]> git.apps.os.sepia.ceph.com Git - ceph.git/log
ceph.git
15 years agoosd: set obs.exists in projected object state during recovery
Sage Weil [Wed, 12 May 2010 18:54:09 +0000 (11:54 -0700)]
osd: set obs.exists in projected object state during recovery

15 years agomds: warn, don't crash, on trailing garbage in dir objects
Sage Weil [Wed, 12 May 2010 18:53:50 +0000 (11:53 -0700)]
mds: warn, don't crash, on trailing garbage in dir objects

15 years agomsgr: put msgvec on heap
Sage Weil [Tue, 11 May 2010 23:52:48 +0000 (16:52 -0700)]
msgr: put msgvec on heap

It can get too big for the stack.

15 years agomsgr: tolerate incoming seq #'s that skip ahead
Sage Weil [Wed, 12 May 2010 04:14:39 +0000 (21:14 -0700)]
msgr: tolerate incoming seq #'s that skip ahead

This is necessary because the kclient may pull messages out of the out/sent
queues, and we can't renumber previously sent (and possibly received)
messages without breaking things entirely.

15 years agoceph: return error code returned by server
Sage Weil [Tue, 11 May 2010 21:46:01 +0000 (14:46 -0700)]
ceph: return error code returned by server

15 years agomds: drop 'closed' bit from MClientReconnect
Sage Weil [Tue, 11 May 2010 20:28:53 +0000 (13:28 -0700)]
mds: drop 'closed' bit from MClientReconnect

15 years agomsgr: set outgoing msg connection before encoding
Sage Weil [Tue, 11 May 2010 20:13:24 +0000 (13:13 -0700)]
msgr: set outgoing msg connection before encoding

This allows encode_payload to adjust behavior based on the target peer's
feature bits.

15 years agopoolop: make new encoding backward compatible
Sage Weil [Sat, 8 May 2010 04:27:57 +0000 (21:27 -0700)]
poolop: make new encoding backward compatible

This makes cea221c64 behave when messages using the old encoding (that is,
older versions of the client talk to us).

15 years agoosd: fix compile error from cmpxattr, cleanup.
Sage Weil [Tue, 11 May 2010 16:36:02 +0000 (09:36 -0700)]
osd: fix compile error from cmpxattr, cleanup.

I think the xattr bufferlist still needs to be null terminated...

15 years agoosd: fix layout return type
Sage Weil [Tue, 11 May 2010 16:24:00 +0000 (09:24 -0700)]
osd: fix layout return type

15 years agomsgr: be less noisy about msgr throttling
Sage Weil [Mon, 10 May 2010 23:21:59 +0000 (16:21 -0700)]
msgr: be less noisy about msgr throttling

15 years agoosd: add cmpxattr op handling
Yehuda Sadeh [Mon, 10 May 2010 23:18:52 +0000 (16:18 -0700)]
osd: add cmpxattr op handling

15 years agodebian: remove pull.sh calls from helper .sh scripts
Sage Weil [Mon, 10 May 2010 23:00:22 +0000 (16:00 -0700)]
debian: remove pull.sh calls from helper .sh scripts

15 years agoatomic: cast away const on read()
Sage Weil [Mon, 10 May 2010 22:53:52 +0000 (15:53 -0700)]
atomic: cast away const on read()

(Only needed to build on lenny, this can go away someday)

15 years agoosd: 'stop' command
Sage Weil [Fri, 7 May 2010 23:40:49 +0000 (16:40 -0700)]
osd: 'stop' command

15 years agomds: fix max_size==1 bug
Sage Weil [Fri, 7 May 2010 22:57:37 +0000 (15:57 -0700)]
mds: fix max_size==1 bug

We should provide a minimum of one stripe unit.

