]>
git.apps.os.sepia.ceph.com Git - ceph.git/log
Sage Weil [Thu, 19 Nov 2009 20:44:55 +0000 (12:44 -0800)]
rgw: fix base64 warnings (move impl to .c file)
Sage Weil [Thu, 19 Nov 2009 19:45:17 +0000 (11:45 -0800)]
mon: only remove_session if !s->closed
Sage Weil [Thu, 19 Nov 2009 19:40:50 +0000 (11:40 -0800)]
atomic: assert valid ref counts
Make sure we catch ref counting errors!
Sage Weil [Thu, 19 Nov 2009 19:38:35 +0000 (11:38 -0800)]
auth: initialize EntityName type in constructor
Sage Weil [Thu, 19 Nov 2009 19:35:33 +0000 (11:35 -0800)]
mon: fix session ref leak
Sage Weil [Thu, 19 Nov 2009 20:42:29 +0000 (12:42 -0800)]
monc: restart auth handshake each time we connect to mon
Sage Weil [Thu, 19 Nov 2009 19:24:48 +0000 (11:24 -0800)]
mds: don't change directory size/rbytes from client cap update
This is wrong, and screws up the recursive accounting.
Sage Weil [Thu, 19 Nov 2009 19:05:17 +0000 (11:05 -0800)]
auth: add ull to large constant
so that we build on 32bit platforms
Sage Weil [Thu, 19 Nov 2009 19:05:04 +0000 (11:05 -0800)]
mds: fix inode rstat debug output
Sage Weil [Thu, 19 Nov 2009 18:57:01 +0000 (10:57 -0800)]
auth: add missing virtual destructors
Sage Weil [Thu, 19 Nov 2009 18:11:19 +0000 (10:11 -0800)]
osdc: use __u64, not size_t
size_t is too small on 32 bit platforms.
Sage Weil [Thu, 19 Nov 2009 17:28:22 +0000 (09:28 -0800)]
mds: do not crash if m->connection == NULL
Sage Weil [Thu, 19 Nov 2009 17:28:02 +0000 (09:28 -0800)]
mds: do not send MMDSMap to ourselves
Sage Weil [Thu, 19 Nov 2009 00:51:24 +0000 (16:51 -0800)]
add ceph_fsid_set helper
Sage Weil [Wed, 18 Nov 2009 20:18:17 +0000 (12:18 -0800)]
filejournal: use single shared zeroed buffer for alignment padding
Avoids lots of mallocs and memsets (one for every journal event).
Sage Weil [Wed, 18 Nov 2009 20:17:35 +0000 (12:17 -0800)]
buffer: add create_static constructor
Yehuda Sadeh [Thu, 19 Nov 2009 00:20:18 +0000 (16:20 -0800)]
vstart.sh: fix -k
Sage Weil [Thu, 19 Nov 2009 00:01:20 +0000 (16:01 -0800)]
Sage Weil [Thu, 19 Nov 2009 00:00:42 +0000 (16:00 -0800)]
vstart: fix errors on auth none
Sage Weil [Thu, 19 Nov 2009 00:00:31 +0000 (16:00 -0800)]
msgr: rev protocol
Yehuda Sadeh [Wed, 18 Nov 2009 23:29:56 +0000 (15:29 -0800)]
auth: fix monclient startup issue
Yehuda Sadeh [Wed, 18 Nov 2009 22:19:01 +0000 (14:19 -0800)]
auth: fix client and server side negotiation
Sage Weil [Wed, 18 Nov 2009 20:25:46 +0000 (12:25 -0800)]
uclient: replace C_Cond with C_SafeCond
Sage Weil [Wed, 18 Nov 2009 19:56:26 +0000 (11:56 -0800)]
todo
Sage Weil [Tue, 17 Nov 2009 19:49:37 +0000 (11:49 -0800)]
todo
Sage Weil [Wed, 18 Nov 2009 18:49:38 +0000 (10:49 -0800)]
osd: allow reinitialization of osd journal
(without hosing osd data)
Greg Farnum [Wed, 18 Nov 2009 18:13:29 +0000 (10:13 -0800)]
rados: Move array to heap. Prevent segfault when you use too-large a write size
Greg Farnum [Wed, 18 Nov 2009 18:08:05 +0000 (10:08 -0800)]
rados: remove unused paramater, add very primitive read bench in verify step
Greg Farnum [Wed, 18 Nov 2009 00:06:24 +0000 (16:06 -0800)]
rados: Cosmetic changes for Sage
Sage Weil [Wed, 18 Nov 2009 18:12:10 +0000 (10:12 -0800)]
auth: change server side negotiation a bit
client is not fixed up yet
Sage Weil [Wed, 18 Nov 2009 00:31:18 +0000 (16:31 -0800)]
Merge branch 'unstable' of ceph.newdream.net:git/ceph into unstable
Sage Weil [Wed, 18 Nov 2009 00:26:33 +0000 (16:26 -0800)]
osd: eval access mode before try_*
The problem is the try_* functions may change the mode, but a start_write
may not follow if the op doesn't execute (for any number of reasons). So
the next time we come around, we may need to reset back to IDLE if
num_wr == 0. Be careful about the wake flag.
