Activity
From 04/11/2018 to 05/10/2018
05/10/2018
- 10:09 PM Bug #24090 (Resolved): mds: fragmentation in QA is slowing down ops enough for WRNs
- http://pulpito.ceph.com/pdonnell-2018-05-08_18:15:09-fs-mimic-testing-basic-smithi/
http://pulpito.ceph.com/pdonnell... - 08:48 PM Bug #24089 (Rejected): mds: print slow requests to debug log when sending health WRN to monitors ...
- Nevermind, it is actually printed earlier in the log. Sorry for the noise.
- 08:46 PM Bug #24089 (Rejected): mds: print slow requests to debug log when sending health WRN to monitors ...
- ...
- 08:22 PM Bug #24088 (Duplicate): mon: slow remove_snaps op reported in cluster health log
- ...
- 04:57 PM Bug #24087 (Duplicate): client: assert during shutdown after blacklisted
- ...
- 11:48 AM Bug #24039: MDSTableServer.cc: 62: FAILED assert(g_conf->mds_kill_mdstable_at != 1)
- the pjd: http://pulpito.ceph.com/pdonnell-2018-05-04_03:45:51-multimds-master-testing-basic-smithi/2475062/...
- 11:25 AM Feature #22446: mds: ask idle client to trim more caps
- Glad to see this :)
- Backport set to mimic,luminous
Thanks. - 09:44 AM Bug #23332: kclient: with fstab entry is not coming up reboot
- I still don't think this is kernel issue. please path kernel with below change and try again....
- 06:44 AM Bug #24073: PurgeQueue::_consume() could return true when there were no purge queue item actually...
- Xuehan Xu wrote:
> In our online clusters, we encountered the bug #19593. Although we cherry-pick the fixing commits... - 04:38 AM Bug #24073: PurgeQueue::_consume() could return true when there were no purge queue item actually...
- https://github.com/ceph/ceph/pull/21923
- 04:38 AM Bug #24073 (Resolved): PurgeQueue::_consume() could return true when there were no purge queue it...
- In our online clusters, we encountered the bug #19593. Although we cherry-pick the fixing commits, the purge queue's ...
- 06:39 AM Bug #24074 (Need More Info): Read ahead in fuse client is broken with large buffer size
- If the read is larger than 128K(e.g. 4M as our object size), fuse client will receive read requests as multiple ll_re...
- 04:10 AM Backport #23984 (In Progress): luminous: mds: scrub on fresh file system fails
- https://github.com/ceph/ceph/pull/21922
- 04:07 AM Backport #23982 (In Progress): luminous: qa: TestVolumeClient.test_lifecycle needs updated for ne...
- https://github.com/ceph/ceph/pull/21921
- 04:05 AM Bug #23826: mds: assert after daemon restart
- Finish context of MDCache::open_undef_inodes_dirfrags() calls rejoin_gather_finish() without check rejoin_gather. I t...
05/09/2018
- 09:29 PM Bug #24072 (Resolved): mds: race with new session from connection and imported session
- ...
- 09:14 PM Feature #22446 (New): mds: ask idle client to trim more caps
- 09:11 PM Documentation #23611: doc: add description of new fs-client auth profile
- Blocked by resolution to #23751.
- 09:08 PM Feature #22370 (In Progress): cephfs: add kernel client quota support
- 09:08 PM Feature #22372: kclient: implement quota handling using new QuotaRealm
- Zheng, what's the status on those patches?
- 09:05 PM Bug #23332 (Need More Info): kclient: with fstab entry is not coming up reboot
- Zheng Yan wrote:
> kexec in dmesgs looks suspicious. client mounted cephfs, then used kexec to load kernel image aga... - 09:02 PM Bug #23350: mds: deadlock during unlink and export
- Well this is aggravating. I think it's time we plan evictions for clients that do not respond to cap release.
- 08:56 PM Bug #23394 (Rejected): nfs-ganesha: check cache configuration when exporting FSAL_CEPH
- 08:52 PM Feature #14456 (Fix Under Review): mon: prevent older/incompatible clients from mounting the file...
- https://github.com/ceph/ceph/pull/21885
- 07:06 PM Bug #23855: mds: MClientCaps should carry inode's dirstat
- Testing; will revert
- 06:57 PM Bug #23291 (Resolved): client: add way to sync setattr operations to MDS
- 06:57 PM Backport #23474 (Resolved): luminous: client: allow caller to request that setattr request be syn...
- 02:55 PM Backport #23474: luminous: client: allow caller to request that setattr request be synchronous
- Nathan Cutler wrote:
> https://github.com/ceph/ceph/pull/21109
merged - 06:56 PM Bug #23602 (Resolved): mds: handle client requests when mds is stopping
- 06:56 PM Backport #23632 (Resolved): luminous: mds: handle client requests when mds is stopping
- 02:54 PM Backport #23632: luminous: mds: handle client requests when mds is stopping
- Nathan Cutler wrote:
> https://github.com/ceph/ceph/pull/21346
merged - 06:56 PM Bug #23541 (Resolved): client: fix request send_to_auth was never really used
- 06:56 PM Backport #23635 (Resolved): luminous: client: fix request send_to_auth was never really used
- 02:54 PM Backport #23635: luminous: client: fix request send_to_auth was never really used
- Nathan Cutler wrote:
> https://github.com/ceph/ceph/pull/21354
merged - 06:13 PM Bug #24040 (Need More Info): mds: assert in CDir::_committed
- 02:14 PM Bug #24040: mds: assert in CDir::_committed
- Thanks for the report - it looks like you're using a 11.x ("kraken") version, which is no longer receiving bug fixes....
- 01:56 PM Bug #24039: MDSTableServer.cc: 62: FAILED assert(g_conf->mds_kill_mdstable_at != 1)
- the fsstress failure: http://pulpito.ceph.com/pdonnell-2018-05-04_03:45:51-multimds-master-testing-basic-smithi/24745...
- 01:12 PM Bug #24039: MDSTableServer.cc: 62: FAILED assert(g_conf->mds_kill_mdstable_at != 1)
- fsstress failure looks like new bug
pjd failure is similar to http://tracker.ceph.com/issues/23327
two dead tasks... - 11:55 AM Bug #23327: qa: pjd test sees wrong ctime after unlink
- http://pulpito.ceph.com/pdonnell-2018-05-04_03:45:51-multimds-master-testing-basic-smithi/2475062/
- 06:08 AM Backport #23951 (In Progress): luminous: mds: stuck during up:stopping
- https://github.com/ceph/ceph/pull/21901
- 03:36 AM Backport #23946: luminous: mds: crash when failover
- Opened backport PR#21900 (https://github.com/ceph/ceph/pull/21900). We need to cherry pick PR#21769 once it gets merg...
- 03:29 AM Backport #23950 (In Progress): luminous: mds: stopping rank 0 cannot shutdown until log is trimmed
- https://github.com/ceph/ceph/pull/21899
05/08/2018
- 10:49 PM Backport #24055 (In Progress): luminous: VolumeClient: allow ceph_volume_client to create 'volume...
- 10:45 PM Backport #24055 (Resolved): luminous: VolumeClient: allow ceph_volume_client to create 'volumes' ...
- https://github.com/ceph/ceph/pull/21897
- 10:43 PM Feature #23695 (Pending Backport): VolumeClient: allow ceph_volume_client to create 'volumes' wit...
- Mimic PR: https://github.com/ceph/ceph/pull/21896
- 10:37 PM Bug #24054 (Resolved): kceph: umount on evicted client blocks forever
- Failed test:
/ceph/teuthology-archive/pdonnell-2018-05-08_01:06:46-kcephfs-mimic-testing-basic-smithi/2494030/teut... - 10:33 PM Bug #24053 (Resolved): qa: kernel_mount.py umount must handle timeout arg
- ...
- 10:27 PM Bug #24052 (Resolved): repeated eviction of idle client until some IO happens
- We see repeated eviction of idle client sessions. We have client_reconnect_stale on the ceph-fuse clients, and these ...
- 08:56 PM Backport #24050 (Resolved): luminous: mds: MClientCaps should carry inode's dirstat
- https://github.com/ceph/ceph/pull/22118
- 08:56 PM Backport #24049 (Resolved): luminous: ceph-fuse: missing dentries in readdir result
- https://github.com/ceph/ceph/pull/22119
- 08:49 PM Bug #23530 (Resolved): mds: kicked out by monitor during rejoin
- 08:49 PM Backport #23636 (Resolved): luminous: mds: kicked out by monitor during rejoin
- 07:47 PM Backport #23636: luminous: mds: kicked out by monitor during rejoin
- Nathan Cutler wrote:
> https://github.com/ceph/ceph/pull/21366
merged - 08:49 PM Bug #23452 (Resolved): mds: assertion in MDSRank::validate_sessions
- 08:48 PM Backport #23637 (Resolved): luminous: mds: assertion in MDSRank::validate_sessions
- 07:46 PM Backport #23637: luminous: mds: assertion in MDSRank::validate_sessions
- Nathan Cutler wrote:
> https://github.com/ceph/ceph/pull/21372
merged - 08:48 PM Bug #23625 (Resolved): mds: sessions opened by journal replay do not get dirtied properly
- 08:48 PM Backport #23702 (Resolved): luminous: mds: sessions opened by journal replay do not get dirtied p...
