Commit Graph

52238 Commits

Author SHA1 Message Date
Greg Farnum
6974ed4254 doc: update cephfs administration page for enable_multiple safety check
Signed-off-by: Greg Farnum <gfarnum@redhat.com>
2016-04-07 16:40:24 -07:00
Greg Farnum
4b3802ed54 mdsmonitor: unify experimental warnings
Signed-off-by: Greg Farnum <gfarnum@redhat.com>
2016-04-07 16:40:23 -07:00
Greg Farnum
ea1b276e7e test: update tests for new flag requirements
Signed-off-by: Greg Farnum <gfarnum@redhat.com>
2016-04-07 16:40:23 -07:00
Greg Farnum
4bc4e74932 MDSMap: lock out multimds clusters and directory fragmentation by default
Add new flags operating the way the snapshot lockout does. Then, actually
block out the directory fragmentation in the MDS; and block out setting
max_mds > 1 in the MDSMonitor.

Signed-off-by: Greg Farnum <gfarnum@redhat.com>
2016-04-07 16:40:23 -07:00
Greg Farnum
15d1fd48ba MDSMap: switch from booleans to flags for feature enablement logging
Signed-off-by: Greg Farnum <gfarnum@redhat.com>
2016-04-07 16:40:23 -07:00
Greg Farnum
cda1c1a984 FSMap: add output for enabled_multiple flags
Signed-off-by: Greg Farnum <gfarnum@redhat.com>
2016-04-07 16:40:22 -07:00
Greg Farnum
313013209b MDSMonitor: make enabling multiple FSes a lot scarier
Output a more explicit warning when the user tries to enable multiple FSes, and
mark the FSMap when the user does so anyway.

Signed-off-by: Greg Farnum <gfarnum@redhat.com>
2016-04-07 16:40:22 -07:00
Greg Farnum
3b9371aa2f FSMap: print out compat instead of erroneously repeating enable_multiple
Signed-off-by: Greg Farnum <gfarnum@redhat.com>
2016-04-07 16:40:22 -07:00
Greg Farnum
2429463ea5 ceph-dencoder: add FSMap
Signed-off-by: Greg Farnum <gfarnum@redhat.com>
2016-04-07 16:40:22 -07:00
Matt Benjamin
15cabdc537 rgw_ldap: make ldap.h inclusion conditional
The feature implementation is conditional, embarassingly, the
header inclusion is not.

Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
2016-04-07 19:39:40 -04:00
Orit Wasserman
d58e5fee64 rgw: fix problem deleting objects begining with double underscore
Fixes: http://tracker.ceph.com/issues/15318

Signed-off-by: Orit Wasserman <owasserm@redhat.com>
2016-04-08 00:48:51 +02:00
Josh Durgin
dfd0856490 Merge pull request #8493 from dillaman/wip-15354
test: fix valgrind memcheck issues for rbd-mirror test cases

Reviewed-by: Josh Durgin <jdurgin@redhat.com>
2016-04-07 14:49:22 -07:00
Josh Durgin
f00ea3f898 Merge pull request #8490 from dillaman/wip-15097
rbd: journal reset should disable/re-enable journaling feature

Reviewed-by: Josh Durgin <jdurgin@redhat.com>
2016-04-07 14:11:10 -07:00
Josh Durgin
43ac60a14c Merge pull request #8477 from dillaman/wip-15364
journal: fix race condition between Future and journal shutdown

Reviewed-by: Josh Durgin <jdurgin@redhat.com>
2016-04-07 14:07:51 -07:00
Jason Dillaman
492a572d60 test: fix memory leaks in rbd-mirror test cases
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
2016-04-07 16:40:45 -04:00
Jason Dillaman
acfc2b182a test: avoid leaking librados connections when creating pools
Fixes: http://tracker.ceph.com/issues/15354
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
2016-04-07 16:39:10 -04:00
Jason Dillaman
88e244c839 rbd: journal reset should disable/re-enable journaling feature
Fixes: http://tracker.ceph.com/issues/15097
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
2016-04-07 15:39:06 -04:00
Samuel Just
b69830fd24 mon: warn if 'sortbitwise' flag is not set and no legacy OSDs are presentReviewed-by: Samuel Just <sjust@redhat.com>
mon: warn if 'sortbitwise' flag is not set and no legacy OSDs are present

