Activity
From 12/22/2011 to 01/20/2012
01/20/2012
- 08:56 PM Revision aea6a305 (ceph): rgw: read_user_buckets() fix redone
- The problem with the original fix is that it wasn't atomic. Going back
to the original inefficient (though atomic) me... - 06:55 PM Revision fdaf91e2 (ceph): osd: implement --dump-journal
- Dump the contents of the journal to stdout in text form. Useful for
debugging.
Signed-off-by: Sage Weil <sage.weil@... - 06:50 PM Revision a52762ac (ceph): rgw: read large bucket directory correctly
- Issue #1955. When there wre too many buckets, we failed reading
the bucket directory.
Signed-off-by: Yehuda Sadeh <y... - 03:48 PM Bug #1958 (Resolved): osd: crash during peering due to receiving an info msg in WaitActingChange
- This happened during a teuthology run with thrashing and reads/writes/deletes.
Logs are in vit:~joshd/bug_1958
<p... - 03:30 PM CephFS Bug #1957: ceph-fuse: have "." and ".." entries consistently
- this is specifically me not know how to handle .. on the root directory with fuse. sshfs does it, though, so it's po...
- 02:56 PM CephFS Bug #1957 (Resolved): ceph-fuse: have "." and ".." entries consistently
- I was cleaning old emails and found this: http://marc.info/?l=ceph-devel&m=130688351921306&w=2
Quick experiment sa... - 12:58 PM rgw Bug #1955 (Resolved): rgw: cannot list user buckets when number of buckets is large
- Issue at rgw code, not librados. read_user_buckets() was broken. Fixed with commit:aea6a305e61c1fa54828b71eff29070c3f...
- 09:29 AM rgw Bug #1955 (Resolved): rgw: cannot list user buckets when number of buckets is large
- could be an issue with librados::tmap_get().
- 11:10 AM Feature #1956 (Resolved): rgw: revisit atomic GET/PUT
- Discuss the following different options to simplify the process:
1. Instead of writing tmp object and then clone i... - 11:02 AM Feature #1944 (Resolved): osd: dump journal
- 12:54 AM Revision 802acb11 (ceph): rgw: refactor acls, separate protocol dependent code
- Does not compile yet, part of swift acls work.
Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
01/19/2012
- 05:11 PM Revision 6c275c81 (ceph): rgw: fix warning
- Signed-off-by: Yehuda Sadeh <yehuda.sadeh@dreamhost.com>
- 01:36 PM Bug #1954 (Resolved): ceph tool: don't create output files when an error occurs
- Doing something like 'ceph osd getmap 1000000 -o osdmap' results in a 0 length file if epoch 1000000 doesn't exist.
- 11:15 AM Bug #1953 (Resolved): teuthology: core files aren't archived when using valgrind
- When daemons crash while running under valgrind, the core file is being saved in the home dir instead of /tmp/cephtes...
- 10:50 AM Bug #1952 (Resolved): rgw: test suite times out
- We have seen a number of instances where the s3-tests test suite (running via jenkins) does not complete after 20 min...
- 10:01 AM rgw Feature #830: rgw: swift per-object ACLs
- Decisions:
(a) buckets used only through S3 see strict S3 behavior
(b) buckets used only through Swift see strict... - 07:51 AM Linux kernel client Bug #1907: rbd: don't reuse device ids while they're still in use elsewhere
- I can ask on fsdevel, but right now I feel the need to understand a
little better what's going in inside rbd in orde... - 04:41 AM Revision 3650fd61 (ceph): Merge remote branch 'gh/wip-op-data-mux'
- Reviewed-by: Greg Farnum <greg.farnum@dreamhost.com>
Reviewed-by: Yehuda Sadeh <yehuda.sadeh@dreamhost.com> - 01:49 AM Revision 29885f3e (ceph): kernel: ignore connection problems while waiting for reboot
01/18/2012
- 11:52 PM Revision e016cca9 (ceph): Convert mount.ceph to use KEY_SPEC_PROCESS_KEYRING
- having mount.ceph use KEY_SPEC_USER_KEYRING to pass keys to the kernel has
several disadvantages:
1) It leaves the k... - 09:16 PM Cleanup #1886 (Resolved): objecter/osd: mux/demux in MOSDOpReply encoding
- 07:46 PM Revision f1f75dd4 (ceph): Merge branch 'wip-rgw-simplelog'
- 07:37 PM Revision 8a9252f9 (ceph): rgw: adjust high level debug level
- setting it to 2 instead of 1
Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> - 07:25 PM Revision b890838c (ceph): Merge remote branch 'gh/wip-rgw-simplelog'
- * gh/wip-rgw-simplelog:
rgw: add timestamp to high level log
rgw: log host_bucket, http status
rgw: simple requ... - 04:05 PM Bug #1849 (Duplicate): directories' timestamps in snapshots sometimes change when directory is mo...
- I believe this is now a duplicate of #1946?
- 03:41 PM Feature #1951 (Resolved): store history of configs and changes
- During a discussion today it occurred to me that there will be situations under which we'll be a lot happier if we ca...
- 02:57 PM rgw Feature #1950 (New): rgw: create S3/Swift ACL interoperability suite
- We need that in order to test swift and s3 ACLs interactions.
- 11:49 AM rgw Feature #1882 (Resolved): rgw: high-level log entries for request state transitions
- Done, merged at commit:f1f75dd4768e017c61088b44e7457bf96916d1a9. Log level 2 now dumps a plain request state with tim...
- 11:45 AM Bug #1949 (Resolved): osd: ENOTEMPTY on collection removal from snaptrimmer
- ...
- 11:27 AM Linux kernel client Bug #1907: rbd: don't reuse device ids while they're still in use elsewhere
- my gut feeling is also that 'echo > /sys/bus/rbd/remove' should return EBUSY (along with rbd unmap). if you can't te...
- 10:40 AM Linux kernel client Bug #1907: rbd: don't reuse device ids while they're still in use elsewhere
- Alex Elder wrote:
> From the linked message:
> > root <at> cephnode3:/# rbd unmap /dev/rbd0
> >
> < -> works wit... - 09:51 AM Linux kernel client Bug #1907: rbd: don't reuse device ids while they're still in use elsewhere
- From the linked message:
> root <at> cephnode3:/# rbd unmap /dev/rbd0
>
< -> works without any error message (sho... - 07:42 AM Revision 148031b7 (ceph): rgw: fix intent log processing
- Intent log processing was completely broken. First, it wasn't
parsing the date correctly (due to failure to initalize... - 07:40 AM Revision 731c8832 (ceph): rgw: initialize tm before calling strptime
- strptime assumes tm is already initialized.
Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> - 05:59 AM Revision 0aab0890 (ceph): objecter: some helpful multiop result debug output
- Signed-off-by: Sage Weil <sage@newdream.net>
- 05:57 AM Linux kernel client Bug #1795 (Resolved): break d_lock > s_cap_lock ordering
- I have verified under UML that no new problems arise with the
fix in place. I have not verified the lockdep warning... - 05:32 AM Revision 6f35e322 (ceph): objecter: make getxattrs set rval on decode error
- Signed-off-by: Sage Weil <sage@newdream.net>
- 05:31 AM Revision f441adfd (ceph): objecter: add stat ops to op vector!
- They work better that way.
Signed-off-by: Sage Weil <sage@newdream.net> - 05:10 AM Revision 1d5c8fd3 (ceph): objecter: gift reply data to outbl _after_ demuxing
- Divvy up the result bl first, then gift the whole shebang to outbl. If
we gift it first, there's nothing to demux (s... - 01:33 AM Revision 2bffed3b (ceph): Merge remote branch 'gh/master' into wip-op-data-mux
- 01:33 AM Revision 905e8d80 (ceph): osd: make in/outdata split/merge helpers static OSDOp methods
- Avoid defining new global functions.
Also add basic doxygen descriptions.
Signed-off-by: Sage Weil <sage.weil@dream...
01/17/2012
- 11:10 PM Revision 1a7c8b49 (ceph): rgw: log_show_next() fix reading of the next buffer
- Bug #1939. Failed reading large logs.
Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> - 11:08 PM Revision 5bb9a9d6 (ceph): Add small cluster thrashing tasks
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 11:05 PM Revision 019a0d4c (ceph): Merge branch 'master' of ssh://github.com/NewDreamNetwork/ceph
- 10:23 PM Revision 956a4b43 (ceph): Merge remote branch 'gh/wip-backfill'
- Reviewed-by: Samuel Just <samuel.just@dreamhost.com>
Conflicts:
src/ceph_mds.cc
src/ceph_osd.cc - 10:21 PM Revision 06e7562f (ceph): filestore: overwrite fsid during --mkfs
- This mainly matters because read_fsid() now looks at the file size to
determine if it's an old- or new-style fsid, an... - 09:42 PM Revision 4c6c4430 (ceph): rgw: reset timestamp when processing starts
- otherwise we'd count also the time waiting for the request.
Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> - 09:00 PM Revision bd8e32d9 (ceph): doc: update control file for setting pg num on pool create
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 09:00 PM Revision a85ea475 (ceph): hadoop: check for valid filehandler, before using in next calls
- In case of nonexistent file, calling Client::replication()
triggers assert.
Signed-off-by: Andrey Stepachev <octo@ya... - 09:00 PM Revision 127bbd17 (ceph): hadoop: fix unix timestamp calculation in hadoop lib
- Hadoop always see wrong dates due of wrong timestamp calculation. Properly
convert nanoseconds to millis when adding.... - 07:43 PM Revision 94f55f48 (ceph): TestRados: fix {min,max}_stride_size initialization
- Signed-off-by: Sage Weil <sage@newdream.net>
- 06:54 PM Revision 0c2f2b76 (ceph): Merge branch 'master' of ssh://ceph.newdream.net/git/ceph
- 06:51 PM Revision 79d19320 (ceph): osd: fix bind error checks
- Signed-off-by: Sage Weil <sage@newdream.net>
- 06:44 PM Revision 7804046d (ceph): Makefile: fix testkeys non-tcmalloc linkage
- Signed-off-by: Sage Weil <sage@newdream.net>
- 05:56 PM Revision 241cbebe (ceph): rgw: add timestamp to high level log
- Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
- 05:56 PM Revision db65295b (ceph): rgw: log host_bucket, http status
- Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
- 05:56 PM Revision 0e8b12cd (ceph): rgw: simple request logging
- Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
- 05:36 PM Revision 63b94b6f (ceph): mds: abort startup if we fail to bind
- Signed-off-by: Sage Weil <sage@newdream.net>
- 05:36 PM Revision 4f70acfa (ceph): osd: abort on startup if we fail to bind to a port
- Signed-off-by: Sage Weil <sage@newdream.net>
- 05:24 PM Revision 45e4c924 (ceph): thrashosds: maxdead default to 0
- This avoids any possibility of blocking peering.
- 04:21 PM Revision 47db4d04 (ceph): ceph: fix "run_uml.sh" script
- Last-minute cleverness prior to checkin broke the "run-uml.sh" script.
Rearange where a few definitions are done to m... - 03:54 PM rgw Bug #1948 (Resolved): rgw: need to read intent log in chunks
- 03:14 PM rgw Bug #1939 (Resolved): rgw: error processing large logs
- Fixed, commit:1a7c8b49f099268ee468877f7f1f7ad747995547.
- 02:58 PM Feature #1658 (Resolved): osd: backfill instead of backlog
- merged in commit:956a4b439759e46424fde3551971cd66b6d682e6
- 02:18 PM Messengers Bug #1942: msgr: Address family not supported by protocol
- commit:dcceb8e835cbf40173c334de18bd68c2cf7f3716 add the osd_fsid to the OSDSuperblock message and reved the version. ...
- 11:27 AM Messengers Bug #1942: msgr: Address family not supported by protocol
- we now report a connection fault instead of asserting. and during initialization we check for bind() errors. afaics...
- 11:47 AM CephFS Bug #1947 (Resolved): mds: SIGBUS during _mark_dirty
- This happened on umount after ffsb with the kernel client.
From teuthology:~teuthworker/archive/nightly_coverage_201... - 11:40 AM Bug #1936: teuthology: github downtime -> failed runs
- Is this going to be okay if the update hangs in the middle and then qa clones the repository?
- 11:25 AM Bug #1936: teuthology: github downtime -> failed runs
- I set up github mirrors for ceph.git, ceph-qa-chef.git, and s3-tests.git. They are at ceph.newdream.net/git, and upd...
- 11:34 AM CephFS Bug #1549: mds: zeroed root CDir* vtable in scatter_writebehind_finish
- Happened again in teuthology:~teuthworker/archive/nightly_coverage_2012-01-15-b/7721/remote/ubuntu@sepia6.ceph.dreamh...
- 11:28 AM Bug #1943 (Need More Info): osd: bad clone transaction on journal replay
- 11:23 AM CephFS Bug #1946: snapshot inherits timestamp/size/etc from modified trunk dir upon mds restart
- It looks to me like the problem is that CInode::old_inodes isn't included in EMetaBlob::fullbit. CInode::pick_old_in...