15 years agoMerge branch 'string' into unstable
Sage Weil [Fri, 7 May 2010 22:00:22 +0000 (15:00 -0700)]
Merge branch 'string' into unstable

Conflicts:
src/client/Client.h
src/messages/MAuthReply.h
src/messages/MClientLease.h
src/osd/ReplicatedPG.cc
src/osd/osd_types.h
src/osdc/Journaler.h

15 years agotypes: clean up printf format warnings
Sage Weil [Fri, 7 May 2010 21:55:31 +0000 (14:55 -0700)]
types: clean up printf format warnings

bah

15 years agorgw: clean up dup #define
Sage Weil [Fri, 7 May 2010 21:30:51 +0000 (14:30 -0700)]
rgw: clean up dup #define

15 years agotypes: standardize on uint64_t
Sage Weil [Fri, 7 May 2010 21:33:42 +0000 (14:33 -0700)]
types: standardize on uint64_t

The problem is that on some platforms __u64 == uint64_t (x86_64), and on
others it's doesn't (ppc64).  Which means we don't know whether to define
different versions of overloaded functions for both types or just one.

So, standardize on uint64_t.  This plays nicer with STL, which defines
hash<uint64_t> on 64 bit arches but not 32 bit.  Which means we can't
standarzie on __u64 or else hash<__u64> won't work.  Bah!

15 years agoendian: simplify __le* type hackery
Sage Weil [Fri, 7 May 2010 20:45:00 +0000 (13:45 -0700)]
endian: simplify __le* type hackery

Instead of preventing linux/types.h from being included, instead name
our types ceph_le*, and remap using #define _only_ when including the
shared kernel/userspace headers.

Also redefine some templates for both __u64 and uint64_t, which are
apparently not the same on x86_64.

15 years agoceph.spec.in: --without-hadoop --without-debug
Sage Weil [Fri, 7 May 2010 19:30:07 +0000 (12:30 -0700)]
ceph.spec.in: --without-hadoop --without-debug

to sync up with f13

15 years agorbdtool: --list on empty pool returns correct message
Yehuda Sadeh [Fri, 7 May 2010 20:48:28 +0000 (13:48 -0700)]
rbdtool: --list on empty pool returns correct message

and also command doesn't hang. This fixes bug #115.

15 years agotime: include _ between date and time
Sage Weil [Fri, 7 May 2010 18:54:07 +0000 (11:54 -0700)]
time: include _ between date and time

15 years agoosd: remove bad assert in get_object_context()
Sage Weil [Fri, 7 May 2010 18:53:03 +0000 (11:53 -0700)]
osd: remove bad assert in get_object_context()

This has always been fishy, and it's original purpose isn't quite clear.
The '!obc->obs.ssc' is clearly the create case, but the head_exists check
is only the case if we are getting the context for a _head, which in
many cases we're not.  In parituclarly, it fails for a snapped object
where the head doesn't exist.  We _do_ have teh ssc (from the _snapdir
object) but there is no head.

Just remove the assert entirely.. it's not clear what it's really catching.

15 years agomds: always release snap cap after flushsnap
Sage Weil [Fri, 7 May 2010 18:58:16 +0000 (11:58 -0700)]
mds: always release snap cap after flushsnap

The old condition makes no sense.  It appears to have been broken since
2008, commit a5d5e679?  Yikes.

The problem is that a flushsnap will 'release' the caps, but still leave
an open Capability on the CInode.  Then, CInode::encode_inodestat() would
& issued() when issuing caps, and a client ceph_open() would get no
caps.  The kclient is fixed up to at least not crash in that case, but even
with the fix it deadlocked.

15 years agoosd: don't put() null ctx->op (message)
Sage Weil [Fri, 7 May 2010 16:34:48 +0000 (09:34 -0700)]
osd: don't put() null ctx->op (message)

15 years agoscripts: fix check_pglog.sh
Sage Weil [Fri, 7 May 2010 16:05:14 +0000 (09:05 -0700)]
scripts: fix check_pglog.sh

15 years agoscript: add check_pglog.sh
Sage Weil [Fri, 7 May 2010 15:52:18 +0000 (08:52 -0700)]
script: add check_pglog.sh

Check pglog for zero segments, normally caused by a race between
pg log replacement (rm log, rewrite log) vs append race.

