]>
git.apps.os.sepia.ceph.com Git - ceph.git/log
Sage Weil [Thu, 11 Jun 2009 16:49:46 +0000 (09:49 -0700)]
librados: make get pool stats work
Sage Weil [Thu, 11 Jun 2009 16:49:39 +0000 (09:49 -0700)]
mon: process MGetPoolStats
Yehuda Sadeh [Wed, 10 Jun 2009 21:26:11 +0000 (14:26 -0700)]
osd: fix encoding/decoding of [sub]op response
Sage Weil [Thu, 11 Jun 2009 04:28:26 +0000 (21:28 -0700)]
kclient: fix typo
Sage Weil [Thu, 11 Jun 2009 04:21:23 +0000 (21:21 -0700)]
buffer: append individual chars
Sage Weil [Thu, 11 Jun 2009 01:36:16 +0000 (18:36 -0700)]
osd: fix MOSDOp printout
Sage Weil [Thu, 11 Jun 2009 00:52:39 +0000 (17:52 -0700)]
rados: add new headers
Sage Weil [Wed, 10 Jun 2009 23:50:14 +0000 (16:50 -0700)]
librados: client side for getting pg stats
No server side yet!
Sage Weil [Wed, 10 Jun 2009 22:59:35 +0000 (15:59 -0700)]
kclient: make /debug/ceph/client#/mdsc filenames work, cleanup
Just print #ino/path for dentries, since that's what goes over the
wire.
Sage Weil [Wed, 10 Jun 2009 21:30:17 +0000 (14:30 -0700)]
radostool: lssnap
Greg Farnum [Wed, 10 Jun 2009 23:04:12 +0000 (16:04 -0700)]
Added a current bandwidth meter to radostool bench.
Prints out once/second or at every write, whichever is slowest.
Sage Weil [Wed, 10 Jun 2009 21:04:52 +0000 (14:04 -0700)]
osd: destroy messengers on shutdown
Sage Weil [Wed, 10 Jun 2009 21:04:35 +0000 (14:04 -0700)]
monclient: avoid hang during startup
Sage Weil [Wed, 10 Jun 2009 20:37:28 +0000 (13:37 -0700)]
osd: fix huge memory leak (MOSDOp)
Sage Weil [Wed, 10 Jun 2009 19:48:41 +0000 (12:48 -0700)]
radostool: select snaps, rm objects
Yehuda Sadeh [Wed, 10 Jun 2009 19:53:35 +0000 (12:53 -0700)]
rados: get rid of wrcall
Yehuda Sadeh [Wed, 10 Jun 2009 19:16:47 +0000 (12:16 -0700)]
rados: rename rdcall to call
Yehuda Sadeh [Wed, 10 Jun 2009 18:18:17 +0000 (11:18 -0700)]
class: add cls_setxattr
Yehuda Sadeh [Wed, 10 Jun 2009 17:50:25 +0000 (10:50 -0700)]
class: set method flags on registration
Yehuda Sadeh [Wed, 10 Jun 2009 17:26:42 +0000 (10:26 -0700)]
osd: don't call may_read/may_write prematurely (revert previous)
Yehuda Sadeh [Wed, 10 Jun 2009 17:04:11 +0000 (10:04 -0700)]
osd: update rwm flags for incoming ops
Sage Weil [Wed, 10 Jun 2009 03:51:07 +0000 (20:51 -0700)]
radostool: clean up bench args, usage
Sage Weil [Wed, 10 Jun 2009 03:33:15 +0000 (20:33 -0700)]
osd: include size in object_info_t
The main advantage here, aside from slightly decluttering the
code, is to avoid a stat() in get_object_context. We just
getxattr and that will implicitly tell us if the object
exists. (If the file exists but the OI_ATTR xattr doesn't,
the object is busted anyway.)
Sage Weil [Wed, 10 Jun 2009 03:29:34 +0000 (20:29 -0700)]
osd: include ondisk magic in encoding for key ondisk state structs
This will eventually allow us to incrementally migrate to
updated ondisk structures by making the decoding conditional
(dependent on the encoded version). At least in some cases.