- 11:18 AM CephFS Bug #1946 (Resolved): snapshot inherits timestamp/size/etc from modified trunk dir upon mds restart
- mkdir .snap/name
ls -ld . .snap/name
# both have the same timestamp
touch .
ls -ld . .snap/name
# now . has a di... - 09:15 AM Bug #1638: Can't create object with large xattrs in a single operation (on extN)
- So now there's an assert on the ENOSPC. I triggered t by running s3tests under teuthology. Adding this here so we kno...
- 09:11 AM CephFS Bug #1945 (Can't reproduce): blogbench hang on caps
- ...
- 12:54 AM Revision 549b7806 (ceph): TestRados: implement max_seconds, reimplement argument parsing
- Signed-off-by: Sage Weil <sage@newdream.net>
- 12:53 AM Revision bf22a4fb (ceph): task/rados: use new usage for radosmodel tool
- 12:22 AM Revision 20f3f686 (ceph): RadosModel: prefix line with m_op
- So we can guage progress...
Signed-off-by: Sage Weil <sage@newdream.net> - 12:16 AM Revision 7b2fd45b (ceph): mds: fix uninitialized value in MClientLease::h
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
01/16/2012
- 11:09 PM Revision b2c07d8a (ceph): add simple thrash workload to regression suite
- 11:05 PM Revision 8fc60869 (ceph): thrashosds: make actions less nonsensical
- Make marking OSD up/down and in/out totally orthogonal.
Signed-off-by: Sage Weil <sage@newdream.net> - 11:05 PM Revision 71390f97 (ceph): thrashosds: fix action selection
- I'm not sure what the old code was trying to do, but I'm pretty sure it
wasn't doing it correctly.. a .1 chance_down ... - 10:38 PM Revision ec6e57c9 (ceph): Merge remote branch 'gh/master' into wip-op-data-mux
- 10:06 PM Revision b5f8de7b (ceph): msgr: move operator<< for sockaddr_storage to msg_types.cc
- tcp.{cc,h} aren't built/linked cleanly.
Signed-off-by: Sage Weil <sage@newdream.net> - 09:26 PM Revision e93999f9 (ceph): qa/workunits/rados/load-gen-mix.sh
- 10k objects, not 100k!
Signed-off-by: Sage Weil <sage@newdream.net> - 09:25 PM Revision ba83e8c6 (ceph): qa: rados load-gen: use rbd pool
- No replay interval.
- 09:18 PM Revision 9419f583 (ceph): ls: include duration, less noise
- 09:18 PM Revision c5bbfffa (ceph): hammer.sh: new -nuke syntax
- 08:39 PM Revision 8fb115fe (ceph): include run duration in summary.yaml
- 07:08 PM Revision 8e126db1 (ceph): mon.0 -> mon.a
- 07:08 PM Revision 43da161d (ceph): mds.0 -> mds.a
- 06:47 PM Revision 7b47e49f (ceph): ls: fix extraneous newline
- 06:40 PM Revision b7a11026 (ceph): rados: load-gen: wake up on reply
- So we can send requests more than once per second.
Signed-off-by: Sage Weil <sage@newdream.net> - 06:40 PM Revision 51e402e3 (ceph): rados: fix load-gen 'max-ops'
- This was mixed up with min/max_op_len. And max_ops wasn't being used
the initial object creation stage, flooding the... - 06:19 PM Revision 7d3b2c41 (ceph): librados: allow ObjectReadOperation::stat() to get time_t mtime
- We can't use the internal utime_t type here.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 06:16 PM Revision ecd6dec6 (ceph): Merge remote branch 'gh/master' into wip-op-data-mux
- Conflicts:
src/librados.cc
src/objclass/class_api.cc
src/rgw/rgw_rados.cc - 05:55 PM Revision b58f9560 (ceph): ceph: ignore all leaks
- unless/until we figure out where the DefinitelyLost records are coming
from.. at first glance they look bogus. - 05:46 PM Revision 706b6910 (ceph): osd: recover_primary_got() -> recover_got()
- This is called on primary and replicas alike.
Signed-off-by: Sage Weil <sage@newdream.net> - 05:34 PM Revision a4e2395f (ceph): osd: clear missing set on replica when restarting backfill
- The primary does the same in PG::activate().
Signed-off-by: Sage Weil <sage@newdream.net> - 05:22 PM Revision 40fb86ff (ceph): ceph: take single arg or list for valgrind args
- 06:54 AM Revision c88ec571 (ceph): combined mon, osd, mds starter functions
- 06:53 AM Revision f8ec23e7 (ceph): rbd: default to all:
- 06:52 AM Revision f7952614 (ceph): lost_unfound: make test work with backfill
- If we backfill, we fail to peer instead of having every object show up as
'unfound'. Avoid that by preventing log tr... - 06:52 AM Revision f70b158c (ceph): show host -> roles mapping on startup
- Less guessing when manually inspecting an in-progress or hung run.
- 06:52 AM Revision fbfa94bb (ceph): teuthology-ls: show pid, last line of output for running jobs
- 06:52 AM Revision 72057a9c (ceph): use local mirrors for (most) github urls
- A cronjob on ceph.newdream.net updates these every 15 minutes. Sigh.
- 06:52 AM Revision 709d9441 (ceph): use local mirrors for (most) github urls
- A cronjob on ceph.newdream.net updates these every 15 minutes. Sigh.
- 05:56 AM Revision a4642946 (ceph): msgr: don't assert on socket(2) failure
- This can happen if we're connecting to an invalid address. Generate an
error message instead of crashing.
See #1942...
01/15/2012
- 10:59 PM Feature #390 (Resolved): Implement bdrv_snapshot_goto (Rollback), bdrv_snapshot_delete
- 10:57 PM Feature #1781 (Resolved): qa: readwrite and roundtrip rgw tests in qa suite
- 10:56 PM rgw Feature #1911 (Closed): rgw: plan handling for large and/or manifest objects, s3 and/or swift
- the plan is to do manifest objects for large s3 objects. that means the pieces won't get a locator and will be distr...
- 10:28 PM Messengers Bug #1942: msgr: Address family not supported by protocol
- Still not sure how the bad address made it into the map (or OSDBoot) message, but at least it won't crash now as of c...
- 05:16 AM Revision a6c06103 (ceph): msgr: uninline operator<< on sockaddr_storage
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
01/14/2012
- 10:01 PM Bug #1928 (Resolved): osd: scrub stat mismatch after fsstress on kernel client
- 09:51 PM Feature #1944: osd: dump journal
- wip-osd-dump-journal
- 06:47 PM Feature #1944 (Resolved): osd: dump journal
- dump text summary of journal contents. this would help debug #1943
- 06:46 PM Bug #1943 (Duplicate): osd: bad clone transaction on journal replay
- Martin got this with v0.40:...
- 02:29 PM Messengers Bug #1942 (Won't Fix): msgr: Address family not supported by protocol
- http://joshp.no-ip.com:8080/20120114-osd-family-error.log.bz2
- 11:01 AM rgw Feature #1941 (Rejected): rgw: revisit bucket removal
- We can try to look again at the steps were doing wen removing buckets, exploring ways to reduce osd operations that c...
- 01:13 AM Revision 6b02f9fa (ceph): osd: rev osd internal cluster protocol
- Prevent backfill code from talking to pre-backfill code.
Signed-off-by: Sage Weil <sage@newdream.net>
01/13/2012
- 11:57 PM Revision 8d271f43 (ceph): Merge branch 'stable'
- 11:08 PM Revision 7f123de8 (ceph): mds: require OSDREPLYMUX feature bit
- We use ObjectOperations now and need a new server to decompose replies
into their constituent components.
Signed-off... - 11:07 PM Revision 012a9855 (ceph): librados: require OSDREPLYMUX feature
- We need this since we now rely on the server telling us rvals and
payload_lens for each OSDOp.
Signed-off-by: Sage W... - 11:07 PM Revision 436f8cac (ceph): define new OSDREPLYMUX feature bit
- This corresponds to the OSDs ability to pass payload_len hints and
return values for each OSDOp in the MSDOOpReply me... - 10:55 PM Linux kernel client Bug #1940: locking cycle in ceph_osdc_start_request
- this was causing teuthology runs to fail.
patch in master, testing! - 10:28 PM Linux kernel client Bug #1940 (Resolved): locking cycle in ceph_osdc_start_request
- ...
- 10:50 PM Revision 9d0476c5 (ceph): objecter: fix add_*() calls to use proper helper
- The helper resizes the other vectors; need that everywhere.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 09:04 PM Revision 42a6cefe (ceph): ReplicatedPG: munge truncate_seq 1/truncate_size -1 to seq 0/size 0
- Truncate with seq 1 and size -1 is a noop.
Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
Reviewed-by: Sage ... - 08:34 PM Revision 0ded7e4d (ceph): ReplicatedPG: munge truncate_seq 1/truncate_size -1 to seq 0/size 0
- Truncate with seq 1 and size -1 is a noop.
Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
Reviewed-by: Sage ... - 08:19 PM Revision 44cb0763 (ceph): rgw: limit object PUT size
- 07:26 PM Revision 3bfa41cf (ceph): Use yaml.safe_dump so unicode doesn't mess up the yaml files.
- In general, yaml.dump is comparable to pickle, and my personal
coding standard says *never* use it. yaml.safe_dump is... - 07:14 PM Linux kernel client Bug #1795: break d_lock > s_cap_lock ordering
- I was hitting some odd behavior while testing. Will try again over
the weekend or early next week.
Also, a note:... - 02:28 PM Linux kernel client Bug #1795: break d_lock > s_cap_lock ordering
- OK, after several iterations and some discussion we
concluded the last two patches (turning these things
into atomi... - 11:54 AM Linux kernel client Bug #1795: break d_lock > s_cap_lock ordering
- I have posted a series of four proposed patches to the list to address
this, along with a few other issues identifie... - 05:06 PM Revision d5753374 (ceph): objecter: fix up stat, getxattrs handlers
- - try/catch
- stat mtime
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 04:36 PM Revision 7eea40ea (ceph): v0.40
- 04:35 PM Revision 224a65a7 (ceph): Merge remote branch 'gh/master' into wip-backfill
- 03:08 PM rgw Bug #1939 (Resolved): rgw: error processing large logs
- 02:06 PM Bug #1928: osd: scrub stat mismatch after fsstress on kernel client
- Looks like the fixes for this introduced a new bug - 3 runs so far today failed with a similar scrub stat mismatch:
... - 10:49 AM Bug #1928 (Closed): osd: scrub stat mismatch after fsstress on kernel client
- 4815cafddf46e968501ac3b96e593c5e8db6218b
- 11:40 AM Bug #1935: teuthology: readwrite/roundtrip jobs run manually, but not in suite
- teuthworker is updated.
- 11:31 AM Bug #1935 (Resolved): teuthology: readwrite/roundtrip jobs run manually, but not in suite
- Alright, I was reading the wrong file. s3readwrite.py and s3roundtrip.py do use the yaml format. The above cleanups w...
- 10:04 AM Bug #1935: teuthology: readwrite/roundtrip jobs run manually, but not in suite
- I'm wrong, ignore me for a while.
- 09:58 AM Bug #1935: teuthology: readwrite/roundtrip jobs run manually, but not in suite
- More specific plan for change:
- change s3tests/functional/__init__.py to read env var S3TEST_YAML, raise if it is... - 09:50 AM Bug #1935: teuthology: readwrite/roundtrip jobs run manually, but not in suite
- s3tests/functional (run via nosetests) reads a .ini -style configuration.
This was not flexible enough for all the... - 11:36 AM CephFS Bug #1938 (Resolved): mds: snaptest-2 doesn't pass with 3 MDS system
- run vstart, mount ceph-fuse, run snaptest-2; mds.a crashes:...
- 09:24 AM Cleanup #1886: objecter/osd: mux/demux in MOSDOpReply encoding
- 08:52 AM Feature #1937 (Resolved): teuthology: --unlock option for -nuke
- this'd make cleanup slightly less painful (no need to check back on old terminal to unlock nuked nodes)
- 08:38 AM CephFS Bug #1682: mds: segfault in CInode::authority
- happened again on /var/lib/teuthworker/archive/nightly_coverage_2012-01-13-a/7335...
- 01:50 AM Revision 81c0ad82 (ceph): librados: make new ObjectReadOperations arguments non-optional
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 01:50 AM Revision 7347538d (ceph): rgw: use new librados ObjectReadOperation method arguments
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 01:47 AM Revision 4815cafd (ceph): ReplicatedPG: Update stat accounting for truncate during write
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
Reviewed-by: Josh Durgin <josh.durgin@dreamhost.com> - 12:39 AM Revision 8ceb3883 (ceph): rgw: wrap cls_cxx_map_* with try/catch around decoding
- 12:23 AM Revision 876d829a (ceph): librados: add ObjectOperation::exec
- 12:23 AM Revision 05d8ecbe (ceph): rgw: bucket index creation and init in a single operation
- 12:14 AM Revision dc628a5b (ceph): secret: move null check before strlen(key_name) deref
- Coverity cid: 98
Signed-off-by: Sage Weil <sage@newdream.net> - 12:10 AM Revision d41ddcdf (ceph): osd: stat op, don't compare in memory state to object
- might be that object is being created by the current compound request.
