ceph/doc/changelog/v0.80.2.txt

   1 commit 198236d68deceb16b6b590d807786a140209bfea (tag: refs/tags/v0.80.2, refs/remotes/gh/last)
   2 Author: Jenkins <jenkins@inktank.com>
   3 Date:   Wed Jul 9 19:32:26 2014 +0000
   4
   5     0.80.2
   6
   7 commit 0b6bd2545925b5e8a80d41de1fda13ffe9d30e2b
   8 Author: Josh Durgin <josh.durgin@inktank.com>
   9 Date:   Wed Jul 9 12:04:49 2014 -0700
  10
  11     Revert "rgw: if extra data pool name is empty, use data pool name instead"
  12
  13     This breaks multipart uploads on firefly, though not on master.
  14     Revert for now so we can release 0.80.2.
  15
  16     This reverts commit 5fd8b0d1639c67e355f0fc0d7e6d7036618d87a1.
  17
  18 commit 9426c4faa9ca327f18f20a57ba772b08d6f069e7
  19 Merge: b81a277 4733178
  20 Author: Sage Weil <sage@inktank.com>
  21 Date:   Tue Jul 8 10:06:01 2014 -0700
  22
  23     Merge pull request #2083 from ceph/wip-8727.firefly
  24
  25     messages: MForward: fix compat version
  26
  27     Reviewed-by: Sage Weil <sage@redhat.com>
  28
  29 commit 4733178c3a8f57c54196f176b20154415783dba0
  30 Author: Joao Eduardo Luis <joao.luis@inktank.com>
  31 Date:   Tue Jul 8 16:46:27 2014 +0100
  32
  33     messages: MForward: fix compat version
  34
  35     A while ago we bumped the head version and reset the compat version to 0.
  36     Doing this so happens to make the messenger assume that the message does
  37     not support the compat versioning and sets the compat version to the head
  38     version -- thus making compat = 2 when it should have been 1.
  39
  40     The nasty side-effect of this is that upgrading from emperor to firefly
  41     will have emperor-leaders being unable to decode forwarded messages from
  42     firefly-peons.
  43
  44     Fixes: #8727
  45
  46     Signed-off-by: Joao Eduardo Luis <joao.luis@inktank.com>
  47     (cherry picked from commit c92feebb81f90344a085f8256f0ed03cdc2f14e1)
  48
  49 commit b81a2777aea636ffe544e737cd8e17dc892ba328
  50 Author: Dan Mick <dan.mick@inktank.com>
  51 Date:   Thu Jul 3 16:11:24 2014 -0700
  52
  53     ceph.spec.in: add bash completion file for radosgw-admin
  54
  55     Signed-off-by: Sandon Van Ness <sandon@inktank.com>
  56     (cherry picked from commit b70096307130bcbac176704493a63c5d039d3edc)
  57
  58 commit 4649d7e90ffec89b452aef69cf833522bdb81f88
  59 Author: Dan Mick <dan.mick@inktank.com>
  60 Date:   Thu Jul 3 16:10:55 2014 -0700
  61
  62     ceph.spec.in: rhel7-related changes:
  63
  64     udev rules: /lib -> /usr/lib
  65     /sbin binaries move to /usr/sbin or %{_sbindir}
  66
  67     Signed-off-by: Sandon Van Ness <sandon@inktank.com>
  68     (cherry picked from commit 235e4c7de8f8efe491edefbdde8e5da4dfc44034)
  69
  70     Conflicts:
  71         ceph.spec.in
  72
  73 commit 38a9566e70d3a2de80c7547fd4e84c8372639f80
  74 Author: Dan Mick <dan.mick@inktank.com>
  75 Date:   Thu Jul 3 16:08:44 2014 -0700
  76
  77     Fix/add missing dependencies:
  78
  79     - rbd-fuse depends on librados2/librbd1
  80     - ceph-devel depends on specific releases of libs and libcephfs_jni1
  81     - librbd1 depends on librados2
  82     - python-ceph does not depend on libcephfs1
  83
  84     Signed-off-by: Sandon Van Ness <sandon@inktank.com>
  85     (cherry picked from commit 7cf81322391b629b241da90181800ca1f138ce78)
  86
  87     Conflicts:
  88         ceph.spec.in
  89
  90 commit 733c89dc6af89321ee58ad78b30bc6a0c4e78291
  91 Author: Dan Mick <dan.mick@inktank.com>
  92 Date:   Thu Jul 3 16:05:00 2014 -0700
  93
  94     ceph.spec.in: whitespace fixes
  95
  96     Signed-off-by: Sandon Van Ness <sandon@inktank.com>
  97     (cherry picked from commit ec8af52a5ede78511423a1455a496d46d580c644)
  98
  99     Conflicts:
 100         ceph.spec.in
 101
 102 commit 21b03bc96ab1a0877235f1e3e4a96899cba2b4fa
 103 Author: Dan Mick <dan.mick@inktank.com>
 104 Date:   Thu Jul 3 16:04:10 2014 -0700
 105
 106     ceph.spec.in: split out ceph-common as in Debian
 107
 108     Move files, postun scriptlet, and add dependencies on ceph-common
 109     where appropriate
 110
 111     Signed-off-by: Sandon Van Ness <sandon@inktank.com>
 112     (cherry picked from commit e131b9d5a5e90e87d8a8346cb96cb5a26135c144)
 113
 114 commit dbee797d1344aec47fedc8d3fdd6c9460e0ffa16
 115 Merge: 6d6039a 9e81fc1
 116 Author: Sage Weil <sage@inktank.com>
 117 Date:   Mon Jul 7 09:40:51 2014 -0700
 118
 119     Merge pull request #2057 from ceph/wip-8593-firefly
 120
 121     mon: backport health check improvements
 122
 123     Reviewed-by: Joao Eduardo Luis <joao.luis@inktank.com>
 124
 125 commit 6d6039a5a56743c006a0d081157cb6ee9e3b7af6
 126 Author: Greg Farnum <greg@inktank.com>
 127 Date:   Thu Jul 3 16:30:49 2014 -0700
 128
 129     mon: check changes to the whole CRUSH map and to tunables against cluster features
 130
 131     When we change the tunables, or set a new CRUSH map, we need to make sure it's
 132     supported by all the monitors and OSDs currently participating in the cluster.
 133
 134     Fixes: #8738
 135
 136     Signed-off-by: Greg Farnum <greg@inktank.com>
 137     (cherry picked from commit 54af8104816ccc76aad251aa47a6efc1bf879e7d)
 138
 139 commit 0a8d194f2c44a73c31ae8c52fabb430d66d4a1b1
 140 Author: Greg Farnum <greg@inktank.com>
 141 Date:   Thu Jul 3 15:24:30 2014 -0700
 142
 143     OSDMonitor: fix quorum_features comparison in check_cluster_features
 144
 145     We need to see if there's a feature which is not in the quorum_features,
 146     not if there are no features in common!