For example, see 686192a03eb7d28c7340dcab6d9671505f9cd986

15 years agomdsmap: lose angle braces, use tabs
Sage Weil [Thu, 6 May 2010 22:30:14 +0000 (15:30 -0700)]
mdsmap: lose angle braces, use tabs

15 years agoosdmap: initially osds don't exist; create them as they boot
Sage Weil [Thu, 6 May 2010 22:27:07 +0000 (15:27 -0700)]
osdmap: initially osds don't exist; create them as they boot

15 years agoosd: resurrect CEPH_OSD_EXISTS flag
Sage Weil [Thu, 6 May 2010 21:04:35 +0000 (14:04 -0700)]
osd: resurrect CEPH_OSD_EXISTS flag

Mainly so we can track how many osds are in the cluster, even if they
don't start at 0.

15 years agologger: be more careful calculating variance
Sage Weil [Thu, 6 May 2010 18:10:14 +0000 (11:10 -0700)]
logger: be more careful calculating variance

15 years agodebian: use 'testing' or 'unstable' in place of 'git' in version suffix
Sage Weil [Wed, 5 May 2010 22:26:45 +0000 (15:26 -0700)]
debian: use 'testing' or 'unstable' in place of 'git' in version suffix

15 years agolibrados: separate callbacks for aio ack/complete and safe
Sage Weil [Wed, 5 May 2010 21:40:59 +0000 (14:40 -0700)]
librados: separate callbacks for aio ack/complete and safe

15 years agoosd: clear pg REPLAY when we reinitiate peering
Sage Weil [Thu, 6 May 2010 18:30:51 +0000 (11:30 -0700)]
osd: clear pg REPLAY when we reinitiate peering

Among other things, a stray/bad REPLAY flag can trigger a false activation
like so:

osd/PG.cc: In function 'void PG::activate(ObjectStore::Transaction&, std::list<Context*, std::allocator<Context*> >&, std::map<int, MOSDPGInfo*, std::less<int>, std::allocator<std::pair<const int, MOSDPGInfo*> > >*)':
osd/PG.cc:1641: FAILED assert(peer_info.count(peer))
 1: (PG::activate(ObjectStore::Transaction&, std::list<Context*, std::allocator<Context*> >&, std::map<int, MOSDPGInfo*, std::less<int>, std::allocator<std::pair<int const, MOSDPGInfo*> > >*)+0x817) [0x75e637]
 2: (OSD::activate_pg(pg_t, utime_t)+0x136) [0x6bfe86]
 3: (OSD::check_replay_queue()+0x134) [0x6c0074]
 4: (OSD::tick()+0x1db) [0x6d96eb]
 5: (OSD::C_Tick::finish(int)+0x1c) [0x74844e]
 6: (SafeTimer::EventWrapper::finish(int)+0x6d) [0x7fa2cf]
 7: (Timer::timer_entry()+0x454) [0x7fb52a]
 8: (Timer::TimerThread::entry()+0x19) [0x647c09]
 9: (Thread::_entry_func(void*)+0x20) [0x659f18]
 10: /lib/libpthread.so.0 [0x7fceb73d473a]
 11: (clone()+0x6d) [0x7fceb65fe69d]

15 years agorgw: admin uses --secret instead of --key (as --key is already taken)
Greg Farnum [Wed, 5 May 2010 21:04:15 +0000 (14:04 -0700)]
rgw: admin uses --secret instead of --key (as --key is already taken)

15 years agoosd: fix critical cap parsing bugs and documentation.
Greg Farnum [Wed, 5 May 2010 20:55:49 +0000 (13:55 -0700)]
osd: fix critical cap parsing bugs and documentation.

15 years agomsgr: fix local_pipe peer_type
Sage Weil [Tue, 4 May 2010 20:57:39 +0000 (13:57 -0700)]
msgr: fix local_pipe peer_type

Need to re-init_local_pipe() after register_entity().

15 years agoMerge branch 'mds_snapids' into unstable
Sage Weil [Tue, 4 May 2010 19:57:32 +0000 (12:57 -0700)]
Merge branch 'mds_snapids' into unstable

15 years agomsgr: existing connection may be OPEN on connection race
Sage Weil [Tue, 4 May 2010 19:51:01 +0000 (12:51 -0700)]
msgr: existing connection may be OPEN on connection race

Yup!

