commit 9411351cc8ce9ee03fbd46225102fe3d28ddf611 Author: Jenkins Build Slave User Date: Mon Dec 5 22:15:20 2016 +0000 10.2.4 commit a9444915a10038c35d5726485e110e56349fe3df Merge: 4d9a28d 4d6f848 Author: Loic Dachary Date: Mon Dec 5 14:50:23 2016 +0100 Merge pull request #12167 from liewegas/wip-osdmap-encoding-jewel jewel: osd: condition OSDMap encoding on features Reviewed-by: Loic Dachary commit 4d9a28d26d84a1dab90d22a09b1a044c193a41ad Merge: 6d17188 d194db8 Author: Loic Dachary Date: Sat Dec 3 10:57:18 2016 +0100 Merge pull request #12067 from SUSE/wip-17953-jewel jewel: mon: OSDMonitor: only reject MOSDBoot based on up_from if inst matches Reviewed-by: Samuel Just commit d194db826bb26483f847235bf062ba7b225be147 Author: Samuel Just Date: Mon Nov 14 11:50:23 2016 -0800 OSDMonitor: only reject MOSDBoot based on up_from if inst matches If the osd actually restarts, there is no guarrantee that the epoch will advance past up_from. If the inst is different, it can't really be a dup. At worst, it might be a queued MOSDBoot from a previous inst, but in that case, the real inst would see itself marked up, and then back down causing it to try booting again. Fixes: http://tracker.ceph.com/issues/17899 Signed-off-by: Samuel Just (cherry picked from commit 033ad5b46c0492134e72a8372e44e3ef1358d2df) commit 6d17188786ef7bd574dd8336d1e97e069db9de74 Merge: b168c26 79be070 Author: Josh Durgin Date: Fri Dec 2 08:16:27 2016 -0800 Merge pull request #12207 from jdurgin/wip-librados-setxattr-overload-jewel librados: remove new setxattr overload to avoid breaking the C++ ABI Reviewed-by: Sage Weil commit b168c262f4c04d748b8a71aa6289e13385398764 Merge: be5c828 00de014 Author: Loic Dachary Date: Fri Dec 2 10:01:39 2016 +0100 Merge pull request #12267 from dachary/wip-17904-jewel jewel: Error EINVAL: removing mon.a at 172.21.15.16:6789/0, there will be 1 monitors Reviewed-by: Samuel Just commit 00de0140410ec8457ca25920866f8409f0d54a10 Author: Joao Eduardo Luis Date: Wed Nov 2 15:38:36 2016 +0000 mon: MonmapMonitor: drop unnecessary 'goto' statements Signed-off-by: Joao Eduardo Luis (cherry picked from commit 20dcb597e35e6961db81831facefbe22cecddec3) commit 25f1b390525d975f90d0c7232b785415431d0f3e Author: Joao Eduardo Luis Date: Wed Nov 2 15:33:52 2016 +0000 mon: MonmapMonitor: return success when monitor will be removed Fixes: http://tracker.ceph.com/issues/17725 Signed-off-by: Joao Eduardo Luis (cherry picked from commit c9d46cfbf2512bc3495c6901de2b8f711bef9bae) commit be5c82829916c581765f7c0130b738395a27720c Merge: 427f357 3cc29c6 Author: Samuel Just Date: Thu Dec 1 11:08:04 2016 -0800 Merge pull request #12001 from dachary/wip-17915-jewel jewel: filestore: can get stuck in an unbounded loop during scrub Reviewed-by: Samuel Just commit 79be070a4d442229d62b168ab87c95b662df1a9c Author: Josh Durgin Date: Mon Nov 28 22:06:56 2016 -0800 librados: remove new setxattr overload to avoid breaking the C++ ABI Fixes: http://tracker.ceph.com/issues/18058 Signed-off-by: Josh Durgin (cherry picked from commit b8ff781ddcf737882163cf56d7c9b11e815fb699) Conflicts: src/include/rados/librados.hpp (trivial namespace change in removed line) commit 4d6f8481b22ecfa5e55b2a6f8f8660f2d0445030 Author: Sage Weil Date: Wed Nov 23 14:15:50 2016 -0500 crush: condition latest tunable encoding on features This avoids throwing hammer OSDMap encodings off. Signed-off-by: Sage Weil (cherry picked from commit 9e5ff86487bd1f5979866b5e16300dd4a3979f97) commit bf96b30eaf552148249953ed4fb654cbb101c3d0 Author: Sage Weil Date: Mon Nov 28 14:35:53 2016 -0500 crush/CrushWrapper: encode with features No behavior change yet; just fixing callers. Signed-off-by: Sage Weil (cherry picked from commit b7c9e055848c8aa951bc48c957cff3ef323ea441) [Updated write_file to use all feaetures] [Updated OSDMonitor.cc to use mon->quorum_features instead of the mon->get_quorum_con_featuers() helper] [trivial conflict from removed write_file and read_file] Conflicts: src/crush/CrushWrapper.h src/mgr/PyModules.cc src/mon/OSDMonitor.cc src/tools/ceph_monstore_tool.cc commit c5f5b94fc2e18b38d05f00fec04d2d7e3c35c54c Author: Sage Weil Date: Mon Nov 28 14:35:24 2016 -0500 crush/CrushWrapper: drop unused 'lean' encode() argument No callers, no users. Signed-off-by: Sage Weil (cherry picked from commit 638a38bbb55c07ad0358a35a56418e66874d1c26) Conflicts: src/crush/CrushWrapper.h [trivial conflict due to removal of write_file and read_file] commit c66c556852b08e18d409e769eb7bd945c35e43cf Author: Sage Weil Date: Wed Nov 23 13:51:59 2016 -0500 osd/osd_types: encode pg_pool_t like hammer if features indicate hammer If the target features are missing the new OSDOp encoding, the first feature we added post-hammer, encode like hammer. Signed-off-by: Sage Weil (cherry picked from commit 2f8cfb632823ba4e63eaff394392d6af7979d7c8) commit 85caf34924bbfb0e872abd096ec571fff73035ae Author: Sage Weil Date: Wed Nov 23 13:48:35 2016 -0500 osd/osd_types: conditional pg_pool_t encoding Align this with decode. Signed-off-by: Sage Weil (cherry picked from commit 01d9e8a20bbc3c039f67b040da95018e2c7b00b6) commit 3cc29c6736007c97f58ba3a77ae149225e96d42a Author: Sage Weil Date: Thu Nov 10 13:56:24 2016 -0500 os/filestore/HashIndex: fix list_by_hash_* termination on reaching end If we set *next to max, then the caller (a few lines up) doesn't terminate the loop and will keep trying to list objects in every following hash dir until it reaches the end of the collection. In fact, if we have an end bound we will never to an efficient listing unless we hit the max first. For one user, this was causing OSD suicides when scrub ran because it wasn't able to list all objects before the timeout. In general, this would cause scrub to stall a PG for a long time and slow down requests. Broken by refactor in 921c4586f165ce39c17ef8b579c548dc8f6f4500. Fixes: http://tracker.ceph.com/issues/17859 Signed-off-by: Sage Weil (cherry picked from commit c5180262a086c2d3895aff4bf0fb0ff9a6666149) commit 427f357f0eed32c9ce17590ae9303a94e8b710e7 Merge: 0c38c46 8b595f5 Author: Loic Dachary Date: Wed Nov 9 20:53:18 2016 +0100 Merge pull request #11822 from SUSE/wip-17816-jewel jewel: Missing comma in ceph-create-keys causes concatenation of arguments Reviewed-by: Loic Dachary commit 8b595f567407be2a759e4987f33ce79e3763ea49 Author: Patrick Donnelly Date: Sun Sep 18 16:26:29 2016 -0400 ceph-create-keys: add missing argument comma The arguments "get" and "client.admin" were being concatenated into "getclient.admin". Found using ceph-ansible + strace: 13031 execve("/usr/bin/ceph", ["ceph", "--cluster=ceph", "--name=mon.", "--keyring=/var/lib/ceph/mon/ceph-ceph-mon0/keyring", "auth", "getclient.admin"], ["PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin", "LANG=en_US.UTF-8", "CLUSTER=ceph", "TCMALLOC_MAX_TOTAL_THREAD_CACHE_BYTES=134217728", "CEPH_AUTO_RESTART_ON_UPGRADE=no"] Signed-off-by: Patrick Donnelly (cherry picked from commit 482022233d845b75876b04ca23fb137281a9f6ab) commit 0c38c464fff2280a6345b470f1c83aa6229776cc Merge: eb67259 eea546f Author: Loic Dachary Date: Mon Nov 7 14:39:48 2016 +0100 Merge pull request #11679 from dachary/wip-17734-jewel jewel: Upgrading 0.94.6 -> 0.94.9 saturating mon node networking Reviewed-by: Kefu Chai commit eea546fafcf22573e33332ce91e3d112202ac207 Author: xie xingguo Date: Sat May 21 14:11:55 2016 +0800 mon: expose require_jewel_osds flag to user Signed-off-by: xie xingguo (cherry picked from commit 83ffc2b761742d563777e50959faa6a6010edae0) commit f8ee076ac4559dc9dbf828121618e78ad11687fd Author: Sage Weil Date: Fri Oct 21 12:25:08 2016 -0400 mon/OSDMonitor: encode OSDMap::Incremental with same features as OSDMap The Incremental encode stashes encode_features, which is what we use later to reencode the updated OSDMap. Use the same features so that the encoding will match! Signed-off-by: Sage Weil (cherry picked from commit 916ca6a0aaa32bd9c2b449e0d7fbd312c29f06e5) commit 1f629b2ba91c793db34614f319d12594646f5ce2 Author: Sage Weil Date: Thu Oct 13 12:16:40 2016 -0400 mon/OSDMonitor: health warn if require_{jewel,kraken} flags aren't set We want to prompt users to set these flags as soon as their upgrades complete. Signed-off-by: Sage Weil (cherry picked from commit 12e508313dbd5d1d38c76859cb7de2ce22404e12) Conflicts: src/mon/OSDMonitor.cc: remove references to kraken if ((osdmap.get_up_osd_features() & CEPH_FEATURE_SERVER_KRAKEN) && !osdmap.test_flag(CEPH_OSDMAP_REQUIRE_KRAKEN)) { string msg = "all OSDs are running kraken or later but the" " 'require_kraken_osds' osdmap flag is not set"; summary.push_back(make_pair(HEALTH_WARN, msg)); if (detail) { detail->push_back(make_pair(HEALTH_WARN, msg)); } } else commit 34555f11b068eb335866d4b536c9e10fe1de62e5 Author: Sage Weil Date: Fri Sep 30 18:02:39 2016 -0400 mon/OSDMonitor: encode canonical full osdmap based on osdmap flags If the JEWEL or KRAKEN flags aren't set, encode the full map without those features. This ensure that older OSDs in the cluster will be able to correctly encode the full map with a matching CRC. At least, that is true as long as the encoding changes are guarded by those feature bits. That appears to be true currently, and we plan to ensure that it is true in the future as well. Signed-off-by: Sage Weil (cherry picked from commit 5e0daf6642011bf1222c4dc20aa284966fa5df9f) Conflicts: src/mon/OSDMonitor.cc: removed reference to kraken if (!tmp.test_flag(CEPH_OSDMAP_REQUIRE_KRAKEN)) { dout(10) << __func__ << " encoding without feature SERVER_KRAKEN" << dendl; features &= ~CEPH_FEATURE_SERVER_KRAKEN; } commit eb672598ad8209dcb7b7454fea31f921d255af1f Merge: 5e079cf d48e603 Author: Loic Dachary Date: Fri Nov 4 15:31:05 2016 +0100 Merge pull request #11742 from tchaikov/wip-17728-jewel jewel: test/ceph_test_msgr: do not use Message::middle for holding transient… Reviewed-by: Loic Dachary commit 5e079cfd46e4649e0eb24b15cb2c45887f5762fe Merge: 7b3ec11 eb30cc5 Author: Loic Dachary Date: Thu Nov 3 15:54:28 2016 +0100 Merge pull request #11746 from liewegas/wip-post-file-key-jewel jewel: ceph-post-file: use new ssh key Reviewed-by: Loic Dachary Reviewed-by: Brad Hubbard commit eb30cc5ad6b439417298b7c615ff8ae15bf00fa2 Author: David Galloway Date: Fri Aug 19 16:11:32 2016 -0400 ceph-post-file: Ignore keys offered by ssh-agent In my case, I had multiple private keys in ssh-agent which resulted in the sftp connection failing despite explicitly specifying the private key to use Signed-off-by: David Galloway (cherry picked from commit a61fcb2eac35a149b49efdc9b2ffa675afb968e8) commit 43282b0657ff19060dad25df79981ce17a76900f Author: Sage Weil Date: Wed Nov 2 09:37:41 2016 -0400 ceph-post-file: migrate to RSA SSH keys DSA keys are being deprecated: http://www.openssh.com/legacy.html drop.ceph.com will continue to allow the old DSA key but eventually, users submitting logs using ceph-post-file will run into issues when OpenSSH completely drops support for the algorithm. Fixes: http://tracker.ceph.com/issues/14267 Signed-off-by: David Galloway (cherry picked from commit ecd02bf3f1c7a07a3271b2736a9e12dd6e897821) # Conflicts: # src/CMakeLists.txt commit d48e603d6c6715fbc127003226d327b79f30713a Author: Sage Weil Date: Sun Oct 23 18:40:57 2016 -0500 msg: adjust byte_throttler from Message::encode Normally we never call encode on a message that has a byte_throttler set because we only use it for messages we received. However, for forwarded messages that we clear_payload() before resending, we *do* reencode, and in that case we need to retake the appropriate number of bytes from the throttler--just like we release them in clear_payload(). Signed-off-by: Sage Weil (cherry picked from commit a9651282f7c16df872757b82d3d2995d92458d5c) commit 1bc616138fea897f36c1c25851f87df91404011c Author: Sage Weil Date: Sun Oct 23 18:10:00 2016 -0500 msg/Message: fix set_middle vs throttler Signed-off-by: Sage Weil (cherry picked from commit e7bf50b27a495ed75def67025d1ceca83861ba35) commit 5d5b5952322bb6a571c9cd58fd3e683eb32a2509 Author: Sage Weil Date: Sat Oct 22 14:01:34 2016 -0400 messages/MForward: reencode forwarded message if target has differing features This ensures we reencode the payload with the appropriate set of features if the client, us, or the target do not have identical features. Otherwise we may forward an encoding with more features than the target can handle. Signed-off-by: Sage Weil (cherry picked from commit a433455e59067a844c3df4a0d6080db2ceb4ec59) commit e068c9206a3d618b3b04975d03f61ca64a92c4d4 Author: Sage Weil Date: Wed Sep 28 11:44:28 2016 -0400 messages/MForward: fix encoding features We were encoding the message with the sending client's features, which makes no sense: we need to encode with the recipient's features so that it can decode the message. The simplest way to fix this is to rip out the bizarre msg_bl handling code and simply keep a decoded Message reference, and encode it when we send. We encode the encapsulated message with the intersection of the target mon's features and the sending client's features. This probably doesn't matter, but it's conceivable that there is some feature-dependent behavior in the message encode/decode that is important. Fixes: http://tracker.ceph.com/issues/17365 Signed-off-by: Sage Weil (cherry picked from commit d4f5e88f36e5388ae9e062c4bc49ac1c684a3f3c) commit 158b003751d8bedafdca60d859aef67e69d9a732 Author: Michal Jarzabek Date: Sat Jun 4 23:24:06 2016 +0100 all: add const to operator<< param Signed-off-by: Michal Jarzabek (cherry picked from commit 0a157e088b2e5eb66177421f19f559ca427240eb) commit 3e1edde98610b11b94c59d23de979d6cd79dd8fe Author: Kefu Chai Date: Sat Oct 29 01:54:58 2016 +0800 test/ceph_test_msgr: do not use Message::middle for holding transient data Message::middle is used for holding encoded data, so we we can not stuff it with payload and leave the "payload" field empty. this change refactors the ceph_test_msgr by introducing a Payload class which encodes all test data in it. Fixes: http://tracker.ceph.com/issues/17728 Signed-off-by: Kefu Chai (cherry picked from commit 56896a7ed20869ce91ade4c77c1d6cbab8d50de1) Conflicts: src/test/msgr/test_msgr.cc: do not use the new-style DENC() framework for implementing the encoder of Payload class. DENC() was introduced after jewel was released. commit 8f75bd64768b6984c537dd15b5d9159070b86d91 Author: Kefu Chai Date: Mon Aug 8 23:20:58 2016 +0800 test/ceph_test_msgr: fix circular locking dependency * do not acquire lock when sending message * remove lock in session * reduce the scope guarded by locks for better performance. Fixes: http://tracker.ceph.com/issues/16955 Signed-off-by: Kefu Chai (cherry picked from commit cf1801c260c42aa93850538eea7a194440ebe350) commit f960db4646a9535bcee6d53740811b84e0678c93 Author: Haomai Wang Date: Thu Jul 7 14:59:51 2016 +0800 ceph_test_msgr: use ceph log infrastructure to output because we want to get the right log sequence which mixes ceph logginer and cerr. Otherwise, cerr output make the logs a little disordered. Signed-off-by: Haomai Wang (cherry picked from commit d1268a6aa895ee93cd9fee6fc6c759317e681a85) commit 7b3ec119cb968a26526ad95355c5bf7525fb5346 Merge: 5efb6b1 779af22 Author: Josh Durgin Date: Tue Nov 1 17:05:27 2016 -0700 Merge pull request #11728 from ceph/wip-librados-upgrade-jewel qa: remove EnumerateObjects from librados upgrade tests commit 5efb6b1c2c9eb68f479446e7b42cd8945a18dd53 Merge: 3c9fe54 39b8e78 Author: Loic Dachary Date: Fri Oct 28 14:29:20 2016 +0200 Merge pull request #11669 from ceph/wip-jewel-11566 rgw: fix put_acls for objects starting and ending with underscore Reviewed-by: Loic Dachary commit 3c9fe545dccf13413bb491098c089d3e4a76bcd2 Merge: 5ca8791 6356664 Author: Loic Dachary Date: Thu Oct 27 21:40:26 2016 +0200 Merge pull request #11472 from dachary/wip-17510-jewel jewel: ERROR: got unexpected error when trying to read object: -2 Reviewed-by: Yehuda Sadeh commit 39b8e783defb930b1dd8eeecdfee68d0d886d03b Author: Orit Wasserman Date: Wed Oct 19 21:51:01 2016 +0200 rgw: fix put_acls for objects starting and ending with underscore Fixes: http://tracker.ceph.com/issues/17625 Signed-off-by: Orit Wasserman (cherry picked from commit 14d4d912c7b47c56b16ae0bdc6bc08d208de3461) commit 5ca879114fcc98b906cac64a5ef5cb2a8568cb60 Merge: ed9a824 dc2ffda Author: Matt Benjamin Date: Thu Oct 27 13:52:22 2016 -0400 Merge pull request #11662 from linuxbox2/jewel-17635 jewel: rgw: handle empty POST condition commit ed9a824f2075e15ac245b9aac683cb28183ecd3a Merge: e08b0f3 f5e37ab Author: Loic Dachary Date: Thu Oct 27 16:48:43 2016 +0200 Merge pull request #11634 from dillaman/wip-17590-jewel jewel: journal: do not prematurely flag object recorder as closed Reviewed-by: Jason Dillaman commit 779af22fc920f1fdfdd3fa2b01d8587088372bd4 Author: Josh Durgin Date: Wed Oct 26 16:33:53 2016 -0700 qa: remove EnumerateObjects from librados upgrade tests These rely on new rados functionality not present in jewel Signed-off-by: Josh Durgin commit e08b0f308f5af45c1b1867ab5b757486bba51333 Merge: 0aee633 4cb83c1 Author: Loic Dachary Date: Wed Oct 26 23:00:35 2016 +0200 Merge pull request #11644 from ceph/wip-17695 jewel: librbd: discard after write can result in assertion failure Reviewed-by: Jason Dillaman commit 4cb83c14dbe09d4b371f7b728d9b5c0549e59f1a Author: Jason Dillaman Date: Tue Oct 25 09:43:06 2016 -0400 