]>
git.apps.os.sepia.ceph.com Git - ceph.git/log
Yehuda Sadeh [Wed, 28 Jul 2010 18:35:14 +0000 (11:35 -0700)]
librados: quiet down some log messages
Wido den Hollander [Wed, 28 Jul 2010 16:48:33 +0000 (18:48 +0200)]
rgw: Support loglevel settings
The gateway now supports changing the verbosity of the logging, this can be achieved by changing the
RGW_LOG_LEVEL envirioment parameter.
For Apache running systems, this can be done with mod_env: SetEnv RGW_LOG_LEVEL 20
This will change the loglevel to the highest level and produce a lot of usefull output for debugging purposes.
For now, all the messages will go to webservers logfile (For Apache: error_log)
Sage Weil [Tue, 27 Jul 2010 20:22:36 +0000 (13:22 -0700)]
qa: consistent snaptest-%d.sh naming
Sage Weil [Tue, 27 Jul 2010 19:53:20 +0000 (12:53 -0700)]
mds: fix uninitialized LeaseStat for null lease
./include/encoding.h: In member function void
Locker::issue_client_lease(CDentry*, client_t, ceph::bufferlist&, utime_t,
Session*):
./include/encoding.h:80: warning: e.LeaseStat::duration_ms may be used
uninitialized in this function
Sage Weil [Tue, 27 Jul 2010 19:18:56 +0000 (12:18 -0700)]
osd: log when we get marked down but aren't
Sage Weil [Tue, 27 Jul 2010 18:20:02 +0000 (11:20 -0700)]
debug: no name symlink when explicit --log-file
Sage Weil [Tue, 27 Jul 2010 17:05:27 +0000 (10:05 -0700)]
client: some whitespace cleanup
Greg Farnum [Tue, 27 Jul 2010 19:13:04 +0000 (12:13 -0700)]
qa: add localized version of Thomas Mueller's snaptest-2.sh
http://github.com/vinzent/ceph-testsuite/blob/master/tests/snaptest-2
Yehuda Sadeh [Tue, 27 Jul 2010 18:18:06 +0000 (11:18 -0700)]
rgw: exit after 5 seconds from SIGUSR1 anyway
Wido den Hollander [Tue, 27 Jul 2010 07:46:37 +0000 (09:46 +0200)]
rgw: fix that we return data when partial content is requested
Yehuda Sadeh [Mon, 26 Jul 2010 23:32:26 +0000 (16:32 -0700)]
rgw: handle sigusr1, get notified for shutdown requests
Yehuda Sadeh [Mon, 26 Jul 2010 22:34:31 +0000 (15:34 -0700)]
rgw: fix returned error code in If-None-Match
Wido den Hollander [Mon, 26 Jul 2010 20:58:26 +0000 (22:58 +0200)]
rgw: Make sure all times are converted to GMT, so we can compare them correctly
Yehuda Sadeh [Mon, 26 Jul 2010 22:22:41 +0000 (15:22 -0700)]
rgw: make sure error structure is initialized
Wido den Hollander [Fri, 23 Jul 2010 22:00:13 +0000 (15:00 -0700)]
rgw: fix date/time format string
Yehuda Sadeh [Fri, 23 Jul 2010 21:07:02 +0000 (14:07 -0700)]
rgw: fix return code for prepare_get_obj() on the fs storage
Yehuda Sadeh [Fri, 23 Jul 2010 21:01:47 +0000 (14:01 -0700)]
rgw: fix error status code
Wido den Hollander [Fri, 23 Jul 2010 14:40:48 +0000 (16:40 +0200)]
We should sent back the Last-Modified header in GMT as specified by RFC2616
Wido den Hollander [Fri, 23 Jul 2010 13:39:49 +0000 (15:39 +0200)]
We shoud always return the Accept-Ranges header, with a GET or HEAD request, even when the client is not requesting for partial content.
This is to let the client know we accept requests for partial content.
On HEAD requests we should also return the Content-Length header.
And we should also return the Last-Modified header with GET and HEAD requests, this way browsers and proxy's can cache content.
Yehuda Sadeh [Thu, 22 Jul 2010 23:48:38 +0000 (16:48 -0700)]
rgw: support partial content (206) response
Couldn't get this one to work yet, as apache (or probably one of
the apache filters) rewrites the content-length value and sets
it to zero.
Greg Farnum [Mon, 26 Jul 2010 23:43:16 +0000 (16:43 -0700)]
mds: Use get_oldest_snap() (not first) in handle_client_lssnap.