15 years agoosd: add CMPXATTR defines
Sage Weil [Tue, 4 May 2010 18:56:15 +0000 (11:56 -0700)]
osd: add CMPXATTR defines

15 years agoosdmaptool: add -h; don't assert on bad filename
Sage Weil [Tue, 4 May 2010 16:51:36 +0000 (09:51 -0700)]
osdmaptool: add -h; don't assert on bad filename

15 years agoceph.spec.in: run configure
Sage Weil [Tue, 4 May 2010 04:06:37 +0000 (21:06 -0700)]
ceph.spec.in: run configure

Reported by MarkN in #ceph on irc.oftc.net

15 years agoMerge remote branch 'origin/msgr_throttle' into unstable
Sage Weil [Tue, 4 May 2010 04:02:47 +0000 (21:02 -0700)]
Merge remote branch 'origin/msgr_throttle' into unstable

15 years agopoolop: do op immediately, instead of waiting for timeout
Yehuda Sadeh [Mon, 3 May 2010 23:57:10 +0000 (16:57 -0700)]
poolop: do op immediately, instead of waiting for timeout

15 years agorbd: update rbd image header format
Yehuda Sadeh [Mon, 3 May 2010 23:56:10 +0000 (16:56 -0700)]
rbd: update rbd image header format

15 years agoMPoolOp: rearrange parsing, for easier kernel implementation
Yehuda Sadeh [Thu, 22 Apr 2010 22:47:59 +0000 (15:47 -0700)]
MPoolOp: rearrange parsing, for easier kernel implementation

15 years agoosd: add removed snapid to removed_snaps for unmanaged snaps
Sage Weil [Mon, 3 May 2010 23:41:30 +0000 (16:41 -0700)]
osd: add removed snapid to removed_snaps for unmanaged snaps

15 years agomds: add removal snap_seq to removed_snaps
Sage Weil [Mon, 3 May 2010 23:41:16 +0000 (16:41 -0700)]
mds: add removal snap_seq to removed_snaps

This prevents lots of fragmentation in removed_snaps interval set.

15 years agomds: acquire anchor_create locks in caller
Sage Weil [Mon, 3 May 2010 22:04:16 +0000 (15:04 -0700)]
mds: acquire anchor_create locks in caller

We're not allowed to call acquire_locks again once we've completed locking.
Do it in the caller.  Add a helper that will also take an rdlock to ensure
the anchor doesn't go away due to a racing operation.

15 years agoobjecter: fix ~Message() calls (use ->put())
Sage Weil [Mon, 3 May 2010 17:55:29 +0000 (10:55 -0700)]
objecter: fix ~Message() calls (use ->put())

15 years agomdsmap: cleanup up braces
Sage Weil [Mon, 3 May 2010 17:21:49 +0000 (10:21 -0700)]
mdsmap: cleanup up braces

15 years agomds: fix EXCL rdlock rule
Sage Weil [Mon, 3 May 2010 21:01:42 +0000 (14:01 -0700)]
mds: fix EXCL rdlock rule

Only the exclusive client should be able to rdlock.  (if anyone!)

No observed problems, caught by inspection.

15 years agomds: fix lock->sync rdlock rule
Sage Weil [Mon, 3 May 2010 21:00:28 +0000 (14:00 -0700)]
mds: fix lock->sync rdlock rule

Observed symptom was touch foo on one node, and missing from ls on another
node.

Since auth rdlock isn't allowed for LOCk, it clearly shouldn't be for
LOCK->SYNC while e.g. wrlock modifications are still being flushed to the
journal and aren't yet applied.

15 years agofilejournal: create if mkfs and journal is small OR dne.
Sage Weil [Mon, 3 May 2010 17:17:53 +0000 (10:17 -0700)]
filejournal: create if mkfs and journal is small OR dne.