Sage Weil [Wed, 10 Jun 2009 03:17:13 +0000 (20:17 -0700)]
rados: remove testradoscio[pp] targets
Sage Weil [Wed, 10 Jun 2009 01:53:16 +0000 (18:53 -0700)]
rados: add helper to header
Sage Weil [Wed, 10 Jun 2009 01:53:00 +0000 (18:53 -0700)]
buffer: remove unused field
Sage Weil [Wed, 10 Jun 2009 01:52:43 +0000 (18:52 -0700)]
osd: fix up OSDOp printing
Greg Farnum [Wed, 10 Jun 2009 00:01:37 +0000 (17:01 -0700)]
radostool now includes benchmarking functionality from testradosciopp.cc
Sage Weil [Tue, 9 Jun 2009 23:48:41 +0000 (16:48 -0700)]
osd: rev osd protocol
Sage Weil [Tue, 9 Jun 2009 23:20:37 +0000 (16:20 -0700)]
kclient: include tid in osdc debugfs dump
Yehuda Sadeh [Tue, 9 Jun 2009 23:33:30 +0000 (16:33 -0700)]
osd: hold op data on the specific op
Sage Weil [Tue, 9 Jun 2009 23:02:15 +0000 (16:02 -0700)]
msgr: calculate data crc after encode_payload, in worker thread
Faster (bc crc calc is in msgr worker thread). Also allows
encode_payload() to manipulate data.
Greg Farnum [Tue, 9 Jun 2009 22:59:54 +0000 (15:59 -0700)]
Removed time_t -> int conversion warnings.
Sage Weil [Tue, 9 Jun 2009 22:22:55 +0000 (15:22 -0700)]
mds: fix dir frag split replay
is_complete() may not be set, but split is still allowed.
Sage Weil [Tue, 9 Jun 2009 22:22:18 +0000 (15:22 -0700)]
osd: don't clobber pg_stats.reported when repairing stats
Sage Weil [Tue, 9 Jun 2009 22:07:24 +0000 (15:07 -0700)]
mds: do not fragment dirs if mds_bal_fragment_interval <= 0
Sage Weil [Tue, 9 Jun 2009 21:58:53 +0000 (14:58 -0700)]
radostool: cleanup
greg [Tue, 9 Jun 2009 21:49:21 +0000 (14:49 -0700)]
Hurray, this file works properly as a standalone.
It takes concurrent writes to maintain, minimum runtime, size of the writes, and whether or not to check for consistency. Reports total runtime, writes made, bandwidth, average & max latency, and time waiting for Rados.
greg [Tue, 9 Jun 2009 00:06:42 +0000 (17:06 -0700)]
Added initial testradosciopp.cc.
greg [Mon, 8 Jun 2009 18:35:57 +0000 (11:35 -0700)]
A properly working testradoscio.c. For the moment it just takes a number of writes to make and does that number in a loop.
Sage Weil [Tue, 9 Jun 2009 21:49:43 +0000 (14:49 -0700)]
osd: simplify op read/write flag setup
Sage Weil [Tue, 9 Jun 2009 21:36:43 +0000 (14:36 -0700)]
osd: kick pg on active set change
Sage Weil [Tue, 9 Jun 2009 21:22:18 +0000 (14:22 -0700)]
common: kill obsolete common/ThreadPool.h
Sage Weil [Tue, 9 Jun 2009 21:18:29 +0000 (14:18 -0700)]
osd: fix osd shutdown vs scrub work queue deadlock
Pause new disk_tp items, kick all (possibly) blocked pgs, and _then_
stop the disk_tp. And make scrub() bail out if the osd is stopping.
Sage Weil [Tue, 9 Jun 2009 20:26:20 +0000 (13:26 -0700)]
mon: fix 'scrub' -> 'repair' in command reply msg
Sage Weil [Tue, 9 Jun 2009 20:24:52 +0000 (13:24 -0700)]
osd: counnt errors fixed during repair, and indicate scrub vs repair
Sage Weil [Tue, 9 Jun 2009 20:03:52 +0000 (13:03 -0700)]
osd: fix pg stats on repair
Sage Weil [Tue, 9 Jun 2009 19:08:05 +0000 (12:08 -0700)]
osd: replica should not check missing map on push
That is, we should let the primary push us objects even if we don't
think they're missing. This is necessary for pg repair to work.
The old check is useless, since we already restart peering when the
pg changes, and ignore pre-change sub_op messages.
