]>
git.apps.os.sepia.ceph.com Git - ceph.git/log
Sage Weil [Mon, 1 Feb 2010 21:34:38 +0000 (13:34 -0800)]
filestore: fix empty cond location
Sage Weil [Sat, 30 Jan 2010 00:37:30 +0000 (16:37 -0800)]
mount: fix compile errors
Sage Weil [Sat, 30 Jan 2010 00:36:10 +0000 (16:36 -0800)]
streamtest: do concurrent ios; async commit AND ack
Sage Weil [Sat, 30 Jan 2010 00:35:04 +0000 (16:35 -0800)]
osd: add mutual exclusion while writes are applied
Sage Weil [Sat, 30 Jan 2010 00:24:39 +0000 (16:24 -0800)]
Merge commit 'origin/unstable' into filestore
Greg Farnum [Fri, 29 Jan 2010 23:48:02 +0000 (15:48 -0800)]
TODO: mark bugs completed
Greg Farnum [Fri, 29 Jan 2010 23:01:53 +0000 (15:01 -0800)]
mount: canonicalize path name for mtab
Greg Farnum [Fri, 29 Jan 2010 21:05:44 +0000 (13:05 -0800)]
mon: possible fix for rare problem where osd boot messages are double-logged.
From Sage.
Greg Farnum [Fri, 29 Jan 2010 21:02:14 +0000 (13:02 -0800)]
mon: don't delete all laggy mds beacons, just the one's you've replaced
Greg Farnum [Fri, 29 Jan 2010 18:15:48 +0000 (10:15 -0800)]
mon: print out proper address in debug log
Greg Farnum [Thu, 28 Jan 2010 20:21:46 +0000 (12:21 -0800)]
mds: unify two if blocks with identical checks
Greg Farnum [Wed, 27 Jan 2010 23:40:09 +0000 (15:40 -0800)]
todo: Slightly out of date
Sage Weil [Fri, 29 Jan 2010 21:27:29 +0000 (13:27 -0800)]
journal: fix padding
Sage Weil [Fri, 29 Jan 2010 19:38:01 +0000 (11:38 -0800)]
osd: streamline writing new maps to disk, somewhat
There is still an apply_transaction that should be replaced with a queue
if/when we are smarter about map caching so that a call to get_map on the
prior epoch won't need to go to disk. Until then, the implicit flush
in apply_transaction is needed.
Sage Weil [Fri, 29 Jan 2010 19:22:45 +0000 (11:22 -0800)]
osd: use flush, not sync_on_flush
Sage Weil [Fri, 29 Jan 2010 19:22:35 +0000 (11:22 -0800)]
filestore: flush() will wait for queued transactions to be readable
Sage Weil [Fri, 29 Jan 2010 18:17:39 +0000 (10:17 -0800)]
journal: add flags field to header, entry_header
Sage Weil [Thu, 28 Jan 2010 05:09:15 +0000 (21:09 -0800)]
journal: allow pre and post padding for journal payload
Replace the header.alignment fixups with an explicit post_pad.
Sage Weil [Thu, 28 Jan 2010 04:22:18 +0000 (20:22 -0800)]
debian: use bash for debian build script
Sage Weil [Thu, 28 Jan 2010 18:55:10 +0000 (10:55 -0800)]
journal: fix thread id printing
Sage Weil [Thu, 28 Jan 2010 18:33:04 +0000 (10:33 -0800)]
finisher: fix flushing, debug output
Sage Weil [Thu, 28 Jan 2010 18:04:28 +0000 (10:04 -0800)]
dout: print thread id hex (to match up with gdb)
Sage Weil [Wed, 27 Jan 2010 22:27:57 +0000 (14:27 -0800)]
filepath: add struct_v
Sage Weil [Wed, 27 Jan 2010 22:27:26 +0000 (14:27 -0800)]
common/include/rgw: Add versioning to structs without it.
