]>
git.apps.os.sepia.ceph.com Git - ceph.git/log
Sage Weil [Thu, 28 Oct 2010 20:14:03 +0000 (13:14 -0700)]
cephfs: remove unused variables
Signed-off-by: Sage Weil <sage@newdream.net>
Sage Weil [Thu, 28 Oct 2010 19:10:29 +0000 (12:10 -0700)]
config: fix signal handler recursion
Avoid having old handler pointer match the new handler.
Avoid calling an old handler if it pointer is null.
Signed-off-by: Sage Weil <sage@newdream.net>
Sage Weil [Thu, 28 Oct 2010 18:53:38 +0000 (11:53 -0700)]
objecter: refactor interface with object_locator_t
This paves the way for a locator that lets the user specify an arbitrary
string to hash for placement (instead of the object name). It also
captures everything that would affect placement (pool, preferred, etc.)
in a single type.
Just the client side; no server side or protocol changes yet.
Signed-off-by: Sage Weil <sage@newdream.net>
Sage Weil [Thu, 28 Oct 2010 18:02:26 +0000 (11:02 -0700)]
mds: pin NEEDSNAPFLUSH only when adding item
This is mainly paranoia.
Signed-off-by: Sage Weil <sage@newdream.net>
Sage Weil [Wed, 27 Oct 2010 21:43:43 +0000 (14:43 -0700)]
mds: store dir inode in separate object; fetch from both. incompat flag.
This avoids setting large xattrs. There's no reason the inode needs to be
on the same object as the dir(frag) data.
Signed-off-by: Sage Weil <sage@newdream.net>
Yehuda Sadeh [Wed, 27 Oct 2010 20:21:11 +0000 (13:21 -0700)]
filestore: can force use of stale snaps
also, overwrite the commit_seq with the current version in case we
forced stale snaps.
Yehuda Sadeh [Wed, 27 Oct 2010 18:38:11 +0000 (11:38 -0700)]
filestore: don't revert to old snapshots on startup
This should fix bug #55
Yehuda Sadeh [Tue, 26 Oct 2010 23:46:31 +0000 (16:46 -0700)]
filestore: read commit_seq before mounting (btrfs ioctls)
Colin Patrick McCabe [Wed, 27 Oct 2010 06:19:19 +0000 (23:19 -0700)]
ceph.cc: delete deadcode
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Colin Patrick McCabe [Wed, 27 Oct 2010 03:51:49 +0000 (20:51 -0700)]
Move ceph.cc to tools/
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Colin Patrick McCabe [Tue, 26 Oct 2010 21:26:12 +0000 (14:26 -0700)]
configure.ac: add ./configure option for gtk2
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Greg Farnum [Tue, 26 Oct 2010 23:28:48 +0000 (16:28 -0700)]
messenger: always unlock existing pipes, even if they're lossy
Sage Weil [Tue, 26 Oct 2010 19:04:17 +0000 (12:04 -0700)]
Merge remote branch 'origin/testing' into unstable
Conflicts:
configure.ac
src/rados.cc
Yehuda Sadeh [Tue, 26 Oct 2010 18:05:55 +0000 (11:05 -0700)]
filestore: some cleanup
Yehuda Sadeh [Mon, 25 Oct 2010 23:35:48 +0000 (16:35 -0700)]
filestore: escape the xattr chunk names
Colin Patrick McCabe [Tue, 26 Oct 2010 17:51:52 +0000 (10:51 -0700)]
osd::Missing: const cleanup
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Colin Patrick McCabe [Tue, 26 Oct 2010 17:42:15 +0000 (10:42 -0700)]
osd: move PG::Missing implementation to PG.cc
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Yehuda Sadeh [Tue, 26 Oct 2010 18:05:55 +0000 (11:05 -0700)]
filestore: some cleanup
Sage Weil [Tue, 26 Oct 2010 04:52:30 +0000 (21:52 -0700)]
Merge branch 'btrfs_snap_ioctls' into unstable
Sage Weil [Tue, 26 Oct 2010 04:52:07 +0000 (21:52 -0700)]
filestore: warn if btrfs_snaps enabled but no async snap create ioctl
Signed-off-by: Sage Weil <sage@newdream.net>
Sage Weil [Mon, 25 Oct 2010 23:44:10 +0000 (16:44 -0700)]
mon: remove pg from deleted pools from pg_map
Signed-off-by: Sage Weil <sage@newdream.net>
Sage Weil [Mon, 25 Oct 2010 22:07:41 +0000 (15:07 -0700)]
mds: fix up mds_bal_frag options
Use the mds_bal_frag option to enable/disable. Make checks consistent.
