Date: Fri Jun 10 21:48:26 2016 -0400 Merge pull request #9579 from Abhishekvrshny/wip-16006-jewel jewel: BackoffThrottle spins unnecessarily with very small backoff while the throttle is full Reviewed-by: Sage Weil commit 2a9fee94f10cba7675f564edbeac35620e8a6602 Merge: ac1b354 4a967eb Author: Sage Weil Date: Fri Jun 10 17:34:19 2016 -0400 Merge pull request #9638 from ceph/wip-16211-jewel Jewel: remove reliance on FLAG_OMAP for reads Reviewed-by: Sage Weil commit bc658ede5b2e68210852b66585bfa5560964f405 Author: Jason Dillaman Date: Fri Jun 10 08:15:19 2016 -0400 rbd-mirror: do not propagate deletions when pool unavailable Fixes: http://tracker.ceph.com/issues/16229 Signed-off-by: Jason Dillaman (cherry picked from commit d321634414fa4b90c66474557ec990dfc935ae29) commit dc38a26e2e48f0a1fabdb51f1819e402e5b214ce Author: Jason Dillaman Date: Fri Jun 10 00:12:06 2016 -0400 rbd-mirror: image deleter properly handles multiple snapshots Fixes: http://tracker.ceph.com/issues/16226 Signed-off-by: Jason Dillaman (cherry picked from commit d72a8cb67fbdc7361f8f9e265c659ddb96dc0f74) commit 0a74552b5a4c297855f5e04336a9d5cf00441d9e Author: Jason Dillaman Date: Thu Jun 9 22:32:33 2016 -0400 rbd-mirror: propagate deletions even if image replayer is stopped If an image deletion which causes the replayer to stop (due to an error) before the deletion is detected, the deletion should still occur. Signed-off-by: Jason Dillaman (cherry picked from commit a0c526a23783fe543737f6445a2d9438cc73b137) commit da6d409ab670e45e34c1dcf7de3f775632423d3f Author: Jason Dillaman Date: Thu Jun 9 21:14:27 2016 -0400 qa/workunits/rbd: improve deletion propagation test case Signed-off-by: Jason Dillaman (cherry picked from commit de15b198cd7b5e94016605c8016a5cf38c07c6f6) commit cb50679bad91fa1a0e360dca0331996b543fdadd Author: Jason Dillaman Date: Wed Jun 8 10:12:04 2016 -0400 librbd: quickly check for duplicate snap name upon create During journal replay, quickly check for a duplicate snap name before allocating a snapshot sequence from the OSDs and reverting due to the duplicate name. Signed-off-by: Jason Dillaman (cherry picked from commit 8374d45c7644666103f88c410a2b1b781f3a0cb8) commit 43f3bbf130c9914f895bf4834305f951d0228954 Author: Jason Dillaman Date: Tue Jun 7 14:45:16 2016 -0400 rbd-mirror: stop external replayer before closing remote journaler Signed-off-by: Jason Dillaman (cherry picked from commit 88bd3002082184c045bb82547929b30da1526cfb) commit f0bf15e9d38e3de63c959bac0ae90b609dafc514 Author: Jason Dillaman Date: Tue Jun 7 12:42:52 2016 -0400 librbd: stop journal recorder before starting external replay Fixes: http://tracker.ceph.com/issues/16165 Signed-off-by: Jason Dillaman (cherry picked from commit 5aa6eb2782bdb5cde6a050a5f7a4e683800414c0) commit 3a2cd16e6c35b7b8ba0dd113b9a8cc586c1086b0 Author: Jason Dillaman Date: Wed Jun 8 12:49:04 2016 -0400 rbd-mirror: refresh image after creating sync point snapshot Fixes: http://tracker.ceph.com/issues/16196 Signed-off-by: Jason Dillaman (cherry picked from commit c9a3e3be7ddc31992167d55de12bbe6849447b43) commit 6b77d2268e7b17911f0a6809045d356d97d1621a Author: Jason Dillaman Date: Wed Jun 8 12:03:33 2016 -0400 rbd-mirror: consistently use dout/derr for debug logging Signed-off-by: Jason Dillaman (cherry picked from commit 03f5e4a917b3c18f8ccdaa5ce676ed9731dd2a8f) commit ac1b35478eea6bad574fac8641b47c0fcaee20f4 Merge: 966186f 03466f0 Author: Jason Dillaman Date: Fri Jun 10 12:32:21 2016 -0400 Merge pull request #9612 from dillaman/wip-16208 jewel: rbd-mirror: support bootstrap canceling Reviewed-by: Jason Dillaman commit 966186f3c20080475d3e692e918cc363cac1ae47 Merge: 49ed4ff bddf944 Author: Jason Dillaman Date: Fri Jun 10 12:32:02 2016 -0400 Merge pull request #9610 from dillaman/wip-16209 jewel: librbd: prevent error messages when journal externally disabled Reviewed-by: Jason Dillaman commit 49ed4ffeb2c17f266559137030ab32182750b321 Merge: 8e533c8 0014a9e Author: Jason Dillaman Date: Fri Jun 10 12:19:47 2016 -0400 Merge pull request #9611 from dillaman/wip-16210 jewel: librbd: journal IO error results in failed assertion in AioCompletion Reviewed-by: Jason Dillaman commit 8e533c877e5d7bac84867d418aada1998425ef72 Merge: a768a1c e424482 Author: Jason Dillaman Date: Fri Jun 10 12:16:49 2016 -0400 Merge pull request #9580 from Abhishekvrshny/wip-15996-jewel jewel: librbd: reuse ImageCtx::finisher and SafeTimer for lots of images case Reviewed-by: Jason Dillaman commit a768a1cb7b0062963714361ce94f8a834db44bd4 Merge: ba5d490 bb279f1 Author: Jason Dillaman Date: Fri Jun 10 12:15:51 2016 -0400 Merge pull request #9609 from dillaman/wip-16216 jewel: librbd: refresh image if needed in mirror functions Reviewed-by: Jason Dillaman commit ba5d490824efa63c63713fee418631cd684293c4 Merge: 41c5c27 3b8cb63 Author: Jason Dillaman Date: Fri Jun 10 12:15:32 2016 -0400 Merge pull request #9608 from dillaman/wip-16189 jewel: cls_rbd: mirror image status summary should read full directory Reviewed-by: Jason Dillaman commit 41c5c273b1f52d0284c6b08f15db3edb17cd674c Merge: 7d217b9 cec6870 Author: Jason Dillaman Date: Fri Jun 10 12:15:08 2016 -0400 Merge pull request #9556 from Abhishekvrshny/wip-16162-jewel jewel: ceph_test_librbd_fsx crashes during journal replay shut down Reviewed-by: Jason Dillaman commit 7d217b9cc55778fa73f0ce4a0c2c5994f5f31c87 Merge: 32f90a7 6c0e202 Author: Jason Dillaman Date: Fri Jun 10 12:14:49 2016 -0400 Merge pull request #9554 from Abhishekvrshny/wip-16084-jewel jewel: librbd: validate image metadata configuration overrides Reviewed-by: Jason Dillaman commit 32f90a7c98ebd5a272ece71b1af7009be2742cb5 Merge: 3715d20c 09200d4 Author: Jason Dillaman Date: Fri Jun 10 12:13:06 2016 -0400 Merge pull request #9555 from Abhishekvrshny/wip-16147-jewel jewel: Object Map is showing as invalid, even when Object Map is disabled for that Image. Reviewed-by: Jason Dillaman commit 3715d20c25c19ec1d1c58d275cef4e7ddb5b7778 Merge: 7ea13df 73464af Author: Jason Dillaman Date: Fri Jun 10 12:12:47 2016 -0400 Merge pull request #9550 from Abhishekvrshny/wip-15969-jewel jewel: Initial python APIs to support mirroring Reviewed-by: Jason Dillaman commit 7ea13dfb9d85613d5c4daf92811a786293ffc2be Merge: 2c22131 3084cf3 Author: Jason Dillaman Date: Fri Jun 10 12:12:37 2016 -0400 Merge pull request #9548 from Abhishekvrshny/wip-15963-jewel jewel: Disabling journaling feature results in "Transport endpoint is not connected" error Reviewed-by: Jason Dillaman commit 2c221318e27fb4c4b68780c0037b9acae5862d6e Merge: 4d68ae0 69a9500 Author: Loic Dachary Date: Fri Jun 10 00:55:47 2016 +0200 Merge pull request #9424 from SUSE/wip-16107-jewel jewel: ceph gid mismatch on upgrade from hammer with pre-existing ceph user (SUSE) Reviewed-by: Samuel Just Reviewed-by: Loic Dachary commit 4d68ae062b90599c184282b807f7d16cfaf2f6ad Merge: 57b4542 b443fdf Author: Loic Dachary Date: Fri Jun 10 00:54:52 2016 +0200 Merge pull request #8802 from SUSE/wip-15606-jewel jewel: ceph-{mds,mon,osd,radosgw} systemd unit files need wants=time-sync.target Reviewed-by: Samuel Just Reviewed-by: Loic Dachary commit 57b45426b5efa7ec4e293f9276f1f4299b5bf3ac Merge: 3cf6533 555cec9 Author: Samuel Just Date: Thu Jun 9 15:44:13 2016 -0700 Merge pull request #9320 from SUSE/wip-16017-jewel jewel: ceph: cli: exception when pool name has non-ascii characters Reviewed-by: Samuel Just commit 3cf653328c988da5cf6be678cf3f0cb4c80c094d Merge: a422d03 663d7c2 Author: Samuel Just Date: Thu Jun 9 15:43:48 2016 -0700 Merge pull request #9270 from Abhishekvrshny/wip-15852-jewel jewel: mon/Monitor: memory leak on Monitor::handle_ping() Reviewed-by: Samuel Just commit a422d0372719841ea4cc63f87e77ddc1dee7d815 Merge: 34b193b 8356021 Author: Samuel Just Date: Thu Jun 9 15:43:04 2016 -0700 Merge pull request #9237 from SUSE/wip-15961-jewel jewel: ceph-osd valgrind invalid reads/writes Reviewed-by: Samuel Just commit 34b193be043f0d88c42c5ae177a660b747e81486 Merge: e2a035e 64f15b3 Author: Samuel Just Date: Thu Jun 9 15:42:32 2016 -0700 Merge pull request #9103 from Abhishekvrshny/wip-15797-jewel jewel: deadlock in OSD::_committed_osd_maps Reviewed-by: Samuel Just commit e2a035eeb19e9043c41e5132d2ea0b8a59ac1bed Merge: 2c35fe7 adfbe95 Author: Samuel Just Date: Thu Jun 9 15:41:52 2016 -0700 Merge pull request #9101 from Abhishekvrshny/wip-15799-jewel jewel: osd boot race with noup being set Reviewed-by: Samuel Just commit 2c35fe7e739e99452cf4d039281185e8c4a40b1c Merge: 6528e43 f4306de Author: Samuel Just Date: Thu Jun 9 15:40:43 2016 -0700 Merge pull request #9104 from Abhishekvrshny/wip-15795-jewel jewel: LibRadosWatchNotifyPPTests/LibRadosWatchNotifyPP.WatchNotify2Timeout/1 segv Reviewed-by: Samuel Just commit 6528e4371e0ab1c385e539646d08f4ea1dded893 Merge: 50143ed b989084 Author: Samuel Just Date: Thu Jun 9 15:37:43 2016 -0700 Merge pull request #9427 from SUSE/wip-16104-jewel jewel: ceph-disk: workaround gperftool hang Reviewed-by: Loic Dachary commit 50143ed004ef198f2161b9f360bfeccd16503954 Merge: b328e99 a32820d Author: Sage Weil Date: Thu Jun 9 14:55:31 2016 -0400 Merge pull request #9614 from ceph/wip-16113-jewel osd: hobject_t::get_max() vs is_max() discrepancy Reviewed-by: Sage Weil commit 03466f096adcfe3322cad49c959fec976d887860 Author: Mykola Golub Date: Wed May 4 15:30:05 2016 +0300 rbd-mirror: don't return