- 07:46 PM Backport #23702: luminous: mds: sessions opened by journal replay do not get dirtied properly
- Nathan Cutler wrote:
> https://github.com/ceph/ceph/pull/21441
merged - 08:47 PM Bug #23582 (Resolved): MDSMonitor: mds health warnings printed in bad format
- 08:47 PM Backport #23703 (Resolved): luminous: MDSMonitor: mds health warnings printed in bad format
- 07:46 PM Backport #23703: luminous: MDSMonitor: mds health warnings printed in bad format
- Nathan Cutler wrote:
> https://github.com/ceph/ceph/pull/21447
merged - 08:47 PM Bug #23380 (Resolved): mds: ceph.dir.rctime follows dir ctime not inode ctime
- 08:47 PM Backport #23750 (Resolved): luminous: mds: ceph.dir.rctime follows dir ctime not inode ctime
- 07:45 PM Backport #23750: luminous: mds: ceph.dir.rctime follows dir ctime not inode ctime
- Nathan Cutler wrote:
> https://github.com/ceph/ceph/pull/21448
merged - 08:46 PM Bug #23764 (Resolved): MDSMonitor: new file systems are not initialized with the pending_fsmap epoch
- 08:46 PM Backport #23791 (Resolved): luminous: MDSMonitor: new file systems are not initialized with the p...
- 07:44 PM Backport #23791: luminous: MDSMonitor: new file systems are not initialized with the pending_fsma...
- Nathan Cutler wrote:
> https://github.com/ceph/ceph/pull/21512
merged - 08:46 PM Bug #23714 (Resolved): slow ceph_ll_sync_inode calls after setattr
- 08:45 PM Backport #23802 (Resolved): luminous: slow ceph_ll_sync_inode calls after setattr
- 07:44 PM Backport #23802: luminous: slow ceph_ll_sync_inode calls after setattr
- Nathan Cutler wrote:
> https://github.com/ceph/ceph/pull/21542
merged - 08:45 PM Bug #23652 (Resolved): client: fix gid_count check in UserPerm->deep_copy_from()
- 08:44 PM Backport #23771 (Resolved): luminous: client: fix gid_count check in UserPerm->deep_copy_from()
- 07:43 PM Backport #23771: luminous: client: fix gid_count check in UserPerm->deep_copy_from()
- Nathan Cutler wrote:
> https://github.com/ceph/ceph/pull/21596
merged - 08:44 PM Bug #23762 (Resolved): MDSMonitor: MDSMonitor::encode_pending modifies fsmap not pending_fsmap
- 08:43 PM Backport #23792 (Resolved): luminous: MDSMonitor: MDSMonitor::encode_pending modifies fsmap not p...
- 07:43 PM Backport #23792: luminous: MDSMonitor: MDSMonitor::encode_pending modifies fsmap not pending_fsmap
- Nathan Cutler wrote:
> https://github.com/ceph/ceph/pull/21732
merged - 08:43 PM Bug #23873 (Resolved): cephfs does not count st_nlink for directories correctly?
- 08:43 PM Bug #19706: Laggy mon daemons causing MDS failover (symptom: failed to set counters on mds daemon...
- I don't have reason to believe use of utime_t caused this issue but it's possible this could fix it: https://github.c...
- 08:43 PM Backport #23987 (Resolved): luminous: cephfs does not count st_nlink for directories correctly?
- 07:42 PM Backport #23987: luminous: cephfs does not count st_nlink for directories correctly?
- Nathan Cutler wrote:
> https://github.com/ceph/ceph/pull/21796
merged - 08:42 PM Bug #23880 (Resolved): mds: scrub code stuck at trimming log segments
- 08:42 PM Backport #23930 (Resolved): luminous: mds: scrub code stuck at trimming log segments
- 07:41 PM Backport #23930: luminous: mds: scrub code stuck at trimming log segments
- Nathan Cutler wrote:
> https://github.com/ceph/ceph/pull/21840
merged - 08:41 PM Bug #23813 (Resolved): client: "remove_session_caps still has dirty|flushing caps" when thrashing...
- 08:41 PM Backport #23934 (Resolved): luminous: client: "remove_session_caps still has dirty|flushing caps"...
- 07:41 PM Backport #23934: luminous: client: "remove_session_caps still has dirty|flushing caps" when thras...
- Nathan Cutler wrote:
> https://github.com/ceph/ceph/pull/21844
merged - 06:25 PM Bug #21777 (New): src/mds/MDCache.cc: 4332: FAILED assert(mds->is_rejoin())
- 01:36 PM Bug #21777 (Fix Under Review): src/mds/MDCache.cc: 4332: FAILED assert(mds->is_rejoin())
- 12:40 PM Bug #21777: src/mds/MDCache.cc: 4332: FAILED assert(mds->is_rejoin())
- -https://github.com/ceph/ceph/pull/21883-
- 03:57 AM Bug #21777 (In Progress): src/mds/MDCache.cc: 4332: FAILED assert(mds->is_rejoin())
- 06:23 PM Bug #24047 (Fix Under Review): MDCache.cc: 5317: FAILED assert(mds->is_rejoin())
- https://github.com/ceph/ceph/pull/21883
- 06:23 PM Bug #24047 (Resolved): MDCache.cc: 5317: FAILED assert(mds->is_rejoin())
- ...
- 04:58 PM Bug #24030: ceph-fuse: double dash meaning
- https://github.com/ceph/ceph/pull/21889
- 04:07 PM Bug #23885 (Resolved): MDSMonitor: overzealous MDS_ALL_DOWN and MDS_UP_LESS_THAN_MAX health warni...
- Mimic PR: https://github.com/ceph/ceph/pull/21888
- 02:39 PM Bug #24039: MDSTableServer.cc: 62: FAILED assert(g_conf->mds_kill_mdstable_at != 1)
- Right, there's something else wrong with the test.
- 01:35 PM Bug #24039: MDSTableServer.cc: 62: FAILED assert(g_conf->mds_kill_mdstable_at != 1)
- These are intentional crashes in table transaction test
- 07:10 AM Backport #23936 (In Progress): luminous: cephfs-journal-tool: segfault during journal reset
- https://github.com/ceph/ceph/pull/21874
05/07/2018
- 11:40 PM Bug #24040 (Need More Info): mds: assert in CDir::_committed
- ...
- 10:56 PM Bug #23894 (Pending Backport): ceph-fuse: missing dentries in readdir result
- Mimic: https://github.com/ceph/ceph/pull/21867
- 10:47 PM Bug #23855 (Pending Backport): mds: MClientCaps should carry inode's dirstat
- Mimic PR: https://github.com/ceph/ceph/pull/21866
- 10:40 PM Bug #21777: src/mds/MDCache.cc: 4332: FAILED assert(mds->is_rejoin())
- <deleted/>
- 08:54 PM Bug #21777 (New): src/mds/MDCache.cc: 4332: FAILED assert(mds->is_rejoin())
- Deleted: see #24047.
- 10:00 PM Bug #24039 (Closed): MDSTableServer.cc: 62: FAILED assert(g_conf->mds_kill_mdstable_at != 1)
- ...
- 02:41 PM Bug #24002 (Resolved): qa: check snap upgrade on multimds cluster
- 01:39 PM Bug #24030: ceph-fuse: double dash meaning
- Jos, please take a crack at fixing this. Thanks!
- 04:38 AM Bug #24030 (Closed): ceph-fuse: double dash meaning
- ...
- 01:37 PM Bug #23994 (Need More Info): mds: OSD space is not reclaimed until MDS is restarted
- 02:47 AM Bug #23994: mds: OSD space is not reclaimed until MDS is restarted
- please try again and dump mds' cache (ceph daemon mds.xxx dump cache /tmp/cachedump.x)
- 05:41 AM Backport #23934 (In Progress): luminous: client: "remove_session_caps still has dirty|flushing ca...
- https://github.com/ceph/ceph/pull/21844
- 04:36 AM Bug #23768 (Fix Under Review): MDSMonitor: uncommitted state exposed to clients/mdss
- https://github.com/ceph/ceph/pull/21842
- 04:02 AM Backport #23931 (In Progress): luminous: qa: test_purge_queue_op_rate: self.assertTrue(phase2_ops...
