]>
git.apps.os.sepia.ceph.com Git - ceph.git/log
Sage Weil [Thu, 31 Jan 2008 00:46:30 +0000 (16:46 -0800)]
improved max_size handling in mds, client. avoid spurious setattr/utime after open in client by ignoring sub-second mtime resolution... weird
Sage Weil [Wed, 30 Jan 2008 21:50:57 +0000 (13:50 -0800)]
Merge commit 'origin/rados' into unstable
Sage Weil [Wed, 30 Jan 2008 21:48:24 +0000 (13:48 -0800)]
mds: flush file (size+mtime) metadata to journal on close
Yehuda Sadeh-Weinraub [Wed, 30 Jan 2008 19:55:06 +0000 (21:55 +0200)]
cleanup
Yehuda Sadeh-Weinraub [Wed, 30 Jan 2008 19:46:19 +0000 (21:46 +0200)]
added inode revalidation
Sage Weil [Wed, 30 Jan 2008 16:31:25 +0000 (08:31 -0800)]
fix mds cap interaction; echo blah > mnt/blah now works
Sage Weil [Tue, 29 Jan 2008 05:24:38 +0000 (21:24 -0800)]
msg: blobhash on entity_name_t causes strange badness... should look into that, but avoid for now
Sage Weil [Tue, 29 Jan 2008 05:23:47 +0000 (21:23 -0800)]
mds: adjust cap issue sequence to avoid unnecessary file_caps messages
Sage Weil [Mon, 28 Jan 2008 23:33:37 +0000 (15:33 -0800)]
debian packing stuffs
Sage Weil [Mon, 28 Jan 2008 20:30:17 +0000 (12:30 -0800)]
more automake. make dist seems to work
Sage Weil [Mon, 28 Jan 2008 19:54:53 +0000 (11:54 -0800)]
autogen.sh
Sage Weil [Mon, 28 Jan 2008 19:39:51 +0000 (11:39 -0800)]
more automake fun
Sage Weil [Mon, 28 Jan 2008 19:05:29 +0000 (11:05 -0800)]
more autoconf- and automake-fu
Sage Weil [Mon, 28 Jan 2008 19:04:46 +0000 (11:04 -0800)]
minor fixups
Sage Weil [Mon, 28 Jan 2008 17:03:43 +0000 (09:03 -0800)]
copy of old Makefile
Sage Weil [Mon, 28 Jan 2008 02:12:20 +0000 (18:12 -0800)]
first pass at autoconf and automake
Sage Weil [Sun, 27 Jan 2008 02:44:09 +0000 (18:44 -0800)]
smarter start.sh
Sage Weil [Sun, 27 Jan 2008 02:33:18 +0000 (18:33 -0800)]
startup: adjust out/* symlinks after forking (from call to daemon())
Sage Weil [Sat, 26 Jan 2008 17:33:13 +0000 (09:33 -0800)]
-d to daemonize, other startup cleanup
Sage Weil [Sat, 26 Jan 2008 16:41:15 +0000 (08:41 -0800)]
allow client to initially bind to INADDR_ANY and learn its ip from the monitor
Sage Weil [Fri, 25 Jan 2008 22:46:40 +0000 (14:46 -0800)]
misc style cleanup
Sage Weil [Fri, 25 Jan 2008 22:14:29 +0000 (14:14 -0800)]
build with 2.6.24 or earlier (well, 2.6.23, at least)
Sage Weil [Sat, 26 Jan 2008 05:46:23 +0000 (21:46 -0800)]
some x86 cleanups
Yehuda Sadeh-Weinraub [Fri, 25 Jan 2008 21:40:02 +0000 (23:40 +0200)]
Merge branch 'rados' of ssh://yehudasa@ceph.newdream.net/git/ceph into myrados
Yehuda Sadeh-Weinraub [Fri, 25 Jan 2008 21:25:44 +0000 (23:25 +0200)]
some reduction in stack space
Yehuda Sadeh-Weinraub [Fri, 25 Jan 2008 20:39:06 +0000 (22:39 +0200)]
protecting inode's i_block
Sage Weil [Fri, 25 Jan 2008 20:10:45 +0000 (12:10 -0800)]