librbd: discard after write can result in assertion failure With journaling enabled, the proper lock is not held when handling a discard after write to overlapping extents. This issue is only present on the jewel branch due to design changes on the master branch. Fixes: http://tracker.ceph.com/issues/17695 Signed-off-by: Jason Dillaman commit dc2ffda7819d2ebeed3526d9e6da8f53221818de Author: Yehuda Sadeh Date: Thu Oct 20 10:17:36 2016 -0700 rgw: handle empty POST condition Fixes: http://tracker.ceph.com/issues/17635 Before accessing json entity, need to check that iterator is valid. If there is no entry return appropriate error code. Signed-off-by: Yehuda Sadeh (cherry picked from commit 23cb642243e09ca4a8e104f62a3bb7b2cbb6ea12) commit 0aee6331ad93a3caf212b84412bb648171758fff Merge: 06f7d7a cd99a64 Author: Loic Dachary Date: Wed Oct 26 16:47:03 2016 +0200 Merge pull request #11657 from dachary/wip-17707-jewel jewel: ceph-disk: using a regular file as a journal fails Reviewed-by: Kefu Chai commit 06f7d7a9b2403cf8b7af5301dae575d6f21c71ce Merge: 7ff2c10 7964187 Author: Loic Dachary Date: Wed Oct 26 14:04:20 2016 +0200 Merge pull request #11321 from linuxbox2/jewel-11051 jewel: rgw: Do not archive metadata by default Reviewed-by: Matt Benjamin commit 7ff2c108ea12451b6dcb25dab4574f2de68162d1 Merge: 0fb486d 2f9a5be Author: Loic Dachary Date: Wed Oct 26 11:55:49 2016 +0200 Merge pull request #11626 from ceph/wip-jewel-11567 jewel: don't loop forever when reading data from 0 sized segment. Reviewed-by: Orit Wasserman commit 0fb486d59e1574e936564e4b048a089dee8b267b Merge: bf5c47a 86eef27 Author: Loic Dachary Date: Wed Oct 26 11:39:40 2016 +0200 Merge pull request #11478 from dachary/wip-17312-jewel jewel: build/ops: allow building RGW with LDAP disabled Reviewed-by: Orit Wasserman commit cd99a642a088c7cd010e45a81d6680cdfca16ca6 Author: Jayashree Candadai Date: Fri Oct 21 13:52:05 2016 -0400 ceph-disk: allow using a regular file as a journal Because of a missing return, ceph-disk prepare would fail if given a regular file as a journal. If the journal file does not exist, ceph-disk will create it but fail to ensure that the ceph user owns it. The symlink to the journal file is not set when the journal file is specified on the command line and the journal file does not exist at all. The ceph-osd daemon will silently create it as a file but it will not be the file given in argument. Add a test case to verify using a regular file as a journal works as expected. Fixes: http://tracker.ceph.com/issues/17662 Signed-off-by: Jayashree Candadai Signed-off-by: Loic Dachary (cherry picked from commit db917d50eb5f86a07a5487e130f46a6b1d27672a) commit a80040230cddc395809d2323392c87a4a1fef923 Author: Anirudha Bose Date: Wed Aug 17 12:19:33 2016 +0530 ceph-disk: PEP8ify Signed-off-by: Anirudha Bose (cherry picked from commit f7a15ccfa2cc214c3688a2b41ac45be1abfb7700) Conflicts: src/ceph-disk/setup.py: python 2.7 restriction not backported commit e200b17016a1a7cd103affde2d2b265916755498 Author: Anirudha Bose Date: Wed Aug 17 12:15:25 2016 +0530 ceph-disk: Set space_symlink to the path, not file object Signed-off-by: Anirudha Bose (cherry picked from commit d290454cf5f660c1681eefd70a38e79da683525f) commit 9191825c06b3d92e2caa26c9efdb69a79cc3964b Author: Anirudha Bose Date: Wed Aug 17 12:13:53 2016 +0530 ceph-disk: Use os.path.isabs instead of .startswith('/') Signed-off-by: Anirudha Bose (cherry picked from commit 6f795356de00dd7c33282183b11a03da800fde7b) commit f1c2de768535ce18259d4fc187cd370766c1e6cf Author: Anirudha Bose Date: Sun Jun 19 05:10:27 2016 +0530 ceph-disk: Use context manager with FileLock acquire and release methods of FileLock are dropped Signed-off-by: Anirudha Bose (cherry picked from commit c1011d514ce0c7d340a5acd6f9c640165e169156) commit 26cb5b6bfbd6eee46bfa3cf005b1cd9f06b0e262 Author: Anirudha Bose Date: Sun Jun 19 04:05:42 2016 +0530 ceph-disk: Fix bug in FileLock Python fcntl.lockf() accepts a file descriptor, not a file object Signed-off-by: Anirudha Bose (cherry picked from commit df9cc2ce938a969f4044b63fd80030d00f64f060) commit 91a9ca374f646966e31b015928a2cc60bc5225aa Author: Anirudha Bose Date: Thu Jun 16 16:37:00 2016 +0530 ceph-disk: Use true integer division in get_dev_size Signed-off-by: Anirudha Bose (cherry picked from commit f1bb72c82806cc03e85e0b19c83d61409c6b2d51) commit 2373ccb7a991069406e344b91fa70c0b2d1aa5e0 Author: Anirudha Bose Date: Tue Jun 14 21:42:02 2016 +0530 ceph-disk: Compatibility fixes for Python 3 ceph-disk: Misc cleanups Signed-off-by: Anirudha Bose (cherry picked from commit d0e29c74f84a2ed3014a516c0106172619314bdc) Conflicts: src/ceph-disk/tox.ini: python3 is not supported in jewel commit 79cf6330d67ad52af5690f4d4efd29aa2722acb2 Author: Shylesh Kumar Date: Thu Jul 7 20:45:57 2016 +0530 ceph-disk: change ownership of init file to ceph:ceph Fixes: http://tracker.ceph.com/issues/16280 Signed-off-by: Shylesh Kumar (cherry picked from commit aab9d03e1b50ba10a383663088400b9fabe306cb) commit 796418746ecd2a4971e365499c6952e131e74913 Author: root Date: Mon Sep 12 14:30:43 2016 +0530 rgw: Do not archive metadata by default Fixes: http://tracker.ceph.com/issues/17256 Signed-off-by: Pavan Rallabhandi Signed-off-by: Matt Benjamin (cherry picked from commit c617ea83e0cca0061af18e0811c7ef8b4e836519) commit bf5c47a28abe6d91dfdace8d6803a7016bbf02fb Merge: 7714689 1eedf18 Author: Loic Dachary Date: Tue Oct 25 15:49:55 2016 +0200 Merge pull request #11642 from tchaikov/wip-17685-jewel jewel: mon: fix missing osd metadata (again) Reviewed-by: Loic Dachary commit 77146891ab254bc796e38e2a45b8500c0119745f Merge: 3ccc17b e8e1acb Author: Loic Dachary Date: Tue Oct 25 15:40:01 2016 +0200 Merge pull request #11467 from dachary/wip-17262-jewel jewel: rbd-nbd IO hang Reviewed-by: Jason Dillaman commit 1eedf181176ce75212a8cfbb87ac927350a826ed Author: John Spray Date: Wed May 25 17:56:51 2016 +0100 mon: fix missing osd metadata (again) The JSON output was getting broken by continuing in the wrong place. Fixes: http://tracker.ceph.com/issues/17685 Signed-off-by: John Spray (cherry picked from commit c5700ce4b45b3a385fe4c2111da852bea7d86da2) commit f5e37abefcb015b2cb58295cfdf109fd6f5833b0 Author: Jason Dillaman Date: Mon Oct 17 09:48:20 2016 -0400 journal: do not prematurely flag object recorder as closed Fixes: http://tracker.ceph.com/issues/17590 Signed-off-by: Jason Dillaman (cherry picked from commit 2be6367cb0119d86dfecfa94eb4a3a102c40162a) commit 2f9a5beb9a903220b70a780b30564d9741e9f0e5 Author: Marcus Watts Date: Thu Oct 13 21:12:36 2016 -0400 Don't loop forever when reading data from 0 sized segment. The 0 sized segment can arise depending on how the client uploads the object in the first place. The cpu loop then happens during a swift `GET'. Signed-off-by: Marcus Watts (cherry picked from commit 46c5f9773246522e66bb2cca49345d0b62a16c42) commit 3ccc17b81d2794406d803ff4210e930a1fa67455 Merge: 112b89d 0b30a1d Author: Loic Dachary Date: Mon Oct 24 12:03:13 2016 +0200 Merge pull request #10757 from dachary/wip-17056-jewel jewel: mon/osdmonitor: decouple adjust_heartbeat_grace and min_down_reporters Reviewed-by: Kefu Chai commit 112b89dc315dd1240d68eb50fba1438057002327 Merge: 7ab1e02 0dcefd2 Author: Loic Dachary Date: Mon Oct 24 12:03:01 2016 +0200 Merge pull request #10759 from dachary/wip-16866-jewel jewel: OSD: ceph osd df does not show summarized info correctly if one or more OSDs are out Reviewed-by: Kefu Chai commit 7ab1e0275101a30c2d318bcd55787f61c687328f Merge: e053b20 4f9e02c Author: Loic Dachary Date: Mon Oct 24 11:55:41 2016 +0200 Merge pull request #11590 from dachary/wip-17642-jewel jewel: TestJournalReplay: sporadic assert(m_state == STATE_READY || m_state == STATE_STOPPING) failure Reviewed-by: Loic Dachary commit 4f9e02c40ac5359f344d29f20332ec6ed8575b8e Author: Jason Dillaman Date: Thu Oct 13 10:04:24 2016 -0400 test: TestJournalReplay test cases need to wait for committed journal event Fixes: http://tracker.ceph.com/issues/17566 Signed-off-by: Jason Dillaman (cherry picked from commit 1bdf7a425eb7436838c9445ea2c8ab42dfd3a1b6) commit e8e1acb1d5154b749d251efa88b45e8ad3edb2bb Author: Jason Dillaman Date: Thu Sep 8 11:51:34 2016 -0400 librbd: ignore cache busy errors when shrinking an image Signed-off-by: Jason Dillaman (cherry picked from commit 4ce663845679dc35f2f15b893c6f988c4a60b25b) Conflicts: src/test/librbd/operation/test_mock_ResizeRequest.cc: when_resize does not have the allow_shrink argument because d1f2c557b2c039730baca9efa3f5244bc19dcb1a has not been backported commit ba2e87e0443069b0552b698cc7a508898a3f585d Author: Jason Dillaman Date: Thu Sep 8 09:59:37 2016 -0400 librbd: invalidate cache before trimming image Any potential writeback outside the extents of a shrunk image would result in orphaned objects. Signed-off-by: Jason Dillaman (cherry picked from commit 3f93a1917437ba7b69b306e4ff971b79e8b79c89) commit d7c0873ea77234b7b736080c3de4012d2f6adaee Author: Jason Dillaman Date: Wed Sep 7 11:49:26 2016 -0400 rbd-nbd: mask out-of-bounds IO errors caused by image shrink Signed-off-by: Jason Dillaman (cherry picked from commit c6cfb616a846959a2cd1c0e540b14668c61a2afd) commit 0ce342d39f359d23c92ab94efb910b84a634094a Author: Jason Dillaman Date: Mon Sep 5 10:01:45 2016 -0400 rbd-nbd: fix kernel deadlock during teuthology testing Fixes: http://tracker.ceph.com/issues/16921 Signed-off-by: Jason Dillaman (cherry picked from commit ce7c1520a8019f011fa34dd898af317f78974577) commit e053b2085b0d0be0e8c9912c82c7142bb0871783 Merge: f6d7290 caf08d7 Author: Casey Bodley Date: Thu Oct 20 11:17:49 2016 -0400 Merge pull request #11519 from dachary/wip-17576-jewel jewel: RGW loses realm/period/zonegroup/zone data: period overwritten if somewhere in the cluster is still running Hammer Reviewed-by: Casey Bodley commit f6d729082359b450950513080c9c08461d6f557c Merge: 9e9a38d 40689d1 Author: Casey Bodley Date: Thu Oct 20 11:17:22 2016 -0400 Merge pull request #11471 from dachary/wip-17511-jewel jewel: s3tests-test-readwrite failing with 500 Reviewed-by: Casey Bodley commit 9e9a38df303749147a638f0df42ad875c829d294 Merge: 9b0fa2b 5a53ffa Author: Casey Bodley Date: Thu Oct 20 11:16:38 2016 -0400 Merge pull request #11469 from dachary/wip-17538-jewel jewel: rgw:user email can modify to empty when it has values Reviewed-by: Casey Bodley commit 9b0fa2b27386dd0c09aead832c9f049ef10b26fb Merge: 0eac635 4babd3f Author: Loic Dachary Date: Thu Oct 20 16:31:10 2016 +0200 Merge pull request #11466 from dachary/wip-17290-jewel jewel: ImageWatcher: use after free within C_UnwatchAndFlush Reviewed-by: Jason Dillaman commit 0eac63545da063a8bbab4b1ea6c9a765023617fc Merge: 51d5ed0 b73356b Author: Loic Dachary Date: Thu Oct 20 16:30:53 2016 +0200 Merge pull request #11464 from dachary/wip-17373-jewel jewel: image.stat() call in librbdpy fails sometimes Reviewed-by: Jason Dillaman commit 51d5ed0ddbe7e1018be1fa46c912d2e0ca07a984 Merge: d2150fb b410c8e Author: Loic Dachary Date: Thu Oct 20 16:30:44 2016 +0200 Merge pull request #11463 from dachary/wip-17384-jewel jewel: helgrind: TestLibRBD.TestIOPP potential deadlock closing an image with read-ahead enabled Reviewed-by: Jason Dillaman commit d2150fb54e894039e03855352708fad6b08d75d0 Merge: f8989ef 32d84e0 Author: Loic Dachary Date: Thu Oct 20 16:30:32 2016 +0200 Merge pull request #11462 from dachary/wip-17404-jewel jewel: update_features API needs to support backwards/forward compatibility Reviewed-by: Jason Dillaman commit f8989ef7a79dd37f58fc9bddbd1e3fcd83e14f42 Merge: ac2ce6d 7d6801b Author: Loic Dachary Date: Thu Oct 20 16:28:18 2016 +0200 Merge pull request #11459 from dachary/wip-17483-jewel jewel: RBD should restrict mirror enable/disable actions on parents/clones Reviewed-by: Jason Dillaman commit ac2ce6dcda27a70a45e17270cab4d5e9a856abc7 Merge: cd48521 391936a Author: Loic Dachary Date: Thu Oct 20 16:27:36 2016 +0200 Merge pull request #11460 from dachary/wip-17482-jewel jewel: Enable/Disable of features is allowed even the features are already enabled/disabled Reviewed-by: Jason Dillaman commit cd48521cde8bb69b6879977f51404a3014571a84 Merge: 410a368 dd93e00 Author: Loic Dachary Date: Thu Oct 20 16:26:53 2016 +0200 Merge pull request #11461 from dachary/wip-17481-jewel jewel: Proxied operations shouldn't result in error messages if replayed Reviewed-by: Jason Dillaman commit 410a368e9bdf31f6512be5e9a783c7990ba9003f Merge: 07a8d8a d66bb7a Author: Loic Dachary Date: Thu Oct 20 12:42:01 2016 +0200 Merge pull request #11431 from ceph/wip-jewel-tarball jewel: build/ops: include more files in "make dist" tarball Reviewed-by: Nathan Cutler Reviewed-by: Boris Ranto commit 07a8d8aedfa101a52e4764cc211fd2a1d8f36582 Merge: 62f7f64 50fd48f Author: Loic Dachary Date: Thu Oct 20 11:48:49 2016 +0200 Merge pull request #11474 from dachary/wip-17350-jewel jewel: rgw:response information is error when geting token of swift account Reviewed-by: Orit Wasserman commit 62f7f646b85f81c0c812f3d5adc1926d4cee3f7d Merge: 638590c 91bd342 Author: Loic Dachary Date: Thu Oct 20 11:47:43 2016 +0200 Merge pull request #11492 from SUSE/wip-17575-jewel jewel: aarch64: Compiler-based detection of crc32 extended CPU type is broken Reviewed-by: Kefu Chai commit 638590c97a5365df118c72aab2e510ffcd2a9470 Merge: c58d626 6575545 Author: Loic Dachary Date: Thu Oct 20 11:28:29 2016 +0200 Merge pull request #11475 from dachary/wip-17349-jewel jewel: Modification for TEST S3 ACCESS section in INSTALL CEPH OBJECT GATEWAY page Reviewed-by: Loic Dachary commit c58d62677449b08023b761762a52cc4ceaf69def Merge: 3cb4bf4 7423661 Author: Loic Dachary Date: Thu Oct 20 11:27:54 2016 +0200 Merge pull request #11473 from dachary/wip-17509-jewel jewel: Config parameter rgw keystone make new tenants in radosgw multitenancy does not work Reviewed-by: Loic Dachary commit 3cb4bf43b923a71872525b0222b4b0fc73664fb0 Merge: fb74b16 de672a0 Author: Loic Dachary Date: Thu Oct 20 11:27:18 2016 +0200 Merge pull request #11201 from ceph/wip-backport-logrotate-jewel jewel: build/ops: backport 'logrotate: Run as root/ceph' Reviewed-by: Loic Dachary commit fb74b1688603c0351e80ed504be4faad320a67e6 Merge: 9d3373d 25a35d4 Author: Loic Dachary Date: Thu Oct 20 11:22:45 2016 +0200 Merge pull request #11126 from tchaikov/wip-17179-jewel jewel: add a tool to rebuild mon store from OSD Reviewed-by: huanwen ren Reviewed-by: Ken Dreyer commit 9d3373d71f8af16a8f0220aa0d0682e2f3bb5a8c Merge: 0c83eb3 eb6c3cb Author: Loic Dachary Date: Thu Oct 20 09:39:15 2016 +0200 Merge pull request #11563 from ceph/wip-jewel-acl-underscore rgw: fix regression with handling double underscore Reviewed-by: Loic Dachary commit eb6c3cbcc6cadd4eff9de0d6332f42c785486f7b Author: Yehuda Sadeh Date: Thu Sep 29 18:35:59 2016 -0700 rgw: set correct instance on the object Fixes: http://tracker.ceph.com/issues/17443 This was broken by commit bc840afafdfe5e528e5c1b711e71420ac3cb5a67 Signed-off-by: Yehuda Sadeh (cherry picked from commit d2ada35c8c0d938a704ecf9974988ea30a9c0105) commit 0c83eb355e989fb6ed38a3b82f9705fd5d700e89 Merge: 7496388 f400ff2 Author: Loic Dachary Date: Wed Oct 19 23:14:23 2016 +0200 Merge pull request #11548 from dachary/wip-17609-jewel jewel: tests: ceph-disk must ignore debug monc Reviewed-by: Sage Weil commit 7496388468688763cab7fab9cc2f61cfaeeb4e7c Merge: f8055a9 d205b74 Author: Loic Dachary Date: Wed Oct 19 23:11:24 2016 +0200 Merge pull request #11411 from dachary/wip-17245-jewel jewel: tests: scsi_debug fails /dev/disk/by-partuuid Reviewed-by: Loic Dachary commit f8055a9453e228079a8c61ba398e93e9fbdcf55a Merge: 3b2d360 dea93dc Author: Loic Dachary Date: Wed Oct 19 23:11:04 2016 +0200 Merge pull request #10884 from dachary/wip-17149-jewel jewel: ceph-disk: expected systemd unit failures are confusing Reviewed-by: Loic Dachary commit 084108e440b7b9b8f0d36282dd4eba64a4b96f4b Author: Orit Wasserman Date: Wed Aug 31 22:25:57 2016 +0200 rgw: fix regression with handling double underscore Fixes: http://tracker.ceph.com/issues/16856 Signed-off-by: Orit Wasserman (cherry picked from commit bc840afafdfe5e528e5c1b711e71420ac3cb5a67) commit 3b2d36039305bc070ac1eaf5142d5b5bfa7accb1 Merge: d15d6dc 483d8c4 Author: Loic Dachary Date: Wed Oct 19 16:50:44 2016 +0200 Merge pull request #11408 from dachary/wip-17345-jewel jewel: Ceph Status - Segmentation Fault Reviewed-by: Kefu Chai commit d15d6dc896d0ac30ddff4354600d7c2ecb61b54a Merge: d4197dc 5ae4f31 Author: Loic Dachary Date: Wed Oct 19 16:50:22 2016 +0200 Merge pull request #11407 from dachary/wip-17360-jewel jewel: ceph-objectstore-tool crashes if --journal-path Reviewed-by: Kefu Chai commit d4197dcae43b073d121c4b12ff7cca2f619745e1 Merge: 54bb909 c94244d Author: Loic Dachary Date: Wed Oct 19 09:47:00 2016 +0200 Merge pull request #11158 from dillaman/wip-rbdmap-jewel jewel: systemd: add install section to rbdmap.service file Reviewed-by: Loic Dachary commit 54bb9092a8d658ab2f352ed8c850d719b2468d83 Merge: dc59575 c3c2910 Author: Loic Dachary Date: Wed Oct 19 09:46:19 2016 +0200 Merge pull request #10862 from dachary/wip-17095-jewel jewel: rpm: ceph installs stuff in %_udevrulesdir but does not own that directory Reviewed-by: Loic Dachary commit f400ff222e531feb00de3ebf1860a0d4bcca736e Author: Loic Dachary Date: Tue Oct 18 17:33:23 2016 +0200 tests: ceph-disk: force debug monc = 0 The sh function will collect