01/12/2012
- 11:44 PM Revision 4f4b79cc (ceph): osd: include return code in OSDOp
- This will expose the per-operation return values to the caller.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 11:44 PM Revision a8558284 (ceph): osd: mux/demux OSDOp::outdata in MOSDOpReply
- Bump encoding, so that we don't try to demux old encoded messages, which
will likely have OSDOp::payload_len == indat... - 11:44 PM Revision ff55d2f3 (ceph): osd: put result data in OSDOp.outdata
- The removes an argument from do_osd_ops() and cleans up the surrounding
code a bit.
Signed-off-by: Sage Weil <sage@n... - 11:44 PM Revision fe077832 (ceph): objecter: specify read return values pointers in ObjectOperatio methods
- This let's Objecter do the demuxing work for compount read operations.
Signed-off-by: Sage Weil <sage.weil@dreamhost... - 11:44 PM Revision 920bd568 (ceph): librados: specify read return value pointers in ObjectReadOperation met...
- This lets librados do the work of parsing the reply from compound
operations, instead of requiring callers to have kn... - 11:09 PM Revision f42c658d (ceph): osd: fill in empty item in peer_missing for strays
- If we search_for_missing() on a host, make a corresponding entry in our
peer_missing map (if it isn't already there).... - 11:06 PM Bug #1936 (Resolved): teuthology: github downtime -> failed runs
- we should use a github mirror for anything on github. maybe we can/should piggyback off whatever carl set up.
- 11:02 PM Revision 10b00316 (ceph): rgw: don't crash when copying a zero sized object
- 10:48 PM Revision 0da44591 (ceph): nuke: take config files from -t argument
- teuthology-lock and teuthology-updatekeys both use -t for this already
- 09:21 PM Revision 80f57f96 (ceph): ReplicatedPG: fix stat accounting error in CEPH_OSD_OP_WRITEFULL
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 09:21 PM Revision 845aa534 (ceph): ReplicatedPG: Do a write even for 0 length operation
- Otherwise, a 0 length write to an offset past the end of the file will
cause the internal accounting to reflect the f... - 09:02 PM Revision 96e89d30 (ceph): kernel: loop reconnecting in case we race with shutdown
- Previously, if we reconnected before shutdown completed we asserted
that the kernel did not boot into the new version... - 08:59 PM Revision cfa39bfb (ceph): qa/client/gen-1774.sh
- Capture Alexandre's script for reproducing #1774 here for posterity, until
we write a properly harnessed test for thi... - 07:46 PM Revision 6cf77532 (ceph): osd: fix PG::Log::copy_up_to() tail
- The tail needs to refer to the entry preceeding the first entry in the
log. This updates copy_up_to() to match the b... - 07:07 PM Revision 805513be (ceph): osd: reset last_complete on backfill restart
- Since last_backfill is hobject_t(), we can set this equal to last_update.
This fixes a problem where last_complete pr... - 06:38 PM Revision 1e56367e (ceph): client: avoid taking inode ref in case of nonexistent dir
- Signed-off-by: Andrey Stepachev <octo@yandex-team.ru>
Signed-off-by: Sage Weil <sage@newdream.net> - 06:35 PM Revision cedd92be (ceph): Merge branch 'wip-makefile'
- 06:03 PM Revision 71131371 (ceph): COPYING: note licenses for all files, not just the default
- This (mostly) copies debian/copyright for now, but there are format
restrictions for that file. Suggestions for a cl... - 06:03 PM Revision 54e0dfc1 (ceph): debian/copyright: note acx_pthread.m4 license
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 05:42 PM Bug #1935 (Resolved): teuthology: readwrite/roundtrip jobs run manually, but not in suite
- I'm not sure what's going on here.. roundtrip and readwrite are failing when scheduled, but when i copy/paste the sam...
- 05:17 PM Revision 6b55b6b3 (ceph): Makefile: Add headers that were omitted in make dist and prevented test...
- Signed-off-by: Kacper Kowalik (Xarthisius) <xarthisius@gentoo.org>
- 05:17 PM Revision c9e028f4 (ceph): Makefile: Handle corner case of crypto++ correctly
- i.e. use c++ while compiling, append to CRYPTO_LIBS instead of LIBS
Signed-off-by: Kacper Kowalik (Xarthisius) <xart... - 05:17 PM Revision c5144eed (ceph): Makefile: Use ACX_PTHREAD in configure.ac and resulting flags in src/Ma...
- instead of hardcoded flags
Signed-off-by: Kacper Kowalik (Xarthisius) <xarthisius@gentoo.org> - 05:17 PM Revision 7bf01b11 (ceph): Makefile: Add recent acx_pthread.m4 that has a fix for nostdlib issue.
- See http://code.google.com/p/protobuf/issues/detail?id=188 for details
Signed-off-by: Kacper Kowalik (Xarthisius) <x... - 04:58 PM Bug #1759: mds/client: truncate size overflow, fails with EINVAL
- Ok, I reproduced this with osd debugging, but not mds unfortunately. The logs are at slider:/home/samuelj/archived_l...
- 04:45 PM Bug #1930 (Resolved): objclass: need to wrap cls_cxx_map_* with try/catch, protect against bad de...
- Fixed, commit:8ceb388396d02daefe53e3bdb68c08b4855ceaf7.
- 09:48 AM Bug #1930 (Resolved): objclass: need to wrap cls_cxx_map_* with try/catch, protect against bad de...
- 04:26 PM Bug #1931 (Resolved): rgw: bucket index should create and init index atomically
- Fixed, commit:05d8ecbe1cea8d7c7bd33f9b53f4cbf06b2c4e61.
- 09:56 AM Bug #1931 (Resolved): rgw: bucket index should create and init index atomically
- 04:25 PM Linux kernel client Bug #1795 (In Progress): break d_lock > s_cap_lock ordering
- Discussed this with Sage. The problem arose because dentry_lease_is_valid()
is using the MDS session's s_cap_lock f... - 04:13 PM Feature #1934 (Closed): Get new Sepia machines into service
- 03:47 PM rgw Bug #1933 (Resolved): rgw: crash in swift copy
- 03:08 PM rgw Bug #1933: rgw: crash in swift copy
- copy of zero sized objects indeed. Affects both S3 and swift.
Fixed, at commit:10b00316b7778f6aecbf46ec0aea2aca8b8... - 01:36 PM rgw Bug #1933: rgw: crash in swift copy
- Might be a copy of a zero sized object.
- 01:24 PM rgw Bug #1933 (Resolved): rgw: crash in swift copy
- ...
- 03:38 PM Bug #1924 (Resolved): teuthology: installing kernels can fail due to reconnecting too soon
- Fixed by 96e89d30ec5f912f3c1b4844328e0966a2266e05 in teuthology.git.
- 03:24 PM Bug #1909: Two mons crash after starting the third one
- I have reinstalled ceph mon like I wrote but it has a different IP address now. Even though I have changed DNS record...
- 01:13 PM Bug #1928: osd: scrub stat mismatch after fsstress on kernel client
- Samuel Just wrote:
> It seems that fstress will do that: 2012-01-11T14:30:04.867 INFO:teuthology.task.workunit.clien... - 01:07 PM Bug #1928: osd: scrub stat mismatch after fsstress on kernel client
- It seems that fstress will do that: 2012-01-11T14:30:04.867 INFO:teuthology.task.workunit.client.0.out:8/17: dwrite f...
- 12:58 PM Bug #1928: osd: scrub stat mismatch after fsstress on kernel client
- One possibility: in CEPH_OSD_OP_WRITE in ReplicatedPG::do_op we pass op.extent.offset and op.extent.length to write_u...
- 11:07 AM Linux kernel client Feature #1922 (Resolved): rbd: annotate for lockdep
- The fix was to initialize the semaphore in rbd_add().
I have verified that this eliminates the lockdep warning.
... - 10:56 AM Bug #1898 (Duplicate): very long scrub blocked write operation
- Even though this is not the same complaint as 1783, we plan on
fixing it with the same changes, so I am calling this... - 10:30 AM Feature #1932 (Resolved): mon: before accepting a new crushmap, monitor should validate and test ...
- 01:04 AM Cleanup #1899: use acx_pthread instead of hardcoding libs and cflags into build system
- Sage Weil wrote:
> Looks good.. can I add your Signed-off-by to this?
Sure
01/11/2012
- 09:50 PM Revision b93bf285 (ceph): PG: gen_prefix should grab a map reference atomically
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 09:46 PM Feature #1929 (Resolved): teuthology: log runtime
- - include run time in summary.yaml
- include run time in teuthology-ls output
this will help a bit in identifying... - 09:37 PM Revision 38b9b503 (ceph): rgw-admin: add pool rm and pools list
- 09:05 PM Revision e2c02543 (ceph): rgw-admin: clean up unused commands
- 09:04 PM Revision ac1e105e (ceph): osd: bound log we send when restarting backfill
- Use the new tunable from b1da5115aa0756aefa4f0aad36395911e82fce28.
Signed-off-by: Sage Weil <sage@newdream.net> - 08:54 PM Revision 6dae2f8a (ceph): thrasher: adjust min_dead default
- Make this 1, not 2. That's a bit more friendly. It doesn't strictly
matter, tho, since we revive osds before waitin... - 08:54 PM Revision 3c0346b4 (ceph): lost_unfound: typo
- 08:54 PM Revision 59369237 (ceph): thrasher: don't mark down osds out; tell monitor same
- Stopping ceph-osd doesn't make it out (immediately). Prevent monitor
from doing this after a delay too so we can kee... - 08:54 PM Revision 50463ffd (ceph): verify all osds start before checking health
- Just checking health isn't good enough, since it races with OSD startup:
we can have a healthy cluster with 0 (or som... - 08:54 PM Revision fb74b901 (ceph): thrasher: add max_dead
- Add max_dead, and revive osds prior to waiting for clean. Otherwise we
can leave too many OSDs down and the cluster ... - 08:22 PM Revision 79085ad0 (ceph): rados.py: avoid getting return value of void function
- rados_ioctx_locator_set_key is void. The return value seems to have
been uninitialized, so the tests failed rarely.
... - 08:12 PM Linux kernel client Feature #1922: rbd: annotate for lockdep
- I believe the problem is that when a new rbd device structure gets
initialized in rbd_add(), the rw_semaphore contai... - 07:13 PM Revision 85552cf8 (ceph): pg: remove unnecessary guard from calc_trim_to()
- The num_objects check doesn't make sense, and could only make trimming
happen more often than it should. Sage did not... - 07:13 PM Revision b1da5115 (ceph): pg: add a configurable lower bound on log size
- This helps prevent problems with retrying requests being detected as
duplicates. The problem occurs when the log is t... - 06:34 PM Revision 8a9dbc47 (ceph): Merge remote branch 'gh/master' into wip-backfill
- 05:29 PM Bug #1928 (Resolved): osd: scrub stat mismatch after fsstress on kernel client
- ...
- 04:24 PM CephFS Bug #1774: client: files become inaccessible in large directories (with snapshots?)
- This script (properly adjusted to actually mount and remount the ceph-fuse tree) should be enough to trigger the bug.
- 02:41 PM Revision 734737f3 (ceph): osd: limit size of log sent to reset backfill targets
- Need to replace magic number with new tunable, once that is merged.
Signed-off-by: Sage Weil <sage@newdream.net> - 01:47 PM Bug #1925 (Closed): osd: segfault during _scan_list
- b93bf285c9f05ab943e8e506ea2125af0f1f97ad should fix it.
- 06:58 AM Bug #1925 (Closed): osd: segfault during _scan_list
- ...
- 01:39 PM rgw Feature #1927 (Resolved): rgw: add radosgw-admin pool list
- Fixed, commit:38b9b5030747349acf657946133ef57736542310.
- 12:50 PM rgw Feature #1927 (Resolved): rgw: add radosgw-admin pool list
- listing the active set of pools.
- 01:38 PM rgw Feature #1926 (Resolved): rgw: add radosgw-admin pool remove
- Fixed, commit:38b9b5030747349acf657946133ef57736542310.
- 12:49 PM rgw Feature #1926 (Resolved): rgw: add radosgw-admin pool remove
- Would allow removing pool from the active set of pools.
- 01:17 PM Cleanup #1899: use acx_pthread instead of hardcoding libs and cflags into build system
- Looks good.. can I add your Signed-off-by to this?
- 04:25 AM Revision 8f9549f0 (ceph): client: start caching readdir results after readdir_start
- Use upper_bound rather than lower_bound to compute the initial pd within
insert_trace, so that we don't attempt to re... - 12:39 AM Revision 5d989608 (ceph): monclient: fix resolve_addrs() call
- This was broken in def36668a13459d9c0851e4d4da440a288f9a34f it looks like.
