Project

General

Profile

Activity

From 05/24/2016 to 06/22/2016

06/22/2016

09:09 PM Bug #16186: kclient: drops requests without poking system calls on reconnect
But if we restart requests from scratch, we're dramatically re-ordering them. We can seemingly send files back in tim... Greg Farnum
09:01 PM Bug #16186: kclient: drops requests without poking system calls on reconnect
I think it is working the way it is supposed to work.
We skip unsafe requests because the mds already got them and...
Sage Weil
08:59 PM Bug #16407: LibCephFS.UseUnmounted failed
You appear to have closed your own PR. And generally speaking we pass around negative error numbers, so readdir() is ... Greg Farnum
08:44 AM Bug #16407: LibCephFS.UseUnmounted failed
https://github.com/ceph/ceph/pull/9860 huanwen ren
07:36 AM Bug #16407 (Rejected): LibCephFS.UseUnmounted failed
2016-06-22T15:03:06.176 INFO:tasks.workunit.client.0.plana146.stdout:[ RUN ] LibCephFS.StripeUnitGran
2016-06-2...
Zezhu Zhang
08:55 PM Support #16043 (Closed): MDS is crashed
Greg Farnum
07:40 PM Feature #16228: Create teuthology task for Samba ping_pong test
(Copied from #16417) See Greg's draft https://github.com/gregsfortytwo/ceph-qa-suite/tree/wip-pingpong John Spray
07:40 PM Feature #16417 (Duplicate): test pingpong on ceph-fuse
John Spray
05:10 PM Feature #16417 (Duplicate): test pingpong on ceph-fuse
See #12653. We should integrate pingpong into our nightly test suite, to verify consistency on the kernel client and ... Greg Farnum
06:10 PM Feature #16419: add statx-like interface to libcephfs
Yeah, that's what I mean. We have ceph_ll_getattr now (afaict), so we need something like a ceph_ll_getattrx (that na... Jeff Layton
06:01 PM Feature #16419: add statx-like interface to libcephfs
Jeff Layton wrote:
> What I'm thinking is that we should add something along the lines of what David Howells has pro...
Sage Weil
05:39 PM Feature #16419: add statx-like interface to libcephfs
What I'm thinking is that we should add something along the lines of what David Howells has proposed for the new stat... Jeff Layton
05:35 PM Feature #16419 (Resolved): add statx-like interface to libcephfs
samba, in particular, can make use of the birthtime for an inode. Have ceph track the btime in the inode and provide ... Jeff Layton
01:01 PM Feature #15615: CephFSVolumeClient: List authorized IDs by share
https://github.com/ceph/ceph/pull/9864 Ramana Raja

06/21/2016

02:03 PM Bug #16397: nfsd selinux denials causing knfs tests to fail
Ahh, hmm -- just noticed the "add name" deinal too. Does the path "/proc/net/rpc/auth.unix.ip/channel" even exist? Ma... Jeff Layton
01:46 PM Bug #16397: nfsd selinux denials causing knfs tests to fail
Looks unrelated to anything ceph-specific. My guess is that this is an selinux policy bug, since rpc.mountd should be... Jeff Layton
11:56 AM Bug #16397 (Resolved): nfsd selinux denials causing knfs tests to fail
http://pulpito.ceph.com/teuthology-2016-06-20_17:35:01-knfs-master-testing-basic-mira/267607/ John Spray
11:26 AM Support #16043: MDS is crashed
I execute... Andrey Matyashov
06:05 AM Support #16043: MDS is crashed
Yes, i try reset journal and sessions.
I run:...
Andrey Matyashov
01:34 AM Support #16043: MDS is crashed
Yep. So looking through the log, I now see
>mds.2.journal ESession.replay sessionmap 0 < 18884 close client.166758...
Greg Farnum
09:36 AM Bug #16396: Fix shutting down mds timed-out due to deadlock
-https://github.com/ceph/ceph/pull/9841- Zhi Zhang
09:31 AM Bug #16396 (Resolved): Fix shutting down mds timed-out due to deadlock
This issue was found in jewel when restarting/stopping mds. It took long time for mds to completely stop until mds th... Zhi Zhang
09:02 AM Bug #16288 (New): mds: `session evict` tell command blocks forever with async messenger (TestVolu...
Oops, I meant to paste to begin with. I think it was this one:
/a/jspray-2016-06-13_14:56:46-fs-wip-jcsp-testing-qu...
John Spray

06/20/2016

08:12 PM Bug #16367: libcephfs: UID parsing breaks root squash (Ganesha FSAL)
Greg Farnum
07:57 PM Bug #16367: libcephfs: UID parsing breaks root squash (Ganesha FSAL)
Yeah, I expect that Frank's report is the root cause, but wanted to see to make sure. :) Greg Farnum
08:56 AM Bug #16367: libcephfs: UID parsing breaks root squash (Ganesha FSAL)
Now easier to read:... Kenneth Waegeman
08:55 AM Bug #16367: libcephfs: UID parsing breaks root squash (Ganesha FSAL)