split-brain error if we still in MIRROR_PEER_STATE_SYNCING Signed-off-by: Mykola Golub (cherry picked from commit 3ac620c5d66d8516e68264fcdb5e97b0e803d0ec) commit dfefd343e35b1c69fc2c938c7ebc30352e1320d4 Author: Mykola Golub Date: Tue May 3 12:48:06 2016 +0300 rbd-mirror: refactor split-brain detection code to make it more readable and extendable. Signed-off-by: Mykola Golub (cherry picked from commit 039cd2d2e3bb6a9c5042c125d2c67ec9adc02c47) commit c94e5b740628b98008e5342b8a3b2b888ff7b794 Author: Mykola Golub Date: Tue May 31 15:00:30 2016 +0300 rbd-mirror: support bootstrap canceling Signed-off-by: Mykola Golub (cherry picked from commit 21f895f48498e518c3675a9e851559f4d071f6dd) commit c8a3e759298c3a2adfa67d932178d284619d460e Author: Mykola Golub Date: Mon May 30 16:24:24 2016 +0300 rbd-mirror: return ECANCELED when ImageReplayer start is canceled Signed-off-by: Mykola Golub (cherry picked from commit f581dbcb2eca108ce9440629b8f00f3229c07b95) commit 0014a9e284c6cfc7678e63b56d707a8ad9b93127 Author: Jason Dillaman Date: Tue Jun 7 01:01:09 2016 -0400 librbd: flush journal commit positions before starting op Ensure all IO has been properly flushed and committed to the journal before starting an op that could affect the IO path. Signed-off-by: Jason Dillaman (cherry picked from commit 636129653c7cac9665d0c423f6aafad5fd0c480a) commit 3cec45a17a3d8bdcd19c63c31dad6b9150a9fbfd Author: Jason Dillaman Date: Tue Jun 7 00:11:34 2016 -0400 librbd: leaked AioCompletion if IO issued after image shut down Signed-off-by: Jason Dillaman (cherry picked from commit d3d4e068396e01c58832e0ea1224ca34bafb808e) commit ef12536830082194da52df3a0dc8e22a3f69cb54 Author: Jason Dillaman Date: Mon Jun 6 19:00:47 2016 -0400 librbd: resize and snap create can hang on journal replay failure Signed-off-by: Jason Dillaman (cherry picked from commit 1032f19b3c9d9c7916669d04af909131574b3cab) commit c8b4cabea0e1a74a62fd9b25ccb5e0b15df38733 Author: Jason Dillaman Date: Mon Jun 6 14:54:23 2016 -0400 librbd: missing journal state for flushing/restarting replay Fixes: http://tracker.ceph.com/issues/16077 Signed-off-by: Jason Dillaman (cherry picked from commit a7f5bc181c9cd677d39f95ad78134e0f85c22acb) commit bc7649be5867fca46c1e86dacb99c1f8b18ad2e8 Author: Jason Dillaman Date: Thu May 12 21:22:56 2016 -0400 librbd: track IO in the order issued Fixes: http://tracker.ceph.com/issues/15034 Signed-off-by: Jason Dillaman (cherry picked from commit eb8dda5296eb5c19aa27e96d6e92eaa0c352cb62) commit 602ffd78f3de00d6655236dd914aa83006a8f862 Author: Jason Dillaman Date: Thu May 12 15:15:04 2016 -0400 librbd: AioCompletion should retrieve CephContext from ImageCtx Signed-off-by: Jason Dillaman (cherry picked from commit 133b1ffb21c982003c7aa9dafd585062f9b2eb51) commit 6e20662a9f5b9eabbd83eb91760532d8171d96b8 Author: Hector Martin Date: Wed May 25 20:41:40 2016 +0900 librbd: only mark async completion as done after callback This change in behavior was introduced in fde9f78. Signed-off-by: Hector Martin (cherry picked from commit 77f7c1cf829e5417eb356ebf3bf1fbc8ff7a25a5) commit de29be73c043c0cd49b4d35a160ddd3910d423e8 Author: Jason Dillaman Date: Mon May 9 18:17:49 2016 -0400 librbd: Journal no longer requires AioCompletion for IO events Signed-off-by: Jason Dillaman (cherry picked from commit 82587a31f1d8ba1765f164391796753cf97a1878) commit 65556aa507e3e70352bb61c1549f340af0c4a9fb Author: Jason Dillaman Date: Mon May 9 17:59:09 2016 -0400 librbd: bubble journal error up via existing AioObjectRequests The journal cannot directly fail the AioCompletion because it might have one-or-more outstanding requests associated with it. Fixes: http://tracker.ceph.com/issues/15791 Signed-off-by: Jason Dillaman (cherry picked from commit 79c934742fb15c803780519b36070408599a74c5) commit bddf944797df5ffb93b587f8ac245846a48bd2e3 Author: Jason Dillaman Date: Mon Jun 6 22:34:30 2016 -0400 librbd: potential duplicate snap removal can result in crash Signed-off-by: Jason Dillaman (cherry picked from commit 276fed6b7078158c2cd04fba14a11531c27898e0) commit 4ce4130d0e6deef315dcbd5dc899faf5de9ab0e7 Author: Jason Dillaman Date: Thu Jun 2 00:43:53 2016 -0400 ObjectCacher: do not merge uncommitted journal bhs Avoid the possibility of an overwrite callback for regions that were never associated with a journal write. Signed-off-by: Jason Dillaman (cherry picked from commit c6060ade16205337de3941a10d05e234b738fa07) commit 6e645985e77bd0597eab6755b6e4071b79967584 Author: Zhiqiang Wang Date: Wed Feb 24 21:54:34 2016 +0800 ObjectCacher: don't loop the extents of OSDRead in map_read Looping the extents of OSDRead is done in _readx. The looping in map_read is not needed. Signed-off-by: Zhiqiang Wang (cherry picked from commit fae912c3856cab6d03183411eecab90c6e6732d2) commit c5f281d64221c62e3736f9630204f85d67328670 Author: Zhiqiang Wang Date: Wed Feb 24 21:49:56 2016 +0800 ObjectCacher: fix the bytes_written in writex The BufferHead returned after map_write may include other dirty data to the left and/or right. The actual length of data been written is in the ObjectExtent. Signed-off-by: Zhiqiang Wang (cherry picked from commit ca6f1681e4baf61682e258c5761536d67bf6bdb8) commit 1c00d813791cdd8c6e685c1ebf6f93ef4aa0adbe Author: Zhiqiang Wang Date: Wed Feb 24 21:45:45 2016 +0800 osdc: don't loop the extents of OSDWrite in map_write We are looping all the extents of the OSDWrite in writex. The looping in map_write is not needed. And if there are same object in these extents,they will be mapped multiple times. Signed-off-by: Zhiqiang Wang (cherry picked from commit 322f1059670898762a890a01af6d3b0dd3ccf826) commit b48cdb21da2ba02ae4ee5398c38b48e857dba8cf Author: Jason Dillaman Date: Thu Jun 2 00:42:30 2016 -0400 librbd: notify image feature update before releasing lock If a client is waiting for the lock, ensure it sees the latest version of the image features. Signed-off-by: Jason Dillaman (cherry picked from commit f601ea835bb78554ac767ccb11789df960765bf9) commit 680efbdcc3d2e28b6de991e7ad5c18e0d71cc551 Author: Jason Dillaman Date: Wed Jun 1 23:19:20 2016 -0400 librbd: refresh image after acquiring exclusive lock It's possible that the object map or journaling features have been disabled while the lock was not owned. Fixes: http://tracker.ceph.com/issues/16114 Signed-off-by: Jason Dillaman (cherry picked from commit 03c54f52d15d6283c630bac6f75427e6829f7d0a) commit bb279f1b288f1069b83d7661588f883716c80d9d Author: Jon Bernard Date: Wed Jun 1 21:01:30 2016 -0400 librbd: refresh image if needed in mirror functions Fixes: http://tracker.ceph.com/issues/16096 Signed-off-by: Jon Bernard (cherry picked from commit a355c8f6d9e792517a8d0e096a99a64474369cfc) commit 3b8cb634b34dd7742bbdea2ca354a358cb0fe089 Author: Jason Dillaman Date: Tue Jun 7 09:36:02 2016 -0400 cls_rbd: mirror image status summary should read full directory Previously only retrieved the status for the first 64 images in the rbd_mirroring directory. Fixes: http://tracker.ceph.com/issues/16178 Signed-off-by: Jason Dillaman (cherry picked from commit 61f86e7ec14233ea0dee5b3370595ce050ee28c3) commit 4a967eb42a7409b9d78d569585af2155414234d2 (refs/remotes/origin/wip-16211-jewel) Author: Samuel Just Date: Thu Jun 9 10:36:44 2016 -0700 Revert "osd/ReplicatedPG: for copy_get get omap, firstly check ob whether has omap." This reverts commit 771c869651132013054599f3decfc1017092a25f. commit fd8f8af2456bc197877793a60d4401751f828032 Author: Samuel Just Date: Thu Jun 9 10:36:39 2016 -0700 Revert "osd/ReplicatedPG: For omap read ops, it should check object wether has omap" This reverts commit 98caba1d0b4dd334748e3bcaf1bfb3f79f75eb65. commit d59ca31a4fcc472809b1f1bd3379fc3521c28fb0 Author: Samuel Just Date: Thu Jun 9 10:36:31 2016 -0700 Revert "osd/ReplicatedPG: When do omapclear, it should check object whether is omap." This reverts commit aeb8141c29cdd337961430bbf3e93a79e3e1bcd3. Conflicts: src/osd/ReplicatedPG.cc commit b328e998b2aa066f5f42577abd0c74a5f3d42694 Merge: b1c4c4c 188318b Author: Nathan Cutler Date: Wed Jun 8 12:00:31 2016 +0200 Merge pull request #8938 from SUSE/wip-15739-jewel jewel: ceph.spec.in: drop support for RHEL<7 and SUSE<1210 in jewel and above Reviewed-by: Boris Ranto Reviewed-by: Ken Dreyer commit f4306de66d6c071cb240a0d77e8f877841f87033 Author: Sage Weil Date: Fri May 6 09:09:43 2016 -0400 osdc/Objecter: upper bound watch_check result This way we always return a safe upper bound on the amount of time since we did a check. Among other things, this prevents us from returning a value of 0, which is confusing. Fixes: http://tracker.ceph.com/issues/15760 Signed-off-by: Sage Weil (cherry picked from commit 2e2ce365866a7569b55aff338a82c132aa927089) commit 64f15b3cfd15faa65e7a36c9779713ea2436cf16 Author: Xinze Chi Date: Wed May 4 09:49:40 2016 +0800 OSD: fix deadlock in OSD::_committed_osd_maps Fixes: http://tracker.ceph.com/issues/15701 Signed-off-by: Xinze Chi (cherry picked from commit 4fbf3891aab16347d02782f7c1c0400f9e0896af) commit adfbe9554be2cb79a6cbccc7c4a556139b4aed50 Author: Sage Weil Date: Mon May 2 23:28:18 2016 -0400 osd: handle boot racing with NOUP set This is a follow-on to 7139a232d26beef441ffbc13bc087baab3505ea8, which handled the NOUP set + clear case when the OSD found out about the flag being cleared. However, it's possible that the flag will get cleared