both stderr and stdout and debug will mess the json parsing. Fixes: http://tracker.ceph.com/issues/17607 Signed-off-by: Loic Dachary (cherry picked from commit 469a53a4adc335ea4ff8e34a958b12f0f222a6c1) commit dc59575b2fec477bdcf152f91a02532fcdfd1a10 Merge: 778d884 c2cf21d Author: Loic Dachary Date: Wed Oct 19 08:56:46 2016 +0200 Merge pull request #10860 from dachary/wip-17121-jewel jewel: the %USED of ceph df is wrong Reviewed-by: Kefu Chai Reviewed-by: Samuel Just commit 778d884512d2cf298c60634cda9a6f66dc2b948b Merge: 82edc5b a25a483 Author: Loic Dachary Date: Wed Oct 19 01:35:13 2016 +0200 Merge pull request #10886 from dachary/wip-17144-jewel jewel: mark_all_unfound_lost() leaves unapplied changes Reviewed-by: David Zafman commit 82edc5bc526e02d7c535d01642d8ef7f00d19939 Merge: c4f4f8a e6ac214 Author: Loic Dachary Date: Tue Oct 18 13:05:43 2016 +0200 Merge pull request #10784 from dachary/wip-17067-jewel jewel: Request exclusive lock if owner sends -ENOTSUPP for proxied maintenance op Reviewed-by: Ilya Dryomov commit 25a35d43a8bf7e1a7d78d5d1d2a5556dff98f5e8 Author: Kefu Chai Date: Mon Oct 10 18:43:39 2016 +0800 doc: fill keyring with caps before passing it to ceph-monstore-tool to make sure the recovered monitor store is ready for use. Signed-off-by: Kefu Chai (cherry picked from commit af8e21163735377071b6832d8a81b035bb835257) commit 73ea9264803bdf53a5da64992c1b91e94633f5e3 Author: Kefu Chai Date: Mon Oct 10 16:32:27 2016 +0800 tools/ceph_monstore_tool: bail out if no caps found for a key we take it as an error if no caps is granted to an entity in the specified keyring file when rebuilding the monitor db. Signed-off-by: Kefu Chai (cherry picked from commit b4bd4004a836121c11b0bb97d8123df54c271f04) commit 4ebf87bccdb17bd04b2c615e6278a7816ae1ff43 Author: Kefu Chai Date: Fri Sep 30 17:58:14 2016 +0800 tools/ceph_monstore_tool: update pgmap_meta also when rebuilding store.db we should rebuild pgmap_meta table from the collected osdmaps Fixes: http://tracker.ceph.com/issues/17400 Signed-off-by: Kefu Chai (cherry picked from commit cdfa7a69f63d047205dcfccd63b5d58ab0d4695b) commit d2deed6bfbc5693e3bd8a10a2f9be8df8ef1b0e5 Author: xie xingguo Date: Sun Sep 18 11:40:56 2016 +0800 tools/rebuild_mondb: kill compiling warning As follow: [ 72%] Building CXX object src/tools/CMakeFiles/ceph-objectstore-tool.dir/RadosDump.cc.o /home/jenkins-build/build/workspace/ceph-pull-requests/src/tools/rebuild_mondb.cc: In function ‘int update_mon_db(ObjectStore&, OSDSuperblock&, const string&, const string&)’: /home/jenkins-build/build/workspace/ceph-pull-requests/src/tools/rebuild_mondb.cc:289:22: warning: ‘crc’ may be used uninitialized in this function [-Wmaybe-uninitialized] if (have_crc && osdmap.get_crc() != crc) { ^ /home/jenkins-build/build/workspace/ceph-pull-requests/src/tools/rebuild_mondb.cc:238:14: note: ‘crc’ was declared here uint32_t crc; Signed-off-by: xie xingguo (cherry picked from commit f16a31476a3f9b44a7c3dabf0dfd2a0d015b11b9) commit 09701269de225e556099b9e5c511faa44acae024 Author: xie xingguo Date: Sun Sep 18 10:33:56 2016 +0800 tools/rebuild_mondb: avoid unnecessary result code cast In general we return negative codes for error cases, so there is no need perform the cast here. Signed-off-by: xie xingguo (cherry picked from commit 6a1c01d334fe65124043aa68a6e0cfaea43836b5) commit 5191b06294cf505716d0c64ac36528e9ea57b0a8 Author: Kefu Chai Date: Sat Oct 1 14:18:55 2016 +0800 doc: add rados/operations/disaster-recovery.rst document the process to recover from leveldb corruption. Fixes: http://tracker.ceph.com/issues/17179 Signed-off-by: Kefu Chai (cherry picked from commit 79a9f299253e24d20547131b3c9c9e0667e3b869) Conflicts: src/tools/rebuild_mondb.cc: remove the code change in this file from this commit. and the code gets removed is added in anther commit. commit 8c8d5ce529fa826bc0d453edf5fb5e98e29294d3 Author: Kefu Chai Date: Mon Aug 29 19:53:11 2016 +0800 tools/ceph_monstore_tool: add "rebuild" command Fixes: http://tracker.ceph.com/issues/17179 Signed-off-by: Kefu Chai (cherry picked from commit d909fa035c8fbbded786b2ca072acc10ea6b6052) commit 1fcb0bb81a8989b31a79bc565354f6817ceb12d7 Author: Kefu Chai Date: Mon Aug 29 19:52:19 2016 +0800 tools/ceph-objectstore-tool: add "update-mon-db" command Fixes: http://tracker.ceph.com/issues/17179 Signed-off-by: Kefu Chai Conflicts: src/tools/CMakeLists.txt: this file was added in master, so update src/CMakeLists.txt instead src/tools/Makefile-server.am: jewel is still using autotools, so update this file also. src/tools/rebuild_mondb.cc: move the code spilled into doc/rados/troubleshooting/troubleshooting-mon.rst by accident back to this commit. (cherry picked from commit 24faea7ce446bbf09cbd4a9d3434dd5444a6c295) commit 416750258fb63064a36eaf53f586a51fc3ea63e0 Author: Kefu Chai Date: Wed Aug 31 13:11:24 2016 +0800 mon/AuthMonitor: make AuthMonitor::IncType public so ceph-objectstore-tool is able to use it when rebuilding monitor db. Fixes: http://tracker.ceph.com/issues/17179 Signed-off-by: Kefu Chai (cherry picked from commit 19ef4f16b3aba04119ac647cd6261c74a57ce829) commit c4f4f8a72421e9c43c09eb9dfb5a3c6fe2123886 Merge: ded7c27 5ef9085 Author: Samuel Just Date: Mon Oct 17 11:56:30 2016 -0700 Merge pull request #10885 from dachary/wip-17145-jewel jewel: PG::choose_acting valgrind error or ./common/hobject.h: 182: FAILED assert(!max || (*this == hobject_t(hobject_t::get_max()))) Reviewed-by: Samuel Just commit ded7c27a2d0fc0ae307321de2bf419e065c36ba1 Merge: bdcdedd 58b7c52 Author: Samuel Just Date: Mon Oct 17 11:54:29 2016 -0700 Merge pull request #10883 from dachary/wip-17141-jewel jewel: PG::_update_calc_stats wrong for CRUSH_ITEM_NONE up set items Reviewed-by: Samuel Just commit bdcdedd32f9e8a71735ae1d4d3ca18b41bde3378 Merge: 757f8e8 3bb2a9e Author: Loic Dachary Date: Mon Oct 17 18:08:46 2016 +0200 Merge pull request #11420 from dachary/wip-17556-jewel jewel: librbd::Operations: update notification failed: (2) No such file or directory Reviewed-by: Jason Dillaman commit 757f8e87e0aa39af05f0bc7b8146c826191e8032 Merge: 4d2bc33 bd63666 Author: Loic Dachary Date: Mon Oct 17 18:08:32 2016 +0200 Merge pull request #10857 from dachary/wip-16984-jewel jewel: Disabling pool mirror mode with registered peers results orphaned mirrored images Reviewed-by: Jason Dillaman commit 4d2bc33101aaadbab73cdae3327fc0888a15e25a Merge: 23d91cf 18a66cb Author: Loic Dachary Date: Mon Oct 17 17:55:10 2016 +0200 Merge pull request #10645 from dachary/wip-16458-jewel jewel: Potential crash during journal::Replay shut down Reviewed-by: Jason Dillaman commit 23d91cfcd1c77766972577639ef72db353076e37 Merge: 51c2794 775c78d Author: Loic Dachary Date: Mon Oct 17 17:54:55 2016 +0200 Merge pull request #10652 from dachary/wip-16951-jewel jewel: ceph 10.2.2 rbd status on image format 2 returns (2) No such file or directory Reviewed-by: Jason Dillaman commit 51c279425e26626884ec54add6043c7ed3873920 Merge: ee32be2 bb81f9d Author: Loic Dachary Date: Mon Oct 17 17:37:41 2016 +0200 Merge pull request #11337 from SUSE/wip-17060-jewel jewel: librbd: cannot disable journaling or remove non-mirrored, non-primary image Reviewed-by: Jason Dillaman commit ee32be2ae3c328238fd2bbb953c18d3b9f0a51b1 Merge: 7d0dd1c 2b27212 Author: Loic Dachary Date: Mon Oct 17 17:37:06 2016 +0200 Merge pull request #10650 from dachary/wip-16868-jewel jewel: Prevent the creation of a clone from a non-primary mirrored image Reviewed-by: Loic Dachary commit 7d0dd1cf6f6568f86fa593de0a7845462a510d31 Merge: f1c21c6 1c76ef4 Author: Loic Dachary Date: Mon Oct 17 17:06:51 2016 +0200 Merge pull request #11433 from dillaman/wip-17416-jewel jewel: rbd-mirror: improve resiliency of stress test case Reviewed-by: Loic Dachary commit f1c21c6f01e20949753198e5f3ad160e0213cc5b Merge: 986a8b5 80e25b0 Author: Loic Dachary Date: Mon Oct 17 17:04:11 2016 +0200 Merge pull request #10796 from dillaman/wip-17059-jewel jewel: rbd: bench io-size should not be larger than image size Reviewed-by: Loic Dachary commit 986a8b5eacbf0d6084121eef8407088a24d27ffd Merge: c7b7f2e 10e603b Author: Loic Dachary Date: Mon Oct 17 14:28:06 2016 +0200 Merge pull request #11414 from dachary/wip-17477-jewel jewel: Crash in Client::_invalidate_kernel_dcache when reconnecting during unmount Reviewed-by: John Spray commit c7b7f2e8c0de596255e651ae5e499c586f1e4493 Merge: ada7edd 3320da0 Author: Loic Dachary Date: Mon Oct 17 14:27:53 2016 +0200 Merge pull request #11412 from dachary/wip-17479-jewel jewel: Duplicate damage table entries Reviewed-by: John Spray commit ada7edd1d21073b71c1d4538ab66b15b978cf418 Merge: 1412b17 3a79db1 Author: Loic Dachary Date: Mon Oct 17 14:27:43 2016 +0200 Merge pull request #11415 from dachary/wip-17476-jewel jewel: Failure in snaptest-git-ceph.sh Reviewed-by: John Spray commit 1412b171c057fabd05abaf9adcf31aa29a4f5b00 Merge: 368c96c 0a17741 Author: Loic Dachary Date: Mon Oct 17 14:27:30 2016 +0200 Merge pull request #11416 from dachary/wip-17474-jewel jewel: Failure in dirfrag.sh Reviewed-by: John Spray commit 368c96c513cd5424d7935fc02edd6297fb680bb7 Merge: 4486e3b 2c4e1c1 Author: Loic Dachary Date: Mon Oct 17 14:27:17 2016 +0200 Merge pull request #11418 from dachary/wip-17246-jewel jewel: Log path as well as ino when detecting metadata damage Reviewed-by: John Spray commit 4486e3b5e00e809bc3391613cc5160810ce5bf1e Merge: b174220 5173563 Author: Loic Dachary Date: Mon Oct 17 14:27:04 2016 +0200 Merge pull request #11419 from dachary/wip-17244-jewel jewel: Failure in snaptest-git-ceph.sh Reviewed-by: John Spray commit b1742204e10d563573bbda82917b843c01b20420 Merge: 54240c4 c0db9fb Author: Loic Dachary Date: Mon Oct 17 14:26:30 2016 +0200 Merge pull request #10877 from dachary/wip-16946-jewel jewel: client: nlink count is not maintained correctly Reviewed-by: John Spray commit 54240c4d3dfeef97c6fba343fe501f0303c73f1b Merge: 8b5aa5d 430ab1b Author: Loic Dachary Date: Mon Oct 17 13:40:13 2016 +0200 Merge pull request #10758 from dachary/wip-17007-jewel jewel: ceph-disk should timeout when a lock cannot be acquired Reviewed-by: Loic Dachary commit caf08d731c7cec40fe6635189de877d1f047e8b4 Author: Orit Wasserman Date: Tue Oct 11 14:18:29 2016 +0200 rgw: avoid corruption when running old radosgw-admin on a newer rgw Fixes:http://tracker.ceph.com/issues/17371 Signed-off-by: Orit Wasserman (cherry picked from commit 8eab9454b4d45ed8092a1292bd1904ef928c2120) commit e5f7854ddfc2a6c123d77b1d33e659b4f870fbe6 Author: Orit Wasserman Date: Tue Oct 11 11:19:01 2016 +0200 rgw: Fix missing master zone for default region conversion Fixes:http://tracker.ceph.com/issues/17371 Signed-off-by: Orit Wasserman (cherry picked from commit cd3bccdef2ecd1d294ab1192ab3b9e731b5592e0) commit ec2fb022307eea93a802f80307b957bfb2581249 Author: Orit Wasserman Date: Tue Oct 11 11:00:00 2016 +0200 rgw: mark configuration as converted to avoid corruption when running older admins Fixes: http://tracker.ceph.com/issues/17371 Signed-off-by: Orit Wasserman (cherry picked from commit ceafa2863dd9c02da8a30c5a5324b472ed5d3404) commit 8b5aa5d94fd24793c160372f02b02735632b5fd0 Merge: 6baac41 2311ab2 Author: Loic Dachary Date: Fri Oct 14 18:13:09 2016 +0200 Merge pull request #10864 from dachary/wip-17131-jewel jewel: Jewel: segfault in ObjectCacher::FlusherThread Reviewed-by: Greg Farnum commit 6baac417eb95df880b97cdc720e6b00f6f8f27d8 Merge: 19922c5 ca8fc6f Author: Sage Weil Date: Fri Oct 14 09:26:32 2016 -0500 Merge pull request #11417 from dachary/wip-17347-jewel jewel: ceph-create-keys: sometimes blocks forever if mds allow is set Reviewed-by: Sage Weil commit 19922c5b818e82c297dfc0d2eb914276945e9a2e Merge: 952a81a 35660d1 Author: Sage Weil Date: Fri Oct 14 09:25:42 2016 -0500 Merge pull request #11272 from aiicore/wip-17402-jewel jewel: OSDMonitor: Missing nearfull flag set Reviewed-by: Sage Weil commit 952a81aed3bae88239e448fcae1ce73e725d9966 Merge: e34b7da 165e5ab Author: Sage Weil Date: Fri Oct 14 09:24:42 2016 -0500 Merge pull request #11193 from SUSE/wip-17377-jewel jewel: LIBRADOS modify Pipe::connect() to return the error code Reviewed-by: Sage Weil commit e34b7dad7e11f1db45f2bd636b1f3ed73742540a Merge: f9c969e 7d92e2e Author: Sage Weil Date: Fri Oct 14 09:23:30 2016 -0500 Merge pull request #10861 from dachary/wip-17135-jewel jewel: ceph mon Segmentation fault after set crush_ruleset ceph 10.2.2 Reviewed-by: Sage Weil commit f9c969eb54fef6f474bc8f784735123039ffeccc Merge: 4bcc21b 117aa35 Author: Sage Weil Date: Fri Oct 14 09:22:05 2016 -0500 Merge pull request #10855 from dachary/wip-16657-jewel jewel: i386 tarball gitbuilder failure on master Reviewed-by: Sage Weil commit 4bcc21b9288ed046a2781e95e781bfc71636b9c1 Merge: 92a3538 5d0e2f8 Author: Loic Dachary Date: Fri Oct 14 15:37:25 2016 +0200 Merge pull request #11409 from dachary/wip-17341-jewel jewel: librados memory leaks from ceph::crypto (WITH_NSS) Reviewed-by: Orit Wasserman commit 92a353847852dbec003e4795e3322d2a92752b9d Merge: 67d0ded 8833c64 Author: Loic Dachary Date: Fri Oct 14 13:45:46 2016 +0200 Merge pull request #11311 from dzafman/wip-scrub-boundary-jewel jewel: osd: adjust scrub boundary to object without SnapSet Reviewed-by: Sage Weil commit 67d0ded8d3ceda56ab7c0aaca25f3c1c51731c72 Merge: 52e596f 9cb45e1 Author: Loic Dachary Date: Fri Oct 14 13:43:27 2016 +0200 Merge pull request #11231 from badone/wip-17376 jewel: common: Log.cc: Assign LOG_INFO priority to syslog calls Reviewed-by: Loic Dachary commit 52e596ffe0b8a83bfeae0ad836227a73d7d6facb Merge: 65e8bbc cca589f Author: Loic Dachary Date: Fri Oct 14 13:30:05 2016 +0200 Merge pull request #9872 from odivlad/fix-init-el7-jewel jewel: remove SYSTEMD_RUN from initscript Reviewed-by: Orit Wasserman commit 91bd342df6de15263004f3a41c285f6658a16d45 Author: Alexander Graf Date: Mon Sep 26 10:26:30 2016 +0200 AArch64: Detect crc32 extension support from assembler The used compiler may or may not be recent enough to recognize the crc32 extended cpu type. However, it does not really have to know about them either, since all we do is pass inline assembly instructions to the assembler. This patch moves the crc cpu extension detection from compiler based to assembler based, so that we can build optimized code even when the compiler does not know about the cpu type yet. Signed-off-by: Alexander Graf (manual backport of e70ab48b7f6d39a281b3ec65098535a55018b681 - manual backport was undertaken because jewel uses autotools) commit 65e8bbccdb73cc6696736367a3df59ea3dcfda1f Merge: 7c9f1c7 086f6e0 Author: Casey Bodley Date: Thu Oct 13 14:52:30 2016 -0400 Merge pull request #10891 from dachary/wip-16793-jewel jewel: rgw: upgrade from old multisite to new multisite fails Reviewed-by: Casey Bodley commit 7c9f1c7940c77451b2d44ef22a278fdfb2938cea Merge: e3839ab 27626ba Author: Casey Bodley Date: Thu Oct 13 14:51:59 2016 -0400 Merge pull request #10889 from dachary/wip-17143-jewel jewel: rgw: rgw file uses too much CPU in gc/idle thread Reviewed-by: Casey Bodley commit e3839ab5ce63abd248d886dcf362f3a6c49f382e Merge: 0561550 416ec6f Author: Casey Bodley Date: Thu Oct 13 14:51:38 2016 -0400 Merge pull request #10868 from dachary/wip-17064-jewel jewel: rgw: radosgw daemon core when reopen logs Reviewed-by: Casey Bodley commit 05615509d946c340d03af7e407a31e3e53447ac8 Merge: abc8e3d f034fd0 Author: Casey Bodley Date: Thu Oct 13 14:51:17 2016 -0400 Merge pull request #10867 from dachary/wip-17118-jewel jewel: rgw: period commit return error when the current period has a zonegroup which doesn't have a master zone Reviewed-by: Casey Bodley commit abc8e3dd0388778d01a19f3de15a1ffcb662549f Merge: 8adc55d 83a91ba Author: Loic Dachary Date: Thu Oct 13 17:43:52 2016 +0200 Merge pull request #11367 from linuxbox2/jewel-s3-versioning jewel: rgw: S3 object versioning fails when applied on a non-master zone Reviewed-by: Orit Wasserman commit 8adc55d0d3f98acde4d91ba6f8ab9fb379ec10d9 Merge: 50404ae 77ea511 Author: Loic Dachary Date: Thu Oct 13 16:43:04 2016 +0200 Merge pull request #11342 from SUSE/wip-17505-jewel jewel: rgw: doc: description of multipart part entity is wrong Reviewed-by: Loic Dachary commit 50404aef1f36cd2035c4384ecdf608910e6cb215 Merge: f1c55ad de0c4e1 Author: Loic Dachary Date: Thu Oct 13 16:42:45 2016 +0200 Merge pull request #11330 from cbodley/wip-17073 jewel: rgw: RGWDataSyncCR fails on errors from RGWListBucketIndexesCR Reviewed-by: Loic Dachary commit f1c55ad7eeeb633e5ad3414e37ebb5dacccb930b Merge: a336110 e2ce857 Author: Loic Dachary Date: Thu Oct 13 16:39:06 2016 +0200 Merge pull request #11139 from cbodley/wip-radosgw-admin-man-jewel jewel: add orphan options to radosgw-admin --help and man page Reviewed-by: Loic Dachary commit a336110785ea06f4404f136dddcce61660d85b2f Merge: 79885ae cf47628 Author: Loic Dachary Date: Thu Oct 13 16:35:57 2016 +0200 Merge pull request #10832 from ceph/jewel-default-quota jewel: rgw - default quota fixes Reviewed-by: Loic Dachary commit 79885ae86be1e444985446ab8315a21ad6d24634 Merge: 9d1f756 54b6b5e Author: Loic Dachary Date: Thu Oct 13 14:58:33 2016 +0200 Merge pull request #10107 from xiaoxichen/wip-16313-jewel jewel: client: FAILED assert(root_ancestor->qtree == __null) Reviewed-by: John Spray Reviewed-by: Yan, Zheng commit d205b74e07ab364378afb4fc63cb08814a8a1f4e Author: Loic Dachary Date: Tue Aug 23 12:17:00 2016 +0200 tests: populate /dev/disk/by-partuuid for scsi_debug The scsi_debug SCSI devices do not have a symlink in /dev/disk/by-partuuid because they are filtered out by 60-persistent-storage.rules. That was worked around by 60-ceph-partuuid-workaround-rules which has been removed by 9f76b9ff31525eac01f04450d72559ec99927496. Add create rules targetting this specific case, only for tests since the problem does not show in real use cases. Fixes: http://tracker.ceph.com/issues/17100 Signed-off-by: Loic Dachary (cherry picked from commit 7cbf1f0a5eced402e6c7242015550668e3b568f3) commit 86eef272304f3896a9cda5cde7e882d09d0269c0 Author: Daniel Gryniewicz Date: Fri Jun 10 11:33:56 2016 -0400 Allow building RGW with LDAP disabled Signed-off-by: Daniel Gryniewicz (cherry picked from commit 38abfcb5d036c050a363533a2c7b658d528d3612) commit 657554574414dcece6126bfa1b4008b038c3f729 Author: la-sguduru Date: Thu May 12 11:53:25 2016 +0530 doc: Modification for "TEST S3 ACCESS" section Signed-off-by: SirishaGuduru sirishaguduru99@gmail.com (cherry picked from commit 27d6cf0c6958de9f0533bedb90a74c4020b53675) commit 50fd48fbdc8f4f0e9151eb405eb112b710304bb9 Author: qiankunzheng Date: Fri Mar 18 13:58:22 2016 -0400 rgw:response information is error when geting token of swift account The header is X-Trans-Id instead of x-amz-request-id in the response header Fixes:#15195 Signed-off-by: Qiankun Zheng (cherry picked from commit 08e909d0b844e8be3cb82d6add3e87d38e19e80e) commit 742366116d368d3adff4b7941a4751d362073b35 Author: SirishaGuduru Date: Mon Sep 19 10:21:06 2016 +0530 doc: Radosgw multitenancy config paramater change Radosgw multitenancy configuration parameter "rgw keystone make new tenants" never works even applied. When gone through the code, itseems this parameter is not used. But "rgw keystone implicit tenants" works as the code looks for this. Modified the configuration parameter in two files mentioned below from "rgw keystone make new tenants" to "rgw keystone implicit tenants" Fixes: http://tracker.ceph.com/issues/17293 Signed-off-by: SirishaGuduru (cherry picked from commit bd9695d50cfd472508a8a704f1ea5fc62dc08faf) commit 635666412c464228edb014b525abb0af3a55ac50 Author: Yang Honggang Date: Tue Oct 4 09:18:09 2016 +0800 rgw: fix versioned object IO error When accessing a copied destination object, its source object's instance ID information is needed, however it's missing now in the destination object's manifest. In order to fix this problem, we can record source object's version_id/instance into dest object's manifest(a new filed 'tail_instance' is added). When creating a new object(not copy), 'tail_instance' should be equal to its instance value. When copy/get a object, 'tail_instance' should always be used to get the right tail objects. Fixes: http://tracker.ceph.com/issues/17111 Signed-off-by: Yang Honggang (cherry picked from commit d43b69e529328f73da6c29cd746557788a989ae0) commit 40689d1acab4da963d05540eb837fd12a4411269 Author: Yehuda Sadeh Date: Fri Sep 30 16:13:00 2016 -0700 rgw: don't fail if lost race when setting acls Fixes: http://tracker.ceph.com/issues/16930 When concurrently setting acls on object/bucket, we could lose in a race. Instead of retry, just return success (same effect as if we won and then other writer overwrote us). Signed-off-by: Yehuda Sadeh (cherry picked from commit 6e9a915b565923081f609048072b8d75716a74ea) commit 5a53ffa8a9e8590ea0d1992fad9a30cde135cbac Author: Yehuda Sadeh Date: Wed Oct 5 11:41:42 2016 -0700 rgw: remove user email index using helper function Signed-off-by: Yehuda Sadeh (cherry picked from commit 10dbfc540aef5c3d45b7e3aabe61090c302f47ab) commit ad54bf9a0dc541cb6d664edec9ca7f3ec6ffbc93 Author: Weijun Duan Date: Sat Feb 27 04:28:14 2016 -0500 rgw:user email can modify to empty Fixes: http://tracker.ceph.com/issues/13286 Signed-off-by: Weijun Duan (cherry picked from commit ebfd713d9a40e1cb33dfdf2198c20d621e387e36) commit 4babd3fa335351bd4327cded47a1ffe5a8cfd897 Author: Jason Dillaman Date: Sat Sep 17 08:29:15 2016 -0400 librbd: corrected use-after-free in ImageWatcher Fixes: http://tracker.ceph.com/issues/17289 Signed-off-by: Jason Dillaman (cherry picked from commit 2f4d4868e3b721c932c35ae7e8f0dd96b36a37fc) commit 1ca4dc6d1bd6aebece500c6e91f6a9871af0e1f1 Author: Jason Dillaman Date: Fri Sep 9 10:42:50 2016 -0400 librbd: possible deadlock if cluster connection closed after image Fixes: http://tracker.ceph.com/issues/17254 Signed-off-by: Jason Dillaman (cherry picked from commit 818c2f2abc972f689acb7d783c2a684bcc9e6e51) commit b73356b500f9ff364c09e6d78e62fc0e79fbfbac Author: Jason Dillaman Date: Tue Sep 20 07:25:36 2016 -0400 librbd: block name prefix might overflow fixed size C-string The issue which resulted in too large v2 image ids was fixed under #16887. Fixes: http://tracker.ceph.com/issues/17310 Signed-off-by: Jason Dillaman (cherry picked from commit 61734d266c6ee476c2f5fcfbbaefc7d0c7939617) commit b410c8ea2164eb53ee7882859f0a977b67cf8b80 Author: Jason Dillaman Date: Tue Sep 20 10:19:45 2016 -0400 librbd: potential deadlock closing image with in-flight readahead Fixes: http://tracker.ceph.com/issues/17198 Signed-off-by: Jason Dillaman (cherry picked from commit c971d58f8a5550b19374b74bb89d69143423479b) commit 32d84e0eed671f0cfe37057d59f0afe1b63f3c4d Author: Jason Dillaman Date: Tue Sep 20 14:47:36 2016 -0400 librbd: update_features should handle Infernalis OSDs Fixes: http://tracker.ceph.com/issues/17330 Signed-off-by: Jason Dillaman (cherry picked from commit 8cb2ccda1b847e0f33c0d34bf57e9ec29bbbb43b) commit b4e13808b004fd10d342e3daa17a3d4e830d4d00 Author: Jason Dillaman Date: Tue Sep 20 14:34:16 2016 -0400 cls/rbd: set_features should adjust the mask to known features Signed-off-by: Jason Dillaman (cherry picked from commit a26c7a5cacebf8814e20a2f3d5b69c20d8798871) commit dd93e009a222e8e86041661742b4df9c92b097f7 Author: Vikhyat Umrao Date: Mon Jun 27 23:36:09 2016 +0530 rbd: cleanup - Proxied operations shouldn't result in error messages if replayed Fixes: http://tracker.ceph.com/issues/16130 Signed-off-by: Vikhyat Umrao (cherry picked from commit d09c9c471f40f15c14f392a93a04353ca30b1c5e) commit 391936a30cc6022f067cb11c2e39bd47b9e58d61 Author: Lu Shi Date: Wed Jun 15 09:24:43 2016 +0800 librbd: enable/disable of features is not allowed when already enabled/disabled Fixes: http://tracker.ceph.com/issues/16079 Signed-off-by: Lu Shi (cherry picked from commit a8a633396a4105991c9643c2b39391621934c26d) commit 7d6801bdd376f8eb5e1fbfa8f3a9f4de5dc710a1 Author: zhuangzeqiang Date: Sat Jun 25 10:21:25 2016 +0800 rbd: restrict mirror enable/disable actions on parents/clones Fixes: http://tracker.ceph.com/issues/16056 Signed-off-by: zhuangzeqiang zhuang.zeqiang@h3c.com (cherry picked from commit 11dee0bbf0a85f2c197192d0560bd486bc2ad6fc) commit cf476284ae9089e73e5fdf5ddfefdd5842246190 Author: root Date: Sun Jun 19 13:06:06 2016 +0530 rgw: Add documentation for RGW default quota Fixes http://tracker.ceph.com/issues/16447 This was modified to use size in kB, rather than size in bytes, since Jewel has not yet been converted to use size in bytes. Signed-off-by: Pavan Rallabhandi Signed-off-by: Daniel Gryniewicz (cherry picked from commit e3ab3476146cf0fe604783e9e5a0d63237271735) commit 22348154ad51c20d73aa7715695244b8d85fac37 Author: root Date: Sun Jun 19 12:48:42 2016 +0530 rgw: Do not bother to check against conf values during quota handling Fixes http://tracker.ceph.com/issues/16447 This was modified to use size in kB, rather than size in bytes, since Jewel has not yet been converted to use size in bytes. Signed-off-by: Pavan Rallabhandi Signed-off-by: Daniel Gryniewicz (cherry picked from commit 17d2c1712a5b72315a47ab0f8380331bfd478c0b) commit 0b8ecce6ac5abafe0175c941dbf55862b1515b1d Author: root Date: Mon May 23 15:34:58 2016 +0530 rgw: Let the default quota settings take effect during user creation Fixes http://tracker.ceph.com/issues/16447 Signed-off-by: Pavan Rallabhandi Signed-off-by: Daniel Gryniewicz (cherry picked from commit 400d7c982f18efd7cf96acfc3a63078791d1ec0a) commit 1c76ef4e3ea7cb401345af0938d6b76652061d05 Author: Mykola Golub Date: Sat Oct 1 11:21:21 2016 +0300 journal: ensure in-flight ops are complete destroying journaler Fixes: http://tracker.ceph.com/issues/17446 Signed-off-by: Jason Dillaman (cherry picked from commit a25b1d7cbebb7b19cebee4cfb362cd744bbb4607) commit 9bc5e444543fade1f6d78788cadeb7bfdd29710b Author: Mykola Golub Date: Thu Sep 29 16:55:22 2016 +0300 journal: complete action only after notification completed Signed-off-by: Mykola Golub (cherry picked from commit 55762cca2bf586d037cb9f32775ec158dc3287c1) commit 74873322da011d0e6221c90ff39d137de80720ce Author: Jason Dillaman Date: Tue Oct 4 21:29:31 2016 -0400 qa/workunits: reduce amount of client IO for rbd-mirror stress Journal IO is faster now, resulting in OSDs quickly filling up under the stress test case. Signed-off-by: Jason Dillaman (cherry picked from commit a6dd6b50d632aa6995a0f4b6fc5e1a1bda23e2a0) commit d51a2f6689391274bba41f9359acb9395574e629 Author: Jason Dillaman Date: Tue Oct 4 12:22:06 2016 -0400 rbd-mirror: potential race allocating tag during shut down Signed-off-by: Jason Dillaman (cherry picked from commit 8c2ff9bce61a8af024150b8be9dee484f339f7df) commit cea000c6225c809b35b6c643919a01d940522c55 Author: Jason Dillaman Date: Fri Sep 30 12:32:37 2016 -0400 librbd: ignore attempts to request lock if already lock owner Signed-off-by: Jason Dillaman (cherry picked from commit 355c79cb7c955e62e3153cf28046a9b8a6d0d25c) commit 912ce3728c09b2bf613da287f5013b97920cc27c Author: Jason Dillaman Date: Fri Sep 30 12:59:16 2016 -0400 journal: clean up object recorder closed/overflow callback Signed-off-by: Jason Dillaman (cherry picked from commit 24faead086a50ea1b9614268d4dd5f3ea7bbe445) commit 310f3f79cf54daeefa8f00aba0d7b692261e4a33 Author: Jason Dillaman Date: Thu Sep 29 08:36:53 2016 -0400 journal: delay moving overflowed buffers until quiesced Signed-off-by: Jason Dillaman (cherry picked from commit 4483531aa3aa3d0f298778062e2b9a339ad05075) commit 9f3614377043efb56c606905a4b4f5c86b3e074e Author: Jason Dillaman Date: Thu Sep 29 08:19:52 2016 -0400 journal: use reverse iterator search to find flush record Signed-off-by: Jason Dillaman (cherry picked from commit a96065815558e50361af4c701c23e5248962dfe0) commit 21502d90ecc72e70faa19b1202f9c7095341aaec Author: Jason Dillaman Date: Wed Sep 28 08:35:36 2016 -0400 journal: avoid holding lock while sending journal append Signed-off-by: Jason Dillaman (cherry picked from commit dc77a629ed353d586b63f0bd8e20f54a7595afba) commit e0de824f7e728b06025a572a0c8213ef9fb1f112 Author: Jason Dillaman Date: Wed Sep 21 15:41:55 2016 -0400 rbd-mirror: wait for in-flight event commit notifications Fixes: http://tracker.ceph.com/issues/17355 Signed-off-by: Jason Dillaman (cherry picked from commit f810c106ad4e9ae94494801fea6c580d81b9156b) commit efa12f7187b6e037c22ba62563030eb883dab5f9 Author: Jason Dillaman Date: Tue Oct 11 13:55:21 2016 -0400 journal: fix AsyncOpTracker compiler error The master branch has an async JournalTrimmer which shouldn't be backported to the Jewel branch yet. This change addresses the missing sync AsyncOpTracker::wait_for_ops method from the backported class. Signed-off-by: Jason Dillaman commit 2460a3dbe7d04f1a4daa44d7367d24d47fa070fb Author: Jason Dillaman Date: Wed Sep 21 13:02:34 2016 -0400 common: move AsyncOpTracker to common library Signed-off-by: Jason Dillaman (cherry picked from commit 72d8992f054a7e36f92fdd2e01278ce3b9ede2eb) Conflicts: src/journal/CMakeLists.txt: doesn't exist in Jewel commit 1748b38da48beb8e25dc5ca831a540d389a963ee Author: Jason Dillaman Date: Thu Jul 28 16:35:48 2016 -0400 librbd: support deleting image by id instead of name The rbd-mirror daemon will use this API to delete images instead of attempting to use the local image name. Fixes: http://tracker.ceph.com/issues/16227 Signed-off-by: Jason Dillaman (cherry picked from commit 87b32d15914207f61595c1c943817d983faceacd) commit bb5f6b6872e81291176543e954bf3654eb88120f Author: Jason Dillaman Date: Wed Jul 13 14:49:06 2016 -0400 rbd-mirror: use pool id + global image id as deletion primary key Fixes: http://tracker.ceph.com/issues/16538 Signed-off-by: Jason Dillaman (cherry picked from commit 25203a8a9d59ff025d223ec1afaeb14946d54993) commit 96d551d7e8f10f98cc0a6ae5d90d33e6bca968a5 Author: Jason Dillaman Date: Thu Sep 15 18:14:15 2016 -0400 rbd-mirror: concurrent access of event might result in heap corruption Fixes: http://tracker.ceph.com/issues/17283 Signed-off-by: Jason Dillaman (cherry picked from commit ac9ad37ab80db8913f9f1149707bd0009f8235c4) commit 8070d6a560044e5b45af7bd6995345ab30489d84 Author: Ricardo Dias Date: Wed Sep 21 23:08:18 2016 +0100 rbd-mirror: test: Fixed timeout problem in rbd_mirror_stress.sh Signed-off-by: Ricardo Dias (cherry picked from commit 5c737038dd6d44bd76605587486ddf9457bc2d96) commit 6ea4cde78ac7db6081e83d2530552946368021ef Author: Ricardo Dias Date: Tue Sep 6 17:28:22 2016 +0100 rbd: Fix race between journal flush and append events Signed-off-by: Ricardo Dias (cherry picked from commit aa959e71fe5a8cec43de75007fc9cef8de5ee3a5) commit 180a86b61498e6a27a59c7673f74e32ce1ae02cf Author: Ricardo Dias Date: Wed Sep 7 15:26:34 2016 +0100 journal: make librados call async in ObjectRecorder Signed-off-by: Ricardo Dias (cherry picked from commit 7b740f5b4ac1c66ac3c80782d2d34e846d00fddd) commit 5edbfe278532225aaaa6bb3fe6bef84a4c693d83 Author: Ricardo Dias Date: Mon Jul 25 17:00:50 2016 +0100 journal: increase concurrency of journal recorder Signed-off-by: Ricardo Dias (cherry picked from commit 5c88edd68a1ee7c77f11e4113251fbe5768b8d99) commit 9ad132e670564c738e448bf719503cc73525fd8a Author: Jason Dillaman Date: Mon Sep 19 11:24:17 2016 -0400 journal: send update notification when tag allocated Signed-off-by: Jason Dillaman (cherry picked from commit f1cd613e3d8ccb6d05b9adfe1956927991a4f4fe) commit f9aae06152e281c271f50201a8dd1852a132447f Author: Jason Dillaman Date: Tue Sep 13 21:54:46 2016 -0400 librbd: new journal listener event for force promotion Fixes: http://tracker.ceph.com/issues/16974 Signed-off-by: Jason Dillaman (cherry picked from commit fd005490e95d7fca85be4cad34344a58986f64d6) commit 4f5ce86349e246e12abe8e0c55380f400bf05ebb Author: Jason Dillaman Date: Tue Sep 13 16:38:51 2016 -0400 librbd: helper class for quiescing in-flight async ops Signed-off-by: Jason Dillaman (cherry picked from commit 39d9e5cc9b38de2ee9ad2faf8e04253314160811) commit 829ff8c5fa9d9470f1e5370cf601509809b39674 Author: Jason Dillaman Date: Tue Sep 13 12:37:53 2016 -0400 librbd: unify journal event callbacks into single interface Signed-off-by: Jason Dillaman (cherry picked from commit dbbcecf4a289ca36b734b7bda9530cc0a59f84ac) commit 41cf3d93356bde801c32dec4b7a21ae065295aa7 Author: Jason Dillaman Date: Tue Sep 13 16:28:50 2016 -0400 journal: expose ability to retrieve partial tag list Signed-off-by: Jason Dillaman (cherry picked from commit 277b6dd9f13a8390cbf7c9ac7a313813ecad4d27) commit 2f9e6412982826fe0712a0b98e004c405b60fac3 Author: Jason Dillaman Date: Wed Sep 14 08:49:39 2016 -0400 qa/workunits/rbd: fix remove mirrored image race conditions Signed-off-by: Jason Dillaman (cherry picked from commit a0f65b968b350629dfad978c191fc878ca26d093) commit 2ec5e93806f8487bdc5f8df23740aab5738cde1f Author: Jason Dillaman Date: Tue Sep 13 12:32:45 2016 -0400 qa/workunits/rbd: new mirroring forced promotion test Signed-off-by: Jason Dillaman (cherry picked from commit a43268a4a30c06233152d531cbf2550224fb8a15) commit 5d1d898e1132325cae7045dc764a533878d56e00 Author: Jason Dillaman Date: Tue Sep 20 13:31:36 2016 -0400 test/rbd: fix possible mock journal race conditions Fixes: http://tracker.ceph.com/issues/17317 Signed-off-by: Jason Dillaman (cherry picked from commit 471898392372ba4c404376410fb56f3af5287c80) commit 775a999e8bac9d0dc02cb40f0206b96c5864b8d1 Author: Mykola Golub Date: Wed Jul 6 12:59:25 2016 +0300 qa/workunits/rbd: before removing image make sure it is not bootstrapped If an image is being bootstrapped, it implies that the rbd-mirror daemon currently has the image open. The removal API will prevent the removal of any image that is opened by another client. Works-around: http://tracker.ceph.com/issues/16555 Signed-off-by: Mykola Golub (cherry picked from commit 9737a8d6cbaf1b79bbc6008249d39acbae883941) commit 3bbd8ba4282ea7f3fa286d0a1944e9e93d321365 Author: Jason Dillaman Date: Tue Sep 27 14:47:02 2016 -0400 librbd: fix journal create compilation error The synchronous journal create method no longer exists on the master branch and the associated change to introduce an asynchronous journal creation state machine should be delayed to provide more testing time on the master branch before being backported to jewel. Signed-off-by: Jason Dillaman commit 89c6618df89dc0d5d3eb1f855f6f93c72be75939 Author: Jason Dillaman Date: Thu Aug 11 21:11:07 2016 -0400 test: fixed unused function warnings in unittest_rbd_mirror Signed-off-by: Jason Dillaman (cherry picked from commit 3545d9ed19be8d3956f0db901ea9d3bb8b10d13d) commit ca94f25aa960bb352043a9d53eee361071d537da Author: Jason Dillaman Date: Thu Aug 11 20:48:27 2016 -0400 rbd-mirror: prevent syncing to non-primary image after failover Signed-off-by: Jason Dillaman (cherry picked from commit