Activity
From 04/13/2016 to 05/12/2016
05/12/2016
- 09:39 PM Fix #5729 (Resolved): ceph-fuse: fix getgroups_cb and gids
- commit:0eb6d0ba1b65ac0aaffc72a82f07857cc65a6b88 fixes it.
- 07:15 PM Bug #15783: client: enable acls by default
- Is there some reason we shouldn't make those the default behaviors at this point?
- 10:53 AM Bug #15862 (Resolved): hadoop test failing on "assert len(masters.remotes) == 1"
- http://qa-proxy.ceph.com/teuthology/teuthology-2016-05-09_22:12:02-hadoop-master---basic-mira/181820/teuthology.log
... - 10:24 AM Bug #15857 (Fix Under Review): qa: test_sessionmap::_get_thread_conn sometimes get short ps outpu...
- https://github.com/ceph/ceph-qa-suite/pull/991
- 10:13 AM Bug #15857: qa: test_sessionmap::_get_thread_conn sometimes get short ps output and fails
- Ah, this will have been broken by Josh's multi-cluster support ("--cluster ceph" in the command line).
05/11/2016
- 07:24 PM Bug #15465 (Resolved): MDSAuthCap parse fails on paths with hyphens
- 07:24 PM Backport #15731 (Resolved): jewel: MDSAuthCap parse fails on paths with hyphens
- 07:24 PM Bug #15591 (Resolved): infernalis to jewel upgrade
- 07:23 PM Bug #15399 (Resolved): MDS incarnation get lost after remove filesystem
- 07:23 PM Backport #15732 (Resolved): jewel: MDS incarnation get lost after remove filesystem
- 07:21 PM Backport #15738 (Resolved): jewel: infernalis to jewel upgrade
- 06:55 PM Bug #15857 (Resolved): qa: test_sessionmap::_get_thread_conn sometimes get short ps output and fails
- http://qa-proxy.ceph.com/teuthology/yuriw-2016-05-10_10:57:23-fs-wip-yuri-testing2-distro-basic-smithi/183134/teuthol...
- 06:49 PM Backport #15855 (Resolved): jewel: CephFSVolumeClient.evict should be limited by path, not just a...
- https://github.com/ceph/ceph/pull/9430
- 06:49 PM Backport #15854 (Resolved): jewel: Make path prefix ("/volumes") in CephFSVolumeClient configurable
- https://github.com/ceph/ceph/pull/9430
- 03:02 PM Feature #15417 (Pending Backport): Make path prefix ("/volumes") in CephFSVolumeClient configurable
- Merged to master.
Would like to backport this to jewel. Although it's not a bugfix, it's aimed at would-be Manila... - 12:24 PM Bug #15045 (Pending Backport): CephFSVolumeClient.evict should be limited by path, not just auth ID
- Merged to master
Needs backport to Jewel... - 08:54 AM Backport #15631 (Resolved): jewel: CephFSVolumeClient should isolate volumes by RADOS namespace
05/10/2016
- 10:19 AM Feature #9903 (Resolved): Recover lost dirfrag via data pool
- This is an old ticket describing one of the things that cephfs-data-scan does today
- 09:58 AM Feature #15066: multifs: Allow filesystems to be assigned RADOS namespace as well as pool for met...
- Note: once we do this, we have to start using two separate messengers in the MDS for rados client vs. cephfs server (...
- 09:22 AM Feature #12258 (Resolved): asok command to initate tagged forward scrub
- ...
- 09:19 AM Feature #12133 (Resolved): cephfs-data-scan: Filter on inodes not touched by forward scrub
- This was done for jewel.
- 09:16 AM Feature #12145 (Resolved): cephfs-data-scan: pgls filter for 0th file objects
- This was done for jewel.
- 02:04 AM Bug #15783: client: enable acls by default
- To enable ACL support, you need to add "--fuse_default_permission=0 --client_acl_type=posix_acl" options to ceph-fuse.