 147
 148     Signed-off-by: Greg Farnum <greg@inktank.com>
 149     (cherry picked from commit 2280c0eaa9f3f733062085a5e7f7dc23c3bbe291)
 150
 151 commit a49c61222f8014db533d8f505ec44d10dc5478e6
 152 Author: Sage Weil <sage@inktank.com>
 153 Date:   Mon Jun 16 12:57:36 2014 -0700
 154
 155     ceph_test_rados_api_tier: disable LibRadosTierECPP::HitSetWrite
 156
 157     Disable this test until hitget-get reliably works on EC pools (currently
 158     it does not, and this test usually passes only because we get the in-memory
 159     HitSet).
 160
 161     Signed-off-by: Sage Weil <sage@inktank.com>
 162     (cherry picked from commit c5b5ed672c039cc4fa04772e065a59d3b1df803a)
 163
 164 commit 828ce4055c8580e42175159c02a9dfa76bc414ff
 165 Merge: b9b498e b1a4a7c
 166 Author: Sage Weil <sage@inktank.com>
 167 Date:   Fri Jul 4 21:18:32 2014 -0700
 168
 169     Merge pull request #2073 from ceph/wip-rgw-firefly-2
 170
 171     rgw: more firefly backports
 172
 173     Reviewed-by: Yehuda Sadeh <yehuda@inktank.com>
 174
 175 commit b1a4a7cb91e164d1f8af8ce9319e3b3c1949858d
 176 Author: Yehuda Sadeh <yehuda@inktank.com>
 177 Date:   Wed Jun 11 16:50:41 2014 -0700
 178
 179     rgw: set a default data extra pool name
 180
 181     Fixes: #8585
 182     Have a default name for the data extra pool, otherwise it would be empty
 183     which means that it'd default to the data pool name (which is a problem
 184     with ec backends).
 185
 186     Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
 187     (cherry picked from commit b2542f8e880e45bb6d052f13eb3ef6a7c61e4cd4)
 188
 189 commit b9b498e6a7794205af8aa99ffe5266e478c0e711
 190 Merge: f285266 053c261
 191 Author: Sage Weil <sage@inktank.com>
 192 Date:   Thu Jul 3 07:00:52 2014 -0700
 193
 194     Merge pull request #2028 from ceph/wip-rgw-firefly
 195
 196     rgw: a couple backports for firefly
 197
 198     Passed the rgw suite, modulo a python bootstrap issue.
 199
 200     sage-2014-07-01_09:50:22-rgw-wip-rgw-firefly-testing-basic-plana/337393
 201
 202     Reviewed-by: Yehuda Sadeh <yehuda@inktank.com>
 203
 204 commit 9f06e2c16e409a9a8284af9964f0cfff6e222164
 205 Author: Sage Weil <sage@inktank.com>
 206 Date:   Mon Jun 30 12:55:45 2014 -0700
 207
 208     rgw: fix uninit ofs in RGWObjManifect::obj_iterator
 209
 210     Valgrind picked this up:
 211
 212       <kind>UninitCondition</kind>
 213       <what>Conditional jump or move depends on uninitialised value(s)</what>
 214       <stack>
 215         <frame>
 216           <ip>0x5145B8</ip>
 217           <obj>/usr/bin/radosgw</obj>
 218           <fn>RGWObjManifest::obj_iterator::seek(unsigned long)</fn>
 219           <dir>/srv/autobuild-ceph/gitbuilder.git/build/out~/ceph-0.82-354-g62027ec/src/rgw</dir>
 220           <file>rgw_rados.cc</file>
 221           <line>562</line>
 222         </frame>
 223         <frame>
 224           <ip>0x5672A4</ip>
 225           <obj>/usr/bin/radosgw</obj>
 226           <fn>list_multipart_parts(RGWRados*, req_state*, std::string const&amp;, std::string&amp;, int, int, std::map&lt;unsigned int, RGWUploadPartInfo, std::less&lt;unsigned int&gt;, std::allocator&lt;std::pair&lt;unsigned int const, RGWUploadPartInfo&gt; &gt; &gt;&amp;, int*, bool*, bool)</fn>
 227           <dir>/srv/autobuild-ceph/gitbuilder.git/build/out~/ceph-0.82-354-g62027ec/src/rgw</dir>
 228           <file>rgw_rados.h</file>
 229           <line>217</line>
 230         </frame>
 231         <frame>
 232           <ip>0x5688EE</ip>
 233           <obj>/usr/bin/radosgw</obj>
 234           <fn>RGWListMultipart::execute()</fn>
 235           <dir>/srv/autobuild-ceph/gitbuilder.git/build/out~/ceph-0.82-354-g62027ec/src/rgw</dir>
 236           <file>rgw_op.cc</file>
 237           <line>2956</line>
 238         </frame>
 239     ...
 240
 241     Fixes: #8699
 242     Backport: firefly
 243     Signed-off-by: Sage Weil <sage@inktank.com>
 244     (cherry picked from commit e0d3b78b6af523ae77b0cee385cd40f67f7a0ab6)
 245
 246 commit 5fd8b0d1639c67e355f0fc0d7e6d7036618d87a1
 247 Author: Yehuda Sadeh <yehuda@inktank.com>
 248 Date:   Fri May 30 14:47:25 2014 -0700
 249
 250     rgw: if extra data pool name is empty, use data pool name instead
 251
 252     Fixes: #8311
 253     An empty pool name could be used as the extra data pool.
 254
 255     Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
 256     (cherry picked from commit b300318113b162522759d4794b1cfa3d1d9398e4)
 257
 258 commit cbad09e1907ceac4feb069f9cf257e580f36da91
 259 Author: Yehuda Sadeh <yehuda@inktank.com>
 260 Date:   Tue May 6 15:35:20 2014 -0700
 261
 262     rgw: extend manifest to avoid old style manifest
 263
 264     In case we hit issue #8269 we'd like to avoid creating an old style
 265     manifest. Since we need to have parts that use different prefix we add a
 266     new rule param that overrides the manifest prefix.
 267
 268     Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
 269     (cherry picked from commit 545d8ad1d2dfe53b967ab1bd17dfeb1780abbb5e)
 270
 271 commit ea68b9372319fd0bab40856db26528d36359102e
 272 Author: Yehuda Sadeh <yehuda@inktank.com>
 273 Date:   Fri May 2 17:06:05 2014 -0700
 274
 275     rgw: don't allow multiple writers to same multiobject part
 276
 277     Fixes: #8269
 278     Backport: firefly, dumpling
 279
 280     A client might need to retry a multipart part write. The original thread
 281     might race with the new one, trying to clean up after it, clobbering the
 282     part's data.
 283     The fix is to detect whether an original part already existed, and if so
 284     use a different part name for it.
 285
 286     Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
 287     (cherry picked from commit bd8e026f88b812cc70caf6232c247844df5d99bf)
 288
 289 commit f285266a320bf22df8b09a669d1c1fd400b05765
 290 Merge: d43e711 6a0dd1b
 291 Author: Sage Weil <sage@inktank.com>
 292 Date:   Tue Jul 1 16:16:09 2014 -0700
 293
 294     Merge pull request #2056 from ceph/wip-8010
 295
 296     Revert "mon: OSDMonitor: Refuse to delete CephFS pools"
 297
 298     Reviewed-by: Sage Weil <sage@redhat.com>
 299
 300 commit 9e81fc180a3999ebec95949a2cb31fea970573fb
 301 Author: Sage Weil <sage@inktank.com>
 302 Date:   Mon Jun 16 17:00:51 2014 -0700
 303
 304     mon: ensure HealthService warning(s) include a summary
 305
 306     The low disk space check would change our status to HEALTH_WARN and include
 307     a detail message, but no summary.  We need both.