quieter startup
Sage Weil [Fri, 25 Jan 2008 20:04:31 +0000 (12:04 -0800)]
write output to default/* by default
Sage Weil [Fri, 25 Jan 2008 20:01:58 +0000 (12:01 -0800)]
kernel: cleaned up mds_client locking, reconnect
Sage Weil [Fri, 25 Jan 2008 05:25:15 +0000 (21:25 -0800)]
map u64 ceph inos into ino_t on <64bit platforms
Sage Weil [Fri, 25 Jan 2008 08:25:27 +0000 (00:25 -0800)]
ebofs: journal replay working; better journal validation on mount
Sage Weil [Fri, 25 Jan 2008 08:15:54 +0000 (00:15 -0800)]
ebofs: avoid buffer cow bug
Sage Weil [Fri, 25 Jan 2008 05:59:38 +0000 (21:59 -0800)]
journal replay, makefile cleanup
Sage Weil [Fri, 25 Jan 2008 04:38:15 +0000 (20:38 -0800)]
ebofs: journal cleanup, tuning params in g_conf
Sage Weil [Fri, 25 Jan 2008 00:33:32 +0000 (16:33 -0800)]
roll directio into FileJournal
Sage Weil [Thu, 24 Jan 2008 10:12:48 +0000 (02:12 -0800)]
hacked dio journal
Sage Weil [Thu, 24 Jan 2008 09:47:03 +0000 (01:47 -0800)]
simplified FileJournal, journal interface
Sage Weil [Wed, 23 Jan 2008 20:59:31 +0000 (12:59 -0800)]
some small cleanups
Sage Weil [Wed, 23 Jan 2008 19:02:58 +0000 (11:02 -0800)]
osd: write sanity check
Sage Weil [Wed, 23 Jan 2008 18:49:54 +0000 (10:49 -0800)]
streamtest.ebofs
Yehuda Sadeh-Weinraub [Wed, 23 Jan 2008 10:59:06 +0000 (12:59 +0200)]
cleanup
Yehuda Sadeh-Weinraub [Wed, 23 Jan 2008 08:05:28 +0000 (10:05 +0200)]
cleanup
Yehuda Sadeh-Weinraub [Wed, 23 Jan 2008 07:57:32 +0000 (09:57 +0200)]
fixed a crash when doing lookup
Sage Weil [Tue, 22 Jan 2008 23:47:43 +0000 (15:47 -0800)]
mdsc: more locking cleanup (not done yet)
Sage Weil [Tue, 22 Jan 2008 21:37:58 +0000 (13:37 -0800)]
rework request alloc in mds_client; use radix_tree_preload
Sage Weil [Tue, 22 Jan 2008 21:36:55 +0000 (13:36 -0800)]
msgr: use radix_tree_preload, and lookup_slot for replacement
Sage Weil [Tue, 22 Jan 2008 21:06:32 +0000 (13:06 -0800)]
msgr: no kmalloc under spinlocks, except make radix tree GFP_ATOMIC
Sage Weil [Tue, 22 Jan 2008 19:51:01 +0000 (11:51 -0800)]
clean up msgr in/out debug output
Sage Weil [Tue, 22 Jan 2008 21:58:51 +0000 (13:58 -0800)]
simplify do_open_request args
Yehuda Sadeh-Weinraub [Tue, 22 Jan 2008 21:23:45 +0000 (23:23 +0200)]
atomic lookup+open
Sage Weil [Tue, 22 Jan 2008 19:46:48 +0000 (11:46 -0800)]
comments
Sage Weil [Tue, 22 Jan 2008 19:38:07 +0000 (11:38 -0800)]
more open refactoring
Sage Weil [Tue, 22 Jan 2008 19:32:43 +0000 (11:32 -0800)]
refactor open a bit; initial pass at ceph_lookup_open
Sage Weil [Tue, 22 Jan 2008 16:17:58 +0000 (08:17 -0800)]
msgr: check for failure to resolve hostname
Sage Weil [Tue, 22 Jan 2008 05:46:13 +0000 (21:46 -0800)]
vfs create() is just OP_MKNOD with rdev=0... make OP_CREATE openc (mknod+open), even tho linux won't use it (yet?)
