1 commit 283c2e7cfa2457799f534744d7d549f83ea1335e (tag: refs/tags/v0.87.1, refs/remotes/gh/giant)
2 Author: Jenkins <jenkins@inktank.com>
3 Date: Mon Feb 23 12:02:04 2015 -0800
7 commit 4178e32dd085adeead84fb168ab8a8a121256259
9 Author: Loic Dachary <ldachary@redhat.com>
10 Date: Tue Feb 17 01:09:54 2015 +0100
12 Merge pull request #3731 from liewegas/wip-10834-giant
14 osd: tolerate sessionless con in fast dispatch path
16 Reviewed-by: Loic Dachary <ldachary@redhat.com>
18 commit 734e9af5df4ae419ded108f5036bee068a9bc2b2
19 Author: Sage Weil <sage@redhat.com>
20 Date: Mon Dec 1 18:15:59 2014 -0800
22 osd: tolerate sessionless con in fast dispatch path
24 We can now get a session cleared from a Connection at any time. Change
25 the assert to an if in ms_fast_dispatch to cope. It's pretty rare, but it
26 can happen, especially with delay injection. In particular, a racing
27 thread can call mark_down() on us.
31 Signed-off-by: Sage Weil <sage@redhat.com>
32 (cherry picked from commit 01df2227125abf94571b4b0c7bccca57098ed2dc)
34 commit ccb0914f76da23acdd7374233cd1939ab80ef3c8
35 Author: Josh Durgin <jdurgin@redhat.com>
36 Date: Mon Feb 2 16:43:35 2015 +0100
38 qa: use correct binary path on rpm-based systems
41 Signed-off-by: Josh Durgin <jdurgin@redhat.com>
42 (cherry picked from commit 05ce2aa1bf030ea225300b48e7914577a412b38c)
44 commit 78c71b9200da5e7d832ec58765478404d31ae6b5
45 Merge: 222aa22 91515e7
46 Author: Loic Dachary <ldachary@redhat.com>
47 Date: Wed Feb 11 00:11:57 2015 +0100
49 Merge pull request #3407 from ceph/wip-9854-giant
51 osdc: Constrain max number of in-flight read requests
53 commit 222aa22ebc0ccb1b04156e0c9d05f4e4733ec290
54 Merge: b9ff170 a5cb39c
55 Author: Loic Dachary <ldachary@redhat.com>
56 Date: Tue Feb 10 22:01:21 2015 +0100
58 Merge pull request #3568 from dachary/wip-10471-rgw-giant
60 rgw: use s->bucket_attrs instead of trying to read obj attrs
62 commit b9ff1708ad85ca5aeb10b4202bcbe197251e3bd8
63 Merge: 34103b6 b1e4882
64 Author: Loic Dachary <ldachary@redhat.com>
65 Date: Tue Feb 10 21:59:40 2015 +0100
67 Merge pull request #3263 from dachary/wip-jerasure-giant
69 erasure-code: update links to jerasure upstream (giant)
71 commit 34103b6355881820aa10b354c2427654bf229e8f
72 Merge: 94889cf d125743
73 Author: Loic Dachary <ldachary@redhat.com>
74 Date: Tue Feb 10 21:59:17 2015 +0100
76 Merge pull request #3191 from ceph/giant-10277
80 commit 94889cf6bef5a542e51bf8434dbe7c68f64604ce
81 Merge: d7b10d8 d28c8e0
82 Author: Loic Dachary <ldachary@redhat.com>
83 Date: Tue Feb 10 21:58:52 2015 +0100
85 Merge pull request #3186 from ceph/wip-giant-mon-backports
87 mon: backports for #9987 against giant
89 commit d7b10d897e17bc3fa690c8484ad2d6f233896237
90 Merge: 11f7d06 16c6d0d
91 Author: Loic Dachary <ldachary@redhat.com>
92 Date: Tue Feb 10 21:58:29 2015 +0100
94 Merge pull request #3185 from dachary/wip-10325-cauchy-giant
96 erasure-code: relax cauchy w restrictions (giant)
98 commit 11f7d064e5d93bc0ed8896750344c6cf6b37aeab
99 Merge: 975be75 636b98f
100 Author: Loic Dachary <ldachary@redhat.com>
101 Date: Tue Feb 10 21:58:05 2015 +0100
103 Merge pull request #3178 from dachary/wip-9998-crush-underfloat-giant
105 crush: fix weight underfloat issue (giant)
107 commit 975be75f4bcea88b232ea76087b49e288d7c29f7
108 Merge: 51fe79d d759e71
109 Author: Loic Dachary <ldachary@redhat.com>
110 Date: Tue Feb 10 21:57:50 2015 +0100
112 Merge pull request #3579 from dachary/wip-9877-rgw-giant
114 rgw: RGWRados::get_obj() returns wrong len if len == 0
116 commit 51fe79d9e63c7df4da547a0ba7a12aa9c6cd7ab2
117 Merge: fca9ead 319f9c9
118 Author: Loic Dachary <ldachary@redhat.com>
119 Date: Tue Feb 10 21:57:18 2015 +0100
121 Merge pull request #3168 from ceph/wip-8797-giant
125 commit fca9eadaf3fcef77e269d4936d4eea86ab6c3faf
126 Merge: 317532b 9886620
127 Author: Loic Dachary <ldachary@redhat.com>
128 Date: Tue Feb 10 21:56:53 2015 +0100
130 Merge pull request #3582 from dachary/wip-10062-rgw-giant
132 rgw: s3 keystone auth fixes
134 commit 317532b70383762f473a910e043c889574eb6087
135 Merge: 3e8f3e3 debc0c5
136 Author: Loic Dachary <ldachary@redhat.com>
137 Date: Tue Feb 10 21:56:37 2015 +0100
139 Merge pull request #3581 from dachary/wip-9918-rgw-giant
141 rgw: update swift subuser perm masks when authenticating
143 commit 3e8f3e38af76fa2cba86aedf962d3230d7979f63
144 Merge: 1d77591 76f9de3
145 Author: Loic Dachary <ldachary@redhat.com>
146 Date: Tue Feb 10 21:56:17 2015 +0100
148 Merge pull request #3580 from dachary/wip-9907-rgw-giant
150 fix can not disable max_size quota
152 commit 1d7759149697242192be05decf7ffafb17b24cbe
153 Merge: b016863 ad04a67
154 Author: Loic Dachary <ldachary@redhat.com>
155 Date: Tue Feb 10 21:55:56 2015 +0100
157 Merge pull request #3083 from dachary/wip-10211-erasure-code-buffer-alignement-giant
159 erasure-code: enforce chunk size alignment (giant)
161 commit b016863ad243388e7571da9ffca3013c8f99237a
162 Merge: bdcc9dc d21f4e3
163 Author: Loic Dachary <ldachary@redhat.com>
164 Date: Tue Feb 10 21:55:23 2015 +0100
166 Merge pull request #3577 from dachary/wip-9587-rgw-giant
168 init-radosgw.sysv: set ulimit -n before starting daemon
170 commit bdcc9dcb8586c91b432c7087e33a2b52ef467b54
171 Merge: df475f9 7b5f746
172 Author: Loic Dachary <ldachary@redhat.com>
173 Date: Tue Feb 10 21:54:58 2015 +0100
175 Merge pull request #3576 from dachary/wip-5595-9576-rgw-giant
177 update object content-length doesn't work correctly
179 commit df475f92a41e3bd5a022335b2c9023ad40c3b47b
180 Merge: db7adf8 b2f6f7f
181 Author: Loic Dachary <ldachary@redhat.com>
182 Date: Tue Feb 10 21:54:13 2015 +0100
184 Merge pull request #3575 from dachary/wip-9479-rgw-giant
186 rgw: send back ETag on S3 object copy
188 commit db7adf8d8ca225fea2d0277ced614e936df086c9
189 Merge: 9b50db9 67ba4d3
190 Author: Loic Dachary <ldachary@redhat.com>
191 Date: Tue Feb 10 21:53:55 2015 +0100
193 Merge pull request #3574 from dachary/wip-9478-rgw-giant
195 rgw: S3 object copy content type fix
197 commit 9b50db97a9552841ed143588e2f63bab56d0aecb
198 Merge: 583fe31 84e9b6c
199 Author: Loic Dachary <ldachary@redhat.com>
200 Date: Tue Feb 10 21:53:33 2015 +0100
202 Merge pull request #3573 from dachary/wip-9254-rgw-giant
204 rgw: http headers need to end with \r\n
206 commit 583fe31681c4eea8b85b413674074445e2b424a6
207 Merge: 5240db5 1cb0955
208 Author: Loic Dachary <ldachary@redhat.com>
209 Date: Tue Feb 10 21:53:15 2015 +0100
211 Merge pull request #3572 from dachary/wip-9973-rgw-giant
213 rgw: remove swift user manifest (DLO) hash calculation
215 commit 5240db588e9017dd8a487b7a9ee16f171fdda1ff
216 Merge: 62e1552 e230fab
217 Author: Loic Dachary <ldachary@redhat.com>
218 Date: Tue Feb 10 21:52:54 2015 +0100
220 Merge pull request #3571 from dachary/wip-8911-rgw-giant
222 rgw: swift GET / HEAD object returns X-Timestamp field
224 commit 62e15528dda20b5419e39744fa9e0c9c4cae053c
225 Merge: 16cd892 c24fab3
226 Author: Loic Dachary <ldachary@redhat.com>
227 Date: Tue Feb 10 21:52:36 2015 +0100
229 Merge pull request #3570 from dachary/wip-10701-rgw-giant
231 rgw: use strict_strtoll() for content length
233 commit 16cd892aab4ffb1dc15b93a4101d9bc209591c94
234 Merge: 028904c 6aef29e
235 Author: Loic Dachary <ldachary@redhat.com>
236 Date: Tue Feb 10 21:52:00 2015 +0100
238 Merge pull request #3569 from dachary/wip-10103-rgw-giant
240 rgw-admin: create subuser if needed when creating user
242 commit 028904cf7c36a1d5342cf29c115bc0437e9b2d74
243 Merge: 520dcf8 425ee8a
244 Author: Loic Dachary <ldachary@redhat.com>
245 Date: Tue Feb 10 21:51:06 2015 +0100
247 Merge pull request #3567 from dachary/wip-10307-rgw-giant
249 rgw: use s->bucket_attrs instead of trying to read obj attrs
251 commit 520dcf8624eeafd694115b382616be83f9b344d3
252 Merge: cae1de2 14cdb9b
253 Author: Loic Dachary <ldachary@redhat.com>
254 Date: Tue Feb 10 21:50:34 2015 +0100
256 Merge pull request #3443 from ceph/wip-10590-giant
258 rbd: ensure aio_write buffer isn't invalidated during image import
260 commit cae1de29922c2183eff021c6fe2b921a87b5f5b2
261 Merge: b346ad3 83a0a2e
262 Author: Loic Dachary <ldachary@redhat.com>
263 Date: Tue Feb 10 21:50:14 2015 +0100
265 Merge pull request #3557 from dachary/wip-10688-boost-157-giant
269 commit b346ad37a84b7cfd0bae84528f2533a8cc4a8e3d
270 Merge: aacd51c 13bb880
271 Author: Loic Dachary <ldachary@redhat.com>
272 Date: Tue Feb 10 21:49:47 2015 +0100
274 Merge pull request #2954 from sponce/giant
276 Fixed trivial locking issue in the trunc method of libradosstriper - Giant branch
278 commit aacd51c74c102d44982421b9bc384d12fc160e3c
279 Merge: 1d97c7c 081f49b
280 Author: Loic Dachary <ldachary@redhat.com>
281 Date: Tue Feb 10 21:49:20 2015 +0100
283 Merge pull request #3405 from ceph/wip-10299-giant
285 librbd: complete all pending aio ops prior to closing image
287 commit 1d97c7c9a3087e7bc98774d9fe2882bdc4a84531
288 Merge: 53dec0e 436923c
289 Author: Loic Dachary <ldachary@redhat.com>
290 Date: Tue Feb 10 21:48:49 2015 +0100
292 Merge pull request #3403 from ceph/wip-10270-giant
294 librbd: gracefully handle deleted/renamed pools
296 commit 53dec0eeee60b315e88acb4ba05666857ae3e0eb
297 Merge: df8285c 1261bf2
298 Author: Loic Dachary <ldachary@redhat.com>
299 Date: Tue Feb 10 21:48:20 2015 +0100
301 Merge pull request #3356 from liewegas/wip-msgr-giant
303 msgr: fast dispatch backports for giant
305 commit df8285c5e5b14d9a4cd42fb9be8e18fe6cdf6f83
306 Author: Greg Farnum <gfarnum@redhat.com>
307 Date: Thu Feb 5 21:12:17 2015 -0800
309 fsync-tester: print info about PATH and locations of lsof lookup
311 We're seeing the lsof invocation fail (as not found) in testing and nobody can
312 identify why. Since attempting to reproduce the issue has not worked, this
313 patch will gather data from a genuinely in-vitro location.
