1 commit 50e863e0f4bc8f4b9e31156de690d765af245185
2 Author: Jenkins Build Slave User <ceph-release-team@redhat.com>
3 Date: Mon Apr 10 11:43:44 2017 +0000
7 commit a64d3e4b33e904aa6585464df8ffff6aafdec10c
9 Author: Ken Dreyer <kdreyer@redhat.com>
10 Date: Fri Mar 31 10:27:07 2017 -0600
12 Merge pull request #14230 from linuxbox2/wip-jewel-expand-argv
14 jewel: rgw_file: expand argv
16 Reviewed-by: Casey Bodley <cbodley@redhat.com>
18 commit c92640a113843a491678c11319d2352f14486f17
19 Merge: 7c35658 fcc3ada
20 Author: Ken Dreyer <kdreyer@redhat.com>
21 Date: Fri Mar 31 10:17:10 2017 -0600
23 Merge pull request #14233 from ktdreyer/wip-19421-jewel
25 jewel: librbd: possible race in ExclusiveLock handle_peer_notification
27 Reviewed-by: Jason Dillaman <dillaman@redhat.com>
29 commit 7c3565899af5f1f474d82d8802d15c6c216ea894
30 Merge: 12467f6 eedb9f7
31 Author: Ken Dreyer <kdreyer@redhat.com>
32 Date: Fri Mar 31 10:16:48 2017 -0600
34 Merge pull request #14231 from linuxbox2/wip-jewel-fhcache
36 jewel: rgw_fh: RGWFileHandle dtor must also cond-unlink from FHCache
38 Reviewed-by: Casey Bodley <cbodley@redhat.com>
40 commit 12467f6c3f346072d81c444ed68eb206402d8c33
41 Merge: 2d2521c 78c8be7
42 Author: Ken Dreyer <kdreyer@redhat.com>
43 Date: Fri Mar 31 10:09:14 2017 -0600
45 Merge pull request #14215 from mdw-at-linuxbox/wip-jewel-rgw-openssl-fix1
49 Reviewed-by: Ken Dreyer <kdreyer@redhat.com>
50 Reviewed-by: Casey Bodley <cbodley@redhat.com>
52 commit 2d2521c0300000421740c611a5133c82d2e4d0f6
53 Merge: 0840cf8 7e4e290
54 Author: Ken Dreyer <kdreyer@redhat.com>
55 Date: Fri Mar 31 10:07:48 2017 -0600
57 Merge pull request #14206 from linuxbox2/wip-gui-jewel
59 jewel: rgw_file: various fixes
61 Reviewed-by: Casey Bodley <cbodley@redhat.com>
63 commit 0840cf86214c793794c6f913b63cc67e0ab3fc94
64 Merge: 0bf6360 b24a8c2
65 Author: Ken Dreyer <kdreyer@redhat.com>
66 Date: Fri Mar 31 10:07:03 2017 -0600
68 Merge pull request #14205 from linuxbox2/wip-dir-orphan-jewel
70 jewel: rgw_file: wip dir orphan
73 src/rgw/rgw_file.cc (whitespace, trivial resolution)
75 Reviewed-by: Casey Bodley <cbodley@redhat.com>
77 commit 0bf636057ba853d1d1a777147014a740d5149b17
78 Merge: 83b5a7f fb85c68
79 Author: Ken Dreyer <kdreyer@redhat.com>
80 Date: Fri Mar 31 10:05:40 2017 -0600
82 Merge pull request #14169 from linuxbox2/wip-18650-jewel
84 jewel: rgw: rgw_file: FHCache residence check should be exhaustive
86 Reviewed-by: Casey Bodley <cbodley@redhat.com>
88 commit 83b5a7f1c589546645e0f377d5d5f7b1f7363a59
89 Merge: 6a175f2 a969c44
90 Author: Ken Dreyer <kdreyer@redhat.com>
91 Date: Fri Mar 31 10:02:58 2017 -0600
93 Merge pull request #13869 from smithfarm/wip-19161-jewel
95 jewel: rgw: rgw_file: fix marker computation
97 Reviewed-by: Casey Bodley <cbodley@redhat.com>
99 commit fb85c68dfbfcfb299cac3a7e6723067f216483ef
100 Author: Matt Benjamin <mbenjamin@redhat.com>
101 Date: Tue Feb 28 15:49:06 2017 -0500
103 rgw_file: use fh_hook::is_linked() to check residence
105 Previously we assumed that !deleted handles were resident--there
106 is an observed case where a !deleted handle is !linked. Since
107 we currently use safe_link mode, an is_linked() check is
108 available, and exhaustive.
110 Fixes: http://tracker.ceph.com/issues/19111
112 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
113 (cherry picked from commit c0aa515f8d8c57ec5ee09e3b60df3cac60453c40)
115 commit fcc3ada733bac74a642ccd0f1ed6301ad7318a30
116 Author: Mykola Golub <mgolub@mirantis.com>
117 Date: Tue Mar 28 15:55:11 2017 +0200
119 jewel: librbd: possible race in ExclusiveLock handle_peer_notification
121 This is a cherry-pick from kraken -- the master diverged after
122 ManagedLock refactoring and is not affected.
124 Fix: http://tracker.ceph.com/issues/19368
125 Signed-off-by: Mykola Golub <mgolub@mirantis.com>
126 (cherry picked from commit df59d6d5f7deb586cf14a6ef6984e6847db08852)
128 commit eedb9f7d2d2a4047b473f23c5a6956c40a3da126
129 Author: Matt Benjamin <mbenjamin@redhat.com>
130 Date: Tue Feb 28 20:24:12 2017 -0500
132 rgw_file: RGWFileHandle dtor must also cond-unlink from FHCache
134 Formerly masked in part by the reclaim() action, direct-delete now
135 substitutes for reclaim() iff its LRU lane is over its high-water
136 mark, and in particular, like reclaim() the destructor is certain
137 to see handles still interned on the FHcache when nfs-ganesha is
138 recycling objects from its own LRU.
140 Fixes: http://tracker.ceph.com/issues/19112
142 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
143 (cherry picked from commit d51a3b1224ba62bb53c6c2c7751fcf7853c35a4b)
145 commit 6c2a40a800277ded302f3183ac6c68b01ca3ed41
146 Author: Matt Benjamin <mbenjamin@redhat.com>
147 Date: Tue Jan 17 11:23:45 2017 -0500
149 rgw_file: split last argv on ws, if provided
151 This is intended to allow an "extra" unparsed argument string
152 containing various cmdline options to be passed as the last argument
153 in the argv array of librgw_create(), which nfs-ganesha is
156 While at it, hook env_args() too.
158 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
159 (cherry picked from commit fbc19e4effc736c98cc1cc283e5c7b131a0fa766)
161 commit d2ca03b391340c6944ba453b106f3e1c45b75f1d
162 Author: Matt Benjamin <mbenjamin@redhat.com>
163 Date: Sun Feb 12 20:18:26 2017 -0500
165 rgw_file: fix hiwat behavior
167 Removed logic to skip reclaim processing conditionally on hiwat,
168 this probably meant to be related to a lowat value, which does
171 Having exercised the hiwat reclaim behavior, noticed that the
172 path which moves unreachable objects to LRU, could and probably
173 should remove them altogether when q.size exceeds hiwat. Now
174 the max unreachable float is lane hiwat, for all lanes.