Sage Weil [Wed, 27 Jan 2010 21:23:10 +0000 (13:23 -0800)]
mds: add v to encoded mds structs
mds: Begin auditing non-struct encodes
mds: add versioning to the events and Anchor.
mds fixes
Sage Weil [Wed, 27 Jan 2010 21:34:46 +0000 (13:34 -0800)]
rgw: struct_v in RGWObjEnt
Sage Weil [Wed, 27 Jan 2010 21:34:27 +0000 (13:34 -0800)]
journaler: struct_v in structs
Greg Farnum [Tue, 26 Jan 2010 23:08:12 +0000 (15:08 -0800)]
mds: remove unused _pad variable
Sage Weil [Wed, 27 Jan 2010 21:33:02 +0000 (13:33 -0800)]
osd: add struct_v to most encoded structs that don't already have one
Sage Weil [Wed, 27 Jan 2010 22:26:11 +0000 (14:26 -0800)]
msg: add type to ceph_entity_addr
Sage Weil [Wed, 27 Jan 2010 19:33:11 +0000 (11:33 -0800)]
osd: set reqid for push/pull subops
This just makes tracking down requests in the debug output easier.
Sage Weil [Wed, 27 Jan 2010 19:24:30 +0000 (11:24 -0800)]
filestore: fix journal flush
Sage Weil [Wed, 27 Jan 2010 19:01:21 +0000 (11:01 -0800)]
osd: use queue_transaction throughout
Only apply_transaction in mkfs/mount/umount, and in handle_map.
Sage Weil [Wed, 27 Jan 2010 18:49:48 +0000 (10:49 -0800)]
filestore: queue onreadable _after_ starting journal item
The onreadable is allowed to delete the Transaction
Sage Weil [Wed, 27 Jan 2010 18:49:30 +0000 (10:49 -0800)]
filestore: queue_transaction and generic C_DeleteTransaction
Sage Weil [Wed, 27 Jan 2010 18:13:39 +0000 (10:13 -0800)]
osd: reply with ACK if client doesn't ask for a COMMIT
Sage Weil [Wed, 27 Jan 2010 18:00:02 +0000 (10:00 -0800)]
journal: include a version field in header
For future ondisk format changes
Sage Weil [Wed, 27 Jan 2010 00:35:11 +0000 (16:35 -0800)]
osd: simplify pg write mode transitions
There is no reason to track client on the per-pg AccessMode, since we only
care about read/write races on a per-object state, and need to track that
per-object anyway due to the new async commit/apply model. (Though this
is semi-broken even without that...)
Currently we will move from delayed to rmw, but there is no rmw->delayed
transition, since the logic for that is a bit more complex.. it'll be
difficult to determine when it is really a win.
In general, though, RMW is probably a better all around policy!
Sage Weil [Tue, 26 Jan 2010 23:26:26 +0000 (15:26 -0800)]
osd: only use DELAYED if no journal or journal is writebehind
Sage Weil [Tue, 26 Jan 2010 22:52:34 +0000 (14:52 -0800)]
osd: make Transaction lifetime match RepModify (not stack)
Sage Weil [Tue, 26 Jan 2010 21:37:35 +0000 (13:37 -0800)]
dout: print thread id hex (to match up with gdb)
Sage Weil [Tue, 26 Jan 2010 21:06:02 +0000 (13:06 -0800)]
wip
Sage Weil [Tue, 26 Jan 2010 00:24:25 +0000 (16:24 -0800)]
finisher: wait_for_empty
Sage Weil [Thu, 7 Jan 2010 22:35:28 +0000 (14:35 -0800)]
filestore: add writeahead journal support
Sage Weil [Thu, 7 Jan 2010 22:35:03 +0000 (14:35 -0800)]
filejournal: optionally block on full journal
Note that there is deadlock potential here if the thing we're trying to
journal is just too big!
Sage Weil [Thu, 7 Jan 2010 21:13:10 +0000 (13:13 -0800)]
filestore: refactor applied/committed completions
Sage Weil [Thu, 17 Dec 2009 19:31:12 +0000 (11:31 -0800)]
osd: refactor op_applied logic
Behavior is still essentially unchanged (for now).