315 Signed-off-by: Greg Farnum <gfarnum@redhat.com>
316 (cherry picked from commit a85051483874ff5b8b0fb50426a3577040457596)
318 commit 91515e750bfe2453ce8ac9ec568b0e314823dd82
319 Author: Jason Dillaman <dillaman@redhat.com>
320 Date: Mon Oct 27 14:47:19 2014 -0400
322 osdc: Constrain max number of in-flight read requests
324 Constrain the number of in-flight RADOS read requests to the
325 cache size. This reduces the chance of the cache memory
326 ballooning during certain scenarios like copy-up which can
327 invoke many concurrent read requests.
330 Backport: giant, firefly, dumpling
331 Signed-off-by: Jason Dillaman <dillaman@redhat.com>
333 commit 98866208c64348ca885335d95a1c737071a17004
334 Author: Abhishek Lekshmanan <abhishek.lekshmanan@gmail.com>
335 Date: Mon Dec 22 19:57:19 2014 +0530
337 rgw: check keystone auth also for s3 post requests
339 This patch adds keystone auth for s3 post requests, once a user fails in
340 cephx authentication, they are checked for keystone if configured.
343 Signed-off-by: Abhishek Lekshmanan <abhishek.lekshmanan@gmail.com>
345 (cherry picked from commit 8b3dfc9472022ea45ad24e02e0aa21dfdad798f8)
347 commit 4e4372b8e551bb1b974f08dc69f5b27bdd22bb4b
348 Author: Abhishek Lekshmanan <abhishek.lekshmanan@gmail.com>
349 Date: Mon Nov 17 17:37:00 2014 +0530
351 rgw: check for timestamp for s3 keystone auth
353 This commit ensures that we check for timestamp of s3 request is within
354 acceptable grace time of radosgw
355 Addresses some failures in #10062
357 Signed-off-by: Abhishek Lekshmanan <abhishek.lekshmanan@gmail.com>
359 (cherry picked from commit 4b35ae067fef9f97b886afe112d662c61c564365)
361 commit debc0c593fb7401d07a34f7916380092ad7285f9
362 Author: Yehuda Sadeh <yehuda@redhat.com>
363 Date: Wed Nov 5 14:38:46 2014 -0800
365 rgw: update swift subuser perm masks when authenticating
368 Backport: firefly, giant
369 It seems that we weren't setting the swift perm mask correctly.
371 Signed-off-by: Yehuda Sadeh <yehuda@redhat.com>
372 (cherry picked from commit 5d9f36f757a7272c24d2c9adc31db1ed5e712992)
374 commit 76f9de330deaa4fafb86d3f6f2754f0d419306b1
375 Author: Lei Dong <leidong@yahoo-inc.com>
376 Date: Mon Oct 27 10:29:48 2014 +0800
378 fix can not disable max_size quota
380 Currently if we enable quota and set max_size = -1, it doesn’t
381 mean max_size is unlimited as expected. Instead, it means object
382 with any size is not allowed to upload because of “QuotaExceeded”.
383 The root cause is the function rgw_rounded_kb which convert max_size
384 to max_size_kb returns 0 for -1 because it takes an unsigned int
385 but we pass an int to it. A simple fix is check max_size before
386 it’s rounded to max_size_kb.
389 1 enable and set quota:
390 radosgw-admin quota enable --uid={user_id} --quota-scope=user
391 radosgw-admin quota set --quota-scope=user --uid={user_id}\
392 --max-objects=100 --max-size=-1
393 2 upload any object with non-zero length
394 it will return 403 with “QuotaExceeded” and return 200 if you apply the fix.
397 Backport: giant, firefly
398 Signed-off-by: Dong Lei leidong@yahoo-inc.com
399 (cherry picked from commit abd3fd3ef9ee9999b99811937af60b7a5e673e35)
401 commit d759e71c8167ea29c8fda9483039a3e491083da5
402 Author: Yehuda Sadeh <yehuda@redhat.com>
403 Date: Tue Nov 4 22:05:03 2014 -0800
405 rgw: RGWRados::get_obj() returns wrong len if len == 0
408 We only updated if len was > 0, should update it if r >= 0. This was the
409 culprit for issue #9877.
410 Backport: giant, firefly
412 Signed-off-by: Yehuda Sadeh <yehuda@redhat.com>
413 (cherry picked from commit fe7bf06366adaf787816d1e68f5e3f68e8c91134)
415 commit d21f4e326eb4821cc9bd38a1b62a0210272277d4
416 Author: Sage Weil <sage@redhat.com>
417 Date: Tue Oct 21 17:59:30 2014 -0700
419 init-radosgw.sysv: set ulimit -n before starting daemon
421 If we do the ulimit inside the daemon command we will have already
422 dropped privs and will fail.
425 Backport: giant, firefly
426 Signed-off-by: Sage Weil <sage@redhat.com>
427 (cherry picked from commit 9803cedf54a7baff45ccd0e0f65d2bc220958a46)
429 commit 7b5f746d2f97c7139f9c31962c107a074bfd1863
430 Author: Yehuda Sadeh <yehuda@redhat.com>
431 Date: Tue Sep 23 12:42:10 2014 -0700
433 rgw: PutObjMetadata, clear bufferlist before appending into it
436 Backport: firefly, dumpling
438 We need to completely rewrite the bufferlist, not append into it.
440 Signed-off-by: Yehuda Sadeh <yehuda@redhat.com>
441 (cherry picked from commit 44cfd88dc65d30f4295743c5834768bb13f7b805)
443 commit e24f27b7b2e2aeb84b14788e8bf2757ecdf8f0c0
444 Author: Yehuda Sadeh <yehuda@redhat.com>
445 Date: Tue Sep 23 12:43:55 2014 -0700
447 rgw: rados->set_attrs() updates bucket index
450 Backport: dumpling, firefly
451 We need to update the bucket index when updating object attrs, otherwise
452 we're missing meta changes that need to be registered. It also
453 solves issue of bucket index not knowing about object acl changes,
454 although this one still requires some more work.
456 Signed-off-by: Yehuda Sadeh <yehuda@redhat.com>
457 (cherry picked from commit f833f12a200ecc2c4f74ddb443d6fa61b7ad14db)
459 commit b2f6f7f6205682aeb09c0785e373ddf5d89c6d04
460 Author: Yehuda Sadeh <yehuda@redhat.com>
461 Date: Wed Nov 5 13:28:02 2014 -0800
463 rgw: send back ETag on S3 object copy
466 Backport: firefly, giant
467 We didn't send the etag back correctly. Original code assumed the etag
468 resided in the attrs, but attrs only contained request attrs.
470 Signed-off-by: Yehuda Sadeh <yehuda@redhat.com>
471 (cherry picked from commit b1bfc3a7e0c9088f01f8ff770ae14f569fbc570d)
473 commit 67ba4d3444f0f64dae6286be28276ba85376ecf6
474 Author: Yehuda Sadeh <yehuda@redhat.com>
475 Date: Wed Nov 5 12:35:42 2014 -0800
477 rgw: S3 object copy content type fix
480 Backport: firefly, giant
481 Content type for S3 object copy response should be set to
484 Signed-off-by: Yehuda Sadeh <yehuda@redhat.com>
485 (cherry picked from commit 31963459a0a869c4d32f55baa629f36df33eaa90)
487 commit 84e9b6c32c6fdc38e2c64f3360c185332e691bf4
488 Author: Yehuda Sadeh <yehuda@redhat.com>
489 Date: Fri Dec 12 11:20:26 2014 -0800
491 rgw: http headers need to end with \r\n
494 Backport: firefly, giant
496 Reported-by: Benedikt Fraunhofer <fraunhofer@traced.net>
497 Signed-off-by: Yehuda Sadeh <yehuda@redhat.com>
498 (cherry picked from commit 7409ab3df18fb312dd6c9f79084f889c523afdce)
500 commit 1cb09555d3fdb568296797cd83eb5557552f056c
501 Author: Yehuda Sadeh <yehuda@redhat.com>
502 Date: Wed Nov 5 13:40:55 2014 -0800
504 rgw: remove swift user manifest (DLO) hash calculation
507 Backport: firefly, giant
509 Previously we were iterating through the parts, creating hash of the
510 parts etags (as S3 does for multipart uploads). However, swift just
511 calculates the etag for the empty manifest object.
513 Signed-off-by: Yehuda Sadeh <yehuda@redhat.com>
514 (cherry picked from commit ef6d3ad964d34bc526dc4435486bd5c8cdc3b230)
516 commit e230fabf29c4660594d19027af49810e57b82e35
517 Author: Yehuda Sadeh <yehuda@redhat.com>
518 Date: Tue Sep 30 14:15:47 2014 -0700
520 rgw: swift GET / HEAD object returns X-Timestamp field
523 Backport: giant, firefly, dumpling
524 Swift clients expect X-Timestamp header, dump it.
526 Signed-off-by: Yehuda Sadeh <yehuda@redhat.com>
527 (cherry picked from commit 5b41d80b7fb9ed96c26801fc42c044191bb18d84)
529 commit c24fab3065ba6d81435981b609f2b69c3d98d21d
530 Author: Yehuda Sadeh <yehuda@redhat.com>
531 Date: Fri Jan 30 10:51:52 2015 -0800
533 rgw: use strict_strtoll() for content length
535 instead of strict_strtol().
537 Backport: giant, firefly
540 Reported-by: Axel Dunkel <ad@dunkel.de>
541 Signed-off-by: Yehuda Sadeh <yehuda@redhat.com>
542 (cherry picked from commit 1c25dbafb45caf1b261cfcec15b868a2ba6b5fef)
544 commit 6aef29e31e9c7c7ccf8e95d573700c08218b2b45
545 Author: Yehuda Sadeh <yehuda@redhat.com>
546 Date: Thu Nov 20 10:36:05 2014 -0800
548 rgw-admin: create subuser if needed when creating user
551 Backport: firefly, giant
552 This turned up after fixing #9973. Earlier we also didn't create the
553 subuser in this case, but we didn't really read the subuser info when it
554 was authenticating. Now we do that as required, so we end up failing the
555 authentication. This only applies to cases where a subuser was created
556 using 'user create', and not the 'subuser create' command.