Signed-off-by: Sage Weil <sage@newdream.net>
Sage Weil [Mon, 25 Oct 2010 22:03:32 +0000 (15:03 -0700)]
add cephfs to deb, rpm
Greg Farnum [Mon, 25 Oct 2010 23:51:38 +0000 (16:51 -0700)]
Merge branch 'objectcacher' into unstable
Greg Farnum [Mon, 25 Oct 2010 21:21:21 +0000 (14:21 -0700)]
client: fix to handle new ObjectCacher pool requirements.
Greg Farnum [Mon, 25 Oct 2010 21:18:02 +0000 (14:18 -0700)]
osdc: Add pool awareness to the ObjectCacher, to prevent unfortunate collisions.
Greg Farnum [Mon, 25 Oct 2010 18:38:41 +0000 (11:38 -0700)]
osdc: Fix release_all so it loops properly
Yehuda Sadeh [Mon, 25 Oct 2010 23:35:48 +0000 (16:35 -0700)]
filestore: escape the xattr chunk names
Colin Patrick McCabe [Mon, 25 Oct 2010 22:22:34 +0000 (15:22 -0700)]
PG::Log::Entry: remove unused snap_t field
The snap_t information is stored in the sobject_t field now.
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Greg Farnum [Mon, 25 Oct 2010 20:42:13 +0000 (13:42 -0700)]
makefile: add cephfs
Colin Patrick McCabe [Mon, 25 Oct 2010 19:24:13 +0000 (12:24 -0700)]
./ceph osd setcrushmap: validate crushmap
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Colin Patrick McCabe [Mon, 25 Oct 2010 19:06:01 +0000 (12:06 -0700)]
crush: improve error handling in map decoding
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Colin Patrick McCabe [Sun, 24 Oct 2010 21:22:00 +0000 (14:22 -0700)]
cap_reconnect_t: ignore embedded NULLs in the path
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Sage Weil [Fri, 22 Oct 2010 23:51:27 +0000 (16:51 -0700)]
v0.22.1
Sage Weil [Fri, 22 Oct 2010 23:33:03 +0000 (16:33 -0700)]
Makefile: add errno.h
Sage Weil [Sat, 23 Oct 2010 17:17:29 +0000 (10:17 -0700)]
mds: be quiet about snaprealm push/pop
Signed-off-by: Sage Weil <sage@newdream.net>
Sage Weil [Fri, 22 Oct 2010 23:06:19 +0000 (16:06 -0700)]
filestore: ignore ENOSPC on setxattr pending a better workaround
This effectively reverts to old behavior (we weren't checking for ENOSPC
errors at all before). Log which object it happened on.
Signed-off-by: Sage Weil <sage@newdream.net>
Yehuda Sadeh [Fri, 22 Oct 2010 22:52:58 +0000 (15:52 -0700)]
filestore: change xattr chunk size to 2048
Yehuda Sadeh [Fri, 22 Oct 2010 17:09:32 +0000 (10:09 -0700)]
filestore: split xattrs to multiple chunks
Yehuda Sadeh [Thu, 21 Oct 2010 21:41:05 +0000 (14:41 -0700)]
rados: add getxattr, setxattr
Yehuda Sadeh [Fri, 22 Oct 2010 22:52:58 +0000 (15:52 -0700)]
filestore: change xattr chunk size to 2048
Yehuda Sadeh [Fri, 22 Oct 2010 21:32:55 +0000 (14:32 -0700)]
librados: add rmxattr
Greg Farnum [Fri, 22 Oct 2010 20:36:30 +0000 (13:36 -0700)]
Revert "messenger: Make sure to unlock existing->pipe_lock. There are a few cases in the "open" section where we can go to fail_unlocked while still holding existing->pipe_lock. So unlock it."
This reverts commit
96692d24c8cdf0fe88260949b67f8580e0c70696 .
This patch accidentally got merged into the tree twice, so
merging issues came up.
