]> git.apps.os.sepia.ceph.com Git - ceph.git/log
ceph.git
15 years agodebian: fix update_pbuilder.sh
Sage Weil [Wed, 11 Aug 2010 17:46:31 +0000 (10:46 -0700)]
debian: fix update_pbuilder.sh

15 years agomon: rename mon_lease_wiggle_room -> mon_allowed_clock_drift
Greg Farnum [Wed, 11 Aug 2010 17:37:14 +0000 (10:37 -0700)]
mon: rename mon_lease_wiggle_room -> mon_allowed_clock_drift

15 years agomon: backoff clock drift warnings
Greg Farnum [Wed, 11 Aug 2010 16:49:29 +0000 (09:49 -0700)]
mon: backoff clock drift warnings

15 years agoosd: write (empty) log, bounds on remove_pg start
Sage Weil [Wed, 11 Aug 2010 17:05:18 +0000 (10:05 -0700)]
osd: write (empty) log, bounds on remove_pg start

This zeros the log, and the bounds, when we start pg removal.  Previously
we just removed the log and didn't write the (empty) bounds, breaking
cosd startup later when the old bounds are totally wrong.

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agorgw: fix content length encoding
Yehuda Sadeh [Tue, 10 Aug 2010 20:37:58 +0000 (13:37 -0700)]
rgw: fix content length encoding

15 years agofilestore: flush using sync(2) hammer
Sage Weil [Sun, 8 Aug 2010 15:59:59 +0000 (08:59 -0700)]
filestore: flush using sync(2) hammer

Since we can't easily detect ext3 (let alone whether we have data=journal),
by default use sync(2) as an overly large hammer to flush all prior applied
ops to disk.  If the option is explicitly enabled, use fsync(2) on a
file to implicitly flush the journal and prior writes.  Admins should only
enable this if they have ext2 in data=journal mode.

15 years agomds: don't try to recover filelocks if the client didn't encode any
Greg Farnum [Fri, 6 Aug 2010 21:58:23 +0000 (14:58 -0700)]
mds: don't try to recover filelocks if the client didn't encode any

15 years agoosd: simplify class waiter list
Sage Weil [Fri, 6 Aug 2010 18:42:33 +0000 (11:42 -0700)]
osd: simplify class waiter list

No reason to break it down by pg_t?

15 years agorgw: make gateway dns name configurable
Yehuda Sadeh [Fri, 6 Aug 2010 21:13:21 +0000 (14:13 -0700)]
rgw: make gateway dns name configurable

15 years agoclient: fill in num_fwd in messages to mds
Greg Farnum [Fri, 6 Aug 2010 20:41:15 +0000 (13:41 -0700)]
client: fill in num_fwd in messages to mds

15 years agoosd: class load shouldn't hang when failed to load
Yehuda Sadeh [Fri, 6 Aug 2010 19:46:07 +0000 (12:46 -0700)]
osd: class load shouldn't hang when failed to load

Fixes bug #332. This happened if either the class existed, and it was either
corrupted, or failed to load from any other reason.

15 years agoqa: echo output
Sage Weil [Fri, 6 Aug 2010 18:27:15 +0000 (11:27 -0700)]
qa: echo output

15 years agoMerge branch 'snap_cap' into unstable
Sage Weil [Fri, 6 Aug 2010 18:34:57 +0000 (11:34 -0700)]
Merge branch 'snap_cap' into unstable

15 years agoMerge branch 'testing' into unstable
Sage Weil [Fri, 6 Aug 2010 18:34:53 +0000 (11:34 -0700)]
Merge branch 'testing' into unstable

15 years agomds: fix send_message_client_counted helper
Sage Weil [Fri, 6 Aug 2010 17:07:46 +0000 (10:07 -0700)]
mds: fix send_message_client_counted helper

Don't use the outgoing message connection (usually there is none!).. use
the connection argument.  Der.  Broken since 6d770abe.