I have ceph mounted under /mnt/nfs/ceph:
[root@test2202 test]# pwd
/mnt/nfs/ceph/test
[root@test2202 test]# ls ...
Kenneth Waegeman
08:08 PM Bug #16288 (Need More Info): mds: `session evict` tell command blocks forever with async messenge...
Greg Farnum
08:08 PM Bug #16288: mds: `session evict` tell command blocks forever with async messenger (TestVolumeClie...
John, do you have any logs? The only failure of this test I can find is http://qa-proxy.ceph.com/teuthology/teutholog... Greg Farnum
07:12 PM Bug #16288 (In Progress): mds: `session evict` tell command blocks forever with async messenger (...
Douglas Fuller
01:31 PM Bug #16042 (In Progress): MDS Deadlock on shutdown active rank while busy with metadata IO
Patrick Donnelly
09:35 AM Support #16043: MDS is crashed
Greg, I sent message with link to my debug log on your email. Service for ceph-post-file working has becomes unstable... Andrey Matyashov

06/17/2016

08:46 PM Bug #16164: mds: enforce a dirfrag limit on entries
PR here: https://github.com/ceph/ceph/pull/9789 Patrick Donnelly
05:50 PM Bug #16367 (Need More Info): libcephfs: UID parsing breaks root squash (Ganesha FSAL)
Greg Farnum
05:49 PM Bug #16367: libcephfs: UID parsing breaks root squash (Ganesha FSAL)
Can you please:
1) run ls -lha on the director you're testing in
2) do your tests
3) run ls -lha on all the releva...
Greg Farnum
03:15 PM Bug #16367 (Resolved): libcephfs: UID parsing breaks root squash (Ganesha FSAL)
Testing with ganesha 2.4-o-dev20 and libcephfs 10.2.1:
I did set root squash on in the ganesha.conf, but as root I c...
Kenneth Waegeman
05:14 PM Support #16043: MDS is crashed
Please set "debug mds = 20" and "debug mds log = 20" in your ceph.conf, turn it on, and then upload the mds log file ... Greg Farnum
04:04 AM Bug #16358 (Fix Under Review): Session::check_access() is buggy
https://github.com/ceph/ceph/pull/9769 Zheng Yan
03:53 AM Bug #16358 (Resolved): Session::check_access() is buggy
It calls CInode::make_path_string(path, false, in->get_projected_parent_dn()). The second argument 'false' makes the ... Zheng Yan

06/16/2016

03:14 PM Bug #16255: ceph-create-keys: sometimes blocks forever if mds "allow" is set
> The loop you're seeing presumably is only occurring when /etc/ceph/ceph.client-admin.keyring has been removed.
e...
Dietmar Maurer
03:05 PM Bug #16255: ceph-create-keys: sometimes blocks forever if mds "allow" is set
The difference between @"allow"@ and @"allow *"@ is that the @"*"@ is necessary in more recent versions to issue 'tel... John Spray
02:39 PM Fix #16276: Update TestSessionMap.test_mount_conn_close for async messenger
NB back out part of https://github.com/ceph/ceph-qa-suite/pull/1054 when fixing this, it's switched back to simple me... John Spray
02:29 PM Fix #16276: Update TestSessionMap.test_mount_conn_close for async messenger
http://pulpito.ceph.com/gregf-2016-06-10_19:20:53-fs-greg-fs-testing-610---basic-mira/250875/ Greg Farnum
02:39 PM Bug #16288: mds: `session evict` tell command blocks forever with async messenger (TestVolumeClie...
NB back out part of https://github.com/ceph/ceph-qa-suite/pull/1054 when fixing this, it's switched back to simple me... John Spray
02:38 PM Bug #16288: mds: `session evict` tell command blocks forever with async messenger (TestVolumeClie...
This deadlocks and lockdep makes it crash in our nightlies; we should fix it quickly! :) Greg Farnum
02:37 PM Feature #14271 (Resolved): directory listing: do not reset when fragmenting
Greg Farnum
02:33 PM Support #16043: MDS is crashed
... Andrey Matyashov
02:31 PM Support #16043: MDS is crashed
I upgraded my cluster to 10.2.2, situation not changed. Andrey Matyashov
01:57 PM Support #16043 (Need More Info): MDS is crashed
This probably isn't an issue any more, but if it is upgrade to 10.2.2 and report back if it's still an issue. Greg Farnum
02:26 PM Feature #11171 (In Progress): Path filtering on "dump cache" asok
Douglas Fuller
02:21 PM Backport #16284 (Resolved): jewel: directory listing: do not reset when fragmenting
This was done as part of #16251. Greg Farnum
11:54 AM Bug #16298 (Resolved): mds: failure in tasks/migration.yaml
John Spray
11:15 AM Bug #16322: ceph mds getting killed for no reason
$gdb /usr/local/bin/ceph-mds
If gdb does not say "no debugging symbols found", the debug package is properly insta...
Zheng Yan
09:45 AM Bug #16322: ceph mds getting killed for no reason
Zheng Yan wrote:
> Your ceph-mds does not contain debuginfo, please install debuginfo package first. then start ceph...
Joao Castro
02:20 AM Bug #16322: ceph mds getting killed for no reason
Your ceph-mds does not contain debuginfo, please install debuginfo package first. then start ceph-mds manually with c... Zheng Yan
07:39 AM Backport #16136: jewel: MDSMonitor fixes
Original description:
These two commits:
https://github.com/ceph/ceph/pull/9418/commits/24b82bafffced97384135e5...
Loïc Dachary

06/15/2016

06:31 PM Bug #16042: MDS Deadlock on shutdown active rank while busy with metadata IO
This is rearing its head in general testing now:
http://pulpito.ceph.com/jspray-2016-06-15_05:28:02-fs-wip-jcsp-test...
John Spray
02:01 PM Bug #16322: ceph mds getting killed for no reason
log: http://95.211.209.196/imgs/ceph-mds.mds01.log Joao Castro
01:48 PM Bug #16322: ceph mds getting killed for no reason
kernel: 4.2.0-36-generic
Joao Castro
01:46 PM Bug #16322: ceph mds getting killed for no reason