Greg Farnum [Fri, 23 Jul 2010 23:29:50 +0000 (16:29 -0700)]
rgw: remove unused label 'done' (compiler warning)
Sage Weil [Mon, 26 Jul 2010 22:56:33 +0000 (15:56 -0700)]
msgr: fix direct send to loopback connection
If we send directly to the local/loopback connection, we need to
queue it for local delivery, not go through the usual send paths.
This fixes lost replies to messages sent to self (e.g., reply to
local anchortable).
Signed-off-by: Sage Weil <sage@newdream.net>
Sage Weil [Mon, 26 Jul 2010 22:03:11 +0000 (15:03 -0700)]
mon: revamp subscribe protocol [backward compatible protocol change]
Before, we would provide "have" and a bool "onetime" flag. The struct was
also screwed up with an extra __le64. Then have=0 was a special case
that meant "give me the latest".
The problem is this is ambiguous between the usual "give me everything
since X" and "give me your latest", because you might actually have 0 and
want 1..current.
Changes protocol and cleans up the struct:
- now "start" and "flags", where only 1 flag (ONETIME) is defined
- clean up sub_want_* methods throughout
- fix all sub_want callers to ask for _start_ (not have) epoch, or 0 for
any/latest
- add a feature bit; talks old clients w/o that bit
Signed-off-by: Sage Weil <sage@newdream.net>
Sage Weil [Mon, 26 Jul 2010 18:44:59 +0000 (11:44 -0700)]
mds: restore CInode::get_oldest_snap()
It's either 'first', for the first for the oldest old_inode.
Sage Weil [Mon, 26 Jul 2010 17:27:50 +0000 (10:27 -0700)]
mds: defer anchortable updates before CLIENTREPLAY, not ACTIVE
Otherwise we hang trying to replay an op that interacts with the
anchortable.
Sage Weil [Mon, 26 Jul 2010 17:24:43 +0000 (10:24 -0700)]
Merge remote branch 'origin/large_objects' into unstable
Sage Weil [Sat, 24 Jul 2010 04:02:47 +0000 (21:02 -0700)]
mds: get rid of old lease mask cruft
Signed-off-by: Sage Weil <sage@newdream.net>
Sage Weil [Sat, 24 Jul 2010 03:51:58 +0000 (20:51 -0700)]
mds: fix lease revocation mask vs type
Sage Weil [Sat, 24 Jul 2010 16:50:31 +0000 (09:50 -0700)]
osd: on dups, reply if committed, or wait until ondisk
Sage Weil [Sat, 24 Jul 2010 16:49:52 +0000 (09:49 -0700)]
osd: helper should reply with ACK on error, op flags on success
Sage Weil [Fri, 23 Jul 2010 23:23:15 +0000 (16:23 -0700)]
osd: map indexed caller_ops back to relevant entry
This let's us look up a version from the osd_reqid_t.
Sage Weil [Fri, 23 Jul 2010 23:19:13 +0000 (16:19 -0700)]
osd: maintain pg last_update_ondisk
Only defined when is_active(). Only printed when it lags (i.e. we are
in the process of replicating and/or writing an update.
Sage Weil [Fri, 23 Jul 2010 23:05:20 +0000 (16:05 -0700)]
initscripts: fix type.id check in ceph.conf
Sage Weil [Fri, 23 Jul 2010 22:53:47 +0000 (15:53 -0700)]
Merge branch 'unstable' into large_objects
Sage Weil [Fri, 23 Jul 2010 19:57:52 +0000 (12:57 -0700)]
osd: make assemble_backlog more tolerant of races
The assemble_backlog is merging data generated while racing with online
updates. It needs to tolerate races with things like delete. For example,
- generate_backlog identifies object A
- client deletes+logs A
- assemble_backlog sees A deletion entry.
We may want to merge the backlog entry in this case.
Sage Weil [Thu, 22 Jul 2010 23:24:54 +0000 (16:24 -0700)]
osd: skip degraded objects we're already recoverying in recover_replicas
Otherwise we'll start_recovery_op, dup the push, etc.
Signed-off-by: Sage Weil <sage@newdream.net>
Sage Weil [Thu, 22 Jul 2010 23:03:53 +0000 (16:03 -0700)]
osd: create missing temp collection/dir
Sage Weil [Thu, 22 Jul 2010 21:58:43 +0000 (14:58 -0700)]
osd: populate in-memory snapset context with just-pulled SnapSet
This can come from the SNAPDIR or NOSNAP (head) object.