15 years agomon: add 'mds compat rm_* id' commands to adjust mdsmap compat set
Sage Weil [Fri, 6 Aug 2010 16:56:34 +0000 (09:56 -0700)]
mon: add 'mds compat rm_* id' commands to adjust mdsmap compat set

This'll be helpful if someone inadvertantly starts up a new cmds, the
compat set updates, and then they can't start old cmds instances.

Of course, that will only work if the new cmds didn't write any new format
metadata!

15 years agomon: automatically incorporate new incompat features into mdsmap
Sage Weil [Thu, 5 Aug 2010 23:49:07 +0000 (16:49 -0700)]
mon: automatically incorporate new incompat features into mdsmap

15 years agomds: include compatset in beacon message
Sage Weil [Thu, 5 Aug 2010 23:48:26 +0000 (16:48 -0700)]
mds: include compatset in beacon message

15 years agomds: add incompat clientranges bit for the format change
Sage Weil [Thu, 5 Aug 2010 23:48:16 +0000 (16:48 -0700)]
mds: add incompat clientranges bit for the format change

15 years agoqa: add snap test for cap writeback
Sage Weil [Thu, 5 Aug 2010 23:16:02 +0000 (16:16 -0700)]
qa: add snap test for cap writeback

This used to fail/hang before the cap cloning business was fixed up.

15 years agomds: non-head requests always rdlock as_anon
Sage Weil [Thu, 5 Aug 2010 23:05:46 +0000 (16:05 -0700)]
mds: non-head requests always rdlock as_anon

This ensures we can't do a snapped GETATTR and rdlock with EXCL on a
directory and miss out on cap metadata still with the client.

15 years agomds: write flushsnap cap metadata into old_inode if appropriate
Sage Weil [Thu, 5 Aug 2010 21:52:40 +0000 (14:52 -0700)]
mds: write flushsnap cap metadata into old_inode if appropriate

Don't have a test case for this yet, unfortunately, but it looks
straightforward at least. :)

15 years agomds: ensure that rdlock on snap filelock flushes data
Sage Weil [Thu, 5 Aug 2010 21:09:53 +0000 (14:09 -0700)]
mds: ensure that rdlock on snap filelock flushes data

Do the head filelock sync 'anonymously', so that we can_rdlock(-1) (no
particular client) and thus can't rdlock in the EXCL state.

15 years agomds: rdlock head when snapped inode is snap->sync
Sage Weil [Thu, 5 Aug 2010 21:01:59 +0000 (14:01 -0700)]
mds: rdlock head when snapped inode is snap->sync

15 years agomds: do not clone caps to snapped inodes
Sage Weil [Thu, 5 Aug 2010 20:08:22 +0000 (13:08 -0700)]
mds: do not clone caps to snapped inodes

Instead, explicitly track which locks need to be flushed (via a FLUSHSNAP)
with a LOCK_SNAP_SYNC lock state.

Restructures the handle_client_caps.

Also changes the client ranges format in the inode to keep a follows for
each client (basically 'flushed through') so that the client ranges can
get cleaned out later when it gets cowed.

15 years agomds: group finishers after eval_cap_gather does all locks
Sage Weil [Thu, 5 Aug 2010 19:23:43 +0000 (12:23 -0700)]
mds: group finishers after eval_cap_gather does all locks

This will reduce some churn, since we can update all the lock states before
worrying about anyone trying to take more than one (and failing, again).

15 years agoconfig: set osd_min_down_reports to 3.
Greg Farnum [Thu, 5 Aug 2010 18:36:15 +0000 (11:36 -0700)]
config: set osd_min_down_reports to 3.

15 years agoosd: only access failure_queue while under heartbeat_lock.
Greg Farnum [Thu, 5 Aug 2010 17:22:39 +0000 (10:22 -0700)]
osd: only access failure_queue while under heartbeat_lock.

15 years agoosd: Adjust failure reporting.
Greg Farnum [Thu, 5 Aug 2010 16:48:41 +0000 (09:48 -0700)]
osd: Adjust failure reporting.

When a failure report is sent to the mon, the failed OSD is added to
pending_failures. If the OSD gets a heartbeat from an OSD in pending_failures,
it sends an MOSDFailure message repealing the previous failure report.
If an OSD is marked as failed but a message hasn't been sent, it's simply
removed from the failed list.