(...)
Loaded symbols for /lib/x86_64-linux-gnu/libnss_files.so.2
Reading symbols from /usr/lib/x86_64-linux-gnu/n...
Joao Castro
01:41 PM Bug #16322: ceph mds getting killed for no reason
I am not very experience with gdb, sorry. Should I use it in ceph-mds ?
I will paste the whole log (it has a lot of ...
Joao Castro
12:16 PM Bug #16322: ceph mds getting killed for no reason
could you enable coredump and use gdb to check which line causes the crash Zheng Yan
11:50 AM Bug #16322: ceph mds getting killed for no reason
add:
2016-06-15 03:15:51.017714 7f582103f700 -1 *** Caught signal (Aborted) **
in thread 7f582103f700 thread_nam...
Joao Castro
11:50 AM Bug #16322 (Can't reproduce): ceph mds getting killed for no reason
Hello,
my ceph mds get killed for no reason (normally they do the active failover).
Log:
ceph version 10.2.1 (...
Joao Castro
10:11 AM Bug #15920: mds/StrayManager.cc: 520: FAILED assert(dnl->is_primary())
John, thank you very much! Yeah, I saw that it was going to miss 10.2.2. Thank you for making this exception! I'll st... Daniel van Ham Colchete
08:06 AM Backport #16320 (Resolved): jewel: fs: fuse mounted file systems fails SAMBA CTDB ping_pong rw te...
https://github.com/ceph/ceph/pull/10108 Nathan Cutler
08:04 AM Backport #16313 (Resolved): jewel: client: FAILED assert(root_ancestor->qtree == __null)
https://github.com/ceph/ceph/pull/10107 Nathan Cutler
02:21 AM Bug #16160 (Resolved): PJD failures on Jewel
http://qa-proxy.ceph.com/teuthology/teuthology-2016-06-13_17:25:02-kcephfs-master-testing-basic-mira/257158/teutholog... Zheng Yan

06/14/2016

04:30 PM Bug #12653 (Pending Backport): fuse mounted file systems fails SAMBA CTDB ping_pong rw test with ...
John Spray
04:29 PM Documentation #16300: doc: fuse_disable_pagecache
NB while doing this would be useful to ask performance team to measure how much impact this really has John Spray
04:28 PM Documentation #16300 (Resolved): doc: fuse_disable_pagecache
http://tracker.ceph.com/issues/12653
https://github.com/ceph/ceph/pull/5521/commits/0f11ec237d4692d313a038ed61aa07a3...
John Spray
04:24 PM Backport #16299 (Resolved): jewel: mds: fix SnapRealm::have_past_parents_open()
https://github.com/ceph/ceph/pull/10499 John Spray
04:22 PM Bug #16298 (Fix Under Review): mds: failure in tasks/migration.yaml
https://github.com/ceph/ceph/pull/9697 John Spray
04:20 PM Bug #16298 (Resolved): mds: failure in tasks/migration.yaml
http://pulpito.ceph.com/jspray-2016-06-14_01:19:46-fs-wip-jcsp-testing-20160610-distro-basic-mira/257906 John Spray
04:19 PM Bug #15920: mds/StrayManager.cc: 520: FAILED assert(dnl->is_primary())

I've pushed a jewel-15920 branch for you with the fix cherry-picked onto it. (don't usually do this, but it's fair...
John Spray
01:56 PM Bug #15920: mds/StrayManager.cc: 520: FAILED assert(dnl->is_primary())
Good morning everyone!
Considering that a backport is done, though not merged yet, is there away for me to get a g...
Daniel van Ham Colchete
02:50 PM Backport #13927 (New): hammer: cephfs-java ftruncate unit test failure
Nathan Cutler
02:50 PM Backport #13927: hammer: cephfs-java ftruncate unit test failure
One attempted backport https://github.com/ceph/ceph/pull/6754 was closed. Nathan Cutler
12:30 PM Bug #16067 (Resolved): client: InvalidWrite in put_qtree
(resolved via http://tracker.ceph.com/issues/16066, track backport there) John Spray
12:30 PM Bug #16066 (Pending Backport): client: FAILED assert(root_ancestor->qtree == __null)
John Spray
10:13 AM Bug #16288 (Resolved): mds: `session evict` tell command blocks forever with async messenger (Tes...

I'm assuming for the moment that this is an MDS bug rather than something getting dropped in the new messenger code...
John Spray
07:19 AM Backport #16284 (Resolved): jewel: directory listing: do not reset when fragmenting
https://github.com/ceph/ceph/pull/9655 Nathan Cutler
12:24 AM Bug #16042: MDS Deadlock on shutdown active rank while busy with metadata IO
I just saw this (or similar shutdown bug) for the first time in an automated test: http://qa-proxy.ceph.com/teutholog... John Spray
12:20 AM Fix #16276 (New): Update TestSessionMap.test_mount_conn_close for async messenger

When the default messenger changed from simple to async, this test started failing[1]. It's because it is using th...
John Spray