Passing uninitialized memory to resolve_ad... - 12:35 AM Revision f09b21ef (ceph): resolve_addrs: return ipv4 and ipv6 addrs
- Fixes: #1891
Signed-off-by: Sage Weil <sage@newdream.net> - 12:22 AM Revision 9e9b5c6f (ceph): ReplicatedPG: fix typo in stats accounting in _rollback_to
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 12:14 AM Revision d4815e5b (ceph): osd: send log with backfill restart
- This makes backfill restart less of a special case: we send an info AND
log, just like we do normally. Code paths ar... - 12:07 AM Revision f4883ebf (ceph): ceph: let the user running ceph-osd remove subvolumes
- This will prevent EPERM when using the SNAP_DESTROY ioctl,
so the filestore will use btrfs snaps.
01/10/2012
- 11:30 PM Revision 2317b9ae (ceph): add rgw readwrite and roundtrip tasks
- 11:28 PM Revision d2fadf9f (ceph): syslog: ignore lockdep non-static key warning
- It looks like this warning was made default in linux 3.2.
This will keep happening until #1922 is done. - 09:23 PM Revision c7d92d1d (ceph): osd: fail to peer if interval lacks any !incomplete replicas
- We need at least one non-incomplete replica during a rw interval in order
to peer. The backfilling/incomplete replic... - 09:00 PM CephFS Bug #1774 (Resolved): client: files become inaccessible in large directories (with snapshots?)
- Yay, that's way better than a log. Merged.
I wanted to make a test case for this but wasn't able to easily reprod... - 07:26 PM Revision 3b81fa58 (ceph): mon: allow specifying pg_num and pgp_num when creating new pools.
- Right now this is only exposed via the monitor command interface:
osd pool create <poolname> [pg_num [pgp_num]]
but i... - 07:20 PM Revision b8916528 (ceph): librados: Make API docs use @note instead of @bug for now.
- Asphyxiate doesn't yet support all of the Doxygen markup.
Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com> - 07:20 PM Revision 741f6d54 (ceph): Fix several doxygen warnings, to minimize noise. Only changes comments.
- Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com>
- 07:20 PM Revision 69aface0 (ceph): auth: Fix Doxygen warnings.
- Match prototype and implementation argument names and types
(textually, that is use std:: prefix).
Signed-off-by: To... - 07:17 PM Revision 1f46e7c9 (ceph): FileStore: assert on ENOSPC even for SETXATTR
- Otherwise we can get corrupt object attributes on ext*.
Signed-off-by: Samuel Just <samuel.just@dreamhost.com> - 06:47 PM Revision 45897b52 (ceph): mds: remove beacon_killer code.
- This no longer does *anything* except print out
useless warning messages.
Signed-off-by: Greg Farnum <gregory.farnum... - 06:47 PM Revision ac9b2d09 (ceph): mds: initiate monitor reconnect if beacon acks take too long
- If it takes 2*mds_beacon_grace (default 30 seconds total) seconds
to get an ack back, maybe it's the monitor and not ... - 05:49 PM Revision ba00f95a (ceph): osd: make less noise when filestore is already up to date
- Signed-off-by: Sage Weil <sage@newdream.net>
- 05:17 PM Bug #1924 (Resolved): teuthology: installing kernels can fail due to reconnecting too soon
- ...
- 04:48 PM Bug #1623 (Can't reproduce): ceph-osd fails to bind socket
- 04:47 PM Bug #1891 (Resolved): monclient: try ipv6 if ipv4 fails
- commit:f09b21eff689c2597c1f006b8a03086d63e1a138
- 04:24 PM Bug #1905 (Resolved): osd: preserve pg log when resetting backfill
- commit:d4815e5bd4727f499643f2bebe2715cc85faaa42
- 04:20 PM Bug #1832 (Closed): osd: size tracking discrepancy (scrub stat mismatch)
- 9e9b5c6f15c6de1af3f804eb10129ffc5d1c9c90 should take care of the more recent occurrence.
- 04:14 PM Tasks #1923 (Resolved): document required properties and features for alternative backend file sy...
- 03:32 PM Linux kernel client Feature #1922 (Resolved): rbd: annotate for lockdep
- This happened when lockdep was turned on:...
- 02:00 PM Feature #1879 (In Progress): osd: track list of in-progress requests, log slow ones
- 01:18 PM Feature #1884 (In Progress): plan encoding strategy to test+facilitate non-disruptive upgrades
- 12:01 PM Feature #390: Implement bdrv_snapshot_goto (Rollback), bdrv_snapshot_delete
- I sent this in patch form to the QEMU list; we'll have to wait on them now...
- 11:45 AM Bug #1921 (Resolved): teuthology: silently continues when len(targets) != len(roles)
- iirc this used to assert, but that probably was lost when the auto-locking was added?
- 11:43 AM Feature #1887 (Resolved): mon: ability to specify pg_num for new pools
- commit:3b81fa5845c158120c66584f3f8d2868cf018e35
- 11:32 AM CephFS Bug #1737: ceph-fuse crash in xlist::remove
- Happened again in teuthology:~teuthworker/archive/nightly_coverage_2012-01-09-b/6741/teuthology.log.
- 11:16 AM CephFS Feature #1912: mds: should time out slow monitors
- Errm, commit:ac9b2d092f1f075621f30f77cdbb49bdf35b9ae5
- 11:13 AM CephFS Feature #1912 (Resolved): mds: should time out slow monitors
- This was actually very easy to do. Ripping out the utterly useless beacon_kill code was actually a larger diff!
- 11:16 AM Bug #1898 (Won't Fix): very long scrub blocked write operation
- We will fix this as part of the incremental scrubbing #1783
- 09:51 AM Bug #1898: very long scrub blocked write operation
- 2012-01-06 16:22:52.45141 another case where osd.220 required 4 minutes to scan the 150000 object pg. osd.97 seems t...
- 11:15 AM Bug #1750 (Closed): xattr errors silently ignored, cause trouble later
- We now assert on ENOSPC for setxattr (1f46e7c91534676f23eaaf8e9740d882928b5685).
- 09:45 AM Linux kernel client Feature #1870 (Resolved): usedcache mount option
- 07:57 AM Bug #1759: mds/client: truncate size overflow, fails with EINVAL
- Hi,
I've triggered this failure mode again with a later snapshot of master (commit: a1252463055e2d6816407bd6465e74... - 03:10 AM Revision ce53447d (ceph): doc: add librados pool creation defaults
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 03:10 AM Revision f99b3e04 (ceph): doc: describe some rados_pool_stat_t members
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 03:10 AM Revision 9490b36e (ceph): doc: add librados C aio example
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 03:10 AM Revision 1e8b8a8d (ceph): doc: standardize rados_tmap_* docs
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 03:10 AM Revision 547e9e34 (ceph): doc: @return -> @returns to match the sphinx output
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 03:10 AM Revision 0e19062c (ceph): doc: clarify librados return codes
- Adding a second @returns for specific error codes makes the sphinx output more readable.
Signed-off-by: Josh Durgin ... - 03:10 AM Revision dd31ff2e (ceph): doc: add short section on documenting code
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 03:10 AM Revision 590520c1 (ceph): doc: fix rados_version todo formatting
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 03:04 AM Revision b464b757 (ceph): doc: move rados_ioctx_get_id to the pool group
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 03:04 AM Revision d9d9e6d3 (ceph): doc: Put rados_ioctx_locator_set_key in a group so it can be cross-refe...
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 03:04 AM Revision 50c9cb16 (ceph): doc: add a prefix to group names in librados.h
- doxygen groups are in a global namespace.
Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com> - 03:03 AM Revision 43952a3b (ceph): doc: add librados C api docs
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 03:03 AM Revision b5759df8 (ceph): doc: add configuration and connecting to librados C api example
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 03:03 AM Revision aca3c621 (ceph): doxygen: Use first sentence as brief description.
- Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com>
- 03:03 AM Revision 78cc07f6 (ceph): librados: Avoid using "crush_rule" as name of function argument.
- "struct crush_rule" exists already, using the same identifier
confuses Doxygen.
Signed-off-by: Tommi Virtanen <tommi... - 03:03 AM Revision c9606416 (ceph): doc: Switch doxygen integration from breathe to asphyxiate.
- TODO: path of librados.h is now just the basename
TODO: no enum support for now
TODO: no @bug support for now
Sign... - 03:03 AM Revision b148bef3 (ceph): doc: fix some typos in librados C API
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 02:13 AM Revision d3e34773 (ceph): ceph: add a new "run_uml.sh" script to manage running a UML client
- This script is used to automate most of what's required to run a
User-Mode Linux (UML) instance. This is mainly of i...
01/09/2012
- 07:59 PM CephFS Bug #1774: client: files become inaccessible in large directories (with snapshots?)
- How about a patch instead of logs? :-)
It turned out that the problem occurred while caching the readdir responses... - 07:40 PM Revision 162ac06f (ceph): rgw: adjust log level
- 07:12 PM Documentation #1815 (Resolved): document librados C api
- 06:37 PM Revision a90f34cd (ceph): rgw: only use plain PUT processor when !chunked_upload
- 06:37 PM Revision 1e0afb62 (ceph): rgw: some cleanup
- 05:50 PM Bug #1898: very long scrub blocked write operation
- It seems that osd.220 took 5 minutes between starting _scan_list and relocking the pg.
- 05:13 PM Revision 1ae42d65 (ceph): Merge remote branch 'gh/master' into wip-backfill
- 04:47 PM Subtask #1920 (Resolved): Update OSD to use ObjectStore tmap implentation (object recovery must c...
- 04:46 PM Subtask #1919 (Resolved): implement the key value interface in terms of leveldb
- 04:39 PM Subtask #1918 (Resolved): create mock key-value store and tests for the related object map implen...
- 04:38 PM Subtask #1917 (Resolved): create interface for backing key-value store and create object map impl...
- 04:37 PM Subtask #1916 (Resolved): add tests ObjectStore level tests for tmap operations
- 04:30 PM Subtask #1915 (Resolved): Create trivial implementation for the object map interface (using curre...
- 04:28 PM Subtask #1914 (Resolved): Create interface for object map implementation
- 04:27 PM Subtask #1913 (Resolved): Add tmap operations to ObjectStore interface
- 03:59 PM CephFS Feature #1912 (Resolved): mds: should time out slow monitors
- Similar to #1841 for the OSDs, the MDS should reconnect to a different monitor if its beacons go too long without get...
- 03:53 PM Bug #1909 (Need More Info): Two mons crash after starting the third one
- can you generate a log with 'debug mon = 20' and 'debug ms = 1' for the existing monitors leading up to the crash?
- 11:32 AM Bug #1909 (Resolved): Two mons crash after starting the third one
- I had three mons. One of them was reinstalled without removing it from the cluster. Now after starting reinstalled mo...
- 03:40 PM Linux kernel client Bug #1871 (Resolved): ceph_client: crash after running xfstests 002
- 03:38 PM rgw Feature #1911 (Closed): rgw: plan handling for large and/or manifest objects, s3 and/or swift
- 03:27 PM rgw Bug #1859 (Resolved): rgw: bucket creation is not atomic
- 03:15 PM CephFS Bug #1047 (In Progress): mds: crash on anchor table query
- All right, looking through a log that Amon got me after I specified the right debug arguments. I see that there are t...
- 02:57 PM Feature #1658 (In Progress): osd: backfill instead of backlog
- 02:46 PM Bug #1903 (Resolved): osd/ReplicatedPG.cc: 3189: FAILED assert(obc->unconnected_watchers.size() =...
- 11:31 AM rbd Feature #1908 (Resolved): rbd: test map/unmap more extensively
- Include unmapping and remapping while reading and writing through an FS or with dd, and simply having an FS mounted.
- 11:23 AM Linux kernel client Bug #1907 (Resolved): rbd: don't reuse device ids while they're still in use elsewhere
- If an FS on top of rbd is mounted, and the rbd device is unmapped, and another one is mapped, the old sysfs entry is ...
- 09:44 AM rgw Bug #1906 (Can't reproduce): rgw: total_time isn't logged consistently
- Just looked at some rgw log, and it seems that the total_time field is usually zero, although there are a few cases w...
- 09:40 AM Bug #1896 (Rejected): osd: FAILED assert(is_up(osd)) from do_queries() during activate_map()
- this was a bug in an early version of commit:a59ee8f91bc879beb614ac10fa2f9a4a284ebfb6
- 01:21 AM Revision c23cc237 (ceph): osd: add OSDOp::outdata
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 01:15 AM Revision 6b389218 (ceph): osd: move OSDOp vector indata split/merge into helpers
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 01:13 AM Revision d8ebbf48 (ceph): osd: OSDOp::data -> indata
- We'll soon add outdata...
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 12:23 AM Revision b17736a1 (ceph): osd: populate_obc_watchers when object pulled to primary
- We don't care about degraded state, only whether the object is on the
primary so that we can load the object_info_t.
...
01/08/2012
- 11:15 PM Revision a59ee8f9 (ceph): osd: handle case where no acceptable info exists
- This happens when the only available replicas has last_backfill != MAX.