176 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
177 (cherry picked from commit b8791b2217e9ca87b2d17b51f283fa14bd68b581)
178 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
180 commit 166cb7f85c240eeaffc70968abf5352d9cd45bd9
181 Author: Matt Benjamin <mbenjamin@redhat.com>
182 Date: Sun Feb 12 18:20:43 2017 -0500
184 rgw_file: refcnt bugfixes
186 This change includes 3 related changes:
188 1. add required lock flags for FHCache updates--this is a crash
189 bug under concurrent update/lookup
191 2. omit to inc/dec refcnt on root filehandles in 2 places--the
192 root handle current is not on the lru list, so it's not
195 3. based on observation of LRU behavior during creates/deletes,
196 update (cohort) LRU unref to move objects to LRU when their
197 refcount falls to SENTINEL_REFCNT--this cheaply primes the
198 current reclaim() mechanism, so very significanty improves
199 space use (e.g., after deletes) in the absence of scans
200 (which is common due to nfs-ganesha caching)
202 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
203 (cherry picked from commit beaeff059375b44188160dbde8a81dd4f4f8c6eb)
204 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
206 commit 31a71be9c9f24e22cf7e6eb390d3b39811ee3577
207 Author: Matt Benjamin <mbenjamin@redhat.com>
208 Date: Sat Feb 11 16:38:05 2017 -0500
210 rgw_file: add refcount dout traces at debuglevel 17
212 These are helpful for checking RGWFileHandle refcnt invariants.
214 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
215 (cherry picked from commit 462034e17f919fb783ee33e2c9fa8089f93fd97d)
216 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
218 commit 1d6c72fe9a59561fb8f33889895b6708342b2856
219 Author: Matt Benjamin <mbenjamin@redhat.com>
220 Date: Fri Feb 10 17:14:16 2017 -0500
222 rgw_file: add pretty-print for RGWFileHandle
224 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
225 (cherry picked from commit ef330f385d3407af5f470b5093145f59cc4dcc79)
226 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
228 commit a969c449272ce88ddcbbd509432134f19520c8c3
229 Author: Matt Benjamin <mbenjamin@redhat.com>
230 Date: Mon Feb 20 15:05:18 2017 -0500
232 rgw_file: fix marker computation
234 Fixes: http://tracker.ceph.com/issues/19018
236 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
237 (cherry picked from commit 4454765e7dd08535c50d24205858e18dba4b454c)
239 commit a70e83c90eb42439ba194ca197f507f972a18c3c
240 Author: Matt Benjamin <mbenjamin@redhat.com>
241 Date: Sun Feb 19 20:34:31 2017 -0500
243 rgw_file: rgw_readdir can't list multi-segment dirs
245 This issue has one root cause in librgw, namely that the marker
246 argument to these requests was incorrectly formatted (though the
247 marker cache was working as intended).
249 Secondarily, for nfs-ganesha users, there is a compounding issue
250 that the RGW fsal was required by "temporary" convention to
251 populate the entire dirent cache for a directory on a single
252 readdir() invocation--the cache_inode/mdcache implementations
253 invariantly pass (before future 2.5 changesets, currently in
254 progress) a null pointer for the start cookie offset, intended
257 Fixes: http://tracker.ceph.com/issues/18991
259 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
260 (cherry picked from commit 2cd60ee9712291b906123aca1704288b18a9742b)
262 commit 209987e1c764cac7574bf0bb969e93fcf55b4361
263 Author: Matt Benjamin <mbenjamin@redhat.com>
264 Date: Sun Feb 19 18:21:06 2017 -0500
266 rgw_file: allow setattr on placeholder directories
268 When a POSIX path <bucket>/foo/ is known only as an implicit path
269 segment from other objects (e.g., <bucket>/foo/bar.txt), a case
270 that would usually arise from S3 upload of such an object, an
271 RGWFileHandle object representing "<bucket>/foo/" will be constructed
272 as needed, with no backing in RGW.
274 This is by design, but subsequently, if a setattr is performed on
275 such a handle, we must be ready to create the object inline with
276 storing the attributes.
278 Fixes: http://tracker.ceph.com/issues/18989
280 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
281 (cherry picked from commit 55eec1c0a0e136736961423b7b6244d0f3693c1a)
283 commit 6a175f27961bd4f0fda8f94c200458f17865c9f9
284 Merge: d32ae21 f3face6
285 Author: Ken Dreyer <kdreyer@redhat.com>
286 Date: Tue Mar 28 16:36:35 2017 -0600
288 Merge pull request #14140 from smithfarm/wip-19341-jewel
290 jewel: 'period update' does not remove short_zone_ids of deleted zones
292 Reviewed-by: Casey Bodley <cbodley@redhat.com>
294 commit d32ae21128d664370f7d07ed14427b75e10da7f8
295 Merge: 4dfeed9 0be4e89
296 Author: Ken Dreyer <kdreyer@redhat.com>
297 Date: Tue Mar 28 16:35:16 2017 -0600
299 Merge pull request #13886 from cbodley/wip-17786
301 jewel: rgw: clear data_sync_cr if RGWDataSyncControlCR fails
303 Reviewed-by: Nathan Cutler <ncutler@suse.com>
305 commit 4dfeed949655a5b5041f612b868225c234e86bba
306 Merge: d69c54a 1985662
307 Author: Ken Dreyer <kdreyer@redhat.com>
308 Date: Tue Mar 28 16:34:57 2017 -0600
310 Merge pull request #13867 from smithfarm/wip-19159-jewel
312 jewel: rgw: multisite: RGWMetaSyncShardControlCR gives up on EIO
314 Reviewed-by: Casey Bodley <cbodley@redhat.com>
316 commit d69c54aca9dde3d64b56e5ee72f1cd29057a80c6
317 Merge: 0830135 dfaaec0
318 Author: Ken Dreyer <kdreyer@redhat.com>
319 Date: Tue Mar 28 16:34:24 2017 -0600
321 Merge pull request #13858 from smithfarm/wip-19152-jewel
323 jewel: rgw_file: restore (corrected) fix for dir partial match (return of FLAG_EXACT_MATCH)
325 Reviewed-by: Casey Bodley <cbodley@redhat.com>
327 commit 0830135f40ac37ac9027bece65b8aac64c2c7ec9
328 Merge: 4dd3a9f 97fed01
329 Author: Ken Dreyer <kdreyer@redhat.com>
330 Date: Tue Mar 28 16:33:18 2017 -0600
332 Merge pull request #13848 from smithfarm/wip-19150-jewel
334 jewel: rgw_file: avoid interning .. in FHCache table and don't ref for them
336 Reviewed-by: Casey Bodley <cbodley@redhat.com>
338 commit 4dd3a9f628053784cff2aa5849ab3c4cdb2acf4e
339 Merge: 5eeba9c d333add
340 Author: Ken Dreyer <kdreyer@redhat.com>
341 Date: Tue Mar 28 16:29:30 2017 -0600
343 Merge pull request #13844 from smithfarm/wip-19148-jewel
345 jewel: rgw: DUMPABLE flag is cleared by setuid preventing coredumps
347 Reviewed-by: Casey Bodley <cbodley@redhat.com>
349 commit 5eeba9cb85e9111ef124acd36dccf77c6f00b3e4
350 Merge: ae632b0 f7ce5df
351 Author: Ken Dreyer <kdreyer@redhat.com>
352 Date: Tue Mar 28 16:26:48 2017 -0600
354 Merge pull request #13823 from asheplyakov/19176-bp-jewel
356 jewel: rgw: fix swift cannot disable object versioning with empty X-Versions-Location
358 Reviewed-by: Casey Bodley <cbodley@redhat.com>
360 commit ae632b0da6cc90a993fcae31b56f97c834d15aa1
361 Merge: 874120f e8041f6
362 Author: Ken Dreyer <kdreyer@redhat.com>
363 Date: Tue Mar 28 16:22:23 2017 -0600
365 Merge pull request #13778 from smithfarm/wip-18811-jewel
367 jewel: librgw: RGWLibFS::setattr fails on directories
369 Reviewed-by: Matt Benjamin <mbenjamin@redhat.com>
371 commit 874120f627f509c75b28d1be699812fce881f4bb
372 Merge: b0108ea e475bfa
373 Author: Ken Dreyer <kdreyer@redhat.com>
374 Date: Tue Mar 28 16:21:44 2017 -0600
376 Merge pull request #13717 from asheplyakov/19115-bp-jewel
378 jewel: rgw_file: ensure valid_s3_object_name for directories, too
380 Reviewed-by: Matt Benjamin <mbenjamin@redhat.com>
382 commit b0108ea3bae1662e90c7fb9a911257f45193798c
383 Merge: 6518d70 db928d6
384 Author: Ken Dreyer <kdreyer@redhat.com>
385 Date: Tue Mar 28 16:19:03 2017 -0600
387 Merge pull request #13596 from dillaman/wip-19038-jewel
389 jewel: rbd-mirror: deleting a snapshot during sync can result in read errors
391 Reviewed-by: Mykola Golub <mgolub@mirantis.com>
393 commit 6518d70b89686b9a4e8854a4c6290381fbcdd8c3
394 Merge: db50938 8941881
395 Author: Ken Dreyer <kdreyer@redhat.com>
396 Date: Tue Mar 28 16:10:43 2017 -0600
398 Merge pull request #13583 from asheplyakov/jewel-bp-18901
