1 commit ad85b8bfafea6232d64cb7ba76a8b6e8252fa0c7
2 Author: Gary Lowell <gary.lowell@inktank.com>
3 Date: Thu Oct 3 22:41:31 2013 +0000
7 commit 5cd66d3b4bca92b402c95ab256fbc3f0329c446f
8 Author: Yehuda Sadeh <yehuda@inktank.com>
9 Date: Fri Sep 20 14:04:47 2013 -0700
11 rgw: fix keystone token expiration test
14 The test was inverted, need expiration to be greater than
15 current time in order for token to be valid.
17 Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
18 Reviewed-by: Sage Weil <sage@inktank.com>
20 commit e0203c61a3f45fdd6d3d3ece26fef6152bdc036d
21 Author: David Zafman <david.zafman@inktank.com>
22 Date: Wed Sep 11 16:55:06 2013 -0700
24 osd/OSD.cc: Use MIN() so that we don't exceed osd_recovery_max_active
26 Caused by 944f3b73531af791c90f0f061280160003545c63
32 Signed-off-by: David Zafman <david.zafman@inktank.com>
33 Reviewed-by: Samuel Just <sam.just@inktank.com>
34 (cherry picked from commit 139a714e13aa3c7f42091270b55dde8a17b3c4b8)
40 commit c376708358cedb5561fbb43e9b9e622df3ea7a58
41 Author: Joao Eduardo Luis <joao.luis@inktank.com>
42 Date: Wed Sep 25 22:08:24 2013 +0100
44 mon: OSDMonitor: do not write full_latest during trim
46 On commit 81983bab we patched OSDMonitor::update_from_paxos() such that we
47 write the latest full map version to 'full_latest' each time the latest
48 full map was built from the incremental versions.
50 This change however clashed with OSDMonitor::encode_trim_extra(), which
51 also wrote to 'full_latest' on each trim, writing instead the version of
52 the *oldest* full map. This duality of behaviors could lead the store
53 to an inconsistent state across the monitors (although there's no sign of
54 it actually imposing any issues besides rebuilding already existing full
55 maps on some monitors).
57 We now stop OSDMonitor::encode_trim_extra() from writing to 'full_latest'.
58 This function will still write out the oldest full map it has in the store,
59 but it will no longer write to full_latest, instead leaving it up to
60 OSDMonitor::update_from_paxos() to figure it out -- and it already does.
66 Signed-off-by: Joao Eduardo Luis <joao.luis@inktank.com>
67 Reviewed-by: Sage Weil <sage@inktank.com>
68 (cherry picked from commit bd0f29a2c28cca496ec830eac932477ebf3182ba)
70 commit de40d0b3e35ab0124cd3c4ebfcaa435ab8abfab9
71 Author: Sage Weil <sage@inktank.com>
72 Date: Tue Oct 1 15:53:42 2013 -0700
74 crush: invalidate rmap on create (and thus decode)
76 If we have an existing CrushWrapper object and decode from a bufferlist,
77 reset build_rmaps so that they get rebuilt.
79 Remove the build_rmaps() all in decode that was useless on a redecode
80 (because have_rmaps == true in that case and it did nothing).
83 Backport: dumpling, maybe cuttlefish
84 Signed-off-by: Sage Weil <sage@inktank.com>
85 Reviewed-by: Joao Eduardo Luis <joao.luis@inktank.com>
86 (cherry picked from commit 9b7a2ae329b6a511064dd3d6e549ba61f52cfd21)
88 commit 32f5233288c47d95b87c0a9cab5f9c2ffcf15417
89 Author: Dan Mick <dan.mick@inktank.com>
90 Date: Mon Sep 30 14:58:11 2013 -0700
92 Invoke python with /usr/bin/env python instead of directly
95 Signed-off-by: Dan Mick <dan.mick@inktank.com>
96 (cherry picked from commit b9000b314b9166845ff302d4a827a996775d9a14)
98 commit 66aeca5a9079be398403bbff67bd5bf68c6fb111
99 Author: Sage Weil <sage@inktank.com>
100 Date: Wed Sep 25 10:10:21 2013 -0700
102 qa/workunits/mon/crush_ops.sh: fix test
107 Signed-off-by: Sage Weil <sage@inktank.com>
108 (cherry picked from commit c8cae87e9e08468cc86145e0fd60c05d12826239)
110 commit beb366302a125dd422c4f092b12eb541cb3bc788
111 Author: Sage Weil <sage@inktank.com>
112 Date: Mon Sep 23 09:04:34 2013 -0700
114 Revert "ceph: parse CEPH_ARGS environment variable"
116 This reverts commit 67a95b9880c9bc6e858150352318d68d64ed74ad.
118 We now put CEPH_ARGS in the actual args we parse in python, which are passed
119 to rados piecemeal later. This lets you put things like --id ... in there
120 that need to be parsed before librados is initialized.