Signed-off-by: Sage Weil <sage@newdream.net>
Sage Weil [Thu, 22 Jul 2010 21:44:24 +0000 (14:44 -0700)]
osd: clean up size learning during pull
Signed-off-by: Sage Weil <sage@newdream.net>
Sage Weil [Fri, 23 Jul 2010 22:41:46 +0000 (15:41 -0700)]
mds: drop remove_client_lease return value
Unused
Sage Weil [Fri, 23 Jul 2010 22:39:53 +0000 (15:39 -0700)]
mds: fix dentry lease mask/lockid disconnect
There is only one lock here, no need to be tricky with the mask.
Sage Weil [Fri, 23 Jul 2010 22:15:40 +0000 (15:15 -0700)]
mds: no return value for issue_client_lease()
Return value is unused. Also drop obsolete inode lease method declaration.
Signed-off-by: Sage Weil <sage@newdream.net>
Sage Weil [Fri, 23 Jul 2010 22:14:43 +0000 (15:14 -0700)]
mds: fix lease issue mask
We define 1 to be the only lease "mask" we currently support (for dentry)
and divorce ourselves from the CEPH_LOCK namespace for this purpose. We
did this in
c8d7c970e864e9f130b7c44b29c322771067026e , but screwed up the
issue function. Fix the caller, and pay attention to @mask here.
Signed-off-by: Sage Weil <sage@newdream.net>
Sage Weil [Fri, 23 Jul 2010 21:54:44 +0000 (14:54 -0700)]
mds: create leases with non-zero seq
We want to avoid the situation:
- mds issues lease, seq 0
- mds times out lease, seq 0
- client sends seq 0 release
- mds issues new lease, seq 0 (again.. poor choice of seq#)
- mds processes client lease seq 0 release (bad)
Sage Weil [Fri, 23 Jul 2010 21:41:07 +0000 (14:41 -0700)]
osd: log/debug hb set changes
Sage Weil [Fri, 23 Jul 2010 21:40:59 +0000 (14:40 -0700)]
osdmap: recalc num_osds after applying incremental
Sage Weil [Fri, 23 Jul 2010 19:59:11 +0000 (12:59 -0700)]
osd: clean_up_local should ignore snapped, snapdir objects
The snap_trimmer is responsible for removing any/all snap objects.
snapdir only exists if snapped objects exist and head doesn't, so it will
get cleaned up when the snapped object(s) do.
Sage Weil [Fri, 23 Jul 2010 18:08:48 +0000 (11:08 -0700)]
filestore: make stat print correct errno
Sage Weil [Thu, 22 Jul 2010 22:55:23 +0000 (15:55 -0700)]
osd: simplify heartbeat checks
- Only check heartbeats when we have heartbeat_lock and osdmap rdlocked,
and thus _know_ heartbeat info and map are in sync. Drop unnecessary
consistency checks.
- Check heartbeats in tick(), since the heartbeat thread may miss it if
it's unlucky.
Sage Weil [Thu, 22 Jul 2010 21:36:00 +0000 (14:36 -0700)]
osd: red is good enough; don't blink
Greg Farnum [Fri, 23 Jul 2010 20:42:07 +0000 (13:42 -0700)]
mds: only try and lssnaps on a dir for as long as it's been around
Greg Farnum [Tue, 20 Jul 2010 22:53:31 +0000 (15:53 -0700)]
qa: runoncfuse shouldn't start cfuse in background any more.
Greg Farnum [Tue, 20 Jul 2010 22:52:43 +0000 (15:52 -0700)]
mds: fix snap lookups in CInode encoding
Sage Weil [Fri, 23 Jul 2010 20:21:36 +0000 (13:21 -0700)]
mon: don't crash on 'osd pool set' command
Sage Weil [Fri, 23 Jul 2010 18:45:26 +0000 (11:45 -0700)]
osd: zero ondisklog pointers when starting pg deletion
This fixes a problem where the osd stops part way through pg cleanup. It
seens the old ondisklog bounds, but then fails to load the log, and crashes
on startup.
We just need to zero the ondisklog bounds when we zero the log.