400 jewel: rgw_file: interned RGWFileHandle objects need parent refs
402 Reviewed-by: Casey Bodley <cbodley@redhat.com>
404 commit db50938013f0b1c1fe8a23247c91bbfc38d36a8c
405 Merge: 7536871 d44263f
406 Author: Ken Dreyer <kdreyer@redhat.com>
407 Date: Tue Mar 28 16:02:43 2017 -0600
409 Merge pull request #13503 from linuxbox2/wip-jewel-10156
411 jewel: rgw: make sending Content-Length in 204 and 304 controllable
413 Reviewed-by: Casey Bodley <cbodley@redhat.com>
414 Reviewed-by: Radoslaw Zarzynski <rzarzynski@mirantis.com>
416 commit 75368712aafc51b062b674b263354cca2dd4b49a
417 Merge: 250071e 2f20328
418 Author: Ken Dreyer <kdreyer@redhat.com>
419 Date: Tue Mar 28 15:53:49 2017 -0600
421 Merge pull request #13232 from ovh/wip-rework-recovery-priorities-jewel
423 jewel: osd: Increase priority for inactive PGs backfill
425 Reviewed-by: Sage Weil <sage@redhat.com>
427 commit 250071eb8dc744237acfed01df2cf04f2d88bc85
428 Merge: 6f57a77 9910eac
429 Author: Ken Dreyer <kdreyer@redhat.com>
430 Date: Tue Mar 28 15:51:46 2017 -0600
432 Merge pull request #13219 from linuxbox2/jewel-rgw-multipart-upload-copy-4
434 jewel: rgw: multipart uploads copy part support
436 Reviewed-by: Casey Bodley <cbodley@redhat.com>
438 commit 6f57a775301edce7a1039a4aa4479bbd8883cbf6
439 Merge: 8c02e54 3fce77a
440 Author: Ken Dreyer <kdreyer@redhat.com>
441 Date: Tue Mar 28 15:51:24 2017 -0600
443 Merge pull request #13108 from vumrao/wip-vumrao-jewel-18657
445 jewel: osd: fix OSD network address in OSD heartbeat_check log message
447 Reviewed-by: Josh Durgin <jdurgin@redhat.com>
449 commit b24a8c2f3890b6fcc60f3b52cba93d573b9d45a2
450 Author: Gui Hecheng <guihecheng@cmss.chinamobile.com>
451 Date: Thu Mar 2 17:21:57 2017 +0800
453 rgw_file: posix style atime,ctime,mtime
455 As an ganesha FSAL backend, rgw_file should properly maintain
456 the atime,ctime,mtime properly against operations such as:
457 (read,write) for file
458 (create,unlink,mkdir,rmdir,rename) for dir
459 (setattr) for file and dir
461 Signed-off-by: Gui Hecheng <guihecheng@cmss.chinamobile.com>
462 (cherry picked from commit ac25da2479b9be876cbdb820560ac46a6e2b17d7)
464 commit b6181833be925b7eb11afffff7f03486bdde2d25
465 Author: Matt Benjamin <mbenjamin@redhat.com>
466 Date: Mon Feb 20 15:05:18 2017 -0500
468 rgw_file: fix marker computation
470 Fixes: http://tracker.ceph.com/issues/19018
472 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
473 (cherry picked from commit 4454765e7dd08535c50d24205858e18dba4b454c)
474 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
476 commit ed9308cba2af227991a50a9535df30cf9bb18a82
477 Author: Matt Benjamin <mbenjamin@redhat.com>
478 Date: Sun Feb 19 20:34:31 2017 -0500
480 rgw_file: rgw_readdir can't list multi-segment dirs
482 This issue has one root cause in librgw, namely that the marker
483 argument to these requests was incorrectly formatted (though the
484 marker cache was working as intended).
486 Secondarily, for nfs-ganesha users, there is a compounding issue
487 that the RGW fsal was required by "temporary" convention to
488 populate the entire dirent cache for a directory on a single
489 readdir() invocation--the cache_inode/mdcache implementations
490 invariantly pass (before future 2.5 changesets, currently in
491 progress) a null pointer for the start cookie offset, intended
494 Fixes: http://tracker.ceph.com/issues/18991
496 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
497 (cherry picked from commit 2cd60ee9712291b906123aca1704288b18a9742b)
498 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
500 commit 26a38e2e1525714978feda018c03698d7af65129
501 Author: Matt Benjamin <mbenjamin@redhat.com>
502 Date: Sun Feb 19 18:21:06 2017 -0500
504 rgw_file: allow setattr on placeholder directories
506 When a POSIX path <bucket>/foo/ is known only as an implicit path
507 segment from other objects (e.g., <bucket>/foo/bar.txt), a case
508 that would usually arise from S3 upload of such an object, an
509 RGWFileHandle object representing "<bucket>/foo/" will be constructed
510 as needed, with no backing in RGW.
512 This is by design, but subsequently, if a setattr is performed on
513 such a handle, we must be ready to create the object inline with
514 storing the attributes.
516 Fixes: http://tracker.ceph.com/issues/18989
518 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
519 (cherry picked from commit 55eec1c0a0e136736961423b7b6244d0f3693c1a)
520 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
522 commit 7e4e29038bd820e171d007360cf383c85f67879b
523 Author: Gui Hecheng <guihecheng@cmss.chinamobile.com>
524 Date: Wed Mar 8 16:23:11 2017 +0800
526 rgw_file: fix reversed return value of getattr
528 When ::getattr returns -ESTALE, rgw_getattr returns ESTALE,
529 which is a not expected postive.
531 Signed-off-by: Gui Hecheng <guihecheng@cmss.chinamobile.com>
532 (cherry picked from commit 39203cf872b8f4af86eb0e4a0f96dffd9cc92b41)
534 commit 3c02ee4fe9ae70f30328a7b015f5cac4a1804c67
535 Author: Gui Hecheng <guihecheng@cmss.chinamobile.com>
536 Date: Mon Mar 20 10:53:46 2017 +0800
538 rgw_file: fix non-negative return code for open operation
540 The nfs-ganesha expects a negative retcode for errors.
542 Signed-off-by: Gui Hecheng <guihecheng@cmss.chinamobile.com>
543 (cherry picked from commit b5f70ef7b066f9ff44770cc8a50ccadaa02f4691)
545 commit e9a4903c6e77caae6f161056e8aaa5ba10ae5ef2
546 Author: Gui Hecheng <guihecheng@cmss.chinamobile.com>
547 Date: Wed Mar 15 15:01:05 2017 +0800
549 rgw_file: fix double unref on rgw_fh for rename
551 Skip unref after unlink to fix the problem.
553 Signed-off-by: Gui Hecheng <guihecheng@cmss.chinamobile.com>
554 (cherry picked from commit bff228734c73b536d2482e2e2fa4ad38b206ebff)
556 commit 9910eac98474930369d694b236c06ffd627fee04
557 Author: Yehuda Sadeh <yehuda@redhat.com>
558 Date: Thu Sep 29 22:20:10 2016 -0700
560 rgw: multipart part copy, fix regression
562 Signed-off-by: Yehuda Sadeh <yehuda@redhat.com>
563 (cherry picked from commit 66fbe9384703c004c01783eb664f55895d310439)
565 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
567 commit a3fdf0e246637ef4145b6b5ba1f4b114ec7ebc62
568 Author: Yehuda Sadeh <yehuda@redhat.com>
569 Date: Thu Sep 29 17:44:08 2016 -0700
571 rgw: minor optimization
573 Signed-off-by: Yehuda Sadeh <yehuda@redhat.com>
574 (cherry picked from commit 4919dc9987c6376d3d4e143702c26417449524c5)
576 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
578 commit 2161376baf413acfbf02df07e404d2918729bfcc
579 Author: Yehuda Sadeh <yehuda@redhat.com>
580 Date: Thu Sep 29 17:43:00 2016 -0700
582 rgw: rgw_obj_key use adjustment in multipart copy part
584 This fixes a case where objects start with double underscore.
586 Signed-off-by: Yehuda Sadeh <yehuda@redhat.com>
587 (cherry picked from commit 29fece3545cc1df404a25eec46706b16f893a5df)
589 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
591 commit 135f84d860d10a2961f430708983113a87ddf899
592 Author: Yehuda Sadeh <yehuda@redhat.com>
593 Date: Thu Sep 29 17:43:35 2016 -0700
595 rgw: multipart copy-part handle versionId
597 Signed-off-by: Yehuda Sadeh <yehuda@redhat.com>
598 (cherry picked from commit 53521efffb1cb92e5f5ce992d4127bf9498d7c33)
600 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
602 commit 9ab65f5501fb59aac17cfdde57371b00af03d84b
603 Author: Yehuda Sadeh <yehuda@redhat.com>
604 Date: Thu Sep 29 14:24:13 2016 -0700
606 rgw: multipart copy part minor parsing cleanup
608 no need for range.size()
610 Signed-off-by: Yehuda Sadeh <yehuda@redhat.com>
611 (cherry picked from commit 6e9b824d5d4017239d58b4752ebc43bfad8f698d)
613 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
615 commit 3eb9dc96ff8e655415500a3595a78cab80739826
616 Author: Yehuda Sadeh <yehuda@redhat.com>
617 Date: Thu Sep 29 14:11:43 2016 -0700
619 rgw: multipart copy, check for empty bucket, improve logging
621 also reduce log level for non critical user errors.