 308
 309     Backport: firefly
 310     Signed-off-by: Sage Weil <sage@inktank.com>
 311     (cherry picked from commit 3ed7f2dd4345633ff34017b201082f4c261ef387)
 312
 313 commit e96637c2e33ec092f28af0ce96bfe2e4d7748ff8
 314 Author: Sage Weil <sage@inktank.com>
 315 Date:   Mon Jun 16 16:58:14 2014 -0700
 316
 317     mon: refactor check_health()
 318
 319     Refactor the get_health() methods to always take both a summary and detail.
 320     Eliminate the return value and pull that directly from the summary, as we
 321     already do with the PaxosServices.
 322
 323     Signed-off-by: Sage Weil <sage@inktank.com>
 324     (cherry picked from commit 82e47db8073b622183a5e33f6e0b999a3a144804)
 325
 326 commit 28584200df36dda5d434d4d7f58585933905de30
 327 Author: Sage Weil <sage@inktank.com>
 328 Date:   Mon Jun 16 16:40:05 2014 -0700
 329
 330     mon: fix typos, punctuation for mon disk space warning(s)
 331
 332     Backport: firefly
 333     Signed-off-by: Sage Weil <sage@inktank.com>
 334     (cherry picked from commit 98883f6308ce72f69a71feab29ef00e13f319cdb)
 335
 336     Conflicts:
 337
 338         src/mon/DataHealthService.cc
 339
 340 commit 4f0666085e2fd553ccc05cef8510aae820849aca
 341 Author: Sage Weil <sage@inktank.com>
 342 Date:   Mon Jun 16 16:27:05 2014 -0700
 343
 344     mon/OSDMonitor: make down osd count sensible
 345
 346     We currently log something like
 347
 348      1/10 in osds are down
 349
 350     in the health warning when there are down OSDs, but this is based on a
 351     comparison of the number of up vs the number of in osds, and makes no sense
 352     when there are up osds that are not in.
 353
 354     Instead, count only the number OSDs that are both down and in (relative to
 355     the total number of OSDs in) and warn about that.  This means that, if a
 356     disk fails, and we mark it out, and the cluster fully repairs itself, it
 357     will go back to a HEALTH_OK state.
 358
 359     I think that is a good thing, and certainly preferable to the current
 360     nonsense.  If we want to distinguish between down+out OSDs that were failed
 361     vs those that have been "acknowledged" by an admin to be dead, we will
 362     need to add some additional state (possibly reusing the AUTOOUT flag?), but
 363     that will require more discussion.
 364
 365     Backport: firefly (maybe)
 366     Signed-off-by: Sage Weil <sage@inktank.com>
 367     (cherry picked from commit 55a97787088b79356c678a909b2410b3924e7f5b)
 368
 369 commit 6a0dd1be909b423f52e2c7f11531b278dc8838ae
 370 Author: John Spray <john.spray@redhat.com>
 371 Date:   Tue Jul 1 22:04:25 2014 +0100
 372
 373     Revert "mon: OSDMonitor: Refuse to delete CephFS pools"
 374
 375     This reverts commit c7d265a9b276aad5caf9b35f3ce8bc7bdd169f0f.
 376
 377     Because this pre-dates the `fs rm` command, this change was
 378     preventing firefly users from ever deleting their filesystem pools.
 379
 380     Fixes: #8010
 381
 382     Signed-off-by: John Spray <john.spray@redhat.com>
 383
 384 commit 053c261e5bfcb4d39770048775515eb6a667f987
 385 Author: Yehuda Sadeh <yehuda@inktank.com>
 386 Date:   Wed Jun 4 09:24:33 2014 -0700
 387
 388     rgw: set meta object in extra flag when initializing it
 389
 390     As part of the fix for 8452 we moved the meta object initialization.
 391     Missed moving the extra flag initialization that is needed. This breaks
 392     setups where there's a separate extra pool (needed in ec backends).
 393
 394     Reported-by: Sylvain Munaut <s.munaut@whatever-company.com>
 395     Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
 396     (cherry picked from commit 703d0eb9bffdf6c553b659728a3a07f068fb15ee)
 397
 398 commit d43e7113dd501aea1db33fdae30d56e96e9c3897
 399 Author: Sage Weil <sage@inktank.com>
 400 Date:   Mon Jun 30 07:05:04 2014 -0700
 401
 402     qa/workunits/suites/fsx.sh: don't use zero range
 403
 404     Zero range is not supported by cephfs.
 405
 406     Fixes: #8542
 407     Signed-off-by: Sage Weil <sage@inktank.com>
 408     (cherry picked from commit 2dec8a810060f65d022c06e82090b4aa5ccec0cb)
 409
 410 commit ee0d322c0b1ebaba2f225b8bed9af2948fdcf4c3
 411 Merge: 3ceafcf a7119c3
 412 Author: Loic Dachary <loic@dachary.org>
 413 Date:   Mon Jun 30 17:01:03 2014 +0200
 414
 415     Merge pull request #1991 from dachary/wip-8307-erasure-code-profile-implicit-creation
 416
 417     erasure code profile implicit creation (firefly backport)
 418
 419     Reviewed-by: Joao Eduardo Luis <joao.luis@inktank.com>
 420
 421 commit a7119c31e9a1b415f56426d2b4cd362357ad8b17
 422 Author: Loic Dachary <loic@dachary.org>
 423 Date:   Wed Jun 11 22:44:57 2014 +0200
 424
 425     erasure-code: pool create must not create profiles
 426
 427     If a non existent profile is provided as an argument to osd pool create,
 428     it must exit on error and not create the profile as a side effect.
 429
 430     http://tracker.ceph.com/issues/8307 refs: #8307
 431
 432     Signed-off-by: Loic Dachary <loic@dachary.org>
 433     (cherry picked from commit ff2eb234e63cd281b40405cb3397da5babda943f)
 434
 435 commit a67c1c1c51b111f40819643530e07f8026227caa
 436 Author: Loic Dachary <loic@dachary.org>
 437 Date:   Wed Jun 11 22:47:49 2014 +0200
 438
 439     erasure-code: OSDMonitor::get_erasure_code is a const
 440
 441     If it is not, the non const version of OSDMap::get_erasure_code_profile
 442     is called and a profile is created as a side effect, which is not
 443     intended.
 444
 445     http://tracker.ceph.com/issues/8307 refs: #8307
 446
 447     Signed-off-by: Loic Dachary <loic@dachary.org>
 448     (cherry picked from commit 3c638111a4943758b6089c63a42aabbf281ac257)
 449
 450 commit 3ceafcf3576efaef58dffbe70654baf1db4f3a1d
 451 Author: Loic Dachary <loic@dachary.org>
 452 Date:   Tue May 27 10:06:46 2014 +0200
 453
 454     mon: fix set cache_target_full_ratio
 455
 456     It was a noop because it was incorrectly using the variable n. Add a
 457     test to protect against regression.
