Commit Graph

51638 Commits

Author SHA1 Message Date
Sage Weil
359d8324a3 qa/workunits/cephtool/test.sh: fix cache mode tests
Signed-off-by: Sage Weil <sage@redhat.com>
2016-03-28 16:57:42 -04:00
Sage Weil
90fe8e3d0b mon/OSDMonitor: require force flag to use obscure cache modes
Only none, writeback, proxy, and readproxy are robust and
supported.  Require a force flag to use any of the others.

Signed-off-by: Sage Weil <sage@redhat.com>
2016-03-28 16:57:28 -04:00
Sage Weil
d7da68848a osd: add 'proxy' cache mode
The 'forward' mode is flawed for the same reason that redirects
are flawed: they allow the reordering of requests.  Add a proxy
mode that can be used instead of forward.  (Later we can consider
dropping forward entirely?)

Fixes: #12814
Signed-off-by: Sage Weil <sage@redhat.com>
2016-03-28 16:57:27 -04:00
Casey Bodley
7f441b1065 Merge pull request #8321 from linuxbox2/jewel-nfs-fixes-2
rgw_file: set owner uid, gid, and Unix mode on new objects
2016-03-28 15:29:41 -04:00
Casey Bodley
cab2b2d2d0 Merge pull request #8106 from linuxbox2/wip-cmake-ssl
cmake: add FindOpenSSL.cmake
2016-03-28 13:24:35 -04:00
Loic Dachary
7327b1ad3d Merge pull request #8312 from SUSE/wip-submittingpatches
doc: amend Fixes instructions in SubmittingPatches

Reviewed-by: Loic Dachary <ldachary@redhat.com>
2016-03-28 19:13:02 +02:00
Matt Benjamin
13bd851840 rgw_file tests: allow override of owner_uid and owner_gid
Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
2016-03-28 12:05:38 -04:00
Casey Bodley
516ad93b2d Merge pull request #8284 from yehudasa/wip-rgw-sync-fixes-5
rgw: fix lockdep false positive
2016-03-28 11:42:05 -04:00
Sage Weil
b83f339f1d Merge pull request #7820 from yuyuyu101/remove-exper
msg/async: remove experiment feature

Reviewed-by: Sage Weil <sage@redhat.com>
2016-03-28 10:48:42 -04:00
Sage Weil
cd3385b59f Merge pull request #8308 from wido/cloudstack-docs
doc: Updated CloudStack RBD documentation
2016-03-28 08:28:45 -04:00
Sage Weil
f394486109 Merge pull request #8327 from vumrao/wip-vumrao-15288
doc: Remove Ceph Monitors do lots of fsync()
2016-03-28 08:23:58 -04:00
Sage Weil
336b082415 Merge pull request #8328 from scienceluo/master
doc: Fixes headline different font size and type
2016-03-28 08:23:10 -04:00
Sage Weil
495e661e4e Merge pull request #8293 from dachary/wip-subman-facts
script: subscription-manager support (part 2)

Reviewed-by: Kefu Chai <kchai@redhat.com>
2016-03-28 08:15:59 -04:00
Vikhyat Umrao
1611dfbf17 doc: Remove Ceph Monitors do lots of fsync()
and change the ligature of "fl" to "f" and "l"

Fixes: #15288

Signed-off-by: Vikhyat Umrao <vumrao@redhat.com>
2016-03-28 13:29:09 +05:30
scienceluo
b3930c51ff Fixes headline different font size and type 2016-03-28 14:10:28 +08:00
Matt Benjamin
81792b3647 rgw_file: set owner uid, gid, and Unix mode on new objects
Unix attributes aren't current materialized, but a future change
will serialize them in a new object attribute.

Also moves create and mkdir into operations on RGWLibFS, which
better encapsulates the behavior.

Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
2016-03-27 17:20:27 -04:00
Sage Weil
eeaab88e31 Merge pull request #8299 from liewegas/wip-15230
messages/MOSDOp: clear reqid inc for v6 encoding
2016-03-27 10:02:26 -04:00
Sage Weil
772ad242fa Merge pull request #8318 from tchaikov/wip-15240
test/system/*: use dynamically generated pool name

Reviewed-by: Sage Weil <sage@redhat.com>
2016-03-27 10:01:14 -04:00
Kefu Chai
239f164a02 test/system/*: use dynamically generated pool name
was using "foo" for the pool name. this works if we are performing
tests in serial. but if we do it in parallel, the tests interfere
with each other.