623 Signed-off-by: Yehuda Sadeh <yehuda@redhat.com>
624 (cherry picked from commit 2bcb3d286b230ef917d5ba96c8276a942f544689)
626 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
628 commit 2588e95d9d422d3b33fd710de1f5884873465483
629 Author: Yehuda Sadeh <yehuda@redhat.com>
630 Date: Thu Sep 29 14:07:14 2016 -0700
632 rgw: multipart copy part, chunked read
634 Don't read the entire range from source object, read it in parts.
636 Signed-off-by: Yehuda Sadeh <yehuda@redhat.com>
637 (cherry picked from commit 4049e47a0cfc1eef6efd502590b68ba7234589d3)
639 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
641 commit bd118b556562176ce2390a935b3bb8c25a62d0c4
642 Author: Javier M. Mellid <jmunhoz@igalia.com>
643 Date: Tue Jul 26 14:56:50 2016 +0200
645 rgw: doc: add multipart uploads copy part feature as supported
647 Signed-off-by: Javier M. Mellid <jmunhoz@igalia.com>
648 (cherry picked from commit 8a7ebeee2ff3f10ceb23b7fa43e43c3c450efe22)
650 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
652 commit b56b719299becc38ec36d427a36b8c29f2416c08
653 Author: Javier M. Mellid <jmunhoz@igalia.com>
654 Date: Tue Jul 26 14:53:44 2016 +0200
656 rgw: multipart uploads copy part support
658 Add multipart uploads copy part feature.
660 Fixes: http://tracker.ceph.com/issues/12790
662 Signed-off-by: Javier M. Mellid <jmunhoz@igalia.com>
663 (cherry picked from commit 949480c2e9760855ed6a0501d364d5e766c8207d)
665 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
667 commit d44263fc91da12ea0ad4fec0cd2877b36ddb9e9f
668 Author: Radoslaw Zarzynski <rzarzynski@mirantis.com>
669 Date: Fri Feb 17 00:56:34 2017 +0100
671 rgw: make sending Content-Length in 204 and 304 controllable
673 This commit introduces a new configurable "rgw print prohibited
674 content length" to let operator decide whether RadosGW complies
675 to RFC 7230 (a part of the HTTP specification) or violates it
676 but follows the Swift's behavior.
678 Fixes: http://tracker.ceph.com/issues/16602
679 Signed-off-by: Radoslaw Zarzynski <rzarzynski@mirantis.com>
680 (cherry picked from commit d8e3e64ec97a3c222a56bb6f510e5e23d7858615)
682 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
684 commit f3face61351a61f3b71dfb8268a4b645c4d92da2
685 Author: Casey Bodley <cbodley@redhat.com>
686 Date: Thu Mar 9 15:24:08 2017 -0500
688 rgw: clear old zone short ids on period update
690 the short ids of old, removed zones were being kept in the period to
691 guard against hash collisions with new zones
693 but for a hash collision to cause a wrong object to sync, that object
694 would have to be uploaded simultaneously to two different zones that had
697 to avoid this, we just have to prevent the period from containing two
698 colliding zones at the same time - we don't have to remember old zone
701 Fixes: http://tracker.ceph.com/issues/15618
703 Signed-off-by: Casey Bodley <cbodley@redhat.com>
704 (cherry picked from commit 9c45633c836c966ab1f75ea2b1ad3fa0a4886600)
706 commit 8c02e54a747644d24feb98dbc15a15cdd55d1afd
707 Merge: 9d222b9 952f34f
708 Author: David Zafman <dzafman@redhat.com>
709 Date: Fri Mar 24 08:13:02 2017 -0700
711 Merge pull request #13146 from dzafman/wip-18502-jewel-zafman
713 jewel: osd: Scrub improvements and other fixes
715 Reviewed-by: Kefu Chai <kchai@redhat.com>
717 commit 9d222b943217a2ec022678e50b34766d231004ce
718 Merge: ce2e989 cb9fae5
719 Author: Nathan Cutler <presnypreklad@gmail.com>
720 Date: Thu Mar 23 16:49:13 2017 +0100
722 Merge pull request #14100 from smithfarm/wip-18781
724 Manually merge PR#13143 which was mistakenly merged to jewel-next
726 Reviewed-by: Loic Dachary <ldachary@redhat.com>
727 Reviewed-by: Ken Dreyer <kdreyer@redhat.com>
729 commit cb9fae56ebdfda28b0606ee89631deb60e5f7de7
730 Merge: ce2e989 ee3c67c
731 Author: Nathan Cutler <ncutler@suse.com>
732 Date: Thu Mar 23 09:46:53 2017 +0100
734 Merge branch 'jewel-next' into jewel
736 Branch "jewel-next" was merged to jewel on Dec 22 (5b402f8a7b), and then PR
737 https://github.com/ceph/ceph/pull/13143 merged to jewel-next much later (Feb
738 1st), and then jewel-next was never merged to jewel again before we tagged
741 This merge brings in a single commit, i.e. the one from
742 https://github.com/ceph/ceph/pull/13143
744 Reported-by: Ken Dreyer <kdreyer@redhat.com>
745 Signed-off-by: Nathan Cutler <ncutler@suse.com>
747 commit ce2e9897c08775cfbe318f637d61e07eb5433df5
748 Merge: 656b5b6 d0a0d2f
749 Author: Zack Cerza <zack@cerza.org>
750 Date: Tue Mar 14 09:19:28 2017 -0600
752 Merge pull request #13952 from smithfarm/wip-fix-merge-error
754 tests: fix merge error in rgw/singleton/all/radosgw-admin.yaml
756 commit 0be4e89419f2083d081ac784891e4653290cd530
757 Author: Casey Bodley <cbodley@redhat.com>
758 Date: Tue Oct 11 15:21:42 2016 -0400
760 rgw: hold a reference on data_sync_cr over run()
762 run() will drop its reference to data_sync_cr, so we need to hold a
763 reference until we can reacquire the lock
765 Signed-off-by: Casey Bodley <cbodley@redhat.com>
766 (cherry picked from commit 4cf0d2a768e7402e71280ca16b47353ca2a68505)
768 commit 18ffdb7baf4aff1fd7f894af1054081f3ee61d28
769 Author: Casey Bodley <cbodley@redhat.com>
770 Date: Tue Oct 11 15:19:37 2016 -0400
772 rgw: clear data_sync_cr if RGWDataSyncControlCR fails
774 async notifications will still try to call wakeup() on RGWDataSyncControlCR
775 if it fails, leading to segfault
777 Fixes: http://tracker.ceph.com/issues/17569
779 Signed-off-by: Casey Bodley <cbodley@redhat.com>
780 (cherry picked from commit 5cc599b9bf2dde31de16a5b2831baf06851d69d5)
783 src/rgw/rgw_data_sync.cc: declaration of 'int r'
785 commit d0a0d2f4bf1dacf7c1f38b968be84a5e905554a8
786 Author: Nathan Cutler <ncutler@suse.com>
787 Date: Tue Mar 14 01:26:23 2017 +0100
789 tests: fix merge error in rgw/singleton/all/radosgw-admin.yaml
791 Introduced by 94d5888c
793 Reported-by: Robin H. Johnson <robin.johnson@dreamhost.com>
794 Signed-off-by: Nathan Cutler <ncutler@suse.com>
796 commit 19856624cd5ea8cb82bc5b46e062beb39674bd34
797 Author: Casey Bodley <cbodley@redhat.com>
798 Date: Mon Feb 20 16:00:01 2017 -0500
800 rgw: RGWMetaSyncShardControlCR retries with backoff on all error codes
802 RGWBackoffControlCR only treats EBUSY and EAGAIN as 'temporary' error
803 codes, with all other errors being fatal when exit_on_error is set
805 to RGWMetaSyncShardControlCR, a 'fatal' error means that no further sync
806 is possible on that shard until the gateway restarts
808 this changes RGWMetaSyncShardControlCR to set exit_on_error to false, so
809 that it will continue to retry with backoff no matter what error code it
812 Fixes: http://tracker.ceph.com/issues/19019
814 Signed-off-by: Casey Bodley <cbodley@redhat.com>
815 (cherry picked from commit 3e4059557fd6cad5d31014327f60832b36d04a6c)
817 commit dfaaec0446c3db458bffe1d725518ab4f7cc2fa8
818 Author: Matt Benjamin <mbenjamin@redhat.com>
819 Date: Thu Feb 23 10:21:38 2017 -0500
821 rgw_file: return of RGWFileHandle::FLAG_EXACT_MATCH
823 Allow callers of rgw_lookup() on objects attested in an
824 rgw_readdir() callback the ability to bypass exact match in
825 RGWLibFS::stat_leaf() case 2, but restore exact match enforcement
828 This preserves required common_prefix namespace behavior, but
829 prevents clients from eerily permitting things like "cd sara0" via
830 partial name match on "sara01."