15 years agofilejournal: return error instead of asserting
Sage Weil [Mon, 3 May 2010 03:50:58 +0000 (20:50 -0700)]
filejournal: return error instead of asserting

15 years agomon: correctly update up map when taking over for creating|starting mds
Sage Weil [Fri, 30 Apr 2010 23:24:34 +0000 (16:24 -0700)]
mon: correctly update up map when taking over for creating|starting mds

Otherwise we can get crap like

in <0>
up <{0=4103}>
failed <>
stopped <>
4300: 192.168.141.103:6800/9446 '0' mds0.0 up:creating seq 2

15 years agoMerge commit 'v0.20' into unstable
Sage Weil [Fri, 30 Apr 2010 22:44:54 +0000 (15:44 -0700)]
Merge commit 'v0.20' into unstable

Conflicts:
src/mon/LogMonitor.cc
src/mon/MDSMonitor.cc
src/mon/Monitor.cc
src/mon/OSDMonitor.cc
src/mon/PGMonitor.cc

15 years agodebian: fix up debian scripts v0.20
Sage Weil [Fri, 30 Apr 2010 22:11:19 +0000 (15:11 -0700)]
debian: fix up debian scripts

15 years agoceph.spec.in: updates
Sage Weil [Fri, 30 Apr 2010 20:51:44 +0000 (13:51 -0700)]
ceph.spec.in: updates

15 years agoceph.spec.in: from Josef
Sage Weil [Fri, 30 Apr 2010 20:39:49 +0000 (13:39 -0700)]
ceph.spec.in: from Josef

15 years agomds: fix trim_dentry on dentry under unlinked dir
Sage Weil [Fri, 30 Apr 2010 18:05:39 +0000 (11:05 -0700)]
mds: fix trim_dentry on dentry under unlinked dir

We can get a dentry that is trimmable (e.g. null) under a new unlinked dir,
which has no subtree.  This will only happen on the auth.  In that case,
having no container is harmless--it's only needed for replicas.

This fixes the following crash:

mds/MDCache.cc: In function 'void MDCache::trim_dentry(CDentry*, std::map<int, MCacheExpire*, std::less<int>, std::allocator<std::pair<const int, MCacheExpire*> > >&)':
mds/MDCache.cc:4797: FAILED assert(con)
 1: (MDCache::trim(int)+0x214) [0x4ffbc4]
 2: (MDS::tick()+0x4c1) [0x48f3b1]
 3: (SafeTimer::EventWrapper::finish(int)+0x269) [0x683a89]
 4: (Timer::timer_entry()+0x819) [0x685909]
 5: (Timer::TimerThread::entry()+0xd) [0x47528d]
 6: (Thread::_entry_func(void*)+0x7) [0x48a8a7]
 7: /lib/libpthread.so.0 [0x7ffe62356fc7]
 8: (clone()+0x6d) [0x7ffe615835ad]
 NOTE: a copy of the executable, or `objdump -rdS <executable>` is needed to interpret this.

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoosd: fix pg_to_acting_osds() calculation to consider pg_temp
Sage Weil [Fri, 30 Apr 2010 17:46:17 +0000 (10:46 -0700)]
osd: fix pg_to_acting_osds() calculation to consider pg_temp

The OSD was using pg_to_up_acting, the client uses pg_to_acting, and their
calculations of 'acting' weren't consistent because the latter did not
consider pg_temp.

Use new helpers to clean this up.

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoobjecter: show osds in laggy debug output
Sage Weil [Fri, 30 Apr 2010 17:11:16 +0000 (10:11 -0700)]
objecter: show osds in laggy debug output

15 years agomds: fix trim_dentry on dentry under unlinked dir
Sage Weil [Fri, 30 Apr 2010 18:05:39 +0000 (11:05 -0700)]
mds: fix trim_dentry on dentry under unlinked dir

We can get a dentry that is trimmable (e.g. null) under a new unlinked dir,
which has no subtree.  This will only happen on the auth.  In that case,
having no container is harmless--it's only needed for replicas.

This fixes the following crash:

mds/MDCache.cc: In function 'void MDCache::trim_dentry(CDentry*, std::map<int, MCacheExpire*, std::less<int>, std::allocator<std::pair<const int, MCacheExpire*> > >&)':
mds/MDCache.cc:4797: FAILED assert(con)
 1: (MDCache::trim(int)+0x214) [0x4ffbc4]
 2: (MDS::tick()+0x4c1) [0x48f3b1]
 3: (SafeTimer::EventWrapper::finish(int)+0x269) [0x683a89]
 4: (Timer::timer_entry()+0x819) [0x685909]
 5: (Timer::TimerThread::entry()+0xd) [0x47528d]
 6: (Thread::_entry_func(void*)+0x7) [0x48a8a7]
 7: /lib/libpthread.so.0 [0x7ffe62356fc7]
 8: (clone()+0x6d) [0x7ffe615835ad]
 NOTE: a copy of the executable, or `objdump -rdS <executable>` is needed to interpret this.

