Project

General

Profile

Activity

From 02/03/2015 to 03/04/2015

03/04/2015

05:20 PM Bug #10982 (Pending Backport): ceph-fuse: remount for dentry trim can fail during unmount
Zheng merged the initial PR; I think I need to backport it. Greg Farnum
03:04 AM Bug #11011 (Resolved): Client doesn't progress after truncate until next tick
Zheng Yan

03/03/2015

08:15 PM Bug #11011 (Fix Under Review): Client doesn't progress after truncate until next tick
https://github.com/ceph/ceph/pull/3865 John Spray
07:59 PM Bug #11011: Client doesn't progress after truncate until next tick
the trouble is that do_open_truncate is giving an early reply, so journal_and_reply isn't bothering to call mdlog->fl... John Spray
07:56 PM Bug #11011 (Resolved): Client doesn't progress after truncate until next tick

Noticed this while trying to reproduce something else: mount a fuse client, and run:...
John Spray
12:42 PM Bug #4280 (Resolved): mds: crash on lookupsnap
fixed by 335f02bf7dde417350dec9cef9c3e6bdb3cfeab5 Zheng Yan
07:05 AM Feature #10504 (Resolved): kclient: include client version in client_metadata
Zheng Yan

03/02/2015

10:01 PM Bug #10997 (Can't reproduce): teuthology tests are using master "cephfs" tool on firefly branch?
... Greg Farnum
09:48 PM Bug #10996 (Can't reproduce): dumpling MDS: failed MDLog assert
This backtrace turned up on the internal lab
http://pulpito.ceph.redhat.com/teuthology-2015-02-24_19:06:01-fs-dumpli...
Greg Farnum
09:22 PM Bug #10863: java.lang.UnsatisfiedLinkError: cephfs_jni (Not found in java.library.path) on RHEL7
They go on the same ticket, thus the "Pending Backport" state. :) Greg Farnum
09:19 PM Bug #10863: java.lang.UnsatisfiedLinkError: cephfs_jni (Not found in java.library.path) on RHEL7
Greg Farnum wrote:
> Noah's started working on the backports too.
I think this ticket is resolved since my last P...
Noah Watkins
07:22 PM Bug #10863: java.lang.UnsatisfiedLinkError: cephfs_jni (Not found in java.library.path) on RHEL7
Noah's started working on the backports too. Greg Farnum
06:29 PM Fix #10874: MDS: file recovery overwrites explicit client timestamps
Mmmm, logging on the local object might work. More likely we'd do it by setting a special xattr (I'm not sure the OSD... Greg Farnum
02:19 PM Fix #10874: MDS: file recovery overwrites explicit client timestamps
What if writes carried the timestamp that ctime and mtime in the object are about to be set to, and the osd set the o... Alexandre Oliva
12:53 PM Bug #10712 (Fix Under Review): TestFlush intermittent failure on scatter_writebehind event
https://github.com/ceph/ceph-qa-suite/pull/358 John Spray
12:52 PM Bug #10949 (Fix Under Review): TestJournalRepair.test_reset fails due to (expected) coredump
John Spray
12:52 PM Bug #10949: TestJournalRepair.test_reset fails due to (expected) coredump
https://github.com/ceph/ceph-qa-suite/pull/357 John Spray
12:43 PM Bug #10948 (Fix Under Review): test_network_death failure with "Another app is currently holding ...
https://github.com/ceph/ceph-qa-suite/pull/356 John Spray
11:11 AM Bug #10948: test_network_death failure with "Another app is currently holding the xtables lock"
Ah, okay, this is because Filesystem._one_or_all executes in parallel, and I was starting to run routinely with multp... John Spray
12:28 PM Bug #10217: old fuse should warn on flock
the failure is likely caused by running ceph-fuse on kernel < 3.15, kernels < 3.15 do not support fcntl.F_OFD_SETLK. Zheng Yan
05:55 AM Bug #10982 (Resolved): ceph-fuse: remount for dentry trim can fail during unmount
http://pulpito.ceph.com/teuthology-2015-02-26_23:04:03-fs-hammer-testing-basic-multi/780938/
http://pulpito.ceph.com...
Greg Farnum

02/28/2015

02:00 PM Bug #10217: old fuse should warn on flock
http://pulpito.ceph.com/loic-2015-02-28_11:45:50-fs-firefly-backports---basic-multi/783093/
http://pulpito.ceph.com/...
Loïc Dachary

