]>
git.apps.os.sepia.ceph.com Git - ceph.git/log
Sage Weil [Fri, 9 Oct 2009 22:15:13 +0000 (15:15 -0700)]
byteorder: more typos
Sage Weil [Fri, 9 Oct 2009 19:19:35 +0000 (12:19 -0700)]
qa: start new qa dir, add some client tests
Sage Weil [Fri, 9 Oct 2009 17:55:32 +0000 (10:55 -0700)]
byteorder: braces!
Sage Weil [Fri, 9 Oct 2009 17:53:37 +0000 (10:53 -0700)]
osd: check is_dup earlier; do not set osdop version if it's a noop
Otherwise, we end up with a bogus version that's not logged and in the
future, which confuses us later (we think it's a replay or something). If
it's a noop, that means the version is either already set, or doesn't need
to be.
Sage Weil [Fri, 9 Oct 2009 05:20:49 +0000 (22:20 -0700)]
osd: use sized int32 in osdmap incremental new_pg_temp
Oops!
Sage Weil [Fri, 9 Oct 2009 04:20:22 +0000 (21:20 -0700)]
msgr: reset session state on lossy_tx fault
Sage Weil [Thu, 8 Oct 2009 22:18:05 +0000 (15:18 -0700)]
byteorder: include endian.h for good measure
Sage Weil [Thu, 8 Oct 2009 21:58:03 +0000 (14:58 -0700)]
Revert "byteorder: properly detect endianness"
Or not!
This reverts commit
a33441c4b274d840432ba553db463cbee1f1a480 .
Sage Weil [Thu, 8 Oct 2009 21:21:08 +0000 (14:21 -0700)]
byteorder: properly detect endianness
Sage Weil [Thu, 8 Oct 2009 21:20:52 +0000 (14:20 -0700)]
osd: fix up handling of dup PG::Info
Ignore dup info. Simplify broken logic.
Sage Weil [Thu, 8 Oct 2009 18:45:27 +0000 (11:45 -0700)]
mon: fix session close races
Make sure we don't use a connection's session if it is closed.
Remove session from connection on reset.
Sage Weil [Thu, 8 Oct 2009 18:44:17 +0000 (11:44 -0700)]
objecter: don't send duplicate onetime sub requests
Sage Weil [Thu, 8 Oct 2009 18:43:43 +0000 (11:43 -0700)]
osd: don't manually set rank addr
This didn't accomplish anything useful.
Sage Weil [Thu, 8 Oct 2009 18:41:32 +0000 (11:41 -0700)]
msgr: include port when printing sockaddrs!
Sage Weil [Thu, 8 Oct 2009 17:35:43 +0000 (10:35 -0700)]
libceph: move to main dir with other main .cc files
Sage Weil [Thu, 8 Oct 2009 17:35:52 +0000 (10:35 -0700)]
monc: no msgr bind needed
Sage Weil [Thu, 8 Oct 2009 17:34:20 +0000 (10:34 -0700)]
libceph: no bind
Sage Weil [Thu, 8 Oct 2009 17:34:09 +0000 (10:34 -0700)]
osd: use fast_fail() policy to connect to osd
Sage Weil [Thu, 8 Oct 2009 17:02:22 +0000 (10:02 -0700)]
msgr: do not require bind() on client apps
Sage Weil [Thu, 8 Oct 2009 16:54:17 +0000 (09:54 -0700)]
msgr: clean out dead connect-side bind code
Sage Weil [Thu, 8 Oct 2009 16:21:39 +0000 (09:21 -0700)]
mds: validate file layouts on file creation, setlayout
Sage Weil [Thu, 8 Oct 2009 16:21:21 +0000 (09:21 -0700)]
ceph: add ceph_file_layout_is_valid function, minimum stripe unit
Sage Weil [Wed, 7 Oct 2009 23:45:02 +0000 (16:45 -0700)]