Reviewed-by: Samuel Just <sjust@redhat.com>
2016-04-07 11:05:47 -07:00
Josh Durgin
9835166b41 Merge pull request #8476 from dillaman/wip-rbd-mirror-lockdep
rbd-mirror: workaround for intermingled lockdep singletons

Reviewed-by: Ricardo Dias <rdias@suse.com>
Reviewed-by: Mykola Golub <mgolub@mirantis.com>
Reviewed-by: Josh Durgin <jdurgin@redhat.com>
2016-04-07 10:14:13 -07:00
Sage Weil
2fa4147084 osd/ReplicatedPG: clean up temp object if copy-from fails
Signed-off-by: Sage Weil <sage@redhat.com>
2016-04-07 11:49:08 -04:00
Sage Weil
a0bb575c19 ceph_test_rados_api_misc: make CopyFrom omap test be a big object
This makes it so we check single-shot path and the one where we have a
temp object to clean up.

Signed-off-by: Sage Weil <sage@redhat.com>
2016-04-07 11:49:08 -04:00
Sage Weil
4d36da6dcd Merge pull request #8456 from dreamhost/wip-authtool-help
authtool: update --help and manpage to match code.

Reviewed-by: Sage Weil <sage@redhat.com>
2016-04-07 11:44:30 -04:00
Sage Weil
66b7ae6610 Merge pull request #8446 from liewegas/wip-15366
osd: fix/clean up full map request handling

Reviewed-by: Kefu Chai <kchai@redhat.com>
2016-04-07 11:36:34 -04:00
Sage Weil
134416a021 Revert "rados: Add new field flags for ceph_osd_op.copy_get."
This reverts commit 2b4acfb1b8.

We fixed this on the receiving end in
fc51ce2a83 a few weeks after this commit
was added.  It's ugly and no longer needed.

This *does* mean that we'll start to copy some data and then abort
partway through, but that's better than an kludgey protocol/interface.

Signed-off-by: Sage Weil <sage@redhat.com>
2016-04-07 11:13:03 -04:00
Orit Wasserman
999e917e5e Merge pull request #8472 from jmunhoz/fix-aws4-subdomain-calling
rgw: aws4 subdomain calling bugfix
Reviewed-by: Orit Wasserman <owasserm@redhat.com>
2016-04-07 16:59:09 +02:00
Jason Dillaman
017d830848 rbd-mirror: workaround for intermingled lockdep singletons
If librados constructs the CephContext, lockdep will be initialized
in both the librados shared address space and also within the app
address space.  This breaks lockdep's functionality and results in
a crash upon shutdown.

Signed-off-by: Jason Dillaman <dillaman@redhat.com>
2016-04-07 10:36:30 -04:00
Sage Weil
98744fdf9b logrotate.conf: poke ceph-fuse after log rotation
This still leaves other loggers with un-reopened log files,
unfortunately.

Signed-off-by: Sage Weil <sage@redhat.com>
2016-04-07 10:27:17 -04:00
Sage Weil
91e0be047c ceph-fuse: reopen log file on SIGHUP
Signed-off-by: Sage Weil <sage@redhat.com>
2016-04-07 10:26:43 -04:00
Matt Benjamin
340fe95b56 Merge pull request #8460 from linuxbox2/jewel-nfs-fixes-3
Jewel nfs fixes 3
2016-04-07 10:26:11 -04:00
Jason Dillaman
60679fc53a librbd: restore out-of-band future callbacks to avoid lock cycles
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
2016-04-07 10:25:24 -04:00
renhwztetecs
e98d046a65 MDS: unregister command add in clean_up_admin_socket
unregister command add in MDSDaemon::clean_up_admin_socket

Signed-off-by: huanwen ren <ren.huanwen@zte.com.cn>
2016-04-07 18:46:28 +08:00
Javier M. Mellid
2f4bc8401d rgw: aws4 subdomain calling bugfix
Fixes: http://tracker.ceph.com/issues/15369

Signed-off-by: Javier M. Mellid <jmunhoz@igalia.com>
2016-04-07 12:39:34 +02:00
weiqiaomiao
dab0b55987 rgw: the map 'headers' is assigned a wrong value
Signed-off-by: weiqiaomiao <wei.qiaomiao@zte.com.cn>
2016-04-07 14:12:51 +08:00
Loic Dachary
c2fd18c00d Merge pull request #8457 from dreamhost/wip-virtualenv-tmpdir
build: Respect TMPDIR for virtualenv.