Colin Patrick McCabe [Fri, 22 Oct 2010 18:48:35 +0000 (11:48 -0700)]
test_lost.sh: put common functions in test_common
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Greg Farnum [Fri, 22 Oct 2010 18:24:24 +0000 (11:24 -0700)]
Merge branch 'msgr' into unstable
Greg Farnum [Fri, 22 Oct 2010 18:16:24 +0000 (11:16 -0700)]
messenger: If we error out of accept() but have messages in our queue, save Pipe.
This can occur if we're replacing another Pipe and hit an error
in the process.
Greg Farnum [Fri, 22 Oct 2010 18:15:22 +0000 (11:15 -0700)]
messenger: If we're replacing an existing Pipe, steal queue when we kill it!
Previously we could fail out after killing existing but before
splicing its queue into our own, which lost messages.
Colin Patrick McCabe [Fri, 22 Oct 2010 18:01:15 +0000 (11:01 -0700)]
PG::peer: introduce prior_set_build flag
Just because we have prior_set.empty() doesn't mean that the prior set
wasn't built. Create a flag to represent this concept.
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Yehuda Sadeh [Wed, 20 Oct 2010 20:49:35 +0000 (13:49 -0700)]
filestore: update btrfs_ioctl.h
Yehuda Sadeh [Mon, 18 Oct 2010 22:57:01 +0000 (15:57 -0700)]
filestore: use different encoding for snap async_create
Sage Weil [Thu, 8 Apr 2010 17:34:50 +0000 (10:34 -0700)]
filestore: use SNAP_DESTROY_ASYNC ioctl if available
Sage Weil [Thu, 18 Mar 2010 18:33:32 +0000 (11:33 -0700)]
filestore: remove stray async_snap_test if present
This cleans up if a prior instance failed to delete its
async_snap_test subvol.
Sage Weil [Tue, 23 Feb 2010 00:03:09 +0000 (16:03 -0800)]
filestore: use new async btrfs ioctls
Sage Weil [Fri, 22 Oct 2010 17:11:21 +0000 (10:11 -0700)]
osd: fix deadlock in map handler
To avoid deadlock,
- we need to drop osd_lock while we flush.
- we need to take map_lock _after_ we flush.
Signed-off-by: Sage Weil <sage@newdream.net>
Yehuda Sadeh [Fri, 22 Oct 2010 17:09:32 +0000 (10:09 -0700)]
filestore: split xattrs to multiple chunks
Yehuda Sadeh [Thu, 21 Oct 2010 21:41:05 +0000 (14:41 -0700)]
rados: add getxattr, setxattr
Sage Weil [Fri, 22 Oct 2010 04:45:27 +0000 (21:45 -0700)]
mds: implement 'fragment_dir path frag by' command
For testing dir fragmentation.
Signed-off-by: Sage Weil <sage@newdream.net>
Sage Weil [Fri, 22 Oct 2010 04:38:21 +0000 (21:38 -0700)]
Merge branch 'testing' into unstable
Colin Patrick McCabe [Fri, 22 Oct 2010 00:20:40 +0000 (17:20 -0700)]
PG::recover_master_log: replace count with find
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Colin Patrick McCabe [Thu, 21 Oct 2010 23:23:21 +0000 (16:23 -0700)]
prior_set_affected: log msg when we see a lost osd
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Colin Patrick McCabe [Thu, 21 Oct 2010 23:12:31 +0000 (16:12 -0700)]
osd: build_prior: clean up started_since_joining
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Colin Patrick McCabe [Thu, 21 Oct 2010 22:17:29 +0000 (15:17 -0700)]
osd: PG::prior_set_affected: fix lost OSD detection
When looking for newly-lost OSDs, we should check prior_set_lost rather
than prior_set. Down OSDs often are in PG::prior_set_down and NOT in
PG::prior_set.
Also update comment for prior_set_down. Sometimes OSDs are in both
PG::prior_set and PG::prior_set_down.
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Colin Patrick McCabe [Thu, 21 Oct 2010 20:59:15 +0000 (13:59 -0700)]
test_lost.sh: ensure that recovery doesn't start.
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Colin Patrick McCabe [Thu, 21 Oct 2010 17:54:08 +0000 (10:54 -0700)]
include/utime.h: should include include/types.h
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Sage Weil [Thu, 21 Oct 2010 00:30:10 +0000 (17:30 -0700)]
errno: add missing common/errno.h
Colin Patrick McCabe [Wed, 20 Oct 2010 22:40:07 +0000 (15:40 -0700)]
Create cpp_strerror to make error reporting easier
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Colin Patrick McCabe [Wed, 20 Oct 2010 20:19:49 +0000 (13:19 -0700)]
FileJournal: fix journal size calculation
If the journal is a raw block device, the user shouldn't need to give a
journal size argument most of the time-- it should default to using the
entire block device. This was the old default but it got changed
erroneously by commit
ad12d5d5be41ce .