Sage Weil [Thu, 17 Dec 2009 00:23:26 +0000 (16:23 -0800)]
filestore: preliminary support for parallel journaling
- fixed issue with normal write-behind journal (op_seq wasn't updated
inside the fs transaction)
- reworked journal interface to support write-behind, parallel, write-ahead
(hopefully)
Sage Weil [Wed, 16 Dec 2009 23:39:48 +0000 (15:39 -0800)]
finish: support non-zero return values
Sage Weil [Mon, 25 Jan 2010 21:39:40 +0000 (13:39 -0800)]
osd: encode struct v for ceph_pg_pool
Sage Weil [Mon, 25 Jan 2010 18:29:24 +0000 (10:29 -0800)]
mon: add struct encoding v to ease future revisions
Sage Weil [Mon, 25 Jan 2010 19:26:34 +0000 (11:26 -0800)]
mds: make CEPH_MDS_OP_CREATE a write op [mds protocol change]
Sage Weil [Mon, 25 Jan 2010 18:34:18 +0000 (10:34 -0800)]
ceph: remove stray #include
Sage Weil [Fri, 15 Jan 2010 19:00:47 +0000 (11:00 -0800)]
librados: implement C versions of snap calls
Greg Farnum [Tue, 19 Jan 2010 20:24:31 +0000 (12:24 -0800)]
Change CompatSet usage to be compatible with new struct
Greg Farnum [Fri, 15 Jan 2010 21:16:46 +0000 (13:16 -0800)]
change CompatSet
Greg Farnum [Fri, 15 Jan 2010 00:37:09 +0000 (16:37 -0800)]
mds: Revert CInode/Journaler changes as we'll move them into MDSMap later.
Greg Farnum [Fri, 15 Jan 2010 00:27:14 +0000 (16:27 -0800)]
add versioning and awareness bla bla
Greg Farnum [Thu, 14 Jan 2010 19:12:23 +0000 (11:12 -0800)]
mon: use CompatSet in disk format
Greg Farnum [Thu, 14 Jan 2010 19:11:51 +0000 (11:11 -0800)]
mds: use CompatSet in disk format
Greg Farnum [Wed, 13 Jan 2010 21:17:18 +0000 (13:17 -0800)]
osd: use CompatSet in the superblock. Only one 'feature' right now
Greg Farnum [Tue, 12 Jan 2010 00:53:11 +0000 (16:53 -0800)]
include: new CompatSet.h
Yehuda Sadeh [Fri, 15 Jan 2010 18:33:57 +0000 (10:33 -0800)]
replace strcat with strncat
Yehuda Sadeh [Fri, 15 Jan 2010 00:07:39 +0000 (16:07 -0800)]
replace sprintf with snprintf all around
Also pass buffer size to relevant functions when required.
Sage Weil [Fri, 15 Jan 2010 00:06:17 +0000 (16:06 -0800)]
msgr: fix mark_down
Beware non-pure virtual methods in the abstract interface!
Sage Weil [Thu, 14 Jan 2010 23:46:26 +0000 (15:46 -0800)]
filestore: don't croak if current/ exists
Sage Weil [Thu, 14 Jan 2010 22:26:19 +0000 (14:26 -0800)]
todo
Yehuda Sadeh [Thu, 14 Jan 2010 22:34:59 +0000 (14:34 -0800)]
logger: use larger buffer, snprinf instead of sprintf
Sage Weil [Thu, 7 Jan 2010 22:33:25 +0000 (14:33 -0800)]
osd: use journal in peek_super
Sage Weil [Fri, 8 Jan 2010 18:25:20 +0000 (10:25 -0800)]
filestore: put data in current/, a btrfs subvol if possible
Sage Weil [Thu, 14 Jan 2010 21:38:44 +0000 (13:38 -0800)]
filestore: rip out usertrans experiment
Sage Weil [Thu, 14 Jan 2010 21:34:59 +0000 (13:34 -0800)]
filestore: rename _apply_transaction
Conflicts:
src/os/FileStore.h
Sage Weil [Thu, 14 Jan 2010 19:52:58 +0000 (11:52 -0800)]
mds: add 'dumpcache [filename]' mds command
Sage Weil [Thu, 14 Jan 2010 19:52:20 +0000 (11:52 -0800)]
mds: disable cache dump on rejoin by default
Greg Farnum [Fri, 8 Jan 2010 23:42:36 +0000 (15:42 -0800)]
TODO
Greg Farnum [Sat, 9 Jan 2010 01:25:53 +0000 (17:25 -0800)]
mon: properly clean up SessionMap on a shutdown.
This fixes failed asserts whenever you tried to do a clean monitor shutdown
instead of killing it.