This fixes osd op hangs (particularly after mds restart, and lots of
failed stats on objects during file size/mtime recovery).
Sage Weil [Tue, 17 Nov 2009 23:22:27 +0000 (15:22 -0800)]
osd: debug pg mode
Sage Weil [Tue, 17 Nov 2009 23:05:41 +0000 (15:05 -0800)]
sepia: valgrind for a while
Sage Weil [Tue, 17 Nov 2009 22:31:19 +0000 (14:31 -0800)]
objecter: be more verbose about laggy requests
Sage Weil [Tue, 17 Nov 2009 22:11:24 +0000 (14:11 -0800)]
msgr: use discard_queue, kill drop_msgs
Sage Weil [Tue, 17 Nov 2009 22:06:07 +0000 (14:06 -0800)]
msgr: don't initiate connect if we keepalive on unconnected peer
Sage Weil [Wed, 18 Nov 2009 00:30:36 +0000 (16:30 -0800)]
msgr: fix possible use-after-free by taking pipe lock during reap
This ensures that whoever called stop() (mark_down, in particular) finished
with the pipe (unlocked it) before we go and free it. Otherwise we might
call p->lock.Unlock() after the reaper deleted the Pipe, mucking up some
other memory.
I don't think this was actually triggerd, tho, since we would have seen
the assert(nlock == 0) in ~Mutex???
Sage Weil [Wed, 18 Nov 2009 00:30:12 +0000 (16:30 -0800)]
msgr: use common helper for reader/writer thread stop and reap queueing; fix locking
We weren't holding lock during ::close(sd)
Sage Weil [Tue, 17 Nov 2009 22:03:36 +0000 (14:03 -0800)]
msgr: rename report_failures -> drop_msgs
Name was from legacy behavior
Sage Weil [Tue, 17 Nov 2009 22:02:57 +0000 (14:02 -0800)]
msgr: small cleanups
Sage Weil [Tue, 17 Nov 2009 22:02:09 +0000 (14:02 -0800)]
monc: small sanity check
Sage Weil [Tue, 17 Nov 2009 19:37:36 +0000 (11:37 -0800)]
uclient: set mode on snapdir properly
Sage Weil [Tue, 17 Nov 2009 19:34:02 +0000 (11:34 -0800)]
uclient: assert we have CAP_PIN
Not sure why that was conditional before??? Hmm!
Sage Weil [Tue, 17 Nov 2009 19:30:23 +0000 (11:30 -0800)]
uclient: always retain CAP_PIN
Sage Weil [Tue, 17 Nov 2009 18:46:32 +0000 (10:46 -0800)]
monc: small cleanup
avoid unnec finish_hunting call on get_monmap_privately
Greg Farnum [Fri, 13 Nov 2009 23:58:14 +0000 (15:58 -0800)]
rados: Update stat calculation and display; change to wait_for_safe
Greg Farnum [Fri, 13 Nov 2009 22:38:50 +0000 (14:38 -0800)]
qa: We want to stop on errors. Add new scripts.
Yehuda Sadeh [Tue, 17 Nov 2009 00:45:53 +0000 (16:45 -0800)]
auth: reuse global_id when requesting tickets
Yehuda Sadeh [Tue, 17 Nov 2009 00:00:23 +0000 (16:00 -0800)]
vstart: fix overwrite conf
Sage Weil [Mon, 16 Nov 2009 22:33:12 +0000 (14:33 -0800)]
mon: fix decode authmon stash on restart
Sage Weil [Mon, 16 Nov 2009 22:25:30 +0000 (14:25 -0800)]
mds: don't crap on !check_from, this doesn't happen with nolaggy
Sage Weil [Mon, 16 Nov 2009 22:25:13 +0000 (14:25 -0800)]
mon: move misplaced parens
Sage Weil [Mon, 16 Nov 2009 21:35:45 +0000 (13:35 -0800)]
todo
Sage Weil [Mon, 16 Nov 2009 19:46:03 +0000 (11:46 -0800)]
msgr: get rid of harmless valgrind error
==7781== Source and destination overlap in memcpy(0x5B97EA8, 0x5B97EA8, 136)
Sage Weil [Mon, 16 Nov 2009 19:45:25 +0000 (11:45 -0800)]
mon: fail on write error
Don't rename badly written file. Also assert success, for now.