Fixes: #15240
Signed-off-by: Kefu Chai <kchai@redhat.com>
2016-03-27 21:21:55 +08:00
Sage Weil
77027813ed Merge pull request #7094 from songbaisen/wipdf
mon: show the pool quota info on ceph df detail command

Reviewed-by: Abhishek Lekshmanan <abhishek@suse.com>
2016-03-26 16:17:14 -04:00
Sage Weil
219995c45c Merge pull request #8275 from xiexingguo/xxg-wip-fixdirtyinfo
osd: fix dirtying info without correctly setting drity_info field

Reviewed-by: Sage Weil <sage@redhat.com>
2016-03-26 16:16:26 -04:00
Sage Weil
c813fea92b Merge pull request #8230 from majianpeng/clear-omap
osd: fix bugs for omap ops

Reviewed-by: Kefu Chai <kchai@redhat.com>
Reviewed-by: Sage Weil <sage@redhat.com>
2016-03-26 16:16:09 -04:00
Sage Weil
77cec5daa9 Merge pull request #8254 from xiexingguo/xxg-wip-osd
osd: fix reference count, rare race condition etc.

Reviewed-by: Kefu Chai <kchai@redhat.com>
2016-03-26 16:14:02 -04:00
Sage Weil
c972d0cd42 Merge pull request #8291 from xiexingguo/xxg-wip-fixhistorymerge
osd: more fixes for incorrectly dirtying info; resend reply for duplicated scrub-reserve req

Reviewed-by: Sage Weil <sage@redhat.com>
2016-03-26 16:13:42 -04:00
Sage Weil
fca1c707d0 Merge pull request #8247 from liewegas/wip-split-creates
mon: do not send useless pg_create messages for split pgs

Reviewed-by: Kefu Chai <kchai@redhat.com>
2016-03-26 16:12:29 -04:00
Sage Weil
63c00cb305 Merge pull request #8208 from liewegas/wip-15027
mon: mark_down_pgs in lockstep with pg_map's osdmap epoch

Reviewed-by: Kefu Chai <kchai@redhat.com>
2016-03-26 16:11:41 -04:00
Loic Dachary
0ab4813092 script: subscription-manager support (part 2)
Strip the .py suffix so that no pyc / pyo files are generated.
Do not install on CentOS.

http://tracker.ceph.com/issues/14972 Fixes: #14972

Signed-off-by: Loic Dachary <loic@dachary.org>
2016-03-26 19:20:57 +01:00
Sage Weil
1cbe2bd9d4 ceph_test_rados_api_misc: debug LibRadosMiscConnectFailure.ConnectFailure
Signed-off-by: Sage Weil <sage@redhat.com>
2016-03-26 09:47:51 -04:00
Sage Weil
bdca28ddfb messages/MOSDOp: clear reqid inc for v6 encoding
For the new v7 encoding, we put the client incarnation in the
reqid so that we can have it at the beginning of the encoding.
However, we do *not* want to do this for the v6 encoding, as that
version of get_reqid() will return reqid if it is not ==
osd_reqid_t().

Fixes: #15230
Signed-off-by: Sage Weil <sage@redhat.com>
2016-03-25 21:56:10 -04:00
Sage Weil
67a61962db Merge pull request #8265 from athanatos/wip-journal-default
config_opts: disable filestore throttle soft backoff by default
2016-03-25 21:21:36 -04:00
Sage Weil
cbec91d2d4 Merge pull request #8306 from tchaikov/wip-doc-formatting
doc/rados/operations/crush: fix the formatting
2016-03-25 16:28:42 -04:00
Sage Weil
91d233e840 Merge pull request #8202 from XinzeChi/wip-objecter-tid
osdc/Objecter: dout log after assign tid

Reviewed-by: Kefu Chai <kchai@redhat.com>
2016-03-25 16:10:35 -04:00
Sage Weil
a0e5aa5c8a Merge remote-tracking branch 'gh/jewel' 2016-03-25 15:49:23 -04:00
Kefu Chai
e82dd05caa Merge pull request #8310 from tchaikov/wip-fix-rbdmap-manpage
makefile: fix rbdmap manpage

Reviewed-by: Sage Weil <sage@redhat.com>
2016-03-26 01:23:41 +08:00
Nathan Cutler
65770059cb doc: amend Fixes instructions in SubmittingPatches
Since the "Fixes: #..." line is interpreted by GitHub as referring to a
pull request, yet the intention is for it to refer to the tracker issue,
change our instructions to use the full tracker issue URL.