15 years agoosdmon: Adjust failure reporting.
Greg Farnum [Wed, 4 Aug 2010 23:02:17 +0000 (16:02 -0700)]
osdmon: Adjust failure reporting.

MOSDFailure can now be a failure or a not-failure report. If it is a
failure, OSDMon will add it to a map of failure reports until a given OSD
exceeds the minimum reports/reporters to qualify as failed, then marks it
down.
Also, only propose new actually marking an osd failed
OSD behavior is not changed, and currently the min reports/reporters are
set to 1 so external OSDMon behavior will also stay the same.

15 years agomakefile: define docdir for old automake versions
Sage Weil [Thu, 5 Aug 2010 14:47:32 +0000 (07:47 -0700)]
makefile: define docdir for old automake versions

Notably centos5.

Reported-by: Thomas Mueller <thomas@chaschperli.ch>
Signed-off-by: Sage Weil <sage@newdream.net>
15 years agomonmaptool: man page typo fix
Haifeng Liu [Thu, 5 Aug 2010 08:16:23 +0000 (17:16 +0900)]
monmaptool: man page typo fix

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoosdmap: only decode internal addresses if map is new enough version
Greg Farnum [Wed, 4 Aug 2010 18:40:36 +0000 (11:40 -0700)]
osdmap: only decode internal addresses if map is new enough version

15 years agoMerge branch 'testing' into unstable
Sage Weil [Wed, 4 Aug 2010 17:37:09 +0000 (10:37 -0700)]
Merge branch 'testing' into unstable

15 years agorados: -i and -o are still there for other commands
Sage Weil [Wed, 4 Aug 2010 17:01:23 +0000 (10:01 -0700)]
rados: -i and -o are still there for other commands

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoceph.spec.in: fix docdir paths
Sage Weil [Wed, 4 Aug 2010 16:56:05 +0000 (09:56 -0700)]
ceph.spec.in: fix docdir paths

_docdir is /usr/share/doc, not /usr/share/doc/ceph (as it is with debian).

15 years agoceph.spec.in: fix sample.ceph.conf chown
Sage Weil [Wed, 4 Aug 2010 16:32:48 +0000 (09:32 -0700)]
ceph.spec.in: fix sample.ceph.conf chown

Reported-by: Xiaoguang Liu <syslxg@gmail.com>
15 years agorados: fix usage
Takuya ASADA [Wed, 4 Aug 2010 16:30:03 +0000 (09:30 -0700)]
rados: fix usage

15 years agoosd: fix heartbeat to/from map updates
Sage Weil [Wed, 4 Aug 2010 16:51:12 +0000 (09:51 -0700)]
osd: fix heartbeat to/from map updates

- always make _inst match _to and _from (don't adjust in note_down_osd)
- only mark down old hb connection in update_heartbeat_peers
- if old peer isn't down, send them a map
- don't print dup new/old messages

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoosd: cover map_in_progress by osd_lock instead of an individual lock.
Greg Farnum [Tue, 3 Aug 2010 23:09:53 +0000 (16:09 -0700)]
osd: cover map_in_progress by osd_lock instead of an individual lock.
The separate lock is no longer needed and removing it reduces lock contention
and makes it easier to avoid races elsewhere.

15 years agoosd: Unlock osd_lock in _dispatch before locking handle_map_lock
Greg Farnum [Tue, 3 Aug 2010 22:21:42 +0000 (15:21 -0700)]
osd: Unlock osd_lock in _dispatch before locking handle_map_lock

15 years agoosd: switching dispatch_running to be a counter
Greg Farnum [Tue, 3 Aug 2010 21:50:55 +0000 (14:50 -0700)]
osd: switching dispatch_running to be a counter
Now it works with multiple messengers dispatching.