filestore: be less excited about btrfs version
Sage Weil [Wed, 7 Oct 2009 23:44:38 +0000 (16:44 -0700)]
mon: take lock in ms_handle_reset before adjusting session map
Sage Weil [Wed, 7 Oct 2009 23:43:49 +0000 (16:43 -0700)]
msgr: no need for sockaddr_storage encoder
Yehuda Sadeh [Wed, 7 Oct 2009 23:29:11 +0000 (16:29 -0700)]
messenger: do not bind socket before connecting
Sage Weil [Wed, 7 Oct 2009 22:36:16 +0000 (15:36 -0700)]
version: git rev-parse, not git-rev-parse
Sage Weil [Wed, 7 Oct 2009 18:58:03 +0000 (11:58 -0700)]
mon: allow {getmap,getcurhsmap} <epoch>
Sage Weil [Wed, 7 Oct 2009 18:54:56 +0000 (11:54 -0700)]
crushtool: show zero weight items in crush map decode/dump
Sage Weil [Wed, 7 Oct 2009 18:54:20 +0000 (11:54 -0700)]
crush: zero weight items get 0 length straws
Sage Weil [Wed, 7 Oct 2009 17:36:58 +0000 (10:36 -0700)]
crush: silence (bogus) item uninit warning
Sage Weil [Wed, 7 Oct 2009 17:33:08 +0000 (10:33 -0700)]
push_to_kclient: fix dest tree check
Sage Weil [Wed, 7 Oct 2009 17:32:11 +0000 (10:32 -0700)]
msgr: fix operator<< for sockaddr_in, too
Sage Weil [Wed, 7 Oct 2009 17:31:31 +0000 (10:31 -0700)]
msgr: uninline tcp_{read,write}
Sage Weil [Wed, 7 Oct 2009 17:29:13 +0000 (10:29 -0700)]
msgr: fix uninitialized memory on sockaddr_storage print
Sage Weil [Tue, 6 Oct 2009 21:00:51 +0000 (14:00 -0700)]
todo
Sage Weil [Tue, 6 Oct 2009 20:55:26 +0000 (13:55 -0700)]
crush: reject on empty bucket (and avoid divide by zero)
Sage Weil [Tue, 6 Oct 2009 20:41:55 +0000 (13:41 -0700)]
Makefile: drop src/kernel/* from Makefile.am
Sage Weil [Tue, 6 Oct 2009 20:41:42 +0000 (13:41 -0700)]
uclient: pull ioctl.h from kernel tree to out tree
Sage Weil [Tue, 6 Oct 2009 20:38:25 +0000 (13:38 -0700)]
uclient: fail gracefully if the mount fails
Sage Weil [Tue, 6 Oct 2009 19:20:01 +0000 (12:20 -0700)]
script to push shared files to kclient repo
Sage Weil [Tue, 6 Oct 2009 19:24:35 +0000 (12:24 -0700)]
debian: don't build ceph-kclient-source
That needs to be built from the ceph-client.git tree instead.
Sage Weil [Tue, 6 Oct 2009 18:41:04 +0000 (11:41 -0700)]
kclient: moving to linux-ceph-client.git
Sage Weil [Tue, 6 Oct 2009 18:38:59 +0000 (11:38 -0700)]
move tests around
Yehuda Sadeh [Tue, 6 Oct 2009 19:34:28 +0000 (12:34 -0700)]
conf: also skip utf8 encoded characters that are not printable in current locale
Greg Farnum [Tue, 6 Oct 2009 18:18:17 +0000 (11:18 -0700)]
Hadoop: Add the patch and a Readme file.
Sage Weil [Tue, 6 Oct 2009 18:08:47 +0000 (11:08 -0700)]
debian: upload tarball before .deb build
Otherwise the final tarball includes binaries and is huuge!
Sage Weil [Tue, 6 Oct 2009 18:03:32 +0000 (11:03 -0700)]
debian: name unstable package version ${vers}git${stamp}
This avoids confusion between 0.16.
2009123123123 vs
0.16.0.
2009123123123 .
