Commit Graph

25479 Commits

Author SHA1 Message Date
Sage Weil
2f0e31f31e Merge pull request #196 from ceph/wip-mon-crush2
Wip mon crush2

Reviewed-by: Dan Mick <dan.mick@inktank.com>
2013-04-03 15:46:11 -07:00
Sage Weil
29f9db1980 mon: fix crush unit tests for idempotency
Signed-off-by: Sage Weil <sage@inktank.com>
2013-04-03 15:45:34 -07:00
Gregory Farnum
32aac00c70 Merge pull request #197 from ceph/wip-3266
mds: verify mds tell 'dumpcache <filename>' target does not exist

Reviewed-by: Greg Farnum <greg@inktank.com>
2013-04-03 15:42:22 -07:00
Sage Weil
336c4ae7ad mds: verify mds tell 'dumpcache <filename>' target does not exist
Open target with O_CREAT|O_EXCL to ensure we don't overwrite some other
important file (like, say, /etc/passwd).  This is irritating because there
is not c++ ofstream equivalent for O_EXCL; kludge around it using
ostringstream instead.

Fixes: #3266
Signed-off-by: Sage Weil <sage@inktank.com>
2013-04-03 15:32:51 -07:00
Sage Weil
e6d4582a68 mon: make 'osd crush unlink ..' idempotent
Signed-off-by: Sage Weil <sage@inktank.com>
2013-04-03 15:04:00 -07:00
Gregory Farnum
3c0debf99d Merge pull request #191 from ceph/wip-4582b
Fixes #4582.

Reviewed-by: Greg Farnum <greg@inktank.com>
2013-04-03 14:32:48 -07:00
Greg Farnum
0d6ddd9264 mds: do not go through handle_mds_failure for oneself
A standby MDS can attempt the handle_mds_failure paths for itself, if
it sees the transition from up to down. This leads it to insert itself
into the resolve_gather set, which is bad. So check if the failed MDS
is the same as whoami, and abort if so. This fixes #4637.

Signed-off-by: Greg Farnum <greg@inktank.com>
Reviewed-by: Yan, Zheng <zheng.z.yan@intel.com>
2013-04-03 12:43:49 -07:00
Josh Durgin
2d19ebda06 Merge pull request #194 from ceph/wip-rbd-diff
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-03 12:16:50 -07:00
Sage Weil
17a38593d6 Merge pull request #195 from dalgaaf/wip-da-fix-make
Makefile.am: fix build of ceph_test_cors

Reviewed-by: Sage Weil <sage@inktank.com>
2013-04-03 11:34:05 -07:00
Danny Al-Gaaf
c8cbec2b48 Makefile.am: fix build of ceph_test_cors
Fix build of ceph_test_cors: use $(CRYPTO_LIBS) instead of -lcryptopp.

Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>
2013-04-03 20:04:02 +02:00
João Eduardo Luís
e8562218b6 Merge pull request #192 from ceph/wip-mon-disk-warn
mon: limit warnings about low mon disk space

Reviewed-by: Joao Eduardo Luis <joao.luis@inktank.com>
2013-04-03 10:59:00 -07:00
Dan Mick
a4c6e2eecb Merge pull request #193 from ceph/wip-mon-crush
mon: make 'osd crush rm|unlink ...' idempotent

Reviewed-by: Dan Mick dan.mick@inktank.com
2013-04-03 10:56:43 -07:00
Sage Weil
e2edb9c646 Merge remote-tracking branch 'gh/wip-4623' 2013-04-03 10:26:30 -07:00
Sam Lang
0ce09fad55 client: Kick waiters for max size
If the mds restarts without successfully logging a max size
cap update, the client waits indefinitely in Client::get_caps
on the waitfor_caps list.  So when the client gets an mds map
indicating a new active mds has replaced a down mds, we need to
kick the caps update request.  This patch mimics the behavior
in the kernel by setting the wanted_max_size
and requested_max_size to 0 and wakes up the waiters.