558 Reviewed-by: Sage Weil <sage@redhat.com>
559 Signed-off-by: Yehuda Sadeh <yehuda@redhat.com>
560 (cherry picked from commit 9ba17a321db06d3d76c9295e411c76842194b25c)
562 commit a5cb39cbb6aee869b92ac20975b5c80a01210b63
563 Author: Yehuda Sadeh <yehuda@redhat.com>
564 Date: Fri Dec 12 17:07:30 2014 -0800
566 rgw: use s->bucket_attrs instead of trying to read obj attrs
569 Backport: firefly, giant
571 This is needed, since we can't really read the bucket attrs by trying to
572 read the bucket entry point attrs. We already have the bucket attrs
575 Signed-off-by: Yehuda Sadeh <yehuda@redhat.com>
576 (cherry picked from commit 5cf193c8686196d5235889e68cb5ea8f1fc8e556)
578 commit 425ee8a07bb8ce12eee124b3c374031f644aa32b
579 Author: Yehuda Sadeh <yehuda@redhat.com>
580 Date: Fri Dec 12 17:07:30 2014 -0800
582 rgw: use s->bucket_attrs instead of trying to read obj attrs
585 Backport: firefly, giant
587 This is needed, since we can't really read the bucket attrs by trying to
588 read the bucket entry point attrs. We already have the bucket attrs
591 Signed-off-by: Yehuda Sadeh <yehuda@redhat.com>
592 (cherry picked from commit 5cf193c8686196d5235889e68cb5ea8f1fc8e556)
594 commit 83a0a2e5b52b18a25009faaf09fa4f48af3c52ee
595 Author: William A. Kennington III <william@wkennington.com>
596 Date: Sat Sep 20 22:52:31 2014 -0700
598 osd: Cleanup boost optionals
600 Signed-off-by: William A. Kennington III <william@wkennington.com>
601 (cherry picked from commit a53ead14c113047567177630b4906136a2109b65)
603 commit eb30631ec3d081fd1bc2cdbd4812a334de9e1282
604 Author: Petr Machata <pmachata@redhat.com>
605 Date: Thu Jan 29 10:15:02 2015 -0700
609 Sometime after 1.55, boost introduced a forward declaration of
610 operator<< in optional.hpp. In 1.55 and earlier, when << was used
611 without the _io having been included, what got dumped was an implicit
614 http://tracker.ceph.com/issues/10688 Refs: #10688
615 Signed-off-by: Ken Dreyer <kdreyer@redhat.com>
616 (cherry picked from commit 85717394c33137eb703a7b88608ec9cf3287f67a)
619 src/include/encoding.h
622 commit 1ccd73a16e1829b5519ec5b83b2554af173ad052
623 Author: Yehuda Sadeh <yehuda@redhat.com>
624 Date: Thu Oct 23 17:39:42 2014 -0700
626 rgw: send http status reason explicitly in fastcgi
628 There are issues in certain versions of apache 2.4, where the reason is
629 not sent back. Instead, just provide the reason explicitly.
631 Backport: firefly, giant
633 Signed-off-by: Yehuda Sadeh <yehuda@redhat.com>
634 (cherry picked from commit a9dd4af401328e8f9071dee52470a0685ceb296b)
636 commit 3bf42af2e932a473b19cb54637e8543a666a4a28
637 Author: Yehuda Sadeh <yehuda@redhat.com>
638 Date: Thu Jan 15 16:31:22 2015 -0800
640 rgw: fix partial GET in swift
643 backport: firefly, giant
645 Don't set the ret code to reflect partial download, just set the
646 response status when needed.
648 Signed-off-by: Yehuda Sadeh <yehuda@redhat.com>
649 (cherry picked from commit 7e1553cedff90fa0fefded65cde87ad068eb5f0c)
651 commit aa038684dce1964d5d23802d23f2bd772458ea11
652 Author: Sage Weil <sage@redhat.com>
653 Date: Mon Dec 15 17:04:32 2014 -0800
655 osd: handle no-op write with snapshot case
657 If we have a transaction that does something to the object but it !exists
658 both before and after, we will continue through the write path. If the
659 snapdir object already exists, and we try to create it again, we will
660 leak a snapdir obc and lock and later crash on an assert when the obc
663 0> 2014-12-06 01:49:51.750163 7f08d6ade700 -1 osd/osd_types.h: In function 'ObjectContext::~ObjectContext()' thread 7f08d6ade700 time 2014-12-06 01:49:51.605411
664 osd/osd_types.h: 2944: FAILED assert(rwstate.empty())
666 Fix is to not recreated the snapdir if it already exists.
669 Signed-off-by: Sage Weil <sage@redhat.com>
670 (cherry picked from commit 02fae9fc54c10b5a932102bac43f32199d4cb612)
672 commit e045ad4a39076547209ac1dc298df5ebffb76669
673 Merge: a463b92 9f865fa
674 Author: Gregory Farnum <greg@gregs42.com>
675 Date: Tue Jan 27 09:40:16 2015 -0800
677 Merge pull request #3502 from ceph/wip-10382-giant
679 [giant backport] mds: handle heartbeat_reset during shutdown
681 Reviewed-by: Greg Farnum <gfarnum@redhat.com>
683 commit 9f865fae095a1fe8a26acb50667f1d774d6020b6
684 Author: John Spray <john.spray@redhat.com>
685 Date: Wed Jan 14 10:35:53 2015 +0000
687 mds: handle heartbeat_reset during shutdown
689 Because any thread might grab mds_lock and call heartbeat_reset
690 immediately after a call to suicide() completes, this needs
691 to be handled as a special case where we tolerate MDS::hb having
692 already been destroyed.
695 Signed-off-by: John Spray <john.spray@redhat.com>
697 commit a463b92e475cd1f4cdb963e402033ebc9d37dbdc
698 Author: Sage Weil <sage@redhat.com>
699 Date: Mon Jan 19 18:28:20 2015 -0800
701 ceph_test_rados_api_misc: do not assert rbd feature match
703 This test fails on upgrades when we (or the server) have new
704 features. Make it less fragile.
707 Signed-off-by: Sage Weil <sage@redhat.com>
708 (cherry picked from commit 9147c62989871cea8b3a85b02c53017825efb55b)
710 commit 14cdb9bb6d27f2017a3a8e6c1f274b9f40fb7456
711 Author: Jason Dillaman <dillaman@redhat.com>
712 Date: Wed Jan 21 14:55:02 2015 -0500
714 rbd: ensure aio_write buffer isn't invalidated during image import
716 The buffer provided to aio_write shouldn't be invalidated until
717 after aio_write has indicated that the operation has completed.
721 Signed-off-by: Jason Dillaman <dillaman@redhat.com>
722 Reviewed-by: Josh Durgin <jdurgin@redhat.com>
723 (cherry picked from commit 4d3b49e9d62bc1040356ca3ebe7f90c181734eb6)
725 commit 081f49b47ca8d7583211f546ab5699b14f773bfc
726 Author: Jason Dillaman <dillaman@redhat.com>
727 Date: Mon Dec 15 10:53:53 2014 -0500
729 librbd: complete all pending aio ops prior to closing image
731 It was possible for an image to be closed while aio operations
732 were still outstanding. Now all aio operations are tracked and
733 completed before the image is closed.
736 Backport: giant, firefly, dumpling
737 Signed-off-by: Jason Dillaman <dillaman@redhat.com>
739 commit 436923c68b77c900b7774fbef918c0d6e1614a36
740 Author: Jason Dillaman <dillaman@redhat.com>
741 Date: Mon Jan 19 10:28:56 2015 -0500
743 librbd: gracefully handle deleted/renamed pools
745 snap_unprotect and list_children both attempt to scan all
746 pools. If a pool is deleted or renamed during the scan,
747 the methods would previously return -ENOENT. Both methods
748 have been modified to more gracefully handle this condition.
751 Backport: giant, firefly
752 Signed-off-by: Jason Dillaman <dillaman@redhat.com>
754 commit 4c8a5cedcb7942e1e01ab4cedfbf03e4c56cc1e4
755 Author: Yehuda Sadeh <yehuda@redhat.com>
756 Date: Fri Dec 12 05:24:01 2014 -0800
758 rgw: change multipart upload id magic
761 Backport: firefly, giant
763 Some clients can't sign requests correctly with the original magic
766 Reported-by: Georgios Dimitrakakis <giorgis@acmac.uoc.gr>
767 Signed-off-by: Yehuda Sadeh <yehuda@redhat.com>
768 (cherry picked from commit 5fc7a0be67a03ed63fcc8408f8d71a31a1841076)
770 commit b10c0d5110547586b2edac53c267391d3d42f974
771 Author: Yehuda Sadeh <yehuda@redhat.com>
772 Date: Thu Dec 11 09:07:10 2014 -0800
774 rgw: url decode http query params correctly
779 This got broken by the fix for #8702. Since we now only url_decode if
780 we're in query, we need to specify that we're in query when decoding
783 Reported-by: Georgios Dimitrakakis <giorgis@acmac.uoc.gr>
784 Signed-off-by: Yehuda Sadeh <yehuda@redhat.com>
785 (cherry picked from commit 21e07eb6abacb085f81b65acd706b46af29ffc03)
787 commit 65bf3b08b572b9b25ad064fb784742e5d6456f06
788 Author: Josh Durgin <jdurgin@redhat.com>
789 Date: Wed Jan 14 15:01:38 2015 -0800
791 qa: ignore duplicates in rados ls
793 These can happen with split or with state changes due to reordering
794 results within the hash range requested. It's easy enough to filter
795 them out at this stage.
797 Backport: giant, firefly
798 Signed-off-by: Josh Durgin <jdurgin@redhat.com>
799 (cherry picked from commit e7cc6117adf653a4915fb7a75fac68f8fa0239ec)
801 commit 1261bf24624f871672002ab0915e23f1c95b0aa5
802 Author: Sage Weil <sage@redhat.com>
803 Date: Tue Oct 14 12:42:40 2014 -0700
805 Revert "Objecter: disable fast dispatch of CEPH_MSG_OSD_OPREPLY messages"
807 This reverts commit 3f23709c474292f9239f77a6cce26309fc86ce29.
809 We have changed mark_down() behavior so that it no longer blocks on
810 fast dispatch. This makes the objecter reply handler safe again.
813 Signed-off-by: Sage Weil <sage@redhat.com>
814 (cherry picked from commit c9f9e72e558521cb90f90538bc27f995f82d76c2)
816 commit 300d4c6ff7e998dba0c67f6dde746dc23d681397
817 Author: Sage Weil <sage@redhat.com>
818 Date: Tue Oct 14 12:41:48 2014 -0700
820 msg/simple: do not stop_and_wait on mark_down
822 We originally blocked in mark_down for fast dispatch threads
823 to complete to avoid various races in the code. Most of these
824 were in the OSD itself, where we were not prepared to get
825 messges on connections that had no attached session. Since
826 then, the OSD checks have been cleaned up to handle this.
827 There were other races we were worried about too, but the
828 details have been lost in the depths of time.