832 Fixes: http://tracker.ceph.com/issues/19059
834 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
835 (cherry picked from commit 70ef7d45e0abf2661bd4e23161d4e70cf5178079)
837 commit 9b3784d924112d9ba42b2088d5fb2656ef74fadc
838 Author: Matt Benjamin <mbenjamin@redhat.com>
839 Date: Sun Feb 19 17:43:17 2017 -0500
841 rgw_file: invalid use of RGWFileHandle::FLAG_EXACT_MATCH
843 The change which introduced this flag also caused it to be
844 given as the flags argument to RGWLibFS::stat_leaf() when called
847 This was incorrect: in particular, when a directory is known only
848 as a common prefix of other objects, the AWS namespace mapping
849 convention requires lookup("foo") to match a non-materialized
850 instance of "foo/" (case 2 in RGWLibFS::stat_leaf's stat loop).
852 Fixes: http://tracker.ceph.com/issues/18992
854 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
855 (cherry picked from commit e31e9eb980f958640150e8d7f17de1b9e5478b1e)
857 commit 7761376c92eb38a2dd3d19f0b3d81895a4b1167c
858 Author: Matt Benjamin <mbenjamin@redhat.com>
859 Date: Wed Feb 22 14:57:59 2017 -0500
861 rgw_file: rgw_lookup: don't ref for "/" or ".."
863 These refs won't be returned by nfs-ganesha, and are sufficiently
864 magical that other consumers should be persuaded to understand
867 Fixes: http://tracker.ceph.com/issues/19060
869 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
870 (cherry picked from commit dea8d1ee373399a21851690a9753388b659b8ede)
872 commit 97fed013f66a89c404bac5145080e5d556ff5c42
873 Author: Matt Benjamin <mbenjamin@redhat.com>
874 Date: Wed Feb 22 10:24:29 2017 -0500
876 rgw_file: avoid stranding invalid-name bucket handles in fhcache
878 To avoid a string copy in the common mkdir path, handles for
879 proposed buckets currently are staged in the handle table, before
880 being rejected. They need to be destaged, not just marked deleted
881 (because deleted objects are now assumed not to be linked, as of
882 beaeff059375b44188160dbde8a81dd4f4f8c6eb).
884 This triggered an unhandled Boost assert when deleting staged
885 handles, as current safe_link mode requires first removing from
888 Fixes: http://tracker.ceph.com/issues/19036
890 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
891 (cherry picked from commit 6cde812c92e5bba9f85fbf8486ebe69b55952370)
893 commit d333addb71a0a92fbac5a7b922dbd69fc9e0604e
894 Author: Brad Hubbard <bhubbard@redhat.com>
895 Date: Mon Feb 27 13:06:59 2017 +1000
897 rgw: set dumpable flag after setuid post ff0e521
899 ff0e521 resolved the issue for the other daemons but not for rgw since
900 it calls setuid (via civetweb) after the new code sets PR_SET_DUMPABLE.
901 Add another prctl call before wait_shutdown.
903 Fixes: http://tracker.ceph.com/issues/19089
905 Signed-off-by: Brad Hubbard <bhubbard@redhat.com>
906 (cherry picked from commit bc458d39630b599e0e1ca9fe25ad7455fcffdd10)
908 commit f7ce5df4064e5538156cb44f9525d7552a0dd098
909 Author: Jing Wenjun <jingwenjun@cmss.chinamobile.com>
910 Date: Wed Feb 8 15:07:43 2017 +0800
912 rgw: fix swift cannot disable object versioning with empty X-Versions-Location
914 we should be able to disable object verioning by removing its X-Versions-Location
915 metadata header by sending an empty key value. this description can be found at
916 No.8 in http://docs.openstack.org/user-guide/cli-swift-set-object-versions.html.
918 Fixes: http://tracker.ceph.com/issues/18852
919 Signed-off-by: Jing Wenjun <jingwenjun@cmss.chinamobile.com>
920 (cherry picked from commit 17c5a0edd2227703cec867f0f588d4eae36dfe1c)
922 commit 2f2032814189a4ecbf8dc01b59bebfae8ab3f524
923 Author: Bartłomiej Święcki <bartlomiej.swiecki@corp.ovh.com>
924 Date: Fri Dec 2 16:54:46 2016 +0100
926 osd: Increase priority for inactive PGs backfill
928 This change does prioritize backfill of PGs which don't
929 have min_size active copies. Such PGs would cause IO stalls
930 for clients and would increase throttlers usage.
932 This change also fixes few subtlle out-of-bounds bugs.
934 Signed-off-by: Bartłomiej Święcki <bartlomiej.swiecki@corp.ovh.com>
935 (cherry picked from commit 6a76adcdb1f92c136841d960aa7cd4e5b94addec)
938 PendingReleaseNotes (removed version number, merged conflicts)
940 commit e8041f627c6a05a59dba29819ed610ea0896f4dd
941 Author: Matt Benjamin <mbenjamin@redhat.com>
942 Date: Fri Feb 3 13:44:45 2017 -0500
944 rgw_file: fix RGWLibFS::setattr for directory objects
946 Fixes: http://tracker.ceph.com/issues/18808
948 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
949 (cherry picked from commit 4ad5a9226852d6d564baf2e63278ed6c4c185ecb)
951 commit e475bfaf7d3a1b0e54172083a92546560219665a
952 Author: Matt Benjamin <mbenjamin@redhat.com>
953 Date: Thu Feb 23 16:02:07 2017 -0500
955 rgw_file: ensure valid_s3_object_name for directories, too
957 The logic in RGWLibFS::mkdir() validated bucket names, but not
958 object names (though RGWLibFS::create() did so).
960 The negative side effect of this was not creating illegal objects
961 (we won't), but in a) failing with -EIO and b) more importantly,
962 not removing up the proposed object from FHCache, which produced a
963 boost assert when recycled.
965 Fixes: http://tracker.ceph.com/issues/19066
967 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
968 (cherry picked from commit eb1cd3b30c0504385f05bf2d2dd5e2251b7efed7)
970 commit 78c8be7a0df3d1c669f8a2a8fd7a5676d0823209
971 Author: Marcus Watts <mdw@linuxbox.com>
972 Date: Thu Feb 23 02:30:52 2017 -0500
974 rgw/openssl fix: xenial autoconf logic problem: gcc/ld got too smart...
976 On xenial, cc -o foo -lssl -lcrypto doesn't always
977 record libssl.so libcrypto.so as runtime library dependencies.
978 It is necessary to actually *use* a function from the library before
979 it gets recorded. The ld(1) options "--as-needed" and "no-as-needed"
980 control this. Evidently the default has changed in xenial.
981 That caused my smart "soname" detecting logic for openssl to
982 stop working. To make it work, the test program has to
983 actually reference routines or variables inside the library.
985 This is a quick fix for xenial / autoconf. There needs to be
986 a better fix for cmake and master.