02/27/2015

02:14 PM Bug #10834: SAMBA VFS module: Timestamps revert back to 01-01-1970
Okay, I've noticed that when I restart the samba process the timestamps are shown correctly as expected, however when... Dennis Kramer
02:05 PM Bug #10834: SAMBA VFS module: Timestamps revert back to 01-01-1970
Sorry, I was wrong.
Even with Samba v4.2.0rc5 the problem exists for "scenario 3" in my comment above.
Dennis Kramer
01:44 PM Bug #10960: Permission/Access errors when using the ceph_vfs from SAMBA (4.x)
Excellent, thank you!
It seems to be working fine now. I'll do some more testing.
Dennis Kramer
12:05 PM Bug #10960: Permission/Access errors when using the ceph_vfs from SAMBA (4.x)
The root cause is that our vfs module has no ACL related callbacks, so samba uses its default ACL callbacks. samba's ... Zheng Yan

02/26/2015

11:59 PM Feature #10872 (Resolved): backport POSIX lock support for ceph-fuse client to firefly
We're doing another quick point release and Sage told me I could just do this, so I have. Merged to firefly branch as... Greg Farnum
11:57 PM Bug #10710 (Resolved): fuse_ll warning
Greg Farnum
11:55 PM Bug #10710: fuse_ll warning
Merged into hammer branch as of commit:5f42b54ab15e77c983cb0d8ac81df383c395f6f0 Greg Farnum
11:55 PM Bug #10542: ceph-fuse cap trimming fails with: mount: only root can use "--options" option
This is at least no longer an issue on kernels older than 3.18 as we aren't doing remount with them as of commit:5f42... Greg Farnum
08:02 PM Bug #10863 (Pending Backport): java.lang.UnsatisfiedLinkError: cephfs_jni (Not found in java.libr...
Okay, I forget whether it was https://github.com/ceph/ceph/pull/3782 or https://github.com/ceph/ceph-qa-suite/pull/34... Greg Farnum
07:08 PM Bug #10960: Permission/Access errors when using the ceph_vfs from SAMBA (4.x)
The root directory is special in various ways (although generally less as time goes on) so it wouldn't surprise me to... Greg Farnum
03:48 PM Bug #10960 (Resolved): Permission/Access errors when using the ceph_vfs from SAMBA (4.x)
Tried both stable as the dev version of SAMBA (4.1.17 & 4.2.0-RC5).
Tested with a Windows 7 & 2008 client.
When I...
Dennis Kramer
06:15 PM Bug #10912 (Resolved): Stuck request during clientreplay
Merged to master by commit:363be2df46f63ddbe0e355af6835aab06b9866c2 and to hammer by commit:a54864a38842def1b4f139e13... Greg Farnum
04:28 PM Bug #10949: TestJournalRepair.test_reset fails due to (expected) coredump
John Spray
06:30 AM Bug #10835 (Resolved): SAMBA VFS module: free/total disk space return values
upstreamed by commit 2501afe08b94a514d8e2f3eeb4a2c4edc9764979 Zheng Yan