- https://github.com/ceph/ceph/pull/21841
- 03:59 AM Backport #23930 (In Progress): luminous: mds: scrub code stuck at trimming log segments
- https://github.com/ceph/ceph/pull/21840
05/06/2018
- 08:44 PM Bug #24028: CephFS flock() on a directory is broken
- I tested flock() logic on different hosts.
on one host:
flock my_dir sleep 1000
on second:
flock my_dir e... - 08:42 PM Bug #24028 (Resolved): CephFS flock() on a directory is broken
- Accroding to man page, flock() semantics must work also on directory. Actually it works with, say, Ext4. It does not ...
05/04/2018
- 02:12 PM Bug #23994: mds: OSD space is not reclaimed until MDS is restarted
- This was on the kernel client. I tried Ubuntu's 4.13.0-39-generic and 4.15.0-15-generic kernels.
With the fuse cli... - 01:44 PM Bug #23994: mds: OSD space is not reclaimed until MDS is restarted
- What client (kernel or fuse), and what version of the client?
- 05:09 AM Bug #23885 (Fix Under Review): MDSMonitor: overzealous MDS_ALL_DOWN and MDS_UP_LESS_THAN_MAX heal...
- https://github.com/ceph/ceph/pull/21810
- 02:07 AM Bug #24002 (Pending Backport): qa: check snap upgrade on multimds cluster
05/03/2018
- 10:33 PM Feature #23695: VolumeClient: allow ceph_volume_client to create 'volumes' without namespace isol...
- https://github.com/ceph/ceph/pull/21808
- 09:27 PM Bug #24004 (Resolved): mds: curate priority of perf counters sent to mgr
- Make sure we have the most interesting statisitcs available for prometheus for dashboard use. Additionally, see if we...
- 08:36 PM Bug #24002 (Fix Under Review): qa: check snap upgrade on multimds cluster
- https://github.com/ceph/ceph/pull/21805
- 08:35 PM Bug #24002 (Resolved): qa: check snap upgrade on multimds cluster
- To get an idea how the snap format upgrade works on a previously multimds cluster. (No need to exercise the two MDS s...
- 07:48 PM Cleanup #24001 (Resolved): MDSMonitor: remove vestiges of `mds deactivate`
- For Nautilus.
- 06:02 PM Backport #23946: luminous: mds: crash when failover
- Will also need: https://github.com/ceph/ceph/pull/21769
- 05:33 PM Feature #23623 (Resolved): mds: mark allow_snaps true by default
- 05:33 PM Documentation #23583 (Resolved): doc: update snapshot doc to account for recent changes
- 01:41 PM Backport #23987 (In Progress): luminous: cephfs does not count st_nlink for directories correctly?
- 10:28 AM Backport #23987 (Resolved): luminous: cephfs does not count st_nlink for directories correctly?
- https://github.com/ceph/ceph/pull/21796
- 01:27 PM Bug #23393 (Fix Under Review): ceph-ansible: update Ganesha config for nfs_file_gw to use optimal...
- 01:26 PM Bug #23393: ceph-ansible: update Ganesha config for nfs_file_gw to use optimal settings
- https://github.com/ceph/ceph-ansible/pull/2556
- 01:02 PM Bug #23994 (Need More Info): mds: OSD space is not reclaimed until MDS is restarted
- With my Luminous test cluster on Ubuntu I ran into a situation where I filled up an OSD by putting files on CephFS, a...
- 10:29 AM Backport #23991 (Resolved): luminous: client: hangs on umount if it had an MDS session evicted
- https://github.com/ceph/ceph/pull/22018
- 10:29 AM Backport #23990 (Rejected): jewel: client: hangs on umount if it had an MDS session evicted
- 10:28 AM Backport #23989 (Resolved): luminous: mds: don't report slow request for blocked filelock request
- https://github.com/ceph/ceph/pull/22782
follow-on fix: https://github.com/ceph/ceph/pull/26048 went into 12.2.11 - 10:27 AM Backport #23984 (Resolved): luminous: mds: scrub on fresh file system fails
- https://github.com/ceph/ceph/pull/21922
- 10:27 AM Backport #23982 (Resolved): luminous: qa: TestVolumeClient.test_lifecycle needs updated for new e...
- https://github.com/ceph/ceph/pull/21921
- 12:00 AM Bug #23958: mds: scrub doesn't always return JSON results
- Zheng Yan wrote:
> recursive scrub is async, it does not return anything
Good point, thanks. Even so, we shoudl r...
05/02/2018
- 11:56 PM Bug #16842 (Can't reproduce): mds: replacement MDS crashes on InoTable release
- 10:57 PM Bug #23975 (Pending Backport): qa: TestVolumeClient.test_lifecycle needs updated for new eviction...
- 07:53 PM Bug #23975 (Fix Under Review): qa: TestVolumeClient.test_lifecycle needs updated for new eviction...
- https://github.com/ceph/ceph/pull/21789
- 06:59 PM Bug #23975 (Resolved): qa: TestVolumeClient.test_lifecycle needs updated for new eviction behavior
- ...
- 08:50 PM Bug #23768 (New): MDSMonitor: uncommitted state exposed to clients/mdss
- Moving this back to fs. This is a different bug Josh.
- 08:44 PM Bug #23768 (Resolved): MDSMonitor: uncommitted state exposed to clients/mdss
- backport is tracked in the fs bug
- 06:06 PM Bug #23972 (New): Ceph MDS Crash from client mounting aufs over cephfs
-
Here is a rough outline of my topology
https://pastebin.com/HQqbMxyj
---
I can reliably crash all (in my case... - 05:02 PM Feature #17230 (In Progress): ceph_volume_client: py3 compatible
- 04:08 PM Bug #10915 (Pending Backport): client: hangs on umount if it had an MDS session evicted
- 02:21 PM Bug #23960 (Pending Backport): mds: scrub on fresh file system fails
- 02:20 PM Bug #23873 (Pending Backport): cephfs does not count st_nlink for directories correctly?
- 02:20 PM Bug #22428 (Pending Backport): mds: don't report slow request for blocked filelock request
- 03:10 AM Bug #23958: mds: scrub doesn't always return JSON results
- recursive scrub is async, it does not return anything
05/01/2018
- 10:42 PM Bug #23960 (In Progress): mds: scrub on fresh file system fails
- https://github.com/ceph/ceph/pull/21762
- 10:21 PM Bug #23960 (Resolved): mds: scrub on fresh file system fails
- In a fresh vstart cluster:...
- 04:01 PM Bug #23958 (Resolved): mds: scrub doesn't always return JSON results
- On a vstart cluster:...
- 06:52 AM Backport #23951 (Resolved): luminous: mds: stuck during up:stopping
- https://github.com/ceph/ceph/pull/21901
- 06:52 AM Backport #23950 (Resolved): luminous: mds: stopping rank 0 cannot shutdown until log is trimmed
- https://github.com/ceph/ceph/pull/21899
- 06:29 AM Bug #23826: mds: assert after daemon restart
- checking MDSMap::is_rejoining() is not required here. If there are recovering mds which haven't entered rejoin state....
- 12:29 AM Bug #23923 (Pending Backport): mds: stopping rank 0 cannot shutdown until log is trimmed
- 12:29 AM Bug #23919 (Pending Backport): mds: stuck during up:stopping
04/30/2018
- 09:05 PM Bug #23448 (Resolved): nfs-ganesha: fails to parse rados URLs with '.' in object name
- Yes.
- 08:51 PM Bug #23448: nfs-ganesha: fails to parse rados URLs with '.' in object name
- Is this resolved?
- 08:00 PM Backport #23946 (Resolved): luminous: mds: crash when failover
- https://github.com/ceph/ceph/pull/21900
- 07:21 PM Bug #23826: mds: assert after daemon restart
- Here's one possible way this could happen I think:
1. All MDS are rejoin or later.
2. A up:rejoin MDS does:
3.... - 07:00 PM Bug #23826: mds: assert after daemon restart
- Adding log from failed MDS.
Looks like it's receiving handle_cache_rejoin_ack message while in replay. - 06:53 PM Bug #23518 (Pending Backport): mds: crash when failover
- 01:43 PM Bug #23883: kclient: CephFS kernel client hang
- v4.9 is quite old at this point, so it would be helpful to know if this is something that has already been fixed in m...
- 06:54 AM Backport #23932 (In Progress): jewel: client: avoid second lock on client_lock
- 04:38 AM Backport #23792 (In Progress): luminous: MDSMonitor: MDSMonitor::encode_pending modifies fsmap no...