15 years agoosd: guard _dispatch to prevent execution of new messages while the map is updating
Greg Farnum [Tue, 3 Aug 2010 21:50:24 +0000 (14:50 -0700)]
osd: guard _dispatch to prevent execution of new messages while the map is updating

15 years agoosd: adjust map_in_progress guarding
Greg Farnum [Tue, 3 Aug 2010 21:39:23 +0000 (14:39 -0700)]
osd: adjust map_in_progress guarding

15 years agocosd: adjust client/cluster messenger setup to be more consistent
Gregory Farnum [Fri, 30 Jul 2010 16:23:46 +0000 (09:23 -0700)]
cosd: adjust client/cluster messenger setup to be more consistent
and poison better

15 years agoconfig: Error out when you can't parse an addr from the config file.
Greg Farnum [Fri, 30 Jul 2010 07:28:52 +0000 (00:28 -0700)]
config: Error out when you can't parse an addr from the config file.

Also, fix up a few other addr parsing cases to break properly!

15 years agoosd: use the client_messenger for the logclient.
Greg Farnum [Thu, 29 Jul 2010 22:28:35 +0000 (15:28 -0700)]
osd: use the client_messenger for the logclient.

15 years agoosdmap: remove unused versions of get_[cluster_]inst
Greg Farnum [Thu, 29 Jul 2010 21:20:31 +0000 (14:20 -0700)]
osdmap: remove unused versions of get_[cluster_]inst

15 years agoOSD: use get_cluster_inst instead of get_inst when checking other OSD instances
Greg Farnum [Thu, 29 Jul 2010 20:58:05 +0000 (13:58 -0700)]
OSD: use get_cluster_inst instead of get_inst when checking other OSD instances

15 years agoosd: when checking OSD identities via require_same_or_newer_map, use cluster_addr
Greg Farnum [Thu, 29 Jul 2010 17:58:52 +0000 (10:58 -0700)]
osd: when checking OSD identities via require_same_or_newer_map, use cluster_addr

15 years agoosd: add Mutex and flag to cover execution of handle_map_lock.
Greg Farnum [Thu, 29 Jul 2010 17:24:14 +0000 (10:24 -0700)]
osd: add Mutex and flag to cover execution of handle_map_lock.
Fixes a crash when the cluster and client messenger both dispatch an
OSDMap and their handling gets intertwined.

15 years agoosd: be more careful with cluster/client messenger use.
Greg Farnum [Thu, 15 Jul 2010 20:52:46 +0000 (13:52 -0700)]
osd: be more careful with cluster/client messenger use.

15 years agoosd: register the OSD on cluster_messenger if needed
Greg Farnum [Thu, 15 Jul 2010 18:26:04 +0000 (11:26 -0700)]
osd: register the OSD on cluster_messenger if needed

15 years agoconfig: correctly notify when OPT_ADDR config options are found.
Greg Farnum [Mon, 12 Jul 2010 18:32:23 +0000 (11:32 -0700)]
config: correctly notify when OPT_ADDR config options are found.

15 years agoconfig: add OPT_ADDR type, move "public_addr"/"cluster_addr" processing there
Greg Farnum [Fri, 9 Jul 2010 18:58:48 +0000 (11:58 -0700)]
config: add OPT_ADDR type, move "public_addr"/"cluster_addr" processing there
Necessary side effect: move public_addr/cluster_addr into g_conf.

15 years agomessages: add cluster_addr to MOSDBoot, use it in OSD and OSDMonitor
Greg Farnum [Thu, 8 Jul 2010 22:51:07 +0000 (15:51 -0700)]
messages: add cluster_addr to MOSDBoot, use it in OSD and OSDMonitor

15 years agocosd: move messenger existence checks to where they're useful
Greg Farnum [Thu, 8 Jul 2010 22:14:05 +0000 (15:14 -0700)]
cosd: move messenger existence checks to where they're useful

15 years agocosd: create separate cluster/client messengers if separate addresses are provided.
Greg Farnum [Thu, 8 Jul 2010 22:12:14 +0000 (15:12 -0700)]
cosd: create separate cluster/client messengers if separate addresses are provided.

