Activity
From 09/23/2012 to 10/22/2012
10/22/2012
- 02:28 PM Bug #1947: mds: SIGBUS during _mark_dirty
- Recent logs: ubuntu@teuthology:/a/teuthology-2012-10-22_00:00:20-regression-next-testing-basic/5426...
10/21/2012
- 02:53 PM Bug #3301 (Resolved): cfuse: chmod not getting respected
- 02:25 PM Bug #1527 (Fix Under Review): mds: assert in handle_cache_expire
- reproduced this with a log. the problem is the preceding conditional not including the EXPORTING stage; testing the ...
- 09:25 AM Bug #3371: java: ldout after ceph_shutdown causes segfault
- 09:25 AM Feature #3368: java: add O_WRONLY open flag
- 09:24 AM Feature #3374: java: add unit tests for CephMount use after shutdown
- 09:24 AM Bug #3375: java: shutdown() doesn't clean-up if unmounted
- 09:21 AM Feature #3372: java: add finalize() for users that don't do explicit clean-up
- 09:19 AM Bug #3373: java: after shutdown() access ceph_* in JNI is not prevented
10/20/2012
- 05:38 PM Bug #3375 (Won't Fix): java: shutdown() doesn't clean-up if unmounted
- shutdown() throws CephNotMountedException if CephMount is not mounted. This means that the native memory allocated by...
10/19/2012
- 05:09 PM Feature #3374 (Rejected): java: add unit tests for CephMount use after shutdown
- There is one unit test per function in CephMount. Pre-test setup is a new CephMount object followed by shutdown(). Ca...
- 04:56 PM Bug #3373: java: after shutdown() access ceph_* in JNI is not prevented
- An alternative semantics which is more natural is to return CephMount to its initial state by performing a shutdown o...
- 04:28 PM Bug #3373 (Won't Fix): java: after shutdown() access ceph_* in JNI is not prevented
- When CephMount is created, a mount context is created to back it. Checks in libcephfs return errors for calls when us...
- 03:53 PM Feature #3372 (Closed): java: add finalize() for users that don't do explicit clean-up
- The JVM will run finalize() before GC, but is not guaranteed to ever do so. Still, it is good to do this in case user...
- 03:37 PM Bug #3371 (Won't Fix): java: ldout after ceph_shutdown causes segfault
- #define dout_subsys ceph_subsys_javaclient
...
ceph_shutdown
ldout(cct, 10) <- blows up - 03:25 PM Bug #3370 (Resolved): All nfsd hung trying to lock page(s) on export of kclient ceph
- Workunit bonnie hung over NFS client with retransmitted NFS read:
ubuntu 2667 2572 0 Oct18 ? 00:00:00 ... - 03:13 PM Bug #3369: journaled two client session close events
- 3369-mds-session-workaround for a bypass. Not sure if we want to merge this or not since it's just a bandaid!
It's ba... - 02:05 PM Bug #3369: journaled two client session close events
- Dur, forgot the file.
- 02:05 PM Bug #3369 (Resolved): journaled two client session close events
- Unfortunately there are no logs of the event occurring, but an MDS crash on startup was reported on the mailing list....
- 01:50 PM Bug #3346 (Resolved): client: we broken open permissions
- 12:16 PM Feature #3368 (Closed): java: add O_WRONLY open flag
- 10:25 AM Bug #3348 (Resolved): java: use libcephfs naming rather than generic ceph
- 10:24 AM Bug #3348: java: use libcephfs naming rather than generic ceph
- Now builds packages:
* libcephfs-java
* libcephfs-jni - 09:36 AM Bug #3367 (Resolved): client: hardlink without original file breaks on unmount
- I'm not able to reproduce this on fuse, but with the libcephfs client, the following:
mkdir dir1
cd dir1
create ...
10/18/2012
- 10:03 AM Bug #3348 (Resolved): java: use libcephfs naming rather than generic ceph
- 08:09 AM Bug #3346: client: we broken open permissions
- I didn't fully understand the (subtle) semantics of open() with O_CREAT and the mode. It expects the file to first b...
- 01:13 AM Bug #3346: client: we broken open permissions
- first two:...
- 12:50 AM Bug #3346 (Resolved): client: we broken open permissions
- ...
10/17/2012
- 06:10 PM Bug #3339 (Resolved): libcephfs: failure on Dir_ls test
- test was broken.
- 10:34 AM Bug #3339 (Resolved): libcephfs: failure on Dir_ls test
- ...
10/16/2012
- 04:20 PM Feature #3338 (Resolved): java: add unit tests for link/unlink/readlink/symlink
- 04:14 PM Bug #3337 (Rejected): java: unit tests don't remove container directory after a failure
- Java unit tests construct a container directory in which lots of temporary files are created and destroyed. The conta...
- 02:56 PM Documentation #3335 (Rejected): doc: Explain kernel dynamic printk debugging
- Under doc/dev/
http://thread.gmane.org/gmane.comp.file-systems.ceph.devel/1809/focus=1810 - 01:03 PM Bug #2681 (Resolved): client: got push without mds session
- commit:db976663a50ce7ab2448c42cb0c6bfd4f58d25a5
- 10:48 AM Bug #3301: cfuse: chmod not getting respected
This time the logs are from doing the following:
root@scout:/tmp/cmnt# echo foo > foo
root@scout:/tmp/cmnt# chm...- 10:19 AM Bug #3301: cfuse: chmod not getting respected
- This log is of the client with only the following operations performed:
echo baz > baz
chmod 400 baz
chmod 644 b... - 09:56 AM Bug #3301: cfuse: chmod not getting respected
- Another issue:
chmod 400 a11
echo "foo" >> a11
< now fails correctly >
chmod 644 a11
ls -la a11
-r-------- 1 ... - 07:37 AM Bug #3318 (Rejected): java: lock access to CephStat, CephStatVFS from native
- Native code needs to synchronize updates to Java structures with the runtime.