02/25/2015

07:23 PM Feature #10872 (Pending Backport): backport POSIX lock support for ceph-fuse client to firefly
It's passed a firefly suite run and been through review. Greg Farnum
07:08 PM Fix #10862 (Resolved): MDSMonitor: do not try and "deactivate" the last MDS
Greg Farnum
02:13 PM Fix #10862 (Fix Under Review): MDSMonitor: do not try and "deactivate" the last MDS
Backport PR here: https://github.com/ceph/ceph/pull/3794 John Spray
06:57 PM Bug #10944 (Resolved): Deadlock, MDS logs "slow request", getattr pAsLsXsFs failed to rdlock
Zheng Yan wrote:
> 3.10 kernel is too old for using cephfs.
Oh, I didn't notice the kernel version, d'oh.
The ...
Greg Farnum
11:28 AM Bug #10944: Deadlock, MDS logs "slow request", getattr pAsLsXsFs failed to rdlock
More experiment.
# create big testfile (30GB)
# pv testfile >testfile_2 (client host 1)
# pv testfile_2 >/dev/null...
Ilja Slepnev
06:09 AM Bug #10944: Deadlock, MDS logs "slow request", getattr pAsLsXsFs failed to rdlock
3.10 kernel is too old for using cephfs. Zheng Yan
06:39 PM Feature #10952 (Resolved): MDS: forward scrub: add functionality to verify disk data is consistent
Design and implement a mechanism that, given a CInode, it will go check the actual disk state of that inode for consi... Zhe Zhang
05:04 PM Bug #10950 (Resolved): Unable to remove MDS host: error handling
Recap:
I have this situation (Ceph 0.87 on Centos 7):
# ceph mds dump
dumped mdsmap epoch 84
epoch 84
flags 0
...
Gianfranco Sigrisi
04:25 PM Bug #10834: SAMBA VFS module: Timestamps revert back to 01-01-1970
Zheng, I've compiled your patches against Samba v4.2.0rc5 and it seems to work now for scenario 3 aswell. So far so g... Dennis Kramer
03:19 PM Bug #10834: SAMBA VFS module: Timestamps revert back to 01-01-1970
After some more testing, I've noticed it still doesn't work perfectly.
My setup is as follows for test purposes:
I ...
Dennis Kramer
12:55 PM Bug #10834: SAMBA VFS module: Timestamps revert back to 01-01-1970
Awesome Zheng! It seems to be working correctly now.
I've only used the patches 0000-, 0001- and 0002-
Thank you.
Dennis Kramer
11:30 AM Bug #10834: SAMBA VFS module: Timestamps revert back to 01-01-1970
Are the previous patches also needed? Dennis Kramer
11:07 AM Bug #10834: SAMBA VFS module: Timestamps revert back to 01-01-1970
they are for 4.3. please apply the prerequisite patch first. Zheng Yan
09:26 AM Bug #10834: SAMBA VFS module: Timestamps revert back to 01-01-1970
It won't patch. Hunk #1 failures.
Is it for SAMBA 4.1.16? I see a couple of discrepancies in the patches.
Dennis Kramer
03:49 AM Bug #10834: SAMBA VFS module: Timestamps revert back to 01-01-1970
please try the new patches Zheng Yan
02:24 PM Bug #10949 (Resolved): TestJournalRepair.test_reset fails due to (expected) coredump
http://pulpito.front.sepia.ceph.com/john-2015-02-25_11:53:35-fs-wip-jcsp-test-testing-basic-multi/777873/
This wor...
John Spray
12:10 PM Bug #10948 (Resolved): test_network_death failure with "Another app is currently holding the xtab...
http://qa-proxy.ceph.com/teuthology/john-2015-02-24_22:58:40-fs-wip-jcsp-test---basic-multi/777299/teuthology.log
...
John Spray
08:07 AM Bug #10912 (Fix Under Review): Stuck request during clientreplay
Zheng Yan

02/24/2015

11:18 PM Bug #10863: java.lang.UnsatisfiedLinkError: cephfs_jni (Not found in java.library.path) on RHEL7
Here Noah, catch! :) Greg Farnum
06:45 PM Bug #10863: java.lang.UnsatisfiedLinkError: cephfs_jni (Not found in java.library.path) on RHEL7
This persists as an issue; we really need to grab the time to look at it. :(
http://pulpito.ceph.com/teuthology-20...
Greg Farnum
11:18 PM Bug #10703: failures in libcephfs-java tests
Maybe we'll start seeing this again once we stop seeing the LinkErrors in #10863, but for now downgrade priority. Greg Farnum
11:15 PM Bug #10912: Stuck request during clientreplay
Can you upload these logs to the appropriate teuthology directory or something?
Also, we drop some caps automatica...
Greg Farnum
11:05 PM Bug #10944 (Need More Info): Deadlock, MDS logs "slow request", getattr pAsLsXsFs failed to rdlock
As it says, the MDS is failing to get a read lock on probably that inode, or maybe on of its ancestors. If you had hi... Greg Farnum
10:01 PM Bug #10944 (Resolved): Deadlock, MDS logs "slow request", getattr pAsLsXsFs failed to rdlock
CephFS path got stuck in directory listing process. No OSD/network activity. Only MDS logs. Client reboot does not he... Ilja Slepnev
07:17 PM Feature #10872: backport POSIX lock support for ceph-fuse client to firefly
Greg Farnum
06:26 AM Bug #10743 (Resolved): mds/MDLog.cc: 259: FAILED assert(!capped) on MDS rank shutdown
Greg Farnum
06:26 AM Fix #10862 (Pending Backport): MDSMonitor: do not try and "deactivate" the last MDS
This was a small patch and will make life simpler for Hammer, can you backport too please. Greg Farnum
06:22 AM Fix #10862 (Resolved): MDSMonitor: do not try and "deactivate" the last MDS
Greg Farnum