- https://github.com/ceph/ceph/pull/21732
- 03:59 AM Backport #23933 (In Progress): luminous: client: avoid second lock on client_lock
04/29/2018
- 08:31 PM Backport #23936 (Resolved): luminous: cephfs-journal-tool: segfault during journal reset
- https://github.com/ceph/ceph/pull/21874
- 08:30 PM Backport #23935 (Resolved): luminous: mds: may send LOCK_SYNC_MIX message to starting MDS
- https://github.com/ceph/ceph/pull/21990
- 08:30 PM Backport #23934 (Resolved): luminous: client: "remove_session_caps still has dirty|flushing caps"...
- https://github.com/ceph/ceph/pull/21844
- 08:30 PM Backport #23933 (Resolved): luminous: client: avoid second lock on client_lock
- https://github.com/ceph/ceph/pull/21730
- 08:30 PM Backport #23932 (Resolved): jewel: client: avoid second lock on client_lock
- https://github.com/ceph/ceph/pull/21734
- 08:30 PM Backport #23931 (Resolved): luminous: qa: test_purge_queue_op_rate: self.assertTrue(phase2_ops < ...
- https://github.com/ceph/ceph/pull/21841
- 08:30 PM Backport #23930 (Resolved): luminous: mds: scrub code stuck at trimming log segments
- https://github.com/ceph/ceph/pull/21840
- 08:07 PM Bug #23815 (Pending Backport): client: avoid second lock on client_lock
- 08:06 PM Bug #23813 (Pending Backport): client: "remove_session_caps still has dirty|flushing caps" when t...
- 08:06 PM Bug #23812 (Pending Backport): mds: may send LOCK_SYNC_MIX message to starting MDS
- 08:06 PM Bug #20549 (Pending Backport): cephfs-journal-tool: segfault during journal reset
- 08:05 PM Bug #23829 (Pending Backport): qa: test_purge_queue_op_rate: self.assertTrue(phase2_ops < phase1_...
- 08:05 PM Bug #23880 (Pending Backport): mds: scrub code stuck at trimming log segments
- 01:44 AM Bug #23919 (Fix Under Review): mds: stuck during up:stopping
- Zheng Yan wrote:
> I think we should call Locker::_readlock_kick in this case.
https://github.com/ceph/ceph/pull/... - 01:15 AM Bug #23927 (Rejected): qa: test_full failure in test_barrier
- https://github.com/ceph/ceph/pull/21668#pullrequestreview-116152567
- 12:54 AM Bug #23927: qa: test_full failure in test_barrier
- Here too: http://pulpito.ceph.com/pdonnell-2018-04-28_06:20:24-fs-wip-pdonnell-testing-20180428.041811-testing-basic-...
- 12:49 AM Bug #23927 (Rejected): qa: test_full failure in test_barrier
- ...
- 12:36 AM Bug #23923 (Fix Under Review): mds: stopping rank 0 cannot shutdown until log is trimmed
- https://github.com/ceph/ceph/pull/21719
04/28/2018
- 06:59 PM Bug #23923 (Resolved): mds: stopping rank 0 cannot shutdown until log is trimmed
- ...
- 03:53 PM Bug #23883: kclient: CephFS kernel client hang
- Hi Wei,
this is a very interesting problem, from your description, i would like to share my thought:
this shoul... - 10:10 AM Bug #23883: kclient: CephFS kernel client hang
- client kernel dmesg:...
- 10:09 AM Bug #23883: kclient: CephFS kernel client hang
- ...
- 08:02 AM Bug #23883: kclient: CephFS kernel client hang
- debug_mds = 10. only for period that mds is recovering
- 07:53 AM Bug #23883: kclient: CephFS kernel client hang
- Zheng Yan wrote:
> please upload mds log
which level?
after setting debug_mds = 20 and debug_ms = 1, log file is... - 05:03 AM Bug #23883: kclient: CephFS kernel client hang
- please upload mds log
- 10:34 AM Bug #22428 (Fix Under Review): mds: don't report slow request for blocked filelock request
- https://github.com/ceph/ceph/pull/21715
- 07:50 AM Bug #23919: mds: stuck during up:stopping
- I think we should call Locker::_readlock_kick in this case.
- 04:02 AM Bug #23919: mds: stuck during up:stopping
- /ceph/tmp/pdonnell/bz1566016/0x20000205a64.log.gz
holds the output of
zgrep -C5 0x20000205a64 ceph-mds.magna05... - 03:52 AM Bug #23919: mds: stuck during up:stopping
- crux of the issue appears to be here:...
- 06:34 AM Bug #23920: Multiple ceph-fuse and one ceph-client.admin.log
- I am using the method you said to modify,After that I found three questions:
1、when I run ceph-fuse ,There will be... - 06:04 AM Bug #23920: Multiple ceph-fuse and one ceph-client.admin.log
- because I have two ceph-fuse,how can i set config use the command as below for each ceph-fuse:
ceph daemon clien... - 05:00 AM Bug #23920 (Rejected): Multiple ceph-fuse and one ceph-client.admin.log
- config issue. you should set log file config option like
log file = /var/log/ceph/ceph-client.$pid.log - 02:28 AM Bug #23920 (Rejected): Multiple ceph-fuse and one ceph-client.admin.log
- I use the command as below:
/usr/bin/ceph-fuse -c /etc/ceph/ceph.conf /nas/test1 -r /test1
/usr/bin/ceph-fu... - 04:54 AM Bug #23894 (Fix Under Review): ceph-fuse: missing dentries in readdir result
- https://github.com/ceph/ceph/pull/21712
- 01:37 AM Bug #23894: ceph-fuse: missing dentries in readdir result
- libcephfs does not handle session stale message properly
steps to reproduce:
1. create two ceph-fuse mounts, mo...
04/27/2018
- 10:27 PM Bug #23919 (Resolved): mds: stuck during up:stopping
- ...
- 10:16 PM Backport #23833: luminous: MDSMonitor: crash after assigning standby-replay daemon in multifs setup
- I think they are separate issues but I will take a look.
- 07:42 PM Backport #23833: luminous: MDSMonitor: crash after assigning standby-replay daemon in multifs setup
- @Patrick - this one looks like it could benefit from being done in a single PR along with http://tracker.ceph.com/iss...
- 05:34 PM Backport #23833: luminous: MDSMonitor: crash after assigning standby-replay daemon in multifs setup
- Travis Nielsen wrote:
> What is the timeline for the backport? Rook would like to see it in 12.2.6. Thanks!
It sh... - 05:16 PM Backport #23833: luminous: MDSMonitor: crash after assigning standby-replay daemon in multifs setup
- What is the timeline for the backport? Rook would like to see it in 12.2.6. Thanks!
- 07:40 PM Backport #23792: luminous: MDSMonitor: MDSMonitor::encode_pending modifies fsmap not pending_fsmap
- @Patrick could you take this one?
- 05:05 PM Bug #23658: MDSMonitor: crash after assigning standby-replay daemon in multifs setup
- When this issue hits, is there a way to recover? For example, to forcefully remove the multiple filesystems that are ...
- 11:58 AM Bug #23873: cephfs does not count st_nlink for directories correctly?
- Peter Mauritius wrote:
> The Dovecot mail server does not work properly, if mailbox files are stored on cephfs and a... - 10:18 AM Bug #23883: kclient: CephFS kernel client hang
- Zheng Yan wrote:
> besides, 4.4/4.9 kernel is too old for using multimds.
It is very difficult to upgrade kernel ... - 03:55 AM Documentation #23897 (In Progress): doc: create snapshot user doc
- Include suggested upgrade procedure: https://github.com/ceph/ceph/pull/21374/commits/e05ebd08ea895626f4a2a52805f17e61...
- 12:50 AM Bug #23894 (Resolved): ceph-fuse: missing dentries in readdir result
- http://lists.ceph.com/pipermail/ceph-users-ceph.com/2018-April/026224.html
04/26/2018
- 11:54 PM Bug #23883: kclient: CephFS kernel client hang
- besides, 4.4/4.9 kernel is too old for using multimds.
- 11:47 PM Bug #23883: kclient: CephFS kernel client hang
- need mds log to check what happened
- 08:09 PM Bug #23883: kclient: CephFS kernel client hang
- 10:19 AM Bug #23883 (New): kclient: CephFS kernel client hang
- ceph: 12.2.4/12.2.5
os: debian jessie
kernel: 4.9/4.4
After restart all mds(6 in total, 5 active, 1 standby), cl... - 10:23 PM Backport #23638 (In Progress): luminous: ceph-fuse: getgroups failure causes exception
- 08:01 PM Bug #23421: ceph-fuse: stop ceph-fuse if no root permissions?
- Jos Collin wrote:
> The hang doesn't exist in the latest code.
>
> The following is my latest finding:
>
> [..... - 10:07 AM Bug #23421: ceph-fuse: stop ceph-fuse if no root permissions?
- The hang doesn't exist in the latest code.
The following is my latest finding:... - 05:20 PM Bug #23873: cephfs does not count st_nlink for directories correctly?