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoosd: fix pg_to_acting_osds() calculation to consider pg_temp
Sage Weil [Fri, 30 Apr 2010 17:46:17 +0000 (10:46 -0700)]
osd: fix pg_to_acting_osds() calculation to consider pg_temp

The OSD was using pg_to_up_acting, the client uses pg_to_acting, and their
calculations of 'acting' weren't consistent because the latter did not
consider pg_temp.

Use new helpers to clean this up.

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoobjecter: show osds in laggy debug output
Sage Weil [Fri, 30 Apr 2010 17:11:16 +0000 (10:11 -0700)]
objecter: show osds in laggy debug output

15 years agofilestore: fail to mount if journal open fails
Sage Weil [Fri, 30 Apr 2010 04:43:07 +0000 (21:43 -0700)]
filestore: fail to mount if journal open fails

Clean up error reporting too.

15 years agofilestore: fail to mount if journal open fails
Sage Weil [Fri, 30 Apr 2010 04:43:07 +0000 (21:43 -0700)]
filestore: fail to mount if journal open fails

Clean up error reporting too.

15 years agomsgr: fix sigint/term signal handler to actually quit (by chaining)
Sage Weil [Fri, 30 Apr 2010 04:12:16 +0000 (21:12 -0700)]
msgr: fix sigint/term signal handler to actually quit (by chaining)

15 years agomsgr: fix sigint/term signal handler to actually quit (by chaining)
Sage Weil [Fri, 30 Apr 2010 04:12:16 +0000 (21:12 -0700)]
msgr: fix sigint/term signal handler to actually quit (by chaining)

15 years agolibrados: fix blank pool stats
Sage Weil [Thu, 29 Apr 2010 22:05:39 +0000 (15:05 -0700)]
librados: fix blank pool stats

15 years agolibrados: fix blank pool stats
Sage Weil [Thu, 29 Apr 2010 22:05:39 +0000 (15:05 -0700)]
librados: fix blank pool stats

15 years agoqa: make pjd clean up
Sage Weil [Thu, 29 Apr 2010 21:26:46 +0000 (14:26 -0700)]
qa: make pjd clean up

15 years agomds: clean up MIN/MAX, client_ranges
Sage Weil [Wed, 28 Apr 2010 20:42:50 +0000 (13:42 -0700)]
mds: clean up MIN/MAX, client_ranges

Avoid double evaluation via macros...

15 years agomsgr: remove pid file on SIGINT, SIGTERM
Sage Weil [Thu, 29 Apr 2010 19:50:34 +0000 (12:50 -0700)]
msgr: remove pid file on SIGINT, SIGTERM

15 years agoqa: make pjd clean up
Sage Weil [Thu, 29 Apr 2010 21:26:46 +0000 (14:26 -0700)]
qa: make pjd clean up

15 years agomds: clean up MIN/MAX, client_ranges
Sage Weil [Wed, 28 Apr 2010 20:42:50 +0000 (13:42 -0700)]
mds: clean up MIN/MAX, client_ranges

Avoid double evaluation via macros...

15 years agomsgr: remove pid file on SIGINT, SIGTERM
Sage Weil [Thu, 29 Apr 2010 19:50:34 +0000 (12:50 -0700)]
msgr: remove pid file on SIGINT, SIGTERM

15 years agoconfig: post-process default config options too
Sage Weil [Thu, 29 Apr 2010 18:56:49 +0000 (11:56 -0700)]
config: post-process default config options too

Previously we were only doing the ones we read out of the .conf.

15 years agoconfig: post-process default config options too
Sage Weil [Thu, 29 Apr 2010 18:56:49 +0000 (11:56 -0700)]
config: post-process default config options too

Previously we were only doing the ones we read out of the .conf.