Sage Weil [Tue, 9 Jun 2009 19:03:02 +0000 (12:03 -0700)]
osd: improved logging for scrub
Sage Weil [Tue, 9 Jun 2009 18:20:46 +0000 (11:20 -0700)]
osd: fix repair_object attr name
Didn't get switched from 'oi' to '_' way back when.
Sage Weil [Tue, 9 Jun 2009 18:18:01 +0000 (11:18 -0700)]
osd: fix pg parsing
Yehuda Sadeh [Tue, 9 Jun 2009 20:24:31 +0000 (13:24 -0700)]
osd: read/write op flags revision (for class methods)
Yehuda Sadeh [Tue, 9 Jun 2009 20:24:03 +0000 (13:24 -0700)]
class: class acl stuff preliminary stuff
Sage Weil [Tue, 9 Jun 2009 18:08:11 +0000 (11:08 -0700)]
mon: don't dup entires in log.err
Sage Weil [Tue, 9 Jun 2009 17:57:59 +0000 (10:57 -0700)]
osd: fix backlog ordering part deux
Remove the old entry from the map!
Sage Weil [Tue, 9 Jun 2009 17:36:03 +0000 (10:36 -0700)]
osd: repair out of order logs in read_log
Works around bug fixed in f1ff93.
Sage Weil [Tue, 9 Jun 2009 00:36:43 +0000 (17:36 -0700)]
osd: fix order of prior_version backlog entries
They need to end up in the log in the proper order. So make 2 passes,
one to generate the entries, and a second to prepend them to the log.
Sage Weil [Mon, 8 Jun 2009 22:52:14 +0000 (15:52 -0700)]
osd: cleanup
Sage Weil [Mon, 8 Jun 2009 22:51:46 +0000 (15:51 -0700)]
mon: handle failures during CLIENTREPLAY state
Sage Weil [Mon, 8 Jun 2009 21:11:22 +0000 (14:11 -0700)]
radostool: 'ls' fixes
Sage Weil [Mon, 8 Jun 2009 21:05:12 +0000 (14:05 -0700)]
librados: fix locking on sync methods
Sage Weil [Mon, 8 Jun 2009 20:53:07 +0000 (13:53 -0700)]
objclass: fix compilation
Yehuda Sadeh [Mon, 8 Jun 2009 20:48:28 +0000 (13:48 -0700)]
class: makefile, include fixes
Yehuda Sadeh [Mon, 8 Jun 2009 20:22:03 +0000 (13:22 -0700)]
conf: avoid parsing daemon related config options when not needed
Sage Weil [Mon, 8 Jun 2009 19:57:39 +0000 (12:57 -0700)]
osd: copy soid out of iterator
A ref into the iterator is inherently risky.
Sage Weil [Mon, 8 Jun 2009 19:08:49 +0000 (12:08 -0700)]
kclient: filemap_write_and_wait_range not exported < 2.6.30
Include a warning that snap flushing may not work.
Sage Weil [Mon, 8 Jun 2009 19:08:09 +0000 (12:08 -0700)]
cosd: put mds on ceph4
Sage Weil [Mon, 8 Jun 2009 19:07:25 +0000 (12:07 -0700)]
osd: fix use-after-free on osd
push_to_replica is called on a sobject_t ref from the missing map,
and then removes the item from said missing map. Reorder to avoid
the soid mangling.
Sage Weil [Mon, 8 Jun 2009 16:54:30 +0000 (09:54 -0700)]
osd: write pg info _after_ updating log
Otherwise the log bounds don't match the PG::Info!
Sage Weil [Mon, 8 Jun 2009 05:20:42 +0000 (22:20 -0700)]
mds: restore session state on load; only save opened sessions
Sage Weil [Sun, 7 Jun 2009 22:21:54 +0000 (15:21 -0700)]
osd: remove bad assertions from IndexedLog::print()
Sage Weil [Mon, 8 Jun 2009 18:03:01 +0000 (11:03 -0700)]
librados: add/fix locking
Sage Weil [Mon, 8 Jun 2009 16:43:53 +0000 (09:43 -0700)]
librados: rename AioCompletion::put -> release
Sage Weil [Mon, 8 Jun 2009 18:01:01 +0000 (11:01 -0700)]
filestore: escape / and leading . in object names
Conflicts:
src/testrados.c
Sage Weil [Sun, 7 Jun 2009 22:18:23 +0000 (15:18 -0700)]
mds: allow online changes in mds_log_max_segments, events
Sage Weil [Sun, 7 Jun 2009 22:17:48 +0000 (15:17 -0700)]
mds: move 'buf' column to right
Sage Weil [Sun, 7 Jun 2009 22:17:25 +0000 (15:17 -0700)]
osd: only update peer_info if !noop
Sage Weil [Sun, 7 Jun 2009 21:28:49 +0000 (14:28 -0700)]
mds: log memory stats to mds%d.mem
Sage Weil [Sat, 6 Jun 2009 22:55:56 +0000 (15:55 -0700)]
librados: basic snap api?