- The Dovecot mail server does not work properly, if mailbox files are stored on cephfs and a mailbox prefix is configu...
- 04:39 AM Bug #23873: cephfs does not count st_nlink for directories correctly?
- Zheng Yan wrote:
> If I remember right, this is not required by POSIX (btrfs does not do this). how NFS behaves depe... - 02:34 AM Bug #23873: cephfs does not count st_nlink for directories correctly?
- If I remember right, this is not required by POSIX (btrfs does not do this). how NFS behaves depends on the exported ...
- 11:10 AM Bug #23885 (Resolved): MDSMonitor: overzealous MDS_ALL_DOWN and MDS_UP_LESS_THAN_MAX health warni...
This is what we currently get when starting with vstart, which creates MDS daemons before creating the filesystem:
...- 10:35 AM Bug #23855 (Fix Under Review): mds: MClientCaps should carry inode's dirstat
- https://github.com/ceph/ceph/pull/21668
- 09:49 AM Bug #23880 (Fix Under Review): mds: scrub code stuck at trimming log segments
- https://github.com/ceph/ceph/pull/21664
- 07:49 AM Bug #23880 (Resolved): mds: scrub code stuck at trimming log segments
- /a/pdonnell-2018-04-25_18:15:51-kcephfs-wip-pdonnell-testing-20180425.144904-testing-basic-smithi/2439034
- 01:26 AM Feature #17854: mds: only evict an unresponsive client when another client wants its caps
- Rishabh Dave wrote:
> I am planning to start working on this feature. How can I get a client to be unresponsive with... - 12:49 AM Bug #23332: kclient: with fstab entry is not coming up reboot
- kexec in dmesgs looks suspicious. client mounted cephfs, then used kexec to load kernel image again. All issues happe...
04/25/2018
- 09:08 PM Feature #17854 (In Progress): mds: only evict an unresponsive client when another client wants it...
- 07:39 PM Feature #17854: mds: only evict an unresponsive client when another client wants its caps
- I am planning to start working on this feature. How can I get a client to be unresponsive without evicting it?
- 08:24 PM Bug #23873 (Fix Under Review): cephfs does not count st_nlink for directories correctly?
- https://github.com/ceph/ceph/pull/21652
- 07:42 PM Bug #23873 (Resolved): cephfs does not count st_nlink for directories correctly?
- Not sure if this behavior is by intention, but if you create a empty directory on cephfs and call stat on the directo...
- 06:09 PM Bug #23332: kclient: with fstab entry is not coming up reboot
- Luis Henriques wrote:
> Actually, the first failure seems to be a bit before:
> [...]
> The client seems to be try... - 04:58 PM Bug #23848 (Rejected): mds: stuck shutdown procedure
- 04:06 AM Bug #23848: mds: stuck shutdown procedure
- ...
- 04:00 AM Bug #23848 (Rejected): mds: stuck shutdown procedure
- The following outputs in an infinite loop:...
- 01:10 PM Bug #23855 (Resolved): mds: MClientCaps should carry inode's dirstat
- inode's dirstat gets updated by request reply, but not by cap message. this is problematic.
For example:
...
MDS... - 08:33 AM Bug #22428: mds: don't report slow request for blocked filelock request
- In case you need more examples, we're seeing this recently on 12.2.4:...
- 02:55 AM Bug #16842: mds: replacement MDS crashes on InoTable release
- make we should mark this as "need more info" or "can't reproduce"
- 02:08 AM Backport #23698: luminous: mds: load balancer fixes
- https://github.com/ceph/ceph/pull/21412
04/24/2018
- 07:50 PM Bug #23829 (Fix Under Review): qa: test_purge_queue_op_rate: self.assertTrue(phase2_ops < phase1_...
- Zheng Yan wrote:
> It's test case issue. The test caused so much trouble. I'd like to drop/disable it
Agreed.
... - 12:24 PM Bug #23829: qa: test_purge_queue_op_rate: self.assertTrue(phase2_ops < phase1_ops * 1.25)
- It's test case issue. The test caused so much trouble. I'd like to drop/disable it
- 07:37 PM Bug #23837 (Fix Under Review): client: deleted inode's Bufferhead which was in STATE::Tx would le...
- 10:44 AM Bug #23837: client: deleted inode's Bufferhead which was in STATE::Tx would lead a assert fail
- fixed by: https://github.com/ceph/ceph/pull/21615
- 09:45 AM Bug #23837 (Resolved): client: deleted inode's Bufferhead which was in STATE::Tx would lead a ass...
- ...
- 07:07 PM Backport #23671 (In Progress): luminous: mds: MDBalancer using total (all time) request count in ...
- https://github.com/ceph/ceph/pull/21412/commits/1a5b7eaac572f1810d0453b053781e6bc8185dd2
- 06:55 PM Tasks #23844 (In Progress): client: break client_lock
- See past efforts on this. Matt Benjamin did some prototyping on Firefly. Those patches will likely be unusable but co...
- 11:19 AM Backport #23835 (In Progress): luminous: mds: fix occasional dir rstat inconsistency between mult...
- https://github.com/ceph/ceph/pull/21617
- 05:48 AM Backport #23835 (Resolved): luminous: mds: fix occasional dir rstat inconsistency between multi-M...
- https://github.com/ceph/ceph/pull/21617
- 11:10 AM Backport #23308 (In Progress): luminous: doc: Fix -d option in ceph-fuse doc
- 08:24 AM Bug #20549 (Fix Under Review): cephfs-journal-tool: segfault during journal reset
- https://github.com/ceph/ceph/pull/21610
- 07:09 AM Feature #23362: mds: add drop_cache command
- https://github.com/ceph/ceph/pull/21566
- 05:47 AM Backport #23834 (Rejected): jewel: MDSMonitor: crash after assigning standby-replay daemon in mul...
- 05:47 AM Backport #23833 (Resolved): luminous: MDSMonitor: crash after assigning standby-replay daemon in ...
- https://github.com/ceph/ceph/pull/22603
- 04:42 AM Bug #23567 (Resolved): MDSMonitor: successive changes to max_mds can allow hole in ranks
- 04:35 AM Bug #23538 (Pending Backport): mds: fix occasional dir rstat inconsistency between multi-MDSes
- 04:34 AM Bug #23658 (Pending Backport): MDSMonitor: crash after assigning standby-replay daemon in multifs...
- 04:33 AM Bug #23799 (Resolved): MDSMonitor: creates invalid transition from up:creating to up:shutdown
- 04:32 AM Bug #23800 (Resolved): MDSMonitor: setting fs down twice will wipe old_max_mds
04/23/2018
- 08:26 PM Bug #23829 (Resolved): qa: test_purge_queue_op_rate: self.assertTrue(phase2_ops < phase1_ops * 1.25)
- ...
- 07:51 PM Bug #20549: cephfs-journal-tool: segfault during journal reset
- Another: http://pulpito.ceph.com/pdonnell-2018-04-23_17:22:02-kcephfs-wip-pdonnell-testing-20180423.033341-testing-ba...
- 05:51 PM Bug #23814 (Rejected): mds: newly active mds aborts may abort in handle_file_lock
- 08:40 AM Bug #23814: mds: newly active mds aborts may abort in handle_file_lock
- I think this is related to #23812. The patch for #23812 makes mds skip sending lock message to 'starting' mds. The sk...
- 05:50 PM Bug #23812: mds: may send LOCK_SYNC_MIX message to starting MDS
- https://github.com/ceph/ceph/pull/21601
- 05:10 PM Backport #22860 (Resolved): luminous: osdc: "FAILED assert(bh->last_write_tid > tid)" in powercyc...
- Looks like a different assertion so perhaps a new bug. I'll create a separate issue for this.
- 03:49 PM Backport #22860 (In Progress): luminous: osdc: "FAILED assert(bh->last_write_tid > tid)" in power...
- ...
- 03:36 PM Backport #23151 (In Progress): luminous: doc: update ceph-fuse with FUSE options
- 01:54 PM Bug #23826 (Duplicate): mds: assert after daemon restart
- ...
- 01:26 PM Backport #23475 (In Progress): luminous: ceph-fuse: trim ceph-fuse -V output
- 11:54 AM Backport #23771 (In Progress): luminous: client: fix gid_count check in UserPerm->deep_copy_from()
- 11:50 AM Backport #23771: luminous: client: fix gid_count check in UserPerm->deep_copy_from()
- https://github.com/ceph/ceph/pull/21596
- 10:45 AM Bug #23813 (Fix Under Review): client: "remove_session_caps still has dirty|flushing caps" when t...