830 Instead, take the other route: make mark_down never block on
831 dispatch. This lets us remove the special case that
832 was added in order to cope with fast dispatch calling
835 Now, the only stop_and_wait() user is the shutdown sequence.
837 Signed-off-by: Sage Weil <sage@redhat.com>
838 (cherry picked from commit 00907e032011b9d2acd16ea588555cf379830814)
840 commit c3335c7aa6c1e6c3f1879c0cd3cd2f13091221be
841 Author: Sage Weil <sage@redhat.com>
842 Date: Fri Oct 31 16:25:09 2014 -0700
844 msg/Pipe: inject delay in stop_and_wait
846 Inject a delay in stop_and_wait. This will mostly affect the connection
847 race Pipe takeover code which currently calls stop_and_wait while holding
848 the msgr->lock. This should make it easier for a racing fast_dispatch
849 method to get stuck on a call that (indirectly) needs the msgr lock.
852 Signed-off-by: Sage Weil <sage@redhat.com>
853 (cherry picked from commit 2fe5c4c305218fdb1771857e4e0ef7c98a8d0fb6)
855 commit 1dbe8f5a6f7bf2b7c86d24f27d569d71e0076ee9
856 Author: Greg Farnum <gfarnum@redhat.com>
857 Date: Tue Oct 28 16:45:43 2014 -0700
859 SimpleMessenger: Pipe: do not block on takeover while holding global lock
861 We previously were able to cause deadlocks:
862 1) Existing pipe is fast_dispatching
863 2) Replacement incoming pipe is accepted
864 *) blocks on stop_and_wait() of existing Pipe
865 3) External things are blocked on SimpleMessenger::lock() while
866 blocking completion of the fast dispatch.
868 To resolve this, if we detect that an existing Pipe we want to take over is
869 in the process of fast dispatching, we unlock our locks and wait on it to
870 finish. Then we go back to the lookup step and retry.
872 The effect of this should be safe:
873 1) We are not making any changes to the existing Pipe in new ways
874 2) We have not registered the new Pipe anywhere
875 3) We have not sent back any replies based on Messenger state to
880 Signed-off-by: Greg Farnum <gfarnum@redhat.com>
881 (cherry picked from commit 2d6980570af2226fdee0edfcfe5a8e7f60fae615)
883 commit 16c023d8fa5575d4dd138aeee4d4fd9b8f32c0f6
884 Author: Sage Weil <sage@redhat.com>
885 Date: Thu Jan 8 13:34:52 2015 -0800
887 osd: requeue PG when we skip handling a peering event
889 If we don't handle the event, we need to put the PG back into the peering
890 queue or else the event won't get processed until the next event is
891 queued, at which point we'll be processing events with a delay.
893 The queue_null is not necessary (and is a waste of effort) because the
894 event is still in pg->peering_queue and the PG is queued.
896 Note that this only triggers when we exceeed osd_map_max_advance, usually
897 when there is a lot of peering and recovery activity going on. A
898 workaround is to increase that value, but if you exceed osd_map_cache_size
899 you expose yourself to crache thrashing by the peering work queue, which
900 can cause serious problems with heavily degraded clusters and bit lots of
903 Backport: giant, firefly
905 Signed-off-by: Sage Weil <sage@redhat.com>
906 (cherry picked from commit 492ccc900c3358f36b6b14a207beec071eb06707)
908 commit 16c6d0d589d53aad7bb2cd0e104300fb920d5caf
909 Author: Loic Dachary <ldachary@redhat.com>
910 Date: Tue Dec 16 13:31:30 2014 +0100
912 erasure-code: relax cauchy w restrictions
914 A restriction that the w parameter of the cauchy technique is limited to
915 8, 16 or 32 was added incorrectly while refactoring parameter parsing in
916 the jerasure plugin and must be relaxed.
918 http://tracker.ceph.com/issues/10325 Fixes: #10325
920 Signed-off-by: Loic Dachary <ldachary@redhat.com>
921 (cherry picked from commit bb80437f247345502203ad87a7e7bbb5b5602b9a)
923 commit 636b98faa6b1c9fd6de1b8653d1d282577b54684
924 Author: Sage Weil <sage@redhat.com>
925 Date: Sun Nov 23 18:50:51 2014 -0800
927 crush/CrushWrapper: fix create_or_move_item when name exists but item does not
929 We were using item_exists(), which simply checks if we have a name defined
930 for the item. Instead, use _search_item_exists(), which looks for an
931 instance of the item somewhere in the hierarchy. This matches what
932 get_item_weightf() is doing, which ensures we get a non-negative weight
933 that converts properly to floating point.
935 Backport: giant, firefly
937 Reported-by: Pawel Sadowski <ceph@sadziu.pl>
938 Signed-off-by: Sage Weil <sage@redhat.com>
939 (cherry picked from commit 9902383c690dca9ed5ba667800413daa8332157e)
941 commit ced2472664fab06d03de03d7b23325f9319163b7
942 Author: Sage Weil <sage@redhat.com>
943 Date: Fri Nov 21 17:47:56 2014 -0800
945 crush/builder: prevent bucket weight underflow on item removal
947 It is possible to set a bucket weight that is not the sum of the item
948 weights if you manually modify/build the CRUSH map. Protect against any
949 underflow on the bucket weight when removing items.
951 Signed-off-by: Sage Weil <sage@redhat.com>
952 (cherry picked from commit 8c87e9502142d5b4a282b94f929ae776a49be1dc)
954 commit adf8798dabb679110c6815af5d73ab6ff20a1af8
955 Author: Sage Weil <sage@redhat.com>
956 Date: Fri Nov 21 17:37:03 2014 -0800
958 crush/CrushWrapper: fix _search_item_exists
960 Reported-by: Pawel Sadowski <ceph@sadziu.pl>
961 Signed-off-by: Sage Weil <sage@redhat.com>
962 (cherry picked from commit eeadd60714d908a3a033aeb7fd542c511e63122b)
964 commit 6caa4fa42c6eaa76b3b9caf37e4ee09844f017a7
965 Author: Warren Usui <warren.usui@inktank.com>
966 Date: Thu Dec 18 20:00:28 2014 -0800
968 If trusty, use older version of qemu
971 Signed-off-by: Warren Usui <warren.usui@inktank.com>
972 (cherry-picked from 46a1a4cb670d30397979cd89808a2e420cef2c11)
974 commit 44c944e96440bd338d22533779e0650b99115a16
975 Merge: abdbbd6 910ec62
976 Author: Sage Weil <sage@redhat.com>
977 Date: Mon Dec 29 10:55:22 2014 -0800
979 Merge pull request #3266 from ceph/giant-10415
981 libcephfs/test.cc: close fd before umount
983 commit b1e48820785a1d3153fc926ad21355b3927b44e9
984 Author: Loic Dachary <ldachary@redhat.com>
985 Date: Sun Dec 28 10:29:54 2014 +0100
987 erasure-code: update links to jerasure upstream
989 It moved from bitbucket to jerasure.org
991 Signed-off-by: Loic Dachary <ldachary@redhat.com>
992 (cherry picked from commit 8e86f901939f16cc9c8ad7a4108ac4bcf3916d2c)
994 commit 910ec624156d26a1830078161f47328a950a4eee
995 Author: Yan, Zheng <zyan@redhat.com>
996 Date: Tue Dec 23 10:22:00 2014 +0800
998 libcephfs/test.cc: close fd before umount
1001 Signed-off-by: Yan, Zheng <zyan@redhat.com>
1002 (cherry picked from commit d3fb563cee4c4cf08ff4ee01782e52a100462429)
1004 commit abdbbd6e846727385cf0a1412393bc9759dc0244
1005 Author: Warren Usui <warren.usui@inktank.com>
1006 Date: Tue Dec 16 22:01:26 2014 -0800
1008 Remove sepia dependency (use fqdn)
1011 Signed-off-by: Warren Usui <warren.usui@inktank.com>
1012 (cherry picked from commit 19dafe164833705225e168a686696fb4e170aba7)
1014 commit d1257436fdf79bad5fe0719a6be71e2abb2d2462 (refs/remotes/gh/giant-10277)
1015 Author: Yan, Zheng <zyan@redhat.com>
1016 Date: Wed Dec 17 15:59:44 2014 +0800
1018 client: use remount to trim kernel dcache
1020 when remounting a file system, linux kernel trims all unused dentry
1024 Signed-off-by: Yan, Zheng <zyan@redhat.com>
1026 commit 9de9901cacd2ed2c8c5f65a938fb6a996efab4cd
1027 Author: Yan, Zheng <zyan@redhat.com>
1028 Date: Wed Dec 17 15:46:49 2014 +0800
1030 client: cleanup client callback registration
1032 Signed-off-by: Yan, Zheng <zyan@redhat.com>
1034 commit d28c8e0fb924fbf36e4e15e19554ad30da3ff8f2 (refs/remotes/gh/wip-giant-mon-backports)
1035 Author: Sage Weil <sage@redhat.com>
1036 Date: Sun Nov 2 08:50:59 2014 -0800
1038 mon/PGMap and PGMonitor: update last_epoch_clean cache from new osd keys
1040 We were only invalidating the cached value from apply_incremental, which
1041 is no longer called on modern clusters.
1043 Fix this by storing the update epoch in the key as well (it is not part
1046 Backport: giant, firefly, dumpling(?)
1048 Signed-off-by: Sage Weil <sage@redhat.com>
1049 (cherry picked from commit 093c5f0cabeb552b90d944da2c50de48fcf6f564)
1051 commit 7646f239476609c96b6baf94dfd5f727fff49502
1052 Author: Sage Weil <sage@redhat.com>
1053 Date: Sun Nov 2 08:49:48 2014 -0800
1055 mon/PGMap: invalidate cached min_last_epoch_clean from new-style pg keys
1057 We were only invalidating the cache from the legacy apply_incremental(),
1058 which is no longer called on modern clusters.
1061 Signed-off-by: Sage Weil <sage@redhat.com>
1062 (cherry picked from commit 3fb731b722c50672a5a9de0c86a621f5f50f2d06)
1064 commit 6ec14b07940ff64d6a121e21a730f691a1a71546
1065 Merge: 758d9cf 7bbf80f
1066 Author: Gregory Farnum <greg@gregs42.com>
1067 Date: Thu Dec 11 17:03:07 2014 -0800
1069 Merge pull request #3159 from ceph/wip-10229-giant
1071 osdc/Filer: use finisher to execute C_Probe and C_PurgeRange [giant backport]
1073 Reviewed-by: Greg Farnum <gfarnum@redhat.com>
1075 commit 7bbf80ff7388f104cf318dd5ac61ca7d35274694
1076 Author: Yan, Zheng <zyan@redhat.com>
1077 Date: Thu Dec 4 12:18:47 2014 +0800
1079 osdc/Filer: use finisher to execute C_Probe and C_PurgeRange
1081 Currently contexts C_Probe/C_PurgeRange are executed while holding
1082 OSDSession::completion_lock. C_Probe and C_PurgeRange may call
1083 Objecter::stat() and Objecter::remove() respectively, which acquire
1084 Objecter::rwlock. This can cause deadlock because there is intermediate
1085 dependency between Objecter::rwlock and OSDSession::completion_lock:
1087 Objecter::rwlock -> OSDSession::lock -> OSDSession::completion_lock
1089 The fix is exexcute C_Probe/C_PurgeRange in finisher thread.
1092 Signed-off-by: Yan, Zheng <zyan@redhat.com>
1093 (cherry picked from commit d3ee89ace660161df7796affbf9a70f3d0dedce1)
1095 commit 758d9cf30bfc7736cc297ba3b047756f7eb8183e
1096 Merge: a8e5638 994dcbb
1097 Author: Gregory Farnum <greg@gregs42.com>
1098 Date: Thu Dec 11 10:47:38 2014 -0800
1100 Merge pull request #3151 from ceph/wip-10288-giant
1102 mon: fix `fs ls` on peons [giant backport]
1104 Reviewed-by: Greg Farnum <gfarnum@redhat.com>
1106 commit 994dcbbef4bea532aea4143c3ac1372ca14d2aea
1107 Author: John Spray <john.spray@redhat.com>
1108 Date: Thu Dec 11 14:00:57 2014 +0000
1110 mon: fix `fs ls` on peons
1112 This was incorrectly using pending_mdsmap instead
1113 of mdsmap. We didn't notice in test because of
1114 single-mon configurations.