 458
 459     http://tracker.ceph.com/issues/8440 Fixes: #8440
 460
 461     Reported-by: Geoffrey Hartz <hartz.geoffrey@gmail.com>
 462     Signed-off-by: Loic Dachary <loic@dachary.org>
 463     (cherry picked from commit c2225f874dcf37222d831b65b5a319d598d2fcd9)
 464
 465 commit 5f66a2c99fdef2c8d609126740ca790ce10e67a4
 466 Author: Yehuda Sadeh <yehuda@inktank.com>
 467 Date:   Mon Jun 16 11:48:24 2014 -0700
 468
 469     rgw: allocate enough space for bucket instance id
 470
 471     Fixes: #8608
 472     Backport: dumpling, firefly
 473     Bucket instance id is a concatenation of zone name, rados instance id,
 474     and a running counter. We need to allocate enough space to account zone
 475     name length.
 476
 477     Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
 478     (cherry picked from commit d2e86a66ca55685e04ffbfaa58452af59f381277)
 479
 480 commit 84d7971c8e50e7d35d7a5fac9353e687f168a1a1
 481 Author: Alfredo Deza <alfredo.deza@inktank.com>
 482 Date:   Fri Jun 20 11:14:25 2014 -0400
 483
 484     log the command that is being run with subprocess
 485
 486     Signed-off-by: Alfredo Deza <alfredo.deza@inktank.com>
 487     (cherry picked from commit e189a668285f9ab73116bc19f9df1cc515473541)
 488
 489 commit 291d448478360a1be0a9826ada58c94ecd643ad0
 490 Author: Yehuda Sadeh <yehuda@inktank.com>
 491 Date:   Fri May 30 16:17:47 2014 -0700
 492
 493     rgw: fetch object attrs on multipart completion
 494
 495     Fixes: #8452
 496     Backport: firefly
 497     This fixes a regression following a code cleanup.
 498
 499     Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
 500     (cherry picked from commit 70afaaabf979d4cc1a14dbc2b772224cbafcac2f)
 501
 502 commit 2fd808fe93fe0cbd8754eb81d281c83254411039
 503 Author: Yehuda Sadeh <yehuda@inktank.com>
 504 Date:   Thu May 22 18:27:58 2014 -0700
 505
 506     rgw: check appropriate entity permission on put_metadata
 507
 508     Fixes: #8428
 509     Backport: firefly
 510
 511     Cannot use verify_object_permission() to test acls, as the operation
 512     here might either be on object or on bucket.
 513
 514     Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
 515     (cherry picked from commit 2ceb13af34bc38b418bb88d81131d770a71159bf)
 516
 517 commit d1afeb38b522baae1b15a7a3548002c830bb7567
 518 Author: Ilya Dryomov <ilya.dryomov@inktank.com>
 519 Date:   Thu Jun 5 10:08:42 2014 +0400
 520
 521     XfsFileStoreBackend: call ioctl(XFS_IOC_FSSETXATTR) less often
 522
 523     No need to call ioctl(XFS_IOC_FSSETXATTR) if extsize is already set to
 524     the value we want or if any extents are allocated - XFS will refuse to
 525     change extsize in that's the case.
 526
 527     Fixes: #8241
 528     Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com>
 529     (cherry picked from commit bc3b30ed09b8f3eb86b61e3a05ccacfd928faa95)
 530
 531 commit 7b8e5c99a4a40ae788ad29e36b0d714f529b12eb
 532 Author: John Spray <john.spray@inktank.com>
 533 Date:   Tue May 20 16:25:19 2014 +0100
 534
 535     mon: Fix default replicated pool ruleset choice
 536
 537     Specifically, in the case where the configured
 538     default ruleset is CEPH_DEFAULT_CRUSH_REPLICATED_RULESET,
 539     instead of assuming ruleset 0 exists, choose the lowest
 540     numbered ruleset.
 541
 542     In the case where an explicit ruleset is passed to
 543     OSDMonitor::prepare_pool_crush_ruleset, verify
 544     that it really exists.
 545
 546     The idea is to eliminate cases where a pool could
 547     exist with its crush ruleset set to something
 548     other than a value ruleset ID.
 549
 550     Fixes: #8373
 551
 552     Signed-off-by: John Spray <john.spray@inktank.com>
 553     (cherry picked from commit 1d9e4ac2e2bedfd40ee2d91a4a6098150af9b5df)
 554
 555     Conflicts:
 556
 557         src/crush/CrushWrapper.h
 558
 559 commit 25a4a874c8e53cd644d6705478fbea4fdb232c4d
 560 Author: Yehuda Sadeh <yehuda@inktank.com>
 561 Date:   Fri May 23 14:58:54 2014 -0700
 562
 563     rgw: calc md5 and compare if user provided appropriate header
 564
 565     Fixes: #8436
 566     Backport: firefly
 567
 568     This was broken in ddc2e1a8e39a5c6b9b224c3eebd1c0e762ca5782. The fix
 569     resurrects and old check that was dropped.
 570
 571     Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
 572     Reviewed-by: Sage Weil <sage@inktank.com>
 573     (cherry picked from commit 9c56c86bdac6bcb8e76c3f04e7d393e4eaadd721)
 574
 575 commit ad18a45a71be34899d9400c8bdef39d0548f9aaf
 576 Author: Yehuda Sadeh <yehuda@inktank.com>
 577 Date:   Mon Apr 21 15:07:12 2014 -0700
 578
 579     rgw: calculate user manifest
 580
 581     Fixes: #8169
 582     Backport: firefly
 583     We didn't calculate the user manifest's object etag at all. The etag
 584     needs to be the md5 of the contantenation of all the parts' etags.
 585
 586     Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
 587     (cherry picked from commit ddc2e1a8e39a5c6b9b224c3eebd1c0e762ca5782)
 588
 589 commit 6aa7f7ee4e082819decc48da1f0eaffc4e616302
 590 Author: Sage Weil <sage@inktank.com>
 591 Date:   Tue Jun 3 11:45:20 2014 -0700
 592
 593     librados: simplify/fix rados_pool_list bounds checks
 594
 595     We were not breaking out of the loop when we filled up the buffer unless
 596     we happened to do so on a pool name boundary.  This means that len would
 597     roll over (it was unsigned).  In my case, I was not able to reproduce
 598     anything particularly bad since (I think) the strncpy was interpreting the
 599     large unsigned value as signed, but in any case this fixes it, simplifies
 600     the arithmetic, and adds a simple test.
 601
 602     - use a single 'rl' value for the amount of buffer space we want to
 603       consume
 604     - use this to check that there is room and also as the strncat length
 605     - rely on the initial memset to ensure that the trailing 0 is in place.