15 years agomsgr: set up bind and Accepter::bind to take in addresses and bind to them
Greg Farnum [Thu, 8 Jul 2010 22:11:01 +0000 (15:11 -0700)]
msgr: set up bind and Accepter::bind to take in addresses and bind to them

15 years agoconfig: rename g_my_addr to g_public_addr, add g_cluster_addr
Greg Farnum [Thu, 8 Jul 2010 22:10:22 +0000 (15:10 -0700)]
config: rename g_my_addr to g_public_addr, add g_cluster_addr

15 years agoosdmap: move osd_cluster_addr decoding, and make sure that size matches osd_addr
Greg Farnum [Thu, 8 Jul 2010 19:06:08 +0000 (12:06 -0700)]
osdmap: move osd_cluster_addr decoding, and make sure that size matches osd_addr

15 years agoosdmap: Make sure osd_cluster_addr matches osd_addr size on decode
Greg Farnum [Thu, 8 Jul 2010 18:30:13 +0000 (11:30 -0700)]
osdmap: Make sure osd_cluster_addr matches osd_addr size on decode

15 years agoosd: remove messenger pointer, fix calls to it in PG and ReplicatedPG
Greg Farnum [Thu, 8 Jul 2010 18:29:46 +0000 (11:29 -0700)]
osd: remove messenger pointer, fix calls to it in PG and ReplicatedPG

15 years agoosd: use cluster/client_messenger as appropriate.
Greg Farnum [Wed, 7 Jul 2010 23:23:43 +0000 (16:23 -0700)]
osd: use cluster/client_messenger as appropriate.

Modify send_incremental_map to choose based on the type of the entity.

15 years agoosdmap: add osd_cluster_addr to full map
Greg Farnum [Wed, 7 Jul 2010 22:15:22 +0000 (15:15 -0700)]
osdmap: add osd_cluster_addr to full map
Include it in lookups and add get_cluster_inst et al to match get_inst et al

15 years agoosd: make new cluster_messenger, client_messenger pointers.
Greg Farnum [Wed, 7 Jul 2010 20:47:32 +0000 (13:47 -0700)]
osd: make new cluster_messenger, client_messenger pointers.

Don't remove old messenger, just alias it to client_messenger.

15 years agoosdmap: make room in incremental for separate cluster/client OSD addrs.
Greg Farnum [Wed, 7 Jul 2010 18:59:45 +0000 (11:59 -0700)]
osdmap: make room in incremental for separate cluster/client OSD addrs.

15 years agoosd: fix replies to dup/already-committed requests
Sage Weil [Tue, 3 Aug 2010 18:07:58 +0000 (11:07 -0700)]
osd: fix replies to dup/already-committed requests

We can safely reply if oldv is _older_ than what's on disk.  Otherwise the
request ends up on the list and never gets removed again.

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoMerge branch 'testing' into unstable
Sage Weil [Tue, 3 Aug 2010 16:43:09 +0000 (09:43 -0700)]
Merge branch 'testing' into unstable

15 years agoassert: only include acconfig.h ifdef __CEPH__
Sage Weil [Tue, 3 Aug 2010 16:42:26 +0000 (09:42 -0700)]
assert: only include acconfig.h ifdef __CEPH__

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agolibrados.hpp fix
Takuya ASADA [Tue, 3 Aug 2010 12:21:10 +0000 (21:21 +0900)]
librados.hpp fix

[ The following text is in the "UTF-8" character set. ]
    [ Your display is set for the "iso-8859-1" character set.  ]
    [ Some characters may be displayed incorrectly. ]

Hi,

I just realized librados.hpp could compile only if "using namespace
std;" is declared before "#include <rados/librados.hpp>", because
"vector" used without std namespace, and I think it's wrong.
It can fix following patch.

syuu

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agomon: fix signed/unsigned compile warning
Sage Weil [Thu, 29 Jul 2010 18:37:17 +0000 (11:37 -0700)]
mon: fix signed/unsigned compile warning

15 years agoosd: fix map sharing with old heartbeat peers
Sage Weil [Tue, 3 Aug 2010 16:36:55 +0000 (09:36 -0700)]
osd: fix map sharing with old heartbeat peers