1119 Signed-off-by: John Spray <john.spray@redhat.com>
1120 (cherry picked from commit 5559e6aea9e9374ecdac0351777dfd6f5f5d1e67)
1122 commit 319f9c9352bfd1b95bd685500922e6cee2199b34 (refs/remotes/gh/wip-8797-giant)
1123 Author: Dan Mick <dan.mick@redhat.com>
1124 Date: Wed Dec 10 13:19:53 2014 -0800
1126 Call Rados.shutdown() explicitly before exit
1128 This is mostly a demonstration of good behavior, as the resources will
1129 be reclaimed on exit anyway.
1131 Signed-off-by: Dan Mick <dan.mick@redhat.com>
1132 (cherry picked from commit b038e8fbf9103cc42a4cde734b3ee601af6019ea)
1134 commit ed8c9af3376aeb6f245cbab694fdbc0ce95634a8
1135 Author: Dan Mick <dan.mick@redhat.com>
1136 Date: Wed Dec 10 13:19:16 2014 -0800
1138 rados.py: remove Rados.__del__(); it just causes problems
1140 Recent versions of Python contain a change to thread shutdown that
1141 causes ceph to hang on exit; see http://bugs.python.org/issue21963.
1142 As it turns out, this is relatively easy to avoid by not spawning
1143 threads on exit, as Rados.__del__() will certainly do by calling
1144 shutdown(); I suspect, but haven't proven, that the problem is
1145 that shutdown() tries to start() a threading.Thread() that never
1146 makes it all the way back to signal start().
1148 Also add a PendingReleaseNote and extra doc comments to clarify.
1151 Signed-off-by: Dan Mick <dan.mick@redhat.com>
1152 (cherry picked from commit 5ba9b8f21f8010c59dd84a0ef2acfec99e4b048f)
1157 commit a8e56380f08cd5940def4cc47cadba699a8ba45d
1158 Merge: 247a6fa e7faed5
1159 Author: Samuel Just <sam.just@inktank.com>
1160 Date: Mon Dec 8 13:19:20 2014 -0800
1162 Merge pull request #3010 from dachary/wip-10018-primary-erasure-code-hinfo-giant
1164 osd: deep scrub must not abort if hinfo is missing (giant)
1166 Reviewed-by: Samuel Just <sjust@redhat.com>
1168 commit 247a6fac54854e92a7df0e651e248a262d3efa05
1169 Merge: 3372060 309fd5f
1170 Author: Gregory Farnum <greg@gregs42.com>
1171 Date: Mon Dec 8 12:36:48 2014 -0800
1173 Merge pull request #3110 from ceph/giant-10263
1175 mds: store backtrace for straydir
1177 Reviewed-by: Greg Farnum <gfarnum@redhat.com>
1179 commit 309fd5f56ef5ea76ffd525fdde6e6fbbc9ef6ef1
1180 Author: Yan, Zheng <zyan@redhat.com>
1181 Date: Fri Nov 7 11:38:37 2014 +0800
1183 mds: store backtrace for straydir
1185 Backport: giant, firefly, emperor, dumpling
1186 Signed-off-by: Yan, Zheng <zyan@redhat.com>
1187 (cherry picked from commit 0d89db5d3e5ae5d552d4058a88a4e186748ab1d2)
1189 commit 3372060894a1da0adef6d36380a131902ca05c5f
1190 Merge: 1f00420 bff6747
1191 Author: Sage Weil <sage@redhat.com>
1192 Date: Sat Dec 6 11:06:20 2014 -0800
1194 Merge pull request #3088 from dachary/wip-10063-hobject-shard-giant
1196 common: do not omit shard when ghobject NO_GEN is set (giant)
1198 commit 1f004209434570337a3f90d7f89741f80dcc7075
1199 Merge: 3b65226 1ec557c
1200 Author: Sage Weil <sage@redhat.com>
1201 Date: Fri Dec 5 17:33:12 2014 -0800
1203 Merge pull request #3095 from dachary/wip-9785-dmcrypt-keys-permissions-giant
1205 ceph-disk: dmcrypt file permissions (giant)
1207 commit 3b65226df806958f6a2f24df6099ee3a86d2a71f
1208 Merge: 691f011 36c7484
1209 Author: Sage Weil <sage@redhat.com>
1210 Date: Fri Dec 5 17:30:31 2014 -0800
1212 Merge pull request #3006 from dachary/wip-9420-erasure-code-non-regression-giant
1214 erasure-code: store and compare encoded contents (giant)
1216 commit 1ec557c0eab94cb898ad3f5448482bd7afc53e09
1217 Author: Loic Dachary <ldachary@redhat.com>
1218 Date: Thu Dec 4 22:21:32 2014 +0100
1220 ceph-disk: dmcrypt file permissions
1222 The directory in which key files are stored for dmcrypt must be 700 and
1225 http://tracker.ceph.com/issues/9785 Fixes: #9785
1227 Signed-off-by: Loic Dachary <ldachary@redhat.com>
1228 (cherry picked from commit 58682d1776ab1fd4daddd887d921ca9cc312bf50)
1230 commit 691f0118ecd051d5f3f61fc696280e3c482de3de
1231 Merge: 81295c5 dabf6f5
1232 Author: Sage Weil <sage@redhat.com>
1233 Date: Fri Dec 5 09:03:54 2014 -0800
1235 Merge pull request #3085 from dachary/wip-10125-radosgw-init-giant
1237 rgw: run radosgw as apache with systemd (giant)
1239 commit bff67475c775914237604ed3374c8ccfe74d0ffd
1240 Author: Loic Dachary <ldachary@redhat.com>
1241 Date: Fri Nov 14 01:16:10 2014 +0100
1243 common: do not omit shard when ghobject NO_GEN is set
1245 Do not silence the display of shard_id when generation is NO_GEN.
1246 Erasure coded objects JSON representation used by ceph_objectstore_tool
1247 need the shard_id to find the file containing the chunk.
1249 Minimal testing is added to ceph_objectstore_tool.py
1251 http://tracker.ceph.com/issues/10063 Fixes: #10063
1253 Signed-off-by: Loic Dachary <ldachary@redhat.com>
1254 (cherry picked from commit dcf09aed121f566221f539106d10283a09f15cf5)
1256 commit dabf6f5f43b53a588bd9fa0cc5aa617ae8128735
1257 Author: Loic Dachary <ldachary@redhat.com>
1258 Date: Tue Dec 2 18:10:48 2014 +0100
1260 rgw: run radosgw as apache with systemd
1264 http://tracker.ceph.com/issues/10125 Fixes: #10125
1266 Signed-off-by: Loic Dachary <loic@dachary.org>
1267 (cherry picked from commit 7b621f4abf63456272dec3449aa108c89504a7a5)
1270 src/init-radosgw.sysv
1272 commit 81295c5ad2befced2e308c1cfb4e036cd5a825a9
1273 Merge: 8046359 3ff94ed
1274 Author: Josh Durgin <jdurgin@redhat.com>
1275 Date: Thu Dec 4 11:32:01 2014 -0800
1277 Merge pull request #3077 from ceph/wip-10030-giant
1279 librbd: don't close an already closed parent image upon failure
1281 Reviewed-by: Josh Durgin <jdurgin@redhat.com>
1283 commit ad04a677cefc1f0a02fbff0c68409fda6874fdc7
1284 Author: Loic Dachary <ldachary@redhat.com>
1285 Date: Tue Dec 2 00:59:08 2014 +0100
1287 common: add bufferlist::rebuild_aligned_size_and_memory
1289 The function bufferlist::rebuild_aligned checks memory and size
1290 alignment with the same variable. It is however useful to separate
1291 memory alignment constraints from size alignment constraints. For
1292 instance rebuild_aligned could be called to allocate an erasure coded
1293 buffer where each 2048 bytes chunk needs to start on a memory address
1294 aligned on 32 bytes.
1296 Signed-off-by: Loic Dachary <loic@dachary.org>
1297 (cherry picked from commit 9ade88e8dacc9b96c042bb668f4452447139a544)
1299 commit cc469b238f42ce989d0efa49154b95612e3d4111
1300 Author: Loic Dachary <ldachary@redhat.com>
1301 Date: Tue Dec 2 01:07:34 2014 +0100
1303 erasure-code: enforce chunk size alignment
1305 Let say the ErasureCode::encode function is given a 4096 bytes
1306 bufferlist made of a 1249 bytes bufferptr followed by a 2847 bytes
1307 bufferptr, both properly starting on SIMD_ALIGN address. As a result the
1308 second 2048 had to be reallocated when bufferlist::substr_of gets the
1309 second 2048 buffer, the address starts at 799 bytes after the beginning
1310 of the 2847 buffer ptr and is not SIMD_ALIGN'ed.
1312 The ErasureCode::encode must enforce a size alignment based on the chunk
1313 size in addition to the memory alignment required by SIMD operations,
1314 using the bufferlist::rebuild_aligned_size_and_memory function instead of
1315 bufferlist::rebuild_aligned.
1317 http://tracker.ceph.com/issues/10211 Fixes: #10211
1319 Signed-off-by: Loic Dachary <loic@dachary.org>
1320 (cherry picked from commit 4e955f41297283798236c505c3d21bdcabb5caa0)
1322 commit 5205d4dacf7ebe2e42d2294bc30cb27f226c8d22
1323 Author: Loic Dachary <ldachary@redhat.com>
1324 Date: Tue Dec 2 02:04:14 2014 +0100
1326 common: allow size alignment that is not a power of two
1328 Do not assume the alignment is a power of two in the is_n_align_sized()
1329 predicate. When used in the context of erasure code it is common
1330 for chunks to not be powers of two.
1332 Signed-off-by: Loic Dachary <loic@dachary.org>
1333 (cherry picked from commit 73ad2d63d479b09037d50246106bbd4075fbce80)
1335 commit 80463596919d28f58010d16ad017b3c5ae6e558c
1336 Merge: 26e8cf1 3dc6298
1337 Author: Sage Weil <sage@redhat.com>
1338 Date: Wed Dec 3 23:02:43 2014 -0800
1340 Merge pull request #3062 from ceph/wip-10123-giant
1342 librbd: protect list_children from invalid child pool IoCtxs
1344 Reviewed-by: Sage Weil <sage@redhat.com>
1346 commit 26e8cf174b8e76b4282ce9d9c1af6ff12f5565a9
1347 Merge: aac7946 7cd8c3f
1348 Author: Gregory Farnum <greg@gregs42.com>
1349 Date: Wed Dec 3 06:44:56 2014 -0800
1351 Merge pull request #3055 from ceph/wip-10135-giant
1353 mon: OSDMonitor: allow adding tiers to FS pools
1355 Reviewed-by: Greg Farnum <gfarnum@redhat.com>
1357 commit 7cd8c3f8a5afa9481b6f6a78d5fb8c04784ef4ca (refs/remotes/gh/wip-10135-giant)
1358 Author: John Spray <john.spray@redhat.com>
1359 Date: Tue Nov 25 16:54:42 2014 +0000
1361 mon: OSDMonitor: allow adding tiers to FS pools
1363 This was an overly-strict check. In fact it is perfectly
1364 fine to set an overlay on a pool that is already in use
1365 as a filesystem data or metadata pool.