- https://github.com/ceph/ceph/pull/21593
- 08:53 AM Bug #23518 (Fix Under Review): mds: crash when failover
- https://github.com/ceph/ceph/pull/21592
- 07:49 AM Bug #23815: client: avoid second lock on client_lock
- supriti singh wrote:
> supriti singh wrote:
> > In function ll_get_stripe_osd client_lock is taken. But its acquire... - 03:52 AM Backport #23818 (In Progress): luminous: client: add option descriptions and review levels (e.g. ...
- https://github.com/ceph/ceph/pull/21589
04/21/2018
- 09:42 PM Backport #23818 (Resolved): luminous: client: add option descriptions and review levels (e.g. LEV...
- https://github.com/ceph/ceph/pull/21589
- 07:52 AM Bug #23815 (Fix Under Review): client: avoid second lock on client_lock
- 07:43 AM Bug #23815: client: avoid second lock on client_lock
- supriti singh wrote:
> In function ll_get_stripe_osd client_lock is taken. But its acquired again in ll_get_inodeno(... - 07:35 AM Bug #23815 (Resolved): client: avoid second lock on client_lock
- In function ll_get_stripe_osd client_lock is taken. But its acquired again in ll_get_inodeno(). Avoid double locking....
- 05:32 AM Bug #23814 (Rejected): mds: newly active mds aborts may abort in handle_file_lock
- ...
- 05:03 AM Bug #23813 (Resolved): client: "remove_session_caps still has dirty|flushing caps" when thrashing...
- While doing a simple copy of /usr with ceph-fuse and thrashing max_mds between 1 and 2, I got these errors from ceph-...
- 12:30 AM Bug #23812 (Fix Under Review): mds: may send LOCK_SYNC_MIX message to starting MDS
- -https://github.com/ceph/ceph/pull/21577-
- 12:28 AM Bug #23812 (Resolved): mds: may send LOCK_SYNC_MIX message to starting MDS
- From mds.0:...
04/20/2018
- 04:36 PM Feature #14456: mon: prevent older/incompatible clients from mounting the file system
- Pre-mimic clients, yes.
- 06:58 AM Feature #14456: mon: prevent older/incompatible clients from mounting the file system
- prevent pre-luminous client to connect?
- 04:33 PM Bug #21848 (Fix Under Review): client: re-expand admin_socket metavariables in child process
- 03:52 AM Bug #21848: client: re-expand admin_socket metavariables in child process
- https://github.com/ceph/ceph/pull/21544
Patrick, could you pls take a look at this new fix? Now it is not only for... - 11:41 AM Bug #23518 (In Progress): mds: crash when failover
- 08:31 AM Bug #23518: mds: crash when failover
- This one is related to http://tracker.ceph.com/issues/23503. #23503 can explain why session was evicted
- 07:29 AM Bug #23327: qa: pjd test sees wrong ctime after unlink
- should close this if it does not happen again
- 05:52 AM Documentation #23583 (In Progress): doc: update snapshot doc to account for recent changes
- by commit "mds: update dev document of cephfs snapshot" in RP https://github.com/ceph/ceph/pull/21374
- 02:31 AM Backport #23802 (In Progress): luminous: slow ceph_ll_sync_inode calls after setattr
- https://github.com/ceph/ceph/pull/21542
04/19/2018
- 11:05 PM Bug #23755 (Resolved): qa: FAIL: test_purge_queue_op_rate (tasks.cephfs.test_strays.TestStrays)
- 10:12 PM Fix #4708 (Rejected): MDS: journaler pre-zeroing is dangerous
- Thanks for explaining Zheng. Closing this.
- 01:44 PM Fix #4708 (Need More Info): MDS: journaler pre-zeroing is dangerous
- I don't think it's still a problem. new mds takes over a rank after it see old mds is blacklisted in osdmap. There is...
- 10:03 PM Backport #23790: luminous: mds: crash during shutdown_pass
- Please just remove the global_snaprealm part of the backport.
- 10:50 AM Backport #23790 (Need More Info): luminous: mds: crash during shutdown_pass
- To backport this PR, we need complete PR#16779 (https://github.com/ceph/ceph/pull/16779) having changes related to mu...
- 05:25 AM Backport #23790 (Resolved): luminous: mds: crash during shutdown_pass
- https://github.com/ceph/ceph/pull/23015
- 10:00 PM Bug #22933 (Pending Backport): client: add option descriptions and review levels (e.g. LEVEL_DEV)
- 08:06 PM Backport #23802 (Resolved): luminous: slow ceph_ll_sync_inode calls after setattr
- https://github.com/ceph/ceph/pull/21542
- 06:59 PM Bug #23800 (Fix Under Review): MDSMonitor: setting fs down twice will wipe old_max_mds
- https://github.com/ceph/ceph/pull/21536
- 06:42 PM Bug #23800 (Resolved): MDSMonitor: setting fs down twice will wipe old_max_mds
- 06:49 PM Bug #23799 (Fix Under Review): MDSMonitor: creates invalid transition from up:creating to up:shut...
- https://github.com/ceph/ceph/pull/21535
- 06:36 PM Bug #23799 (Resolved): MDSMonitor: creates invalid transition from up:creating to up:shutdown
- ...
- 06:11 PM Bug #23714 (Pending Backport): slow ceph_ll_sync_inode calls after setattr
- 03:31 PM Bug #23797 (Can't reproduce): qa: cluster [WRN] Health check failed: 1 osds down (OSD_DOWN)
- This is v12.2.5 QE validation
Run: http://pulpito.ceph.com/yuriw-2018-04-17_21:20:41-knfs-luminous-testing-basic-s... - 09:52 AM Bug #23332: kclient: with fstab entry is not coming up reboot
- Actually, the first failure seems to be a bit before:...
- 08:12 AM Backport #23791 (In Progress): luminous: MDSMonitor: new file systems are not initialized with th...
- https://github.com/ceph/ceph/pull/21512
- 05:25 AM Backport #23791 (Resolved): luminous: MDSMonitor: new file systems are not initialized with the p...
- https://github.com/ceph/ceph/pull/21512
- 05:25 AM Backport #23792 (Resolved): luminous: MDSMonitor: MDSMonitor::encode_pending modifies fsmap not p...
- https://github.com/ceph/ceph/pull/21732
- 03:24 AM Bug #23658 (Fix Under Review): MDSMonitor: crash after assigning standby-replay daemon in multifs...
- 02:48 AM Bug #23658: MDSMonitor: crash after assigning standby-replay daemon in multifs setup
- https://github.com/ceph/ceph/pull/21510
04/18/2018
- 09:43 PM Feature #20606 (Resolved): mds: improve usability of cluster rank manipulation and setting cluste...
- 09:42 PM Subtask #20864 (Resolved): kill allow_multimds
- 09:42 PM Feature #20610 (Resolved): MDSMonitor: add new command to shrink the cluster in an automated way
- 09:41 PM Feature #20608 (Resolved): MDSMonitor: rename `ceph fs set <fs_name> cluster_down` to `ceph fs se...
- 09:41 PM Feature #20609 (Resolved): MDSMonitor: add new command `ceph fs set <fs_name> down` to bring the ...
- 09:40 PM Bug #23764 (Pending Backport): MDSMonitor: new file systems are not initialized with the pending_...
- 09:39 PM Bug #23766 (Pending Backport): mds: crash during shutdown_pass
- 09:38 PM Bug #23762 (Pending Backport): MDSMonitor: MDSMonitor::encode_pending modifies fsmap not pending_...
- 06:36 PM Feature #3244 (New): qa: integrate Ganesha into teuthology testing to regularly exercise Ganesha ...
- Jeff, fixed the wording to be clear.
- 06:14 PM Feature #3244 (Rejected): qa: integrate Ganesha into teuthology testing to regularly exercise Gan...
- I'm going to suggest that we just close this bug. We're doing this as a matter of course with the current work to cle...
- 05:43 PM Bug #23421 (Need More Info): ceph-fuse: stop ceph-fuse if no root permissions?
- Jos, please get hte client logs so we can diagnose.
- 01:01 PM Bug #23714: slow ceph_ll_sync_inode calls after setattr
- Thanks, dongdong! That seems to resolve the problem. Pull request is up here:
https://github.com/ceph/ceph/pull/21... - 11:53 AM Backport #23770 (In Progress): luminous: ceph-fuse: return proper exit code
- https://github.com/ceph/ceph/pull/21495
- 08:49 AM Documentation #23775: PendingReleaseNotes: add notes for major Mimic features
- FYI: https://github.com/ceph/ceph/pull/21374 already include mds upgrade process
04/17/2018
- 11:10 PM Documentation #23775 (Resolved): PendingReleaseNotes: add notes for major Mimic features
- mds upgrade process, snapshots, kernel quotas, etc.
- 11:08 PM Bug #23421 (Closed): ceph-fuse: stop ceph-fuse if no root permissions?
- 08:31 AM Bug #23421: ceph-fuse: stop ceph-fuse if no root permissions?