05/09/2016
- 06:05 PM Bug #15783 (New): client: enable acls by default
- I found while doing some SAMBA testing using Jewel on both a kernel mounted and fuse mounted Ceph File system that AC...
- 02:30 PM Bug #15508 (Fix Under Review): client: simultaneous readdirs are very racy
- last commit of https://github.com/ceph/ceph/pull/8739
- 12:19 PM Bug #15508: client: simultaneous readdirs are very racy
- I found that seekdir can also trigger this issue. I'm working on fixing it.
05/06/2016
- 02:31 PM Backport #15738 (In Progress): jewel: infernalis to jewel upgrade
- 02:28 PM Backport #15732 (In Progress): jewel: MDS incarnation get lost after remove filesystem
- 02:25 PM Backport #15731 (In Progress): jewel: MDSAuthCap parse fails on paths with hyphens
05/04/2016
- 06:33 PM Backport #15738 (Resolved): jewel: infernalis to jewel upgrade
- https://github.com/ceph/ceph/pull/8971
- 03:31 PM Feature #15614: CephFSVolumeClient: read-only authorization for volumes
- 03:28 PM Feature #15614: CephFSVolumeClient: read-only authorization for volumes
- https://github.com/ceph/ceph/pull/8929
https://github.com/ceph/ceph-qa-suite/pull/983 - 02:28 PM Bug #15591 (Pending Backport): infernalis to jewel upgrade
- 02:09 PM Backport #15732 (Resolved): jewel: MDS incarnation get lost after remove filesystem
- https://github.com/ceph/ceph/pull/8970
- 02:09 PM Backport #15731 (Resolved): jewel: MDSAuthCap parse fails on paths with hyphens
- https://github.com/ceph/ceph/pull/8969
- 10:50 AM Bug #15465 (Pending Backport): MDSAuthCap parse fails on paths with hyphens
- 10:49 AM Bug #15399 (Pending Backport): MDS incarnation get lost after remove filesystem
- 09:29 AM Bug #15723 (Fix Under Review): client: fstat cap release
- open PR https://github.com/ceph/ceph/pull/8925
- 04:47 AM Bug #15723 (Resolved): client: fstat cap release
- From mailing list. The patch below lets fstat take cap. Seems to work just fine with a single client (it gets the cap...
05/03/2016
- 12:16 PM Bug #15702 (Fix Under Review): mds: wrongly treat symlink inode as normal file/dir when symlink i...
- 09:47 AM Bug #15702: mds: wrongly treat symlink inode as normal file/dir when symlink inode is stale on kc...
- I have tried to backport Yan's following patch, which can only resolve this issue when symlink dentry is invalid, but...
- 09:41 AM Bug #15702: mds: wrongly treat symlink inode as normal file/dir when symlink inode is stale on kc...
- https://github.com/ceph/ceph/pull/8906
- 09:39 AM Bug #15702 (Resolved): mds: wrongly treat symlink inode as normal file/dir when symlink inode is ...
- This issue should happen on kcephfs only. My kernel version is 3.10.94. I think it might also happen on other 3.x ver...
- 12:03 PM Bug #15705 (Fix Under Review): ceph status mds output ignores active MDS when there is a standby ...
- https://github.com/ceph/ceph/pull/8908
- 12:01 PM Bug #15705 (Resolved): ceph status mds output ignores active MDS when there is a standby replay
- Because standby replay mds_info_t now has rank set for daemons in state standby replay, the by_rank dict in print_sum...
- 03:22 AM Feature #15634 (Fix Under Review): Enable fuse_use_invalidate_cb by default
- https://github.com/ceph/ceph/pull/8890
05/02/2016
- 09:25 PM Bug #15689 (Fix Under Review): Confusing MDS log message when shut down with stalled journaler reads
- https://github.com/ceph/ceph/pull/8887
- 09:21 PM Bug #15689 (Resolved): Confusing MDS log message when shut down with stalled journaler reads
- https://www.mail-archive.com/ceph-users@lists.ceph.com/msg28758.html
Sending kill to MDS while it is stuck trying ... - 06:24 PM Feature #15634: Enable fuse_use_invalidate_cb by default
- Drat, we haven't been testing with it on. That's annoying.