but the OSD won't get a map update (because it hasn't subscribed and is not doing any work). This means that it is *more* likely than before that we will restart the boot process even though the OSD did successfully mark us up. However, as before, it is unavoidable because there is no notification of whether our boot request succeeds or not. And it is still mostly harmless (an extra mark down + up cycle). Fixes: http://tracker.ceph.com/issues/15678 Signed-off-by: Sage Weil (cherry picked from commit 11e4242fbdb2f2f6f654d4cb3a7c95d5b38a88c2) commit e424482d0e79ad0a0c44b14f0977cfa69d1b7f72 Author: Haomai Wang Date: Sun Feb 28 23:35:17 2016 +0800 TaskFinisher: use singleton SafeTimer and Finisher Fix #13938 Signed-off-by: Haomai Wang (cherry picked from commit dc6d6ce8b29de51b69242e569517f69acfe226f4) commit 99ff1fc34215da3339ccd227c4f17caf8294a32f Author: Samuel Just Date: Wed May 18 12:09:10 2016 -0700 BackoffThrottle: wait() if past max Otherwise, we risk spinning on wait_for() with a small delay. Fixes: http://tracker.ceph.com/issues/15953 Signed-off-by: Samuel Just (cherry picked from commit 6f835726a7df5b29a96849fc1362186fd2116d9b) commit 5ce43eb7682535deeee149600f1226b87e41defd Author: Samuel Just Date: Wed May 18 11:49:19 2016 -0700 BackoffThrottle: use wait_for instead of wait_until On some platforms, wait_until won't surrender the lock with a negative argument. Signed-off-by: Samuel Just (cherry picked from commit 846808ff3f3837b1772f293ae71dba5eda609cc4) commit cec6870572621ed9abbe3791ba6e1a460333dfc0 Author: Jason Dillaman Date: Thu Jun 2 10:41:18 2016 -0400 test/rbd: fsx needs to shut down the journal Fixes: http://tracker.ceph.com/issues/16123 Signed-off-by: Jason Dillaman (cherry picked from commit d10ffaafbd2a5831154757b848ed8504dad99069) commit 09200d4b5002f8ef9e75a9218fe85aaa5da33a8b Author: xinxin shu Date: Thu Jun 2 14:13:09 2016 +0800 remove invalid objectmap flag when objectmap is disabled Fixes: http://tracker.ceph.com/issues/16076 Signed-off-by: xinxin shu (cherry picked from commit b2d475686ee7617bb2023d753941e3d6952f0878) commit 6c0e202573110de522d531e0aa8978e843b5167e Author: zhuangzeqiang Date: Sat Oct 7 16:52:07 2017 +0800 rbd: check value of config override when setting image-meta Fixes: http://tracker.ceph.com/issues/15522 Signed-off-by: zhuangzeqiang (cherry picked from commit 4538f8152d0e73bdefc09874113c87467bdda622) commit e15bfff2e822e15e0a34d09825e04cee381d8737 Author: Mykola Golub Date: Mon May 23 09:18:34 2016 +0300 pybind/rbd: create/clone/copy: default to None for features param If None is specified don't set the features so that the defaults is used. Fixes: http://tracker.ceph.com/issues/15982 Signed-off-by: Mykola Golub (cherry picked from commit b8b3a9757f7d517210a91f2893b0961b6b35afab) commit f9e32ac68a83932945165612dc64c307f0411b19 Author: Mykola Golub Date: Mon May 23 09:16:38 2016 +0300 librbd: clone: default to parent features Fixes: http://tracker.ceph.com/issues/15982 Signed-off-by: Mykola Golub (cherry picked from commit d305eaad8d57a6f64e6d80eebcd0839a65c9a668) commit 6af7b223f3d0a8d51cd472cb7024a5ec70d30f61 Author: Mykola Golub Date: Thu May 12 09:15:53 2016 +0300 qa: dynamic_features.sh: return error only if it failed on alive QEMU Fixes: #15500 Signed-off-by: Mykola Golub (cherry picked from commit 8a71a79d24f2e1f999a5fc0a4154a10401d5a261) commit 73464af844b45137049b0fff294449ed3a6f13bb Author: Mykola Golub Date: Wed May 11 14:33:00 2016 +0300 test: initial python APIs to support mirroring Signed-off-by: Mykola Golub (cherry picked from commit bfad0ca33b9d65dcec65bc1a18596d35fb3098f3) commit 5c312669ba2d208acaa8e819812e50ce23418208 Author: Mykola Golub Date: Wed May 11 14:29:17 2016 +0300 rbd: initial python APIs to support mirroring Fixes: http://tracker.ceph.com/issues/15656 Signed-off-by: Mykola Golub (cherry picked from commit ef0ea8ee3c926a31b54e410c18e887415f6ea3cf) commit 3084cf32463a07cecec66befa1562a187bc1734f Author: Yuan Zhou Date: Mon May 16 16:18:59 2016 +0800 rbd: close journal before removing Otherwise there is some unwanted error message. Fixes: http://tracker.ceph.com/issues/15863 Signed-off-by: Yuan Zhou (cherry picked from commit ecdad4acca635d4b9f0b407889dde9985ab3a506) commit a32820d15fd5d2f817bc5fc94b731f4e7adafbf3 (refs/remotes/origin/wip-16113-jewel) Author: Samuel Just Date: Thu Jun 2 17:13:09 2016 -0700 src/: remove all direct comparisons to get_max() get_max() now returns a special singleton type from which hobject_t's can be assigned and constructed, but which cannot be directly compared. This patch also cleans up all such uses to use is_max() instead. This should prevent some issues like 16113 by preventing us from checking for max-ness by comparing against a sentinel value. The more complete fix will be to make all fields of hobject_t private and enforce a canonical max() representation that way. That patch will be hard to backport, however, so we'll settle for this for now. Fixes: http://tracker.ceph.com/issues/16113 Signed-off-by: Samuel Just (cherry picked from commit 7c5f2acaa57bd6caaf4b13b48154df3ad6fbe84d) commit f869594332fb8ffa4d29aae22573afd094c39f54 Author: Samuel Just Date: Thu Jun 2 17:39:09 2016 -0700 PG::replica_scrub: don't adjust pool on max object Signed-off-by: Samuel Just (cherry picked from commit 8b7b7c7a2379ce77173bbdfd90c98c561d786155) commit 1737ff3f49080904ef07a7519d200ee608fab446 Author: Samuel Just Date: Thu Jun 2 17:36:21 2016 -0700 hobject: compensate for non-canonical hobject_t::get_max() encodings This closes a loop-hole that could allow a non-canonical in memory hobject_t::get_max() object which would return true for is_max(), but false for *this == hobject_t::get_max(). Fixes: http://tracker.ceph.com/issues/16113 Signed-off-by: Samuel Just (cherry picked from commit 42fad4b76def50a0984bf3de06f78ed434d56954) commit b1c4c4c456d1d20ccc91de4f762dbe46d8e339a3 Merge: 24404c9 5d9ee88 Author: John Spray Date: Fri Jun 3 12:37:00 2016 +0100 Merge pull request #9430 from ceph/wip-manila-backports jewel: backports needed for Manila Reviewed-by: John Spray commit 24404c928b878d28a34439b8adaf43854547d317 Merge: 1aa81ce 208659f Author: Jason Dillaman Date: Thu Jun 2 12:03:12 2016 -0400 Merge pull request #9423 from dillaman/wip-16101 jewel: rbd-mirror: replicate cloned images Reviewed-by: Jason Dillaman commit 5d9ee88e2a0ee57455581350bafe613fef8e3da2 Author: Xiaoxi Chen Date: Tue Apr 26 17:51:48 2016 -0500 pybind: configurable cephfs_vol_client prefix and ns_prefix. Which is useful if we would like to have multiple backends in manila and potentially set different layout based on prefix. And also make ns_prefix configurable also in order to support multiple backend. Fixes #15417 Signed-off-by: Xiaoxi Chen (cherry picked from commit d1a21b07ede77ab3f2be7a753426d76de0343086) commit 470605c38f772e5bcc466e71ac232294b41be276 Author: Ramana Raja Date: Wed Apr 13 14:03:51 2016 +0530 ceph_volume_client: evict client also based on mount path Evict clients based on not just their auth ID, but also based on the volume path mounted. This is needed for the Manila use-case, where the clients using an auth ID are denied further access to a share. Fixes: http://tracker.ceph.com/issues/15855 Signed-off-by: Ramana Raja (cherry picked from commit 68714b9dda8ed6010cad6730e0ddf54b3219150d) (cherry picked from commit aaa0e9aea9fcf2f63e73e544bd40c10a0d694408) commit 726292eb305d58661993803dea613e9d98a64f4f Author: John Spray Date: Wed May 11 13:18:23 2016 +0100 client: report root's quota in statfs When user is mounted a quota-restricted inode as the root, report that inode's quota status as the filesystem statistics in statfs. This allows us to have a fairly convincing illusion that someone has a filesystem to themselves, when they're really mounting a restricted part of the larger global filesystem. Fixes: http://tracker.ceph.com/issues/15599 Signed-off-by: John Spray (cherry picked from commit b6d2b6d1a51969c210ae75fef93c71ac21f511a6) commit 46c2bd0864f8f7ace80a8cad07b873d9aa57ead9 Author: John Spray Date: Fri Apr 1 14:27:31 2016 +0100 pybind: fix unicode handling in CephFSVolumeClient::purge os.path.join is sensitive to string encoding, but just doing a straight substitution should not be. Signed-off-by: John Spray (cherry picked from commit e590f4d05fdb46747e83e35e66a26d9f4aa0314d) commit b989084dfc28659445fe08847537f8e5ad275cdd Author: Loic Dachary Date: Thu May 26 12:55:51 2016 +0200 ceph-disk: workaround gperftool hang Temporary workaround: if ceph-osd --mkfs does not complete within 5 minutes, assume it is blocked because of https://github.com/gperftools/gperftools/issues/786 References http://tracker.ceph.com/issues/13522 Signed-off-by: Loic Dachary (cherry picked from commit c092321c24b9ca493d90a9ccc8f0b4b9a38677f1) commit 69a9500d900b2d5b9bdb0e989f83fe60ddddeda0 Author: Nathan Cutler Date: Thu May 12 19:23:31 2016 +0200 rpm: unconditionally set ceph user's primary group to ceph (SUSE) This commit brings the user/group creation into greater semantic alignment with the Debian packaging. Fixes: http://tracker.ceph.com/issues/15869 Signed-off-by: Nathan Cutler (cherry picked from commit b3dfa8d90962e60b6841555db830c8ea635f4cce) commit 208659fdebc888cab35239ec848f27320477420f Author: Jason Dillaman Date: Thu May 26 12:49:45 2016 -0400 qa/workunits/rbd: basic