988 Signed-off-by: Marcus Watts <mwatts@redhat.com>
990 commit db928d6b3f983b3a1ccc07440fcd3680700a1188
991 Author: Jason Dillaman <dillaman@redhat.com>
992 Date: Tue Feb 21 15:33:01 2017 -0500
994 rbd-mirror: retry object copy after -ENOENT error
996 Fixes: http://tracker.ceph.com/issues/18990
997 Signed-off-by: Jason Dillaman <dillaman@redhat.com>
998 (cherry picked from commit b4f36d5dc3f4f3cbb23f61cbb945b222248a50df)
1001 src/test/librados_test_stub/MockTestMemIoCtxImpl.h: sparse reads not supported
1002 src/test/rbd_mirror/image_sync/test_mock_ObjectCopyRequest.cc: sparse reads not supported
1004 commit a643fa80e03b41edcc720ff77b69ebaf24a23f3e
1005 Author: Jason Dillaman <dillaman@redhat.com>
1006 Date: Tue Feb 21 13:09:39 2017 -0500
1008 rbd-mirror: object copy should always reference valid snapshots
1010 If a remote snapshot is deleted while an image sync is in-progress,
1011 associate the read request against the most recent, valid remote
1012 snapshot for a given snapshot object clone.
1014 Signed-off-by: Jason Dillaman <dillaman@redhat.com>
1015 (cherry picked from commit 9a91efc3047963364944f8be91cee8e8f6afc49a)
1017 commit 37bbc95d56f73e4ec9c6e13ddbae199b14b104b4
1018 Author: Jason Dillaman <dillaman@redhat.com>
1019 Date: Tue Feb 21 11:52:00 2017 -0500
1021 rbd-mirror: replace complex object op tuple with struct
1023 Signed-off-by: Jason Dillaman <dillaman@redhat.com>
1024 (cherry picked from commit 0c181527c0e151784a0f7c466aaa70b0772f91b1)
1027 src/tools/rbd_mirror/image_sync/ObjectCopyRequest.cc: sparse reads not supported
1028 src/tools/rbd_mirror/image_sync/ObjectCopyRequest.h: sparse reads not supported
1030 commit 8941881e8986af0b2871c15d188d94d49e95dc02
1031 Author: Matt Benjamin <mbenjamin@redhat.com>
1032 Date: Fri Dec 30 23:30:16 2016 -0500
1034 rgw_file: interned RGWFileHandle objects need parent refs
1036 RGW NFS fhcache/RGWFileHandle operators assume existence of the
1037 full chain of parents from any object to the its fs_root--this is
1038 a consequence of the weakly-connected namespace design goal, and
1041 This change ensures the invariant by taking a parent ref when
1042 objects are interned (when a parent ref is guaranteed). Parent
1043 refs are returned when objects are destroyed--essentially by the
1044 invariant, such a ref must exist.
1046 The extra ref is omitted when parent->is_root(), as that node is
1047 not in the LRU cache.
1049 Fixes: http://tracker.ceph.com/issues/18650
1051 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
1052 (cherry picked from commit 0e5299f3f43e633a5d8a9360893b4b11f6217d81)
1054 commit 952f34f39cdd0438e4a4fb369ea8ca20b26488a8
1055 Author: David Zafman <dzafman@redhat.com>
1056 Date: Mon Dec 12 15:53:25 2016 -0800
1058 test: Update for new error message when doing scrub with deep-scrub errors
1060 Signed-off-by: David Zafman <dzafman@redhat.com>
1061 (cherry picked from commit 85e0774a7dded4aa6c67f237416041e25a7680bc)
1063 commit b655b98e48989ae954dc150fcb3d8976978cd90d
1064 Author: David Zafman <dzafman@redhat.com>
1065 Date: Fri Dec 9 12:14:06 2016 -0800
1067 osd: Add "trigger_scrub" admin socket command
1069 This allows testing to fake a schedule scrub (must_scrub not set)
1071 Signed-off-by: David Zafman <dzafman@redhat.com>
1072 (cherry picked from commit c8dc945260ee2ee841aca00fbc802d41036033d6)
1074 commit 94c958d2c9570e55305384ac86185e328746d2ff
1075 Author: David Zafman <dzafman@redhat.com>
1076 Date: Thu Dec 8 23:00:13 2016 -0800
1078 test: Add test for keeping deep-scrub information
1080 Signed-off-by: David Zafman <dzafman@redhat.com>
1081 (cherry picked from commit 64a7012e986ec88994c073b738fd08e8958395c3)
1083 commit 42eb24f18d5114410bd2a3e84e9219584d9e165e
1084 Author: David Zafman <dzafman@redhat.com>
1085 Date: Thu Dec 1 16:39:42 2016 -0800
1087 osd: When deep-scrub errors present upgrade regular scrubs
1089 Previously, if a weekly deep-scrub found errors the next daily scrub
1090 would cause the deep-scrub information to be removed.
1092 Signed-off-by: David Zafman <dzafman@redhat.com>
1093 (cherry picked from commit 532a759dca466181f37a329f86045c34f1a2506f)
1095 commit 76a457aa7c411dc5ed22f171954b8ed3c4661845
1096 Author: David Zafman <dzafman@redhat.com>
1097 Date: Wed Oct 19 17:10:29 2016 -0700
1099 tasks/scrub_test.py: Make test deterministic by updating digests
1101 ceph_test_rados: Get coverage of digest updates in deep-scrub/repair
1103 Signed-off-by: David Zafman <dzafman@redhat.com>
1104 (cherry picked from commit e000ab2f46b7b98a223176a2a25bb7195e601af6)
1106 commit bd1f1983129d6e1787b8b5eb4884c5e908f7b274
1107 Author: David Zafman <dzafman@redhat.com>
1108 Date: Wed Oct 19 17:10:07 2016 -0700
1110 repair_test, scrub_test: Fix whitelists for scrub changes
1112 Signed-off-by: David Zafman <dzafman@redhat.com>
1113 (cherry picked from commit 5fe8522df47f56842d227e08b2912623504afe24)
1115 commit 28106f06404407df7d05b35f9f570a80e785e635
1116 Author: David Zafman <dzafman@redhat.com>
1117 Date: Wed Aug 31 12:02:31 2016 -0700
1119 scrub_test: Fix for list-inconsistent-obj output changes
1121 Signed-off-by: David Zafman <dzafman@redhat.com>
1122 (cherry picked from commit 5ae0f5c75a8236d4a026f9ddcb5ff59964d90579)
1124 commit abcefc1395f76e5b05a988b970457a7f6dac6e8e
1125 Author: David Zafman <dzafman@redhat.com>
1126 Date: Tue Aug 30 12:11:44 2016 -0700
1128 doc, test: Add schemas for list-inconsistent-* rados command output
1130 If jsonschema cmd is available use it to test output against schema
1132 Signed-off-by: David Zafman <dzafman@redhat.com>
1133 (cherry picked from commit eb73dd473600fbbb45fad00194f7c46b565d6b81)
1135 commit 2fa2b64676555a91883a5aabb4e4237124308629
1136 Author: David Zafman <dzafman@redhat.com>
1137 Date: Tue Aug 30 12:22:55 2016 -0700
1139 test: Update testing for new list-inconsistent-obj output
1141 Signed-off-by: David Zafman <dzafman@redhat.com>
1142 (cherry picked from commit b7bacd219b000869b9c93e21edba4f8a3ace60d9)
1144 Adapted for difference in Jewel object_info
1146 commit 304f697ff1106695188b572e0da2415437040a6c
1147 Author: David Zafman <dzafman@redhat.com>
1148 Date: Tue Aug 30 12:22:29 2016 -0700
1150 rados, osd: Improve attrs output of list-inconsistent-obj
1152 Persist the user_version and shard id of scrubbed obj
1153 Rados command dump inconsistent obj's version and shard-id
1154 so they can be passed to repair command
1155 Rados list-inconsistent-obj output of attrs
1156 Make attrs an array since there are more than one
1157 Use base64 encode for values with non-printable chars
1158 Add indication if base64 encoding used
1159 Add checking for ss_attr_missing and ss_attr_corrupted
1160 Rename attr errors to attr_key_mismatch and attr_value_mismatch
1161 Add missing size_mismatch_oi scrub checking
1162 For erasure coded pools add ec_size_error and ec_hash_error not just read_error