In that case, revert to up, and then set the... - 10:39 PM Revision b354ce4e (ceph): run: put pid in archive dir
- This will make it easy for teuthology-ls to show you the running process's
pid (if it's still running). Or for other... - 06:15 PM Revision c5affd6c (ceph): Merge remote branch 'gh/wip-osd-retry-attempt'
- 04:16 PM Revision 5abe49d6 (ceph): Merge remote branch 'gh/wip-admin-socket'
- 03:27 PM Bug #1904 (Resolved): osd: calc_acting bad iterator deref
- commit:a59ee8f91bc879beb614ac10fa2f9a4a284ebfb6
- 08:39 AM Bug #1904 (Resolved): osd: calc_acting bad iterator deref
- ...
- 10:46 AM Bug #1905 (Resolved): osd: preserve pg log when resetting backfill
- If we use the pg log to detect dups, we need to preserve some pg log history when we backfill.
- 10:45 AM Feature #1883 (Resolved): admin socket: string based protocol
- 08:36 AM Bug #1896: osd: FAILED assert(is_up(osd)) from do_queries() during activate_map()
- again,...
01/07/2012
- 06:16 PM Revision fbf79121 (ceph): do not put monitors on the same nodes as clients
- Otherwise, for kernel clients (rbd or kclient), ceph-mon can cause a deadlock when it calls sync(2).
- 10:22 AM Bug #1903 (Resolved): osd/ReplicatedPG.cc: 3189: FAILED assert(obc->unconnected_watchers.size() =...
- ...
- 10:21 AM CephFS Bug #1902 (Won't Fix): mds: unittest_interval_tree bad memory access
- after a segfault:...
- 07:25 AM Bug #1901 (Resolved): Missing files in ceph packages results in build failure of tests
- make dist in src/Makefile.am is missing several header files that are used by tests, as result ceph fails to build.
- 07:23 AM Bug #1900 (Resolved): Fix detection and build issues with libcrypto++
- Currently ceph's build system uses AC_SEARCH_LIBS (in some cases) to search for libcrypto++. As a result C++ library ...
- 07:18 AM Cleanup #1899 (Resolved): use acx_pthread instead of hardcoding libs and cflags into build system
- Hardcoding -lphtread is not the most portable way of including threads support. Please use upstream macros instead.
... - 04:54 AM Revision 92ca3ef7 (ceph): perfcounters: fix unittest for new admin_socket interface
- Broken by b389685afa1be00b5147855bf71c50042bfbfa6c.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 04:39 AM Revision d8e54994 (ceph): Makefile: disable untitest_interval_tree
- Segfaults. Valgrind errors. Accessing uninitialized memory.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 04:38 AM Revision bcf21467 (ceph): unittest_interval_tree: make it compile
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 01:21 AM Revision 13445d23 (ceph): ceph_manager: a booting osd is no longer automatically marked in
- as of ceph.git commit 96b7b0d83e5fe70a4efb4e284e18b4b40840bfec
- 01:18 AM Revision a774d500 (ceph): osd: clean up src_oid, src_obc map key calculation
- Be consistent about how we generate the src_oid and src_oloc, so that we
feed good value into find_object_context and... - 12:55 AM Revision 3c60e804 (ceph): osd: read op should claim_append data instead of claim
- 12:53 AM Revision 0d175cd6 (ceph): rgw: remove object before writing both xattrs and data
- otherwise we'll leak xattrs from previous incarnation
- 12:53 AM Revision 469f3eb4 (ceph): rgw: create plain processor for small objects
- 12:53 AM Revision 75acc0a3 (ceph): rgw: fix multipart PUT
- latest revamp broke it, missed calling RGWPutObjProcessor::prepare(s)
where needed. - 12:53 AM Revision 26b54ae5 (ceph): rgw: rearrange PutObj::execute()
- groundwork for different handling of small object PUTs
- 12:53 AM Revision a0b55397 (ceph): rgw: different atomic handling for small objects
- 12:44 AM Revision 66f38254 (ceph): Merge remote branch 'gh/master' into wip-backfill
- 12:22 AM Revision 199b14d8 (ceph): mon: fix uninitialized cluster_logger_registered
- Signed-off-by: Sage Weil <sage@newdream.net>
01/06/2012
- 11:12 PM Revision 001701a0 (ceph): mon_recovery: need n/2 + 1 monitors for quorum
- 11:08 PM Revision cfeaef45 (ceph): move multimon failure thrashing tests into regression
- We need to test these nightly.
- 09:36 PM Revision da921077 (ceph): ceph: don't skip monitor ports
- We can use the same port multiple times if they are on a different hosts.
- 09:00 PM CephFS Bug #1682: mds: segfault in CInode::authority
- hit this again:...
- 08:50 PM Revision bebd393b (ceph): objecter: ignore replies from old request attempts
- If we know the request attempt, ignore old attempts.
If we do not know the attempt (because the server is old), acce... - 08:49 PM Revision ac177d78 (ceph): osd: encode retry attempt in MOSDOp[Reply]
- In addition to the boolean flag, also encode the exact retry attempt.
Return -1 if we don't know.
Signed-off-by: Sa... - 08:23 PM Revision b501efda (ceph): mon: document quorum_status, mon_status
- Fixes: #1824
Signed-off-by: Sage Weil <sage@newdream.net> - 08:22 PM Revision ca8df7ea (ceph): mon: fix misplaced else
- Broken by 435c29448a10ec343f5a2b7195d94c72de5b1a25.
Signed-off-by: Sage Weil <sage@newdream.net> - 06:20 PM Revision 13f1debb (ceph): Merge remote branch 'gh/wip-mon-timeouts'
- 05:31 PM Revision b389685a (ceph): admin_socket: string commands
- Commands are strings. Old __be32 works too. 'help' to list available
commands.
Signed-off-by: Sage Weil <sage@newd... - 05:31 PM Revision 5a5dece1 (ceph): admin_socket: fix, extend admin_socket unit tests
- Signed-off-by: Sage Weil <sage@newdream.net>
- 05:31 PM Revision 643b9dbd (ceph): ceph: speak new admin socket protocol
- Signed-off-by: Sage Weil <sage@newdream.net>
- 05:19 PM Bug #1896: osd: FAILED assert(is_up(osd)) from do_queries() during activate_map()
- ...
- 03:48 PM Bug #1896 (Rejected): osd: FAILED assert(is_up(osd)) from do_queries() during activate_map()
- ...
- 05:18 PM Bug #1897 (Resolved): osd: failed assert assert(src_obc) under rgw workload
- commit:a774d5002132cffb7b408e7de3d75ee841301fbf
- 03:53 PM Bug #1897 (Resolved): osd: failed assert assert(src_obc) under rgw workload
- ...
- 04:32 PM Bug #1898 (Duplicate): very long scrub blocked write operation
- On cephstore6235 we saw a write operation get blocked for 4 minutes by scrub. The log is available in /var/log/ceph/l...
- 04:23 PM Feature #1895: osd: detect duplicate requests by tracking per-client last_acked_tid instead of us...
- my first thought would be something like:
- set<osd_reqid_t> in the Session
- on session open, load above set fro... - 04:18 PM Feature #1895: osd: detect duplicate requests by tracking per-client last_acked_tid instead of us...
- Naively this data can just go in the OSD::Session struct. However, it might be a bit of a hassle dealing with thrashi...
- 03:15 PM Feature #1895 (Rejected): osd: detect duplicate requests by tracking per-client last_acked_tid in...
- Currently duplicate request detection uses the PG log, which may be trimmed too much to contain actual duplicates. Th...
- 03:16 PM Bug #1490: cfuse assert failure: assert(ob->last_commit_tid < tid)
- One server side cause of this (I'm not sure there aren't more) is duplicate request tracking. Currently requests are ...
- 12:42 PM Bug #1835 (Resolved): Monclient crash when keyring is not readable
- Thanks, Wido!
- 12:38 PM Bug #1835: Monclient crash when keyring is not readable
- Confirmed, this fixed it for me.
- 12:41 PM rgw Tasks #1824 (Resolved): ceph monitor status should be available and documented
- documented in commit:b501efdab9798e30b8bf7990e9a58d076553cd2f
the in-quorum monitors don't keep track of out-of-qu... - 11:53 AM rgw Bug #1859: rgw: bucket creation is not atomic
- resolved, as of commit:bd3ccf7f7b8370bd0e8d52f9f74fcccc7e62c902.
- 10:48 AM CephFS Bug #1435 (Need More Info): mds: loss of layout policies upon mds restart
- Alexandre-
Same situation here. If you can produce a full mds log that includes the creation of the initial layou... - 10:35 AM CephFS Bug #1318: directories disappear across multiple rsyncs
- Alexandre, have you seen this recently? We haven't turned it up in our testing.
Could this be the same as #1774? - 10:31 AM Feature #1808: filestore: gracefully handle EMFILE
- I'm going to call this a feature. The more pressing problem is to find any fd leaks (#1788), and to limit msgr conne...
- 10:28 AM Bug #1831 (Resolved): mon: should not accept (and should disconnect) session when not in quorum
- commit:13f1debbf054612fbb2c9f4dafbe12c8f937cf14
- 10:27 AM Bug #1804 (Closed): filestore: unexpected EINVAL
- ok, i'm going to assume this was the mds trucnate thing and close it out. we can reopen later if it crops up again!
- 10:10 AM CephFS Bug #1774 (New): client: files become inaccessible in large directories (with snapshots?)
- 10:09 AM CephFS Bug #1774 (Need More Info): client: files become inaccessible in large directories (with snapshots?)
- Alexandre-
We're heavily focusing on rados for the next couple of weeks, so I don't have time to try to reproduce ... - 10:07 AM CephFS Bug #1586 (Can't reproduce): failed pjd chmod test 00 on kclient
- haven't seen this since...
- 09:52 AM Bug #1842 (Can't reproduce): osd: failed authorizations leak memory somehow?
- I instrumented the rados client to send bad authenticators and hammered ceph-osd, but massif showed no leaks. I thin...
- 09:43 AM Feature #1798 (Rejected): qa: add rados/librados tests (RadosModel)
- 01:27 AM Revision 561f06cf (ceph): suite: make email-on-success the default behavior
- This way you can tell when a run is complete, instead of wondering if
it's stuck in the queue.
01/05/2012
- 11:42 PM Revision 435c2944 (ceph): mon: instrument elector so you can stop participating in the quorum
- Add new monitor commands "quorum exit" and "quorum enter" to use it.
Signed-off-by: Greg Farnum <gregory.farnum@drea... - 11:42 PM Revision 14a49433 (ceph): mon: elector needs to reset leader_acked on every election start
- Otherwise you never reset the leader_acked after a failed
election attempt, so if mon 0 is available on the first rou... - 11:41 PM Revision 99e5f85e (ceph): mon: kill client sessions when we're not in quorum
- After a timeout of 2*mon_lease length (ie, two election rounds),
kill existing client sessions so they can reconnect ... - 09:46 PM Revision c83b2a0b (ceph): OCF RA: fix variable name
- 09:46 PM Revision b3f8b55d (ceph): debian: build ceph-resource-agents
- 06:51 PM Revision 49a96fa7 (ceph): osd: parameterize min/max values for backfill scanning
- For local scans, use the optimal value for the local filestore.
For remote scans, make it configurable, so we can co... - 06:46 PM Revision 98881a11 (ceph): admin_socket: refactor
- Combine AdminSocketConfigObs with AdminSocket so that we can interact
with it via the cct. Simpler class structure. ... - 06:19 PM Revision b6c43d2a (ceph): rbd: add a command to delete all snapshots of an image
- This makes deleting images with many snapshots easier.
Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com> - 05:39 PM Feature #1879 (New): osd: track list of in-progress requests, log slow ones
- 03:17 PM Feature #1879 (Duplicate): osd: track list of in-progress requests, log slow ones
- 05:38 PM rgw Tasks #1823 (Rejected): radosgw should have internal timeouts
- 02:44 PM rgw Tasks #1823: radosgw should have internal timeouts
- Sage suggests that this can more properly be detected in the OSD:
- add request to tail list when started
- remove ... - 05:32 PM Revision a94b7314 (ceph): admin_socket: whitespace
- Signed-off-by: Sage Weil <sage@newdream.net>
- 05:30 PM Revision 96b7b0d8 (ceph): common: default 'mon osd auto mark in = false'
- This way an osd that was explicitly marked out will stay out, even when
it is restarted.
Signed-off-by: Sage Weil <s... - 05:26 PM Revision a71a0d36 (ceph): osd: log backfill restart
- This is interesting, particularly in determining when a peer that was
partially backfilled needs to be restarted.
Si... - 05:08 PM Bug #1490: cfuse assert failure: assert(ob->last_commit_tid < tid)
- It looks like there are at least two bugs here: one client side, and one server side. I'm reproducing with more logs ...
- 04:46 PM Bug #1831: mon: should not accept (and should disconnect) session when not in quorum
- Updated the branch, it now includes monitor commands and instrumentation so you can drop a monitor out of the quorum....
- 02:21 PM Bug #1831 (In Progress): mon: should not accept (and should disconnect) session when not in quorum
- A basic stab at this is staggeringly boring — pushed to wip-mon-timeouts. I want to discuss instrumenting Monitor to ...