1369 Signed-off-by: John Spray <john.spray@redhat.com>
1370 (cherry picked from commit 17b5fc9a40440e76dd1fa64f7fc19577ae3b58ce)
1372 commit 3ff94ed73ff27af2c8ea215ab693d815e285a27f
1373 Author: Jason Dillaman <dillaman@redhat.com>
1374 Date: Thu Nov 6 05:01:38 2014 -0500
1376 librbd: don't close an already closed parent image upon failure
1378 If librbd is not able to open a child's parent image, it will
1379 incorrectly close the parent image twice, resulting in a crash.
1382 Backport: firefly, giant
1383 Signed-off-by: Jason Dillaman <dillaman@redhat.com>
1384 (cherry picked from commit 61ebfebd59b61ffdc203dfeca01ee1a02315133e)
1386 commit aac794616580ed0bb00608c5867199b12d4e1d94
1387 Merge: 65f6814 c8b46d6
1388 Author: John Spray <jcspray@gmail.com>
1389 Date: Tue Dec 2 11:35:59 2014 +0000
1391 Merge pull request #2990 from ceph/wip-10151-giant
1393 mon: fix MDS health status from peons
1395 Reviewed-by: Greg Farnum <gfarnum@redhat.com>
1397 commit 3dc629822adeee961d78208b46b9bd7ef1200890
1398 Author: Jason Dillaman <dillaman@redhat.com>
1399 Date: Mon Nov 17 21:49:26 2014 -0500
1401 librbd: protect list_children from invalid child pool IoCtxs
1403 While listing child images, don't ignore error codes returned
1404 from librados when creating an IoCtx. This will prevent seg
1405 faults from occurring when an invalid IoCtx is used.
1408 Backport: giant, firefly, dumpling
1409 Signed-off-by: Jason Dillaman <dillaman@redhat.com>
1410 (cherry picked from commit 0d350b6817d7905908a4e432cd359ca1d36bab50)
1412 commit 65f6814847fe8644f5d77a9021fbf13043b76dbe
1413 Merge: 28e2708 9158326
1414 Author: Gregory Farnum <greg@gregs42.com>
1415 Date: Mon Dec 1 17:59:19 2014 -0800
1417 Merge pull request #3047 from ceph/wip-10011-giant
1419 osdc: fix Journaler write error handling [giant backport]
1421 Reviewed-by: Greg Farnum <gfarnum@redhat.com>
1423 commit 9158326eeb69312283a6e8174352f36ea30d0cbf
1424 Author: John Spray <john.spray@redhat.com>
1425 Date: Thu Nov 6 11:46:29 2014 +0000
1427 osdc: fix Journaler write error handling
1429 Since we started wrapping the write error
1430 handler in a finisher, multiple calls to
1431 handle_write_error would hit the assert()
1432 on the second call before the actual
1433 handler had been called (at the other end
1434 of the finisher) from the first call.
1436 The symptom was that the MDS was intermittently
1437 failing to respawn on blacklist, seen in #10011.
1439 Signed-off-by: John Spray <john.spray@redhat.com>
1440 (cherry picked from commit 762eda88a18ba707bd5410f38e21e95c4a6b3a46)
1442 commit 28e27080e25f95851039a0cc4e1c1d06b2cd597d
1443 Merge: 37ffccb cb1d681
1444 Author: Sage Weil <sage@redhat.com>
1445 Date: Tue Nov 25 21:18:59 2014 -0800
1447 Merge pull request #3005 from dachary/wip-9665-ceph-disk-partprobe-giant
1449 ceph disk zap must call partprobe
1451 commit e7faed5d903cf7681d77a6af53cf8137eeb2fc69
1452 Author: Loic Dachary <loic-201408@dachary.org>
1453 Date: Thu Nov 6 17:11:20 2014 +0100
1455 osd: deep scrub must not abort if hinfo is missing
1457 Instead it should set read_error.
1459 http://tracker.ceph.com/issues/10018 Fixes: #10018
1461 Signed-off-by: Loic Dachary <loic-201408@dachary.org>
1462 (cherry picked from commit 9d84d2e8309d26e39ca849a75166d2d7f2dec9ea)
1464 commit 36c7484c18fd072ba2f7b176403414dd32fbe92b
1465 Author: Loic Dachary <loic-201408@dachary.org>
1466 Date: Thu Sep 25 14:46:07 2014 +0200
1468 erasure-code: erasure_code_benchmark exhaustive erasure exploration
1470 Add the --erasure-generation exhaustive flag to try all combinations of
1471 erasures, not just one at random.
1473 Signed-off-by: Loic Dachary <loic-201408@dachary.org>
1474 (cherry picked from commit 2d7adb23bc52e7c0753f4571fecd8eefa209ef02)
1477 src/test/erasure-code/ceph_erasure_code_benchmark.h
1479 commit 32daa9b0f4d39f8a49512b18d5c19437aca5fec6
1480 Author: Loic Dachary <loic-201408@dachary.org>
1481 Date: Mon Sep 29 11:17:13 2014 +0200
1483 erasure-code: add erasure_code_benchmark --verbose
1485 Signed-off-by: Loic Dachary <loic-201408@dachary.org>
1486 (cherry picked from commit 3ff2816b3eecfb7277295583387549dac5429628)
1489 src/test/erasure-code/ceph_erasure_code_benchmark.cc
1490 src/test/erasure-code/ceph_erasure_code_benchmark.h
1492 commit da9a7f07787d7f8c20b0c3e7a53fcaf95ed7ca20
1493 Author: Loic Dachary <loic-201408@dachary.org>
1494 Date: Tue Sep 23 14:37:57 2014 +0200
1496 erasure_code: implement ceph_erasure_code to assert the existence of a plugin
1498 This is handy when scripting in the context of teuthology and only
1499 conditionally run tests for the isa plugin, for instance.
1501 Signed-off-by: Loic Dachary <loic-201408@dachary.org>
1502 (cherry picked from commit efe121d9f2028c312eef2650d32ccf0cbc828edb)
1504 commit c855f3958fb8c10bd824075c1739f40799f6d74b
1505 Author: Loic Dachary <loic-201408@dachary.org>
1506 Date: Tue Sep 23 14:36:08 2014 +0200
1508 erasure-code: ceph_erasure_code does not need to avoid dlclose
1510 The only reason for not dlclosing plugins at exit is for callgrind but
1511 ceph_erasure_code has no workload that would require callgrind.
1513 Signed-off-by: Loic Dachary <loic-201408@dachary.org>
1514 (cherry picked from commit 49613cb2aab6e73e3ea50fa164735b55e80121cd)
1516 commit ba8ceb1f067e0f9f6419358435ed0008b61fa438
1517 Author: Loic Dachary <loic-201408@dachary.org>
1518 Date: Tue Sep 23 11:38:09 2014 +0200
1520 erasure-code: add corpus verification to make check
1522 Signed-off-by: Loic Dachary <loic-201408@dachary.org>
1523 (cherry picked from commit 6fdbdff2ad1b55d4a37dcb95cfbb06c4454cdaf2)
1525 commit ca4c2702139cc7fd8f2e3fa2ee5cda4094ecad79
1526 Author: Loic Dachary <loic-201408@dachary.org>
1527 Date: Sat Sep 13 12:58:27 2014 +0200
1529 erasure-code: Makefile.am cosmetics
1531 Cluster benchmark related lines together.
1533 Signed-off-by: Loic Dachary <loic-201408@dachary.org>
1534 (cherry picked from commit 10c88c8f27080a8e25f128b7065cee5c2f68e91b)
1536 commit 208a5ee1c9975adaa8b09b1bf541aff0d8551c63
1537 Author: Loic Dachary <loic-201408@dachary.org>
1538 Date: Sat Sep 13 12:55:26 2014 +0200
1540 erasure-code: s/alignement/alignment/ typos in jerasure
1542 The jerasure-per-chunk-alignment prameter was mispelled and while
1543 useable that would lead to confusion.
1545 Signed-off-by: Loic Dachary <loic-201408@dachary.org>
1546 (cherry picked from commit 2c84d0b1db57d918840e669a17bbd8c5ddca9747)
1548 commit 1def82d530965bd0441e4f7f6aa032666984f17d
1549 Author: Loic Dachary <loic-201408@dachary.org>
1550 Date: Sat Sep 13 13:36:09 2014 +0200
1552 erasure-code: workunit to check for encoding regression
1554 Clone the archive of encoded objects and decode all archived objects, up
1555 to and including the current ceph version.
1557 http://tracker.ceph.com/issues/9420 Refs: #9420
1559 Signed-off-by: Loic Dachary <loic-201408@dachary.org>
1560 (cherry picked from commit 7638b15f23976c3265cf766e16cf93af1a7e0091)
1562 commit 1b7fc7e49e6edf0d0f7d1d6d9f9447c42067d8b8
1563 Author: Loic Dachary <loic-201408@dachary.org>
1564 Date: Sat Sep 13 10:16:31 2014 +0200
1566 erasure-code: store and compare encoded contents
1568 Introduce ceph_erasure_code_non_regression to check and compare how an
1569 erasure code plugin encodes and decodes content with a given set of
1570 parameters. For instance:
1572 ./ceph_erasure_code_non_regression \
1574 --parameter technique=reed_sol_van \
1577 --stripe-width 3181 \
1581 Will create an encoded object (--create) and store it into a directory
1582 along with the chunks, one chunk per file. The directory name is derived
1583 from the parameters. The content of the object is a random pattern of 31
1584 bytes repeated to fill the object size specified with --stripe-width.
1586 The check function (--check) reads the object back from the file,
1587 encodes it and compares the result with the content of the chunks read
1588 from the files. It also attempts recover from one or two erasures.
1590 Chunks encoded by a given version of Ceph are expected to be encoded
1591 exactly in the same way by all Ceph versions going forward.
1593 http://tracker.ceph.com/issues/9420 Refs: #9420
1595 Signed-off-by: Loic Dachary <loic-201408@dachary.org>
1596 (cherry picked from commit f5901303dbf50e9d08f2f1e510a1936a20037909)
1598 commit cb1d68113477cf9c2028a65372d2d4a3e6a8bdc1
1599 Author: Loic Dachary <loic-201408@dachary.org>
1600 Date: Thu Oct 9 18:52:17 2014 +0200
1602 ceph-disk: run partprobe after zap
1604 Not running partprobe after zapping a device can lead to the following:
1606 * ceph-disk prepare /dev/loop2
1607 * links are created in /dev/disk/by-partuuid
1608 * ceph-disk zap /dev/loop2
1609 * links are not removed from /dev/disk/by-partuuid
1610 * ceph-disk prepare /dev/loop2
1611 * some links are not created in /dev/disk/by-partuuid
1613 This is assuming there is a bug in the way udev events are handled by
1614 the operating system.