Along the way, I split the FileJournal::_open code into multiple
functions, and added some additional checks. We no longer try block
device ioctls on regular files. We now check for block devices that are
too small.
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Sage Weil [Wed, 20 Oct 2010 20:47:16 +0000 (13:47 -0700)]
makefile: simplify cdebugpack install rule
Sage Weil [Tue, 19 Oct 2010 16:11:25 +0000 (09:11 -0700)]
debian: 0.22-4
Colin Patrick McCabe [Wed, 20 Oct 2010 00:38:32 +0000 (17:38 -0700)]
TestTimers: test SafeTimer as well as Timer
Test SafeTimer as well as Timer. Test timer shutdown.
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Colin Patrick McCabe [Wed, 20 Oct 2010 00:02:15 +0000 (17:02 -0700)]
Timer.cc: add testtimers
Add testtimers to test the timer code.
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Colin Patrick McCabe [Tue, 19 Oct 2010 21:45:14 +0000 (14:45 -0700)]
test_lost.sh: update timeout, fix payload
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Colin Patrick McCabe [Tue, 19 Oct 2010 17:20:21 +0000 (10:20 -0700)]
OSDMap: const cleanup
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Colin Patrick McCabe [Tue, 19 Oct 2010 17:09:29 +0000 (10:09 -0700)]
PG::build_prior: update comment
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Colin Patrick McCabe [Mon, 18 Oct 2010 20:44:51 +0000 (13:44 -0700)]
osd: track prior_set_lost
In the placement group code, track prior_set_lost. This fixes a bug
where a new OSDMap updates an OSD's lost_at time, but the PG code does
not update the PG data structures.
When clearing the peering state, call clear_prior() rather than manually
clearing every prior set.
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Colin Patrick McCabe [Mon, 18 Oct 2010 17:31:25 +0000 (10:31 -0700)]
osd: fix spacing in OSDMap::print
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Colin Patrick McCabe [Fri, 15 Oct 2010 19:16:41 +0000 (12:16 -0700)]
OSDMap::print: print osd_info_t using ostream op
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Colin Patrick McCabe [Fri, 15 Oct 2010 19:16:41 +0000 (12:16 -0700)]
osd mon: validate arguments before marking lost
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Colin Patrick McCabe [Fri, 15 Oct 2010 01:14:27 +0000 (18:14 -0700)]
osd: test: Add script to test LOST state
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Greg Farnum [Mon, 18 Oct 2010 18:23:50 +0000 (11:23 -0700)]
messenger: Make sure to unlock existing->pipe_lock. There are a few cases in the "open" section where we can go to fail_unlocked while still holding existing->pipe_lock. So unlock it.
Sage Weil [Thu, 21 Oct 2010 23:16:34 +0000 (16:16 -0700)]
mds: show readdir frag
Signed-off-by: Sage Weil <sage@newdream.net>
Sage Weil [Thu, 21 Oct 2010 23:16:20 +0000 (16:16 -0700)]
client: reset fg after _readdir_get_frag
The _readdir_get_frag may remap our frag; update the local variable
accordingly.
Signed-off-by: Sage Weil <sage@newdream.net>
Sage Weil [Thu, 21 Oct 2010 23:15:52 +0000 (16:15 -0700)]
client: fix skipped dentry on readdir chunk boundaries
The at_cache_name is the last name successfully passed to the caller.
Signed-off-by: Sage Weil <sage@newdream.net>
Sage Weil [Thu, 21 Oct 2010 23:15:03 +0000 (16:15 -0700)]
client: fix dcache removal during multiple frags
We remove unexpected dentries from our cache while processing mds results.
Results are ordered within a frag, but not between them. Since we can
have multiple frags, only remove results for the current frag, to avoid
removing items from earlier frags.
Signed-off-by: Sage Weil <sage@newdream.net>
Sage Weil [Thu, 21 Oct 2010 23:13:35 +0000 (16:13 -0700)]
client: show file offsets in hex
This makes it easy to pick out frags and offsets.