beaef377d69f555277b706afff944a15086da28e) commit 7a434842ac2a2799b611aa87422009c244418922 Author: Jason Dillaman Date: Mon Aug 8 23:45:46 2016 -0400 rbd-mirror: demote/promote in same cluster results in split-brain Fixes: http://tracker.ceph.com/issues/16855 Signed-off-by: Jason Dillaman (cherry picked from commit a6901ca1a065419426b3ad704e27e43ba8d591b8) commit f8f3bbd7246d9a1d5a82f6b0b112f185323a5fb3 Author: Jason Dillaman Date: Thu Aug 11 19:39:21 2016 -0400 librbd: normalize journal promotion/demotion events A non-primary image's commit possition won't accurately reflect the current demotion/promotion chain. Therefore, directly specify the predecessor for promotion events. Signed-off-by: Jason Dillaman (cherry picked from commit ac590e85a472533005cad73c82b2d61db161ab7a) commit 4baea6ace14bd834a3cb2c1d9d5202e94546e264 Author: Jason Dillaman Date: Thu Aug 11 19:09:09 2016 -0400 librbd: block RPC requests when demoting the image Signed-off-by: Jason Dillaman (cherry picked from commit 8b195e1fc8fe70a0e5417934302d5831b1f8dfb3) commit 8ccdad6d9d09e0f09431df6d3ebb2fbed47040c6 Author: Jason Dillaman Date: Tue Aug 9 12:24:19 2016 -0400 librbd: separate journal::TagData predecessor data into new struct Signed-off-by: Jason Dillaman (cherry picked from commit 718befdae711141ef4a1e2e9f5e9aca97f1b5513) commit 13daaffb451e7187d5492a82bb85488c1bfe55d2 Author: Jason Dillaman Date: Mon Aug 8 20:57:45 2016 -0400 rbd-mirror: include tag tid in bootstrap debug log messages Signed-off-by: Jason Dillaman (cherry picked from commit 47d1e625006d554164f020e496a847735240ec95) commit d775680dc4eb53c08a692e746dd65c57a560496d Author: Jason Dillaman Date: Mon Aug 8 10:39:01 2016 -0400 qa/workunits/rbd: demote/promote image on same cluster Signed-off-by: Jason Dillaman (cherry picked from commit e6aa18ea0df6dc4d1add597bc2d972c79699bf4c) commit f7ffbfa71d6f8c417b140ce434b7714cd0053b09 Author: Mykola Golub Date: Wed Aug 10 13:46:46 2016 +0300 rbd-mirror: option to automatically resync after journal client disconnect Signed-off-by: Mykola Golub (cherry picked from commit 77fd6a1c2016262d734b0bb5387e6b6a41232e8b) Conflicts: src/common/config_opts.h: trivial resolution commit 0b402390381dfbfcf4d3810f3179f90c4aa995ef Author: Mykola Golub Date: Wed Jul 27 13:45:32 2016 +0300 rbd-mirror: stop replay when client is disconnected Signed-off-by: Mykola Golub (cherry picked from commit 330dba00ba3153ba2862eef52714e0dceae05192) commit 80aa7e0b4fae6f71a34e28ad1225e3540f9606ed Author: Mykola Golub Date: Wed Aug 10 11:22:16 2016 +0300 rbd-mirror: resync was possible only when image replayer start had succeeded Signed-off-by: Mykola Golub (cherry picked from commit 4bf6912f3c75560b89324fc29286028750f122c1) commit 170476e408308fad622063eddbd282a38a28ed89 Author: Mykola Golub Date: Tue Jul 26 16:10:30 2016 +0300 rbd-mirror: decode_client_meta should return false on error Signed-off-by: Mykola Golub (cherry picked from commit cd5eb36e98f46a1d84bddeafa4e7dcad415aa4a2) commit 7382e1a5a2b801cdfcbf7cda109343c365d005f3 Author: Mykola Golub Date: Wed Aug 3 14:19:51 2016 +0300 rbd: new command to disconnect journal client Signed-off-by: Mykola Golub (cherry picked from commit fc3ba54b3c101498a08a3f34ac8f7eab0152ad7c) commit d3ad2ff9dff192d93d2fbef82ec0ccd37809d2ca Author: Mykola Golub Date: Wed Jul 27 14:06:42 2016 +0300 librbd: optionally flag "laggy" journal clients disconnected Fixes: http://tracker.ceph.com/issues/14738 Signed-off-by: Mykola Golub (cherry picked from commit b8eafefba9f2221a0ce927568795cb1c3ac0fa9c) commit 4056e360117b3aacfba2ae98cd4ecb60e356730c Author: Mykola Golub Date: Wed Jul 13 15:49:40 2016 +0300 journal: allow to trim journal for "laggy" clients Signed-off-by: Mykola Golub (cherry picked from commit 0b8b1aaedc10f7f46e91bf6ad809414feb770c8d) commit 3aec576572066a4329488c0b4420fe863cbbeeb2 Author: Mykola Golub Date: Wed Jul 27 13:42:19 2016 +0300 cls/journal: add async client_update_state method Signed-off-by: Mykola Golub (cherry picked from commit 58b8c66d5bfa60e6dd3ad2ec79360c2eca165c58) commit d66bb7a991affb8b77eb9440d7c8874cc0b3bbf4 Author: Ken Dreyer Date: Mon Oct 3 14:43:25 2016 -0600 build: include more files in "make dist" tarball Include all the documentation, qa, and CMakeLists.txt files in the "make dist" tarball. Kraken and newer releases will already do this. This change is restricted to releases that still use autotools. The purpose of this change is to make it easier to apply downstream patches on an unpacked upstream tarball, because often those patches are cherry-picks of commits that touch files under doc/, qa/, or CMakeLists.txt. Signed-off-by: Ken Dreyer commit 9d1f756dfa1775338c2fa1cf6bfafd45b842b75f Merge: 59f0f4c 57f08fe Author: Loic Dachary Date: Tue Oct 11 12:30:44 2016 +0200 Merge pull request #11389 from jcsp/wip-17466-jewel jewel: mon: don't crash on invalid standby_for_fscid Reviewed-by: John Spray commit 59f0f4c7cc81b20fc584b8e6fff84887ec9c26dd Merge: b747903 c2d4239 Author: Loic Dachary Date: Tue Oct 11 12:30:25 2016 +0200 Merge pull request #10997 from batrick/i17105-backport jewel: multimds: allow_multimds not required when max_mds is set in ceph.conf at startup Reviewed-by: John Spray commit b7479037472c7ce633e2130fb6dbc302dbb3affb Merge: db2e822 cf211d7 Author: Loic Dachary Date: Tue Oct 11 12:30:03 2016 +0200 Merge pull request #10958 from ukernel/jewel-16764 jewel: client: fix shutdown with open inodes Reviewed-by: John Spray commit db2e8224853dc344b71871317204f13a0a53d808 Merge: 4769027 1bc047b Author: Loic Dachary Date: Tue Oct 11 12:29:36 2016 +0200 Merge pull request #10921 from jcsp/wip-client-lock-backport jewel: client: add missing client_lock for get_root Reviewed-by: John Spray commit 4769027f0c83cb22f25186356b56e86b49579b8b Merge: f49bac2 494687b Author: Loic Dachary Date: Tue Oct 11 12:29:05 2016 +0200 Merge pull request #11400 from ceph/wip-rbd-cli-jewel jewel: krbd-related CLI patches Reviewed-by: Loic Dachary commit f49bac2cbbda6aefef676b03f6656a98ec390b87 Merge: 2a348d2 5ab5e82 Author: Loic Dachary Date: Tue Oct 11 11:03:40 2016 +0200 Merge pull request #11171 from ceph/wip-rasize-doc-jewel jewel: doc: fix description for rsize and rasize Reviewed-by: Loic Dachary commit 3bb2a9ed9602b0d280a357f53bc90c6cf83ffe32 Author: Jason Dillaman Date: Mon Oct 10 11:57:03 2016 -0400 librbd: ignore notify errors on missing image header The rename op on v1 images fails since the header no longer exists. In the general case, the removal of the header object will also fail the watcher which has its own recovery path. Fixes: http://tracker.ceph.com/issues/17549 Signed-off-by: Jason Dillaman (cherry picked from commit fe3e2eafa087f81c9ab59f3fbc39600d6adaa9c2) commit 5173563fbfc2799cc2328468bb197d65a94b4d7a Author: Yan, Zheng Date: Fri Sep 2 16:19:29 2016 +0800 client: properly set inode number of created inode in replay request Fixes: http://tracker.ceph.com/issues/17172 Signed-off-by: Yan, Zheng (cherry picked from commit e59385f16afc607ec700397b0bea5229ce69df30) commit 2c4e1c1cd1d01e51d5e4ad500d6b31832591d201 Author: John Spray Date: Tue Sep 6 13:16:04 2016 +0100 mds: log path with CDir damage messages Previously you just got the inode number, which wasn't terribly useful for e.g. a missing fragment object, as you couldn't readily resolve the parent path. Fixes: http://tracker.ceph.com/issues/16973 Signed-off-by: John Spray (cherry picked from commit 439cd5e1b4725e594786785189a37184243828d9) Conflicts: src/mds/CDir.cc: the go_bad() prototype which is part of the context of the patch has changed. commit d52f190d449a2e68baed0659367795ebfb0dcb3d Author: John Spray Date: Tue Sep 6 13:07:38 2016 +0100 mds: s/used_ions/used_inos/ Signed-off-by: John Spray (cherry picked from commit bd3645fb07714d43fb7bd3d66b7e34b33768a8e8) commit 494687b7f11d0c9ca56d96b6b4480de96847bec2 Author: Ilya Dryomov Date: Fri Oct 7 12:32:43 2016 +0200 rbd: expose rbd unmap options Reuse rbd map -o infrastructure to expose rbd unmap options in a similar fashion. Currently it's just one bool option, but we may need more in the future. Signed-off-by: Ilya Dryomov (cherry picked from commit 620f5e1455fdcb05cd3873c1e260141849829e35) Conflicts: doc/man/8/rbd.rst [ PR #9151 ("rbd: add methods to set and get snapshot limits") not in jewel ] commit 6b0226774e2f09f98751e6fa3b0125a960417cc0 Author: Ilya Dryomov Date: Fri Oct 7 11:14:16 2016 +0200 rbd: fix up terminology in help texts Signed-off-by: Ilya Dryomov (cherry picked from commit ba8f9ee791d55c4cf25ad751213ea6ba770b7434) commit 071bd4778547930d9aab9340dcabbb6d9ffdd890 Author: Ilya Dryomov Date: Mon Oct 3 14:09:11 2016 +0200 rbd: recognize lock_on_read option Signed-off-by: Ilya Dryomov (cherry picked from commit e857b7896527b676155d6e01c78567337dc33b1c) commit 7d0714e17557ad281bd92907e10462894a482ce9 Author: Ilya Dryomov Date: Tue Oct 4 10:21:36 2016 +0200 doc: clarify rbd size units It wasn't clear from the man page that --size defaults to M. Signed-off-by: Ilya Dryomov (cherry picked from commit d38dc29ed588b8933da3b66e46e1f0dd4cbb2cf4) Conflicts: doc/man/8/rbd.rst [ PR #9066 ("rbd: introduce a simple bench for read") not in jewel ] commit ca8fc6fc8baae2e4da842e16e25b91d304702cb5 Author: John Spray Date: Sat Jul 23 22:56:45 2016 +0100 ceph-create-keys: fix existing-but-different case We just have to refrain from calling get-or-create if the named key already exists, to avoid potentially having an error when the default creation args don't match the key as it has already been created, such as on certain upgrades. Fixes: http://tracker.ceph.com/issues/16255 Signed-off-by: John Spray (cherry picked from commit 278196d86c52b0be3cb9c17ed7f8f3e3502a217a) commit 0a17741861faa1643bc338459397072268e469bd Author: Yan, Zheng Date: Tue Sep 20 18:07:56 2016 +0800 client: fix readdir vs fragmentation race following sequence of events tigger the race client readdir frag 0* -> got item 'A' MDS merges frag 0* and frag 1* client send readdir request (frag 1*, offset 2, readdir_start 'A') MDS reply items (that are after item 'A') in frag * Fixes: http://tracker.ceph.com/issues/17286 Signed-off-by: Yan, Zheng (cherry picked from commit feb63776d4f2b03ece189715fdd75da5cae2afc2) commit 3a79db10f259d66ca17c2152cf639a63f245505d Author: Yan, Zheng Date: Wed Sep 14 18:22:01 2016 +0800 mds: remove fail-safe queueing replay request MDSRank::queue_one_replay() does not create active request immediately, it just queues corresponding C_MDS_RetryMessage for execution. So the fail-safe code can queue an extra replay request. This can cause replay requests be processed out-of-order Fixes: http://tracker.ceph.com/issues/17271 Signed-off-by: Yan, Zheng (cherry picked from commit d755a4a78242293dd3103bdd0748292df3034563) commit 10e603b1e3024c760cd5567dbbc73fbb3c1281f3 Author: Yan, Zheng Date: Wed Sep 21 21:57:25 2016 +0800 client: fix segment fault in Client::_invalidate_kernel_dcache(). when umounting, root can be NULL Fixes: http://tracker.ceph.com/issues/17253 Signed-off-by: Yan, Zheng (cherry picked from commit ddea9a5b78151c0abe834e4a9872f7cf5f918ebc) commit 3320da0adb346af15a5d50a45e0aea5ca234eb79 Author: John Spray Date: Mon Sep 19 15:18:24 2016 +0100 mds: catch duplicates in DamageTable There was an implicit assumption in the code that callers wouldn't hit the notify_*damaged paths twice because they would have checked is_*_damaged paths first. However, that's not really true in all cases, e.g. scrub code isn't required to respect existing damage entries when trying to load a CDir. Simply fix this by having the DamageTable notify* functions check the key they're inserting doesn't already exist. Fixes: http://tracker.ceph.com/issues/17173 Signed-off-by: John Spray (cherry picked from commit c9cfaef104e9aaefad55583d7e54f8b4665904b3) commit 5d0e2f8690b08418fc404dc43df7a3a16ac83309 Author: Casey Bodley Date: Fri Sep 2 14:13:19 2016 -0400 common: only call crypto::init once per CephContext Fixes: http://tracker.ceph.com/issues/17205 Signed-off-by: Casey Bodley (cherry picked from commit 9dfc1537564c4ce65e4d0230a83f0d8d47503b16) commit 483d8c41a241aa917878284765aef2ba2ced0bb3 Author: Brad Hubbard Date: Tue Jun 14 17:34:44 2016 +1000 cephx: Fix multiple segfaults due to attempts to encrypt or decrypt an empty secret and a null CryptoKeyHandler Fixes: http://tracker.ceph.com/issues/16266 Signed-off-by: Brad Hubbard (cherry picked from commit 009e777fbd18602e5fd66f97bdad95e977e6fecc) commit 5ae4f316b2cf37ce5d13b1ced4b65962c5583882 Author: Kefu Chai Date: Tue Sep 20 17:39:24 2016 +0800 os/filestore/FileJournal: fail out if FileJournal is not block device or regular file otherwise JournalingFileStore will assert when deleting FileJournal which still has the non block/regular file opened. Fixes: http://tracker.ceph.com/issues/17307 Signed-off-by: Kefu Chai (cherry picked from commit 7431eec6fd24cd08ca6c76a9893e3f6e8c63a916) commit 2a348d23e5a047bf6a9c80140cf31ba2d6d4113f Merge: da04ac0 29a8701 Author: Loic Dachary Date: Mon Oct 10 20:30:54 2016 +0200 Merge pull request #10812 from SUSE/wip-17094-jewel jewel: build/ops: ceph-osd-prestart.sh fails confusingly when data directory does not exist Reviewed-by: Loic Dachary commit da04ac055188e2e84d21b29feee3069bb95e5ebe Merge: a80aacf 1a7faefd Author: Matt Benjamin Date: Mon Oct 10 10:46:33 2016 -0400 Merge pull request #11335 from linuxbox2/jewel-nfs jewel: rgw: nfs backports commit a80aacf048c2809bc2b7fd7e2f41e12b5470c5ab Merge: e795450 a10d700 Author: Matt Benjamin Date: Mon Oct 10 10:45:07 2016 -0400 Merge pull request #11332 from linuxbox2/jewel-ldap jewel: rgw: combined LDAP backports commit c2d4239331e397e6a48b1ee7a4c69dad0c951f94 Author: Patrick Donnelly Date: Mon Aug 29 16:25:10 2016 -0400 mds: remove max_mds config option It is now required that all changes to max_mds use the run-time `ceph fs set max_mds` command. The rationale for this change is that it is confusing to have a configuration for max_mds which is only observed at file system creation. Fixes: http://tracker.ceph.com/issues/17105 Signed-off-by: Patrick Donnelly (cherry picked from commit 52f4e545e0ce4782d5f2239d31080d392c12eba2) commit e7954500dab4f3e3bead824ba0da66105940e36a Merge: eac384a 0b93a9a Author: Sage Weil Date: Mon Oct 10 08:54:02 2016 -0500 Merge pull request #10761 from dachary/wip-16377-jewel jewel: msgr/async: Messenger thread long time lock hold risk Reviewed-by: Sage Weil commit eac384a5a158191e3276940d340ee2fe0b4b4b73 Merge: c21e1de d94e5bc Author: Sage Weil Date: Mon Oct 10 08:52:59 2016 -0500 Merge pull request #10278 from SUSE/wip-16667-jewel jewel: incorrect value of CINIT_FLAG_DEFER_DROP_PRIVILEGES Reviewed-by: Sage Weil commit 57f08fe72269b573d95791b44f25de044d40fba6 Author: John Spray Date: Fri Sep 30 21:57:35 2016 +0100 mds: make mds_role_t member order match constructor Previously this was mildy confusing. Signed-off-by: John Spray (cherry picked from commit 63d711d4e9decd2940c55aa697e6a588593da23a) commit c19326cef28ca3764340434cacc1f77e5899b887 Author: John Spray Date: Mon Oct 3 07:33:27 2016 +0100 mon: don't crash on invalid standby_for_fscid Fixes: http://tracker.ceph.com/issues/17466 Signed-off-by: John Spray (cherry picked from commit 92fdcf36bd7d57b97f5d887d0370a181860e388c) commit cbd9a39e83258a7fa98d3d33aea652def17d3906 Author: John Spray Date: Mon Oct 3 07:40:12 2016 +0100 messages: fix MMDSBeacon constructor This was leaving garbage in some fields during decode when decoding messages with an older version. Fixes: http://tracker.ceph.com/issues/17466 Signed-off-by: John Spray (cherry picked from commit f1fb667dfd21cd241f4f45ce9ef3f8393d114c60) commit c21e1de83294fea247bd2a458143639544f91f6d Merge: 8a84877 f7d23d5 Author: Loic Dachary Date: Mon Oct 10 12:26:02 2016 +0200 Merge pull request #10794 from SUSE/wip-17082-jewel jewel: disable LTTng-UST in openSUSE builds Reviewed-by: Loic Dachary commit 8a8487711339596f2f22da572e59a194c6166c69 Merge: 4c526a4 47605a2 Author: Loic Dachary Date: Mon Oct 10 12:21:59 2016 +0200 Merge pull request #9739 from vumrao/wip-vumrao-16337 jewel: osd: add peer_addr in heartbeat_check log message Reviewed-by: Samuel Just commit 4c526a4211f29a5151c5d79d3f0aa221f3d24221 Merge: c362c71 5ffee35 Author: Loic Dachary Date: Mon Oct 10 12:20:30 2016 +0200 Merge pull request #9388 from vumrao/wip-vumrao-16069 jewel: mon: Display full flag in ceph status if full flag is set Reviewed-by: Loic Dachary commit c362c71ef34b136112852cce7a769a8bac30626c Merge: f707d1d 79e2acb Author: Loic Dachary Date: Mon Oct 10 12:03:51 2016 +0200 Merge pull request #10496 from Abhishekvrshny/wip-16583-jewel jewel: mon crash: crush/CrushWrapper.h: 940: FAILED assert(successful_detach) Reviewed-by: Kefu Chai commit f707d1d4f1cae66927e630da9a4524d2af6c9a11 Merge: 84adafe f337a07 Author: Loic Dachary Date: Mon Oct 10 09:31:29 2016 +0200 Merge pull request #11018 from ceph/wip-17223-jewel jewel: krbd qa scripts and concurrent.sh test fix Reviewed-by: Loic Dachary commit cca589fdb18147ea61e79b8d98b886ef30cd65d5 Author: Vladislav Odintsov Date: Mon Jul 4 15:29:22 2016 +0300 init-radosgw: do not use systemd-run in sysvinit `systemd-run` logic in initscripts was introduced because of ticket http://tracker.ceph.com/issues/7627. If we have systemd-based operating system, we should use systemd unit files from systemd directory to start/stop ceph