Greg Farnum [Fri, 8 Jan 2010 21:42:46 +0000 (13:42 -0800)]
csyn: properly use multiple clients
Sage Weil [Fri, 8 Jan 2010 19:19:23 +0000 (11:19 -0800)]
don't keep configs in git
Sage Weil [Fri, 8 Jan 2010 18:13:50 +0000 (10:13 -0800)]
todo
Greg Farnum [Fri, 8 Jan 2010 01:35:55 +0000 (17:35 -0800)]
botched merge
Greg Farnum [Fri, 8 Jan 2010 01:15:39 +0000 (17:15 -0800)]
Merge branch 'unstable' of ceph.newdream.net:/git/ceph into unstable
Conflicts:
src/msg/SimpleMessenger.cc
Greg Farnum [Fri, 8 Jan 2010 01:10:37 +0000 (17:10 -0800)]
msgr: make use of references and const slightly more consistent
Greg Farnum [Fri, 8 Jan 2010 01:10:21 +0000 (17:10 -0800)]
monClient: fix potential naming confusion with messenger
Greg Farnum [Fri, 8 Jan 2010 00:56:28 +0000 (16:56 -0800)]
msgr: rename endpoint_stopped to destination_stopped
Greg Farnum [Fri, 8 Jan 2010 00:53:51 +0000 (16:53 -0800)]
msgr: rank is just a bad name for local SimpleMessengers now
Sage Weil [Fri, 8 Jan 2010 00:10:23 +0000 (16:10 -0800)]
msgr: remove erank
Obsolete, now that we have only a single entity per SimpleMessenger
instance. Avoid changing protocol by making it a reserved (and ignored)
field.
Greg Farnum [Thu, 7 Jan 2010 23:35:26 +0000 (15:35 -0800)]
msgr: should be destroyed in the same thread it's created
Sage Weil [Thu, 17 Dec 2009 20:11:16 +0000 (12:11 -0800)]
filestore: don't bother masking signals
The OSD already catches SIGINT, SIGTERM. It's not really
effective anyway.
Greg Farnum [Thu, 7 Jan 2010 20:05:38 +0000 (12:05 -0800)]
msgr: Remove erroneous rank->destroy calls I inserted in 9a96bf
Greg Farnum [Thu, 7 Jan 2010 01:15:44 +0000 (17:15 -0800)]
msgr: get and put inside DispatchThread, not when you create/end DispatchThread
Greg Farnum [Thu, 7 Jan 2010 00:14:45 +0000 (16:14 -0800)]
msgr: Remove unregister_entity since there's only ever one now.
Move needed functionality into the end of dispatch_entry.
Greg Farnum [Wed, 6 Jan 2010 22:13:05 +0000 (14:13 -0800)]
msgr: Make usage compatible with last commit
Greg Farnum [Wed, 6 Jan 2010 01:09:41 +0000 (17:09 -0800)]
msgr: Remove all references to Endpoint
Greg Farnum [Wed, 6 Jan 2010 00:51:01 +0000 (16:51 -0800)]
msgr: Remove declarations of unimplemented functions:
rename_entity and new_entity
Greg Farnum [Wed, 6 Jan 2010 00:48:17 +0000 (16:48 -0800)]
msgr: SimpleMessenger takes responsibility for Messenger functions.
Greg Farnum [Tue, 5 Jan 2010 20:02:59 +0000 (12:02 -0800)]
msgr: All SimpleMessenger uses are now on the heap for appropriate destruction
Greg Farnum [Tue, 5 Jan 2010 00:04:28 +0000 (16:04 -0800)]
msgr: SimpleMessenger is now a Messenger itself -- sort of
Greg Farnum [Mon, 4 Jan 2010 18:01:41 +0000 (10:01 -0800)]
msgr: remove rank number from endpoint, since there's only one.
Remove Endpoint's wait() method as it's unused.
Sage Weil [Thu, 7 Jan 2010 00:08:29 +0000 (16:08 -0800)]
todo
Sage Weil [Thu, 7 Jan 2010 00:03:31 +0000 (16:03 -0800)]
osd: fix MASKTRUNC
This was looking at the wrong osd_op for the truncate_seq, which made it
essentially broken. This was problematic any time the client SETTRUNC
reached an object (on a write after truncate) before the MDS TRIMTRUNC did.