02/23/2015

04:30 PM Bug #10912: Stuck request during clientreplay
Hmm, so it's not the trim_cache_for_reconnect that's doing it, as that happens before we compose the reconnect messag... John Spray
03:19 PM Feature #10792: qa: enable thrasher for MDS cluster size (vary max_mds)
NB while adding cluster size tests, also check the mon commands e.g. #10862 (don't allow deactivate rank 0) John Spray
02:26 PM Fix #10862 (Fix Under Review): MDSMonitor: do not try and "deactivate" the last MDS
https://github.com/ceph/ceph/pull/3775 John Spray
01:51 PM Feature #10649 (Fix Under Review): Store MDS tables without size limit
John Spray
01:46 PM Bug #10712: TestFlush intermittent failure on scatter_writebehind event
Updating the test to tolerate this condition, as it's a valid outcome from flush John Spray
12:01 PM Bug #10834: SAMBA VFS module: Timestamps revert back to 01-01-1970
When I use the local filesystem or even cephfs with the kernel module (instead of ceph vfs) everything is working fin... Dennis Kramer
12:20 AM Bug #10834: SAMBA VFS module: Timestamps revert back to 01-01-1970
please check how samba behave when using local file system as backbend. I think this patch make ceph backend behave t... Zheng Yan
05:00 AM Fix #10874: MDS: file recovery overwrites explicit client timestamps
Transactions like that would be *much* more expensive than e.g. adding backtraces to file data objects, which I seem ... Greg Farnum
04:58 AM Bug #10915: client: hangs on umount if it had an MDS session evicted
Mmmm, that should be a pretty easy change MDS-side; I'm trying to figure out if it could get us in trouble though. An... Greg Farnum

02/20/2015

01:28 PM Bug #10834: SAMBA VFS module: Timestamps revert back to 01-01-1970
Replaced it and it builds, but the problem still exists.
The created time == modified time after editing.
Dennis Kramer
12:22 PM Bug #10834: SAMBA VFS module: Timestamps revert back to 01-01-1970
try replacing "lp_fake_directory_create_times" in the patch with "lp_fake_dir_create_times" Zheng Yan
11:21 AM Fix #10874: MDS: file recovery overwrites explicit client timestamps
I'd be happy enough if the mds wouldn't mess with timestamps of files of clients that are still up, and sync would ma... Alexandre Oliva

02/19/2015

10:41 AM Bug #10916 (New): ceph-fuse doesn't terminate properly if in the middle of a remount

This is probably related to some of the mysterious "device or resource busy" failures to umount in tests.
Just f...
John Spray
10:24 AM Bug #10915 (Resolved): client: hangs on umount if it had an MDS session evicted
Seen like this with fuse client:
* Start 2 active MDSs
* Do some activity such that sessions are open with both M...
John Spray

02/18/2015

08:08 PM Bug #10863: java.lang.UnsatisfiedLinkError: cephfs_jni (Not found in java.library.path) on RHEL7
http://pulpito.ceph.redhat.com/teuthology-2015-02-16_23:04:01-fs-giant-distro-basic-magna/48617/ Greg Farnum
07:17 PM Bug #10912: Stuck request during clientreplay
Could this be a result of the client dumping all caps it doesn't care about, and the MDS somehow thinking the client ... Greg Farnum
07:13 PM Bug #10912: Stuck request during clientreplay
... John Spray
07:10 PM Bug #10912: Stuck request during clientreplay
Logs at teuthology:~/jcsp/10912 John Spray
07:09 PM Bug #10912 (Resolved): Stuck request during clientreplay
Looks like a client bug, MDS has a stuck unlink which can't get the wrlock it wants -- it has sent a revoking MClient... John Spray
05:04 PM Bug #10834: SAMBA VFS module: Timestamps revert back to 01-01-1970
I will :-)
Thanks for the reply Greg!
Dennis Kramer
04:56 PM Bug #10834: SAMBA VFS module: Timestamps revert back to 01-01-1970
He's on vacation for Chinese New Year, and the rest of us are traveling. This'll probably require some work to get it... Greg Farnum
04:44 PM Bug #10834: SAMBA VFS module: Timestamps revert back to 01-01-1970
Hi Zheng, I was wondering if there is any progress with the above? Dennis Kramer