- Patrick Donnelly wrote:
> Jos, can you get more detailed debug logs when this happens? It is probably not related to... - 07:00 PM Backport #23771 (Resolved): luminous: client: fix gid_count check in UserPerm->deep_copy_from()
- https://github.com/ceph/ceph/pull/21596
- 07:00 PM Backport #23770 (Resolved): luminous: ceph-fuse: return proper exit code
- https://github.com/ceph/ceph/pull/21495
- 04:29 PM Bug #23755 (Fix Under Review): qa: FAIL: test_purge_queue_op_rate (tasks.cephfs.test_strays.TestS...
- 04:29 PM Bug #23755 (Pending Backport): qa: FAIL: test_purge_queue_op_rate (tasks.cephfs.test_strays.TestS...
- 01:11 PM Bug #23755 (Fix Under Review): qa: FAIL: test_purge_queue_op_rate (tasks.cephfs.test_strays.TestS...
- https://github.com/ceph/ceph/pull/21472
- 04:16 PM Bug #23768 (Resolved): MDSMonitor: uncommitted state exposed to clients/mdss
- e.g.
https://github.com/ceph/ceph/pull/21458#discussion_r182041693
and
https://github.com/ceph/ceph/pull/214... - 02:08 PM Backport #23704 (In Progress): luminous: ceph-fuse: broken directory permission checking
- https://github.com/ceph/ceph/pull/21475
- 01:51 PM Bug #23665: ceph-fuse: return proper exit code
- backporter note: please include https://github.com/ceph/ceph/pull/21473
- 11:59 AM Feature #23623 (Fix Under Review): mds: mark allow_snaps true by default
- by one commit in https://github.com/ceph/ceph/pull/21374
- 03:50 AM Bug #23762 (Fix Under Review): MDSMonitor: MDSMonitor::encode_pending modifies fsmap not pending_...
- https://github.com/ceph/ceph/pull/21458
- 01:50 AM Bug #23762: MDSMonitor: MDSMonitor::encode_pending modifies fsmap not pending_fsmap
- 03:16 AM Bug #23766 (Fix Under Review): mds: crash during shutdown_pass
- https://github.com/ceph/ceph/pull/21457
- 03:13 AM Bug #23766 (Resolved): mds: crash during shutdown_pass
- ...
- 01:49 AM Bug #23764 (Fix Under Review): MDSMonitor: new file systems are not initialized with the pending_...
- https://github.com/ceph/ceph/pull/21456
- 01:41 AM Bug #23764 (In Progress): MDSMonitor: new file systems are not initialized with the pending_fsmap...
- 01:41 AM Bug #23764 (Resolved): MDSMonitor: new file systems are not initialized with the pending_fsmap epoch
- Problem here: https://github.com/ceph/ceph/blob/60e8a63fdc21654d7f199a67f3f410a9e33c8134/src/mds/FSMap.cc#L234
FSM...
04/16/2018
- 08:10 PM Bug #23762 (In Progress): MDSMonitor: MDSMonitor::encode_pending modifies fsmap not pending_fsmap
- 08:07 PM Bug #23762 (Resolved): MDSMonitor: MDSMonitor::encode_pending modifies fsmap not pending_fsmap
- https://github.com/ceph/ceph/blob/60e8a63fdc21654d7f199a67f3f410a9e33c8134/src/mon/MDSMonitor.cc#L162-L166
- 02:13 PM Backport #23750 (In Progress): luminous: mds: ceph.dir.rctime follows dir ctime not inode ctime
- https://github.com/ceph/ceph/pull/21448
- 02:09 PM Backport #23703 (In Progress): luminous: MDSMonitor: mds health warnings printed in bad format
- https://github.com/ceph/ceph/pull/21447
- 12:20 PM Backport #23703: luminous: MDSMonitor: mds health warnings printed in bad format
- I'm on it.
- 01:50 PM Bug #23714: slow ceph_ll_sync_inode calls after setattr
- I tried to roll a standalone testcase for this, but it didn't stall out in the same way. I'm not quite sure what caus...
- 11:36 AM Feature #21156: mds: speed up recovery with many open inodes
- Patrick Donnelly wrote:
> Very unlikely because of the new structure in the metadata pool adds unacceptable risk for... - 10:44 AM Backport #23702 (In Progress): luminous: mds: sessions opened by journal replay do not get dirtie...
- https://github.com/ceph/ceph/pull/21441
- 03:29 AM Bug #23652 (Pending Backport): client: fix gid_count check in UserPerm->deep_copy_from()
- 03:29 AM Bug #23665 (Pending Backport): ceph-fuse: return proper exit code
- 03:25 AM Bug #23755 (Resolved): qa: FAIL: test_purge_queue_op_rate (tasks.cephfs.test_strays.TestStrays)
- ...
- 03:16 AM Bug #22933 (Fix Under Review): client: add option descriptions and review levels (e.g. LEVEL_DEV)
- https://github.com/ceph/ceph/pull/21434
04/15/2018
- 06:30 PM Bug #23751: mon: use fs-client profile for fs authorize mon caps
- Actually I think this gives blanket permission to read from OSDs for all pools so we may actually want to remove this...
- 06:27 PM Bug #23751 (New): mon: use fs-client profile for fs authorize mon caps
- This is simpler and consistent.
- 05:40 PM Backport #23750 (Resolved): luminous: mds: ceph.dir.rctime follows dir ctime not inode ctime
- https://github.com/ceph/ceph/pull/21448
- 05:39 PM Bug #23724 (Resolved): qa: broad snapshot functionality testing across clients
- Ganesha FSAL
ceph-fuse
kclient - 02:02 PM Bug #23714: slow ceph_ll_sync_inode calls after setattr
- I think patrick is right, maybe we should call flush_mdlog_sync to make mds do the mdlog flush before we wait on the ...
- 03:35 AM Bug #23714: slow ceph_ll_sync_inode calls after setattr
- Sounds like Ganesha is blocked on a journal flush by the MDS.
- 02:50 AM Bug #23723 (New): qa: incorporate smallfile workload
- Add smallfile workload
https://github.com/distributed-system-analysis/smallfile
to fs:workloads suite.
04/14/2018
- 03:30 PM Bug #23715: "Scrubbing terminated -- not all pgs were active and clean" in fs-jewel-distro-basic-ovh
- Patrick, any requirements against running k* suites on ovh ?
- 07:10 AM Bug #23715: "Scrubbing terminated -- not all pgs were active and clean" in fs-jewel-distro-basic-ovh
- Yuri, I have not seen this on smithi, so I assume it only happens in virtual environments.
- 12:27 AM Cleanup #23718 (Resolved): qa: merge fs/kcephfs suites
- and remove redundant tests (e.g. inline on/off with administrative tests like changing max_mds).
- 12:15 AM Cleanup #23717 (New): cephfs: consider renaming max_mds to a better name
- It is no longer considered a "max" and having fewer ranks than max_mds is considered a bad configuration which genera...
- 12:12 AM Bug #23567 (Fix Under Review): MDSMonitor: successive changes to max_mds can allow hole in ranks
- https://github.com/ceph/ceph/pull/16608
QA and fix here.
04/13/2018
- 09:53 PM Bug #23715: "Scrubbing terminated -- not all pgs were active and clean" in fs-jewel-distro-basic-ovh
- Nathan, I am not sure if you have seen this.
Suspect also in http://pulpito.ceph.com/teuthology-2018-04-11_04:15:02-... - 09:52 PM Bug #23715 (Closed): "Scrubbing terminated -- not all pgs were active and clean" in fs-jewel-dist...
- Run: http://pulpito.ceph.com/teuthology-2018-04-11_04:10:03-fs-jewel-distro-basic-ovh/
Jobs: 40 jobs
Logs: teutholo... - 09:40 PM Feature #21156: mds: speed up recovery with many open inodes
- Webert Lima wrote:
> Hi, thank you very much for this.
>
> I see this
> > Target version: Ceph - v13.0.0
>
> ... - 08:33 PM Feature #21156: mds: speed up recovery with many open inodes
- Hi, thank you very much for this.
I see this
> Target version: Ceph - v13.0.0
So I'm not even asking for a bac... - 09:32 PM Backport #23698 (In Progress): luminous: mds: load balancer fixes
- 09:07 AM Backport #23698: luminous: mds: load balancer fixes
- https://github.com/ceph/ceph/pull/21412
- 08:54 AM Backport #23698 (New): luminous: mds: load balancer fixes
- backport https://github.com/ceph/ceph/pull/19220 to luminous
- 08:41 AM Backport #23698 (Pending Backport): luminous: mds: load balancer fixes
- 02:26 AM Backport #23698 (Resolved): luminous: mds: load balancer fixes
- https://github.com/ceph/ceph/pull/21412
- 06:49 PM Bug #23714: slow ceph_ll_sync_inode calls after setattr
- I'll see if I can cook up libcephfs standalone testcase for this.