1616 http://tracker.ceph.com/issues/9665 Fixes: #9665
1618 Signed-off-by: Loic Dachary <loic-201408@dachary.org>
1619 (cherry picked from commit fed3b06c47a5ef22cb3514c7647544120086d1e7)
1621 commit d9c04b880d8bf867aa454132117119be5bd550ad
1622 Author: Loic Dachary <loic-201408@dachary.org>
1623 Date: Fri Oct 10 10:26:31 2014 +0200
1625 ceph-disk: use update_partition in prepare_dev and main_prepare
1627 In the case of prepare_dev the partx alternative was missing and is not
1628 added because update_partition does it.
1630 http://tracker.ceph.com/issues/9721 Fixes: #9721
1632 Signed-off-by: Loic Dachary <loic-201408@dachary.org>
1633 (cherry picked from commit 23e71b1ee816c0ec8bd65891998657c46e364fbe)
1635 commit 5c9cd3c2a292ae540fd0a487defaf4d712e41e62
1636 Author: Loic Dachary <loic-201408@dachary.org>
1637 Date: Fri Oct 10 10:23:34 2014 +0200
1639 ceph-disk: encapsulate partprobe / partx calls
1641 Add the update_partition function to reduce code duplication.
1642 The action is made an argument although it always is -a because it will
1643 be -d when deleting a partition.
1645 Use the update_partition function in prepare_journal_dev
1647 Signed-off-by: Loic Dachary <loic-201408@dachary.org>
1648 (cherry picked from commit 922a15ea6865ef915bbdec2597433da6792c1cb2)
1650 commit c8b46d68c71f66d4abbda1230741cc4c7284193b
1651 Author: John Spray <john.spray@redhat.com>
1652 Date: Mon Nov 24 11:00:25 2014 +0000
1654 mon: fix MDS health status from peons
1656 The health data was there, but we were attempting
1657 to enumerate MDS GIDs from pending_mdsmap (empty on
1658 peons) instead of mdsmap (populated from paxos updates)
1663 Signed-off-by: John Spray <john.spray@redhat.com>
1664 (cherry picked from commit 0c33930e3a90f3873b7c7b18ff70dec2894fce29)
1667 src/mon/MDSMonitor.cc
1669 commit 37ffccbd57e7e441e0eb1499e5c173aa9c375d35
1670 Merge: b13a56a 65c5657
1671 Author: Josh Durgin <jdurgin@redhat.com>
1672 Date: Thu Nov 20 13:13:33 2014 -0800
1674 Merge pull request #2975 from ceph/wip-9936-giant
1676 rbd: Fix the rbd export when image size more than 2G
1678 Reviewed-by: Josh Durgin <jdurgin@redhat.com>
1680 commit b13a56afe99c091600392a2fc15befa9438d59c9
1681 Merge: 828c1a2 46bd344
1682 Author: Loic Dachary <loic-201408@dachary.org>
1683 Date: Wed Nov 19 02:40:47 2014 +0100
1685 Merge pull request #2963 from ceph/wip-10114-giant
1689 Reviewed-by: Loic Dachary <ldachary@redhat.com>
1691 commit 828c1a2bcd81a49264f8a81ca7f1070169037820
1692 Merge: ccfd241 6cb9a24
1693 Author: David Zafman <david.zafman@inktank.com>
1694 Date: Tue Nov 18 15:48:16 2014 -0800
1696 Merge pull request #2958 from ceph/wip-10128-giant
1698 ceph_objectstore_tool: When exporting to stdout, don't cout messages
1700 Reviewed-by: Loic Dachary <ldachary@redhat.com>
1702 commit 46bd3441b00c22ba78f04617fd77f0231ccc698d
1703 Author: Dan Mick <dan.mick@redhat.com>
1704 Date: Tue Nov 18 15:21:30 2014 -0800
1706 erasure-code isa-l: remove duplicated lines (fix warning)
1708 06a245a added a section def to assembly files; I added it twice to
1709 this file. There's no damage, but a compiler warning (on machines with
1712 Signed-off-by: Dan Mick <dan.mick@redhat.com>
1713 (cherry picked from commit 10f6ef185a9d09e396e94036ec90bfe8a0738ce9)
1715 commit 1bba3887038aade137a808d751561cc02002f4bf
1716 Author: Dan Mick <dan.mick@redhat.com>
1717 Date: Fri Nov 14 17:59:57 2014 -0800
1719 Add annotation to all assembly files to turn off stack-execute bit
1721 See discussion in http://tracker.ceph.com/issues/10114
1723 Building with these changes allows output from readelf like this:
1725 $ readelf -lW src/.libs/librados.so.2 | grep GNU_STACK
1726 GNU_STACK 0x000000 0x0000000000000000 0x0000000000000000 0x000000
1729 (note the absence of 'X' in 'RW')
1732 Signed-off-by: Dan Mick <dan.mick@redhat.com>
1733 (cherry picked from commit 06a245a9845c0c126fb3106b41b2fd2bc4bc4df3)
1735 commit 6cb9a2499cac2645e2cc6903ab29dfd95aac26c7
1736 Author: David Zafman <dzafman@redhat.com>
1737 Date: Mon Nov 17 23:02:50 2014 -0800
1739 ceph_objectstore_tool: When exporting to stdout, don't cout messages
1744 Signed-off-by: David Zafman <dzafman@redhat.com>
1745 (cherry picked from commit 0d5262ac2f69ed3996af76a72894b1722a27b37d)
1747 commit 13bb880b2a04ab354506eb183d2907b9054bf937
1748 Author: Sebastien Ponce <sebastien.ponce@cern.ch>
1749 Date: Tue Nov 18 10:30:36 2014 +0100
1751 Fixed locking issue in the trun method of libradosstriper leading to potential race conditions - Fixes: #10129
1753 Signed-off-by: Sebastien Ponce <sebastien.ponce@cern.ch>
1754 (cherry picked from commit 8613984373de946e1815cc84d50bbd4437a3f7a7)
1756 commit 65c565701eb6851f4ed4d2dbc1c7136dfaad6bcb
1757 Author: Vicente Cheng <freeze.bilsted@gmail.com>
1758 Date: Wed Oct 29 12:21:11 2014 +0800
1760 rbd: Fix the rbd export when image size more than 2G
1762 When using export <image-name> <path> and the size of image is more
1763 than 2G, the previous version about finish() could not handle in
1764 seeking the offset in image and return error.
1766 This is caused by the incorrect variable type. Try to use the correct
1767 variable type to fixed it.
1769 I use another variable which type is uint64_t for confirming seeking
1770 and still use the previous r for return error.
1772 uint64_t is more better than type int for handle lseek64().
1774 Signed-off-by: Vicente Cheng <freeze.bilsted@gmail.com>
1775 (cherry picked from commit 4b87a81c86db06f6fe2bee440c65fc05cd4c23ce)
1777 commit ccfd2414c68afda55bf4cefa2441ea6d53d87cc6
1778 Author: Sage Weil <sage@redhat.com>
1779 Date: Wed Nov 12 17:11:10 2014 -0800
1781 osd/OSD: use OSDMap helper to determine if we are correct op target
1783 Use the new helper. This fixes our behavior for EC pools where targetting
1784 a different shard is not correct, while for replicated pools it may be. In
1785 the EC case, it leaves the op hanging indefinitely in the OpTracker because
1786 the pgid exists but as a different shard.
1789 Signed-off-by: Sage Weil <sage@redhat.com>
1790 (cherry picked from commit 9e05ba086a36ae9a04b347153b685c2b8adac2c3)
1792 commit 963947718a954f63f351ce4034bf97380421ab7c
1793 Author: Sage Weil <sage@redhat.com>
1794 Date: Wed Nov 12 17:04:35 2014 -0800
1796 osd/OSDMap: add osd_is_valid_op_target()
1798 Helper to check whether an osd is a given op target for a pg. This
1799 assumes that for EC we always send ops to the primary, while for
1800 replicated we may target any replica.
1802 Signed-off-by: Sage Weil <sage@redhat.com>
1803 (cherry picked from commit 89c02637914ac7332e9dbdbfefc2049b2b6c127d)
1805 commit 0cb32c157c6c11b26607521a20c6f320c5170516
1806 Author: Josh Durgin <jdurgin@redhat.com>
1807 Date: Tue Nov 11 18:16:02 2014 -0800
1809 qa: allow small allocation diffs for exported rbds
1811 The local filesytem may behave slightly differently. This isn't
1812 foolproof, but seems to be reliable enough on rhel7 rootfs, where
1813 exact comparison was failing.
1816 Signed-off-by: Josh Durgin <jdurgin@redhat.com>
1817 (cherry picked from commit e94d3c11edb9c9cbcf108463fdff8404df79be33)
1819 commit fe705c8fdef2371d3f5b11eb73f87a0cf6ef0f9e
1820 Author: Adam Crume <adamcrume@gmail.com>
1821 Date: Thu Sep 18 16:57:27 2014 -0700
1823 common: Add cctid meta variable
1826 Signed-off-by: Adam Crume <adamcrume@gmail.com>
1827 (cherry picked from commit bb45621cb117131707a85154292a3b3cdd1c662a)
1829 commit 5fc659a0d52094a4c595ca8b33b407ecdefc180a
1830 Merge: b27f5db a6c02a1
1831 Author: Sage Weil <sage@redhat.com>
1832 Date: Tue Nov 11 08:28:19 2014 -0800
1834 Merge pull request #2804 from ceph/wip-9301-giant
1836 mon: backport paxos off-by-one bug (9301) to giant
1838 commit b27f5dba8677ca48c9819980e3c90b76f5f04267
1839 Merge: 97e423f fc5354d
1840 Author: Gregory Farnum <greg@gregs42.com>
1841 Date: Mon Nov 10 22:41:19 2014 -0800
1843 Merge pull request #2887 from ceph/wip-9977-backport
1845 tools: skip up to expire_pos in journal-tool
1847 Reviewed-by: Greg Farnum <gfarnum@redhat.com>
1849 commit 97e423f52155e2902bf265bac0b1b9ed137f8aa0
1850 Author: Yan, Zheng <ukernel@gmail.com>
1851 Date: Thu Sep 11 09:36:44 2014 +0800
1853 client: trim unused inodes before reconnecting to recovering MDS
1855 So the recovering MDS does not need to fetch these ununsed inodes during
1856 cache rejoin. This may reduce MDS recovery time.
1858 Signed-off-by: Yan, Zheng <zyan@redhat.com>
1859 (cherry picked from commit 2bd7ceeff53ad0f49d5825b6e7f378683616dffb)
1861 Reviewed-by: Greg Farnum <gfarnum@redhat.com>
1863 commit 387efc5fe1fb148ec135a6d8585a3b8f8d97dbf8
1864 Author: John Spray <john.spray@redhat.com>
1865 Date: Mon Oct 27 12:02:17 2014 +0000
1867 client: allow xattr caps in inject_release_failure
1869 Because some test environments generate spurious
1870 rmxattr operations, allow the client to release
1871 'X' caps. Allows xattr operations to proceed
1872 while still preventing client releasing other caps.