02/17/2015

06:36 PM Feature #1398 (In Progress): qa: multiclient file io test
I'll grab this again because I need to get a new wip branch out there. Anonymous
06:24 PM Feature #1398: qa: multiclient file io test
I have been waylaid by QE work. I'll get a wip branch pushed again. Anonymous
02:26 PM Bug #10744 (Fix Under Review): MDS gets stuck in 'stopping' when strays exist
https://github.com/ceph/ceph/pull/3752 John Spray
02:26 PM Bug #10743 (Fix Under Review): mds/MDLog.cc: 259: FAILED assert(!capped) on MDS rank shutdown
https://github.com/ceph/ceph/pull/3753 John Spray
11:35 AM Feature #10388 (Resolved): Add MDS perf counters for stray/purge status
... John Spray
11:08 AM Bug #10834: SAMBA VFS module: Timestamps revert back to 01-01-1970
It doesn't build after patching:... Dennis Kramer

02/16/2015

04:22 PM Bug #10876 (Duplicate): Cluster seems ok but CEPHFS is hanging
See #9458. This is due to a bug in the kernel CephFS client. (The critical line I used in tracing that is "failed to ... Greg Farnum
04:15 PM Fix #5268: mds: fix/clean up file size/mtime recovery code
From #10875:... Greg Farnum
04:11 PM Fix #10874: MDS: file recovery overwrites explicit client timestamps
I'm not quite sure the best way to go about fixing this. I think we might have discussed just dropping the mtime reco... Greg Farnum
01:58 PM Bug #10743 (In Progress): mds/MDLog.cc: 259: FAILED assert(!capped) on MDS rank shutdown
John Spray
12:50 PM Bug #10743: mds/MDLog.cc: 259: FAILED assert(!capped) on MDS rank shutdown
The trouble seems to be that rank 1 is sending lock messages to revoke rank 0's replica of ~mds1, then proceeding to ... John Spray

02/14/2015

07:15 AM Feature #10872: backport POSIX lock support for ceph-fuse client to firefly
Greg Farnum
02:39 AM Bug #10834: SAMBA VFS module: Timestamps revert back to 01-01-1970
most linux filesystems (extN,xfs,btrfs) do not have birth time. When dealing with linux local filesystem, samba fakes... Zheng Yan
12:24 AM Bug #10834: SAMBA VFS module: Timestamps revert back to 01-01-1970
I wonder how other Linux filesystems keep track of creation times. Are they just storing them in an xattr or somethin... Greg Farnum

02/13/2015

11:36 PM Bug #10834: SAMBA VFS module: Timestamps revert back to 01-01-1970
ceph inode does not store birth time. ceph vfs module assigns 'modified time' to 'birth time' Zheng Yan
02:49 PM Bug #10834: SAMBA VFS module: Timestamps revert back to 01-01-1970
Tried it. Patched successful. But not entirely fixed though, it seems that the timestamp "created timestamp" also cha... Dennis Kramer
01:50 PM Bug #10834: SAMBA VFS module: Timestamps revert back to 01-01-1970
it's for samba 4.3. please try the new one Zheng Yan
12:41 PM Bug #10834: SAMBA VFS module: Timestamps revert back to 01-01-1970
It won't patch on samba 4.1.16 (latest) source. Did you grab a different version? Dennis Kramer
12:18 PM Bug #10834: SAMBA VFS module: Timestamps revert back to 01-01-1970
could you try the attached patch Zheng Yan
11:18 PM Feature #10872 (In Progress): backport POSIX lock support for ceph-fuse client to firefly
Much to my surprise, a straightforward cherry-pick of the ceph-fuse flock branch to firefly does build after some mer... Greg Farnum
09:50 AM Feature #10872 (Resolved): backport POSIX lock support for ceph-fuse client to firefly
Description of problem:
In Firefly, the userspace version of the cephfs client does not support POSIX file locking. ...
Martin Schuppert
04:31 PM Bug #10876 (Duplicate): Cluster seems ok but CEPHFS is hanging
Hi,
I started some test running rsync to a cephfs mount using a cache layer, and at first the data was written, bu...
Kenneth Waegeman
02:33 PM Fix #10874 (New): MDS: file recovery overwrites explicit client timestamps
I think I've had an open bug about this for a very long time, but I couldn't find it, and I think it had been regarde... Alexandre Oliva
11:12 AM Bug #10835: SAMBA VFS module: free/total disk space return values
Awesome! It works.
Thank you.
Dennis Kramer
09:52 AM Bug #10835: SAMBA VFS module: free/total disk space return values
could you try the attached patch Zheng Yan