Signed-off-by: Sage Weil <sage@newdream.net>
Yehuda Sadeh [Fri, 23 Jul 2010 18:43:25 +0000 (11:43 -0700)]
librados: remove use of snapid_t
Sage Weil [Fri, 23 Jul 2010 16:42:51 +0000 (09:42 -0700)]
objecter: default debug level 0 (no 'laggy' messages)
Sage Weil [Thu, 22 Jul 2010 21:27:08 +0000 (14:27 -0700)]
initscripts: use $type.$id, not $name, when possible
Sage Weil [Thu, 22 Jul 2010 21:26:49 +0000 (14:26 -0700)]
initscripts: accept osd0 or osd.0, regardless of what conf uses
Sage Weil [Thu, 22 Jul 2010 21:17:56 +0000 (14:17 -0700)]
Merge remote branch 'origin/large_objects' into unstable
Sage Weil [Thu, 22 Jul 2010 19:01:11 +0000 (12:01 -0700)]
mds: be careful obeying REQRDLOCK
Only do a simple_sync() if we are stable, auth, and not already sync. The
client request can race with other state changes, so be careful. The
client will also retry on any state change, so we can safely ignore if
things don't look quite right.
This fxes crash:
mds/Locker.cc: In function 'bool Locker::simple_sync(SimpleLock*, bool*)':
mds/Locker.cc:2592: FAILED assert(0)
1: (Locker::handle_file_lock(ScatterLock*, MLock*)+0x2a8) [0x57d328]
2: (MDS::_dispatch(Message*)+0x2165) [0x4a11d5]
3: (MDS::ms_dispatch(Message*)+0x6d) [0x4a169d]
4: (SimpleMessenger::dispatch_entry()+0x733) [0x47c903]
5: (SimpleMessenger::DispatchThread::entry()+0x1c) [0x474bac]
6: (Thread::_entry_func(void*)+0xa) [0x48748a]
7: /lib/libpthread.so.0 [0x7f8af92ff73a]
8: (clone()+0x6d) [0x7f8af852569d]
Sage Weil [Thu, 22 Jul 2010 17:53:14 +0000 (10:53 -0700)]
debian: fix typo in libcrush description
Reported-by: Wido
Sage Weil [Tue, 20 Jul 2010 22:59:11 +0000 (15:59 -0700)]
mds: initialize snaprealm created, current_parent_since on creation
Need to initialize created and current_parent_since on new snaprealms
when they are created, or else we get incorrect results from the likes of
SnapRealm::get_snap_info() (e.g., parent snaps from before we were
created).
Sage Weil [Wed, 21 Jul 2010 20:30:02 +0000 (13:30 -0700)]
Merge remote branch 'origin/debian' into unstable
Yehuda Sadeh [Wed, 21 Jul 2010 18:03:03 +0000 (11:03 -0700)]
rgw: fix calls to finish_get_obj() with uninitialized handle
Sage Weil [Wed, 21 Jul 2010 16:31:57 +0000 (09:31 -0700)]
debian: drop unused *.modules.in
Sage Weil [Wed, 21 Jul 2010 04:40:34 +0000 (21:40 -0700)]
debian: no more ceph-kclient-soruce
Sage Weil [Wed, 21 Jul 2010 04:40:02 +0000 (21:40 -0700)]
debian: cleanup from Clint Byrum <clint@ubuntu.com>
With a few changes by me (changelog, copyright clarification)
Sage Weil [Wed, 21 Jul 2010 04:25:46 +0000 (21:25 -0700)]
fuse.cc: make license GPL
Originally based on Miklos' fusexmp.c, which is GPL. Keep that license.
LGPL makes no sense here anyway.
Sage Weil [Tue, 20 Jul 2010 22:02:59 +0000 (15:02 -0700)]
mon: specify default new pool size/rule/pg_num in conf
Sage Weil [Tue, 20 Jul 2010 21:59:33 +0000 (14:59 -0700)]
clean up new pool crush rule values [backward compatible protocol change]
Default to -1 (not 0!), which means 'server default'.
Sage Weil [Tue, 20 Jul 2010 20:24:26 +0000 (13:24 -0700)]
osd: infer missing map is empty when last_complete==last_update
Sage Weil [Tue, 20 Jul 2010 20:24:08 +0000 (13:24 -0700)]
mon: set/clear pause{rd,wr} osdmap flags
Sage Weil [Tue, 20 Jul 2010 20:23:51 +0000 (13:23 -0700)]
osd: clear failure_queue when marked down
This prevents bleed through of failures (due to not getting hearbeats, due
to us being marked down) so they don't get sent after we are marked back
up again.
This fixes one possible source of up/down flapping...