1875 Signed-off-by: John Spray <john.spray@redhat.com>
1876 (cherry picked from commit 5691c68a0a44eb2cdf0afb3f39a540f5d42a5c0c)
1878 Reviewed-by: Greg Farnum <gfarnum@redhat.com>
1880 commit fc5354dec55248724f8f6b795e3a96882c33b490 (refs/remotes/gh/wip-9977-backport)
1881 Author: John Spray <john.spray@redhat.com>
1882 Date: Mon Nov 3 19:19:45 2014 +0000
1884 tools: skip up to expire_pos in journal-tool
1886 Previously worked for journals starting from an
1887 object boundary (i.e. freshly created filesystems)
1890 Signed-off-by: John Spray <john.spray@redhat.com>
1891 (cherry picked from commit 65c33503c83ff8d88781c5c3ae81d88d84c8b3e4)
1894 src/tools/cephfs/JournalScanner.cc
1896 commit 9680613141b3eef62f35a6728e654efa5f6ba8e8
1897 Merge: fd4363d a5984ba
1898 Author: Gregory Farnum <greg@gregs42.com>
1899 Date: Fri Nov 7 16:26:54 2014 -0800
1901 Merge pull request #2876 from ceph/giant-readdir-fix
1905 commit fd4363d1bd49f73e1b3c22516686c7b7e1745b57
1906 Merge: f66bf31 7166ff8
1907 Author: Gregory Farnum <greg@gregs42.com>
1908 Date: Fri Nov 7 14:10:40 2014 -0800
1910 Merge pull request #2879 from ceph/wip-10025-giant
1912 #10025/giant -- tools: fix MDS journal import
1914 Reviewed-by: Greg Farnum <gfarnum@redhat.com>
1916 commit 7166ff83f6343d31d52a58363e2767434554505c
1917 Author: John Spray <john.spray@redhat.com>
1918 Date: Fri Nov 7 11:34:43 2014 +0000
1920 tools: fix MDS journal import
1922 Previously it only worked on fresh filesystems which
1923 hadn't been trimmed yet, and resulted in an invalid
1924 trimmed_pos when expire_pos wasn't on an object
1929 Signed-off-by: John Spray <john.spray@redhat.com>
1930 (cherry picked from commit fb29e71f9a97c12354045ad2e128156e503be696)
1932 commit a5984ba34cb684dae623df22e338f350c8765ba5
1933 Author: Yan, Zheng <zyan@redhat.com>
1934 Date: Mon Oct 27 13:57:16 2014 -0700
1936 client: fix I_COMPLETE_ORDERED checking
1938 Current code marks a directory inode as complete and ordered when readdir
1939 finishes, but it does not check if the directory was modified in the middle
1940 of readdir. This is wrong, directory inode should not be marked as ordered
1941 if it was modified during readddir
1943 The fix is introduce a new counter to the inode data struct, we increase
1944 the counter each time the directory is modified. When readdir finishes, we
1945 check the counter to decide if the directory should be marked as ordered.
1948 Signed-off-by: Yan, Zheng <zyan@redhat.com>
1949 (cherry picked from commit a4caed8a53d011b214ab516090676641f7c4699d)
1951 commit b5ff4e99c87958211e4b7716b59084fc3417ec17
1952 Author: Yan, Zheng <ukernel@gmail.com>
1953 Date: Tue Sep 9 17:34:46 2014 +0800
1955 client: preserve ordering of readdir result in cache
1957 Preserve ordering of readdir result in a list, so that the result of cached
1958 readdir is consistant with uncached readdir.
1960 As a side effect, this commit also removes the code that removes stale dentries.
1961 This is OK because stale dentries does not have valid lease, they will be
1962 filter out by the shared gen check in Client::_readdir_cache_cb()
1964 Signed-off-by: Yan, Zheng <zyan@redhat.com>
1965 (cherry picked from commit 346c06c1647658768e927a47768a0bc74de17b53)
1967 commit 0671c1127015311d9894f15e2493805f93432910
1968 Author: Yan, Zheng <ukernel@gmail.com>
1969 Date: Tue Sep 9 14:06:06 2014 +0800
1971 client: introduce a new flag indicating if dentries in directory are sorted
1973 When creating a file, Client::insert_dentry_inode() set the dentry's offset
1974 based on directory's max offset. The offset does not reflect the real
1975 postion of the dentry in directory. Later readdir reply from real postion
1976 of the dentry in directory. Later readdir reply from MDS may change the
1977 dentry's position/offset. This inconsistency can cause missing/duplicate
1978 entries in readdir result if readdir is partly satisfied by dcache_readdir().
1980 The fix is introduce a new flag indicating if dentries in directory are
1981 sorted. We use _readdir_cache_cb() to handle readdir only when the flag is
1982 set, clear the flag after creating/deleting/renaming file.
1985 Signed-off-by: Yan, Zheng <zyan@redhat.com>
1986 (cherry picked from commit 600af25493947871c38214aa370e2544a7fea399)
1988 commit f66bf31b6743246fb1c88238cf18101238dee3a4
1989 Author: Greg Farnum <gfarnum@redhat.com>
1990 Date: Thu Nov 6 17:48:01 2014 -0800
1992 qa: use sudo even more when rsyncing /usr
1994 Signed-off-by: Greg Farnum <gfarnum@redhat.com>
1995 (cherry picked from commit 3aa7797741f9cff06053a2f31550fe6929039692)
1997 commit f7ec448d6579f965eec184416a97b47ae27ab47a
1998 Merge: f410d76 f111bc8
1999 Author: Loic Dachary <loic-201408@dachary.org>
2000 Date: Wed Nov 5 08:51:18 2014 +0100
2002 Merge pull request #2858 from ceph/wip-9909
2004 tools: rados put /dev/null should write() and not create()
2006 Reviewed-by: Loic Dachary <loic-201408@dachary.org>
2008 commit f111bc8eac8a521b13340f4a75418d839725e010
2009 Author: Loic Dachary <loic-201408@dachary.org>
2010 Date: Thu Oct 2 09:23:55 2014 +0200
2012 tools: rados put /dev/null should write() and not create()
2014 In the rados.cc special case to handle put an empty objects, use
2015 write_full() instead of create().
2017 A special case was introduced 6843a0b81f10125842c90bc63eccc4fd873b58f2
2018 to create() an object if the rados put file is empty. Prior to this fix
2019 an attempt to rados put an empty file was a noop. The problem with this
2020 fix is that it is not idempotent. rados put an empty file twice would
2021 fail the second time and rados put a file with one byte would succeed as
2024 Signed-off-by: Loic Dachary <loic-201408@dachary.org>
2025 (cherry picked from commit 50e80407f3c2f74d77ba876d01e7313c3544ea4d)
2027 commit f410d764d2e6795389cb320b4436cff3607927bd
2028 Author: Yehuda Sadeh <yehuda@redhat.com>
2029 Date: Thu Oct 9 10:20:27 2014 -0700
2031 rgw: set length for keystone token validation request
2034 Backport: giany, firefly
2035 Need to set content length to this request, as the server might not
2036 handle a chunked request (even though we don't send anything).
2038 Tested-by: Mark Kirkwood <mark.kirkwood@catalyst.net.nz>
2039 Signed-off-by: Yehuda Sadeh <yehuda@redhat.com>
2040 (cherry picked from commit 3dd4ccad7fe97fc16a3ee4130549b48600bc485c)
2042 commit dba7defc623474ad17263c9fccfec60fe7a439f0
2043 Merge: 6a201f8 e0b0441
2044 Author: Sage Weil <sage@redhat.com>
2045 Date: Fri Oct 31 08:35:42 2014 -0700
2047 Merge pull request #2846 from dachary/wip-9752-past-intervals-giant
2049 osd: past_interval display bug on acting
2051 commit e0b04414b92018277a0d3b9d82e72ea7529f4ef5
2052 Author: Loic Dachary <loic-201408@dachary.org>
2053 Date: Fri Oct 31 00:49:21 2014 +0100
2055 osd: past_interval display bug on acting
2057 The acting array was incorrectly including the primary and up_primary.
2059 http://tracker.ceph.com/issues/9752 Fixes: #9752
2061 Signed-off-by: Loic Dachary <loic-201408@dachary.org>
2062 (cherry picked from commit c5f8d6eded52da451fdd1d807bd4700221e4c41c)
2064 commit 6a201f89b1aa6c2197383c29919cdeb4a8353d1b
2065 Merge: ebe1637 905aba2
2066 Author: Yan, Zheng <ukernel@gmail.com>
2067 Date: Thu Oct 30 17:01:12 2014 -0700
2069 Merge pull request #2841 from ceph/giant-9869
2071 Backport "client: cast m->get_client_tid() to compare to 16-bit Inode::flushing_cap_tid"
2073 commit 905aba2f3d847933f98124f3ea8d1d76d644edb4
2074 Author: Greg Farnum <greg@inktank.com>
2075 Date: Wed Oct 22 17:16:31 2014 -0700
2077 client: cast m->get_client_tid() to compare to 16-bit Inode::flushing_cap_tid
2079 m->get_client_tid() is 64 bits (as it should be), but Inode::flushing_cap_tid
2080 is only 16 bits. 16 bits should be plenty to let the cap flush updates
2081 pipeline appropriately, but we need to cast in the proper direction when
2082 comparing these differently-sized versions. So downcast the 64-bit one
2086 Backport: giant, firefly, dumpling
2088 Signed-off-by: Greg Farnum <greg@inktank.com>
2089 (cherry picked from commit a5184cf46a6e867287e24aeb731634828467cd98)
2091 commit ebe16373e966917ca8cb03ebeac974bdff7b7685
2092 Merge: c51c8f9 b704f0d
2093 Author: Sage Weil <sage@redhat.com>
2094 Date: Thu Oct 30 10:05:22 2014 -0700
2096 Merge pull request #2838 from ceph/wip-9945-giant
2098 messages: fix COMPAT_VERSION on MClientSession
2100 Reviewed-by: Sage Weil <sage@redhat.com>
2102 commit b704f0dd888aacb10c32cdb63cdbf9f06296fc18
2103 Author: John Spray <john.spray@redhat.com>
2104 Date: Thu Oct 30 16:43:21 2014 +0000
2106 messages: fix COMPAT_VERSION on MClientSession
2108 This was incorrectly incremented to 2 by omission
2109 of an explicit COMPAT_VERSION value.
2113 Signed-off-by: John Spray <john.spray@redhat.com>
2114 (cherry picked from commit 1eb9bcb1d36014293efc687b4331be8c4d208d8e)
2116 commit a6c02a18d6249ea62cf7a74710c8d0192b6eecaa
2117 Author: Sage Weil <sage@redhat.com>
2118 Date: Thu Sep 18 14:23:36 2014 -0700
2120 mon: re-bootstrap if we get probed by a mon that is way ahead
2122 During bootstrap we verify that our paxos commits overlap with the other
2123 mons we will form a quorum with. If they do not, we do a sync.
2125 However, it is possible we pass those checks, then fail to join a quorum
2126 before the quorum moves ahead in time such that we no longer overlap.
2127 Currently nothing kicks up back into a probing state to discover we need
2128 to sync... we will just keep trying to call or join an election instead.
2130 Fix this by jumping back to bootstrap if we get a probe that is ahead of
2131 us. Only do this from non probe or sync states as these will be common;
2132 it is only the active and electing states that matter (and probably just
2136 Backport: giant, firefly
2137 Signed-off-by: Sage Weil <sage@redhat.com>
2138 (cherry picked from commit c421b55e8e15ef04ca8aeb47f7d090375eaa8573)
2140 commit 92d2a38efd458f9e8f4da228ea1e94df08dc8222
2141 Author: Sage Weil <sage@redhat.com>
2142 Date: Thu Sep 18 14:11:24 2014 -0700
2144 mon/Paxos: fix off-by-one in last_ vs first_committed check
2146 peon last_committed + 1 == leader first_committed is okay. Note that the
2147 other check (where I clean up whitespace) gets this correct.
2149 Fixes: #9301 (partly)
2150 Signed-off-by: Sage Weil <sage@redhat.com>
2151 (cherry picked from commit d81cd7f86695185dce31df76c33c9a02123f0e4a)