06/13/2016

10:42 AM Feature #14271 (Pending Backport): directory listing: do not reset when fragmenting
John Spray
08:27 AM Backport #16252 (Resolved): jewel: Client: reports that readahead is not working
Greg Farnum
05:05 AM Backport #16252 (In Progress): jewel: Client: reports that readahead is not working
Nathan Cutler
05:00 AM Backport #16252 (Resolved): jewel: Client: reports that readahead is not working
https://github.com/ceph/ceph/pull/9656 Nathan Cutler
08:27 AM Bug #16024 (Resolved): Client: reports that readahead is not working
Greg Farnum
08:23 AM Backport #16251 (Resolved): jewel: client: simultaneous readdirs are very racy
Greg Farnum
04:55 AM Backport #16251 (In Progress): jewel: client: simultaneous readdirs are very racy
Nathan Cutler
04:54 AM Backport #16251 (Resolved): jewel: client: simultaneous readdirs are very racy
https://github.com/ceph/ceph/pull/9655 Nathan Cutler
08:23 AM Bug #15508 (Resolved): client: simultaneous readdirs are very racy
Greg Farnum
05:41 AM Bug #16255 (Resolved): ceph-create-keys: sometimes blocks forever if mds "allow" is set
The documentations at:
http://docs.ceph.com/docs/master/dev/mon-bootstrap/
tells to create the client.admin key...
Dietmar Maurer

06/12/2016

09:45 PM Bug #16024 (Pending Backport): Client: reports that readahead is not working
Backport PR: https://github.com/ceph/ceph/pull/9656 Greg Farnum
09:35 PM Bug #15508 (Pending Backport): client: simultaneous readdirs are very racy
Backport PR: https://github.com/ceph/ceph/pull/9655 Greg Farnum

06/10/2016

09:55 AM Feature #16228 (New): Create teuthology task for Samba ping_pong test

The Samba ping_pong test validates the interaction between multiple clients accessing the same data.
Related:
h...
John Spray

06/09/2016

09:03 PM Bug #16067: client: InvalidWrite in put_qtree
Greg: yes, I expect the big quotatree patch will fix both. John Spray
04:32 PM Bug #16067: client: InvalidWrite in put_qtree
Any chance this is because of #16066, or at least resolved by the associated PR? Greg Farnum
07:40 PM Feature #16219 (New): test: smallfile benchmark tool
Run this metadata tester in our nightlies.
https://github.com/bengland2/smallfile
>smallfile is a python-based ...
Greg Farnum
05:38 PM Backport #16215 (Resolved): jewel: client: crash in unmount when fuse_use_invalidate_cb is enabled
https://github.com/ceph/ceph/pull/10106 Nathan Cutler
09:49 AM Cleanup #15922 (Resolved): MDS: remove TMAP support from CDir
John Spray
09:36 AM Bug #16137 (Pending Backport): client: crash in unmount when fuse_use_invalidate_cb is enabled
John Spray

06/08/2016

08:29 PM Bug #16186: kclient: drops requests without poking system calls on reconnect
I'm concerned to hear that; I thought those patches had been zapped from the queue. If we disconnect and reconnect, r... Greg Farnum
01:49 AM Bug #16186: kclient: drops requests without poking system calls on reconnect
Client only drops unsafe MDS requests after session reset. It also tries re-sending outstanding requests
please ...
Zheng Yan
01:04 AM Bug #16186 (Duplicate): kclient: drops requests without poking system calls on reconnect
If I'm understanding the way things currently work:
*) kernel client loses network connection
*) MDS times out kern...
Greg Farnum
08:18 PM Bug #16024: Client: reports that readahead is not working
Test is here: https://github.com/ceph/ceph-qa-suite/pull/1046
Greg is adding that to his test branch.
Patrick Donnelly
01:42 PM Bug #16164: mds: enforce a dirfrag limit on entries
He's got more info in http://tracker.ceph.com/issues/16177.
Basically, a CephFS user created directories large eno...
Greg Farnum
05:10 AM Bug #16164: mds: enforce a dirfrag limit on entries
Greg Farnum wrote:
> Hmm, I was talking to m0zes (whose situation kicked off this bug) and it turns out the objects ...
Xiaoxi Chen
12:41 PM Cleanup #16195 (Resolved): mds: Don't spam log with standby_replay_restart messages
... John Spray
10:25 AM Bug #16066 (Fix Under Review): client: FAILED assert(root_ancestor->qtree == __null)
Zheng Yan
10:24 AM Bug #16066: client: FAILED assert(root_ancestor->qtree == __null)
https://github.com/ceph/ceph/pull/9591 Zheng Yan