Fixes #4582.
Signed-off-by: Sam Lang <sam.lang@inktank.com>
2013-04-03 11:36:39 -05:00
Sam Lang
f9b7dff70a client: log inode when waiting on max size
Logging the specific inode that the get_caps()
is waiting for helps during debugging of stuck waiters.

Signed-off-by: Sam Lang <sam.lang@inktank.com>
2013-04-03 11:31:52 -05:00
Sage Weil
85a77dfda6 mon: limit warnings about low mon disk space
Only warn once per percentage point per epoch.

Signed-off-by: Sage Weil <sage@inktank.com>
2013-04-03 08:37:50 -07:00
David Zafman
ea7c6c6db9 librados: Fixes to list_snaps test changes
Add back assert for overlap
Fix snap size and remove unnecessary comments
Fix failing test cases

Signed-off-by: David Zafman <david.zafman@inktank.com>
2013-04-03 08:25:53 -07:00
Gary Lowell
a793853850 ceph-disk: CalledProcessError has no output keyword on 2.6
Signed-off-by: Gary Lowell  <gary.lowell@inktank.com>
2013-04-02 23:01:12 -07:00
Sage Weil
b48d6b4a09 librados: test for EINVAL on bad list_snaps() ioctx snapid
Signed-off-by: Sage Weil <sage@inktank.com>
2013-04-02 20:59:34 -07:00
David Zafman
65a6975014 librados: set SNAP_DIR on listsnaps command
Signed-off-by: David Zafman <david.zafman@inktank.com>
2013-04-02 20:59:17 -07:00
David Zafman
ccb47fd050 osd: fail with EINVAL if bad snapid set on list_snaps
Signed-off-by: David Zafman <david.zafman@inktank.com>
2013-04-02 20:58:51 -07:00
David Zafman
ae67d37515 librados: fix list_snaps() tests
The ioctx must be set to SNAP_DIR, and the head does not include any
clones.

Bug #4623

Signed-off-by: David Zafman <david.zafman@inktank.com>
2013-04-02 20:57:34 -07:00
Sage Weil
8d3c7ffdfd mon: make 'osd crush rm|unlink ...' idempotent
Return 0 if the item is not there, not -ENOENT.

Fixes: #4624
Signed-off-by: Sage Weil <sage@inktank.com>
2013-04-02 18:39:37 -07:00
Sage Weil
4e847e8b2c librbd: simplify diff_iterate calls to list_snaps
We don't need the size.  Use the simpler API call.

Signed-off-by: Sage Weil <sage@inktank.com>
2013-04-02 18:13:01 -07:00
David Zafman
edc46a7fbb Fix comment typos
Signed-off-by: David Zafman <david.zafman@inktank.com>
2013-04-02 17:51:40 -07:00
Sage Weil
8ebf26ae18 Merge pull request #185 from dalgaaf/wip-da-fix-misc-2
Bunch of fixes for issues from SCA
2013-04-02 17:41:50 -07:00
Sage Weil
7f19915f30 Merge pull request #186 from dalgaaf/wip-da-pylint
Fix smaller python issues
2013-04-02 17:41:16 -07:00
Sage Weil
04fff0dc71 Merge pull request #187 from imjustmatthew/imjustmatthew_docs2
Adds "mds fail 0" command to operations commmand reference.
2013-04-02 17:40:50 -07:00
Dan Mick
17a55660e8 Merge pull request #188 from dmick/wip-test-config-key
test_mon_config_key.py: fix 'del' to clean up correctly internally

Reviewed-by: Joao Eduardo Luis <joao.luis@inktank.com>
2013-04-02 16:38:12 -07:00
Dan Mick
cfd32fff7c test_mon_config_key.py: fix 'del' to clean up correctly internally 2013-04-02 15:09:43 -07:00
Greg Farnum
3842ff7d67 Merge remote-tracking branch 'origin/wip-4619'
Reviewed-by: Greg Farnum <greg@inktank.com>
2013-04-02 14:38:44 -07:00
Sage Weil
5cf0c492e7 mds: initialize tableservers/clients on mds creation
The handle_mds_recovery(who) path initializes the anchorclients by having
the server send a 'ready' message on recovery when the server is active
and a peer becomes active.  Similarly, recovery_done() does the same when
the server becomes active.  However, this misses the creation path.  Handle
that explicitly in boot_create.