cloned image test Signed-off-by: Jason Dillaman (cherry picked from commit 943ff829874f96f720d6c6299e1212e4da24c6c5) commit af6be1bc15dfc2757cf74772078dd8516ed889be Author: Jason Dillaman Date: Thu May 26 08:22:16 2016 -0400 rbd-mirror: copy snapshot parent image settings Signed-off-by: Jason Dillaman (cherry picked from commit f5fd09373714ebe8886bcf51bb97732038fffe55) commit 8405c2c377c7f9e755a349bfc90494f7ee7cc477 Author: Jason Dillaman Date: Thu May 26 12:01:15 2016 -0400 cls_rbd: asynchronous helper method to set parent Signed-off-by: Jason Dillaman (cherry picked from commit fcfdebb471650a1730306f6a309f1775ba7a5eee) commit 26f4edcbb3b24eed98f111a2c77a5380cbf16927 Author: Jason Dillaman Date: Wed May 25 17:38:06 2016 -0400 rbd-mirror: add support for cloning images from mirrored parent Fixes: http://tracker.ceph.com/issues/14937 Signed-off-by: Jason Dillaman (cherry picked from commit ae6e03dfb1fa3be61e6deac4fe39982e7e7c18aa) commit b8b01c57949955b1ff6a6a35dcae05108a2dda58 Author: Jason Dillaman Date: Wed May 25 17:36:06 2016 -0400 librbd: extend internal API to clone from open parent image Signed-off-by: Jason Dillaman (cherry picked from commit bfaa112c30da7091fd2d539b80a4a211176a6031) commit 9081c58a31aa8fa6f966f3b682b9870e2b6f8297 Author: Jason Dillaman Date: Wed May 25 17:13:59 2016 -0400 cls_rbd: asynchronous helper methods to retrieve mirror image metadata Signed-off-by: Jason Dillaman (cherry picked from commit 5a26080447cfa4b7bfdacdc13242569258a64076) commit 2c2f4b251d0c8b870728293d06e9c853856c8092 Author: Jason Dillaman Date: Wed May 25 15:16:28 2016 -0400 rbd-mirror: helper state machine to open remote/read-only images Signed-off-by: Jason Dillaman (cherry picked from commit 3731c3c34c15f8dcfbab444bdbf3c05976b03431) commit 1aa81ce5ba492e8eeea2b20f08b77a18a94cd738 Merge: 54bfefb 0bcc295 Author: Jason Dillaman Date: Tue May 31 13:11:22 2016 -0400 Merge pull request #9409 from dillaman/wip-16068 jewel: rbd-mirror: support multiple replicated pools Reviewed-by: Jason Dillaman commit 0bcc295363d9ca20d43fd313084370158513bc62 Author: Jason Dillaman Date: Fri May 27 11:40:58 2016 -0400 rbd-mirror: cluster-level asok commands need to support multiple pools Signed-off-by: Jason Dillaman (cherry picked from commit f4339ee8cb64d5f1a164e017fea942600ffb0ea9) commit e4547c036ef641d48ab03808f59e035b56136745 Author: Jason Dillaman Date: Thu May 26 23:26:20 2016 -0400 rbd-mirror: replayer should only handle a single pool The pool watcher now only needs to scan a single pool for its associated replayer since a peer is inherently tied to a single pool. Fixes: http://tracker.ceph.com/issues/16045 Signed-off-by: Jason Dillaman (cherry picked from commit c5a8b780edeb9b62a601f0718459837f3d66e63d) commit 6fc5328e93b7d005479f925cca1268e7e7e10234 Author: Jason Dillaman Date: Thu May 26 16:03:59 2016 -0400 rbd-mirror: group peers by local pools The peer structure is automatically unique per pool due to its UUID, so grouping local pools by a single peer doesn't work. Signed-off-by: Jason Dillaman (cherry picked from commit 020f0282a0771c0d321732a62f53cee10a193715) Conflicts: src/tools/rbd_mirror/ClusterWatcher.cc: trivial resolution commit 0745cc3db35a349799891c8397249bf8d088ef83 Author: Jason Dillaman Date: Wed May 25 14:00:34 2016 -0400 rbd-mirror: stop stale replayers before starting new replayers If the connection details are tweaked for a remote peer, stop the existing replayer before potentially starting a new replayer against the same remote. Signed-off-by: Jason Dillaman (cherry picked from commit 6a2adca2a47b384f64c20996b9fc77010a70bc22) commit 91f8373613a38d48d71f2762e75e21ec817305be Author: Jason Dillaman Date: Thu May 26 13:29:49 2016 -0400 rbd-mirror: normalize debug log message prefix Signed-off-by: Jason Dillaman (cherry picked from commit d52843190d84d533ab50a1702cf4a829b71ad68b) commit 5f0d89a9258fdf2f7392c63ca5d6488c4e3f210d Author: Jason Dillaman Date: Thu May 26 13:26:57 2016 -0400 qa/workunits/rbd: create secondary replicated pool Signed-off-by: Jason Dillaman (cherry picked from commit 839544ee6446b959f01290667592e44fc5ee6991) commit 54bfefb7d108667058eda56fda24b0029a560746 Merge: 522c7f9 61542c1 Author: Jason Dillaman Date: Tue May 31 11:51:05 2016 -0400 Merge pull request #9373 from dillaman/wip-16055 jewel: journal: support asynchronous shutdown Reviewed-by: Jason Dillaman commit 522c7f9432e029f4db2b7455fb83b4aeb3ff4a4e Merge: f04972c 6375c78 Author: Jason Dillaman Date: Tue May 31 11:49:39 2016 -0400 Merge pull request #9372 from dillaman/wip-15995 jewel: rbd-mirror: Delete local image mirror when remote image mirroring is disabled Reviewed-by: Jason Dillaman commit 61542c14d34c4192f1468e2d7865fe913cce21b8 Author: Jason Dillaman Date: Wed May 25 02:31:11 2016 -0400 rbd-mirror: image replay now uses asynchronous journal shutdown Signed-off-by: Jason Dillaman (cherry picked from commit a1b0a1b1b9a237ae363a52c7b4038b19e87052a5) commit bf4c45830330b025bf818286e13693b837f1f2c5 Author: Jason Dillaman Date: Wed May 25 00:21:14 2016 -0400 librbd: integrate with async journaler shutdown API Signed-off-by: Jason Dillaman (cherry picked from commit ad297850b1be8ed38f77b86913c6821748f3368b) commit 36cf42a3f47c23db82a8aca7a5560857cc3e0027 Author: Jason Dillaman Date: Tue May 24 16:12:16 2016 -0400 journal: extend asynchronous shutdown to facade Fixes: http://tracker.ceph.com/issues/14530 Signed-off-by: Jason Dillaman (cherry picked from commit f40c595255a6828e27abfe63b4209ebae03e237b) commit 1b0422d7272788561253479a43f5b3dcdbb6282b Author: Jason Dillaman Date: Tue May 24 15:53:12 2016 -0400 journal: trimmer now has asynchronous shutdown Signed-off-by: Jason Dillaman (cherry picked from commit 59ca9916ba2e2b144eff5ee471a314b9c4391936) commit f913946ff74924fc14c7336fef6c73a4c6edfdfa Author: Jason Dillaman Date: Tue May 24 15:36:17 2016 -0400 journal: metadata init and shutdown now asynchronous Signed-off-by: Jason Dillaman (cherry picked from commit 0543e1b29eb253bf1d9e9af7af6f22fa394dd854) commit 1a9d60f8244d6ba8697b2254d967907fdb618cdb Author: Jason Dillaman Date: Tue May 24 12:06:26 2016 -0400 journal: player shutdown is now handled asynchronously Fixes: http://tracker.ceph.com/issues/15949 Signed-off-by: Jason Dillaman (cherry picked from commit 79b41346678b3acdeb547fe07e44d8d0543d37a7) commit 4af5bb1d2d86de68be0cef3713e09dfe0f1a9030 Author: Jason Dillaman Date: Mon May 23 22:26:05 2016 -0400 test: use randomized write sizes for rbd-mirror stress test Signed-off-by: Jason Dillaman (cherry picked from commit b9edff2174ff04bb8b6353dd2d2218c64757aae1) commit df2b8d9f0f399dcf9bb0e3eba3a39e14de504c49 Author: Jason Dillaman Date: Mon May 23 22:21:33 2016 -0400 journal: eliminate watch delay for object refetches The randomized write sizes of the modified rbd-mirror stress test results in a lot of journal object with few entries. Immediately fetch objects when performing a refetch check prior to closing an empty object. Signed-off-by: Jason Dillaman (cherry picked from commit d7b511b854c8ab633dcdb1c935014218b83f4e09) commit c65614f155ceccebd1960d6042d55e13b31e62ba Author: Jason Dillaman Date: Mon May 23 14:57:03 2016 -0400 journal: keep active tag to assist with pruning watched objects It's possible that there might be additional entries to prune in objects that haven't been prefetched yet. Keep the active tag to allow these entries to be pruned after they have been loaded. Signed-off-by: Jason Dillaman (cherry picked from commit 0fb730e6be632e25d76695e1f37eba542162d620) commit 30c7f4dbe0a7c515d14be98f3f59ba8a8cdb1dcf Author: Jason Dillaman Date: Mon May 23 12:15:49 2016 -0400 journal: update commit entry object number upon overflow Otherwise the recorded object positions might point to an older object that doesn't contain the actual entry. Signed-off-by: Jason Dillaman (cherry picked from commit 5fbf5f82de8214c5c676d38feae7682a006fcab1) commit fa08330dadb9ce973db866ad641a79d5d2dd290c Author: Jason Dillaman Date: Mon May 23 11:01:05 2016 -0400 journal: cleanup watch refetch flag handling Clear the refetch required flag while scheduling the watch and remove the stale object after the watch completes if still empty. Previously, it was possible for the flag to become out-of-sync with whether or not it was actually refreshed and pruned. Fixes: http://tracker.ceph.com/issues/15993 Signed-off-by: Jason Dillaman (cherry picked from commit ff2cc27ae592646b495bf1b614d35bd50c091a3d) commit 6375c7831de551c52f1dbf74bf9f65be3b6d8adf Author: Ricardo Dias Date: Fri May 13 16:47:25 2016 +0100 rbd-mirror: test: added image-deleter test case when mirroring is disabled Signed-off-by: Ricardo Dias (cherry picked from commit d85e312ead1da7c35d5ee4fe4be3d6e69a68b572) commit 1ba650539c1e90b82727a6bd9f1c16796b92b821 Author: Ricardo Dias Date: Fri May 13 16:44:53 2016 +0100 rbd-mirror: Unregister clients from non-primary images journal A non-primary image may have registered clients on its journal (for instance a primary image that was later demoted). We must unregister the clients when disabling image mirroring with the force option. Signed-off-by: Ricardo Dias (cherry picked from commit 9cb0b180bc8a12ea2ebb0d7e32279b696ddd127e) commit a53cf28ab95a704b78cfa46cf758c2eec8058619 Author: Ricardo Dias Date: Tue Apr 19 12:35:49 2016 +0100 rbd-mirror: test: Added unit test