daemons. Otherwise, `daemon()` from `/etc/init.d/functions` on systemd distro starts service in `system.slice` and everything works well. `systemd-run` can not be run on non-systemd distros, so it's not needed on SysV systems. also, ceph-disk is now able to run the "systemctl" to enable and start the ceph-osd, and ceph-deploy is also now playing well with systemd when it comes to ceph-mon and ceph-mds Fixes: http://tracker.ceph.com/issues/16440 Signed-off-by: Vladislav Odintsov (cherry picked from commit 1fd4f92a025e80092b8d08d9b7da2e0b73a52f0f) commit 83a91baff05f03f907fb80702ae1c68194b5308b Author: Yehuda Sadeh Date: Thu Jun 30 17:36:16 2016 -0700 rgw: forward input data when forwarding set_bucket_version to master Fixes: http://tracker.ceph.com/issues/16494 Needed to keep input data around to be forwarded correctly. Also, master does not send any data back, so don't try to parse anything. Signed-off-by: Yehuda Sadeh (cherry picked from commit 1eec00bef1e5540bf3e31d1f8fb1645eb64b1e62) commit 0986aff2aafdf64847a7e9608abb84edaea59eb1 Author: Yehuda Sadeh Date: Thu Jun 30 17:34:38 2016 -0700 rgw: set args when rebuilding req_info Was missing args, so it didn't sign requests with subresources correctly when forwarding. Signed-off-by: Yehuda Sadeh (cherry picked from commit dd1068c54e50671d8904b99189823418a5a5bc07) commit bb81f9dd83da76704a5b6ba3e718fef3ac918290 Author: Jason Dillaman Date: Fri Aug 12 12:39:45 2016 -0400 librbd: permit disabling journaling if in corrupt state Fixes: http://tracker.ceph.com/issues/16740 Signed-off-by: Jason Dillaman (cherry picked from commit 6afb884d755697528684fad54dbb69b15d4386f1) commit b14d7f148c8f64d2e2da533fe2b35661536810c4 Author: Jason Dillaman Date: Fri Aug 12 12:21:02 2016 -0400 librbd: new journal policy to disable initializing the journal This will be used in the case where the journal is being disabled. Signed-off-by: Jason Dillaman (cherry picked from commit 73e4c65c809a1e4161229f49285b21b2cfc623ca) Conflicts: src/librbd/Makefile.am (no CreateRequest.h or RemoveRequest.h in jewel) commit 77ea51138f05b4fa1c603ade0e62f961ec708441 Author: weiqiaomiao Date: Wed Aug 10 14:11:43 2016 +0800 doc/radosgw: fix description of response elements 'Part' Signed-off-by: weiqiaomiao (cherry picked from commit 9c3fe46a7eb89aedc52582737c200e58d63738a5) commit 1a7faefd035aa8f31ab11ddfe1046989d0e69aac Author: Matt Benjamin Date: Wed Sep 21 17:18:20 2016 -0400 rgw_file: pre-assign times Set unix timestamps based on RGW values for creation, modification time for buckets, objects in stat requests. Regard any saved value of these in unix attributes as an overlay. Fixes: http://tracker.ceph.com/issues/17367 Signed-off-by: Matt Benjamin (cherry picked from commit 969750e96380859a350e25015bc8cbf9bdf88512) Fixes: http://tracker.ceph.com/issues/17394 commit 0c932b9078a249bcb9dbb8c425788787f1cd48f2 Author: Matt Benjamin Date: Tue Sep 20 17:32:03 2016 -0400 rgw_file: fix set_attrs operation The effective part of this change is to always set a value for RGW_ATTR_UNIX_KEY1 (because it is expected later). Secondarily, do not pass the address of the to-set attributes buffer::list as remove attrs--this is confusing. Signed-off-by: Matt Benjamin (cherry picked from commit 08afb40b2812270a6af3cd1965e8c03c10392ace) Fixes: http://tracker.ceph.com/issues/17393 commit 0de33909012f2a8305ff2038906669c78070cbcd Author: Matt Benjamin Date: Thu Apr 14 19:18:37 2016 -0400 rgw_file: implement rgw_setattr Introduce a new RGWSetattrs RGWOp descendant, to create or replace sets of attrs on buckets or objects. This version of the change uses the standard RGWRADOS::set_attrs op (we want attribute changes to (e.g.) sync with other changes). Previous versions of this changed incorrectly masked the values of st->st_ino in RGWFileHandle::stat(), now fixed. Signed-off-by: Matt Benjamin (cherry picked from commit 4de1c3c260265f821ebee842d49cb35bf49d8e4e) Fixes: http://tracker.ceph.com/issues/17311 Fixes: http://tracker.ceph.com/issues/17332 commit 00f3fbe7dc940d8c9176a8da35fccd42e6ddc10f Author: zhangweibing Date: Fri Aug 26 10:49:48 2016 +0800 rgw: remove duplicated calls to getattr Signed-off-by: Weibing Zhang (cherry picked from commit 87b550ed2fba472d4ef2e1f771bb5f9fd248804f) Fixes: http://tracker.ceph.com/issues/17327 commit c53992856d6058b71bed55337c3830dbaefb6823 Author: Min Chen Date: Fri Aug 19 01:02:50 2016 -0400 rgw: rgw file fix bug of rgw_lookup can not exact match file name bug reproduce steps: in nfs-client (connect to nfs-ganesha server): 1. mv file file.rename 2. stat file // get file attrs, not return with -ENOENT the reason is that: RGWStatLeafRequest does not exact match the file name, just take file name as a prefix filter Signed-off-by: Min Chen (cherry picked from commit 9d813bafc8e197507457c58ab4f365ccdb7f3589) Fixes: http://tracker.ceph.com/issues/17326 commit 9f5c7377143c56b169a62a96dba5c6ff800e4d5c Author: Yan Jun Date: Tue Aug 23 12:51:10 2016 +0800 rgw: fix the wrong return value Here `parent->readdir` will return negative numbers When there is an error. so we should just return it. Signed-off-by: Yan Jun (cherry picked from commit c979be7810793534715c24c6a9f7c8b0ed31de8d) Fixes: http://tracker.ceph.com/issues/17325 commit cf1353c24917ecb9056272da7f6b3e73bba11fa8 Author: Matt Benjamin Date: Wed Aug 17 10:28:48 2016 -0400 rgw_file: restore local definition of RGWLibFS gc interval Signed-off-by: Matt Benjamin (cherry picked from commit 31936caa09b46e86700faad969adfa9d43176206) Fixes: http://tracker.ceph.com/issues/17323 commit 7a648ed991630541b5875a329820b67b269af570 Author: Matt Benjamin Date: Tue Aug 23 16:50:42 2016 -0400 rgw_file: explain semantics of RGWFileHandle::commit Add a comment explaining why the method currently returns 0 unconditionally. Signed-off-by: Matt Benjamin (cherry picked from commit b1da1354320cc5f69277b225a293a03cc2a5054f) commit 0135da74e2a08fccc2e2da9f1b7e55478bf79157 Author: Matt Benjamin Date: Tue Aug 23 15:57:06 2016 -0400 ceph_timer: prefer using directive to typedef This change replaces the existing member-hook typedef as well as the new set-type typedefs, so committed separately. Signed-off-by: Matt Benjamin (cherry picked from commit f7d4d418658d2091652033045abdb4bfbeb637aa) commit 2a9233f1fbff38c12d378d5b6528a7e576a62e3e Author: Matt Benjamin Date: Tue Aug 23 14:18:43 2016 -0400 librgw: add API version defines for librgw and rgw_file This change borrows the major, minor+extra format used by libcephfs. The version numbering is starting at 1,1,0 on the theory that the implicit version at Jewel is 1,0,0. Signed-off-by: Matt Benjamin (cherry picked from commit 355ccc5ded7f0b459bca24bc8b504b41807c583d) commit 2d6966f7e1f4c6dab12f7500746f1021b52d91af Author: Matt Benjamin Date: Wed Aug 17 11:24:01 2016 -0400 rgw file: allow RGWLibFS::write_completion_interval_s to be set in conf Signed-off-by: Matt Benjamin (cherry picked from commit 70dad0fba561edaf0bb4e746e29b92a4e9b014b2) commit 7071845f2e477648c322f0072e54490c47ed4154 Author: Matt Benjamin Date: Wed Aug 17 11:00:22 2016 -0400 rgw_file: log on write completion events at level 10 Signed-off-by: Matt Benjamin (cherry picked from commit d1e512d8c0e20df5a96ccb5f775a7a2ca123efbe) commit 95310c7f960eead2e22bc02c0e9d94ec863cafd6 Author: Matt Benjamin Date: Wed Aug 17 10:52:58 2016 -0400 rgw_file: add RGW_OPEN_FLAG_STATELESS as a synonym for V3 Signed-off-by: Matt Benjamin (cherry picked from commit da596eb3af69c2cdf191893eb44a8947662634cf) commit 697d4ef7837061bb6b5ebaad07ae86836cdbe126 Author: Matt Benjamin Date: Tue Aug 16 18:17:53 2016 -0400 rgw_file: implement stateless write completion timer Implements a temporal mechanism to enforce write completion for setups which lack open state tracking (e.g., NFS3). Signed-off-by: Matt Benjamin (cherry picked from commit 2c83ed4bd43fcb5d5497151a157b1dc08022fed9) commit 4a31509c5ae7e40dd1538b2075287d8cc202f159 Author: Matt Benjamin Date: Fri Aug 12 16:52:51 2016 -0400 rgw_file: fix/update rgw_create flag arguments Allow passing POSIX open flags as well as api call flags. Needed for NFS3 support. Signed-off-by: Matt Benjamin (cherry picked from commit 2d0ea2b4649e76cbbb87f44c57754d7ea9c46dce) commit 73f9a0b147914f3ef4f7b69a3b57899034406bca Author: Matt Benjamin Date: Fri Aug 12 10:39:21 2016 -0400 rgw_file: fix/update rgw_open flag arguments Allow passing POSIX open flags as well as api call flags. Needed for NFS3 support. Signed-off-by: Matt Benjamin (cherry picked from commit ada29f7f8323be2301588941bcd5c60190b32a4b) commit 4ceb70eba4d777ecc34cce9d1d10002807a6e770 Author: Matt Benjamin Date: Tue Aug 9 16:49:41 2016 -0400 rgw file: remove busy-wait in RGWLibFS::gc() This is a background thread. However, CPU is wasted. Signed-off-by: Matt Benjamin (cherry picked from commit ca33241286f52d849dbde8092507131b8b1108cc) Fixes: http://tracker.ceph.com/issues/17321 Signed-off-by: Matt Benjamin commit 8e1515b068991567136042fb11384fb0fd0620fe Author: Matt Benjamin Date: Tue Aug 9 14:23:00 2016 -0400 rgw_file: unlock() must precede out label In lookup_handle(...). Signed-off-by: Matt Benjamin (cherry picked from commit 9f64f34e481278a683e962b17c7aa811005783a8) commit 88eeaa967494ba6b536d493d06bc4bdbe92e452f Author: Matt Benjamin Date: Sun Aug 7 18:39:33 2016 -0400 rgw nfs: fix write continuations Signed-off-by: Matt Benjamin (cherry picked from commit b4d079f2b5a4d49b2ba7576b272a2a3f3b0d66c8) commit f0b1799910564706766fccb8ec2b1c2c27a10467 Author: Matt Benjamin Date: Fri Aug 5 10:03:33 2016 -0400 rgw nfs: don't leak fh->mtx in lookup_handle() This change fixes a serious latent locking problem, noticed after updating the ganesha/rgw driver invalidation after renames. Signed-off-by: Matt Benjamin (cherry picked from commit d74d46170d7104a6553674f111bbdbe3a116cf54) commit f16ac90f9c097352de2763215f0c83812503e815 Author: Matt Benjamin Date: Mon Aug 8 10:18:35 2016 -0400 rgw file: refuse to rename directories The representation of paths as an aggregate involving any number of objects argues against ever permitting such operations. Signed-off-by: Matt Benjamin (cherry picked from commit 297220fd2a96c0cf34701127b8cf724e24e6865f) commit 23b749f93ed9a7de93aabc871df0adeeebac4e0f Author: Matt Benjamin Date: Wed Aug 3 13:53:15 2016 -0400 rgw_file: refuse ops on deleted targets Detect these illegal cases, as they are indicate various incorrect behaviors/bugs. Signed-off-by: Matt Benjamin (cherry picked from commit 34224ecd8360e03603a1a5dd1c1c9f0a31f1c1d5) commit 345de45cbf5932174222b733bf71820e818fbfd5 Author: Matt Benjamin Date: Thu Apr 14 19:18:37 2016 -0400 rgw_file: implement rgw_setattr Introduce a new RGWSetattrs RGWOp descendant, to create or replace sets of attrs on buckets or objects. This version of the change uses the standard RGWRADOS::set_attrs op (we want attribute changes to (e.g.) sync with other changes). Previous versions of this changed incorrectly masked the values of st->st_ino in RGWFileHandle::stat(), now fixed. Signed-off-by: Matt Benjamin (cherry picked from commit 4de1c3c260265f821ebee842d49cb35bf49d8e4e) commit 88b5027850618f14c562d07ba25a664fd4c22771 Author: Matt Benjamin Date: Wed Jul 13 10:16:59 2016 -0400 rgw_file: refuse partial, out-of-order writes A single file object may be opened only once per gateway instance, and writes to that object must be complete, and in-order. Enforce this. If an invalid write is seen, deletes the current write transaction. Signed-off-by: Matt Benjamin (cherry picked from commit 12aded803e24539266ce9698c678088e2158a82a) commit be243695c1309acfc293f7080cf213edbc558d14 Author: Matt Benjamin Date: Tue Jul 12 17:43:43 2016 -0400 rgw_file: fix rename cases and unify unlink Rather ordinary rename cases failed in the unlink step. Fix this, unifying the unlink path while at it. Signed-off-by: Matt Benjamin (cherry picked from commit 05af1fbb870a905ea8903594dfc607d346c9dd31) commit a10d700697d8f432aca2b1bff6766e2478ac060d Author: Harald Klein Date: Wed Aug 31 17:41:26 2016 +0200 add ldap auth custom search filter feature patch - http://tracker.ceph.com/issues/17185 Signed-off-by: Harald Klein (cherry picked from commit c935885ae9d5fd413495448a0b0e5fce899c9b73) Fixes: http://tracker.ceph.com/issues/17185 commit 7486638563c1eeda7781dcf58b0c536d11f17a0b Author: Matt Benjamin Date: Thu Aug 18 10:54:16 2016 -0400 rgw ldap: protect rgw::from_base64 from non-base64 input Also adds unit tests for: 1. empty output from from_base64 (turns out to be harmless) 2. random and specific non-base64 and sort strings (modified from upstream to avoid alteration of src/test/test_rgw_token.cc) Signed-off-by: Matt Benjamin (cherry picked from commit 0a4c91ec7652d02673a9b156cd16144d778a3844) Fixes: http://tracker.ceph.com/issues/17324 commit 68788fb898118c826b136b7cd7f60265a6dfef79 Author: Matt Benjamin Date: Fri Aug 5 10:02:03 2016 -0400 rgw ldap: enforce simple_bind w/LDAPv3 Found by Harald Klein . Signed-off-by: Matt Benjamin (cherry picked from commit 08d54291435e4d1cb5e02cda3951bc6e8510b0e2) commit 871caeb5011b06d54bd41370ee20ba38e84c9ac0 Author: Matt Benjamin Date: Mon Aug 1 17:36:17 2016 -0400 rgw: add reinit/rebind logic (ldap) Gracefully handle stale LDAP connections by rebinding--verified with MS AD. Rebind is attempted once per request until rebound-- not worse than misconfiguring the environment. Signed-off-by: Matt Benjamin (cherry picked from commit 9b8fccf68789ba5c5986766c544b23aeada0e87e) commit de0c4e1281b5b67af1459aa11fbb292a0cae2ab2 Author: Casey Bodley Date: Tue Aug 16 16:58:51 2016 -0400 rgw: RGWDataSyncCR fails on errors from RGWListBucketIndexesCR Fixes: http://tracker.ceph.com/issues/17073 Signed-off-by: Casey Bodley (cherry picked from commit e77a523f1d74768f4fef58c05cc167705d219036) commit 8833c64459edb77fa0d394b2eda2f79cd0f6dba9 Author: Samuel Just Date: Mon Oct 3 10:34:51 2016 -0700 PG: block writes and scan log for scrub using inclusive upper bound See comment in commit. Signed-off-by: Samuel Just commit 0e2338e5ec5441e5fc173a7af69343c775b02a13 Author: David Zafman Date: Mon Oct 3 14:34:19 2016 -0700 osd_types,PG: force the map used in _scrub to be sorted correctly Previously, ScrubMap::objects was always sorted bitwise (nibblewise before the comparator change was made. It didn't matter because we always scrubbed whole hash values. Now, we need the objects in the objectstore ordering because we may be missing objects at the tail of the scanned range and need them to show up at the tail of the ScrubMap::objects mapping. We don't need to do anything else to handle the upgrade process since the actual objects *in* the map were determined by the objectstore ordering. Signed-off-by: Samuel Just (cherry picked from commit 6d410e97232afdad4f226151333b209d8b50f4ed) commit 16f9d95204f3eca6908ae241e5bacc48b006cb23 Author: Samuel Just Date: Wed Sep 28 13:24:56 2016 -0700 src/osd: relax the requirement that we scrub a whole hash value Previously, we needed to scrub all objects in clones in a single hash value mainly to ensure that _scrub had access to all clones of a single object at the same time. Instead, just avoid letting head or snapdir be a boundary (see the comment in the commit for details). Signed-off-by: Samuel Just (cherry picked from commit 27bdc8ce6d0a7d8ae47f29540f281ba417e16b4c) commit 2176c847d588fbec654653db48c4ad156a87aac2 Author: Samuel Just Date: Wed Sep 28 13:25:42 2016 -0700 hobject: clarify is_snap and has_snapset for max Signed-off-by: Samuel Just (cherry picked from commit a2c27c9423b43b8c5633fb0af95f28b0de94b365) commit 84adafe3fe6f57d49b4bdf1e036e1b686e6b0f55 Merge: 5246f81 fb36dd0 Author: Matt Benjamin Date: Fri Sep 30 14:02:43 2016 -0400 Merge pull request #10656 from dachary/wip-16564-jewel jewel: cors auto memleak commit 35660d1aabd8b27bf3ba71e50d00ba1cf1cecf00 Author: Igor Podoski Date: Wed Sep 14 07:36:43 2016 -0400 mon: OSDMonitor: Missing nearfull flag set Output from 'ceph -s -f json-pretty' wans't showing nearfull flag set. Signed-off-by: Igor Podoski (cherry picked from commit 754887b575795ef96cda095f7f0d8c898e20f00f) commit 5246f8135124dd49466268ead3c8c10d241c94f9 Merge: 9c464a5 3cb0a5e Author: Josh Durgin Date: Thu Sep 29 15:54:05 2016 -0700 Merge pull request #11252 from ceph/wip-offline-split-jewel ceph-objectstore-tool: add a way to split filestore directories offline Reviewed-by: Josh Durgin commit 9c464a55c8ffe7278450c88cc208f2f2e7bce1a0 Merge: 085e9e3 457d78f Author: Yehuda Sadeh Date: Thu Sep 29 11:34:44 2016 -0700 Merge pull request #10866 from dachary/wip-17122-jewel jewel: COPY broke multipart files uploaded under dumpling commit 085e9e369c5c6fbe48f888009e3997d86dcaf968 Merge: cb46489 23d73dc Author: Yehuda Sadeh Date: Thu Sep 29 11:32:09 2016 -0700 Merge pull request #10890 from dachary/wip-17140-jewel jewel: rgw: period commit loses zonegroup changes: region_map converted repeatedly commit cb46489b92b14c5722368e88a4c66712ecc1623d Merge: a0fce63 cefd6f5 Author: Yehuda Sadeh Date: Thu Sep 29 10:51:09 2016 -0700 Merge pull request #11021 from cbodley/wip-16794 jewel: rgw: back off bucket sync on failures, don't store marker commit a0fce6302013b2e9ddad8466d7ab34c370d3786e Merge: 4e61538 92581a3 Author: Yehuda Sadeh Date: Thu Sep 29 10:49:28 2016 -0700 Merge pull request #10657 from dachary/wip-16792-jewel jewel: metadata sync can skip markers for failed/incomplete entries commit 