- 06:19 AM Feature #3315 (Resolved): client: Add acl support
Add acl support to the client. This will probably require a library to parse the acl extended attributes, and mods...- 06:15 AM Feature #3314 (Resolved): client: client interfaces should take a set of group ids
Right now each of the client interfaces takes only uid and gid parameters. Extend the interface to take a set of g...
10/15/2012
- 09:55 AM Bug #3301: cfuse: chmod not getting respected
- yeah, apparently fuse isn't doing the check itself... :/ we should enforce this in Client::_open(), methinks. after...
- 07:17 AM Bug #3301 (Resolved): cfuse: chmod not getting respected
- echo "foo" > foo
chmod 400 foo
echo "bar" >> foo
Should fail with EPERM, but it succeeds.
Attached is a clien...
10/11/2012
- 01:21 PM Feature #3289 (Resolved): ceph-fuse: somehow exert pressure on the VFS to remove dentries from th...
- We think this is the cause of some ballooning caches in ceph-fuse and the MDS that have been reported. The VFS will h...
10/10/2012
10/09/2012
- 06:27 PM Feature #3245 (Fix Under Review): qa: test knfsd reexport in teuthology
- Pushed to the wip-nfs-dzafman branch of the teuthology repo.
- 01:36 PM Feature #3245 (In Progress): qa: test knfsd reexport in teuthology
- 01:36 PM Feature #3242 (In Progress): samba: push plugin upstream
- 09:57 AM Feature #3248 (Resolved): libcephfs: move testceph.cc tests into gtest framework
- 09:10 AM Feature #3248 (Fix Under Review): libcephfs: move testceph.cc tests into gtest framework
- The wip-gtest-libcephfs branch contains the libcephfs tests now part of the gtest framework
- 09:13 AM Bug #2285 (Resolved): libcephfs: failure with empty name components
- Fix merged to master
10/05/2012
- 08:20 AM Bug #3267: Multiple active MDSes stall when listing freshly created files
- Greg Farnum wrote:
> I'll try and take a look at this, but multi-MDS setups are known to be pretty unstable at this ...
10/04/2012
- 01:03 PM Bug #3267: Multiple active MDSes stall when listing freshly created files
- I'll try and take a look at this, but multi-MDS setups are known to be pretty unstable at this point. Have you tried ...
- 11:54 AM Bug #3267 (Closed): Multiple active MDSes stall when listing freshly created files
- The output from ceph-debugpack can be found at the following location: [[http://cumulonim.biz/mds.tar.gz]] We were ru...
- 10:52 AM Bug #3266 (Resolved): "ceph mds tell 0 dumpcache /etc/passwd" is not cool
- This might have plenty of security implications.
- 10:33 AM Feature #3245: qa: test knfsd reexport in teuthology
- * knfsd.py task to export an existing client
* nfs.py task to mount another client
so you can do something like
...
10/03/2012
- 03:57 PM Bug #3261: mds crashes in EMetaBlob::replay
- aww. i had debug ms = 20 in my ceph.conf. sorry.
the new one is attached - 03:49 PM Bug #3261 (Need More Info): mds crashes in EMetaBlob::replay
- can you put 'debug mds = 20' in the ceph.conf, restart ceph-mds, and then attach the resulting log (assuming it crash...
- 03:46 PM Bug #3261 (Rejected): mds crashes in EMetaBlob::replay
- while testing cephfs using the debian wheezy packages on a fairly large volume (2TB) i ran into random unreproduceabl...
- 05:09 AM Bug #3254 (Resolved): mds: Replica inode's parent snaprealms are not open
- When an inode with snaprealm get replicated, its parent snaprealms are not open on the dest MDS. This is problematic ...
10/02/2012
10/01/2012
- 09:51 PM Feature #3248 (Resolved): libcephfs: move testceph.cc tests into gtest framework
- 12:48 PM Feature #3245 (Resolved): qa: test knfsd reexport in teuthology
- 12:48 PM Feature #3244 (In Progress): qa: integrate Ganesha into teuthology testing to regularly exercise ...
- 12:48 PM Feature #3243 (Resolved): qa: test samba reexport via libcephfs vfs plugin in teuthology
- 12:47 PM Feature #3242 (Resolved): samba: push plugin upstream
09/27/2012
- 05:33 PM Fix #2215: ceph-fuse does not invalidate page cache
- 02:38 PM Bug #3184 (Resolved): ceph-fuse should not segfault without a keyring
09/26/2012
- 05:06 PM Bug #3186 (Resolved): mds: mds crash on client shutdown
- 08:25 AM Bug #1677 (Resolved): mds interval_set.h: 385: FAILED assert(p->first <= start)
09/25/2012
- 06:24 PM Bug #2285 (Fix Under Review): libcephfs: failure with empty name components
- This was probably caused by the memory error in filepath (encoded member not getting initialized), which may explain ...
- 06:22 PM Bug #3186 (Fix Under Review): mds: mds crash on client shutdown
- Proposed fix in wip-3186, specifically c92e1dd7b0106fa986985d8179a8ed66c61e6709. Needs review.
- 06:20 PM Bug #3184 (Fix Under Review): ceph-fuse should not segfault without a keyring
09/24/2012
- 05:57 PM Bug #3210 (Resolved): mds crashed and segfault at unlink_local_finish
- Logs: /a/teuthology-2012-09-23_19:00:07-regression-master-testing-gcov/28428...
- 10:00 AM Bug #3184: ceph-fuse should not segfault without a keyring
- client->init() wasn't getting checked for error. Proposed fix in wip-3184 branch.
Also available in: Atom