Not actually implemented.
Sage Weil [Sat, 6 Jun 2009 21:44:07 +0000 (14:44 -0700)]
radostool: add 'get' and 'put'
Sage Weil [Sat, 6 Jun 2009 21:38:17 +0000 (14:38 -0700)]
radostool: add simple radostool shell
So far just 'lspools' works.
Sage Weil [Sat, 6 Jun 2009 21:04:08 +0000 (14:04 -0700)]
add Completion class
This should eventually replace Context. The main difference is instead
of
context->finish(r);
delete context;
you do
completion->complete(r);
and the completion should delete this as needed. It's more flexible
because it doesn't require a new heap allocation for each caller
reference. And 'delete this' is legal, so long as you do not access any
member fields after.
Sage Weil [Sat, 6 Jun 2009 04:34:58 +0000 (21:34 -0700)]
objclass: trivialmap
Sage Weil [Sat, 6 Jun 2009 04:33:32 +0000 (21:33 -0700)]
objclass: add support for c++ methods
Sage Weil [Sat, 6 Jun 2009 04:15:38 +0000 (21:15 -0700)]
librados: clean up snap and ctx a bit
Define PoolCtx constructor, methods.
Sage Weil [Sat, 6 Jun 2009 04:14:47 +0000 (21:14 -0700)]
buffer: add ability to claim ownership of memory allocated via new or malloc()
Sage Weil [Sat, 6 Jun 2009 00:45:50 +0000 (17:45 -0700)]
librados: aio completion notification via a callback
greg [Fri, 5 Jun 2009 23:54:06 +0000 (16:54 -0700)]
Added testradoscio.
Yehuda Sadeh [Fri, 5 Jun 2009 20:24:25 +0000 (13:24 -0700)]
rados: can set snapshot sequence
Yehuda Sadeh [Thu, 4 Jun 2009 19:00:44 +0000 (12:00 -0700)]
rados: fixes for dynamic oids
Sage Weil [Fri, 5 Jun 2009 20:07:06 +0000 (13:07 -0700)]
memorymodel: fix crash on weird input
Sage Weil [Fri, 5 Jun 2009 18:56:39 +0000 (11:56 -0700)]
rados: rados.initialize returns negative on error
Sage Weil [Fri, 5 Jun 2009 18:54:57 +0000 (11:54 -0700)]
rados: rados_initialize returns negative on error
Sage Weil [Fri, 5 Jun 2009 18:52:43 +0000 (11:52 -0700)]
osd: add 'osd pool rmsnap poolname snapname' command
Sage Weil [Fri, 5 Jun 2009 18:46:17 +0000 (11:46 -0700)]
mon: create pool snaps
Sage Weil [Fri, 5 Jun 2009 17:52:14 +0000 (10:52 -0700)]
osd: use the pool snapc if client doesn't specify one
Sage Weil [Fri, 5 Jun 2009 17:50:21 +0000 (10:50 -0700)]
osd: handle removal of pool snaps
Use helper to build effective removed_snaps set (regardless of
whether they are explicitly specified in the pg_pool_t or implicity
via the pool snaps). Keep a copy of that set in the OSD's PGPool
so we can track changes.
Sage Weil [Fri, 5 Jun 2009 17:43:06 +0000 (10:43 -0700)]
osd: purge snaps on a per-pool basis
Maintain a refcounted PGPool object in the OSD to track pool-wide
state (such as the latest SnapContext, snap trimming info).
Doesn't trim per-pool snaps just yet.
Sage Weil [Fri, 5 Jun 2009 04:10:57 +0000 (21:10 -0700)]
encoding: kill T* templates
These are a bit sneaky because they match things like 'const char *'
in a non-obvious way. We aren't using them, so turn them off for now!