for testing image-deleter thread Signed-off-by: Ricardo Dias (cherry picked from commit ac9514941b0accb6c738ce4aa0953c36cdcf69a6) commit 78b13a504c32685090bb5b6512be9c050df2d8b6 Author: Ricardo Dias Date: Mon Apr 18 12:22:19 2016 +0100 rbd-mirror: test: Fixed test_PoolWatcher to support the new changes in PoolWatcher class Signed-off-by: Ricardo Dias (cherry picked from commit e3677f115e1f369a49227fba51a1d34505d487ac) commit fef7456f8bc85f32cbdb28ce81198a9eb7d56d3b Author: Ricardo Dias Date: Tue Apr 5 09:47:32 2016 +0100 rbd-mirror: Replayer: bootstrap existing mirrored images Signed-off-by: Ricardo Dias (cherry picked from commit 3c2615aa6da837bfef2365f701f56a34c49b536f) commit 97eb85f971a74678e4075b16cb5a8e836a985b56 Author: Ricardo Dias Date: Thu Apr 28 11:49:43 2016 +0100 rbd-mirror: replayer: Added image-deleter thread to replayer Fixes: http://tracker.ceph.com/issues/14421 Signed-off-by: Ricardo Dias (cherry picked from commit a6d698829d925894e4afcdec0ee42b1fc1205db0) commit 24406c5f8ea12d2c19573116ea84c90ba7446f8b Author: Ricardo Dias Date: Thu Apr 28 11:48:05 2016 +0100 rbd-mirror: Added image-deleter thread to rbd-mirror Signed-off-by: Ricardo Dias (cherry picked from commit 011984d40bce448fe0febfc6a9fcf57d2c7b4768) commit 0780230b4af7e4ca923dab5f77735d0a11409202 Author: Ricardo Dias Date: Wed Apr 13 15:17:40 2016 +0100 rbd-mirror: Added implementation of image-deleter thread image-deleter thread will handle the deletion of local images that have been disabled for mirroring by the primary site. Signed-off-by: Ricardo Dias (cherry picked from commit 6cc6e663ed2ec336a9475d93ab2cb605644e6ee0) commit 38514adc847a227dbd28318650931a8c2a7df7db Author: runsisi Date: Wed May 11 20:42:37 2016 +0800 rbd-mirror: librbd::mirror_peer_list never returns -ENOENT no need to check -ENOENT specially Signed-off-by: runsisi (cherry picked from commit 81ff42253369a9f57118bd26b29115c86a1748c5) commit 4e7233ad43d368a5f6e812870cec77545e034186 Author: runsisi Date: Wed May 11 20:52:38 2016 +0800 rbd-mirror: fix typo Signed-off-by: runsisi (cherry picked from commit 25bfd4c786d0701fe6ae06e3d120098bd82f8aa0) commit c3b1bf1e530a4cf55585886a4b4038b9d2862471 Author: Mykola Golub Date: Wed May 18 10:07:55 2016 +0300 rbd-mirror: calculate behind_master only if mirror tag is not newer than master Fixes: http://tracker.ceph.com/issues/15916 Signed-off-by: Mykola Golub (cherry picked from commit cbd8d526f94523c9de3c575d426063e63d7e1802) commit f04972c8be28fc6d3133363b5474f22ebac5fd15 Merge: c194c5d cb950fc Author: Jason Dillaman Date: Tue May 31 11:48:06 2016 -0400 Merge pull request #9377 from dillaman/wip-16063 jewel: test: failure in journal.sh workunit test Reviewed-by: Jason Dillaman commit cb950fca12395a1c62421d4b0bce64b4a52b2fd4 Author: Mykola Golub Date: Wed May 25 21:54:16 2016 +0300 test: workaround failure in journal.sh With the changes to ensure that the commit position of a new client is initialized to the minimum position of other clients, the 'journal inspect/export' commands return zero records because the master client has committed all of its entries. Workaround this by restoring the initial commit position after writing to the image. Fixes: http://tracker.ceph.com/issues/16011 Signed-off-by: Mykola Golub (cherry picked from commit e9ed8ac33b1884c6ed5cdfb23023094ec790b116) commit f92c2a5604d4a8e83504b7283114d3d59bc45f38 Author: Mykola Golub Date: Wed May 25 14:51:43 2016 +0300 cls::journal: treat empty commit position as minimal Signed-off-by: Mykola Golub (cherry picked from commit 16846e8085b0c594f065322a768e6dfcbd5d2444) commit c194c5d8d454d0b387e191d03fec9d916271ff7f Merge: 4d3b668 8fbb555 Author: Jason Dillaman Date: Tue May 31 11:46:30 2016 -0400 Merge pull request #9376 from dillaman/wip-15994 jewel: cls_journal: initialize empty commit position upon client register Reviewed-by: Jason Dillaman commit 4d3b66842d4477ec4c4a418a91723fc9b800101c Merge: dafe375 e5ebb51 Author: Jason Dillaman Date: Tue May 31 11:45:43 2016 -0400 jewel: librbd: write-after-write might result in an inconsistent replicated image jewel: librbd: write-after-write might result in an inconsistent replicated image Reviewed-by: Jason Dillaman commit dafe375640967e10d1a94b02970d399c51980076 Merge: 409fc64 305ebbc Author: Jason Dillaman Date: Tue May 31 11:30:47 2016 -0400 Merge pull request #9318 from Abhishekvrshny/wip-15957-jewel jewel: Metadata config overrides are applied synchronously Reviewed-by: Jason Dillaman commit 409fc64799aec36e183e6ba2bb25c3bf7e3d603b Merge: cd6ed4d 0c99028 Author: Jason Dillaman Date: Tue May 31 11:29:45 2016 -0400 Merge pull request #9317 from Abhishekvrshny/wip-15956-jewel jewel: rbd-mirror should disable the rbd cache for local images Reviewed-by: Jason Dillaman commit cd6ed4deb0f4b66a10ea7bd5e027daee3a93d1f4 Merge: 9fa8596 be9e85d Author: Kefu Chai Date: Tue May 31 18:48:58 2016 +0800 Merge pull request #9403 from dachary/wip-16088-jewel jewel: tests: rm -fr /tmp/*virtualenv* Reviewed-by: Kefu Chai commit be9e85d83f3d025cec16b7da85d8a2ce1f64e8fa Author: Loic Dachary Date: Fri May 20 13:19:07 2016 +0200 tests: rm -fr /tmp/*virtualenv* Signed-off-by: Loic Dachary (cherry picked from commit 8f0eefba00d1c6ad656302a608217fc3395d0401) commit 9fa85963a8accd669b0f0fbb17f4b721290432c6 Merge: a046d2a 3674341 Author: Kefu Chai Date: Tue May 31 13:46:31 2016 +0800 Merge pull request #8968 from Abhishekvrshny/wip-15728-jewel jewel: osd: acting_primary not updated on split Reviewed-by: Kefu Chai commit a046d2a8c37182652d5f300f58673f70c8485114 Merge: 954af78 8bbb5ad Author: Kefu Chai Date: Tue May 31 13:42:28 2016 +0800 Merge pull request #9100 from Abhishekvrshny/wip-15856-jewel jewel: Hammer (0.94.3) OSD does not delete old OSD Maps in a timely fashion (maybe at all?) Reviewed-by: Kefu Chai commit 954af787526a77b923fe85ed1282ba98277738e4 Merge: e8e572c fae360f Author: Yehuda Sadeh Date: Mon May 30 13:53:45 2016 +0300 Merge pull request #9245 from dreamhost/wip-15974-jewel rgw: fix manager selection when APIs customized Reviewed-by: Yehuda Sadeh commit e8e572c6454816119448240fa7813be7995fd250 Merge: c11db6f ffd545b Author: Orit Wasserman Date: Mon May 30 11:15:18 2016 +0200 Merge pull request #9239 from yehudasa/wip-15886-jewel rgw: keep track of written_objs correctly Reviewed-by: Orit Wasserman commit 8fbb555befb1415ef5218c8aaa149ca44017d73d Author: Venky Shankar Date: Thu May 19 22:45:34 2016 +0530 cls_journal: Select min commit position for new clients When a new client registers, its commit position is initialized to the least commit position of all other clients. Fixes: http://tracker.ceph.com/issues/15757 Signed-off-by: Venky Shankar (cherry picked from commit dd2e3dae8adb3778e7e381db5dbfca0948571c55) commit 576ff0ca804779d1eb0b1c44d6ce38e0658b6a05 Author: runsisi Date: Mon May 9 15:15:31 2016 +0800 cls_journal: remove duplicated key generation Signed-off-by: runsisi (cherry picked from commit 715e99c83e1ebc01ceb9eff0f66da8e2b598766a) commit fae360f0e419266b80f001a25a6864668e772d08 Author: Robin H. Johnson Date: Fri May 20 16:00:33 2016 -0700 rgw: fix manager selection when APIs customized When modifying rgw_enable_apis per RGW instance, such as for staticsites, you can end up with RESTManager instance being null in some cases, which returns a HTTP 405 MethodNotAllowed to all requests. Example configuration to trigger the bug: rgw_enable_apis = s3website Backport: jewel X-Note: Patch from Yehuda in private IRC discussion, 2016/05/20. Fixes: http://tracker.ceph.com/issues/15973 Fixes: http://tracker.ceph.com/issues/15974 Signed-off-by: Robin H. Johnson (cherry picked from commit 7c7a465b55f7100eab0f140bf54f9420abd1c776) commit 8bbb5add6208a552979270baf4623430cf45e1b3 Author: Kefu Chai Date: Fri May 13 11:26:31 2016 +0800 osd/OpRequest: reset connection upon unregister this helps to free the resources referenced by the connection, among other things, in the case of MOSDOp, the OSD::Session and OSDMap. this helps to free the resource earlier and trim the osdmaps in time. Fixes: http://tracker.ceph.com/issues/13990 Signed-off-by: Kefu Chai (cherry picked from commit 87850e96ea6fa45a8368bacabee50f9e95b40ae9) commit e97cc2d1e9bcee0a43f54ed592d3078182571874 Author: Kefu Chai Date: Thu May 12 20:28:11 2016 +0800 osd: reset session->osdmap if session is not waiting for a map anymore we should release the osdmap reference once we are done with it, otherwise we might need to wait very long to update that reference with a newer osdmap ref. this appears to be an OSDMap leak: it is held by an quiet OSD::Session forever. the osdmap is not reset in OSD::session_notify_pg_create(), because its only caller is wake_pg_waiters(), which will call dispatch_session_waiting() later. and dispatch_session_waiting() will check the session->osdmap, and will also reset the osdmap if session->waiting_for_pg.empty(). Fixes: http://tracker.ceph.com/issues/13990 Signed-off-by: Kefu Chai (cherry picked from commit 82b0af7cedc3071cd83ee53479f834c23c62b7d0) commit 555cec9a92568666880d70b9ae8f6e9748eca204 Author: Ricardo Dias Date: Tue May 17 18:04:28 2016 +0100 ceph.in: fix exception when pool name has non-ascii characters When deleting a pool without the --i-really-really-mean-it option, if the pool name has non-ascii characters, the format of the command message raises a