02/12/2015

04:54 PM Bug #10863 (Resolved): java.lang.UnsatisfiedLinkError: cephfs_jni (Not found in java.library.path...
http://pulpito.ceph.redhat.com/teuthology-2015-02-06_23:04:01-fs-giant-distro-basic-magna/42028/
http://pulpito.ceph...
Greg Farnum
04:23 PM Fix #10862 (Resolved): MDSMonitor: do not try and "deactivate" the last MDS
STATE_STOPPING makes the MDS try and export its data to another MDS, but for the final MDS there's no target for offl... Greg Farnum
03:19 PM Bug #10861 (Resolved): mysterious wrapped dd failure
Originally reported in #10465,
http://qa-proxy.ceph.com/teuthology/teuthology-2014-12-26_23:04:02-fs-master-testing-...
Greg Farnum
12:09 PM Feature #10649 (In Progress): Store MDS tables without size limit
John Spray
09:46 AM Bug #10835: SAMBA VFS module: free/total disk space return values
It's not a fix, its a workaround to get the smb.conf parameter "dfree command" working. It seems like if the ".disk_f... Dennis Kramer
06:42 AM Bug #10835: SAMBA VFS module: free/total disk space return values
remove which line of code of vfs_ceph.c can fix the issue? Zheng Yan
06:38 AM Bug #10542: ceph-fuse cap trimming fails with: mount: only root can use "--options" option
Ah, I managed to get it with the "-d" flag. I was under the apparently mistaken impression that if it ran as a proces... Greg Farnum
06:36 AM Bug #10703: failures in libcephfs-java tests
Argh, sorry for the bad pattern matching. I wonder if that's another thing not working properly on RHEL; I'll check i... Greg Farnum
06:29 AM Bug #10703: failures in libcephfs-java tests
It's not the same as previous one. All failure are caused by "java.lang.UnsatisfiedLinkError: cephfs_jni (Not found i... Zheng Yan
05:42 AM Bug #10703: failures in libcephfs-java tests
Those who have access can check out http://pulpito.ceph.redhat.com/teuthology-2015-02-06_23:04:01-fs-giant-distro-bas... Greg Farnum
04:16 AM Bug #10413 (Resolved): samba: coredumps after tests run
commit 81464daea71e5fa3067ec7d5f5c69c890c0f7949 smbd: Stop using vfs_Chdir after SMB_VFS_DISCONNECT" should fix this ... Zheng Yan

02/11/2015

10:40 PM Bug #10791 (Resolved): ceph mds deactivate sometimes fails silently
Greg Farnum
01:33 PM Bug #10542: ceph-fuse cap trimming fails with: mount: only root can use "--options" option
It comes out of stdout/stderr (i.e. not the ceph log). Was visible for me in a ./ceph-fuse -d terminal. John Spray
11:24 AM Bug #10835 (Resolved): SAMBA VFS module: free/total disk space return values
The return values for free/total disk space is not working correctly when you mount a share in Windows. My 340TB clus... Dennis Kramer
11:18 AM Bug #10834 (Closed): SAMBA VFS module: Timestamps revert back to 01-01-1970
I've used the ceph_vfs.c upstream module from samba v4.1.16 for our cephfs cluster, but i've noticed a bug where time... Dennis Kramer

02/10/2015

11:17 PM Bug #10791 (Fix Under Review): ceph mds deactivate sometimes fails silently
I think what we really want to do is have the state_seq go in both directions and force the MDS to acknowledge the tr... Greg Farnum
09:14 PM Bug #10791 (In Progress): ceph mds deactivate sometimes fails silently
Yep, that's a pretty clear race in the source. Greg Farnum
10:07 PM Bug #10277 (Resolved): ceph-fuse: Consistent pjd failure in getcwd
Loïc Dachary
07:33 PM Bug #10710 (Fix Under Review): fuse_ll warning
https://github.com/ceph/ceph/pull/3681
Needs better testing too.
Greg Farnum
05:13 PM Bug #10710 (In Progress): fuse_ll warning
Greg Farnum
06:16 PM Bug #10702 (Resolved): ceph-qa-suite: hung client-recovery task in nightlies
Sage will merge the branches as long as he's doing so for Ceph. Greg Farnum
04:59 PM Bug #10702 (Pending Backport): ceph-qa-suite: hung client-recovery task in nightlies
Merged to master in commit:0dce67d6bb83495b9d7f0c6cdfd9cd4bf193c749. We might also need to "backport" it to hammer, b... Greg Farnum
12:25 PM Bug #10702: ceph-qa-suite: hung client-recovery task in nightlies
Ah, having seen this fail in an interactive run, I now also realise that the /sys/fs/fuse/connections/*/abort' is bog... John Spray
05:47 PM Bug #10542: ceph-fuse cap trimming fails with: mount: only root can use "--options" option
I tested this on Fedora 20 (rex004) and am not seeing any warnings. Where did it show up for you? Greg Farnum
05:00 PM Feature #10826: MDSAuthCaps should have gids field as well as uid field
This may need to be more complicated and allow a list of UIDs as well — depending on deployment scenarios, it's easy ... Greg Farnum
02:47 PM Feature #10826 (Resolved): MDSAuthCaps should have gids field as well as uid field