Now that github pull request numbers are in the 8000s, there are cases when
a PR is wrongly closed by a "Fixes: #..." line in a commit message. See

https://github.com/ceph/ceph/pull/8286#event-603120191

for one such case.

Signed-off-by: Nathan Cutler <ncutler@suse.com>
2016-03-25 17:39:39 +01:00
Kefu Chai
3b551e6db2 cmake: conditionally install man pages
this matches its counterpart in automake
* install fuse related man pages only if HAVE_LIBFUSE
* install rbd related man pages only if WITH_RBD

Signed-off-by: Kefu Chai <kchai@redhat.com>
2016-03-26 00:03:19 +08:00
Kefu Chai
1287846c66 autotools: add rbdmap to dist_man_MANS
this fixes the rpm/deb build failure
was introduced by 20c8650

Signed-off-by: Kefu Chai <kchai@redhat.com>
2016-03-25 23:53:52 +08:00
Sage Weil
1ac8b5f972 Merge pull request #8038 from badone/wip-crushtool-crash
crushtool: Don't crash when called on a file that isn't a crushmap

Reviewed-by: Kefu Chai <kchai@redhat.com>
2016-03-25 11:22:43 -04:00
Kefu Chai
816c902ee5 Merge pull request #8309 from tchaikov/wip-15240
test/system/rados_list_parallel: print oid if rados_write fails

Reviewed-by: Sage Weil <sage@redhat.com>
2016-03-25 23:21:31 +08:00
Sage Weil
9f47a5b68e xxHash: fix submodule commit
This was inadvertantly rewinded in b9a14312ed

Signed-off-by: Sage Weil <sage@redhat.com>
2016-03-25 11:10:06 -04:00
xie xingguo
294cab1171 osd/pg: publish stats to osd if pg is currently marked as down
Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
2016-03-25 22:14:24 +08:00
xie xingguo
c0ceb8b7a6 osd/pg: publish stats to osd if we recovered from degraded
Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
2016-03-25 22:14:23 +08:00
xie xingguo
c6a4f34940 osd/pg: avoid publishing pg_stat twice in a short interval
Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
2016-03-25 22:14:21 +08:00
xie xingguo
be3ebd8fe5 osd/pg: fold sanity check into one single "if" qualification
Just as "boost::statechart::result
PG::RecoveryState::WaitRemoteBackfillReserved::react(const RemoteBackfillReserved &evt)"
does.

Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
2016-03-25 22:14:20 +08:00
xie xingguo
81d692922b osd/pg: drop unnecessary update_heartbeat_peers()
The proc_replica_info() method will do the same job in a
smarter way, so the succeeding call to update_heartbeat_peers()
is unnecessary and cost expensive.

Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
2016-03-25 22:14:19 +08:00
Sage Weil
a431d04acb Merge pull request #8211 from liewegas/wip-jewel
doc: draft notes for jewel
2016-03-25 09:55:30 -04:00
Sage Weil
bc3dcd22bb Merge pull request #8044 from dzafman/wip-8885
osd: fix dump_ops_in_flight races

Reviewed-by: Samuel Just <sjust@redhat.com>
2016-03-25 09:53:56 -04:00
Sage Weil
7a199ba536 Merge pull request #8156 from liewegas/wip-pg-query
osd/PG: indicate in pg query output whether ignore_history_les would help

Revewed-by: Samuel Just <sjust@redhat.com>
2016-03-25 09:52:35 -04:00
Sage Weil
badb37ddd7 Merge pull request #8177 from majianpeng/signal-handle
global/signal_handler: print thread name in signal handler

Reviewed-by: Kefu Chai <kchai@redhat.com>
2016-03-25 09:52:21 -04:00
Kefu Chai
71c41be3b1 test/system/rados_list_parallel: print oid if rados_write fails
Signed-off-by: Kefu Chai <kchai@redhat.com>
2016-03-25 21:43:24 +08:00