06/07/2016

09:57 PM Bug #16164: mds: enforce a dirfrag limit on entries
Hmm, I was talking to m0zes (whose situation kicked off this bug) and it turns out the objects actually causing the i... Greg Farnum
04:15 PM Bug #15266 (Resolved): ceph_volume_client purge failing on non-ascii filenames
Ramana Raja
02:47 PM Bug #16022: MDSMonitor::check_subs() is very buggy
Abhishek: yes, that should be backported too. It's not strictly necessary but is worthwhile. John Spray
02:35 PM Bug #16022: MDSMonitor::check_subs() is very buggy
Does https://github.com/ceph/ceph-qa-suite/pull/1018 need to be packported to Jewel too? Please confirm. Abhishek Varshney
02:47 PM Backport #16152 (In Progress): jewel: fs: client: fstat cap release
Abhishek Varshney
02:44 PM Backport #16136 (In Progress): jewel: MDSMonitor fixes
Abhishek Varshney
02:41 PM Backport #16135 (In Progress): jewel: MDS: fix getattr starve setattr
Abhishek Varshney
02:38 PM Backport #16041 (In Progress): jewel: mds/StrayManager.cc: 520: FAILED assert(dnl->is_primary())
Abhishek Varshney
02:23 PM Backport #15999 (In Progress): jewel: CephFSVolumeClient: read-only authorization for volumes
Abhishek Varshney
02:06 PM Backport #15898 (In Progress): jewel: Confusing MDS log message when shut down with stalled journ...
Abhishek Varshney
01:31 PM Backport #15968 (Fix Under Review): jewel: ceph status mds output ignores active MDS when there i...
https://github.com/ceph/ceph/pull/9547 John Spray
01:28 PM Backport #15971 (Resolved): jewel: ceph_volume_client purge failing on non-ascii filenames
John Spray
01:24 PM Bug #16160: PJD failures on Jewel
Cool, I'll close this when our nightlies are passing again. John Spray
01:21 PM Bug #16160: PJD failures on Jewel
VFS maintainer proposed a fix. The issue should be fixed in 4.7-rc3 kernel Zheng Yan
01:53 AM Bug #16160: PJD failures on Jewel
Sorry, I mean 4.7-rc1,it's the newest rc kernel. Zheng Yan

06/06/2016

08:52 PM Bug #16160: PJD failures on Jewel
The failing test file is pretty short; the important bit is... Greg Farnum
12:32 PM Bug #16160: PJD failures on Jewel
This issue happens only on 3.7.0-rc1 kernel Zheng Yan
11:41 AM Bug #16160 (Resolved): PJD failures on Jewel
http://qa-proxy.ceph.com/teuthology/jspray-2016-06-03_06:24:38-fs:basic-jewel---basic-mira/232657/teuthology.log
htt...
John Spray
06:49 PM Bug #16164 (In Progress): mds: enforce a dirfrag limit on entries
Patrick Donnelly
04:36 PM Bug #16164: mds: enforce a dirfrag limit on entries
I'm taking a look at this one. Patrick Donnelly
01:49 PM Bug #16164 (Resolved): mds: enforce a dirfrag limit on entries
- add a new config option to cap the number of entries in a difrag
- set the limit an order of magnitude higher than...
Sage Weil
11:45 AM Feature #15417 (Resolved): Make path prefix ("/volumes") in CephFSVolumeClient configurable
Nathan Cutler
11:44 AM Backport #15854 (Resolved): jewel: Make path prefix ("/volumes") in CephFSVolumeClient configurable
Nathan Cutler
11:42 AM Feature #15599 (Resolved): quota: Generate client df from quota, when using subdirectory mount
Nathan Cutler
11:38 AM Backport #16065 (Resolved): jewel: quota: Generate client df from quota, when using subdirectory ...
Nathan Cutler
11:38 AM Backport #16065: jewel: quota: Generate client df from quota, when using subdirectory mount
h3. original description
This is an enabler for Manila, creating this ticket to track backport to Jewel after land...
Nathan Cutler
09:58 AM Bug #16066: client: FAILED assert(root_ancestor->qtree == __null)
The quota code expects that directory is alway connected to the FS hierarchy. But this is not true, we can create dis... Zheng Yan
08:55 AM Bug #16137 (Fix Under Review): client: crash in unmount when fuse_use_invalidate_cb is enabled
https://github.com/ceph/ceph/pull/9509 Zheng Yan

06/03/2016

05:44 PM Backport #16135: jewel: MDS: fix getattr starve setattr
h3. original description
To backport this fix (https://github.com/ceph/ceph/pull/8965) to Jewel
Nathan Cutler
10:44 AM Backport #16135 (Resolved): jewel: MDS: fix getattr starve setattr
https://github.com/ceph/ceph/pull/9560 John Spray
05:38 PM Bug #16154 (Resolved): mds: lock waiters are not finished in the same order that they were added
https://github.com/ceph/ceph/pull/8965 Nathan Cutler
04:54 PM Backport #16152 (Resolved): jewel: fs: client: fstat cap release
https://github.com/ceph/ceph/pull/9562 Nathan Cutler
02:55 PM Cleanup #16144 (Resolved): Remove cephfs-data-scan tmap_upgrade
This was something that was only for jewel only (the last release to have tmaps in rados).
We can remove it for Kr...
John Spray
12:57 PM Bug #9904 (Resolved): Don't crash MDS on clients sending messages with bad seq
https://github.com/ceph/ceph/pull/9214 John Spray
12:56 PM Bug #8255 (New): mds: directory with missing object cannot be removed
John Spray
12:56 PM Feature #11950: Strays enqueued for purge cause MDCache to exceed size limit
I'm reverting state to verified because although we've merged the patch for this, it still needs more attention to ha... John Spray
12:08 PM Bug #15045 (Resolved): CephFSVolumeClient.evict should be limited by path, not just auth ID
Nathan Cutler
12:07 PM Backport #15855 (Resolved): jewel: CephFSVolumeClient.evict should be limited by path, not just a...
Nathan Cutler
11:32 AM Bug #16137 (Resolved): client: crash in unmount when fuse_use_invalidate_cb is enabled
Seen in testing of https://github.com/ceph/ceph/pull/8890... John Spray
10:51 AM Bug #15723 (Pending Backport): client: fstat cap release
John Spray
10:46 AM Backport #16136 (Resolved): jewel: MDSMonitor fixes