- 06:48 PM Bug #23714: slow ceph_ll_sync_inode calls after setattr
- We recently added some calls to ceph_ll_sync_inode in ganesha, to be done after a setattr request. Testing with cthon...
- 06:40 PM Bug #23714 (Resolved): slow ceph_ll_sync_inode calls after setattr
- 04:55 PM Bug #23697 (Pending Backport): mds: load balancer fixes
- 08:56 AM Bug #23697: mds: load balancer fixes
- Sorry for the confused edits. The "backport-create-issue" script is much better at this than I am. It's enough to cha...
- 08:36 AM Bug #23697 (New): mds: load balancer fixes
- 02:23 AM Bug #23697 (Resolved): mds: load balancer fixes
- https://github.com/ceph/ceph/pull/19220
- 09:48 AM Bug #21848: client: re-expand admin_socket metavariables in child process
- Hi Patrick,
Sorry for missing this for a long time. I will take a look recently to see whether there is a better f... - 08:35 AM Backport #23705 (Rejected): jewel: ceph-fuse: broken directory permission checking
- 08:35 AM Backport #23704 (Resolved): luminous: ceph-fuse: broken directory permission checking
- https://github.com/ceph/ceph/pull/21475
- 08:35 AM Backport #23703 (Resolved): luminous: MDSMonitor: mds health warnings printed in bad format
- https://github.com/ceph/ceph/pull/21447
- 08:34 AM Backport #23702 (Resolved): luminous: mds: sessions opened by journal replay do not get dirtied p...
- https://github.com/ceph/ceph/pull/21441
- 02:07 AM Feature #17434: qa: background rsync task for FS workunits
- Current work on this by Ramakrishnan: https://github.com/ceph/ceph/pull/12503
- 01:26 AM Bug #23509 (Pending Backport): ceph-fuse: broken directory permission checking
- 01:25 AM Bug #23582 (Pending Backport): MDSMonitor: mds health warnings printed in bad format
- 01:24 AM Bug #23625 (Pending Backport): mds: sessions opened by journal replay do not get dirtied properly
- 01:10 AM Feature #22417 (Resolved): support purge queue with cephfs-journal-tool
04/12/2018
- 10:50 PM Feature #20608 (In Progress): MDSMonitor: rename `ceph fs set <fs_name> cluster_down` to `ceph fs...
- Doug, thinking about this more, I'd like to keep "cluster_down" (as "joinable" or not) because it simplifies qa testi...
- 09:40 PM Feature #23695 (Resolved): VolumeClient: allow ceph_volume_client to create 'volumes' without nam...
- https://bugzilla.redhat.com/show_bug.cgi?id=1566194
to address the needs of
https://github.com/kubernetes-incub... - 07:39 PM Documentation #23568 (Resolved): doc: outline the steps for upgrading an MDS cluster
- 07:39 PM Backport #23634 (Resolved): luminous: doc: outline the steps for upgrading an MDS cluster
- 07:27 PM Backport #23634: luminous: doc: outline the steps for upgrading an MDS cluster
- Nathan Cutler wrote:
> https://github.com/ceph/ceph/pull/21352
merged - 07:19 PM Bug #23665 (Fix Under Review): ceph-fuse: return proper exit code
- https://github.com/ceph/ceph/pull/21396
- 12:53 AM Bug #23665 (Resolved): ceph-fuse: return proper exit code
- from mailling list...
- 05:00 PM Feature #23689 (New): qa: test major/minor version upgrades
- We should verify the upgrade process [1] works and that older clients with parallel I/O still function correctly.
... - 10:19 AM Backport #23637 (In Progress): luminous: mds: assertion in MDSRank::validate_sessions
- https://github.com/ceph/ceph/pull/21372
- 04:17 AM Backport #23636 (In Progress): luminous: mds: kicked out by monitor during rejoin
- https://github.com/ceph/ceph/pull/21366
- 01:35 AM Backport #23671 (Resolved): luminous: mds: MDBalancer using total (all time) request count in loa...
- https://github.com/ceph/ceph/pull/21412
- 01:34 AM Backport #23669 (Resolved): luminous: doc: create doc outlining steps to bring down cluster
- https://github.com/ceph/ceph/pull/22872
04/11/2018
- 07:05 PM Feature #20611 (New): MDSMonitor: do not show cluster health warnings for file system intentional...
- Doug, I was just thinking about this and a valid reason to not want a HEALTH_ERR is if you have dozens or hundreds of...
- 01:42 PM Feature #20611 (Fix Under Review): MDSMonitor: do not show cluster health warnings for file syste...
- 01:42 PM Feature #20611: MDSMonitor: do not show cluster health warnings for file system intentionally mar...
- See https://github.com/ceph/ceph/pull/16608, which implements the opposite of this behavior. Whenever a filesystem is...
- 06:54 PM Feature #20607 (Rejected): MDSMonitor: change "mds deactivate" to clearer "mds rejoin"
- This is rejected in favor of removing `mds deactivate`.
- 05:31 PM Documentation #23427 (Pending Backport): doc: create doc outlining steps to bring down cluster
- 05:29 PM Bug #23658 (Resolved): MDSMonitor: crash after assigning standby-replay daemon in multifs setup
- From: https://github.com/rook/rook/issues/1027...
- 05:19 PM Bug #23567: MDSMonitor: successive changes to max_mds can allow hole in ranks
- Doug, I tested with master but I believe it also happened with your PR. I can't remember.
- 03:02 PM Bug #23567 (Need More Info): MDSMonitor: successive changes to max_mds can allow hole in ranks
- Was this before or after https://github.com/ceph/ceph/pull/16608 ?
- 03:14 PM Bug #23652 (Fix Under Review): client: fix gid_count check in UserPerm->deep_copy_from()
- https://github.com/ceph/ceph/pull/21341
- 03:10 PM Bug #23652 (Resolved): client: fix gid_count check in UserPerm->deep_copy_from()
- Fix gid_count check in UserPerm->deep_copy_from(). Allocate gids only if gid_count > 0.
- 02:34 PM Backport #23635 (In Progress): luminous: client: fix request send_to_auth was never really used
- https://github.com/ceph/ceph/pull/21354
- 01:46 PM Feature #20609 (In Progress): MDSMonitor: add new command `ceph fs set <fs_name> down` to bring t...
- https://github.com/ceph/ceph/pull/16608 overhauls this behavior, and re-implements the cluster_down flag for this fun...
- 01:45 PM Feature #20606 (Fix Under Review): mds: improve usability of cluster rank manipulation and settin...
- https://github.com/ceph/ceph/pull/16608
- 01:44 PM Feature #20610 (Fix Under Review): MDSMonitor: add new command to shrink the cluster in an automa...
- https://github.com/ceph/ceph/pull/16608
- 01:43 PM Feature #20608 (Rejected): MDSMonitor: rename `ceph fs set <fs_name> cluster_down` to `ceph fs se...
- This behavior is overhauled in https://github.com/ceph/ceph/pull/16608 .
- 01:30 PM Backport #23634 (In Progress): luminous: doc: outline the steps for upgrading an MDS cluster
- https://github.com/ceph/ceph/pull/21352
- 01:30 PM Bug #23393 (Resolved): ceph-ansible: update Ganesha config for nfs_file_gw to use optimal settings
- 01:26 PM Bug #23643 (Resolved): qa: osd_mon_report_interval typo in test_full.py
- 10:27 AM Backport #23632 (In Progress): luminous: mds: handle client requests when mds is stopping
- https://github.com/ceph/ceph/pull/21346
- 09:09 AM Backport #23632: luminous: mds: handle client requests when mds is stopping
- I am on it.
- 05:14 AM Bug #21745 (Pending Backport): mds: MDBalancer using total (all time) request count in load stati...
- 01:00 AM Feature #22372: kclient: implement quota handling using new QuotaRealm
- by following commits in testing branch
ceph: quota: report root dir quota usage in statfs …
ceph: quota: add cou... - 12:47 AM Bug #18730 (Closed): mds: backtrace issues getxattr for every file with cap on rejoin
- should be resolved by open file table https://github.com/ceph/ceph/pull/20132
- 12:44 AM Fix #5268 (Closed): mds: fix/clean up file size/mtime recovery code
- current code does parallel object checks.
- 12:39 AM Bug #4212 (Closed): mds: open_snap_parents isn't called all the times it needs to be
- with the new snaprealm format, there is no need to open past parent
- 12:37 AM Bug #21412 (Closed): cephfs: too many cephfs snapshots chokes the system
- 12:37 AM Bug #21412: cephfs: too many cephfs snapshots chokes the system
- this is actually osd issue. I talk to josh at cephalocon. He said it has already been fixed
Also available in: Atom