Sage Weil [Tue, 6 Oct 2009 17:44:36 +0000 (10:44 -0700)]
v0.16.1
Sage Weil [Tue, 6 Oct 2009 17:30:26 +0000 (10:30 -0700)]
rgw: fix build errors on i386
Sage Weil [Tue, 6 Oct 2009 17:26:06 +0000 (10:26 -0700)]
Makefile: include buffer.c in tarball, kclient package
Sage Weil [Tue, 6 Oct 2009 17:05:57 +0000 (10:05 -0700)]
rgw: encode time_t, size_t using fixed-size types
Sage Weil [Tue, 6 Oct 2009 16:48:12 +0000 (09:48 -0700)]
test_ioctls: add missing open arg
Sage Weil [Mon, 5 Oct 2009 23:46:32 +0000 (16:46 -0700)]
kclient: include fs/{Kconfig,Makefile} in patchset
Greg Farnum [Mon, 5 Oct 2009 20:29:53 +0000 (13:29 -0700)]
mds: Don't perform proper rebalancing if set to thrash
Greg Farnum [Mon, 5 Oct 2009 17:51:44 +0000 (10:51 -0700)]
mds: rename do_rebalance to prep_rebalance
Greg Farnum [Mon, 5 Oct 2009 06:32:11 +0000 (23:32 -0700)]
uclient: Only connect to offload targets if the mds goes down.
Sage Weil [Mon, 5 Oct 2009 22:06:50 +0000 (15:06 -0700)]
kclient: checkpatch cleanups
Sage Weil [Mon, 5 Oct 2009 22:04:17 +0000 (15:04 -0700)]
mds: fix default layout settings
Sage Weil [Mon, 5 Oct 2009 21:53:35 +0000 (14:53 -0700)]
v0.16
Sage Weil [Fri, 2 Oct 2009 22:29:31 +0000 (15:29 -0700)]
uclient: fix null deref in unlink() on null dn
Sage Weil [Fri, 2 Oct 2009 21:27:48 +0000 (14:27 -0700)]
todo
Sage Weil [Fri, 2 Oct 2009 22:16:25 +0000 (15:16 -0700)]
mon: fix offload_targets
Greg Farnum [Fri, 2 Oct 2009 21:26:08 +0000 (14:26 -0700)]
uclient: Put some guards up around the Inode->cap map accesses
Greg Farnum [Thu, 1 Oct 2009 22:56:56 +0000 (15:56 -0700)]
msg: handle MMDSLoadTargets
Greg Farnum [Thu, 1 Oct 2009 21:30:36 +0000 (14:30 -0700)]
mon: Handle MMDSLoadTarget messages
Greg Farnum [Thu, 1 Oct 2009 21:30:09 +0000 (14:30 -0700)]
mds/message: MMDSLoadTargets should use a set<int32_t> not a map
Greg Farnum [Thu, 1 Oct 2009 20:51:22 +0000 (13:51 -0700)]
mds: Pull my_target checking out into a function targets_safe
Greg Farnum [Thu, 1 Oct 2009 20:30:00 +0000 (13:30 -0700)]
mds: Don't do rebalancing until the target mds' are in the MDSMap.
This will break rebalancing until the monitor code to
add targets is in place.
Greg Farnum [Thu, 1 Oct 2009 20:28:02 +0000 (13:28 -0700)]
messages: Add a completely useless and unused MMDSOffloadComplete message.
Greg Farnum [Thu, 1 Oct 2009 18:34:56 +0000 (11:34 -0700)]
mon: Put in placeholder code to handle MDS target messages
Greg Farnum [Thu, 1 Oct 2009 00:00:58 +0000 (17:00 -0700)]
mds: Break apart rebalance function into prepwork and offload
Greg Farnum [Tue, 29 Sep 2009 23:37:28 +0000 (16:37 -0700)]
messages: Add an MMDSLoadTargets message for mds->monitor info.
Greg Farnum [Tue, 29 Sep 2009 23:16:46 +0000 (16:16 -0700)]
Message: add constants for MDS offload messages, and clean up the alignment
Greg Farnum [Wed, 30 Sep 2009 18:23:19 +0000 (11:23 -0700)]
Check for invalid layout input in uclient and MDS, replace with sane if needed
Greg Farnum [Wed, 30 Sep 2009 16:55:18 +0000 (09:55 -0700)]
libceph: Actually implement ceph_get_local_osd so that libhadoop works!