 606
 607     Fixes: #8447
 608     Signed-off-by: Sage Weil <sage@inktank.com>
 609     (cherry picked from commit 3ec32a6bb11d92e36a0e6381b40ce2fd1fbb016a)
 610
 611 commit 2ba03e08f1bce0a8af87014583d635eb5c0aa376
 612 Merge: 50dc246 3ad451e
 613 Author: Sage Weil <sage@inktank.com>
 614 Date:   Wed Jun 25 12:42:11 2014 -0700
 615
 616     Merge pull request #1982 from accelazh/firefly-fix-issue-8256
 617
 618     Make <poolname> in "ceph osd tier --help" clearer (fix issue 8256).
 619
 620     Reviewed-by: Loic Dachary <loic@dachary.org>
 621
 622 commit 50dc2462ab96c3457b08c50b4cc07676b50dbf72
 623 Author: Samuel Just <sam.just@inktank.com>
 624 Date:   Tue Jun 3 16:14:15 2014 -0700
 625
 626     OSD::calc_priors_during: handle CRUSH_ITEM_NONE correctly
 627
 628     Fixes: #8507
 629     Signed-off-by: Samuel Just <sam.just@inktank.com>
 630     (cherry picked from commit 0bd6f6797c69af9aff851033c57c42121671c684)
 631
 632     Conflicts:
 633         src/osd/OSD.cc
 634
 635 commit a53bcf2357c7f351ad520039774d895b8c27987f
 636 Author: Samuel Just <sam.just@inktank.com>
 637 Date:   Tue Jun 3 16:11:32 2014 -0700
 638
 639     OSD::calc_priors_during: fix confusing for loop bracing (cosmetic)
 640
 641     Confusing lack of braces is confusing.
 642
 643     Signed-off-by: Samuel Just <sam.just@inktank.com>
 644     (cherry picked from commit d76936b92300be5cc403fd5a36616a2424c7877d)
 645
 646     Conflicts:
 647         src/osd/OSD.cc
 648
 649 commit 178e6f0f422c9f1b8a51d0e9c364773397a4555f
 650 Author: Samuel Just <sam.just@inktank.com>
 651 Date:   Tue Jun 24 10:11:21 2014 -0700
 652
 653     rados.cc: fix pool alignment check
 654
 655     Only check pool alignment if io_ctx is initialized.
 656
 657     Introduced in 304b08a23a3db57010078046955a786fe3589ef8
 658     Fixes: #8652
 659     Signed-off-by: Samuel Just <sam.just@inktank.com>
 660
 661     (cherry picked from commit d7350a3741bf4cdb270c6361e68090fe280cf36d)
 662
 663     Conflicts:
 664         src/tools/rados/rados.cc
 665
 666 commit a9f97f8cae5b9d0245e0b0842c5c581cdacbb2da
 667 Author: Sage Weil <sage@inktank.com>
 668 Date:   Tue Jun 17 13:33:14 2014 -0700
 669
 670     osd: fix filestore perf stats update
 671
 672     Update the struct we are about to send, not the (unlocked!) one we will
 673     send the next time around.
 674
 675     Backport: firefly, dumpling
 676     Signed-off-by: Sage Weil <sage@inktank.com>
 677     (cherry picked from commit 4afffb4a10a0bbf7f2018ef3ed6b167c7921e46b)
 678
 679 commit b758fd5f0ebcccdb1641a852d40ee880bad01107
 680 Author: Greg Farnum <greg@inktank.com>
 681 Date:   Thu Apr 24 15:34:24 2014 -0700
 682
 683     FileStore: set XATTR_NO_SPILL_OUT when creating new files.
 684
 685     Fixes: #8205
 686     Backport: firefly
 687
 688     Signed-off-by: Greg Farnum <greg@inktank.com>
 689     (cherry picked from commit e3b995e1567f3ccc6d00ae27ab2aa99ca157228a)
 690
 691 commit d6f302314fc3580dfc746c9d82bd2f7f0bc87d42
 692 Author: Haomai Wang <haomaiwang@gmail.com>
 693 Date:   Sat Jun 7 14:32:23 2014 +0800
 694
 695     FileStore: make _clone() copy spill out marker
 696
 697     Previously we were not doing so, and that resulted in unpredictable loss
 698     of xattrs from the client's perspective.
 699
 700     Signed-off-by: Haomai Wang <haomaiwang@gmail.com>
 701     Reviewed-by: Greg Farnum <greg@inktank.com>
 702     (cherry picked from commit 239476a92849159d2a8966d90ca055c116bee91e)
 703
 704 commit bb2ed23d8129c8db628475d744656b9b43f839c0
 705 Author: Loic Dachary <loic@dachary.org>
 706 Date:   Wed Jun 18 17:01:54 2014 +0200
 707
 708     erasure-code: verify that rados put enforces alignment
 709
 710     http://tracker.ceph.com/issues/8622 refs: #8622
 711
 712     Signed-off-by: Loic Dachary <loic@dachary.org>
 713     (cherry picked from commit b46c4056014dd6de5e3bd736f2c41f096ea708b4)
 714
 715 commit dece65064d949b5afcc359cd408615883b5e002a
 716 Author: Lluis Pamies-Juarez <lluis@pamies.cat>
 717 Date:   Wed Jun 18 10:00:09 2014 -0700
 718
 719     enforce rados put aligment
 720
 721     Signed-off-by: Lluis Pamies-Juarez <lluis.pamies-juarez@hgst.com>
 722     (cherry picked from commit 304b08a23a3db57010078046955a786fe3589ef8)
 723
 724 commit 203318b8eb3753723277ec7a459239c81097f090
 725 Author: Sage Weil <sage@inktank.com>
 726 Date:   Fri Jun 6 13:31:29 2014 -0700
 727
 728     osd/OSDMap: do not require ERASURE_CODE feature of clients
 729
 730     Just because an EC pool exists in the cluster does not mean tha tthe client
 731     has to support the feature:
 732
 733     1) The way client IO is initiated is no different for EC pools than for
 734        replicated pools.
 735     2) People may add an EC pool to an existing cluster with old clients and
 736        locking those old clients out is very rude when they are not using the
 737        new pool.
 738     3) The only direct client user of EC pools right now is rgw, and the new
 739        versions already need to support various other features like CRUSH_V2
 740        in order to work.  These features are present in new kernels.
 741
 742     Fixes: #8556
 743     Backport: firefly
 744     Signed-off-by: Sage Weil <sage@inktank.com>
 745     (cherry picked from commit 3fe1699f9620280d0070cfe6f01cfeb2332e7470)
 746
 747 commit 6340b14b9c9a9c786b5448f2ea2a5556055a02e9
 748 Author: Sage Weil <sage@inktank.com>
 749 Date:   Thu Jun 12 16:44:53 2014 -0700
 750
 751     osd/OSDMap: make get_features() take an entity type
 752
 753     Make the helper that returns what features are required of the OSDMap take
 754     an entity type argument, as the required features may vary between
 755     components in the cluster.