04/30/2016
- 06:22 PM Bug #15399 (Fix Under Review): MDS incarnation get lost after remove filesystem
- https://github.com/ceph/ceph/pull/8484
- 06:15 PM Bug #15591 (Fix Under Review): infernalis to jewel upgrade
- Added the fix for this to https://github.com/ceph/ceph/pull/8300
- 04:52 PM Backport #13932 (In Progress): hammer: kclient: intermittent log warnings "client.XXXX isn't resp...
04/28/2016
- 09:29 PM Bug #15591 (In Progress): infernalis to jewel upgrade
- 08:50 AM Backport #15512 (In Progress): hammer: Double decreased the count to trim caps which will cause f...
04/27/2016
- 09:03 PM Feature #15417 (Fix Under Review): Make path prefix ("/volumes") in CephFSVolumeClient configurable
- https://github.com/ceph/ceph/pull/8771
and test is
https://github.com/ceph/ceph-qa-suite/pull/975 - 08:59 PM Feature #15615: CephFSVolumeClient: List authorized IDs by share
- @Rama Raja: Would you mind if I could help on this feature?
- 08:04 PM Bug #15502: files read or written with cephfs (fuse or kernel) on client drop all their page cach...
- Created #15634 to enable the config value.
- 08:03 PM Feature #15634 (Resolved): Enable fuse_use_invalidate_cb by default
- Apparently we don't yet. Whoops! Make sure fuse_use_invalidate_cb=true has been in our testing for a while, then swit...
- 05:52 PM Backport #15631 (In Progress): jewel: CephFSVolumeClient should isolate volumes by RADOS namespace
- 05:30 PM Backport #15631: jewel: CephFSVolumeClient should isolate volumes by RADOS namespace
- https://github.com/ceph/ceph/pull/8787
- 03:40 PM Backport #15631 (Resolved): jewel: CephFSVolumeClient should isolate volumes by RADOS namespace
- https://github.com/ceph/ceph/pull/8787
- 02:31 PM Feature #15614 (In Progress): CephFSVolumeClient: read-only authorization for volumes
04/26/2016
- 08:29 PM Feature #15619 (Resolved): Repair InoTable during forward scrub
- If an inode's number is not marked as used in the InoTable, fix that during forward scrub when repair is enabled.
... - 08:12 PM Feature #15617 (New): CephFSVolumeClient: OSD blacklisting on deauthorize
This might just mean waiting for http://tracker.ceph.com/issues/9754, or we could implement it in CephFSVolumeClien...- 05:34 PM Feature #15615 (Resolved): CephFSVolumeClient: List authorized IDs by share
- For implementing update_access efficiently in Manila. Will require storing the per-share metadata on disk.
Relate... - 05:31 PM Feature #15614 (Resolved): CephFSVolumeClient: read-only authorization for volumes
I don't think we have tests for how well setting "r" mds auth caps works for clients, hopefully well enough that th...- 05:23 PM Feature #15400 (Pending Backport): CephFSVolumeClient should isolate volumes by RADOS namespace
- 05:17 PM Feature #15400: CephFSVolumeClient should isolate volumes by RADOS namespace
- Test is here,
https://github.com/ceph/ceph-qa-suite/pull/959 - 03:13 PM Bug #15502: files read or written with cephfs (fuse or kernel) on client drop all their page cach...
- Its interesting. The kernel changelogs reference the page cache invalidation change as
* Fri Mar 04 2016 Rafa... - 01:21 PM Bug #15502: files read or written with cephfs (fuse or kernel) on client drop all their page cach...
- I checked again. 3.10.0-327.el7.x86_64 does not contain the fix. 3.10.0-375.el7 kernel does.