Backported as described : https://github.com/ceph/ceph/pull/9561
John Spray

06/02/2016

10:40 AM Cleanup #15922 (Fix Under Review): MDS: remove TMAP support from CDir
https://github.com/ceph/ceph/pull/9443 John Spray

06/01/2016

10:15 PM Backport #16065 (In Progress): jewel: quota: Generate client df from quota, when using subdirecto...
https://github.com/ceph/ceph/pull/9430 John Spray
10:15 PM Backport #15971 (In Progress): jewel: ceph_volume_client purge failing on non-ascii filenames
https://github.com/ceph/ceph/pull/9430 John Spray
04:57 PM Bug #90: mds: don't sync log on every clientreplay request
Yeah. So, I think there was a reason for this. We definitely need to flush out any clientreplay requests before exiti... Greg Farnum
08:26 AM Bug #90: mds: don't sync log on every clientreplay request

The code that flush journal. https://github.com/ceph/ceph/blob/8ce337e3552004fc4853c0c94f33235da4caa5df/src/mds/Ser...
Zheng Yan
02:39 AM Feature #3575: ceph-fuse: Add support for forget_multi
the kernel sends forget_multi request to libfuse even we don't provide the forget_multi callback. Zheng Yan

05/31/2016

09:30 PM Bug #90: mds: don't sync log on every clientreplay request
It might be fixed; we should check. But once upon a time (I think maybe still now?) then when in clientreplay set the... Greg Farnum
11:04 AM Bug #90: mds: don't sync log on every clientreplay request
Is this definitely still an issue? I'm not seeing a particular place in the code where we do something different wit... John Spray
09:28 PM Feature #83: mds: rename over old files should flush data or revert to old contents?
I think it does still exist?
We have an existing foo.conf, inode x
Write to foo.conf.tmp, inode y
rename foo.conf....
Greg Farnum
10:59 AM Feature #83: mds: rename over old files should flush data or revert to old contents?
The original ticket is describing a real bug (that no longer exists), right? It's not clear to me that there's still... John Spray
08:01 PM Feature #3575: ceph-fuse: Add support for forget_multi
Hmm, doesn't the syscall overhead make enough of a difference to be noticeable in some of our bigger invalidate ops? Greg Farnum
11:43 AM Feature #15067 (Fix Under Review): mon: client: multifs: enable clients to map a filesystem name ...
https://github.com/ceph/ceph/pull/8386
https://github.com/ceph/ceph-qa-suite/pull/924
John Spray
11:41 AM Feature #7326: qa: fix flock tests
Me neither. Sage? John Spray
11:39 AM Feature #2097 (Rejected): mds: 'ceph mds activate <gid>'
I think we don't. John Spray
11:39 AM Feature #9466: kclient: Extend CephFSTestCase tests to cover kclient
I think most or all of the hooks are there, should be a case of taking suites/fs/recovery/tasks/client-limits.yaml an... John Spray
11:35 AM Feature #7764 (Resolved): InoTable/SessionMap/ manipulator (cephfs-table-tool)
So cephfs-table-tool has resetting tables, and it has consuming inodes from InoTable.
There probably are more thin...
John Spray
11:34 AM Feature #7762 (Rejected): journal-tool: backwards-search after corrupt regions
Hmm, iirc this ticket was about doing something smarter than scanning forward for a sentinel, where we would potentia... John Spray
11:19 AM Feature #15400 (Resolved): CephFSVolumeClient should isolate volumes by RADOS namespace
Loïc Dachary
11:18 AM Feature #7760 (Resolved): journal-tool: implement splice
Yep! John Spray
11:18 AM Feature #7758 (Resolved): journal-tool: complete filtering
Nope! John Spray
11:16 AM Feature #7318 (Duplicate): qa: ceph-fuse + sync mode
Hmm, #4022 seems to be about globally disabling cacher (as an alternative way to test direct IO paths) whereas this o... John Spray
11:11 AM Backport #16065 (New): jewel: quota: Generate client df from quota, when using subdirectory mount
Switching back to New because there does not seem to be a pull request for this yet. Loïc Dachary
09:51 AM Backport #16083 (In Progress): jewel: mds: wrongly treat symlink inode as normal file/dir when sy...
Nathan Cutler
07:45 AM Backport #16083 (Resolved): jewel: mds: wrongly treat symlink inode as normal file/dir when symli...
https://github.com/ceph/ceph/pull/9405 Loïc Dachary
09:50 AM Backport #16082 (In Progress): hammer: mds: wrongly treat symlink inode as normal file/dir when s...
Nathan Cutler
07:45 AM Backport #16082 (Resolved): hammer: mds: wrongly treat symlink inode as normal file/dir when syml...
https://github.com/ceph/ceph/pull/9404 Loïc Dachary

05/30/2016

12:28 PM Feature #3575: ceph-fuse: Add support for forget_multi
If there is not forget_multi() callback, libfuse calls the forget() callback in a loop. I don't think implementing fo... Zheng Yan