1163 Use oi_attr_missing and oi_attr_corrupted just like list-inconsistent-snap does
1164 Pick an object info based on version and use that to find specific shards in error
1165 Check for object info inconsistency which should be rare
1166 Make all errors based on comparing shards to each other object errors
1167 We don't want give the impression that we've picked the correct one
1169 Signed-off-by: Kefu Chai <kchai@redhat.com>
1170 Signed-off-by: David Zafman <dzafman@redhat.com>
1171 (cherry picked from commit df3ff6dafeadb3822b35c424a890db9a14d7f60f)
1173 commit ea52f527e530b111b1bb26d10824c0230d662c4a
1174 Author: David Zafman <dzafman@redhat.com>
1175 Date: Thu Sep 1 14:45:01 2016 -0700
1177 osd: Fix logging to help with diagnostics
1179 These messages were not outputing the intended information
1181 Signed-off-by: David Zafman <dzafman@redhat.com>
1182 (cherry picked from commit b39001ec6ff0996860bcb5a3578bc2c66355c781)
1184 commit 5e8f8a2288aa1a9f1de86d3f7d0f9c66c795113e
1185 Author: David Zafman <dzafman@redhat.com>
1186 Date: Tue Jun 21 18:05:25 2016 -0700
1188 test: Fix use of wait_for_clean()
1190 Signed-off-by: David Zafman <dzafman@redhat.com>
1191 (cherry picked from commit d0503a52d0fe5505bdb38dfd613c03a20500c05d)
1193 commit 2230e6adc66ca3e5dbe81aecccdb4a435e93e256
1194 Author: David Zafman <dzafman@redhat.com>
1195 Date: Tue Aug 30 10:56:06 2016 -0700
1197 common: Change cleanbin() to use base64 encoding, update ceph-objectstore-tool
1199 Signed-off-by: David Zafman <dzafman@redhat.com>
1200 (cherry picked from commit 92e982c71995b863466d83671468f84761cb1793)
1202 commit 83ea077ee560d31a5c302a62b55451a2571fda8d
1203 Author: David Zafman <dzafman@redhat.com>
1204 Date: Tue Aug 30 11:05:16 2016 -0700
1206 common: Move cleanbin() function to common/util.cc
1208 Signed-off-by: David Zafman <dzafman@redhat.com>
1209 (cherry picked from commit 5c79074ffaee34b2956d9dfc67b1eff9f39b47f3)
1212 src/tools/CMakeLists.txt (changes goes in src/CMakeLists.txt)
1214 commit ba84ca9fa391d963d1d4e46fbf27ce3dbff397be
1215 Author: David Zafman <dzafman@redhat.com>
1216 Date: Wed Jun 8 09:48:00 2016 -0700
1218 test: Add test support for deep-scrub
1220 Signed-off-by: David Zafman <dzafman@redhat.com>
1221 (cherry picked from commit 907e79e2b77835d1aca9e8ba2fae5fa2fd437e5a)
1223 commit 6a421d19d0be4d59beed5c69bb0aa4477d65a14e
1224 Author: David Zafman <dzafman@redhat.com>
1225 Date: Sat Jun 18 17:58:36 2016 -0700
1227 common: Fix indentation
1229 Signed-off-by: David Zafman <dzafman@redhat.com>
1230 (cherry picked from commit a74418a29e4e469117fc0c4edd80f78b62944c98)
1232 commit c7c3e070336dc898460d4338b4d78b8963467c8a
1233 Author: David Zafman <dzafman@redhat.com>
1234 Date: Fri Aug 12 12:06:31 2016 -0700
1236 osd: Handle corrupt attributes in get_object_context()
1238 Signed-off-by: David Zafman <dzafman@redhat.com>
1239 (cherry picked from commit 30f8b0d6593718dd10bcf6ff459c95b4bb68c05c)
1241 commit 8006ba7e86ebb8201a040c427cc95197901064be
1242 Author: Kefu Chai <kchai@redhat.com>
1243 Date: Fri Nov 11 21:37:50 2016 +0800
1245 ReplicatedPG::failed_push: release read lock on failure
1247 and requeue the blocked ops.
1249 Fixes: http://tracker.ceph.com/issues/17857
1250 Signed-off-by: Kefu Chai <kchai@redhat.com>
1251 (cherry picked from commit b3224a18f6acc7ed54c2162b140a33b6146a16be)
1253 commit 5ca69d57dbe2ee7acc64d28ca35bb390bf463199
1254 Author: David Zafman <dzafman@redhat.com>
1255 Date: Fri Oct 14 16:17:55 2016 -0700
1257 test.sh: Make check for flags more robust
1259 Low space broke test, saw "flags nearfull,pauserd,pausewr...."
1261 Signed-off-by: David Zafman <dzafman@redhat.com>
1262 (cherry picked from commit b4c080b1b4378d659c1ca8a17811cd6f84595166)
1264 commit 1be38e5ebd2d01d5d527b05e64b026df955ffe69
1265 Author: David Zafman <dzafman@redhat.com>
1266 Date: Thu Jul 21 17:36:34 2016 -0700
1268 test: Remove extra objectstore_tool call which causes a recovery
1270 Caused by: 70e000a9a42c50eda48f5d3b9e432ffc3a70f75b
1272 Signed-off-by: David Zafman <dzafman@redhat.com>
1273 (cherry picked from commit 6904529d1b53993304de2927500937ba0d493e9e)
1275 commit 026f181e72f2e09e769a8821dfe2c99f6213e6a3
1276 Author: David Zafman <dzafman@redhat.com>
1277 Date: Tue Jun 14 20:09:15 2016 -0700
1279 test: Handle object removals in a non-racey way
1281 Signed-off-by: David Zafman <dzafman@redhat.com>
1282 (cherry picked from commit bfe3ebd94fdc1ef36ebe6e8f9d38acea322eca22)
1284 commit c5ef385c25086dc1582ebe5343481d05283b0cc6
1285 Author: David Zafman <dzafman@redhat.com>
1286 Date: Thu May 26 22:09:42 2016 -0700
1288 osd: Fix hang on unfound object after mark_unfound_lost is done
1290 Signed-off-by: David Zafman <dzafman@redhat.com>
1291 (cherry picked from commit 73a27533beba3587f8447b4d41d200427c45042b)
1294 src/osd/ReplicatedPG.cc (trivial)
1296 commit 6284f745157458439699c76e8616455c50d1eb71
1297 Author: David Zafman <dzafman@redhat.com>
1298 Date: Fri May 20 15:20:18 2016 -0700
1300 osd: Handle recovery read errors
1302 Fixes: http://tracker.ceph.com/issues/13937
1304 Signed-off-by: David Zafman <dzafman@redhat.com>
1305 (cherry picked from commit c51d70e1e837c972e42ddd5fa66f7ca4477b95cc)
1308 src/osd/ReplicatedPG.h (trivial)
1310 commit 27393a8c05d4656f342ecd32817307e558a2e400
1311 Author: David Zafman <dzafman@redhat.com>
1312 Date: Fri May 20 18:19:42 2016 -0700
1314 osd: Fix log messages
1316 Signed-off-by: David Zafman <dzafman@redhat.com>
1317 (cherry picked from commit b40ec3fe890679b4c73778815e1bedd492cb264b)
1319 commit cbf66f3b16f194dd0c67b46e0fec247e02e84134
1320 Author: David Zafman <dzafman@redhat.com>
1321 Date: Fri May 20 13:58:32 2016 -0700
1323 osd: CLEANUP: Remove unused pending_read member
1325 Remove unused struct
1327 Signed-off-by: David Zafman <dzafman@redhat.com>
1328 (cherry picked from commit 36fd68c96653e83f87767feb08530a9fc6e841b0)
1330 commit 228b91d72a15b62adc10591604c4e8a849df53d5
1331 Author: David Zafman <dzafman@redhat.com>
1332 Date: Tue Oct 11 18:08:12 2016 -0700
1334 test/osd-scrub-repair.sh: Use test case specific object names to help with diagnostics
1336 Signed-off-by: David Zafman <dzafman@redhat.com>
1337 (cherry picked from commit 0bf4da589155ee50969812492cfbc66368efb54c)
1339 commit ee3c67c9cbfdeeb3e628bac34b708cf150b3862e
1340 Merge: 3dbf0c9 39848e4
1341 Author: Nathan Cutler <presnypreklad@gmail.com>
1342 Date: Wed Feb 1 23:28:22 2017 +0100
1344 Merge pull request #13143 from linuxbox2/jewel-mdw-rgw-lf
1346 jewel: rgw: radosgw/swift: clean up flush / newline behavior.
1348 Reviewed-by: Nathan Cutler <ncutler@suse.com>
1350 commit 39848e41b7c517cc5faab1ccf77c2804fd7d2628
1351 Author: Marcus Watts <mwatts@redhat.com>
1352 Date: Wed Jan 11 00:06:15 2017 -0500
1354 radosgw/swift: clean up flush / newline behavior.