Sage Weil [Tue, 22 Jan 2008 05:26:33 +0000 (21:26 -0800)]
mdsc: dput after d_find_alias
Sage Weil [Tue, 22 Jan 2008 05:13:45 +0000 (21:13 -0800)]
kernel msgr: fix connection lookup radix tree bucketing bugs
Sage Weil [Tue, 22 Jan 2008 04:34:24 +0000 (20:34 -0800)]
a few client/mds reconnect fixes
Sage Weil [Tue, 22 Jan 2008 04:06:34 +0000 (20:06 -0800)]
mon,msg: small fd cleanups
Sage Weil [Tue, 22 Jan 2008 01:27:48 +0000 (17:27 -0800)]
silly_write
Sage Weil [Mon, 21 Jan 2008 20:33:18 +0000 (12:33 -0800)]
utime_t cleanup
Patience Warnick [Tue, 22 Jan 2008 01:32:07 +0000 (17:32 -0800)]
simplied TCP_CLOSE state, added some prints
Patience Warnick [Tue, 22 Jan 2008 01:30:52 +0000 (17:30 -0800)]
removed some unneeded connection states
Patience Warnick [Tue, 22 Jan 2008 01:29:20 +0000 (17:29 -0800)]
Fixed up fault handling some, and some other bugs..
Merge branch 'rados' of ssh://patience@ceph.newdream.net/home/sage/ceph.newdream.net/git/ceph into
rados
Conflicts:
src/kernel/messenger.c
Sage Weil [Mon, 21 Jan 2008 16:58:52 +0000 (08:58 -0800)]
more todos
Sage Weil [Mon, 21 Jan 2008 16:47:44 +0000 (08:47 -0800)]
todos
Yehuda Sadeh-Weinraub [Mon, 21 Jan 2008 10:41:29 +0000 (12:41 +0200)]
added create (without open)
Yehuda Sadeh-Weinraub [Sun, 20 Jan 2008 21:55:27 +0000 (23:55 +0200)]
change lookup operation to use lstat and not stat (to match expected behavior)
Yehuda Sadeh-Weinraub [Sun, 20 Jan 2008 21:22:02 +0000 (23:22 +0200)]
hash inodes only when required. This fixes hang that occurs when searching for an inode with a hash value that equals to the hash value of the root inode.
Also added default implementation for getattr.