(As sage pointed out last week)
Currently we have a place to put a UID limit (i.e. only allow client to act as t...
John Spray
02:39 PM Bug #10720 (Rejected): MDS: valgrind leaks
Dur, we aren't valgrind clean yet anyway...these failures are all due to the OSD failures associated with them. Greg Farnum
07:46 AM Bug #10740: teuthology: nfs test getting EBUSY on umount
http://qa-proxy.ceph.com/teuthology/teuthology-2015-02-08_23:10:02-knfs-hammer-testing-basic-multi/745286/ Greg Farnum
06:40 AM Bug #10465 (Resolved): Audit and fix ceph-qa-suite exec tasks
Master branch has been tested several times since this went in, so I backported it to giant,firefly,dumpling qa suites. Greg Farnum
06:38 AM Bug #10807 (Rejected): blogbench: hung MDS requests
Awesome. Greg Farnum
06:30 AM Bug #10807: blogbench: hung MDS requests
For some unknown reason, plana09 and plana47 couldn't reach burnupi40. Not cephfs bug. Zheng Yan
05:22 AM Bug #10417: snaptest-2.sh is failing
Hopefully fixed by the snapshot branch merge; it was popping up in many (all?) the nightly runs so we'll know if it's... Greg Farnum
03:46 AM Bug #10417 (Can't reproduce): snaptest-2.sh is failing
Zheng Yan
03:43 AM Bug #10344 (Resolved): qa/workunits/snaps/snaptest-git-ceph.sh fails
I forget the detail, but can't reproduce it anymore Zheng Yan
03:41 AM Bug #10315 (Resolved): set last snapid according to removed snaps in data pools
Zheng Yan
03:41 AM Bug #10312 (Resolved): creating snapshot makes parent snapshot lost
fix by 5e5052b3b016066b38e1ac2daccb9f70872a4c21 "mds: update current_parent_since for newly snapshotted directory" Zheng Yan
03:39 AM Bug #10343 (Resolved): qa/workunits/snaps/snaptest-xattrwb.sh fails
Zheng Yan
03:38 AM Bug #10323 (Resolved): lock get stuck in snap->sync state
Zheng Yan

02/09/2015

07:23 PM Bug #10703: failures in libcephfs-java tests
This is buggy on giant as well (and given what Zheng found out, probably Firefly?). Greg Farnum
06:56 PM Bug #10807 (Rejected): blogbench: hung MDS requests
http://pulpito.ceph.com/teuthology-2015-02-06_23:04:01-fs-master-testing-basic-multi/743177/
The whole run hung, s...
Greg Farnum
06:24 PM Bug #10720 (Need More Info): MDS: valgrind leaks
There's one DefinitelyLost in addition to a whole bunch of PossiblyLost. It's unfortunately a Message, so the origina... Greg Farnum

02/07/2015

10:30 PM Feature #10792 (Resolved): qa: enable thrasher for MDS cluster size (vary max_mds)

Rather than just killing MDSs within a fixed size cluster, have a thrasher that varies max_mds and deactivates MDSs...
John Spray
08:38 PM Bug #10744 (In Progress): MDS gets stuck in 'stopping' when strays exist
have a working patch for this, PR in due course... John Spray
02:01 PM Bug #10791 (Resolved): ceph mds deactivate sometimes fails silently

I suspect this is due to the targeted MDS sending a beacon at the same moment as the deactivate, which flips the pe...
John Spray

02/06/2015

06:13 AM Bug #10744: MDS gets stuck in 'stopping' when strays exist
Is it possible your hack to avoid purging anything is busting more than you realize? Greg Farnum
06:12 AM Bug #10743: mds/MDLog.cc: 259: FAILED assert(!capped) on MDS rank shutdown
Probably means there's some point where we run shutdown_pass() and assume it succeeds. (That's what calls MDLog::cap().) Greg Farnum