 756
 757     Backport: firefly
 758     Signed-off-by: Sage Weil <sage@inktank.com>
 759     (cherry picked from commit 250677c965365edf3ecd24ef73700fc6d992ea42)
 760
 761 commit 400fa88f6cf8b1e5b4b15d9e36b1de8c3d1c3970
 762 Author: Haomai Wang <haomaiwang@gmail.com>
 763 Date:   Wed May 21 18:12:22 2014 +0800
 764
 765     Avoid extra check for clean object
 766
 767     We needn't to check clean object via buffer state, skip the clean object.
 768
 769     Signed-off-by: Haomai Wang <haomaiwang@gmail.com>
 770     (cherry picked from commit f51e33bd9c5a8e1cfc7065b30785696dc45918bc)
 771
 772 commit 3ad451ef57e2ab3fc644c94d44c2de7a17cced45
 773 Author: Accela Zhao <accelazh@gmail.com>
 774 Date:   Wed Jun 18 17:17:03 2014 +0800
 775
 776     Make <poolname> in "ceph osd tier --help" clearer.
 777
 778     The ceph osd tier --help info on the left always says <poolname>.
 779     It is unclear which one to put <tierpool> on the right.
 780
 781     $ceph osd tier --help
 782     osd tier add <poolname> <poolname> {--   add the tier <tierpool> to base pool
 783     force-nonempty}                          <pool>
 784     osd tier add-cache <poolname>            add a cache <tierpool> of size <size>
 785     <poolname> <int[0-]>                     to existing pool <pool>
 786     ...
 787
 788     This patch modifies description on the right to tell which <poolname>:
 789
 790     osd tier add <poolname> <poolname> {--   add the tier <tierpool> (the second
 791      force-nonempty}                          one) to base pool <pool> (the first
 792                                                one)
 793     ...
 794
 795     Fix: http://tracker.ceph.com/issues/8256
 796
 797     Signed-off-by: Yilong Zhao <accelazh@gmail.com>
 798
 799 commit f290a021bc8ec31c18b91cb37323345692f2f198
 800 Merge: 5722419 d02d46e
 801 Author: Sage Weil <sage@inktank.com>
 802 Date:   Mon Jun 16 09:25:32 2014 -0700
 803
 804     Merge pull request #1962 from dachary/wip-8599-ruleset-firefly
 805
 806     mon: pool set <pool> crush_ruleset must not use rule_exists (firefly)
 807
 808     Reviewed-by: Sage Weil <sage@inktank.com>
 809
 810 commit d02d46e25080d5f7bb8ddd4874d9019a078b816b
 811 Author: John Spray <john.spray@inktank.com>
 812 Date:   Tue May 20 16:50:18 2014 +0100
 813
 814     mon: pool set <pool> crush_ruleset must not use rule_exists
 815
 816     Implement CrushWrapper::ruleset_exists that iterates over the existing
 817     rulesets to find the one matching the ruleset argument.
 818
 819     ceph osd pool set <pool> crush_ruleset must not use
 820     CrushWrapper::rule_exists, which checks for a *rule* existing, whereas
 821     the value being set is a *ruleset*. (cherry picked from commit
 822     fb504baed98d57dca8ec141bcc3fd021f99d82b0)
 823
 824     A test via ceph osd pool set data crush_ruleset verifies the ruleset
 825     argument is accepted.
 826
 827     http://tracker.ceph.com/issues/8599 fixes: #8599
 828
 829     Backport: firefly, emperor, dumpling
 830     Signed-off-by: John Spray <john.spray@inktank.com>
 831     Signed-off-by: Loic Dachary <loic@dachary.org>
 832
 833 commit 572241942706e2f845a147f19717bf6c487f23f4
 834 Author: Sage Weil <sage@inktank.com>
 835 Date:   Sun Jun 8 20:18:49 2014 -0700
 836
 837     init-ceph: continue after failure doing osd data mount
 838
 839     If we are starting many daemons and hit an error, we normally note it and
 840     move on.  Do the same when doing the pre-mount step.
 841
 842     Fixes: #8554
 843     Signed-off-by: Sage Weil <sage@inktank.com>
 844     (cherry picked from commit 6a7e20147cc39ed4689809ca7d674d3d408f2a17)
 845
 846 commit e5a291ef69316519b105b57193753f0bf8ac8e62
 847 Author: Steve Taylor <steve@DES-U1404-STETAY.stc.local>
 848 Date:   Tue Jun 10 12:42:55 2014 -0600
 849
 850     Fix for bug #6700
 851
 852     When preparing OSD disks with colocated journals, the intialization process
 853     fails when using dmcrypt. The kernel fails to re-read the partition table after
 854     the storage partition is created because the journal partition is already in use
 855     by dmcrypt. This fix unmaps the journal partition from dmcrypt and allows the
 856     partition table to be read.
 857
 858     Signed-off-by: Stephen F Taylor <steveftaylor@gmail.com>
 859     (cherry picked from commit 673394702b725ff3f26d13b54d909208daa56d89)
 860
 861 commit 34d9b2bc659c4107b30e1ebe7a55315512a0222a
 862 Author: John Wilkins <john.wilkins@inktank.com>
 863 Date:   Thu Jun 5 11:29:20 2014 -0700
 864
 865     doc: Added Disable requiretty commentary.
 866
 867     Signed-off-by: John Wilkins <john.wilkins@inktank.com>
 868
 869 commit 54a89bab15a945d47b3f44448910521a2ebf41c6
 870 Author: Samuel Just <sam.just@inktank.com>
 871 Date:   Fri May 16 16:56:33 2014 -0700
 872
 873     ReplicatedPG::start_flush: fix clone deletion case
 874
 875     dsnapc.snaps will be non-empty most of the time if there
 876     have been snaps before prev_snapc.  What we really want to
 877     know is whether there are any snaps between oi.snaps.back()
 878     and prev_snapc.
 879
 880     Fixes: 8334
 881     Backport: firefly
 882     Signed-off-by: Samuel Just <sam.just@inktank.com>
 883     (cherry picked from commit 29f615b7ac9e92f77cdef9927070727fee9d5e33)
 884
 885 commit 308e2ef5c107bf04d3c5623f3c6bdffa10cf8c21
 886 Author: Samuel Just <sam.just@inktank.com>
 887 Date:   Mon May 12 15:08:07 2014 -0700
 888
 889     ReplicatedPG::start_flush: send delete even if there are no snaps
 890
 891     Even if all snaps for the clone have been removed, we still have to
 892     send the delete to ensure that when the object is recreated the
 893     new snaps aren't included in the wrong clone.
 894
 895     Signed-off-by: Samuel Just <sam.just@inktank.com>
 896     (cherry picked from commit 2ec2182745fa7c22526a7cf3dedb25bc314c9db4)
 897
 898 commit 8d04a7ece364c0a95b087990d6365401819b3efa
 899 Author: Samuel Just <sam.just@inktank.com>
 900 Date:   Thu May 15 20:53:27 2014 -0700
 901
 902     HashIndex: in cleanup, interpret missing dir as completed merge
 903
 904     If we stop between unlinking the empty subdir and removing the root
 905     merge marker, we get ENOENT on the get_info.  That's actually fine.