UnicodeEncodeError exception. Fixes: http://tracker.ceph.com/issues/15913 Signed-off-by: Ricardo Dias (cherry picked from commit 805873fed0135616a4cf8b499d1ba0b10dac004c) commit 305ebbc36473cf8bfd22513ffe55d3a4a99d5ef1 Author: Jason Dillaman Date: Wed May 18 19:19:24 2016 -0400 librbd: metadata retrieval added to open image state machine Fixes: http://tracker.ceph.com/issues/15928 Signed-off-by: Jason Dillaman (cherry picked from commit b64cb31f8e7796175b4709929c017b3236649462) commit 5c9ecea87d3eb90201a967d8da827018cd728f0d Author: Jason Dillaman Date: Wed May 18 17:50:07 2016 -0400 cls_rbd: async version of metadata_list helper method Signed-off-by: Jason Dillaman (cherry picked from commit 985cb38211c51c95d84479df231c4f53847cb2ec) commit 0c99028582326c4ac142a03f168ab1bcf0ddd3c2 Author: Jason Dillaman Date: Wed May 18 20:53:26 2016 -0400 rbd-mirror: disable librbd caching for replicated images Each image has its own cache and each cache uses its own thread. With a large replicated cluster, this could result in thousands of extra threads and gigabytes of extra memory. Fixes: http://tracker.ceph.com/issues/15930 Signed-off-by: Jason Dillaman (cherry picked from commit ea35f148257282fe3f3ae02fe7a26cf245cda952) commit 188318b7b567f206d60dfb754fedbb0bfc74ef91 Author: Boris Ranto Date: Mon May 23 17:06:19 2016 +0200 rpm: Restore /var/run/ceph context manually The fixfiles command won't fix label for /var/run/ceph (/run is probably excluded from relabel), we need to restore the context manually. Signed-off-by: Boris Ranto (cherry picked from commit 6de39ce73737f79d4511dfc4e1d10d7f3733ab1b) commit ab1f65dd0429647c30fdae1dd97ff6f393dbc1a8 Author: Boris Ranto Date: Fri May 20 08:09:42 2016 +0200 rpm: Mute semodule in postun script Currently, we don't mute the semodule output in postun script. This results in the following message when removing ceph-selinux package: libsemanage.semanage_direct_remove_key: Removing last ceph module (no other ceph module exists at another priority). The fix is to simply mute the output of the command. Signed-off-by: Boris Ranto (cherry picked from commit 3c104b106540fabdba43e0d04a3828d77f83e89d) commit 587fb3df704973071114a30bc55288768dbc2720 Author: Nathan Cutler Date: Wed Apr 27 11:13:06 2016 +0200 install-deps.sh: systemd-rpm-macros must be installed on SUSE Otherwise the script breaks. Fixes: http://tracker.ceph.com/issues/15627 Signed-off-by: Nathan Cutler (cherry picked from commit a7f5ad5f6f9c01ef81f90a7071b6e2243b6da4e7) commit 69470d406d2b539e8021afd4def7de931ac91857 Author: Boris Ranto Date: Tue May 10 18:20:13 2016 +0200 rpm: Drop SELinux priority setting Older versions of semodule binary that are in Centos/RHEL do not support priority setting, dropping it. Fixes: #15822 Signed-off-by: Boris Ranto (cherry picked from commit 7e606f29f499eac83ecb7053ea6874fcfd092924) commit 128dbc8af0095622d48b962ddd9476aa4140bc16 Author: Boris Ranto Date: Wed May 4 09:09:47 2016 +0200 rpm: Fix SELinux relabel on fedora The SELinux userspace utilities stopped providing versions when they switched to CIL language. We need to use a different technique to relabel the files. Fixes: #15725 Signed-off-by: Boris Ranto (cherry picked from commit bd26641d705a34a1148a781eea08d203e81b750d) commit 8182b9a4d2750bb8f0f035d84e9d8476abc62c3c Author: Nathan Cutler Date: Tue Apr 12 15:33:45 2016 +0200 Test-enable XIO messenger and IB/RDMA support Fixes: http://tracker.ceph.com/issues/15472 Signed-off-by: Lars Marowsky-Bree Signed-off-by: Nathan Cutler (cherry picked from commit d4b74c56d6cbb2c6852bf66f2f34b8a877d7a9ef) commit 50626cec7a55637764ba379d6cb71e0b485290a8 Author: Nathan Cutler Date: Thu Apr 28 19:29:00 2016 +0200 rpm: do a full make check when --with-tests is given The check-local target tests the CLI tools only. Signed-off-by: Nathan Cutler (cherry picked from commit e1514809ae3d2df08a0c47882f9b3bd0ed68d200) commit fb6858c283d96de15c70fbe8ffc33b5dada82739 Author: Nathan Cutler Date: Tue Apr 26 17:03:28 2016 +0200 rpm: drop sysvinit bits from relabel_files function Signed-off-by: Nathan Cutler (cherry picked from commit 20df58799a4853ed3798501176988ffaf888936b) commit db4f2d677df5c9fc00592fc9e003bccb44af6da6 Author: Nathan Cutler Date: Sun Oct 11 18:27:45 2015 +0200 rpm: replace literal paths with RPM macros The only place we should write out literal paths is in the RPM scriptlets, and there only for things that are not installed by this package. Signed-off-by: Nathan Cutler (cherry picked from commit e0346b36fc0e6e17c80c047b13997448ccf1c76e) commit 3abf89517b0b2142698610debbaa3c6b535424f3 Author: Nathan Cutler Date: Sun Oct 11 18:05:12 2015 +0200 rpm: drop udev/95-ceph-osd-alt.rules This udev rules file was needed on older RHEL platforms, which are unsupported as of jewel. Signed-off-by: Nathan Cutler (cherry picked from commit ee4acb31593dd33da99138998fd0f30ebc2f1a6e) commit a7e01dd4ff08607fc6d99e41189e3db1a3c1d4da Author: Nathan Cutler Date: Sun Oct 11 17:20:46 2015 +0200 rpm: global replace $RPM_BUILD_ROOT with %{buildroot} Signed-off-by: Nathan Cutler (cherry picked from commit bed92b6a86cdef784e89c38952fa779ce5fc5a91) commit d81cec99421e544e2cf7d212680b04d51ef675f3 Author: Nathan Cutler Date: Sun Oct 11 10:37:47 2015 +0200 rpm: put dependencies in alphabetical order Signed-off-by: Nathan Cutler (cherry picked from commit 66042c03df012754215ed848e475f2195ee988a0) commit e8338a5844aea6dad8e61f00dbd94ca4e04df7be Author: Nathan Cutler Date: Sun Oct 11 10:32:46 2015 +0200 rpm: drop sysvinit-specific dependencies Obsolete in jewel and above. Signed-off-by: Nathan Cutler (cherry picked from commit c655cb896ad10e001b95848f4a4b64dd1ef6984f) commit b18d8ac0d65f91929fc956980d3b5977bb2970e6 Author: Nathan Cutler Date: Sun Oct 11 10:32:05 2015 +0200 rpm: move boost-random dependency to appropriate section Signed-off-by: Nathan Cutler (cherry picked from commit 46b16c19f73d82cc357b16e6201926a3479046f2) commit 94efaef9c28347baebc374e27d3e81f143e642e0 Author: Nathan Cutler Date: Sat Apr 23 19:57:49 2016 +0200 rpm: move unified python-sphinx build dependency Now that the python-sphinx build dependency is unified, move it to the proper section of the spec file. Signed-off-by: Nathan Cutler (cherry picked from commit 1d897da861c6134778136fa2b8b33e36c08faff9) commit 54dc8bf5703e2fb8962a97f7737a6114132b9b20 Author: Nathan Cutler Date: Sat Apr 23 19:41:06 2016 +0200 rpm: drop init-ceph.in-fedora.patch Signed-off-by: Nathan Cutler (cherry picked from commit e8cf501d0f22c757a0860a0c91dcb1d7a472ed34) commit 641e5bb0808a14d2b59cba141bad4acf5bbfb63a Author: Nathan Cutler Date: Sun Oct 11 10:22:37 2015 +0200 rpm: drop dead conditionals This commit drops conditionals that no longer serve any purpose, since jewel and above do not support the distro versions they are checking for. Signed-off-by: Nathan Cutler (cherry picked from commit 1332fac70e537014b5e03810aa983e2c9210d8dc) commit 4317a17997f21019e1dec59606928a2a9f4d94b9 Author: Nathan Cutler Date: Sun Oct 11 10:18:33 2015 +0200 rpm: drop python-argparse dependency This was only necessary for older (now unsupported) distro versions. Signed-off-by: Nathan Cutler (cherry picked from commit 6c42066d488d03142a56f92cff99d9c29d106214) commit 70466ffd386795082f02d718e804ad9e9c494401 Author: Nathan Cutler Date: Sun Oct 11 10:02:40 2015 +0200 rpm: drop _with_systemd Ceph versions jewel and above only support systemd. Signed-off-by: Nathan Cutler (cherry picked from commit 2ca6dd250b6a049f27fa17c5e8ca11674acb63f6) commit 928a8158aa71012db9f451a4302e8661783b5d37 Author: Nathan Cutler Date: Sun Oct 11 09:43:06 2015 +0200 rpm: drop python_sitelib/sitearch conditional This conditional was required to support older versions of RHEL/CentOS that are no longer supported in infernalis and above. Signed-off-by: Nathan Cutler (cherry picked from commit 98d52120f89af3e93896d0518dbea4c4f660af48) commit 9f76b9ff31525eac01f04450d72559ec99927496 Author: Sage Weil Date: Mon Apr 18 09:16:02 2016 -0400 udev: remove 60-ceph-partuuid-workaround-rules These were added to get /dev/disk/by-partuuid/ symlinks to work on wheezy. They are no longer needed for the supported distros (el7+, jessie+, trusty+), and they apparently break dm by opening devices they should not. Fixes: http://tracker.ceph.com/issues/15516 Signed-off-by: Sage Weil (cherry picked from commit 9f77244b8e0782921663e52005b725cca58a8753) commit 1ad390197af21d8cd7e507e2959f803028d3d09d Author: Nathan Cutler Date: Wed Apr 20 17:21:56 2016 +0200 rpm: refrain from installing systemd unit files twice These files are already installed by systemd/Makefile.am Fixes: http://tracker.ceph.com/issues/15549 Signed-off-by: Nathan Cutler (cherry picked from commit 6d3cbaa509525f1807ac34bf5fed0fe99255f40f) commit 9c81f25b76ef2211910e453ba2bc4577fc921d7f Author: Boris Ranto Date: Tue Feb 23 05:57:14 2016 +0100 test/ec: build the libs only when 'make check' Currently, we are always building the erasure code libraries while we need them only when 'make check' is run. Moving the test libraries to check_LTLIBRARIES should fix this for us. We no longer need to remove the libec libs manually, remove the lines that do that. Signed-off-by: Boris Ranto (cherry picked from commit 0bbfea5cf13c138bbb2f0993607fa0dbf756281b) commit 663d7c2901ddf0b6e4184ec344bf0af00582eaf2 Author: xie xingguo Date: Wed May 4 