The addr is in old_inst, since it's an old peer.  Fixes a crash like

osd/OSD.cc: In function 'void OSD::_share_map_outgoing(const entity_inst_t&)':
osd/OSD.cc:1791: FAILED assert(inst.name.is_osd())
 1: (OSD::update_heartbeat_peers()+0x1d3f) [0x4da66f]
 2: (OSD::activate_map(ObjectStore::Transaction&, std::list<Context*, std::allocator<Context*> >&)+0x8ee) [0x4db61e]
 3: (OSD::handle_osd_map(MOSDMap*)+0x233a) [0x4e506a]
 4: (OSD::_dispatch(Message*)+0x230) [0x4ef400]
 5: (OSD::ms_dispatch(Message*)+0x39) [0x4efe39]
 6: (SimpleMessenger::dispatch_entry()+0x749) [0x461fa9]
 7: (SimpleMessenger::DispatchThread::entry()+0x1c) [0x458f6c]
 8: (Thread::_entry_func(void*)+0xa) [0x46cf1a]
 9: (()+0x69ca) [0x7f978d6fb9ca]
 10: (clone()+0x6d) [0x7f978c91b6cd]

introduced by 9bfb8da9f925642bca46528a999124cd8b28ba2a.

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agomds: fix cap encoding skipping when snaprealm differs
Sage Weil [Mon, 2 Aug 2010 23:27:18 +0000 (16:27 -0700)]
mds: fix cap encoding skipping when snaprealm differs

We were avoiding adding a new cap, but _any_ issue is problematic, not just
adds (which we can't really be sure about anyway due to races with client
cap release).

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoosdmap: show flags
Sage Weil [Mon, 2 Aug 2010 23:01:44 +0000 (16:01 -0700)]
osdmap: show flags

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoMerge branch 'locks' into unstable
Greg Farnum [Mon, 2 Aug 2010 22:36:26 +0000 (15:36 -0700)]
Merge branch 'locks' into unstable

15 years agomsgr: Add CEPH_FEATURE_FLOCK to default features set, remove extraneous data from...
Greg Farnum [Mon, 2 Aug 2010 17:39:40 +0000 (10:39 -0700)]
msgr: Add CEPH_FEATURE_FLOCK to default features set, remove extraneous data from protocol

15 years agoceph_fs: use __le64 instead of __u64 for over-the-wire data
Greg Farnum [Thu, 24 Jun 2010 00:11:42 +0000 (17:11 -0700)]
ceph_fs: use __le64 instead of __u64 for over-the-wire data

15 years agomds: new low-level debug output, and turn down previous debugs
Greg Farnum [Tue, 18 May 2010 18:45:26 +0000 (11:45 -0700)]
mds: new low-level debug output, and turn down previous debugs

15 years agodebian: add ceph-client-tools dep to ceph
Thomas Mueller [Mon, 2 Aug 2010 20:27:16 +0000 (13:27 -0700)]
debian: add ceph-client-tools dep to ceph

Signed-off-by: Thomas Mueller <thomas@chaschperli.ch>
Signed-off-by: Sage Weil <sage@newdream.net>
15 years agodebian: move utils rbd,rados,csyn to ceph-client-tools
Thomas Mueller [Mon, 2 Aug 2010 16:27:22 +0000 (09:27 -0700)]
debian: move utils rbd,rados,csyn to ceph-client-tools

Signed-off-by: Thomas Mueller <thomas@chaschperli.ch>
Signed-off-by: Sage Weil <sage@newdream.net>
15 years agodebian: move mount.ceph to ceph-client-tools
Thomas Mueller [Mon, 2 Aug 2010 16:25:20 +0000 (09:25 -0700)]
debian: move mount.ceph to ceph-client-tools

Signed-off-by: Thomas Mueller <thomas@chaschperli.ch>
Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoMerge branch 'testing' into unstable
Sage Weil [Mon, 2 Aug 2010 20:26:23 +0000 (13:26 -0700)]
Merge branch 'testing' into unstable

15 years agodebian: move fuse-utils dep to ceph-fuse
Sage Weil [Mon, 2 Aug 2010 20:26:09 +0000 (13:26 -0700)]
debian: move fuse-utils dep to ceph-fuse

Reported-by: Thomas Mueller <thomas@chaschperli.ch>
Signed-off-by: Sage Weil <sage@newdream.net>
15 years agocclass: fix .so loading file
Christian Brunner [Mon, 2 Aug 2010 19:58:30 +0000 (21:58 +0200)]
cclass: fix .so loading file

The only thing I've found, is that the *.so files are included in the -devel
rpm. This leads to the fact that the 'find' in cclass doesn't return any
files. I'd suggest the attached change in the script.