Sage Weil [Fri, 2 Oct 2009 20:59:07 +0000 (13:59 -0700)]
kclient: add missing include for in[46]_pton
Sage Weil [Fri, 2 Oct 2009 20:43:59 +0000 (13:43 -0700)]
msgr: use getnameinfo to format socket addresses
Sage Weil [Fri, 2 Oct 2009 20:39:17 +0000 (13:39 -0700)]
use sockaddr_storage; and some ipv6 support groundwork
Sage Weil [Fri, 2 Oct 2009 17:27:30 +0000 (10:27 -0700)]
kclient: don't choke on missing snaprealm
Sage Weil [Thu, 1 Oct 2009 22:43:28 +0000 (15:43 -0700)]
kclient: no system.* xattrs, yet
Sage Weil [Thu, 1 Oct 2009 22:38:42 +0000 (15:38 -0700)]
kclient: kill generic_*xattr calls until we implement xattr_handler
Sage Weil [Thu, 1 Oct 2009 22:34:01 +0000 (15:34 -0700)]
kclient: clean up module startup
Sage Weil [Thu, 1 Oct 2009 21:25:10 +0000 (14:25 -0700)]
Merge branch 'unstable' of ceph.newdream.net:/git/ceph into unstable
Sage Weil [Thu, 1 Oct 2009 21:19:49 +0000 (14:19 -0700)]
kclient: fix ra_pages setup after bdi initiation
Sage Weil [Thu, 1 Oct 2009 21:16:01 +0000 (14:16 -0700)]
kclient: single thread wq for invalidate, truncate
Keep multiple threads for the writeback.
Sage Weil [Thu, 1 Oct 2009 21:12:45 +0000 (14:12 -0700)]
kclient: use KMEM_CACHE macro where possible
We can't for the inode cache bc we have a init_once method.
Sage Weil [Thu, 1 Oct 2009 20:58:00 +0000 (13:58 -0700)]
kclient: mark initialization functions __init
Sage Weil [Thu, 1 Oct 2009 20:45:21 +0000 (13:45 -0700)]
kclient: fix maintainers record
Sage Weil [Thu, 1 Oct 2009 20:30:47 +0000 (13:30 -0700)]
kclient: backport sync_page_range change
Sage Weil [Thu, 1 Oct 2009 20:28:30 +0000 (13:28 -0700)]
todo
Sage Weil [Thu, 1 Oct 2009 20:16:28 +0000 (13:16 -0700)]
kclient: add to MAINTAINERS
Sage Weil [Thu, 1 Oct 2009 20:11:45 +0000 (13:11 -0700)]
kclient: sync_page_range gone in 2.6.32; use vfs_fsync_range
This is what generic_write_sync does, but we also want to do the
sync if we are NEARFULL.
Sage Weil [Thu, 1 Oct 2009 20:05:36 +0000 (13:05 -0700)]
kclient: current_is_pdflush() gone in 2.6.32
Sage Weil [Thu, 1 Oct 2009 19:57:39 +0000 (12:57 -0700)]
todo
Sage Weil [Thu, 1 Oct 2009 19:57:17 +0000 (12:57 -0700)]
Revert "kclient: add smp_mb to sensitive bitops in msgr"
Don't think these are needed after all...
This reverts commit
fe11285c4b00fb9148251cfaaf28aa71a70b31c8 .
Sage Weil [Thu, 1 Oct 2009 18:21:54 +0000 (11:21 -0700)]
kclient: use smp_mb in place of spinlock for testing flag bit
Sage Weil [Thu, 1 Oct 2009 18:07:09 +0000 (11:07 -0700)]
kclient: add smp_mb to sensitive bitops in msgr
Sage Weil [Thu, 1 Oct 2009 17:59:10 +0000 (10:59 -0700)]
uclient: set new file layout to 0, which lets mds choose its default
Greg Farnum [Wed, 30 Sep 2009 17:58:49 +0000 (10:58 -0700)]
uclient: Use much saner/not insane defaults for layout settings