16:53:18 2016 +0800 mon/Monitor: use CEPH_MON_PORT for CRC output The default monitor port will be changed from 6789 to 3300, which has been assigned to us by IANA. So update this accordingly for better tracing and understanding. Signed-off-by: xie xingguo (cherry picked from commit 2b3d5686d76ee65fceca31d0d3343c2eaa4f2364) commit 64b948c21a4608a7680012e247f23dd2a3393f3f Author: xie xingguo Date: Wed May 4 15:03:55 2016 +0800 mon/Monitor: fix memory leak Fixes: http://tracker.ceph.com/issues/15793 Signed-off-by: xie xingguo (cherry picked from commit ae5184ac07523a050584a3d621954baeb0dca01a) commit 9c3c6d57552a0a2cd448f0e60ded2f96ef9802ef Author: xie xingguo Date: Wed May 4 14:54:32 2016 +0800 mon/Monitor: use VOID_TEMP_FAILURE_RETRY wrapper for fd close(). This is safer and much more robust. Signed-off-by: xie xingguo (cherry picked from commit bfa40363b2d8fc25b4aed0164579d55137ac9e8b) commit 544c538d3720d7f18fca29ba15fd47200d3e05fc Author: xie xingguo Date: Wed May 4 10:02:01 2016 +0800 mon/Monitor: add "host" to tracked_conf_key The get_tracked_conf_keys() method gets the specified keys registered and works as filters for configuration changing, so we can track these key configuration changing and react properly. Right now graylog prints host in its log entry, so Monitor::update_log_clients() also reads from conf.host and pass the host read from conf to graylog. That is why we shall get "host" to the tracked_conf_key too. Signed-off-by: xie xingguo (cherry picked from commit c79fc3305be42bd6126e8a60ae1f220d86754830) commit e5ebb51f22c43d49d716129ffcb5f87390b7acab Author: Jason Dillaman Date: Thu May 19 15:52:16 2016 -0400 journal: replay position might change after pruning stale tags Signed-off-by: Jason Dillaman (cherry picked from commit 0f5f1c375d8ef7b301cfd20b6848bdc81a761746) commit 9ecc3dcb76cc8f6d2904118d6b246d77d10cd9cf Author: Jason Dillaman Date: Thu May 19 15:50:04 2016 -0400 librbd: delay commit of overwritten journal event With the cache enabled and write-after-write IOs to the same object extents, it was possible for the overwritten journal event to be committed before the overwriter journal event was written to disk. If a client crash occurs before the event is written, the image will be inconsistent on replay. Fixes: http://tracker.ceph.com/issues/15938 Signed-off-by: Jason Dillaman (cherry picked from commit e8bf64cc85ffe3d2dda23eab1834f7a5f104f6fe) commit 070dc7c98daaacaa05f72f50561d47ce459984da Author: Jason Dillaman Date: Thu May 19 14:13:46 2016 -0400 ObjectCacher: pass new journal tid when overwriting extent librbd needs to ensure the new journal event has been safely committed to disk before it can mark an overwritten journal event as committed. Signed-off-by: Jason Dillaman (cherry picked from commit 0051316effe0b25804378975b117691d7ad08a3b) commit 0dfc787ef58fe9aa8177742f7c6e294e3b5d4b89 Author: Jason Dillaman Date: Thu May 19 13:48:22 2016 -0400 qa/workunits/rbd: record rbd CLI debug messages during mirror stress The debug messages from 'rbd bench-write' and 'rbd snap create', in addition to the existing debug messages from rbd-mirror, make it possible to determine the source of any image inconsistency. Signed-off-by: Jason Dillaman (cherry picked from commit 714ed5e64e65fed9ac886537becf3a85ae6ae7fd) commit ffd545bf173c5a076c47541cbe3889175188d73e Author: Yehuda Sadeh Date: Mon May 16 14:35:12 2016 -0700 rgw: keep track of written_objs correctly Fixes: http://tracker.ceph.com/issues/15886 Only add a rados object to the written_objs list if the write was successful. Otherwise if the write will be canceled for some reason, we'd remove an object that we didn't write to. This was a problem in a case where there's multiple writes that went to the same part. The second writer should fail the write, since we do an exclusive write. However, we added the object's name to the written_objs list anyway, which was a real problem when the old processor was disposed (as it was clearing the objects). Signed-off-by: Yehuda Sadeh (cherry picked from commit 8fd74d11682f9d0c9085d2dc445fc3eb5631f6e0) commit 8356021ef3d3d21fd632d3caa660ba9a03cf80d3 Author: Samuel Just Date: Thu May 12 16:57:49 2016 -0700 Pipe: take a ref to existing while we are waiting Otherwise, if it is reaped while we are waiting, it'll be a use-after-free. Fixes: http://tracker.ceph.com/issues/15870 Signed-off-by: Samuel Just (cherry picked from commit b224912d249453d754fc0478d3680f8cfa1a5c22) commit c11db6f103b2dd0049894a606916a84d66933833 Merge: 7834070 e0dfc55 Author: Jason Dillaman Date: Fri May 20 14:57:00 2016 -0400 Merge pull request #9232 from jdurgin/wip-mirror-workunit-jewel jewel: qa/workunits/rbd: fixed rbd_mirror teuthology runtime errors Reviewed-by: Jason Dillaman commit e0dfc55ac874f2531bd0b88aa1c81156d7f7aec9 Author: Jason Dillaman Date: Wed May 18 14:17:13 2016 -0400 qa/workunits/rbd: fixed rbd_mirror teuthology runtime errors Signed-off-by: Jason Dillaman (cherry picked from commit 8ef09c4d8d3c7c6fbc9b6f7ef17cca0179e73468) commit 7834070fd8ead19e49fb8c766ea3ce8bfe63f032 Merge: 9873fc1 5f09b9a Author: Jason Dillaman Date: Fri May 20 09:02:51 2016 -0400 Merge pull request #9226 from dillaman/wip-15950 jewel: rbd-mirror: potential crash during image status update Reviewed-by: Jason Dillaman commit 9873fc13d5647e94ec6f0152f903dd67b2882b0d Merge: aee6f7e 6c2973c Author: Jason Dillaman Date: Fri May 20 09:02:06 2016 -0400 Merge pull request #9069 from dillaman/fixup-mirror_image_status_list-jewel jewel: cls::rbd: mirror_image_status_list returned max 64 items Reviewed-by: Jason Dillaman commit 5f09b9aef8f30d9dc7e5ec934b228d885e6456e2 Author: Jason Dillaman Date: Thu May 19 10:11:12 2016 -0400 journal: reset watch step after pruning expired tag Signed-off-by: Jason Dillaman (cherry picked from commit 48c2f4e079000021b6ebe807bb1dba0dc1a6a11d) commit 2ee1e0a583d0ca851db4671b561eff9d493ca7d3 Author: Jason Dillaman Date: Wed May 18 11:01:22 2016 -0400 rbd-mirror: additional debug messages during image replayer start/stop Signed-off-by: Jason Dillaman (cherry picked from commit afc891f7fcb000d106b1d722386f6d81ddeddab6) commit f9e6bd8f1b6750d500fba3dd134257b7baac08c3 Author: Jason Dillaman Date: Wed May 18 00:55:01 2016 -0400 rbd-mirror: ensure proper handling of status updates during shutdown Previously, several shutdown race conditions could occur due to the use of the async work queue for scheduling updates. Fixes: http://tracker.ceph.com/issues/15909 Signed-off-by: Jason Dillaman (cherry picked from commit 9ae4edc05660665c3a135536a7c51d4070d28a40) commit f789d83cbb9dbb1a6daa15585e2be2980ea19616 Author: Jason Dillaman Date: Tue May 17 23:31:02 2016 -0400 rbd-mirror: track bootstrap state within image status Signed-off-by: Jason Dillaman (cherry picked from commit e5dd97b4bf4a993911febf34dce90470a2a91d59) commit e0b3879692d38bcc01d928cef09f74bc850cef86 Author: Jason Dillaman Date: Tue May 17 23:25:25 2016 -0400 rbd-mirror: combine ImageReplayer stopped and uninitialized states Signed-off-by: Jason Dillaman (cherry picked from commit a8f6dde4dbae8039380f31e026030a22a89fb4bd) commit 9c7977da45aef44160efadb48cb2d606568d4a7b Author: Jason Dillaman Date: Tue May 17 16:14:42 2016 -0400 rbd-mirror: lock ordering issue in status update callback Signed-off-by: Jason Dillaman (cherry picked from commit 5c04c1361074ca9e9df1bc1c2dd82dfafc2a1304) commit 66f7f2d7bda7f2db6871870f5622594d6391b921 Author: Mykola Golub Date: Wed May 4 15:23:11 2016 +0300 rbd-mirror: don't unregister asok commands if image replayer start failed Signed-off-by: Mykola Golub (cherry picked from commit 2fd6fdd3dc9ecb915362f5e3aa8a224210273540) commit 569f06f1241309a710a0570d4ede950c1dc6e5d1 Author: Mykola Golub Date: Wed May 4 15:25:04 2016 +0300 rbd-mirror: avoid potential deadlock Signed-off-by: Mykola Golub (cherry picked from commit 46bf727d43af4975beeb514630a99d59cfb0a406) commit aee6f7ed791abd63ff9d1158a258d51db6c5ad28 Merge: 2d0669e 0d19390 Author: Jason Dillaman Date: Fri May 20 09:01:50 2016 -0400 Merge pull request #9217 from dillaman/wip-15945 jewel: journal: live replay might skip entries from previous object set Reviewed-by: Jason Dillaman commit 0d193905b1d552600b542a466759af054c80e498 Author: Jason Dillaman Date: Mon May 16 21:17:09 2016 -0400 qa/workunits/rbd: rbd-mirror daemon stress test This test repeatedly runs rbd bench-write, kills the process randomly to create an unclean journal shutdown, and verifies that the image content replicates correctly. Signed-off-by: Jason Dillaman (cherry picked from commit e2ab3128a37f4f36c9a1de6ddf2c0fe2e48aa7e2) commit d020b6ab1d10daa92fc99d1506222987b4306b26 Author: Jason Dillaman Date: Mon May 16 18:08:35 2016 -0400 journal: helper method to detect newer tags Signed-off-by: Jason Dillaman (cherry picked from commit 0a8a6126ea35344e85af7eb64ffc490938edba51) commit 56515070710a6723ae4adc3e3b166e9ae9a35fdf Author: Jason Dillaman Date: Sun May 15 09:52:41 2016 -0400 journal: skip partially complete tag entries during playback If a journal client does not fully write out its buffered entries before quiting, replay should skip over all remaining out-of- sequence entries for the tag. Fixes: http://tracker.ceph.com/issues/15864 Signed-off-by: Jason Dillaman (cherry picked from commit 9454f7e4c62437b1c288f371009feba1fd374584) commit 