- 02:30 PM RADOS Feature #1894 (New): mon: implement internal heartbeating
- Right now the monitor doesn't really detect if it starts breaking. It should (probably using the HeartbeatMap thingy)...
- 02:03 PM Bug #1893 (Rejected): crushtool can't decompile crushmap
- This is just the cephtool writing excess information to stdout.
- 01:56 PM Bug #1893 (Rejected): crushtool can't decompile crushmap
- This happens on 0.39 and master....
- 02:03 PM Bug #1892 (Rejected): osdmaptool can't decode osdmap
- This is just the cephtool writing excess information to stdout.
- 01:55 PM Bug #1892 (Rejected): osdmaptool can't decode osdmap
- This happens on 0.39 and master....
- 12:58 PM Bug #1891: monclient: try ipv6 if ipv4 fails
- I wonder if changing MonClient::build_initial_monmap() to put both the A and AAAA records in the search pool is suffi...
- 11:30 AM Bug #1891 (Resolved): monclient: try ipv6 if ipv4 fails
- When a hostname is specified, and it has an A and AAAA record, only the ipv4 address is tried.
If this fails, the... - 10:50 AM Bug #1771 (Resolved): rbd: delete snapshots when image is deleted
- Made image deletion fail when snapshots are present (commit:bd2339102f0c650d87203fdc2336f9533a18b755), and added a co...
- 10:08 AM Bug #1832: osd: size tracking discrepancy (scrub stat mismatch)
- This happened in the other direction with python rbd tests:...
- 10:04 AM Bug #1758 (Need More Info): OSD segfault in SimpleMessenger::send_message
- 09:10 AM Feature #1890 (Resolved): log: async log writeout
- dump logs asynchronously (to a file, syslog, or whatever other sync)
make log level and emit level independently a... - 09:09 AM Feature #1889 (Resolved): log: structure log records
- break out some structure from the log entries. for starters,
- threadid
- timestamp
- log level
- subsystem
... - 09:07 AM Feature #1888 (Rejected): log: per-thread ring buffer
- use per thread buffer for log messages to reduce lock contention by logging code
- 08:47 AM Feature #1887 (Resolved): mon: ability to specify pg_num for new pools
- 01:46 AM Revision bd233910 (ceph): librbd: don't remove an image that still has snapshots
- Return -EBUSY instead. After the header is removed, the snapshots
can't be removed or read, so make sure they're gone... - 01:34 AM Revision 4728f4f8 (ceph): SimpleMessenger: clarify when ms_bind_ipv6 is used
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 01:11 AM Revision c8a13f2b (ceph): qa: fix mdstable script for proper injectargs use.
- This script is fairly primitive, but somebody might find it useful...
Signed-off-by: Greg Farnum <gregory.farnum@dre... - 01:11 AM Revision 4ea8ad43 (ceph): qa: add a slightly more stressful anchortable test
- This creates more than 8 links.
Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com> - 12:38 AM Revision 3935551d (ceph): Merge remote branch 'gh/master' into wip-backfill
- 12:25 AM Revision 61c3a918 (ceph): rados: fix run-length option parsing for rados load-gen
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
01/04/2012
- 11:57 PM Revision 857b243b (ceph): osdmap: include state names in dump()
- Signed-off-by: Sage Weil <sage@newdream.net>
- 11:57 PM Revision b8be3382 (ceph): mon: rev cluster protocol
- The OSDMap NEW and AUTOOUT bit additions subtely change the decoding of
the incremental maps in a reasonably harmless... - 11:57 PM Revision 9986553c (ceph): msgr: explicitly specify internal cluster protocol
- Replace case statement based on my_type.
Signed-off-by: Sage Weil <sage@newdream.net> - 11:57 PM Revision 0fb88b6f (ceph): move cluster protocol definitions out of ceph_fs.h
- Among other things, we don't recompile the whole system when we touch
these.
Signed-off-by: Sage Weil <sage@newdream... - 11:57 PM Revision 9510f9c9 (ceph): mon: track auto-marked-out osds
- Mark OSDs that were automatically marked OUT by the monitor because they
were down for too long. Clear the bit as so... - 11:57 PM Revision fcb87701 (ceph): mon: independently control whether AUTOOUT OSDs are marked in on boot
- Add separate config option to control whether the monitor will mark
AUTOOUT OSDs in on boot.
Signed-off-by: Sage Wei... - 11:57 PM Revision af535077 (ceph): mon: maintain CEPH_OSD_NEW bit for new, unused OSDs
- Signed-off-by: Sage Weil <sage@newdream.net>
- 11:57 PM Revision 80d90109 (ceph): mon: separately control auto-mark-in of new OSDs
- Signed-off-by: Sage Weil <sage@newdream.net>
- 11:56 PM Revision adea084a (ceph): osd: mark degraded only when < desired replica count
- Having extra replicas is not 'degraded' per se. Although it's weird that
we ever do that!
Signed-off-by: Sage Weil ... - 11:56 PM Revision becb71b0 (ceph): osd: don't add all strays in calc_acting()
- We weren't counting up usable strays, which meant we added all of them.
This could result in acting sets with more ac... - 11:56 PM Revision d3395335 (ceph): osd: fix backfill reset on activate
- Look at peer's info, now our own!
Signed-off-by: Sage Weil <sage@newdream.net> - 11:56 PM Revision 3166b373 (ceph): osd: avoid querying missing set from (full) backfill target
- If we are doing a complete backfill, we don't care about missing; it will
clearly all be below last_backfill anwyay a... - 11:53 PM Revision 6a918feb (ceph): Merge pull request #8 from kylemarsh/master
- Remove cloudfiles requirement from obsync.
- 10:21 PM Revision 8658f0d5 (ceph): qa: load-gen-mix-small-long
- 30 minutes
Signed-off-by: Sage Weil <sage@newdream.net> - 10:14 PM Revision 5cebc740 (ceph): obsync: make obsync run without cloudfiles installed
- Cloudfiles probably shouldn't be a requirement for running obsync, so this
commit makes it optional. - 09:59 PM Revision 4bcdb37c (ceph): osd: do not use incomplete peer for best info/log
- For one, their stats are incomplete; if we use them we'll screw up everyone
else. For another, it doesn't do us any ... - 09:59 PM Revision dcd1ebab (ceph): osd: initialize backfill_pos on activate
- Handling of writes depends on backfill_pos being initialized (to know what
is between the leading and trailing edge o... - 09:59 PM Revision 6a1cac92 (ceph): osd: initialize backfill_target; include in PG operator<<
- Signed-off-by: Sage Weil <sage@newdream.net>
- 06:46 PM Revision a5d99add (ceph): osd: fix misdirect check for requests with old epochs
- get_map() assumes the epoch passed is valid. Check here in the caller.
Signed-off-by: Sage Weil <sage@newdream.net> - 06:44 PM Revision 172dd3e0 (ceph): osd: check that we're supposed to be getting a PG before waitlisting re...
- This was broken in fa722de6708d3e92037df6289cc29ece12c8ea66.
Fix it by checking if the mapping was correct in the se... - 06:40 PM Revision 54f36f0b (ceph): rados: gracefully report errors from 'ls'
- Catch the exception thrown by the iterator when the OSD returns errors.
Signed-off-by: Sage Weil <sage@newdream.net> - 06:38 PM Revision ed9a4a09 (ceph): osd: return EINVAL on bad PGLS[_FILTER] handle
- Fixes: #1875
Signed-off-by: Sage Weil <sage@newdream.net> - 05:25 PM CephFS Bug #1047: mds: crash on anchor table query
- After looking into this a lot today, none of my ideas panned out. I also tried a few simple tests to reproduce locall...
- 04:50 PM Bug #1858 (Resolved): OSD needs to check for misdirected ops before putting non-existent PGs on hold
- Sage merged a modified form of this in commit: 172dd3e0e40ead65349904e604962eb0724bc230.
- 04:35 PM Cleanup #1886 (Resolved): objecter/osd: mux/demux in MOSDOpReply encoding
- - osd: return read results via OSDOp, not odata, in do_osd_ops()
- MOSDOpReply: mux/demux based on payload_len
- ob... - 03:37 PM Feature #1885 (Resolved): identify top 10 expected failures and process to diagnose
- - peering failures
- unfound objects
- 03:36 PM Feature #1884 (Resolved): plan encoding strategy to test+facilitate non-disruptive upgrades
- - best practices for encoding structures, like
- single encoding version vs compat+incompat version value
- pr... - 03:35 PM Feature #1883 (Resolved): admin socket: string based protocol
- switch admin socket from a u32 based binary protocol to a string based one. e.g., commands like "perfcounter_dump\n"...
- 03:33 PM rgw Feature #1882 (Resolved): rgw: high-level log entries for request state transitions
- log request start and transition through high-level stages (start, authenticate, read, write, etc.)
probably just ... - 03:32 PM Feature #1881 (Resolved): objecter: expose in-progress request state via admin socket
- an admin socket command that will dump current in-progress requests, similar to cat /sys/kernel/debug/ceph/*/osdc
- 03:31 PM Feature #1880 (Rejected): osd: optionally log all request latencies
- start/stop+dump via admin socket?
need something like this to evaluate distirbution of latencies (e.g. 99 percenti... - 03:31 PM Feature #1879 (Resolved): osd: track list of in-progress requests, log slow ones
- - add request to tail list when started
- remove when complete
- periodically scan start of list and log slow reque... - 11:32 AM Feature #1422: libvirt: rbd storage pool
- I've done some work on this, see: http://permalink.gmane.org/gmane.comp.file-systems.ceph.devel/4846
- 10:48 AM Bug #1875 (Resolved): osd: ReplicatedPG::do_op
- commit:ed9a4a0996a9879bb2798a0771b263312f5d88fc
- 09:55 AM Bug #1875: osd: ReplicatedPG::do_op
- The PGLS iterator handle format was recently changed, and this crashed while decoding it. My guess is an old binary ...
- 05:35 AM Bug #1875 (Resolved): osd: ReplicatedPG::do_op
- I just noticed two OSD's (osd.11 and osd.20) go down in my cluster.
The backtrace of both OSD's:... - 10:47 AM Bug #1831: mon: should not accept (and should disconnect) session when not in quorum
- Last night a test hit this. The MDS got stuck connected to an out-of-quorum mon, and never stopped being laggy.
- 10:33 AM CephFS Bug #1874: Running `git gc` on a bare git repository hosted by ceph results in a bus error.
- Hi,
Drat, I was hoping this would be a simple-to-reproduce case. Never mind, here are some more details:
Kerne... - 10:06 AM CephFS Bug #1874 (Need More Info): Running `git gc` on a bare git repository hosted by ceph results in a...
- 10:06 AM CephFS Bug #1874: Running `git gc` on a bare git repository hosted by ceph results in a bus error.
- Which version of the kernel client and server are you running?
Can you attach an strace -f of the 'git gc' run s... - 03:27 AM CephFS Bug #1874 (Can't reproduce): Running `git gc` on a bare git repository hosted by ceph results in ...
- When @git gc@ is run on a bare git repository hosted by a local test ceph filesystem mounted via the kernel client, i...
- 10:19 AM CephFS Bug #1878 (Resolved): ceph.ko doesn't setattr (lchown, utimes) on symlinks
- rsync -a /my/symlink/pool/ /mnt/ceph.ko/pool/ silently fails to set times and ownership of symlinks, whereas the same...
- 10:02 AM CephFS Bug #1877 (Can't reproduce): ceph.ko (3.1.6) oopses upon cephfs set_layout of a symlink to a dir
- I moved elsewhere a directory that had a layout policy set. Later on, next time the mds lost directory policy inform...
- 09:57 AM Cleanup #1876 (Resolved): osd: EINVAL on all client command decoding errors
- There are various other places in the osd (besides #1875) where we decode data that is provided as part of the user c...
- 01:28 AM Revision a97aca74 (ceph): rados.py: use uint64_t for auids
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 01:28 AM Revision 2f1720d0 (ceph): librados: return int64_t pool ids
- 468e28ee60ee2fe625d2680c792a4bcb9ef19951 missed the get_id() functions.
Signed-off-by: Josh Durgin <josh.durgin@drea...
01/03/2012
- 10:08 PM Revision 8e56e99c (ceph): radosgw-admin: add eol following info
- 10:07 PM Revision ec3a3a96 (ceph): rados: fix example config
- 09:55 PM Revision 71d5bcbb (ceph): Adjust rados model workloads for new config format
- 09:10 PM Revision ec6530df (ceph): RadosModel: make object write ranges configurable
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 09:10 PM Revision f03e770f (ceph): RadosModel: allow TestOps to pass data to their finish methods
- This will allow nested writes to keep track of which write actually
completed. Also remove finish() and _finish() fr... - 09:10 PM Revision 91124051 (ceph): RadosModel: check for out of order replies within WriteOps
- A single WriteOp already does multiple aio_writes. Each aio_write
gets a unique tid that is checked upon completion. ... - 09:10 PM Revision 0e470c50 (ceph): testrados: replace testreadwrite and testsnaps with testrados
- testrados can act as testreadwrite or testsnaps by changing the
command line options for the weight of each operation... - 09:02 PM Revision 0176c9ab (ceph): Remove unused mon.0 variables.