 906
 907     Backport: firefly
 908     Fixes: 8332
 909     Signed-off-by: Samuel Just <sam.just@inktank.com>
 910     (cherry picked from commit 5ff95dbdd2dbb533d344f37fea722ca4f140e670)
 911
 912 commit b2bf8cb21761d88557fb2d5c0935c4a3678ee01f
 913 Author: Alfredo Deza <alfredo@deza.pe>
 914 Date:   Wed May 28 11:48:12 2014 -0400
 915
 916     add backport of collections.Counter for python2.6
 917
 918     Using Raymond Hettinger's MIT backport
 919
 920     Signed-off-by: Alfredo Deza <alfredo.deza@inktank.com>
 921     (cherry picked from commit 23b75b550507438c79b3aa75e06721e5f7b134a4)
 922
 923 commit 0b00443eec4c372f2ef245f85dc7c1f497ba3778
 924 Author: Ailing <ailzhang@users.noreply.github.com>
 925 Date:   Wed May 28 12:37:48 2014 -0700
 926
 927     rest-api: key missing for per "rx" and "rwx"
 928
 929     commit 85a1cf31e6 and db266a3fb2 introduce new per "rx" and "rwx", but key missing for per "rx" and "rwx" in permmap
 930
 931     Signed-off-by: Ailing Zhang <ailzhang@cisco.com>
 932     (cherry picked from commit 0b5a67410793ec28cac47e6e44cbbcf5684d77e7)
 933
 934 commit a898ee56e01f971c12c0f7bebc22e59f850f27c0
 935 Author: Greg Farnum <greg@inktank.com>
 936 Date:   Wed May 21 21:41:23 2014 -0700
 937
 938     cephfs-java: build against older jni headers
 939
 940     Older versions of the JNI interface expected non-const parameters
 941     to their memory move functions. It's unpleasant, but won't actually
 942     change the memory in question, to do a cast_const in order to satisfy
 943     those older headers. (And even if it *did* modify the memory, that
 944     would be okay given our single user.)
 945
 946     Signed-off-by: Greg Farnum <greg@inktank.com>
 947     (cherry picked from commit 4d4b77e5b6b923507ec4a0ad9d5c7018e4542a3c)
 948
 949 commit 8a12a940817d957063c21ac71a6128f991801573
 950 Author: Ilya Dryomov <ilya.dryomov@inktank.com>
 951 Date:   Fri May 16 19:03:13 2014 +0400
 952
 953     OSDMonitor: set next commit in mon primary-affinity reply
 954
 955     Commit 8c5c55c8b47e ("mon: set next commit in mon command replies")
 956     fixed MMonCommand replies to include the right version, but the
 957     primary-affinity handler was authored before that.  Fix it.
 958
 959     Backport: firefly
 960     Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com>
 961     (cherry picked from commit a78b14ec1769ef37bef82bfda6faabb581b4cd7d)
 962
 963 commit 6ba57da94a6e00040677d6e43abf10a36ed9a931
 964 Author: Dmitry Smirnov <onlyjob@member.fsf.org>
 965 Date:   Mon May 12 14:08:44 2014 +1000
 966
 967     prioritise use of `javac` executable (gcj provides it through alternatives).
 968
 969      On Debian this fixes FTBFS when gcj-jdk and openjdk-7-jdk are installed at
 970      the same time because build system will use default `javac` executable
 971      provided by current JDK through `update-alternatives` instead of blindly
 972      calling GCJ when it is present.
 973
 974     Signed-off-by: Dmitry Smirnov <onlyjob@member.fsf.org>
 975     (cherry picked from commit 8b682d167e4535df582f1c77542e2b1ea0981228)
 976
 977 commit 9e335a494c6eff209cff64bd663b5aa2a7bbc242
 978 Author: Dmitry Smirnov <onlyjob@member.fsf.org>
 979 Date:   Mon May 12 14:02:53 2014 +1000
 980
 981     pass '-classpath' option (gcj/javah ignores CLASSPATH environment variable).
 982
 983      This should not affect OpenJDK which understands '-classpath' as well.
 984
 985      With gcj-jdk we still get FTBFS later:
 986
 987      ~~~~
 988         java/native/libcephfs_jni.cc:2878:55: error: invalid conversion from 'const jbyte* {aka const signed char*}' to 'jbyte* {aka signed char*}' [-fpermissive]
 989                      reinterpret_cast<const jbyte*>(rawAddress));
 990                                                                ^
 991         In file included from java/native/libcephfs_jni.cc:27:0:
 992         /usr/lib/gcc/x86_64-linux-gnu/4.8/include/jni.h:1471:8: error:   initializing argument 4 of 'void _Jv_JNIEnv::SetByteArrayRegion(jbyteArray, jsize, jsize, jbyte*)' [-fpermissive]
 993            void SetByteArrayRegion (jbyteArray val0, jsize val1, jsize val2, jbyte * val3)
 994                 ^
 995         make[5] *** [java/native/libcephfs_jni_la-libcephfs_jni.lo] Error 1
 996      ~~~~
 997
 998     Signed-off-by: Dmitry Smirnov <onlyjob@member.fsf.org>
 999     (cherry picked from commit 89fe0353582bde7e2fbf32f1626d430a20002dd0)
1000
1001 commit fcd30500790eac0266ffb70d83ee17b23aeaf209
1002 Author: Dmitry Smirnov <onlyjob@member.fsf.org>
1003 Date:   Mon May 12 13:57:20 2014 +1000
1004
1005     look for "jni.h" in gcj-jdk path, needed to find "jni.h" with gcj-jdk_4.9.0
1006
1007     Signed-off-by: Dmitry Smirnov <onlyjob@member.fsf.org>
1008     (cherry picked from commit 0f4120c0115e7977ae7c03458addcc2b2916db07)
1009
1010 commit de43f6d2d813d2c7e5cf915c1b9037fe662d6afc
1011 Author: Sage Weil <sage@inktank.com>
1012 Date:   Thu May 8 08:52:51 2014 -0700
1013
1014     ceph-disk: partprobe before settle when preparing dev
1015
1016     Two users have reported this fixes a problem with using --dmcrypt.
1017
1018     Fixes: #6966
1019     Tested-by: Eric Eastman <eric0e@aol.com>
1020     Signed-off-by: Sage Weil <sage@inktank.com>
1021     (cherry picked from commit 0f196265f049d432e399197a3af3f90d2e916275)
1022
1023 commit fe1fbfb7a43d6590efce99711e47f986249f0956
1024 Author: Greg Farnum <greg@inktank.com>
1025 Date:   Tue May 13 13:15:28 2014 -0700
1026
1027     test: fix some templates to match new output code
1028
1029     Signed-off-by: Greg Farnum <greg@inktank.com>
1030     Reviewed-by: Samuel Just <sam.just@inktank.com>
1031     (cherry picked from commit 00225d739cefa1415524a3de45fb9a5a2db53018)
1032
1033 commit 5c8afaa8861345efbcc5488e0336327a8a38d3bc
1034 Author: Greg Farnum <greg@inktank.com>
1035 Date:   Thu May 15 16:50:43 2014 -0700
1036
1037     OSD: fix an osdmap_subscribe interface misuse
1038
1039     When calling osdmap_subscribe, you have to pass an epoch newer than the
1040     current map's. _maybe_boot() was not doing this correctly -- we would
1041     fail a check for being *in* the monitor's existing map range, and then
1042     pass along the map prior to the monitor's range. But if we were exactly
1043     one behind, that value would be our current epoch, and the request would
1044     get dropped. So instead, make sure we are not *in contact* with the monitor's
1045     existing map range.