02/05/2015

03:03 PM Feature #10764: optimize memory usage of MDSCacheObject
We've scoped this out some before:
#4499, #4500, #4501, #4502, #4503, #4504, #4535 may be of interest. At least keep...
Greg Farnum
02:52 PM Feature #10764 (In Progress): optimize memory usage of MDSCacheObject
the attached file is dump of layouts of various classes. Zheng Yan
05:32 AM Bug #10620 (Resolved): TestFlush fails on formatting mistake
Haven't seen any issues. Greg Farnum

02/04/2015

06:05 PM Bug #10368: Assertion in _trim_expired_segments
We saw this on our internal lab install as well.
[ubuntu@magna125 ~]$ zless /var/log/ceph/ceph-mds.magna125.log-20...
Greg Farnum
05:30 PM Bug #4920 (Resolved): client: does not respect O_NOFOLLOW
Thanks! Greg Farnum
09:32 AM Bug #4920: client: does not respect O_NOFOLLOW
With pleasure. :-)
I corrected the typo in comment and squashed wip-4920 branch into two commits.
PR is available h...
Radoslaw Zarzynski
05:41 AM Bug #4920: client: does not respect O_NOFOLLOW
Looks good! Can you submit it as a PR against the main Ceph repo? :) Greg Farnum
01:02 PM Bug #10744: MDS gets stuck in 'stopping' when strays exist
Yes, I know -- multiple MDSs is what I'm testing here :-) John Spray
12:36 PM Bug #10744: MDS gets stuck in 'stopping' when strays exist
looks like you are running multiple MDS. When a MDS gets stuck in 'stopping' state, restart the whole MDS cluster, th... Zheng Yan
11:13 AM Bug #10744 (Resolved): MDS gets stuck in 'stopping' when strays exist
The migrate_stray part happens, but we end up stuck like this:... John Spray
12:32 PM Bug #10737 (Fix Under Review): ceph-fuse permits layout changes to files with data
https://github.com/ceph/ceph/pull/3617 Zheng Yan
11:10 AM Bug #10743 (Resolved): mds/MDLog.cc: 259: FAILED assert(!capped) on MDS rank shutdown

Hacked MDCache::eval-stray to never purge if my rank >0
Vstart cluster with 2 MDSs
mkdir ALPHA
dmesg> ALPHA/t...
John Spray
10:10 AM Bug #10720: MDS: valgrind leaks
This is a global string constant, which valgrind apparently doesn't understand:... John Spray
04:12 AM Bug #10740 (Rejected): teuthology: nfs test getting EBUSY on umount
ubuntu@teuthology:/a/teuthology-2015-02-01_23:10:01-knfs-next-testing-basic-multi/735724
This was previously #8576...
Greg Farnum

02/03/2015

10:17 PM Feature #1398: qa: multiclient file io test
Hmm. I have a branch with this commit: f93e531de6fa69a3ad32117c613094fc0aa0283e
(same as the one mentioned a few u...
Anonymous
09:39 PM Bug #10737 (Resolved): ceph-fuse permits layout changes to files with data
With two data pools having IDs 3 and 4:... John Spray
06:44 PM Bug #10720: MDS: valgrind leaks
Merged. Do you think there might be more leaks, John? Your comment was ambiguous. :) Greg Farnum
11:22 AM Bug #10720: MDS: valgrind leaks
Here's one: https://github.com/ceph/ceph/pull/3598 John Spray
04:58 AM Bug #10720: MDS: valgrind leaks
I know there are some OSD leaks, but there are records in the MDS too. Unless I mis-parsed the valgrind logs and they... Greg Farnum
03:32 AM Bug #10720: MDS: valgrind leaks
both failures are leaks in OSD Zheng Yan
06:39 PM Feature #10498: ObjectCacher: order wakeups when write calls block on throttling
It requires some dev work and testing, which might need to be scheduled. It goes in the feature tracker even if it's ... Greg Farnum
12:18 PM Feature #10498: ObjectCacher: order wakeups when write calls block on throttling
I vote that this is restored as "Bug" instead of "Feature". http://pulpito.ceph.com/loic-2015-01-29_15:41:06-rados-du... Loïc Dachary
03:49 PM Bug #10702: ceph-qa-suite: hung client-recovery task in nightlies
It's weird that the fusermount -u is failing -- usually if you have an offline MDS and an idle fuse mount and you fus... John Spray
02:51 AM Bug #10702: ceph-qa-suite: hung client-recovery task in nightlies
no, the first umount supposed to success. but it fails with error "Device or resource busy".
I think the simplest ...
Zheng Yan
02:44 AM Bug #10703: failures in libcephfs-java tests
... Zheng Yan
 

Also available in: Atom