121 (cherry picked from commit 97f462be4829f0167ed3d65e6694dfc16f1f3243)
123 commit b475ff9576f145d31c053213c699e13df76d2bcb
124 Author: Benoît Knecht <benoit.knecht@fsfe.org>
125 Date: Mon Sep 23 15:58:42 2013 +0200
127 Add CEPH_ARGS at the end of sys.argv
129 This allows, for instance, to pass a different client name to ceph by
130 exporting CEPH_ARGS="--id client_id".
132 Signed-off-by: Benoît Knecht <benoit.knecht@fsfe.org>
133 Signed-off-by: Sage Weil <sage@inktank.com>
134 (cherry picked from commit 30abe3244c86cbbe1f5b005850c29c9c0eafcad4)
136 commit 94548b4b67cca37366c7d8719209a6d2e7956811
137 Author: Sage Weil <sage@inktank.com>
138 Date: Tue Sep 24 15:26:03 2013 -0700
140 mon/OSDMonitor: fix 'ceph osd crush reweight ...'
142 The adjust method returns a count of adjusted items.
148 Signed-off-by: Sage Weil <sage@inktank.com>
149 Reviewed-by: Dan Mick <dan.mick@inktank.com>
150 (cherry picked from commit 3de32562b55c6ece3a6ed783c36f8b9f21460339)
152 commit 00ff7f5c20e13869d0694379739ba4e61d44b97c
153 Author: Joao Eduardo Luis <joao.luis@inktank.com>
154 Date: Tue Sep 10 00:20:41 2013 +0100
156 qa: workunits: mon: crush_ops: test 'ceph osd crush move'
158 Signed-off-by: Joao Eduardo Luis <joao.luis@inktank.com>
159 (cherry picked from commit 3bc618b7b46496c5110edde0da9cae5d3e68e0e1)
161 commit 0ff5b4a96833681e92cc41f019a569134474f4cf
162 Author: Loic Dachary <loic@dachary.org>
163 Date: Tue Sep 24 19:04:23 2013 +0200
165 osd: change warn_interval_multiplier to uint32_t
167 to prevent overflow in OpTracker::check_ops_in_flight when
168 multiplying warn_interval_multiplier *= 2
170 Backport: cuttlefish, dumpling
172 http://tracker.ceph.com/issues/6370 fixes #6370
174 Signed-off-by: Loic Dachary <loic@dachary.org>
175 (cherry picked from commit 1bce1f009bffd3e28025a08775fec189907a81db)
177 commit fb15040b6cec6221baa550ddfffade823f784c4a
178 Author: David Zafman <david.zafman@inktank.com>
179 Date: Mon Sep 9 13:01:12 2013 -0700
181 crushtool: do not dump core with non-unique bucket IDs
183 Return -EEXIST on duplicate ID
184 BUG FIX: crush_add_bucket() mixes error returns and IDs
185 Add optional argument to return generated ID
189 Signed-off-by: David Zafman <david.zafman@inktank.com>
190 Reviewed-by: Sage Weil <sage@inktank.com>
191 (cherry picked from commit 8c76f3a0f9cf100ea2c941dc2b61c470aa5033d7)
193 commit 410db3f30c6eb54b807908c1f251ad4026e7d446
194 Author: Joao Eduardo Luis <jecluis@gmail.com>
195 Date: Fri Sep 20 17:06:30 2013 +0100
197 qa: workunits: cephtool: check if 'heap' commands are parseable
199 Signed-off-by: Joao Eduardo Luis <jecluis@gmail.com>
200 (cherry picked from commit b1eeaddd5f214c1b0883b44fc8cae07c649be7c4)