- 09:02 PM Revision cdd5c456 (ceph): nuke-on-error: only unlock if this run locked the machines
- 09:02 PM Revision 2e9b1c75 (ceph): rados: use testrados instead of testsnaps and testreadwrite
- 07:37 PM Revision a66d90e3 (ceph): osd: add a monitor timeout via MPGStatsAck messages
- Keep track of when we have outstanding updates, and while we do, make
sure the monitor responds within a timeout (def... - 05:55 PM Feature #1863 (Resolved): qa: tester for osd op reply order
- 05:13 PM Bug #1873 (Won't Fix): crush_rule type is inconsistent
- Here's a table of crush_rule's type in various places:...
- 04:49 PM rgw Feature #1872: rgw: only use shadow objects for large objects
- Once a race has been detected, operation needs to be restarted (unless we already have all requested data).
- 04:29 PM rgw Feature #1872: rgw: only use shadow objects for large objects
- This will also require being careful to check both current and new sizes — the new object might be < chunk size while...
- 04:23 PM rgw Feature #1872 (Resolved): rgw: only use shadow objects for large objects
- Currently we use shadow objects for every write that overwrites an object. We can avoid that by assuming that objects...
- 04:07 PM Feature #1836: filejournal: use async directio to write to the journal
- Presumably AIO writes can be combined or reordered by the block device/interfaces, right? So having a bunch of them i...
- 09:27 AM Feature #1836: filejournal: use async directio to write to the journal
- Say we have a cap of N aio ops, to prevent a stream of small ops resulting in a stream of tiny aio writes to the jour...
- 04:04 PM Revision f4b0cda1 (ceph): Fix invalid docdir_SCRIPTS usage with >=automake-1.11.2
- 04:04 PM Bug #1865 (Duplicate): mon: need to disconnect clients when we drop out of quorum
- Adding active ping requirements to the monitors is contrary to the direction we want to take them with clients, thoug...
- 04:02 PM Bug #1841 (Resolved): OSDs should disconnect from Monitor before their MOSDPGStat timeouts happen
- Sage merged this into master.
- 03:43 PM CephFS Bug #1047: mds: crash on anchor table query
- This log is less illuminating than I'd hoped since it's just replay commits. :(
However, it did give me one idea t... - 02:33 PM Linux kernel client Bug #1866 (Duplicate): null pointer dereference after osd went down
- same as #1793
- 01:10 PM Linux kernel client Feature #1870: usedcache mount option
- maybe call it use_dcache? usedcache is easy to misread as 'used cache'.
- 11:14 AM Linux kernel client Feature #1870 (Resolved): usedcache mount option
- 12:16 PM rgw Bug #1867 (Resolved): rgw crash
- should be fixed by commit:68ec8d8ee900642cdb594c67b7d2c416d55bec80.
- 11:46 AM Linux kernel client Bug #1871 (Resolved): ceph_client: crash after running xfstests 002
- Running xfstests under UML against a ceph filesystem, I get a
client crash due to dereferencing a null pointer in ce... - 11:00 AM Linux kernel client Bug #1767 (Resolved): osd_client: send_request() cannot fail
- commit:24e08cacc999503a069003364565116c923342b9
- 10:53 AM Linux kernel client Bug #1762 (Resolved): i_lock vs s_cap_lock vs inodes_with_caps_lock lock ordering
- merged into mainline for 3.2
- 10:47 AM Linux kernel client Bug #1652 (Resolved): rbd: rollback correctly after resizing
- we removed rollback functionality from the kernel.
- 10:42 AM Linux kernel client Bug #1812 (Resolved): iput scheduling while atomic
- 10:42 AM Linux kernel client Bug #1812: iput scheduling while atomic
- fixed by commit:aab26905f6a1df8e6a14f41a32a93b9af0c8b7c5
- 09:16 AM Bug #1868 (Resolved): Ceph client crashed after shutting down one mds and osd
- This bug was fixed by commit:935b639a049053d0ccbcf7422f2f9cd221642f58 in v3.1.
You should have better luck with th... - 08:37 AM Bug #1869 (Resolved): automake fails with >=automake-1.11.2 due to "docdir" usage
- applied, commit:f4b0cda17875c27d8b945be6cf5db9b356bb2dab
Thanks! - 07:40 AM Bug #1869 (Resolved): automake fails with >=automake-1.11.2 due to "docdir" usage
- doc_SCRIPTS is no longer allowed in >=automake-1.11.2 [1], as a result ceph fails to configure:
configure.ac:40: A... - 08:34 AM Bug #1759: mds/client: truncate size overflow, fails with EINVAL
- David McBride wrote:
> Unfortunately, my Ceph cluster which was presenting these symptoms has now died -- the remain... - 07:45 AM Bug #1759: mds/client: truncate size overflow, fails with EINVAL
- Unfortunately, my Ceph cluster which was presenting these symptoms has now died -- the remaining set of OSDs didn't h...
01/02/2012
- 08:01 PM CephFS Bug #1682: mds: segfault in CInode::authority
- This probably isn't all that useful for anyone who knows the code well, but I threw together a quick run down of plac...
- 02:24 PM Bug #1868 (Resolved): Ceph client crashed after shutting down one mds and osd
- Here is my cluster configuration before shutting down ceph components on n2cc (2.2.2.2)....
- 05:47 AM CephFS Bug #1047: mds: crash on anchor table query
- Here is what MDS logs with debug 20. No idea if it really helps. The cluster
is still in the broken state, should I...
12/31/2011
- 11:09 PM Revision f8929bad (ceph): osd: trigger RecoveryFinished event on recovery completion
- Unconditionally trigger the RecoveryFinished event when start_recvoery_ops
thinks it may be done. This lets us trigg... - 02:49 PM CephFS Bug #1774: client: files become inaccessible in large directories (with snapshots?)
- Oh, interesting! With a debug client = 20, debug ms = 1 log from ceph-fuse this should be pretty straightforward to ...
- 01:04 AM Revision a1252463 (ceph): librados: take lock in rollback
- We're poking through the osdmap; need to hold the lock here.
Signed-off-by: Sage Weil <sage@newdream.net> - 01:04 AM Revision 4c23e9e4 (ceph): objecter: assert lock held in op_submit
- Signed-off-by: Sage Weil <sage@newdream.net>
- 01:04 AM Revision 68ec8d8e (ceph): librados: call aio_operate() with lock held
- Signed-off-by: Sage Weil <sage@newdream.net>
- 12:45 AM Revision 0692bed8 (ceph): cmp: fix 5-uple operator==
- Doh!
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 12:45 AM Revision 1c754187 (ceph): osd: be a bit more verbose during backfill
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
12/30/2011
- 11:51 PM Revision 3dcaf6c3 (ceph): osd: do not backfill if any objects are missing on the primary
- Someday we need to do something smarter so that a single unfound object
doesn't hold up replication of other objects.... - 10:37 PM Revision cdf142b5 (ceph): rados: fix documentation format
- 10:37 PM Revision 6df4ce50 (ceph): rados: fix references to testrados
- 10:37 PM Revision 0af9c0a2 (ceph): rados: clean up argument construction
- Only the client id varies, so it can be done outside the loop. Also
handle coredumps and coverage, and use LD_LIBRARY... - 10:37 PM Revision 932257fb (ceph): rados: remove unused variable
- 10:37 PM Revision 2f71f03f (ceph): misc: simplify reconnect logic
- Ignore all errors until the timeout expires so we don't have to worry
about whitelisting them. - 10:31 PM Revision 949f24d5 (ceph): rgw: create default constructors for some structs
- this will silence valgrind a bit
- 08:23 PM Revision 251fc3d5 (ceph): osd: handle backfill_target for pick_newest_available
- The it may not be missing on the backfill_target if it is after the
last_backfill marker.
Signed-off-by: Sage Weil <... - 08:19 PM Revision a3525891 (ceph): osd: return EINVAL if multi op specified with no src object name
- This avoids crashing later in do_osd_ops() with something like
osd/ReplicatedPG.cc: In function 'int ReplicatedPG::d... - 07:39 PM Revision e686c1b6 (ceph): hobject_t: fix operator==, !=
- These weren't comparing key.
While we're at it, clean this up by using generic macros for writing
these operators, s... - 07:38 PM Revision 063ab2e4 (ceph): cmp.h: define macros for creating comparison operators
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 06:43 PM Revision 7969cc4f (ceph): Merge remote branch 'gh/master' into wip-backfill
- 06:32 PM Revision 6687ccf5 (ceph): workunits: update rbd test for new error format
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 05:50 PM Revision 85719b0e (ceph): config: use autoconf $libdir for default rados class dir
- Fixes: #1722
Signed-off-by: Sage Weil <sage@newdream.net> - 05:17 PM Revision 0d9507c2 (ceph): .gitignore: src/ocf/ceph
- Signed-off-by: Sage Weil <sage@newdream.net>
- 05:00 PM Revision 9b6422db (ceph): Spec: conditionally build ceph-resource-agents package
- Put OCF resource agents in a separate subpackage,
to be enabled with a separate build conditional
(--with ocf).
Make... - 05:00 PM Revision 92cfad42 (ceph): Add OCF-compliant resource agent for Ceph daemons
- Add a wrapper around the ceph init script that makes
MDS, OSD and MON configurable as Open Cluster Framework
(OCF) co... - 04:31 PM rgw Bug #1867: rgw crash
- This is the same assert as #1737. It may not be related, tho.. the bug may just be unlocked concurrent access to the...
- 04:16 PM rgw Bug #1867 (Resolved): rgw crash
- logs were turned off, so not much of that, but here's the backtrace. Happened on 7fc97e6 during a performance test (r...
- 04:06 PM Revision 66170633 (ceph): mon: fix full ratio updates
- - update them independently
- only if we are leader
- fix type for nearfull_ratio
Signed-off-by: Sage Weil <sage@new... - 04:06 PM Revision f2e41097 (ceph): mon: don't ignore first full ratio update callback
- We get a callack on startup. Don't ignore it.
Signed-off-by: Sage Weil <sage@newdream.net> - 03:45 PM Revision a693438e (ceph): mon: only update full_ratio if we're the leader
- Signed-off-by: Sage Weil <sage@newdream.net>
- 03:42 PM Revision 47d02271 (ceph): Merge remote branch 'gh/wip-cleanup'
- 03:28 PM Linux kernel client Bug #1793: NULL pointer dereference at try_write+0x627/0x1060
- looks like a null pointer dereference.. look for a struct member that's 0x48 bytes in?
- 02:59 PM Linux kernel client Bug #1793: NULL pointer dereference at try_write+0x627/0x1060
- This happened again with the same workload on sepia81.
- 10:35 AM CephFS Bug #1549: mds: zeroed root CDir* vtable in scatter_writebehind_finish
- hit this again, nightly_coverage_2011-12-29-b/5388...
- 09:59 AM Bug #1722 (Resolved): osd_class_dir must reflect autoconf libdir
- fixed by commit:85719b0ed81a38c3bd36c6be411f29181c969cda. duh
- 08:47 AM Bug #1865: mon: need to disconnect clients when we drop out of quorum
- the ceph-mon is deadlocked by...
- 08:44 AM Bug #1865: mon: need to disconnect clients when we drop out of quorum
- the kernel client is repeated reconnecting to a down osd and resendig its' requests because its osdmpa is out of date...
- 06:52 AM Bug #1862: filestore: EINVAL on replay
- Hello Sage,
Int64_t do the trick and now the osd are back online again!
Thank you
Marco
- 01:15 AM Revision df84594f (ceph): mon: make full ratio config change callback safe
- We can't propose_pending() from any context; do this in the tick() thread,
with the proper locking. Among other thin...
12/29/2011
- 11:43 PM Revision 585fb5ce (ceph): clitests: update for new error format
- This was changed in 1f434da8a3ca4db830d1f3b0d87e5df941d85f2d
Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com> - 11:28 PM Revision cec2692e (ceph): clitests: update monmaptool test
- e93961c11119942eae3a4cd14a79f779a5a4d277 changed output format.
Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com> - 09:09 PM Revision f04e2955 (ceph): teuthology rgw-admin: annotated test cases for inventory
- this is not a nose suite, so I simply added test case
descriptions in csv format, and put a file to extract
the... - 08:00 PM Revision 48df71c8 (ceph): init script: be LSB compliant for exit code on status
- An exit code of 1 on status is defined in LSB as
"program is dead, but pid file exists". Check for existence
of this ... - 07:58 PM Revision 3b2ca7cf (ceph): keyring: print more useful errors to log/err
- Signed-off-by: Sage Weil <sage@newdream.net>
- 07:57 PM Revision eba235f2 (ceph): common: trigger all observers on startup
- Among other things, this makes err-to-stderr and friends initialize
properly in the DoutStreamBuf.
Signed-off-by: Sa... - 07:24 PM Revision 1f434da8 (ceph): common: make cpp_strerror output prettier
- Signed-off-by: Sage Weil <sage@newdream.net>
- 07:24 PM Revision 04c8db00 (ceph): librados: check for monclient::init() error
- I think this fixes #1835.