05/28/2016

10:46 PM Bug #16066: client: FAILED assert(root_ancestor->qtree == __null)

http://pulpito.ceph.com/jspray-2016-05-28_13:42:42-fs-wip-jcsp-testing-20160527b---basic-mira/220171
+other instan...
John Spray
10:42 PM Bug #16066 (Resolved): client: FAILED assert(root_ancestor->qtree == __null)
Seen on test branch where client quota was enabled by default (https://github.com/ceph/ceph/pull/9346) -- presumably ... John Spray
10:44 PM Bug #16067 (Resolved): client: InvalidWrite in put_qtree
Like #16066, this seen in test branch with client quota enabled by default.
http://pulpito.ceph.com/jspray-2016-05...
John Spray
07:32 AM Cleanup #15922 (In Progress): MDS: remove TMAP support from CDir
Vishal Kanaujia
06:55 AM Backport #16065 (In Progress): jewel: quota: Generate client df from quota, when using subdirecto...
Xiaoxi Chen
06:53 AM Backport #16065 (Resolved): jewel: quota: Generate client df from quota, when using subdirectory ...
https://github.com/ceph/ceph/pull/9430
Xiaoxi Chen
06:51 AM Feature #15599 (Pending Backport): quota: Generate client df from quota, when using subdirectory ...
Xiaoxi Chen

05/27/2016

11:37 PM Feature #81 (Resolved): mds: do authentication checks
Hey, we do that now! Greg Farnum
11:36 PM Feature #3314: client: client interfaces should take a set of group ids
Sounds like we'll need this for NFS, if nothing else requires it sooner. Greg Farnum
11:35 PM Feature #3315 (Resolved): client: Add acl support
Hey, we support POSIX ACLs now, and already have a ticket/code for RichACLs whenever that's done. Greg Farnum
11:34 PM Feature #118: kclient: clean pages when throwing out dirty metadata on session teardown
Zheng, do you have any idea what this is about? Greg Farnum
11:21 PM Feature #83: mds: rename over old files should flush data or revert to old contents?
I don't suppose POSIX says anything about this...
I think I vote for requiring a flush before renaming over existi...
Greg Farnum
11:17 PM Bug #90: mds: don't sync log on every clientreplay request
This'll probably kill us in a cluster of any size...? Greg Farnum
11:16 PM Feature #13999: client: richacl support
Greg Farnum
11:15 PM Feature #9880: mds: more gracefully handle EIO on missing dir object
I think this'll be caught in John's damaged stuff now. Sage, did you have any thing specifically you wanted to see ge... Greg Farnum
11:02 PM Feature #7318: qa: ceph-fuse + sync mode
Is this different from #4022? Greg Farnum
11:00 PM Feature #7322: qa: inline data + thrashing
We enable inline data, but I'm not sure if it thrashes as well. Greg Farnum
10:59 PM Feature #7333: client: evaluate multiple O_APPEND writers
Do we have some known issues with how we handle EOF and synchronous IO? Greg Farnum
10:57 PM Feature #7758: journal-tool: complete filtering
Was there something in this that isn't done yet? Greg Farnum
10:56 PM Feature #7760: journal-tool: implement splice
This is done now, right? Greg Farnum
10:56 PM Feature #7762: journal-tool: backwards-search after corrupt regions
We already have commands to harvest the latest versions of dentries and things, and I think we can punch holes and sk... Greg Farnum
10:55 PM Feature #7764: InoTable/SessionMap/ manipulator (cephfs-table-tool)
John, do you think we're done with this based on the table reset commands, or is there more to usefully do? Greg Farnum
10:50 PM Feature #9466: kclient: Extend CephFSTestCase tests to cover kclient
What still needs to get done for this, John? Greg Farnum
10:49 PM Feature #2097: mds: 'ceph mds activate <gid>'
Do we really want to be able to do this? Greg Farnum
10:49 PM Feature #1237 (Resolved): mds caps limit mount to some subdir
We did this last summer! Check out the MDS cephx cap configuration docs. Greg Farnum
10:37 PM Feature #8636 (Resolved): mds/libcephfs: read only mount
I believe this is done: we can now restrict clients to read-only caps on the MDS and the OSD. Greg Farnum
10:35 PM Feature #10393 (Rejected): client: remove mount prefix shenanigans for quota
Rejecting because we're not sure about subvolumes yet. I think we handle the security stuff properly, but not sure ab... Greg Farnum
10:34 PM Feature #10392 (Rejected): mds: refactor subvolume vs snaprealm, capture quota trees
This is a possibility but not something we're sure we want to do. Greg Farnum
10:00 PM Feature #12671: Enforce cache limit during dirfrag load during open_ino (during rejoin)
The naive solution to this seems pretty bad as well. If we only load the needed dentries, in a serial fashion, we'll ... Greg Farnum
09:36 PM Bug #15702 (Pending Backport): mds: wrongly treat symlink inode as normal file/dir when symlink i...
Greg Farnum
09:34 PM Feature #15617: CephFSVolumeClient: OSD blacklisting on deauthorize
This is a feature, not a bug, right? ;) Greg Farnum
09:09 PM Feature #603 (Resolved): mds: repair directory hierarchy
Another fsck ticket that doesn't contribute past what we already have. :) Greg Farnum
09:08 PM Feature #4145 (Resolved): MDS: design and implement a backwards-scanning fsck
It looks a little different now, and we have other tickets to improve stuff, but cephfs-data-scan shoudl qualify this... Greg Farnum
09:07 PM Feature #86 (Resolved): mds: implement fsck
Not complete, but we have recovery and repair tools. Let's call this ticket done.
http://docs.ceph.com/docs/master/c...
Greg Farnum
09:06 PM Feature #4799 (Resolved): Client Security for CephFS
Done last year! http://docs.ceph.com/docs/master/cephfs/client-auth/ Greg Farnum
05:55 PM Bug #16024 (Fix Under Review): Client: reports that readahead is not working
https://github.com/ceph/ceph/pull/9374 Greg Farnum
10:16 AM Bug #15921: segfault in cephfs-journal-tool (TestJournalRepair failure)
And again:
http://pulpito.ceph.com/jspray-2016-05-26_17:59:25-fs-wip-jcsp-testing-20160527---basic-mira/217709
John Spray
09:43 AM Support #16043: MDS is crashed
Today all MDS in my cluster is died. Andrey Matyashov