202 commit 062060a38bb26ff260cc51accc534413d726de49
203 Author: Joao Eduardo Luis <jecluis@gmail.com>
204 Date: Fri Sep 20 17:50:27 2013 +0100
206 osd: OSD: add 'heap' command to known osd commands array
208 Must have been forgotten during the cli rework.
212 Signed-off-by: Joao Eduardo Luis <jecluis@gmail.com>
213 (cherry picked from commit 296f2d0db31e9f5a59a3a62a1e95b6c440430fa3)
215 commit 3f32f57b98e0224a1d30b2a81d7d260be0f53800
216 Author: Joao Eduardo Luis <jecluis@gmail.com>
217 Date: Fri Sep 20 16:43:27 2013 +0100
219 mds: MDS: pass only heap profiler commands instead of the whole cmd vector
221 The heap profiler doesn't care, nor should it, what our command name is.
222 It only cares about the commands it handles.
226 Signed-off-by: Joao Eduardo Luis <jecluis@gmail.com>
227 (cherry picked from commit 238fe272c6bdb62d4e57fd8555c0136de99c8129)
229 commit 46dcc46617d8f35ab8433540b22343ddcbcc3716
230 Author: Joao Eduardo Luis <jecluis@gmail.com>
231 Date: Fri Sep 20 16:41:14 2013 +0100
233 perfglue/heap_profiler.cc: expect args as first element on cmd vector
235 We used to pass 'heap' as the first element of the cmd vector when
236 handling commands. We haven't been doing so for a while now, so we
239 Not expecting 'heap' also makes sense, considering that what we need to
240 know when we reach this function is what command we should handle, and
241 we should not care what the caller calls us when handling his business.
246 Signed-off-by: Joao Eduardo Luis <jecluis@gmail.com>
247 (cherry picked from commit c98b910d49bd2b46ceafdc430044a31524c29f5b)
249 commit 9dc5f15fbae22244ad1f62925e17c9d81e856e55
250 Author: Yehuda Sadeh <yehuda@inktank.com>
251 Date: Mon Sep 16 14:35:25 2013 -0700
253 rgw: destroy get_obj handle in copy_obj()
257 We take different code paths in copy_obj, make sure we close the handle
258 when we exit the function. Move the call to finish_get_obj() out of
259 copy_obj_data() as we don't create the handle there, so that should
260 makes code less confusing and less prone to errors.
261 Also, note that RGWRados::get_obj() also calls finish_get_obj(). For
262 everything to work in concert we need to pass a pointer to the handle
263 and not the handle itself. Therefore we needed to also change the call
266 Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
267 Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
268 (cherry picked from commit 9e98620e4325d15c88440a890b267131613e1aa1)
270 commit 471233e98a9f64ad513a4a196b7661b80534cb00
271 Author: Joao Eduardo Luis <joao.luis@inktank.com>
272 Date: Mon Sep 9 23:14:11 2013 +0100
274 mon: MonCommands: expect a CephString as 1st arg for 'osd crush move'
278 Signed-off-by: Joao Eduardo Luis <joao.luis@inktank.com>
279 (cherry picked from commit 7d3799fde19138f957f26ec6be10a8a0000fc1f0)
281 commit 2908225092bd2aa1b8afcb7848c1cdac5bd9e638
282 Author: Sage Weil <sage@inktank.com>
283 Date: Mon Sep 23 16:23:33 2013 -0700
285 osd: revert 'osd max xattr size' limit
287 Set it to 0 (unlimited) for now.
291 Signed-off-by: Sage Weil <sage@inktank.com>
292 Reviewed-by: Yehuda Sadeh <yehuda@inktank.com>
293 (cherry picked from commit abb88d70643c3a76435b7a9d5b04ff29f7502361)
295 commit b3d3b3747c1eef695138dac828e5fcb435309c7b
296 Author: Greg Farnum <greg@inktank.com>
297 Date: Wed Sep 11 16:24:32 2013 -0700
299 mds: be more careful about decoding LogEvents
301 We need to wrap the full decode section or we can abort the process
302 if there's an issue (which we may want to just skip by).
304 Signed-off-by: Greg Farnum <greg@inktank.com>
305 Reviewed-by: Sage Weil <sage@inktank.com>
306 (cherry picked from commit 73289b34b0be5b6612e38944794d59b5e789f841)
308 commit 06c58132199ed22413b509dfa751321ccdb24225
309 Author: Joao Eduardo Luis <joao.luis@inktank.com>
310 Date: Tue Sep 17 17:58:20 2013 +0100
312 mon: OSDMonitor: multiple rebuilt full maps per transaction
314 Signed-off-by: Joao Eduardo Luis <joao.luis@inktank.com>
315 (cherry picked from commit 0d20cae0be701c5b6151a26ee5e4fe24d89aa20a)
317 commit 65bbcaf4b68790dae4506c1f5db237077e1ff0ae
318 Author: Joao Eduardo Luis <joao.luis@inktank.com>
319 Date: Sun Sep 15 21:03:50 2013 +0100
321 mon: OSDMonitor: update latest_full while rebuilding full maps
323 Not doing so will make the monitor rebuild the osdmap full versions, even
324 though they may have been rebuilt before, every time the monitor starts.