Signed-off-by: Sage Weil <sage@newdream.net> - 05:59 PM Revision 1a59405c (ceph): rgw: turn on cache by default
- 05:59 PM Revision 37013b6f (ceph): qa: load-gen-mix-small.sh
- 05:41 PM Revision 959fd71f (ceph): osd: explicitly track leading edge of backfill
- backfill_pos is the leading edge; last_backfill is the trailing edge.
Anything inbetween is either pushed, doesn't ex... - 05:31 PM CephFS Bug #1682: mds: segfault in CInode::authority
- Not sure if this is the same underlying problem, but here's another CInode::authority crash from teuthology:~teuthwor...
- 05:09 PM Revision d24ea235 (ceph): mds: assert if we get an EINVAL on our truncate
- Signed-off-by: Sage Weil <sage@newdream.net>
- 05:00 PM Revision 47013c28 (ceph): osd: get fsid from monmap, not osdmap
- We may not have a valid OSDMap in all of these cases (notably, during
boot). Always take the fsid from the monmap, w... - 04:59 PM Revision 05cc4eb9 (ceph): monc: get latest monmap during authentication
- Tell the monitor which monmap version we have in our initial auth message.
Make the monitor send the latest monmap if... - 04:44 PM Revision 5d5c9b6f (ceph): osdmap: add const markers to some unfixed functions
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 04:44 PM Revision 300c7584 (ceph): osd: catch authenticate error on startup
- Signed-off-by: Sage Weil <sage@newdream.net>
- 03:30 PM Linux kernel client Bug #1866 (Duplicate): null pointer dereference after osd went down
- This was during a kernel_untar_build workunit on rbd:...
- 12:15 PM Bug #1835: Monclient crash when keyring is not readable
- should be fixed by commit:04c8db001a4ed02ef7335ed01ce73ce9ab28dc9d .. can you verify, Wido?
- 11:16 AM Feature #1863 (In Progress): qa: tester for osd op reply order
- 11:14 AM Bug #1865 (Duplicate): mon: need to disconnect clients when we drop out of quorum
- From sepia4:/tmp/cephtest/archive/log/osd.0.log:...
- 10:59 AM CephFS Bug #1549: mds: zeroed root CDir* vtable in scatter_writebehind_finish
- Happened again in teuthology:~teuthworker/archive/nightly_coverage_2011-12-29-a/5318/remote/ubuntu@sepia60.ceph.dream...
- 10:33 AM Bug #1490: cfuse assert failure: assert(ob->last_commit_tid < tid)
- Happened again in teuthology:~teuthworker/archive/nightly_coverage_2011-12-28-b/5258/teuthology.log.
- 10:04 AM Bug #1862: filestore: EINVAL on replay
- Marco Aroldi wrote:
> Hmmm
> I have another problem: i've tried the patch in #1759 but I have a error at compile ti... - 10:01 AM Bug #1862: filestore: EINVAL on replay
- Hmmm
I have another problem: i've tried the patch in #1759 but I have a error at compile time:
CXX libos_l... - 08:33 AM Bug #1862 (Duplicate): filestore: EINVAL on replay
- Aha, this is actually #1759. If you apply the patch in that bug report it'll get your OSDs up and running again. Th...
- 03:33 AM Bug #1862: filestore: EINVAL on replay
- Hi,
I've downloaded and compiled the latest code from the git repository.
I've issued a "ceph-osd -i 1 --debug_ms 2... - 09:53 AM Bug #1804: filestore: unexpected EINVAL
- My money is that this is caused by #1759. Which hopefully means that the qa suite will eventually trigger the new as...
- 09:17 AM Bug #1741 (Can't reproduce): teuthology: failed to untar
- 09:16 AM Bug #1759 (Need More Info): mds/client: truncate size overflow, fails with EINVAL
- The OSD now returns EINVAL, the MDS asserts if it gets EINVAL, and we have some MDS-side assertions that should catch...
- 09:10 AM Bug #1846 (Resolved): Mds crash immediately after start (segmentation fault)
- Great!
- 06:05 AM Bug #1846: Mds crash immediately after start (segmentation fault)
- I have built debian package from master branch and upgraded ceph on both servers. Mds and osd started properly. Thank...
- 09:08 AM Bug #1848 (Resolved): osd got zeroed out fsid
- 09:08 AM Bug #1848: osd got zeroed out fsid
- fixed by commit:47013c289e6ad6638b0f77152dafbc9f4723c032 and commit:05cc4eb93ce6d193c6aea4918144006fb4d1c187
- 01:00 AM Revision e18b1c97 (ceph): rgw: removing swift user index when removing user
- 12:50 AM Revision 997e35ae (ceph): rgw-admin: remove subuser index when required
- 12:42 AM Revision 1f40031f (ceph): osd: fix push completion check
- Only check backfill if we pushed to the backfill target. And avoid teh hash
lookup in the general case.
Signed-off-... - 12:34 AM Revision 2dc90d03 (ceph): rgw: clone operation should only update index for main category
- 12:33 AM Revision bb52b187 (ceph): rgw: fix cache interface (was not overloading method)
12/28/2011
- 11:10 PM Revision 0db9a423 (ceph): rgw: fix bucket creation
- 06:43 PM Feature #1709 (Resolved): specfile: merge suse spec file changes
- 06:42 PM Feature #1678 (Resolved): rados tool: ability to specify object locator
- 06:42 PM Bug #1683 (Resolved): librados: list objects should also return locator key
- 06:41 PM Bug #1508 (Can't reproduce): iozone stuck on kernel rbd mount
- haven't seen this recently
- 05:06 PM Bug #1848: osd got zeroed out fsid
- Happened again in teuthology:~teuthworker/archive/nightly_coverage_2011-12-28/5238/remote/ubuntu@sepia56.ceph.dreamho...
- 05:05 PM rgw Bug #1854 (Resolved): Deletion of an rgw user that has a subuser with a swift key leaves behind a...
- Fixed, commit:e18b1c9734e88e3b779ba2d70cdd54f8fb94743d.
- 03:26 PM Bug #1846: Mds crash immediately after start (segmentation fault)
- Oh, I think Henry sent in a fix for this. Can you apply commit:bfbeae68c045de76ede86ca4f72d2a760a19c84b (or use late...
- 02:45 PM Bug #1846: Mds crash immediately after start (segmentation fault)
- ...
- 08:12 AM Bug #1846: Mds crash immediately after start (segmentation fault)
- I looked at the attached monmap and didn't see anything odd. This fully reproducible, I take it? That's good news.
... - 02:35 PM rgw Bug #1864 (Resolved): rgw: atomic bucket info
- 01:46 PM Bug #1828 (Resolved): osd: preserve write order when ops wait for recovery of src_oids
- 01:45 PM Feature #1863 (Resolved): qa: tester for osd op reply order
- Out of order osd replies current trigger an ObjectCacher assert. Presumably there are lots more of them than the one...
- 09:14 AM Bug #1862: filestore: EINVAL on replay
- Hi Sage,
I'm sorry but I don't understand the steps requested.
Please, could you explain a little bit more? - 08:19 AM Bug #1862 (Need More Info): filestore: EINVAL on replay
- Can you try running the latest master code and restart ceph-osd? Specifically, commit:7133a2faf0ae0710b7cbd9801c6476...
- 07:25 AM Bug #1862 (Duplicate): filestore: EINVAL on replay
- Hello,
I'm testing ceph 0.39 on two VM (Ubuntu 11.10) on Hyper-V with all Linux Integration Components installed.
2...
12/27/2011
- 03:18 PM Bug #1846: Mds crash immediately after start (segmentation fault)
- Do you have any suggestions how to temporary workaround this problem?
12/23/2011
- 08:47 PM Revision 4ac04e89 (ceph): rgw: write bucket info in one operation
- 05:56 PM Revision 60bbf688 (ceph): Objecter: fix local reads one more time.
- Document it a little since we've gotten it wrong so often.
Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com> - 04:59 PM Bug #1841: OSDs should disconnect from Monitor before their MOSDPGStat timeouts happen
- Pushed a wip-osd-mon-communication branch that implements this. It's untested, though!
- 02:50 PM RADOS Feature #1861 (New): qa: test OSD handling of misdirected operations
- This will probably require some new facilities in a client in order to do automatically, but we had a regression (I t...
- 02:48 PM Messengers Bug #1803 (New): msgr: behave better when ending TCP connections
- 02:47 PM Bug #1858: OSD needs to check for misdirected ops before putting non-existent PGs on hold
- Pushed the branch osd-misdirected-checks. So far it's untested.
- 07:56 AM Bug #1858: OSD needs to check for misdirected ops before putting non-existent PGs on hold
- we should also include a read-from-replicas workload in the qa suite.. probably combined with osd thrashing. that ma...
- 10:56 AM rados-java Feature #1860 (New): qa: write tests for local reads and random replica reads
- We need to test the client options that randomize the read load and attempt to read from local replicas. The local re...
- 10:07 AM rgw Bug #1859 (Resolved): rgw: bucket creation is not atomic
- the backing bucket object is being created in two opearions: create and write. We need to combine these into a single...
- 01:41 AM Revision eb37637f (ceph): Merge remote branch 'upstream/master' into wip-backfill
- Conflicts:
src/include/object.h
Signed-off-by: Samuel Just <samuel.just@dreamhost.com> - 01:38 AM Revision 1f02e34c (ceph): ReplicatedPG: objects currently begin backfilled are degraded
- pending_stat_updates has also been renamed to pending_backfill_updates.
Signed-off-by: Samuel Just <samuel.just@drea... - 01:24 AM Revision d2eb119a (ceph): ReplicatedPG: fill in backfill_peer in on_activate
- Previously, there was a race between issue_repop/do_op and
start_recovery_ops.
Signed-off-by: Samuel Just <samuel.ju... - 01:17 AM Revision 517ddf84 (ceph): ReplicatedPG: only pull in one backfill peer at a time
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
12/22/2011
- 11:25 PM Revision 855e93b6 (ceph): filestore: fix config observer
- Actually, I don't think this was fully implemented to begin with, so it's
not a 'fix' per se. This will let you use ... - 11:18 PM Revision decdc363 (ceph): MOSDPGRepScrub: Fix typo in MOSDPGRepScrub
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 10:33 PM Revision db04d680 (ceph): ReplicatedPG: update last_backfill when pushes complete
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 10:33 PM Revision 3b90df0d (ceph): ReplicatedPG: init backfill infos to last_backfill
- We can scan starting from last_backfill to avoid rescanning portions
of the collection recovered by normal recovery. ... - 10:33 PM Revision 298b1349 (ceph): PG: backfill info should be cleared on recovery reset
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 10:33 PM Revision 8b8aab84 (ceph): PG: update stats from master only if not backfilling
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 10:33 PM Revision fa6bd38a (ceph): ReplicatedPG: simplify recover_backfill
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 10:00 PM Revision 9fc060c0 (ceph): Merge branch 'wip-signal'
- 09:48 PM Bug #1858 (Resolved): OSD needs to check for misdirected ops before putting non-existent PGs on hold
- Right now it doesn't. If it had, then diagnosing Noah's local reads problem would have been much, much faster. :(
- 08:33 PM Revision c7fee72d (ceph): MOSDRepScrub: use header.version for payload version
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 08:16 PM Revision 3cb53cc9 (ceph): Merge branch 'stable'
- 08:15 PM Revision e93961c1 (ceph): monmap: iterate over addr_name when printing summary
- The rank is now ordered by IP address. We should iterate over
addr_name.
Signed-off-by: Henry C Chang <henry.cy.chan... - 08:15 PM Revision bfbeae68 (ceph): monmap: clear addr_name map on calculating ranks
- We should clear addr_name before filling it. Otherwise, the removed
mon will stay there and cause incorrect rank assi... - 08:15 PM Revision ea9f2f62 (ceph): interval_set: fix truncation of _size
- _size is type of int64_t. Use int to store the value of _size
will cause value truncation.
Signed-off-by: Henry C Ch... - 04:28 PM CephFS Bug #1047: mds: crash on anchor table query
- Unfortunately there's not enough info in this log either. We're going to need a log with (at minimum) level 10 mds de...
- 04:27 PM Bug #1850 (Duplicate): mds sometimes crashes removing trees with plenty of hardlinks
- I'm pretty sure you're looking at #1047 here. :)
- 12:45 PM Fix #1857 (Resolved): osd: reimplement shutdown()
- on sigterm, go through OSD::shutdown() and try to clean things up in an orderly fashion. This will be useful for lea...
- 11:02 AM rgw Bug #1856 (Resolved): It is possible to look up an rgw user by a subuser that does not exist as l...
- 11:00 AM rgw Bug #1855 (Resolved): Creation of a subuser that appears to own an s3 key is possible, and removi...
- 10:58 AM rgw Bug #1854 (Resolved): Deletion of an rgw user that has a subuser with a swift key leaves behind a...
- Creating an rgw user with a subuser and swift key and then deleting the user appears to orphan the object for that su...
Also available in: Atom