1356 The current code emits a newline after swift errors, but fails
1357 to account for it when it calculates 'content-length'. This results in
1358 some clients (go github.com/ncw/swift) producing complaints about the
1359 unsolicited newline such as this,
1360 Unsolicited response received on idle HTTP channel starting with "\n"; err=<nil>
1362 This logic eliminates the newline on flush. This makes the content length
1363 calculation correct and eliminates the stray newline.
1365 There was already existing separator logic in the rgw plain formatter
1366 that can emit a newline at the correct point. It had been checking
1367 "len" to decide if previous data had been emitted, but that's reset to 0
1368 by flush(). So, this logic adds a new per-instance variable to separately
1369 track state that it emitted a previous item (and should emit a newline).
1371 Fixes: http://tracker.ceph.com/issues/18473
1372 Signed-off-by: Marcus Watts <mwatts@redhat.com>
1373 Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
1374 (cherry picked from commit 5f229d6a33eae4906f22cdb90941835e47ee9f02)
1376 commit 3fce77ab3662496368e25cbbf6d1b37d3c140db9
1377 Author: Vikhyat Umrao <vumrao@redhat.com>
1378 Date: Wed Jan 25 21:54:27 2017 +0530
1380 osd: jewel fix OSD network address in OSD heartbeat_check log message
1382 Fixes: http://tracker.ceph.com/issues/18657
1384 Signed-off-by: Vikhyat Umrao <vumrao@redhat.com>
1386 commit 14a6aabe22f68436ea3297ce0851700f86ee5b12
1387 Author: Marcus Watts <mwatts@redhat.com>
1388 Date: Wed Aug 3 17:36:55 2016 -0400
1390 rgw: Handle multiple listening addreses w/ optional ssl "correctly" with civetweb.
1392 For civetweb: accept a range of port numbers joined with '+'.
1393 Port numbers may include an ipaddress: prefix and 's' suffix.
1394 Additionally, use "mg_get_local_addr" to correctly deduce host port per
1395 incoming connection.
1397 civetweb can accept connections on multiple ports, some of which
1398 might have SSL turned on and some not. Both s3 and swift have various
1399 authorization protocols in which the port number matters. In the generic
1400 radosgw frontend process, each frontend only has one port number, but
1401 we should want to have both ssl and non-ssl connections managed within
1402 one rgw frontend, because the thread pool is also per front-end, and
1403 that *is* a scarce resource.
1405 So, this patch enables the use of multiple ports with a single civetweb
1406 frontend. To indicate https: append an 's' to portno. To use multiple
1407 ports, use +. So 80+443s indicates use of the usual default http ports.
1408 The parsed port is not stored in the frontend structure,
1410 So instead, this patch adds logic to use the results of
1411 mg_get_local_addr() on a per-connection basis insetad of the generic
1412 front-end port number. This will affect "v4" s3 authorization, and also
1413 affect swift pre-signed URLs.
1415 mg_get_local_addr() is a new customization to civetweb; that submodule
1416 was updated (in a temporary repository) by the previous commit to this.
1418 Signed-off-by: Marcus Watts <mwatts@redhat.com>
1419 (cherry picked from commit 8bc6decc0774fae9ac881f91e73da55deebe3360)
1421 commit 698250563ccc4c69e5ca5aebf65dc352d80a8bdd
1422 Author: Marcus Watts <mwatts@redhat.com>
1423 Date: Tue Dec 20 00:22:02 2016 -0500
1425 rgw: s3: secure_port should override port, also apply ssl default right.
1427 Without https, only port is set. With https, secure_port and port are
1428 both set to the same value. The previous logic looked at port first and
1429 had overly simplified conditional logic which was liable to try to apply
1430 both non-default cases. The correct behavior is: look secure_port first,
1431 and if secure_port is set, then only check to see if it's a non-default
1434 Signed-off-by: Marcus Watts <mwatts@redhat.com>
1435 (cherry picked from commit a113cf5ff5a642d2ee4cc83f5c7001b4bfe0a5df)
1437 commit 28f2841c8b9a832e486b9f89c574a4591bf3d448
1438 Author: Marcus Watts <mwatts@redhat.com>
1439 Date: Wed Nov 2 00:18:48 2016 -0400
1441 rgw: Get civetweb ssl enhancement: wip-listen3 = mg_get_local_addr
1443 The logic inside of radosgw that computes aws v4 signatures wants to know
1444 what server port the client connected. The patch to civetweb patch adds a
1445 call mg_get_local_addr() which will permit that code to actually find out
1446 on what address a connection was received, rather than merely guessing
1447 based on configuration as it previously did.
1449 Signed-off-by: Marcus Watts <mwatts@redhat.com>
1450 (cherry picked from commit 46ced9ddd2795f00f014e22e5637070b49e7a6d5)
1452 commit 8d83dfb4176ede7490d0cab589e9566bc7d4e387
1453 Author: Marcus Watts <mwatts@redhat.com>
1454 Date: Wed Mar 23 17:32:14 2016 -0400
1456 rgw: Document that radosgw now supports SSL.
1458 This includes information on file format and configuration file syntax.
1460 Signed-off-by: Marcus Watts <mwatts@redhat.com>
1461 (cherry picked from commit d4e72dfed30274b3cfbad4ac58c0746a98c0148b)
1463 commit e3f80c9d0ed6762a39fc242561f5ea26f0f26546
1464 Author: Marcus Watts <mwatts@redhat.com>
1465 Date: Fri Jan 13 03:39:57 2017 -0500
1467 rgw: civetweb/openssl: automagic: load libssl.so and libcrypto.so by soname.
1469 If building with radosgw, always look for openssl library (even when
1470 building with nss). Then, use objdump to fetch SONAME from the copies
1471 of libssl and libcrypto that were found. When building civetweb; pass
1472 the library soname values in as the libraries to load with "dlopen".
1474 The issue reported here against master
1475 http://tracker.ceph.com/issues/16535
1476 reflects stuff that doesn't seem to have made it into jewel
1477 (which had: -lssl -lcrypto hard-wired into it.)
1478 Still, since people were pretty riled about making that change,
1479 this puts things in congruence to the final result in master.
1480 (which is: runtime shared object load of ssl crypto by soname.)
1482 Fixes: http://tracker.ceph.com/issues/11239
1484 Signed-off-by: Marcus Watts <mwatts@redhat.com>
1485 (inspired by commit 7caa0bd002110b62514da83a37a2a3deb841267a)
1487 commit 045551485415309ba9bad77e1aee28a0008881ca
1488 Author: Marcus Watts <mwatts@redhat.com>
1489 Date: Mon Jul 11 17:54:07 2016 -0400
1491 rgw: civetweb/openssl: Load libssl.so and libcrypto.so by soname.
1493 If building with radosgw, always look for openssl library (even when
1494 building with nss). Then, use objdump to fetch SONAME from the copies
1495 of libssl and libcrypto that were found. When building civetweb; pass
1496 the library soname values in as the libraries to load with "dlopen".
1498 This is a problem that went away for a bit, but came back with some
1500 http://tracker.ceph.com/issues/16535
1502 Fixes: https://bugzilla.redhat.com/show_bug.cgi?id=1341775
1503 Fixes: https://bugzilla.redhat.com/show_bug.cgi?id=1258961
1505 Fixes: http://tracker.ceph.com/issues/11239
1507 Signed-off-by: Marcus Watts <mwatts@redhat.com>
1508 (cherry picked from commit 7caa0bd002110b62514da83a37a2a3deb841267a)
1510 commit 386640865dee30d38f17e55fc87535e419bc3cb5
1511 Author: Marcus Watts <mwatts@redhat.com>
1512 Date: Thu Nov 3 23:33:44 2016 -0400
1514 rgw: cmake: remove useless civetweb include path side effect.
1516 For 'target_include_directories" for the cmake object library
1517 'civetweb_common_objs', change from PUBLIC to PRIVATE. This doesn't
1518 break anything, so it wasn't doing anything useful. If it has it
1519 any effect, it would be to cause everything that linked against this
1520 "library" to also use the indictated include path. Which would be great
1521 except everything in ceph wants to include "civetweb/civetweb.h" and
1522 not "civetweb.h". We already make separate arrangements elsewhere for
1523 that to work. Additionally, static object libraries in cmake aren't
1524 really libraries, so I'm not entirely sure this even does anything.
1525 So: making this public is not useful, and could be harmful. Making it
1526 private makes this only take effect for building civetweb.c itself,
1527 exactly the effect we we require, and no more.
1529 Signed-off-by: Marcus Watts <mwatts@redhat.com>
1530 (cherry picked from commit 8308a13b0257c9460fd2a721c20b0c37cb9e7c57)