18f663d834edd03b42e2c08b9428e72fdc6bae9d Author: Jason Dillaman Date: Sat May 14 18:58:41 2016 -0400 journal: close, advance, and open object set ordering Flush in-flight appends to open objects before advancing the active object set. Additionally, don't start recording to the new objects until after advancing the active set. Signed-off-by: Jason Dillaman (cherry picked from commit de830057d0f7286914e019540c6263423cb60428) commit 27fba6ad7e7cb566d32c6e63f35aff966fc60bd7 Author: Jason Dillaman Date: Sat May 14 18:13:38 2016 -0400 journal: new ObjectRecorder closed callback The callback will be invoked if there were in-flight appends when the close was requested. Signed-off-by: Jason Dillaman (cherry picked from commit bba91437dbe3b7a9b6da8a61ccc4c597858c8efc) commit 870c21c738e42d2cf34478bfa0631d602175dadc Author: Jason Dillaman Date: Fri May 13 16:34:44 2016 -0400 journal: do not flag append as full if already known to be full Once an object has overflowed or a close is in-progress, re-attempting to close the object is not needed since the async process is already underway. Signed-off-by: Jason Dillaman (cherry picked from commit 45620dc312ca2aeaf084638144aaa4783b622c4a) commit 1cdebc7b896dd36bfee30f12aa3abd602c15af3e Author: Jason Dillaman Date: Fri May 13 16:28:50 2016 -0400 journal: delay object overflow event until in-flight appends settled Signed-off-by: Jason Dillaman (cherry picked from commit ee8d311a1fcdf7a85cdc168e1740fbaa252e87d3) commit 821d92af2fcc43ff560f01c78ff699ad7df33daa Author: Jason Dillaman Date: Fri May 13 16:17:37 2016 -0400 journal: ignore flush on closed/overflowed object The journal would be in-progress on transitioning to a new object recorder in a newer object set. Once the records re-attach to the new object player they will automatically flush. Signed-off-by: Jason Dillaman (cherry picked from commit a13cb821f71b6b1b3378464bcc2d8318eb0da48a) commit 3d860d5db37f77c1e8c76d214529266062caaa75 Author: Jason Dillaman Date: Fri May 13 16:10:11 2016 -0400 journal: implicitly detach future's flush handler on append If the future is already in-flight, there is no purpose served by requesting the future be flushed to disk. Signed-off-by: Jason Dillaman (cherry picked from commit 1cb9b1a1b25d0f4c2d9a9e22661636e195fc70a0) commit f98e1289ce10aeb059a0b7d7cc8d17d6f1d60219 Author: Jason Dillaman Date: Fri May 13 15:22:30 2016 -0400 journal: async callback for advancing the active object set Signed-off-by: Jason Dillaman (cherry picked from commit ca89abfd9e6f0e8edc3234afea387757ef694826) commit 86e51ebcb301d823fb5c8b30bdc07107704679b4 Author: Jason Dillaman Date: Fri May 13 14:49:07 2016 -0400 journal: re-fetch active object before advancing set during replay During a live replay, it's possible that an append and and overflow into the next object could race with the live playback of the same object. Re-fetch an "empty" object at least once before advancing to next set to ensure all records have been read. Fixes: http://tracker.ceph.com/issues/15665 Signed-off-by: Jason Dillaman (cherry picked from commit 6056f8c45c99bd37cb18933a37cc238c7e9a7c7d) commit 2d0669e172c37d38f118332574a474a534d27020 Merge: f3979cd e885f1e Author: Yehuda Sadeh Date: Thu May 19 13:08:00 2016 -0700 Merge pull request #9210 from yehudasa/wip-15926-jewel radosgw-admin: fix 'period push' handling of --url commit e885f1e3be0f9a0c6457c5387f1c38b6e8b6a492 Author: Casey Bodley Date: Mon May 16 09:37:05 2016 -0400 radosgw-admin: fix 'period push' handling of --url was calling send_to_remote_gateway(), but passing 'url' instead of 'remote'. now uses send_to_remote_or_url() to accept either Fixes: http://tracker.ceph.com/issues/15926 Signed-off-by: Casey Bodley commit f3979cd1d2fe910185db6dc3894722dd385379bd Merge: 1356d29 37ecfa6 Author: Boris Ranto Date: Thu May 19 12:56:15 2016 +0200 Merge pull request #9194 from ceph/wip-jewel-no-lttng-global-link Do not link lttng into libglobal Reviewed-by: Boris Ranto commit 37ecfa6c754683cac1c6f4c66a1a491d2fd5db1d Author: Karol Mroz Date: Mon May 2 14:01:27 2016 +0200 global: don't link lttng into libglobal Rely on dynamic initialization instead. Linking lttng in this way had the unfortunate side effect of causing radosgw to segfault (when daemonized) during sigterm processing (ie. during lttng_ust_exit()). This was originally removed in 638738f, but accidentally re-added via 5f61d36. Signed-off-by: Karol Mroz commit 1356d29c70afedcbd55599d29aab93d2f79a02ad Merge: 15c4ad4 16ba13c Author: Jason Dillaman Date: Wed May 18 13:23:39 2016 -0400 Merge pull request #9180 from dillaman/wip-15746-jewel jewel: doc: update mirroring guide to include pool/image status commands Reviewed-by: Jason Dillaman commit 16ba13c83d678656723173c061318dd83805768a Author: Mykola Golub Date: Thu May 12 11:15:34 2016 +0300 doc: update mirroring guide to include pool/image status commands Fixes: http://tracker.ceph.com/issues/15746 Signed-off-by: Mykola Golub (cherry picked from commit 4381e341c435c80793b847115eed616b8359c4e6) commit 818166e25d2ccac5d9bb3541f6166fcb8c25fd64 Author: Mykola Golub Date: Thu May 12 11:11:55 2016 +0300 doc: fixup: "rbd-mirror daemon" instead of "rbd-daemon" Signed-off-by: Mykola Golub (cherry picked from commit ddd6345f8f7edcff638ed435aef316ba3e0e40dd) commit 15c4ad44010c798af7804e287ba71dcc289f806f Merge: 3a66dd4 7be281d Author: Sage Weil Date: Wed May 18 05:11:55 2016 -0400 Merge pull request #8815 from ktdreyer/wip-15646-jewel-system-targets jewel: debian: install systemd target files Reviewed-by: Kefu Chai commit 7be281dab96c15b53fcfc3c2ffe6d17aec84b315 Author: Sage Weil Date: Thu Apr 28 09:13:50 2016 -0400 debian/control: dh_systemd_start is in the dh-systemd package Fixes: http://tracker.ceph.com/issues/15573 Signed-off-by: Sage Weil (cherry picked from commit 89154d1e01dbc58dc92f37741031f7ddb0448d2e) commit e463aa8b92dabdc7c13e150585eb09790da638ba Author: Kefu Chai Date: Fri Apr 22 21:07:58 2016 +0800 debian: install systemd target files * enable it using dh_systemd_enable * start the target using dh_systemd_start * move the dh_installinit, dh_systemd_enable, dh_systemd_start calls down, so they can identify the service files if they care about them. Fixes: http://tracker.ceph.com/issues/15573 Signed-off-by: Kefu Chai (cherry picked from commit f950a0a7f5b35e13668379bc69dcb5d98a2dfc1f) commit 72c9b6f70752faf779b9272724f428e68d6499bb Author: Kefu Chai Date: Mon May 9 14:15:36 2016 +0800 osd: remove all stale osdmaps in handle_osd_map() in a large cluster, there are better chances that the OSD fails to trim the cached osdmap in a timely manner. and sometimes, it is just unable to keep up with the incoming osdmap if skip_maps, so the osdmap cache can keep building up to over 250GB in size. in this change * publish_superblock() before trimming the osdmaps, so other osdmap consumers of OSDService.superblock won't access the osdmaps being removed. * trim all stale osdmaps in batch of conf->osd_target_transaction_size if skip_maps is true. in my test, it happens when the osd only receives the osdmap from monitor occasionally because the osd happens to be chosen when monitor wants to share a new osdmap with a random osd. * always use dedicated transaction(s) for trimming osdmaps. so even in the normal case where we are able to trim all stale osdmaps in a single batch, a separated transaction is used. we can piggy back the commits for removing maps, but we keep it this way for simplicity. * use std::min() instead MIN() for type safety Fixes: http://tracker.ceph.com/issues/13990 Signed-off-by: Kefu Chai (cherry picked from commit 369db9930887d75b498927da9c97733bff4472b6) commit 6c2973cd08503a3d91eeb881732c79213e42599f Author: Mykola Golub Date: Wed May 11 14:14:46 2016 +0300 cls::rbd: mirror_image_status_list returned max 64 items Signed-off-by: Mykola Golub (cherry picked from commit e925ce86243c91e6d64ff8d7b45166e4e3f2650b) commit 36743419ba49df4f97cdbe9ac25d9b08b0685108 Author: Sage Weil Date: Fri Apr 29 15:24:09 2016 -0400 osd/PG: update info.stats.* mappings on split These are updated in the init and start_peering_interval paths, but not on split. Fixes: http://tracker.ceph.com/issues/15523 Signed-off-by: Sage Weil (cherry picked from commit 8b42294d98dd70715184991da5ba495fbe93d009) commit b443fdf9402df621400ab7ca66799f597e8b2c23 Author: Nathan Cutler Date: Sat Apr 23 11:42:25 2016 +0200 ceph.spec.in: recommend ntp-daemon on SUSE SUSE has settled on "ntp-daemon" as the generic package name. The "ntp" and "chrony" etc. packages have "Provides: ntp-daemon" in their respective spec files. References: http://tracker.ceph.com/issues/15419 Signed-off-by: Nathan Cutler (cherry picked from commit 23108547454f53b95e193086d82d433570312ad4) commit 299f84ccdbf4d91412e216042fb248fd802af89d Author: Nathan Cutler Date: Thu Apr 7 20:54:21 2016 +0200 deb: make ceph-base and radosgw recommend time-daemon References: http://tracker.ceph.com/issues/15419 Signed-off-by: Nathan Cutler (cherry picked from commit f27cda03b7a961e445cce4ad35db6417e9e55959) commit 2da7cb2c7daad3ce2f9d2614e019291f75d4737d Author: Nathan Cutler Date: Thu Apr 7 20:17:44 2016 +0200 systemd: make Ceph daemon units "want" time-sync.target Fixes: http://tracker.ceph.com/issues/15419 Signed-off-by: Nathan Cutler (cherry picked from commit df893f395e62c8b1db18a277b38b44cab8b2016f)