Yehuda Sadeh-Weinraub [Sun, 20 Jan 2008 07:41:39 +0000 (09:41 +0200)]
simplified script
Sage Weil [Sat, 19 Jan 2008 16:25:47 +0000 (08:25 -0800)]
strip_trailing_whitespace.sh
Sage Weil [Sat, 19 Jan 2008 16:18:14 +0000 (08:18 -0800)]
kernel: setattr; moved fill_trace to inode.c
Sage Weil [Fri, 18 Jan 2008 15:49:41 +0000 (07:49 -0800)]
drop caps granted by mds on inodes we dont have; use send_cap_ack helper
Sage Weil [Sat, 19 Jan 2008 00:44:18 +0000 (16:44 -0800)]
todos
Sage Weil [Sat, 19 Jan 2008 00:21:14 +0000 (16:21 -0800)]
crush: check overload map for forcefed values
Sage Weil [Fri, 18 Jan 2008 23:58:54 +0000 (15:58 -0800)]
fixed missing iput; cleanup
Sage Weil [Fri, 18 Jan 2008 23:38:05 +0000 (15:38 -0800)]
close mds sessions on umount
Yehuda Sadeh-Weinraub [Fri, 18 Jan 2008 21:17:55 +0000 (23:17 +0200)]
Merge branch 'rados' of ssh://yehudasa@ceph.newdream.net/git/ceph into myrados
Conflicts:
src/kernel/super.c
Yehuda Sadeh-Weinraub [Fri, 18 Jan 2008 21:15:22 +0000 (23:15 +0200)]
Merge branch 'rados' of ssh://yehudasa@ceph.newdream.net/git/ceph into myrados
Conflicts:
src/kernel/client.c
src/kernel/inode.c
Sage Weil [Fri, 18 Jan 2008 20:29:13 +0000 (12:29 -0800)]
fixed inode cap init bug
Sage Weil [Fri, 18 Jan 2008 20:17:21 +0000 (12:17 -0800)]
renew, resume, etc. caps from kernel client
Sage Weil [Fri, 18 Jan 2008 20:16:55 +0000 (12:16 -0800)]
fix double-delete bug in SimpleMessenger
Sage Weil [Fri, 18 Jan 2008 01:08:27 +0000 (17:08 -0800)]
ebofs: oops, be quiet about csums
Sage Weil [Fri, 18 Jan 2008 01:07:25 +0000 (17:07 -0800)]
Merge branch 'caps' into rados
- fixed up session handler along the way
Sage Weil [Fri, 18 Jan 2008 00:56:48 +0000 (16:56 -0800)]
ebofs: fix superblock csum calc
sage [Fri, 18 Jan 2008 00:34:48 +0000 (16:34 -0800)]
Merge branch 'master' of ssh://ceph.newdream.net/git/ceph into rados
Sage Weil [Fri, 18 Jan 2008 00:21:15 +0000 (16:21 -0800)]
kernel todos
Sage Weil [Fri, 18 Jan 2008 00:20:41 +0000 (16:20 -0800)]
fixed calc_csum alignment on superblock
Sage Weil [Fri, 18 Jan 2008 00:01:36 +0000 (16:01 -0800)]
link caps into a per-mds-session list
Sage Weil [Thu, 17 Jan 2008 22:08:08 +0000 (14:08 -0800)]
fixed add_cap inode ref count bug
Sage Weil [Thu, 17 Jan 2008 21:00:36 +0000 (13:00 -0800)]
resurrected ldceph.so, sort of
Sage Weil [Thu, 17 Jan 2008 20:53:42 +0000 (12:53 -0800)]
cleaned up mds map requesting
Sage Weil [Thu, 17 Jan 2008 19:54:53 +0000 (11:54 -0800)]
clean up mds session open; request mdsmap when necessary
Sage Weil [Tue, 8 Jan 2008 22:00:18 +0000 (14:00 -0800)]
note client reconnect todo
Sage Weil [Thu, 17 Jan 2008 18:27:12 +0000 (10:27 -0800)]
note seq on export
Sage Weil [Thu, 17 Jan 2008 02:26:21 +0000 (18:26 -0800)]
msgr todos
Sage Weil [Thu, 17 Jan 2008 01:52:13 +0000 (17:52 -0800)]
mark_down. some cleanup.
Sage Weil [Thu, 17 Jan 2008 01:44:08 +0000 (17:44 -0800)]
backoff/retry seems to work
Sage Weil [Thu, 17 Jan 2008 00:43:42 +0000 (16:43 -0800)]
rework messenger work queueing, retry
- take a reference whene queueing work; use helpers
- use a READABLE/READING type latches on both try_read and try_write
- move some code around to avoid prototypes
- simplify lossy send_fault case
Sage Weil [Wed, 16 Jan 2008 21:28:24 +0000 (13:28 -0800)]
fix mds session handling