1046
1047     Signed-off-by: Greg Farnum <greg@inktank.com>
1048     Reviewed-by: Samuel Just <sam.just@inktank.com>
1049     (cherry picked from commit 290ac818696414758978b78517b137c226110bb4)
1050
1051 commit eb98f4ab3a27aa8e00e0eb7e504ca7bec7c48a06
1052 Author: Sage Weil <sage@inktank.com>
1053 Date:   Mon May 19 10:32:12 2014 -0700
1054
1055     osd: skip out of order op checks on tiered pools
1056
1057     When we send redirected ops, we do not assign a new tid, which means that
1058     a given client's ops for a pool may not have strictly ordered tids.  Skip
1059     this check if the pool is tiered to avoid false positives.
1060
1061     Fixes: #8380
1062     Signed-off-by: Sage Weil <sage@inktank.com>
1063     (cherry picked from commit cf2b172c843da0599164901956b66c306a59e570)
1064
1065 commit 29ee6faecb9e16c63acae8318a7c8f6b14367af7
1066 Author: Samuel Just <sam.just@inktank.com>
1067 Date:   Tue May 6 11:50:14 2014 -0700
1068
1069     ReplicatedPG: block scrub on blocked object contexts
1070
1071     Fixes: #8011
1072     Backport: firefly
1073     Signed-off-by: Samuel Just <sam.just@inktank.com>
1074     (cherry picked from commit 7411477153219d66625a74c5886530029c516036)
1075
1076 commit 6078726e0b89d0202c3a4bc594be11fc346bed9d
1077 Author: Guang Yang <yguang@yahoo-inc.com>
1078 Date:   Fri May 9 09:21:23 2014 +0000
1079
1080     msg: Fix inconsistent message sequence negotiation during connection reset
1081
1082     Backport: firefly, emperor, dumpling
1083
1084     Signed-off-by: Guang Yang (yguang@yahoo-inc.com)
1085     Reviewed-by: Greg Farnum <greg@inktank.com>
1086     (cherry picked from commit bdee119076dd0eb65334840d141ccdf06091e3c9)
1087
1088 commit 02557f8bc05ff28013e241611cb046b28f97b4c6
1089 Author: Samuel Just <sam.just@inktank.com>
1090 Date:   Tue Apr 15 12:55:47 2014 -0700
1091
1092     Objecter::_op_submit: only replace the tid if it's 0
1093
1094     Otherwise, redirected ops will suddenly have a different tid
1095     and will become uncancelable.
1096
1097     Fixes: #7588
1098     Signed-off-by: Samuel Just <sam.just@inktank.com>
1099     (cherry picked from commit 76568aa0db4e16ac1af8fe6405edade1e61cbc81)
1100
1101 commit 788b5fad515c48d4028f5a67d63052e373c81e66
1102 Author: Sage Weil <sage@inktank.com>
1103 Date:   Thu May 8 10:42:42 2014 -0700
1104
1105     mon/OSDMonitor: force op resend when pool overlay changes
1106
1107     If a client is sending a sequence of ops (say, a, b, c, d) and partway
1108     through that sequence it receives an OSDMap update that changes the
1109     overlay, the ops will get send to different pools, and the replies will
1110     come back completely out of order.
1111
1112     To fix this, force a resend of all outstanding ops any time the overlay
1113     changes.
1114
1115     Fixes: #8305
1116     Signed-off-by: Sage Weil <sage@inktank.com>
1117     (cherry picked from commit 63d92ab0969357f78fdade749785136a509bc81b)
1118
1119 commit 268df4862d95bf506ef6f594fdfa8294671babc3
1120 Author: Sage Weil <sage@inktank.com>
1121 Date:   Thu May 8 10:50:51 2014 -0700
1122
1123     osd: discard client ops sent before last_force_op_resend
1124
1125     If an op is sent before last_force_op_resend, and the client's feature is
1126     present, drop the op because we know they will resend.
1127
1128     Signed-off-by: Sage Weil <sage@inktank.com>
1129     (cherry picked from commit 45e79a17a932192995f8328ae9f6e8a2a6348d10)
1130
1131 commit 8bc4992f434ed0b180d5e4a293d8191f4846358c
1132 Author: Sage Weil <sage@inktank.com>
1133 Date:   Thu May 8 10:52:11 2014 -0700
1134
1135     osdc/Objecter: resend ops in the last_force_op_resend epoch
1136
1137     If we are a client, and process a map that sets last_force_op_resend to
1138     the current epoch, force a resend of this op.
1139
1140     If the OSD expects us to do this, it will discard our previous op.  If the
1141     OSD is old, it will process the old one, this will appear as a dup, and we
1142     are no worse off than before.
1143
1144     Signed-off-by: Sage Weil <sage@inktank.com>
1145     (cherry picked from commit dd700bdf7115223cb3e517b851f462d75dd76a2b)
1146
1147 commit 83d03f0e1c65385dcf465108d544da84f8bfad97
1148 Author: Sage Weil <sage@inktank.com>
1149 Date:   Thu May 8 10:40:10 2014 -0700
1150
1151     osd/osd_types: add last_force_op_resend to pg_pool_t
1152
1153     Signed-off-by: Sage Weil <sage@inktank.com>
1154     (cherry picked from commit 3152faf79f498a723ae0fe44301ccb21b15a96ab)
1155
1156 commit c94ac44ab41fc62add3fc93c5c7ce06950eed3f1
1157 Author: Sage Weil <sage@inktank.com>
1158 Date:   Fri May 9 09:20:34 2014 -0700
1159
1160     osd: handle race between osdmap and prepare_to_stop
1161
1162     If we get a MOSDMarkMeDown message and set service.state == STOPPING, we
1163     kick the prepare_to_stop() thread.  Normally, it will wake up and then
1164     set osd.state == STOPPING, and when we process the map message next we
1165     will not warn.  However, if dispatch() takes the lock instead and processes
1166     the map, it will fail the preparing_to_stop check and issue a spurious
1167     warning.
1168
1169     Fix by checking for either preparing_to_stop or stopping.
1170
1171     Fixes: #8319
1172     Backport: firefly, emperor, dumpling
1173     Signed-off-by: Sage Weil <sage@inktank.com>
1174     (cherry picked from commit 6b858be0676f937a99dbd51321497f30c3a0097f)
1175
1176 commit 22e2e7cff42970b00dc352513f1f138132f78d04
1177 Author: Sage Weil <sage@inktank.com>
1178 Date:   Sat May 10 10:29:11 2014 -0700
1179
1180     osd/ReplicatedPG: do not queue NULL dup_op
1181
1182     We call start_flush() with a NULL op in a couple different places.  Do not
1183     put a NULL pointer on the dup_ops list or we will crash later.
1184
1185     Fixes: #8328
1186     Signed-off-by: Sage Weil <sage@inktank.com>
1187     (cherry picked from commit 0d67f9b0695765824bdc4a65fbed88edf8ea232e)