Reviewed-by: Loic Dachary <ldachary@redhat.com>
2016-04-07 07:24:45 +02:00
Robin H. Johnson
f01261f371 authtool: fix test output validation & minor whitespace.
Signed-off-by: Robin H. Johnson <robin.johnson@dreamhost.com>
2016-04-06 21:31:44 +00:00
Jason Dillaman
408964eb09 journal: fix context memory leak when shutting down live replay
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
2016-04-06 17:20:30 -04:00
Jason Dillaman
f931066cf5 journal: Future does not require metadata shared pointer
Avoid keeping the metadata object alive through the lifespan of the
future.  Callers can expect to receive a re-entrant callback if the
future is already safe and flush/wait is invoked.

Fixes: http://tracker.ceph.com/issues/15364
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
2016-04-06 17:18:33 -04:00
Sage Weil
d3dbd8581b mon: warn if 'sortbitwise' flag is not set and no legacy OSDs are present
If the user has upgraded, they should set sortbitwise.

Signed-off-by: Sage Weil <sage@redhat.com>
2016-04-06 17:03:53 -04:00
Sage Weil
297f5f2ce7 Merge pull request #8418 from liewegas/wip-blk-probe
os/ObjectStore: make device uuid probe output something friendly
2016-04-06 16:57:11 -04:00
Sage Weil
82e9f94523 Merge pull request #8428 from liewegas/wip-rest-mds
ceph-rest-api: fix fs/flag/set
2016-04-06 16:10:53 -04:00
Sage Weil
ed31ad64a4 Merge pull request #7981 from liewegas/wip-14364
osdc/Objecter: fix narrow race with tid assignment

Reviewed-by: Josh Durgin <jdurgin@redhat.com>
2016-04-06 15:52:10 -04:00
Sage Weil
b3d27f8ca7 Merge pull request #8403 from dx9/wip-ceph-dencoder-esessions-fix
mds: Add cmapv to ESessions default constructor initializer list

Reviewed-by: Sage Weil <sage@redhat.com>
2016-04-06 15:51:40 -04:00
Sage Weil
683a46cd5a Merge pull request #8419 from adamemerson/wip-32bit-time
common: fix time_t cast in decode

Reviewed-by: Kefu Chai <kchai@redhat.com>
Reviewed-by: Sage Weil <sage@redhat.com>
2016-04-06 15:49:22 -04:00
Sage Weil
78d4fada4b Merge pull request #8431 from liewegas/wip-bluestore
os/bluestore: revamp BlueFS bdev management and add perfcounters
2016-04-06 15:48:45 -04:00
Orit Wasserman
f64a9e3b42 Merge pull request #8445 from jmunhoz/fix-aws4-uri-encoding
rgw: aws4 uri encoding bugfix
Review-by: Orit Wasserman <owasserm@redhat.com>
2016-04-06 20:37:58 +02:00
John Spray
d544e447df mds: validate file layouts during replay
Instead of letting potentially bogus stuff into
the cache, go damaged().

Signed-off-by: John Spray <john.spray@redhat.com>
2016-04-06 16:23:16 +01:00
Sage Weil
7b1ed5dd14 Merge pull request #8450 from javacruft/tasksmax-infinity
systemd: drop any systemd imposed process/thread limits

Reviewed-by: Sage Weil <sage@redhat.com>
2016-04-06 09:36:15 -04:00
Sage Weil
9414befb89 debian/rules: include ceph-mds-*.conf upstart files in ceph-mds
These were lost by a typo in 0cbe3dea69

Fixes: http://tracker.ceph.com/issues/15395
Signed-off-by: Sage Weil <sage@redhat.com>
2016-04-06 08:55:49 -04:00
John Spray
45a0bc103c mds: add operator<< for file_layout_t
Signed-off-by: John Spray <john.spray@redhat.com>
2016-04-06 13:05:17 +01:00
John Spray
693f46a75a mds: skip size recovery on non-file inodes
Previously even directories were getting added to the
list of inodes to check in rejoin.  This was a (small)
waste of time, with these dir inodes getting passed
all the way down into check_inode_max_size for no purpose.

Signed-off-by: John Spray <john.spray@redhat.com>
2016-04-06 13:05:17 +01:00