Christian

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoosd: send latest osdmap when we stop sending heartbeats to peer
Sage Weil [Mon, 2 Aug 2010 20:11:22 +0000 (13:11 -0700)]
osd: send latest osdmap when we stop sending heartbeats to peer

Otherwise, they may still expect them, and may mark us down.

15 years agoosd: avoid most of handle_osd_map if there are no new maps in message
Sage Weil [Mon, 2 Aug 2010 19:57:36 +0000 (12:57 -0700)]
osd: avoid most of handle_osd_map if there are no new maps in message

We don't want to take the time to flush things to disk and shut down all
the worker threads if there is nothing new in the osdmap update (as is
usually the case).

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoosd: send heartbeat request from update_heartbeat_peers
Sage Weil [Mon, 2 Aug 2010 20:12:29 +0000 (13:12 -0700)]
osd: send heartbeat request from update_heartbeat_peers

This keeps our from_stamps in a consistent state, and sends the hb request
sooner rather than later.

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agomon: drop redundant pg_temp mappings
Sage Weil [Fri, 30 Jul 2010 20:16:24 +0000 (13:16 -0700)]
mon: drop redundant pg_temp mappings

15 years agomon: make backward-compat mon id's start at '0', not 'a'
Sage Weil [Thu, 29 Jul 2010 21:16:47 +0000 (14:16 -0700)]
mon: make backward-compat mon id's start at '0', not 'a'

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoosd: add asserts when indexing pg log by reqid
Sage Weil [Mon, 2 Aug 2010 19:32:06 +0000 (12:32 -0700)]
osd: add asserts when indexing pg log by reqid

This should help catch callers that submit dup entries with the same
request id, causing other problems down the line.

15 years agomds: recover filelocks properly on client reconnect
Greg Farnum [Tue, 18 May 2010 16:02:27 +0000 (09:02 -0700)]
mds: recover filelocks properly on client reconnect

15 years agomds: debug output and some comments.
Greg Farnum [Mon, 17 May 2010 22:18:24 +0000 (15:18 -0700)]
mds: debug output and some comments.

15 years agomds: locking. Compare iterator to right list!
Greg Farnum [Mon, 17 May 2010 22:17:28 +0000 (15:17 -0700)]
mds: locking. Compare iterator to right list!

15 years agomds: locking. Fix waiting lock count map deletion
Greg Farnum [Mon, 17 May 2010 22:14:58 +0000 (15:14 -0700)]
mds: locking. Fix waiting lock count map deletion

15 years agomds: drop distributed locks on waiting filelock messages, so they don't block
Greg Farnum [Mon, 17 May 2010 22:14:03 +0000 (15:14 -0700)]
mds: drop distributed locks on waiting filelock messages, so they don't block

15 years agomds: better printing on lock messages
Greg Farnum [Fri, 14 May 2010 23:53:12 +0000 (16:53 -0700)]
mds: better printing on lock messages

15 years agomds: replace accidnetal __u64 with uint64_t
Greg Farnum [Fri, 14 May 2010 20:42:07 +0000 (13:42 -0700)]
mds: replace accidnetal __u64 with uint64_t

15 years agomds: Determine if there are file locks based on flockbl size, not dirty cap bit
Greg Farnum [Thu, 13 May 2010 21:20:17 +0000 (14:20 -0700)]
mds: Determine if there are file locks based on flockbl size, not dirty cap bit

15 years agomds: need to size values read from a bufferlist
Greg Farnum [Thu, 13 May 2010 19:04:43 +0000 (12:04 -0700)]
mds: need to size values read from a bufferlist