Sage Weil [Fri, 13 Nov 2009 18:40:02 +0000 (10:40 -0800)]
monc: fix race/deadlock for get_monmap_privately
Sage Weil [Sat, 14 Nov 2009 00:27:26 +0000 (16:27 -0800)]
uclient: use Inode::auth_cap to find mds to write back snap cap metadata to
Sage Weil [Sat, 14 Nov 2009 00:26:59 +0000 (16:26 -0800)]
uclient: keep MClientRequest local to send_request
Sage Weil [Fri, 13 Nov 2009 23:06:31 +0000 (15:06 -0800)]
mds: journal open_files based on is_any_caps_wanted(), not is_any_caps()
Actually we're a bit conservative in a few places since the wanted check
is a bit more expensive. We always do a full check in try_to_expire, so
much of the time we can do the quick check only.
Sage Weil [Fri, 13 Nov 2009 22:57:39 +0000 (14:57 -0800)]
mds: don't rejournal files with caps that are unwanted
If they're unwanted, it's no biggie to fail to reconnect the cap. And
Locker::adjust_cap_wanted() already adjusts the open_file logseg lists in
this way, so let's just totally consistent.
Sage Weil [Fri, 13 Nov 2009 22:44:22 +0000 (14:44 -0800)]
mds: don't croak on open_files without caps
We can get a capless inode here if we replay an open file, the client
fails to reconnect it, but does REPLAY an open request (that adds it
to the logseg). AFAICS it's ok for the client to replay an open on a
file it doesn't have in it's cache anymore.
Greg Farnum [Fri, 13 Nov 2009 19:00:55 +0000 (11:00 -0800)]
rados: status printouts are now threaded
Sage Weil [Fri, 13 Nov 2009 19:10:36 +0000 (11:10 -0800)]
mkmonfs: cleanup monmap fn arg
Sage Weil [Fri, 13 Nov 2009 18:55:31 +0000 (10:55 -0800)]
Makefile: include all the headers we need to build
Sage Weil [Fri, 13 Nov 2009 18:47:52 +0000 (10:47 -0800)]
config: accept list of possible conf file locations; include ./ceph.conf
This makes './ceph -w' work with './vstart.sh -x'.
Sage Weil [Fri, 13 Nov 2009 18:40:10 +0000 (10:40 -0800)]
vstart: -x to start up with cephx
Sage Weil [Fri, 13 Nov 2009 18:40:02 +0000 (10:40 -0800)]
monc: fix race/deadlock for get_monmap_privately
Sage Weil [Fri, 13 Nov 2009 18:09:31 +0000 (10:09 -0800)]
mds: don't try to send mdsmap to clients that need to reconnect
It won't work, clients must connect to us.
Also stop sending the map on recovery finish; they'll also (currently, at
least) get that from the monitor.
Sage Weil [Fri, 13 Nov 2009 18:09:12 +0000 (10:09 -0800)]
mds: ignore/warn on late client reconnect attempts
In the future maybe we can do better (i.e. best effort attempt to reconnect
after the normal interval has passed)
Yehuda Sadeh [Fri, 13 Nov 2009 17:34:51 +0000 (09:34 -0800)]
auth: don't try to read keys if cephx not supported
Yehuda Sadeh [Fri, 13 Nov 2009 00:53:46 +0000 (16:53 -0800)]
auth: use user admin as the default user
Greg Farnum [Fri, 13 Nov 2009 00:45:05 +0000 (16:45 -0800)]
Merge branch 'unstable' of ceph.newdream.net:/git/ceph into unstable
Greg Farnum [Fri, 13 Nov 2009 00:44:35 +0000 (16:44 -0800)]
qa: prepare for a hierarchical test script system
Sage Weil [Fri, 13 Nov 2009 00:12:20 +0000 (16:12 -0800)]
vstart: some cleanup
Sage Weil [Thu, 12 Nov 2009 23:29:36 +0000 (15:29 -0800)]
conf: default to auth type 'none'
Sage Weil [Thu, 12 Nov 2009 23:29:26 +0000 (15:29 -0800)]
auth: fix compile errors, warnings
Sage Weil [Thu, 12 Nov 2009 23:24:50 +0000 (15:24 -0800)]
auth: kill residual kernel/ files
Sage Weil [Thu, 12 Nov 2009 23:23:24 +0000 (15:23 -0800)]
Merge branch 'unstable' into auth
Sage Weil [Tue, 10 Nov 2009 00:35:05 +0000 (16:35 -0800)]
uclient: fewer msgs
Sage Weil [Thu, 12 Nov 2009 23:12:18 +0000 (15:12 -0800)]
todo
Sage Weil [Thu, 12 Nov 2009 22:55:26 +0000 (14:55 -0800)]
Merge branch 'unstable' of ssh://ceph.newdream.net/git/ceph into unstable
Sage Weil [Thu, 12 Nov 2009 22:17:18 +0000 (14:17 -0800)]
put testing ceph.conf's in git
Sage Weil [Thu, 12 Nov 2009 22:16:31 +0000 (14:16 -0800)]
osd: print useful mode wait message
Sage Weil [Thu, 12 Nov 2009 22:16:17 +0000 (14:16 -0800)]
filestore: skip sync_file_range if the full commit has started
Presumably btrfs commit_transaction is more effective than our per-file
flushing. And we don't want to fight each other.