05/26/2016

01:25 PM Bug #16022: MDSMonitor::check_subs() is very buggy
Created reproducer in TestMultiFilesystems https://github.com/ceph/ceph-qa-suite/pull/1018 John Spray
09:48 AM Bug #16022 (Pending Backport): MDSMonitor::check_subs() is very buggy
John Spray
12:59 PM Support #16043 (Closed): MDS is crashed
I updated ceph from hammer to jewel. After restart ceph daemons, 2 mds (from 5) not started.
I execute commands on...
Andrey Matyashov
12:48 PM Bug #16042 (Resolved): MDS Deadlock on shutdown active rank while busy with metadata IO

Report and log here: https://bugzilla.redhat.com/show_bug.cgi?id=1340004
Log looks like one of the MDLog threads...
John Spray
12:02 PM Backport #16041 (Resolved): jewel: mds/StrayManager.cc: 520: FAILED assert(dnl->is_primary())
https://github.com/ceph/ceph/pull/9559 Nathan Cutler
12:01 PM Backport #16037 (Resolved): jewel: MDSMonitor::check_subs() is very buggy
https://github.com/ceph/ceph/pull/10103 Nathan Cutler
09:55 AM Cleanup #16035 (Resolved): Remove "cephfs" CLI
IIRC we talked about this in standup the other day, and this pull request (https://github.com/ceph/ceph/pull/9338) re... John Spray
09:46 AM Bug #15920 (Pending Backport): mds/StrayManager.cc: 520: FAILED assert(dnl->is_primary())
John Spray

05/25/2016

02:50 PM Bug #16024: Client: reports that readahead is not working
<spoiler> John Spray
02:12 PM Bug #16024 (Resolved): Client: reports that readahead is not working
We've had one or two previous reports that ceph-fuse is reading slowly and that readahead may not be working, but we ... Greg Farnum
02:32 PM Feature #3426 (New): ceph-fuse: build/run on os x
Nobody owns it right now, and we never made it official. Greg Farnum
10:31 AM Feature #3426: ceph-fuse: build/run on os x
Cleaning up "In progress" tickets: should this be "resolved" or go back to new? I know various people have worked on... John Spray
02:30 PM Feature #4022: client: qa: test non-cached operation (force sync mode)
I don't think we have anything right now, no. Greg Farnum
10:31 AM Feature #4022 (New): client: qa: test non-cached operation (force sync mode)
Greg: did we end up with a test that corresponds with this ticket or should it still be open? John Spray
12:30 PM Bug #16022 (Resolved): MDSMonitor::check_subs() is very buggy
https://github.com/ceph/ceph/pull/9323 Zheng Yan
11:19 AM Feature #16016 (Resolved): Populate DamageTable from forward scrub
We detect metadata damage two ways: when we try to read things off disk during normal operation, and when we try to v... John Spray
11:03 AM Bug #11255 (Resolved): nfs: mount failures on ceph-backed NFS share
The linked PR merged and this is not appearing currently in master. John Spray
10:32 AM Bug #15900 (Resolved): TestSessionMap.test_mount_conn_close failure: AssertionError: 1 not greate...
John Spray
10:30 AM Bug #7422 (Resolved): client/barrier.h uses boost's interval set library, which is not available ...
John Spray
10:29 AM Bug #9904 (Fix Under Review): Don't crash MDS on clients sending messages with bad seq
John Spray
10:29 AM Feature #11859 (Resolved): MDS "damage table" for recording scrub/fetch errors
DamageTable went in for Jewel John Spray
10:28 AM Bug #14195 (Won't Fix): test_full_fclose fails (tasks.cephfs.test_full.TestClusterFull)
This was only a test race, and it only happens on pathologically slow clusters. John Spray
10:27 AM Bug #15920 (Fix Under Review): mds/StrayManager.cc: 520: FAILED assert(dnl->is_primary())
John Spray
08:39 AM Bug #16013: Failing file operations on kernel based cephfs mount point leaves unaccessible file b...
Thanks for the fast reply.
I had to restart the MDS server due to pending locks (which also blocked the ceph-fuse ...
Burkhard Linke
08:07 AM Bug #16013: Failing file operations on kernel based cephfs mount point leaves unaccessible file b...
The warning is caused by bad inode size for symlink, commit https://github.com/ceph/ceph-client/commit/8e876900de5196... Zheng Yan
08:00 AM Bug #16013: Failing file operations on kernel based cephfs mount point leaves unaccessible file b...
another addition:
- mounted cephfs using ceph-fuse
- directory is listed correctly:
# ls -al bin
total 1
drwxr...
Burkhard Linke
07:39 AM Bug #16013: Failing file operations on kernel based cephfs mount point leaves unaccessible file b...
Additional information:
- ceph cluster reported client failing to release capabilities prior to the current state
...
Burkhard Linke
07:28 AM Bug #16013 (Resolved): Failing file operations on kernel based cephfs mount point leaves unaccess...
After some number of operations on files (which could not be traced for reproduction), we end up with a broken direct... Burkhard Linke
 

Also available in: Atom