326 This mostly happens when the cluster is left in an unhealthy state for
327 a long period of time and incremental versions build up. Even though we
328 build the full maps on update_from_paxos(), not updating 'full_latest'
329 leads to the situation initially described.
333 Signed-off-by: Joao Eduardo Luis <joao.luis@inktank.com>
334 (cherry picked from commit 81983bab3630520d6c7ee9b7e4a747bc17b8c5c3)
336 commit 9b9edb04581cca15e67c567332529f5b3f426743
337 Author: Joao Eduardo Luis <joao.luis@inktank.com>
338 Date: Sun Sep 15 21:00:55 2013 +0100
340 mon: OSDMonitor: smaller transactions when rebuilding full versions
342 Otherwise, for considerably sized rebuilds, the monitor will not only
343 consume vast amounts of memory, but it will also have troubles committing
344 the transaction. Anyway, it's also a good idea to adjust transactions to
345 the granularity we want, and to be fair we care that each rebuilt full map
346 gets to disk, even if subsequent full maps don't (those can be rebuilt
351 Signed-off-by: Joao Eduardo Luis <joao.luis@inktank.com>
352 (cherry picked from commit 4ac1570c5cdcd6556dc291cc6d7878fd92d343ae)
354 commit 298811f7a15541b9ec1015c416ad2aa075be5691
355 Author: Joao Eduardo Luis <jecluis@gmail.com>
356 Date: Wed Aug 28 15:51:01 2013 +0100
358 mon: OSDMonitor: check if pool is on unmanaged snaps mode on mk/rmsnap
363 Signed-off-by: Joao Eduardo Luis <jecluis@gmail.com>
364 (cherry picked from commit fab79543c54c2e446d3f76520d7906645c6b0075)
366 commit a992664435db9dde3745eb7f354cce3fc5400a47
367 Author: Yehuda Sadeh <yehuda@inktank.com>
368 Date: Thu Sep 12 14:32:17 2013 -0700
370 lru_map: don't use list::size()
372 replace list::size() with map::size(), which should have
373 a constant time complexity.
375 Reviewed-by: Sage Weil <sage@inktank.com>
376 Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
377 (cherry picked from commit 7c1d2ded8fa8061bf3f14932800998b963745dd1)
379 commit 788546ea71c994ff35323747294ed9c177fe7020
380 Author: Yehuda Sadeh <yehuda@inktank.com>
381 Date: Thu Sep 12 14:30:19 2013 -0700
383 common/lru_map: rename tokens to entries
385 This code was originally used in a token cache, now
386 as a generic infrastructure rename token fields.
388 Reviewed-by: Sage Weil <sage@inktank.com>
389 Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
390 (cherry picked from commit 532e41a9985a16b35a6e49cdcba38af0ad166fa8)
392 commit babeb00c42af760b3e7575166479e95365cfcc0a
393 Author: Yehuda Sadeh <yehuda@inktank.com>
394 Date: Wed Sep 18 10:37:21 2013 -0700
396 rgw: use bufferlist::append() instead of bufferlist::push_back()
398 push_back() expects char *, whereas append can append a single char.
399 Appending a NULL char to push_back is cast as a NULL pointer which is
402 Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
403 Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
404 (cherry picked from commit 08fe028bad13096d482454a2f303158727c363ff)
406 commit daf85c45dd4d158bc7c33a2fb784857bc7db35cd
407 Author: Yehuda Sadeh <yehuda@inktank.com>
408 Date: Wed Sep 11 13:46:31 2013 -0700
410 rgw: NULL terminate buffer before parsing it
414 We get a buffer off the remote gateway which might
415 not be NULL terminated. The JSON parser needs the
416 buffer to be NULL terminated even though we provide
417 a buffer length as it calls strlen().