Fixes: #4619
Signed-off-by: Sage Weil <sage@inktank.com>
2013-04-02 13:05:49 -07:00
Sage Weil
1054c87c19 Revert "mds: trigger tableserver active/recovery hook even for self"
This reverts commit 968c6c0c94.

This will trigger the 'ready' message twice when we restart, because we
will trigger in both recovery_done() and handle_mds_recovery().

Signed-off-by: Sage Weil <sage@inktank.com>
2013-04-02 13:05:46 -07:00
Dan Mick
8a6438fefa README: fix dependency lists
1) automake appeared twice
2) make apt-get command match the list
2013-04-02 13:01:04 -07:00
Matthew Roy
0c16b31db7 Adds "mds fail 0" command to operations commmand reference.
Partially fixes #2206, though better documentation will eventually be needed.
2013-04-02 13:57:53 -04:00
Sage Weil
6072f8f7c0 Merge pull request #184 from dachary/wip-4617
explain what an inline xattr is and how it relates to omap
2013-04-02 09:35:05 -07:00
Sage Weil
968c6c0c94 mds: trigger tableserver active/recovery hook even for self
The tableserver now sends a READY message to clients when they go active;
we need to do this even for our own local tableclients, or else they do
not initialize and hang on first use after bringing up a fresh cluster.

Signed-off-by: Sage Weil <sage@inktank.com>
Reviewed-by: Greg Farnum <greg@inktank.com>
2013-04-02 08:58:35 -07:00
Danny Al-Gaaf
c4eb7e6ddd ceph-disk: fix some (local) variable names
Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>
2013-04-02 17:54:53 +02:00
Danny Al-Gaaf
3dd8b46121 ceph-disk: fix naming of local variable in is_mounted()
Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>
2013-04-02 17:36:37 +02:00
Danny Al-Gaaf
eaf31bf9f9 ceph-disk: merge twice defined function is_mounted(dev)
Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>
2013-04-02 17:33:08 +02:00
Danny Al-Gaaf
4c6d6442a8 ceph-disk: define exception type
Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>
2013-04-02 17:26:12 +02:00
Danny Al-Gaaf
058eb923c5 ceph-disk: fix Redefining name 'uuid' from outer scope
Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>
2013-04-02 17:17:38 +02:00
Danny Al-Gaaf
0080d1df7c ceph-disk: add missing space after comma
Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>
2013-04-02 17:14:23 +02:00
Danny Al-Gaaf
b99b8cc8a2 rgw/rgw_user.cc: reduce scope of variable
Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>
2013-04-02 17:01:07 +02:00
Danny Al-Gaaf
ee7eae3699 rgw/rgw_user.h: remove twice initialized purge_keys
Remove twice initialized purge_keys from RGWUserAdminOpState();

Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>
2013-04-02 16:50:41 +02:00
Danny Al-Gaaf
6e9f4781d3 test_cors.cc: prefer prefix ++operator for iterator
Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>
2013-04-02 16:36:17 +02:00
Danny Al-Gaaf
b4845a6975 test_cors.cc: use static_cast instead of C-Style cast
Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>
2013-04-02 16:30:57 +02:00
Danny Al-Gaaf
e2c3511b7a test_cors.cc: use %u to format unsigned in sprintf()
Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>
2013-04-02 16:25:22 +02:00
Danny Al-Gaaf
bda21f7264 rgw/rgw_user.cc: use 'true' directly instead of variable
Instead of passing 'true' via bool defer_user_update variable
in RGWUser::execute_modify() to keys.add() use it directly.

Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>
2013-04-02 16:17:52 +02:00
Danny Al-Gaaf
84b1914b78 rgw/rgw_user.cc: reduce scope of same_email in execute_modify()
Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>
2013-04-02 16:15:27 +02:00