Signed-off-by: Sage Weil <sage@newdream.net>
Greg Farnum [Thu, 21 Oct 2010 20:27:35 +0000 (13:27 -0700)]
messenger: a 0 timeout on ::poll really means don't wait
(as opposed to -1, which waits until an event occurs).
So, set the default timeout to -1, and convert ms_tcp_read_timeout
to -1 if it's set at 0.
Sage Weil [Thu, 21 Oct 2010 20:07:38 +0000 (13:07 -0700)]
mds: fix inodestat encoding when frags are present
Also simplify the max_size check calculation.
Signed-off-by: Sage Weil <sage@newdream.net>
Sage Weil [Thu, 21 Oct 2010 19:48:07 +0000 (12:48 -0700)]
mds: do not finish_scatter_gather_update_accounted on dirfraglock
This needs to match finish_scatter_gather_update, and we don't
update/project the dirfrag there.
Signed-off-by: Sage Weil <sage@newdream.net>
Sage Weil [Thu, 21 Oct 2010 18:37:45 +0000 (11:37 -0700)]
objecter: reconnect on osd disconnect
If the connection closes to an OSD, we need to reconnect and resubmit our
ops. Otherwise we just hang. This is problematic if it is a transient
error, since we'll only retry if the OSDMap reflects a change, and that
won't happen for transient network/socket errors and such.
Signed-off-by: Sage Weil <sage@newdream.net>
Yehuda Sadeh [Thu, 21 Oct 2010 16:53:37 +0000 (09:53 -0700)]
rgw: return 204 on successful removal of bucket/object
Jim Schutt [Thu, 21 Oct 2010 16:32:09 +0000 (10:32 -0600)]
init-ceph: Make sure daemon_is_running() checks the correct instance
When starting multiple instances of a daemon on a single host,
for unknown reasons /var/run/ceph/$type.$id.pid can hold a pid
for which /proc/$pid/cmdline identifies the right type of daemon,
but the wrong instance. When this happens, all the configured
instances of a daemon are not running, but repeated invocations
of "init-ceph start" do not start the missing instances.
So, check for the correct daemon instance id as well as type when
testing if the daemon is up.
Signed-off-by: Jim Schutt <jaschut@sandia.gov>
Signed-off-by: Sage Weil <sage@newdream.net>
Sage Weil [Thu, 21 Oct 2010 17:59:49 +0000 (10:59 -0700)]
objecter: pause writes when FULL flag is set
Also, subscribe to all osdmap updates while FULL flag is set, so that we
discover when it is unset.
Signed-off-by: Sage Weil <sage@newdream.net>
Sage Weil [Thu, 21 Oct 2010 18:05:55 +0000 (11:05 -0700)]
objecter: always set READ or WRITE flag
We should set either (or both). Assert if we don't.
Signed-off-by: Sage Weil <sage@newdream.net>
Colin Patrick McCabe [Thu, 21 Oct 2010 17:54:08 +0000 (10:54 -0700)]
include/utime.h: should include include/types.h
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Yehuda Sadeh [Thu, 21 Oct 2010 16:53:37 +0000 (09:53 -0700)]
rgw: return 204 on successful removal of bucket/object
Sage Weil [Thu, 21 Oct 2010 00:30:10 +0000 (17:30 -0700)]
errno: add missing common/errno.h
Colin Patrick McCabe [Wed, 20 Oct 2010 22:40:07 +0000 (15:40 -0700)]
Create cpp_strerror to make error reporting easier
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Colin Patrick McCabe [Wed, 20 Oct 2010 20:19:49 +0000 (13:19 -0700)]
FileJournal: fix journal size calculation
If the journal is a raw block device, the user shouldn't need to give a
journal size argument most of the time-- it should default to using the
entire block device. This was the old default but it got changed
erroneously by commit
ad12d5d5be41ce .
Along the way, I split the FileJournal::_open code into multiple
functions, and added some additional checks. We no longer try block
device ioctls on regular files. We now check for block devices that are
too small.
Signed-off-by: Colin McCabe <colinm@hq.newdream.net>
Sage Weil [Wed, 20 Oct 2010 20:47:16 +0000 (13:47 -0700)]
makefile: simplify cdebugpack install rule
Sage Weil [Tue, 19 Oct 2010 16:11:25 +0000 (09:11 -0700)]
debian: 0.22-4