419 Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
420 Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
421 (cherry picked from commit e7f7483192cddca1159aba439ce62b1e78669d51)
423 commit c73040a5518971813b9ebaae1624c5bacef315d0
424 Author: Yehuda Sadeh <yehuda@inktank.com>
425 Date: Wed Sep 11 22:30:12 2013 -0700
427 rgw: don't call list::size() in ObjectCache
430 Use an external counter instead of calling list::size()
432 Reviewed-by: Sage Weil <sage@inktank.com>
433 Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
434 (cherry picked from commit 31e3a51e933429d286104fe077e98ea883437ad6)
436 commit a855aba9d18936e9a060119e041518790cd4b831
437 Author: Yehuda Sadeh <yehuda@inktank.com>
438 Date: Tue Sep 10 12:18:55 2013 -0700
440 rgw: drain pending requests before completing write
443 When doing aio write of objects (either regular or multipart parts) we
444 need to drain pending aio requests. Otherwise if gateway goes down then
445 object might end up corrupted.
447 Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
448 Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
449 (cherry picked from commit 626669afaa333d73707553a85f5c874e99e9cbd8)
451 commit 670db7e80ddc9c26c43a4f66907a5996ce207c4d
452 Author: Yehuda Sadeh <yehuda@inktank.com>
453 Date: Fri Sep 6 22:33:38 2013 -0700
455 rgw: fix get cors, delete cors
457 Remove a couple of variables that overrode class member. Not
458 really clear how it was working before, might have been a bad
461 Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
462 (cherry picked from commit 13872785aeeddbe1b8dd97e49fd6a2d879514f8d)
464 commit a304016fa01b02efd500135c00b9bf3407a9999c
465 Merge: 408cd61 ac0a30f
466 Author: Yehuda Sadeh <yehuda@inktank.com>
467 Date: Wed Sep 11 09:47:10 2013 -0700
469 Merge branch 'wip-6078-dumpling' into dumpling
471 Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
473 commit ac0a30feb8c64a3b80d9c519a7b561213403afab
474 Author: Yehuda Sadeh <yehuda@inktank.com>
475 Date: Wed Aug 28 21:25:20 2013 -0700
477 rgw: fix certain return status cases in CORS
479 Change return values in certain cases, reorder
482 Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
484 commit 13b28cc3f1eb8ef42875b630c485ee0105cd244a
485 Author: Yehuda Sadeh <yehuda@inktank.com>
486 Date: Wed Aug 28 21:24:36 2013 -0700
488 rgw: add COPY method to be handled by CORS
490 Was missing this http method.
492 Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
494 commit d45c87ea738807487e72c0719b0d3d459cbe19e9
495 Author: Yehuda Sadeh <yehuda@inktank.com>
496 Date: Tue Aug 27 19:38:45 2013 -0700
498 rgw: fix CORS rule check
500 Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
502 commit 986fa92a7a1d88111ba28457160adfcfdaabc5d2
503 Author: Yehuda Sadeh <yehuda@inktank.com>
504 Date: Tue Aug 27 19:38:18 2013 -0700
506 rgw: don't handle CORS if rule not found (is NULL)
508 Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
510 commit 71873aba6553492d3ad71596cefd7c841030a277
511 Author: Yehuda Sadeh <yehuda@inktank.com>
512 Date: Thu Aug 22 13:38:55 2013 -0700
514 rgw: tie CORS header response to all relevant operations
516 Have the CORS responses on all relevant operations. Also add headers
519 Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
521 commit 94e7b594d85dbd26e58d823b41f418032e9f163f
522 Author: Yehuda Sadeh <yehuda@inktank.com>
523 Date: Thu Aug 22 10:00:53 2013 -0700
525 rgw: add a generic CORS response handling
527 Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
529 commit c3385d8a102faf5379559bb98cf89637ceda1579
530 Author: Yehuda Sadeh <yehuda@inktank.com>
531 Date: Wed Aug 21 17:22:46 2013 -0700
533 rgw: OPTIONS request doesn't need to read object info
535 This is a bucket-only operation, so we shouldn't look at the
536 object. Object may not exist and we might respond with Not
537 Exists response which is not what we want.
539 Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
541 commit a5fdd44e5d8ce4b8d82273d83e27aea19e63aa7c
542 Author: Yehuda Sadeh <yehuda@inktank.com>
543 Date: Wed Aug 21 14:43:28 2013 -0700
545 rgw: remove use of s->bucket_cors
547 Some old code still tried to use s->bucket_cors, which was
548 abandoned in a cleanup work.
550 Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>