15 years agomon: use (robust) buffer::list::write_fd() to write to disk
Sage Weil [Thu, 29 Apr 2010 17:43:24 +0000 (10:43 -0700)]
mon: use (robust) buffer::list::write_fd() to write to disk

15 years agomon: use (robust) buffer::list::write_fd() to write to disk
Sage Weil [Thu, 29 Apr 2010 17:43:24 +0000 (10:43 -0700)]
mon: use (robust) buffer::list::write_fd() to write to disk

15 years agomon: m->get_session() may return null if session has closed
Sage Weil [Thu, 29 Apr 2010 17:27:02 +0000 (10:27 -0700)]
mon: m->get_session() may return null if session has closed

because the session close clears connection->priv.  We need to check at
each site anyway, either for null session, or for session->closed.  So
check for null session.

Conflicts:

src/mon/Monitor.cc
src/mon/OSDMonitor.cc
src/mon/PGMonitor.cc

15 years agomon: m->get_session() may return null if session has closed
Sage Weil [Thu, 29 Apr 2010 17:22:13 +0000 (10:22 -0700)]
mon: m->get_session() may return null if session has closed

because the session close clears connection->priv.  We need to check at
each site anyway, either for null session, or for session->closed.  So
check for null session.

15 years agoconfigure: remove second AM_INIT_AUTOMAKE line
Sage Weil [Wed, 28 Apr 2010 22:55:22 +0000 (15:55 -0700)]
configure: remove second AM_INIT_AUTOMAKE line

15 years agomsgr: fix segfault sending msg on closed Connection/pipe
Sage Weil [Wed, 28 Apr 2010 21:07:20 +0000 (14:07 -0700)]
msgr: fix segfault sending msg on closed Connection/pipe

15 years agostring fix
Sage Weil [Wed, 28 Apr 2010 21:18:05 +0000 (14:18 -0700)]
string fix

15 years agomsgr: fix segfault sending msg on closed Connection/pipe
Sage Weil [Wed, 28 Apr 2010 21:07:20 +0000 (14:07 -0700)]
msgr: fix segfault sending msg on closed Connection/pipe

15 years agomsgr: add message throttling; with a default max of 100MB waiting for dispatch
Greg Farnum [Wed, 28 Apr 2010 20:56:43 +0000 (13:56 -0700)]
msgr: add message throttling; with a default max of 100MB waiting for dispatch

15 years agoreplace nstring with std::string
Sage Weil [Wed, 28 Apr 2010 20:45:55 +0000 (13:45 -0700)]
replace nstring with std::string

15 years agothrottle: add fairness
Sage Weil [Wed, 28 Apr 2010 17:36:30 +0000 (10:36 -0700)]
throttle: add fairness

This assumes that SignalOne() will wake people up in the right order.  Not
sure if that's true...

15 years agomds: fix snaprealm split on capless inode
Sage Weil [Mon, 26 Apr 2010 17:14:58 +0000 (10:14 -0700)]
mds: fix snaprealm split on capless inode

15 years agomds: fix snaprealm split on capless inode
Sage Weil [Mon, 26 Apr 2010 17:14:58 +0000 (10:14 -0700)]
mds: fix snaprealm split on capless inode

15 years agobuffer: clean up namespace pollution
Sage Weil [Fri, 23 Apr 2010 23:21:05 +0000 (16:21 -0700)]
buffer: clean up namespace pollution

15 years agopoolop: rename and move definitions to ceph_fs.h
Yehuda Sadeh [Sat, 24 Apr 2010 17:10:00 +0000 (10:10 -0700)]
poolop: rename and move definitions to ceph_fs.h

This was done so that we can use it in the kclient

15 years agoMakefile: define __CEPH__ in ceph builds. use it.
Sage Weil [Fri, 23 Apr 2010 22:50:29 +0000 (15:50 -0700)]
Makefile: define __CEPH__ in ceph builds. use it.

Avoid breakage when headers are used outside of ceph build (librados-dev)

15 years agomds: fix req use-after-free to get connection to direct reply
Sage Weil [Fri, 23 Apr 2010 22:22:29 +0000 (15:22 -0700)]
mds: fix req use-after-free to get connection to direct reply

15 years agoatomic: reverse if test so we don't break things on separately-compiled librados
Greg Farnum [Fri, 23 Apr 2010 22:13:35 +0000 (15:13 -0700)]
atomic: reverse if test so we don't break things on separately-compiled librados