3cb0a5e0b0517701ad3b44bdf6d4b60d6a51e8ee Author: Josh Durgin Date: Fri Aug 5 11:45:00 2016 -0700 ceph-objectstore-tool: add a way to split filestore directories offline Use the usual split code, and split each dir that meets the usual split criteria. This can be run with lower than usual split settings, to avoid costly online splits. To make sure the directories are not merged again, use a load merge threshold (e.g. filestore merge threshold = 1), and adjust the split multiplier accordingly. Fixes: http://tracker.ceph.com/issues/17220 Signed-off-by: Josh Durgin (cherry picked from commit e7b0428e0e8d8f5459311dc698d94a3ac4f04684) Conflicts: src/os/filestore/FileStore.cc src/tools/ceph_objectstore_tool.cc (trivial resolution) commit de672a09f1e2b29de1e0e6e15d26be99315c41b9 Author: Boris Ranto Date: Fri Aug 5 11:14:53 2016 +0200 logrotate: Run as root/ceph Currently, we run the logrotate scripts as ceph/ceph but that way we cannot rotate the scripts created by qemu (they are root/ceph and 644). The original su line was introduced in commit 73d7bed9 because logrotate complained that the directory was writable by a non-root group and it needed a su line to supress that error. This way, we keep the group settings and we can access and rotate the qemu logs as well. Signed-off-by: Boris Ranto commit 4e61538d9f1b372f28e61ad94c368c782be09d92 Merge: ecc2377 e72e7a2 Author: Sage Weil Date: Tue Sep 27 15:37:17 2016 -0500 Merge pull request #10888 from dachary/wip-17062-jewel jewel: tests: fix TestClsRbd.mirror_image failure in upgrade:jewel-x-master-distro-basic-vps commit 9cb45e11ec392da7c22578539b99619bdbf37e35 Author: Brad Hubbard Date: Tue May 10 16:44:44 2016 +1000 log: Log.cc: Assign LOG_INFO priority to syslog calls LOG_DEBUG prio messages are not logged by a default syslog configuration so log at LOG_INFO instead. Fixes: http://tracker.ceph.com/issues/15808 Signed-off-by: Brad Hubbard (cherry picked from commit 3ab5a660a45506d6c4c41bfeb5babdf188e62b3d) commit 165e5abdbf6311974d4001e43982b83d06f9e0cc Author: Vikhyat Umrao Date: Fri Apr 1 16:55:36 2016 +0530 librados: modify Pipe::connect() to return the error code Fixes: http://tracker.ceph.com/issues/15308 Signed-off-by: Vikhyat Umrao (cherry picked from commit cad38dca0c90fb2ee4b589f336f8272122f50ad6) commit 5ab5e8268174bd34a095a1e96a1a969ad9e3531b Author: Andreas Gerstmayr Date: Thu Sep 15 22:02:23 2016 +0200 doc: fix description for rsize and rasize Signed-off-by: Andreas Gerstmayr (cherry picked from commit a004254a02cafdee848a382f893f89c3c5a9845a) commit c94244d3b132b6cf1f8ad309644cf48739282f04 Author: Jelle vd Kooij Date: Thu Sep 1 00:42:34 2016 +0200 Add Install section to systemd rbdmap.service file Signed-off-by: Jelle vd Kooij (cherry picked from commit 57b6f656e17124a5ab4cd1400840d9c7c87a3cc3) commit e2ce8576f547b1e073c169ef331c7063319d9783 Author: tserlin Date: Mon Sep 19 14:40:12 2016 -0400 Add two options to radosgw-admin.rst manpage Add '--job-id' and '--max-concurrent-ios' to Orphan Search Options Fixes: http://tracker.ceph.com/issues/17281 Signed-off-by: Thomas Serlin (cherry picked from commit 697f30d86f49b73c981c06375ab2937570b1db01) commit 4e66f9e9d1179204a6fdf08e43608a852d2b7fd3 Author: Ken Dreyer Date: Thu Sep 15 12:56:39 2016 -0600 radosgw-admin: add "--orphan-stale-secs" to --help The radosgw-admin --help did not include the description of the `--orphan-stale-secs` option of the `orphans find` command. The option sets the number of seconds to wait before declaring an object to be an orphan. Fixes: http://tracker.ceph.com/issues/17280 Signed-off-by: Ken Dreyer (cherry picked from commit 354059ae43b4f4cc797da1669715399cd96a4738) commit bfa90a152867e0277ae05c64b6bdc429b659b9a5 Author: Ken Dreyer Date: Thu Sep 15 11:40:14 2016 -0600 doc: add "--orphan-stale-secs" to radosgw-admin(8) The radosgw-admin(8) manual page did not include the description of the `--orphan-stale-secs` option of the `orphans find` command. The option sets the number of seconds to wait before declaring an object to be an orphan. Fixes: http://tracker.ceph.com/issues/17280 Signed-off-by: Ken Dreyer (cherry picked from commit a676c516069cc448591018ecf4f7d1f7f7bc3bfd) commit cefd6f51068df3ed04c755b1167dd146b85a0741 Author: Yehuda Sadeh Date: Wed Jul 20 12:59:50 2016 -0700 rgw: fix collection of object sync errors Signed-off-by: Yehuda Sadeh (cherry picked from commit aacc31f72cef6f151459e4a0543850edeeaf1938) commit aa369817c9bf10cffa420bfb88a27957407fb625 Author: Yehuda Sadeh Date: Wed Jul 20 12:43:48 2016 -0700 rgw: fix marker tracker completion handling Was not tracking high markers correctly. Could only work if there was a single hole in the completion range. Just keep a map of all the complete entries. Signed-off-by: Yehuda Sadeh (cherry picked from commit f5801c726efdc2f3067a071e6bb5ac83fd0cd147) commit bce19a39b6f6ff23b6373f82ed2e6e12d65a77ef Author: Yehuda Sadeh Date: Wed Jul 20 12:42:52 2016 -0700 rgw: collect() stops if error encoutered and returns true if needed to be called again Signed-off-by: Yehuda Sadeh (cherry picked from commit 5528932e4c327a0fb0ce34a676eb59707a575325) commit f82b59363d12e38446b4f06f554b932a3cbdb1fb Author: Yehuda Sadeh Date: Tue Jul 19 15:32:03 2016 -0700 rgw: back off bucket sync on failures, don't store marker Fixes: http://tracker.ceph.com/issues/16742 If we fail on any single entry in bucket, skip updating the marker tracker so that next time we'll go over that entry, and back off. This will trigger a report to the data sync error repo and eventually a retry on the failing object. Signed-off-by: Yehuda Sadeh (cherry picked from commit 1f3fec807043fd313ef7c66ff48f18b82c8bfa66) commit f337a0720713329b5305ddd9e5c18e8d2b11b475 Author: Ilya Dryomov Date: Fri Jun 3 17:24:48 2016 +0200 krbd: don't segfault if images are unmapped concurrently "rbd map c" can die from a NULL dereference on any of this_pool, this_image or this_snap in wait_for_udev_add(): rbd map c rbd map b rbd unmap a rbd unmap b However unlikely, this segfault is triggered by the rbd/concurrent.sh workunit on a regular basis. Similarly, "rbd showmapped" can die if an image to be listed is unmapped. Signed-off-by: Ilya Dryomov (cherry picked from commit 2bfecb1c0a0b2a314a5d137e6ca2cccd9ddc9b54) commit 2399d4ae54780cb18c313a8da79b697a70a87146 Author: Ilya Dryomov Date: Fri May 13 17:57:12 2016 +0200 qa: rbd/simple_big.sh: drop extraneous sudo Signed-off-by: Ilya Dryomov (cherry picked from commit 6643f4d9723e5d9c5195680ac9abb9915ef42a56) commit 472cb29c5712c4fbbeaef26cdfddc1afd5f2945d Author: Ilya Dryomov Date: Fri May 13 15:05:53 2016 +0200 qa: rbd/concurrent.sh: suppress rbd map output rbd_map_image() echoes the id of the image and is used in command substitution: id=$(rbd_map_image "${image}") Output from rbd map isn't consumed and clobbers the return. Signed-off-by: Ilya Dryomov (cherry picked from commit a7d4189587e343a581660f1a118f6667c26e5bd4) commit 667d42aad7e19e36200ee7cd979efdd19154876a Author: Ilya Dryomov Date: Thu May 12 15:07:12 2016 +0200 qa: rbd: don't modprobe, chown sysfs files or udevadm settle This is a followup to 38a572011dbb ("qa: rbd/map-snapshot-io.sh: don't chown sysfs files") for the rest of the workunits: - /sys/bus/rbd/add_single_major is used if present, so chown'ing /sys/bus/rbd/add doesn't help. Use sudo on rbd map instead. - Don't modprobe or udevadm settle - rbd CLI tool takes care of that. Signed-off-by: Ilya Dryomov (cherry picked from commit 6bddef8ecff97f732bf707d9ba6d2386bc147f5a) commit 866c3e594b754d10bf960375f8498ba573c531dd Author: Ilya Dryomov Date: Wed May 11 17:26:39 2016 +0200 qa: rbd/map-snapshot-io.sh: don't chown sysfs files /sys/bus/rbd/add_single_major is used if present, so chown'ing /sys/bus/rbd/add doesn't help. Use sudo on rbd map instead. Signed-off-by: Ilya Dryomov (cherry picked from commit 38a572011dbbf8298cb7ba8ebf43909967eec689) commit cf211d7dc2eef46cb409b425caa72aac1b24bd7c Author: John Spray Date: Sun Jul 24 17:19:55 2016 +0100 client: fix shutdown with open inodes This piece of code was dereferencing an invalid iterator (invalidated by call to erase()) Fixes: http://tracker.ceph.com/issues/16764 Signed-off-by: John Spray (cherry picked from commit d642b4faec3266f609e4871ccdccdcd73707dc23) commit 1bc047b1c6ccc1a1757c97e52da5e4db5bc8f801 Author: Patrick Donnelly Date: Wed Jun 29 15:45:30 2016 -0400 client: add missing client_lock for get_root A segmentation fault was observed in a recent test failure (test_client_cache_size) [1]. An analysis of the core dump revealed that the client->root inode was NULL. The client log [2] showed that the root inode was deleted prior to the segfault: 2016-06-27 14:24:13.358500 7fe75cff9700 20 client.4161 _ll_put 0x7fe76000ba40 100000003e9 1 -> 0 2016-06-27 14:24:13.358502 7fe75cff9700 10 client.4161 put_inode on 100000003e9.head(faked_ino=0 ref=1 ll_ref=0 cap_refs={} open={} mode=41777 size=0/0 mtime=2016-06-27 14:24:10.700562 caps=pAsLsXsFs(0=pAsLsXsFs) 0x7fe76000ba40) 2016-06-27 14:24:13.358513 7fe75cff9700 15 inode.put on 0x7fe76000ba40 100000003e9.head now 0 2016-06-27 14:24:13.358515 7fe75cff9700 10 client.4161 remove_cap mds.0 on 100000003e9.head(faked_ino=0 ref=0 ll_ref=0 cap_refs={} open={} mode=41777 size=0/0 mtime=2016-06-27 14:24:10.700562 caps=pAsLsXsFs(0=pAsLsXsFs) 0x7fe76000ba40) 2016-06-27 14:24:13.358538 7fe75cff9700 15 client.4161 remove_cap last one, closing snaprealm 0x7fe76000c1b0 2016-06-27 14:24:13.358544 7fe75cff9700 20 client.4161 put_snap_realm 1 0x7fe76000c1b0 1002 -> 1001 2016-06-27 14:24:13.358546 7fe75cff9700 10 client.4161 put_inode deleting 100000003e9.head(faked_ino=0 ref=0 ll_ref=0 cap_refs={} open={} mode=41777 size=0/0 mtime=2016-06-27 14:24:10.700562 caps=- 0x7fe76000ba40) After looking through the reference count inc/dec, I noticed this mismatched inc/dec: 2016-06-27 14:24:13.352622 7fe75dffb700 3 client.4161 ll_forget 100000003e9 1 2016-06-27 14:24:13.352601 7fe7767fc700 1 -- 172.21.15.58:0/3762258427 >> 172.21.15.57:6804/11121 conn(0x7fe789bf3b60 sd=20 :-1 s=STATE_OPEN_MESSAGE_READ_FOOTER_AND_DISPATCH pgs=7 cs=1 l=0). == rx == mds.0 seq 1473 0x7fe764037a80 client_reply(???:470 = 0 (0) Success safe) v1 2016-06-27 14:24:13.352665 7fe75dffb700 20 client.4161 _ll_put 0x7fe76000ba40 100000003e9 1 -> 1 2016-06-27 14:24:13.352687 7fe75d7fa700 3 client.4161 ll_lookup 0x7fe76000ba40 testdir 2016-06-27 14:24:13.352699 7fe75d7fa700 20 client.4161 _lookup have dn testdir mds.-1 ttl 0.000000 seq 0 2016-06-27 14:24:13.352705 7fe75d7fa700 15 inode.get on 0x7fe760011da0 100000003ea.head now 537 2016-06-27 14:24:13.352693 7fe76effd700 1 -- 172.21.15.58:0/3762258427 <== mds.0 172.21.15.57:6804/11121 1473 ==== client_reply(???:470 = 0 (0) Success safe) v1 ==== 27+0+0 (3458149698 0 0) 0x7fe764037a80 con 0x7fe789bf3b60 2016-06-27 14:24:13.352708 7fe75d7fa700 10 client.4161 _lookup 100000003e9.head(faked_ino=0 ref=3 ll_ref=1 cap_refs={} open={} mode=41777 size=0/0 mtime=2016-06-27 14:24:10.700562 caps=pAsLsXsFs(0=pAsLsXsFs) 0x7fe76000ba40) testdir = 100000003ea.head(faked_ino=0 ref=537 ll_ref=999 cap_refs={} open={} mode=40755 size=0/0 mtime=2016-06-27 14:24:13.325297 caps=pAsLsXsFsx(0=pAsLsXsFsx) COMPLETE parents=0x7fe750001f50 0x7fe760011da0) This sequence of ll_forget/ll_lookup is repeated in this test several hundred times. This is the prior sequence (in one thread): 2016-06-27 14:24:13.324896 7fe75dffb700 3 client.4161 ll_forget 100000003e9 1 2016-06-27 14:24:13.324904 7fe75dffb700 20 client.4161 _ll_put 0x7fe76000ba40 100000003e9 1 -> 1 2016-06-27 14:24:13.324915 7fe75dffb700 3 client.4161 ll_lookup 0x7fe76000ba40 testdir 2016-06-27 14:24:13.324921 7fe75dffb700 20 client.4161 _lookup have dn testdir mds.-1 ttl 0.000000 seq 0 2016-06-27 14:24:13.324931 7fe75dffb700 15 inode.get on 0x7fe760011da0 100000003ea.head now 621 2016-06-27 14:24:13.324904 7fe7767fc700 1 -- 172.21.15.58:0/3762258427 >> 172.21.15.57:6804/11121 conn(0x7fe789bf3b60 sd=20 :-1 s=STATE_OPEN_MESSAGE_READ_FOOTER_AND_DISPATCH pgs=7 cs=1 l=0). == rx == mds.0 seq 1387 0x7fe764026720 client_reply(???:385 = 0 (0) Success safe) v1 2016-06-27 14:24:13.324934 7fe75dffb700 10 client.4161 _lookup 100000003e9.head(faked_ino=0 ref=3 ll_ref=2 cap_refs={} open={} mode=41777 size=0/0 mtime=2016-06-27 14:24:10.700562 caps=pAsLsXsFs(0=pAsLsXsFs) 0x7fe76000ba40) testdir = 100000003ea.head(faked_ino=0 ref=621 ll_ref=998 cap_refs={} open={} mode=40755 size=0/0 mtime=2016-06-27 14:24:13.321021 caps=pAsLsXsFsx(0=pAsLsXsFsx) COMPLETE parents=0x7fe750001f50 0x7fe760011da0) The key line is `_lookup 100000003e9.head(...)`. ll_ref=2 is the expected count after this sequence but, as we see at 14:24:13.352708, ll_ref=1. This racing behavior is not serialized by client_lock because Client::get_root() does not lock the client_lock. Additionally, this race is not easy to identify in the logs because Client::get_root() does not use Client::_ll_get to increment the reference, which hides the increments in the logs. Instead it directly increments the reference using Inode:ll_get(). This commit adds the client_lock for Client::get_root() and Client::get_root_ino() (which may not strictly be necessary but can't hurt). [1] http://pulpito.ceph.com/pdonnell-2016-06-27_06:54:23-fs-wip-pdonnell-i16164---basic-smithi/280361/ [2] http://qa-proxy.ceph.com/teuthology/pdonnell-2016-06-27_06:54:23-fs-wip-pdonnell-i16164---basic-smithi/280361/remote/smithi058/log/ceph-client.0.23448.log.gz Signed-off-by: Patrick Donnelly (cherry picked from commit 208e3b6fd58237b309aae15b18dcd82672b133af) commit 086f6e0c07c807dc5001e40d18a72d53c5a10f66 Author: Orit Wasserman Date: Wed Jul 20 13:02:03 2016 +0200 rgw: fix upgrade from old multisite to new multisite configuration We need to store the updated current period after adding the old converted regions Fixes: http://tracker.ceph.com/issues/16751 Signed-off-by: Orit Wasserman (cherry picked from commit 6d7841c6f050f6bd813d54df65d38e7f6934d22d) commit 23d73dca4e8fd9472b8e39d9219fda90323478fb Author: Casey Bodley Date: Tue Aug 23 15:10:44 2016 -0400 rgw: delete region map after upgrade to zonegroup map convert_regionmap() reads the region map and uses it to initialize the zonegroup map. but it doesn't remove the region_map afterwards, so radosgw (and some radosgw-admin commands) will keep doing this on startup, overwriting any changes made to the period/zonegroup map Fixes: http://tracker.ceph.com/issues/17051 Signed-off-by: Casey Bodley (cherry picked from commit 215cd6f6ff7b8cb74df6dc3b94a5928f56ab9540) commit 27626ba9cc7495e312bf5caba6be8d6ee2f1ea64 Author: Matt Benjamin Date: Wed Aug 17 10:28:48 2016 -0400 rgw_file: restore local definition of RGWLibFS gc interval Signed-off-by: Matt Benjamin (cherry picked from commit 31936caa09b46e86700faad969adfa9d43176206) commit 91b65e7b4976103ea45c923bac9d21321d962773 Author: Matt Benjamin Date: Tue Aug 9 16:49:41 2016 -0400 rgw file: remove busy-wait in RGWLibFS::gc() This is a background thread. However, CPU is wasted. Signed-off-by: Matt Benjamin (cherry picked from commit ca33241286f52d849dbde8092507131b8b1108cc) commit e72e7a26f22022e6520232481f66f509367073a1 Author: Jason Dillaman Date: Fri Aug 12 14:44:09 2016 -0400 qa/workunits: support filtering cls_rbd unit test cases Fixes: http://tracker.ceph.com/issues/16529 Signed-off-by: Jason Dillaman (cherry picked from commit 0891ede57cd254aa75c66301fde4a2eb6957a3d6) commit a25a4835ef91f2c249107a3f402c299d8c5c40da Author: Samuel Just Date: Fri Jun 3 14:41:13 2016 -0700 ReplicatedPG: call op_applied for submit_log_entries based repops Signed-off-by: Samuel Just (cherry picked from commit 8bde0ae122f384817f4e522604379d1e9b21d8d9) commit 5ef90856ce26761b31cfd4eb1bea4dfbe88cfb13 Author: Tao Chang Date: Mon May 9 09:50:17 2016 -0400 Cleanup: delete find_best_info again After called find_best_info find a auth_pg_shard, it must be not incomplete, so it will not enter find_best_info again. Signed-off-by: Tao Chang (cherry picked from commit a25cfc4d46c03d8d78e0254c728ea5c29e2246e1) commit dea93dc35859ebf31366dfe5f2b8c2ebebc4f6a6 Author: Boris Ranto Date: Mon Jun 27 11:48:17 2016 +0200 ceph-disk: do not activate device that is not ready If the journal (or data) device is not ready when we are activating the data (or journal) device, just print an info message and exit with 0 so that the ceph-disk systemd service won't fail in this case. Fixes: http://tracker.ceph.com/issues/15990 Signed-off-by: Boris Ranto (cherry picked from commit 73a7a65f65b1f6e90c4cae101bd43fabaeac3045) commit 58b7c52db2df0105aea3fbc36a0b2bc0d299415f Author: Samuel Just Date: Thu Aug 11 08:57:51 2016 -0700 PG: use upset rather than up for _update_calc_stats Fixes: http://tracker.ceph.com/issues/16998 Signed-off-by: Samuel Just (cherry picked from commit 200cae8c9db99b3dede377f74413cc0b15033e1f) commit 4e2b4917c30b431f87b0da0d090664f2210f7ad2 Author: Samuel Just Date: Thu Aug 11 08:57:35 2016 -0700 PG: introduce and maintain upset Signed-off-by: Samuel Just (cherry picked from commit 80a5c21d97116e7a66351a0e1f5d9b561f8663ac) commit c0db9fb831dc67ae138056c2f03c31dcc7f6adbf Author: Jeff Layton Date: Mon Aug 1 09:01:15 2016 -0400 ceph: don't fudge the ctime in stat() unless it's really older than the mtime The current code just looks at the sec field, but the difference may only be evident in the nsec field. Signed-off-by: Jeff Layton (cherry picked from commit 14ee7bcbf0bdd6553dc1b78898ae429aae644100) commit 66cd43bfedddbbc82e3c303554f089f20eabf0f6 Author: Jeff Layton Date: Mon Aug 1 09:01:15 2016 -0400 client: only skip querying the MDS in _lookup when we have the necessary caps If we don't have the caps we'll need later, then we must reissue the lookup to get them regardless of whether we have a lease on the dentry. Fixes: http://tracker.ceph.com/issues/16668 Signed-off-by: Jeff Layton (cherry picked from commit 9e8476743eef195a0a9bc6ef82392a8781e82549) commit fb4a9398a8d60f40ed4c8f57965de54dac1df983 Author: Jeff Layton Date: Mon Aug 1 09:01:14 2016 -0400 client: plumb a mask argument into _lookup ...and attempt to pass in a sane value there, based on what we'll do with the resulting inode. Signed-off-by: Jeff Layton (cherry picked from commit f3605d39e53b3ff777eb64538abfa62a5f98a4f2) commit b5cbd5766fbf2a9ff5ee0e58b4ea42c706f4505a Author: Jeff Layton Date: Mon Aug 1 09:01:14 2016 -0400 client: add mask parameter to _do_lookup We need to allow callers to specify caps to acquire during a lookup, as they may need to scrape certain info out of the inode later. Allow them to pass in a mask. For now, _lookup just passes in 0 for the mask, but verify_reply_trace passes in the regetattr_mask to match what we'd request in the _getattr request if there were a traceless reply. Signed-off-by: Jeff Layton (cherry picked from commit a2ce16f8bfdb16ac485b8c4ad9a51ade5c256a5b) commit 416ec6ff91c602f385385acd8ac458ac58f13222 Author: weiqiaomiao Date: Tue Aug 16 14:19:20 2016 +0800 rgw: fix radosgw daemon core when reopen logs the rgw_user_st_syn thread use the pointer meta_mgr in RGWUserStatsCache::sync_all_users() fuction, so we should close the thread before delete the pointer in RGWRados::finalize() function when close storage before reopen_logs. Fixes: http://tracker.ceph.com/issues/17036 Signed-off-by: weiqiaomiao (cherry picked from commit ea5d7fdd7fa978deadeee42bc0ea33fb11662e59) commit f034fd0a7b959dc3553fb0708da989ba6ac77db3 Author: weiqiaomiao Date: Wed Aug 24 09:38:03 2016 +0800 rgw: fix period update --commit return error period commit return error when the current period has a zonegroup which doesn't have a master zone Fixes: http://tracker.ceph.com/issues/17110 Signed-off-by: weiqiaomiao (cherry picked from commit fcedda6e5dfbaff3975b4474d8546cf77f78cc83) commit 457d78f4f497fdaa0f0f116228ebf135a1ce7fb6 Author: Yehuda Sadeh Date: Thu Jul 7 18:01:54 2016 -0700 rgw: adjust manifest head object adjust the manifest head object: - when reading manifest, set the head object to the object we read the manifest from (and its size). Some manifests are broken and point at a different object - when copying multipart object, set the manifest head object to point at the new head object Signed-off-by: Yehuda Sadeh (cherry picked from commit 419264586fc46337322f0d60b0ffcdcce3bb5c5a) commit 77e497da5c618a30fb908cb3915aa1ca7279ce7f Author: Yehuda Sadeh Date: Thu Jul 7 15:49:07 2016 -0700 rgw: adjust objs when copying obj with explicit_objs set If the head object contains data, need to set it in the list of objs (in addition to being pointed at by head_obj). Signed-off-by: Yehuda Sadeh (cherry picked from commit 1e012112b99004745952accc6fb11ada5e8e0045) commit c1250ee6cbc4a7710399e603ff3ee0e46aa92d49 Author: Yehuda Sadeh Date: Thu Jul 7 15:36:33 2016 -0700 rgw: patch manifest to handle explicit objs copy issue Fixes: http://tracker.ceph.com/issues/16435 Signed-off-by: Yehuda Sadeh (cherry picked from commit b51476466e5bb03bbaa6e428bb6bb189a259b9fd) commit 2311ab24049805ea274451607949ba9b54028470 Author: Yan, Zheng Date: Fri Jul 15 11:32:18 2016 +0800 ObjectCacher: fix last_write check in bh_write_adjacencies() Make the last_write check in bh_write_adjacencies() match corresponding check in flush(). Fixes: http://tracker.ceph.com/issues/16610 Signed-off-by: Yan, Zheng (cherry picked from commit ce166c2cd706b8573deabd331d29544e75e85972) commit c3c2910fbe8c6cc206e1cd70d267503f8ab36e74 Author: Nathan Cutler Date: Sun Aug 7 23:08:48 2016 +0200 rpm: proper packaging of udev directories The issue here is that ceph.spec.in does not package the directories /usr/lib/udev and /usr/lib/udev/rules.d. The problem was not showing because hdparm, which is brought in as a build dependency, packages these directories. However, in SUSE a recent update to hdparm changes that and the problem manifests. This PR addresses the issue by adding udev as a build dependency, which should cover /usr/lib/udev, and by adding explicit "%dir %{_udevrulesdir}" to the packages that put files in this directory. Fixes: http://tracker.ceph.com/issues/16949 Signed-off-by: Nathan Cutler (cherry picked from commit 1a164bc94d661b9d83b4a80e957712ed6f5cd122) commit 7d92e2ea1855a21f9a444c716c590e90dcb9f978 Author: songbaisen Date: Mon Apr 25 10:34:42 2016 +0800 crush: When consider rule size use the correct method. Signed-off-by: song baisen (cherry picked from commit 4cf4791ea1200528ccc0d65b3d4a0fbc234c5df0) commit c2cf21dcce9f481a53c8309800707c80c77754e1 Author: Kefu Chai Date: Fri Aug 5 14:27:43 2016 +0800 mon/PGMonitor: calc the %USED of pool using used/(used+avail) we were using "the raw space used by this pool" / "the raw space of the whole cluster" as the %USED. but it's wrong: - there is chance that not all OSDs are assigned to the pool in question - we can not utilize all the raw space for the pool: there is overhead. and the size of available space for a pool is capped by the assigned OSD with minimal free space. so we should use USED / (USED + AVAIL) as the %USED. so once we have no AVAIL space left, %USED will be 100%. Fixes: http://tracker.ceph.com/issues/16933 Signed-off-by: Kefu Chai (cherry picked from commit 08532ea166dfd97c548d9f1fb478e163021cdda3) commit 109da6fd2fa8a6f8e56a1fe2daa6a738bb2a1fde Author: Kefu Chai Date: Sun Aug 7 00:58:04 2016 +0800 mon/PGMonitor: mark dump_object_stat_sum() as static so we can test this method without create an instance of PGMonitor. Signed-off-by: Kefu Chai (cherry picked from commit af3d3e25823ca9b40ddf09835edb82795ac68f33) commit bd636662a2ff26d9af13955b08f9ab20f29de771 Author: Jason Dillaman Date: Fri Aug 12 09:52:21 2016 -0400 librbd: fix possible inconsistent state when disabling mirroring Fixes: http://tracker.ceph.com/issues/16984 Signed-off-by: Jason Dillaman (cherry picked from commit 7cfedb54ea0cf496cc4b55d08a787abc2d6a4bbe) commit 117aa35094c059dbf5770b01ac13a583471e54aa Author: Kefu Chai Date: Sun Jun 26 01:02:03 2016 +0800 common: instantiate strict_si_cast not strict_si_cast this fixes the build on armf. on 32bit platforms, cstdint is very likely to typedef long long int int64_t; this results in compilation error like `common/strtol.cc:190:75: error: duplicate explicit instantiation of 'T strict_si_cast(const char, std::string) [with T = long long int; std::string = std::basic_string]' [-fpermissive] template int64_t strict_si_cast(const char *str, std::string *err); ^` we can address this by instantiate the primitive type of `long long` instead of `in64_t`. Fixes: http://tracker.ceph.com/issues/16398 Signed-off-by: Kefu Chai (cherry picked from commit 31db4c5f9f725e13e38f3c90744e299e023d02a4) commit d93eda88048d2bcefe4be3ea0aaa6ca0289eabbf Author: Vikhyat Umrao Date: Thu May 26 23:30:25 2016 +0530 common: add int64_t template for strict_si_cast() Signed-off-by: Vikhyat Umrao (cherry picked from commit 8e429d05370fbe7935212d0ae9608e7547f39860) commit f7cd28460147530cfd265a593b32d02adb93abe6 Author: Kefu Chai Date: Sat Apr 30 18:31:37 2016 +0800 common/config: cast OPT_U32 options using uint32_t the OPT_U32 options was translated using strict_si_cast(), and then cast the converted result to uint32_t. this could cause integer underflow. we could have lifted the burden of checking invalid input from the user of this option to the strict_si_cast<>() function. so in this change, we use strict_si_cast() instead, before casting the converted value into `uint32_t`. Signed-off-by: Kefu Chai (cherry picked from commit b7babd6aa671d688eef0af61ca17fd11eec22773) commit 518883d939f34ec0afa03aea1bac35960fb579f2 Author: Loic Dachary Date: Thu Aug 25 09:09:40 2016 +0200 Revert "common: add int64_t template for strict_si_cast()" This reverts commit e3a99c082e3ebd56d5b40d7d94d98e35629df81e. commit 29a87012f560c213e34b893cd8dd9dedfd0b11da Author: Nathan Cutler Date: Mon Aug 22 11:52:45 2016 +0200 ceph-osd-prestart.sh: check existence of OSD data directory Fixes: http://tracker.ceph.com/issues/17091 Signed-off-by: Nathan Cutler (cherry picked from commit d6ffa770784f0efb88827ee7c65ff2e5f9029434) commit 80e25b00554e3955f7bee2e03eba6c55a955999b Author: Jason Dillaman Date: Fri Aug 12 08:26:49 2016 -0400 rbd: bench io-size should not be larger than image size Fixes: http://tracker.ceph.com/issues/16967 Signed-off-by: Jason Dillaman (cherry picked from commit 113775eb583fa28a669c6dc1f8dbf47a8ecd789c) Conflicts: src/tools/rbd/action/BenchWrite.cc: trivial resolution commit f7d23d59535dee004686c5baa79c7b0815ab610a Author: Michel Normand Date: Fri Aug 5 15:19:50 2016 +0200 LTTng-UST disabled for openSUSE LTTng-UST not yet supported in openSUSE so do not enable lltng for it. The (1) is where is defined "is_opensuse" Remove value for test of suse_version in spec file and change related comment from SLES12 to SLE as per comment in https://github.com/ceph/ceph/pull/10592 (1) https://en.opensuse.org/openSUSE:Build_Service_cross_distribution_howto#Detect_a_distribution_flavor_for_special_code Fixes: http://tracker.ceph.com/issues/16937 Signed-off-by: Michel Normand (cherry picked from commit 7da19b6db7eaefb5fe159375cc5a465a722d3897) commit e6ac214f8a6d8773de032c83f5fdd7c65b68d53f Author: Jason Dillaman Date: Thu Jul 28 15:09:53 2016 -0400 librbd: request exclusive lock if current owner cannot execute op The initial krbd implementation will not support executing maintenance ops and instead will return -EOPNOTSUPP. In this case, librbd can take the lock and execute the operation. Fixes: http://tracker.ceph.com/issues/16171 Signed-off-by: Jason Dillaman (cherry picked from commit d2d2d90d64663905c2b81f7809f1d636db6b7fb1) commit 0b93a9a1e7929bc0709ec139fbc6686783f99044 Author: Wei Jin Date: Thu Apr 28 19:00:48 2016 +0800 fast dispatch heartbeat message Signed-off-by: Wei Jin (cherry picked from commit 6f1f717446314618db7a165c75dfd6a76a01f55e) commit 0dcefd2352c06b719960a2df970ac3cbdcce8f52 Author: xie xingguo Date: Mon Jul 18 13:57:08 2016 +0800 mon/osdmonitor: fix incorrect output of "osd df" due to osd out If an osd is automatically marked as out, the output of "osd df" is not right, as follow: -5 10.00999 - 5586G 2989G 2596G 0 0 host ceph192-9-9-8 11 0.90999 1.00000 931G 542G 388G 58.25 0.99 osd.11 14 0.90999 1.00000 931G 530G 400G 56.97 0.97 osd.14 20 0.90999 1.00000 931G 716G 214G 76.99 1.31 osd.20 22 0.90999 1.00000 931G 477G 453G 51.29 0.87 osd.22 26 0.90999 0 0 0 0 0 0 osd.26 28 0.90999 1.00000 931G 587G 343G 63.09 1.07 osd.28 30 0.90999 1.00000 931G 602G 328G 64.75 1.10 osd.30 16 0.90999 1.00000 931G 589G 341G 63.34 1.08 osd.16 18 0.90999 1.00000 931G 530G 400G 56.93 0.97 osd.18 24 0.90999 1.00000 931G 202G 728G 21.77 0.37 osd.24 32 0.90999 1.00000 931G 477G 454G 51.23 0.87 osd.32 Two problems are identified from the above output: 1. the total capacity(total, total used, total avial) only includes osd.32, osd.24, osd.18, osd.16, osd.30, osd.28, and other healthy osds such as osd.11, osd.14 etc. are excluded. 2. the average utilization/deviation are forced resetted. Fixes: http://tracker.ceph.com/issues/16706 Signed-off-by: xie xingguo (cherry picked from commit 1e4735440ca329f4b95d9bd6f58d9efab7d28d20) commit ac27352743a4ac1e4a4bec422d41a0ad8f17e41a Author: xie xingguo Date: Fri Jul 15 16:08:33 2016 +0800 mon/osdmonitor: initialize local variable "kb_avail_i" Signed-off-by: xie xingguo (cherry picked from commit 98f50cebe1d1708adeeb6fc6d0aa2cdc85ac942c) commit 430ab1b83e67dfb697b034e669b06b7a600bcc6b Author: Loic Dachary Date: Tue Jul 12 16:56:52 2016 +0200 ceph-disk: timeout ceph-disk to avoid blocking forever When ceph-disk runs from udev or init script, it is in the background and should it block for any reason, it may keep a lock forever. All calls to ceph-disk in these context are changed to timeout. The TimeoutStartSec= and TimeoutStopSec= which are both set via TimeoutSec= do not apply to Type=oneshot services. https://www.freedesktop.org/software/systemd/man/systemd.service.html Fixes: http://tracker.ceph.com/issues/16580 Signed-off-by: Loic Dachary (cherry picked from commit bed1a5cc05a9880b91fc9ac8d8a959efe3b3d512) commit 0b30a1d210710593678462b287ff33f22de52323 Author: Zengran Zhang Date: Mon Apr 25 22:45:56 2016 -0400 mon/osdmonitor: decouple adjust_heartbeat_grace and min_down_reporters if cancel mon_osd_adjust_heartbeat_grace, we must set mon_osd_min_down_reporters to zero. otherwise the next checking for reporters_by_subtree.size() >= g_conf->mon_osd_min_down_reporters will allways be fail.the two options should take effect respectively... Signed-off-by: Zengran Zhang (cherry picked from commit 7ac5ca5698cfdaab019bf282b537bcc897f9444c) commit 92581a388462039a7f4dc748e8318c4226f068d3 Author: Yehuda Sadeh Date: Wed Jul 20 12:43:48 2016 -0700 rgw: fix marker tracker completion handling Was not tracking high markers correctly. Could only work if there was a single hole in the completion range. Just keep a map of all the complete entries. Signed-off-by: Yehuda Sadeh (cherry picked from commit f5801c726efdc2f3067a071e6bb5ac83fd0cd147) commit fb36dd0c7a72ce5f3affd69d1f77e6222e26de94 Author: Yan Jun Date: Tue Jun 28 16:17:58 2016 +0800 rgw: fix potential memory leaks in RGWPutCORS_ObjStore_S3::get_params We should free `data` before return or jump to `done_err` label. Signed-off-by: Yan Jun (cherry picked from commit 69c6cf551754e51debb4e645716a89e29517f4ee) commit 775c78d4bb31640a006d255ec0cb3818b6a5d23a Author: Jason Dillaman Date: Thu Aug 4 20:40:10 2016 -0400 librbd: prevent creation of v2 image ids that are too large The librbd API is capped at 24 characters for expressing the object prefix for data blocks (including trailing null byte). Fixes: http://tracker.ceph.com/issues/16887 Signed-off-by: Jason Dillaman (cherry picked from commit 4de7c8d0a7222c83268d03b99015c6b9d25f124d) commit 2b272126de6ea6e75b9dcf4fedac5c216093e36d Author: Mykola Golub Date: Mon Jul 4 13:54:32 2016 +0300 librbd: prevent creation of clone from non-primary mirrored image Fixes: http://tracker.ceph.com/issues/16449 Signed-off-by: Mykola Golub (cherry picked from commit ba849e3b04a5c513849d40a7fe4151375265302a) commit 18a66cb375bf6540a3908f843b004c652eacc2dc Author: Jason Dillaman Date: Wed Jun 22 18:09:29 2016 -0400 librbd: journal::Replay no longer holds lock while completing callback Fixes: http://tracker.ceph.com/issues/16433 Signed-off-by: Jason Dillaman (cherry picked from commit 3112a93b49c24f9ae101a7f252c8e708e0d3e260) commit 79e2acb9a585ede5b97d804550b5394f277e7cc4 Author: Kefu Chai Date: Fri Jul 1 20:44:35 2016 +0800 crush: reset bucket->h.items[i] when removing tree item * crush: so we don't see the reference after the removing, this keeps check_item_loc() happy, and move_bucket() use check_item_loc() to see if the removed bucket disappears after the removal. * test: also add unittest_crush_wrapper::CrushWrapper.insert_item Fixes: http://tracker.ceph.com/issues/16525 Signed-off-by: Kefu Chai (cherry picked from commit a7069c7aab6b3f605f3d8f909af96f87977e8698) commit d94e5bc0903a40d8ea353327ddaa7b8567b95f5f Author: Casey Bodley Date: Wed Apr 27 16:53:26 2016 -0400 common: fix value of CINIT_FLAG_DEFER_DROP_PRIVILEGES 0x16 (binary 10110) was overlapping two other flags Signed-off-by: Casey Bodley (cherry picked from commit dfb897ad576e4861a3cb75d85dffbbeaad5e980e) commit 54b6b5e42bca2b7f05b9ff77315de61d3ba14daf Author: Yan, Zheng Date: Wed Jun 8 17:47:58 2016 +0800 client: kill QuotaTree Multiple clients can modify cephfs at the same time. It is very tricky to keep QuotaTree consistant with the global FS hiberarchy. This patch kills the quota tree. After removing the quota tree, we traverse inode's path to find quota root. Fixes: http://tracker.ceph.com/issues/16066 Fixes: http://tracker.ceph.com/issues/16067 Signed-off-by: Yan, Zheng (cherry picked from commit 161954bbfeb13d5588668637d5258221948128ea) commit 47605a2fdde65265e3c4dc60aac206c6ae712be5 Author: Sage Weil Date: Fri May 6 09:46:06 2016 -0400 msg/msg_types: update sockaddr, sockaddr_storage accessors Return sockaddr_storage by value. New sockaddr-based accessors. Signed-off-by: Sage Weil Signed-off-by: Vikhyat Umrao (cherry picked from commit 0038d362d38181bd348881bc0fbe717211eb8d15) commit d5c77c8ff6ad0e4b4242669e4f218f607f433310 Author: Vikhyat Umrao Date: Fri May 20 15:47:23 2016 +0530 osd: add peer_addr in heartbeat_check log message Fixes: http://tracker.ceph.com/issues/16337 Signed-off-by: Vikhyat Umrao (cherry picked from commit f235b9cc1541d33c210e2d56e8061e8908b91ac9) commit 5ffee353cb947bfb79925ca4d62daec248c3ba4a Author: Vikhyat Umrao Date: Fri May 20 23:09:46 2016 +0530 mon : Display full flag in ceph status if full flag is set Fixes: http://tracker.ceph.com/issues/16069 Signed-off-by: Vikhyat Umrao (cherry picked from commit 6b1c894b2b083bf6cead21e9f96d304b2eb7887d)