Sage Weil [Tue, 20 Jul 2010 19:39:42 +0000 (12:39 -0700)]
osd: clean up pulling first/complete business
Sage Weil [Tue, 20 Jul 2010 19:30:53 +0000 (12:30 -0700)]
osd: push in chunks
Signed-off-by: Sage Weil <sage@newdream.net>
Yehuda Sadeh [Tue, 20 Jul 2010 19:12:39 +0000 (12:12 -0700)]
rgw: fix content-length reporting
and also some other various issues that came up while at it
Sage Weil [Tue, 20 Jul 2010 17:38:26 +0000 (10:38 -0700)]
debug: fix log append
ios::ate does not work as advertised!
Signed-off-by: Sage Weil <sage@newdream.net>
Sage Weil [Tue, 20 Jul 2010 17:36:35 +0000 (10:36 -0700)]
debug: say 'append' or 'new' when opening log
Signed-off-by: Sage Weil <sage@newdream.net>
Sage Weil [Tue, 20 Jul 2010 17:25:34 +0000 (10:25 -0700)]
osd: wait for non-existant pools (w/o crashing); clean up handle_op()
Make sure the pool exists before calling OSDMap::raw_pg_to_pg(). Fixes
crash on operations shortly after pg pools are created.
Signed-off-by: Sage Weil <sage@newdream.net>
Yehuda Sadeh [Tue, 20 Jul 2010 00:20:53 +0000 (17:20 -0700)]
rgw: fix copy
Yehuda Sadeh [Mon, 19 Jul 2010 23:50:43 +0000 (16:50 -0700)]
rgw: get/put read and write by chunks
Sage Weil [Mon, 19 Jul 2010 23:22:52 +0000 (16:22 -0700)]
osd: refactor push code
- send_push_op() does a push, nothing else
- push_start() starts a primary->replica push, tracks state
- push_to_replica() ensures we push head first, calculates cloning, etc.
Yehuda Sadeh [Mon, 19 Jul 2010 23:20:34 +0000 (16:20 -0700)]
rgw: lay groundwork for chunky read/write
Sage Weil [Mon, 19 Jul 2010 21:44:09 +0000 (14:44 -0700)]
osd: recover degraded objects _before_ modifying it
This will slow down writes to degraded objects because we will wait for it
to recover before applying the write. OTOH it will be robust in the case
of large objects. We can optimize the small object update (and overwrite)
cases later.
Signed-off-by: Sage Weil <sage@newdream.net>
Sage Weil [Mon, 19 Jul 2010 20:40:50 +0000 (13:40 -0700)]
osd: make client op version actual at_version
I.e. this should take into account any cloning. So set it after we call
prepare_transaction, not before.
Sage Weil [Mon, 19 Jul 2010 20:39:18 +0000 (13:39 -0700)]
osd: fix subop version in snaptrimmer
The at_version should end equal to last_update, not +1.
Sage Weil [Mon, 19 Jul 2010 20:08:10 +0000 (13:08 -0700)]
osd: observe osd_recovery_max_chunk when pulling objects to primary
Sage Weil [Mon, 19 Jul 2010 20:06:48 +0000 (13:06 -0700)]
intervalset: implement span_of()
Sage Weil [Mon, 19 Jul 2010 17:48:48 +0000 (10:48 -0700)]
osd: clean out temp pg on startup
Sage Weil [Mon, 19 Jul 2010 17:44:11 +0000 (10:44 -0700)]
osd: create temp collection during mkfs
Sage Weil [Mon, 19 Jul 2010 17:40:12 +0000 (10:40 -0700)]
osd: give coll_t an explicit type in {meta,temp,pg}
Sage Weil [Mon, 19 Jul 2010 17:14:56 +0000 (10:14 -0700)]
osd: add osd_recovery_max_chunk
Sage Weil [Mon, 19 Jul 2010 17:13:05 +0000 (10:13 -0700)]
mon: check command length on osdmon
Fixes crash on 'ceph osd tell'.
Signed-off-by: Sage Weil <sage@newdream.net>
Yehuda Sadeh [Mon, 19 Jul 2010 17:20:01 +0000 (10:20 -0700)]
rbd: add removal of snapshots
Greg Farnum [Fri, 16 Jul 2010 23:12:38 +0000 (16:12 -0700)]
librados: don't read attributes until the callback is done!
Greg Farnum [Fri, 16 Jul 2010 19:10:16 +0000 (12:10 -0700)]
rgw: rgw_rest: Properly return the content-length.
From wido.
Greg Farnum [Fri, 16 Jul 2010 18:13:31 +0000 (11:13 -0700)]
rgw: rgw_rados: initialize the position in list_buckets_init!
Sage Weil [Fri, 16 Jul 2010 21:03:30 +0000 (14:03 -0700)]
monc: be quiet about 'continuing hunt' too