Sage Weil [Thu, 12 Nov 2009 22:54:22 +0000 (14:54 -0800)]
mds: recommit after commit if waiting for newer version
If there are waiters for a later version of the dir to hit
disk, then we need to recommit as soon as the prior commit
completes. We auth_pin on adding the first waiter, and do
not unpin until removing the last waiter, so this doesn't
break auth_pin rules.
Previously we could stall because we didn't finish the
waiter (on the later version) but also never started the
commit. Sometimes we would get lucky and someone else would
ask for a commit, but sometimes not. We would then see old
LogSegments that would never get fully expired.
Yehuda Sadeh [Thu, 12 Nov 2009 22:31:36 +0000 (14:31 -0800)]
auth: fix mds messages filtering
Greg Farnum [Thu, 12 Nov 2009 20:06:57 +0000 (12:06 -0800)]
rados: fix usage() and -p option checking
Sage Weil [Thu, 12 Nov 2009 18:45:54 +0000 (10:45 -0800)]
mds: warn, don't crash, on rfiles/rsubdirs underflow
This doesn't fix the bug, but lets the mds at least start up.
Yehuda Sadeh [Thu, 12 Nov 2009 17:51:29 +0000 (09:51 -0800)]
msg: don't die unconditionally when got unknown message type
Sage Weil [Thu, 12 Nov 2009 00:10:14 +0000 (16:10 -0800)]
uclient: increase cache size
Sage Weil [Thu, 12 Nov 2009 00:09:52 +0000 (16:09 -0800)]
filestore: flusher thread; commit snaps (disabled)
Sage Weil [Wed, 11 Nov 2009 18:20:12 +0000 (10:20 -0800)]
osd: avoid truncate, remove ops we know will fail
Now that we check return codes, these cause problems.
Sage Weil [Wed, 11 Nov 2009 18:19:28 +0000 (10:19 -0800)]
osd: don't requeue pg removal if already removing
Sage Weil [Wed, 11 Nov 2009 23:52:10 +0000 (15:52 -0800)]
todo
Sage Weil [Wed, 11 Nov 2009 23:47:28 +0000 (15:47 -0800)]
mds: force rdlock on any snapped inodes
When the client has an excl lock on an inode, and it's
stating a snapped version of it, we can't expect it to
put 2 and 2 together and look at it's head metadata. If
the cap does not follow the snapid we're trying to stat, do
the full rdlock to force the snapped values back to the
mds so we can do the cow.
If there is nothing cow, the cap will get reissued with an
accurate follows value, and we won't have to do this again.
Greg Farnum [Wed, 11 Nov 2009 00:44:07 +0000 (16:44 -0800)]
sessionmap is an object, not a pointer
Sage Weil [Wed, 11 Nov 2009 00:32:01 +0000 (16:32 -0800)]
mds: fix typo; also only suicide if we have clients
If we are the last MDS and have no clients we should be
able to stop cleanly....
Sage Weil [Wed, 11 Nov 2009 00:26:44 +0000 (16:26 -0800)]
mds: underwater is function of _loaded_ version, not in core version
We may load a dir version off disk that is older than the
in-core version (because we got newer data from the
journal, say). When marking underwater items clean, do
so based on the _loaded_ version, not out in-core version.
Sage Weil [Mon, 9 Nov 2009 21:32:07 +0000 (13:32 -0800)]
todo