- 01:05 PM Bug #15502: files read or written with cephfs (fuse or kernel) on client drop all their page cach...
- Greg Farnum wrote:
> That was just because our callback logic was broken though, right? I think it's time to enable ...
04/25/2016
- 10:35 PM Feature #15599 (Fix Under Review): quota: Generate client df from quota, when using subdirectory ...
- https://github.com/ceph/ceph/pull/8747
https://github.com/ceph/ceph-qa-suite/pull/974 - 10:31 PM Feature #15599 (Resolved): quota: Generate client df from quota, when using subdirectory mount
This is an enabler for Manila, creating this ticket to track backport to Jewel after landing in master.- 04:01 PM Bug #15591: infernalis to jewel upgrade
- ...
- 03:06 PM Bug #15591 (Resolved): infernalis to jewel upgrade
- ubuntu trusty
3.13.0-68-generic
# dpkg -l | egrep '(ceph|rbd)'
ii ceph 10.2.0-1... - 02:59 PM Bug #15502: files read or written with cephfs (fuse or kernel) on client drop all their page cach...
- I've verified by adding:
fuse_use_invalidate_cb=true
to the /etc/ceph/ceph.config file that mounting with fu... - 02:33 PM Bug #15502: files read or written with cephfs (fuse or kernel) on client drop all their page cach...
- That was just because our callback logic was broken though, right? I think it's time to enable by default (I think it...
- 09:52 AM Feature #15589: Ceph status outputs mds slow request summary for better monitoring
- https://github.com/ceph/ceph/pull/8737
- 09:48 AM Feature #15589 (Resolved): Ceph status outputs mds slow request summary for better monitoring
- Currently ceph status or ceph health won't output mds slow request summary. It is not very friendly and timely for op...
04/22/2016
- 09:22 AM Bug #15378 (Rejected): ceph_volume_client: hasty removal of OSD caps during deauthorization
- 09:06 AM Bug #15378: ceph_volume_client: hasty removal of OSD caps during deauthorization
- This issue would no longer be a concern once each volume (CephFS directory) gets its own RADOS namespace http://track...
04/20/2016
- 02:00 AM Bug #15502: files read or written with cephfs (fuse or kernel) on client drop all their page cach...
- here is my test result....
04/19/2016
- 02:43 PM Bug #15502: files read or written with cephfs (fuse or kernel) on client drop all their page cach...
- I was able to make this happen with kernel mode as well. Does that tunable noted in #6 need to be implemented for fu...
- 07:24 AM Bug #15502 (Need More Info): files read or written with cephfs (fuse or kernel) on client drop al...
- 07:30 AM Bug #15045: CephFSVolumeClient.evict should be limited by path, not just auth ID
- tested by https://github.com/ceph/ceph-qa-suite/pull/968
04/18/2016
- 12:52 PM Feature #13999: client: richacl support
- code is done; waiting for richacl get merged into upstream
- 12:49 PM Feature #13998 (Resolved): posix acl support
- https://github.com/ceph/ceph/pull/5658
- 12:45 PM Feature #11950 (Fix Under Review): Strays enqueued for purge cause MDCache to exceed size limit
- https://github.com/ceph/ceph/pull/8582
- 11:30 AM Bug #15379: ceph mds continiously crashes and going into laggy state (stray purging problems)
- wip-zyan-testing branch includes that fix. http://gitbuilder.ceph.com/ceph-rpm-centos7-x86_64-basic/ref/wip-zyan-test...
- 08:22 AM Bug #15379: ceph mds continiously crashes and going into laggy state (stray purging problems)
- Are there rpms builded for this pr? I looked at http://gitbuilder.ceph.com/ceph-rpm-centos7-x86_64-basic/ but I did n...
- 07:37 AM Bug #15502: files read or written with cephfs (fuse or kernel) on client drop all their page cach...
- I can't reproduce this on 3.10.0-327.el7 kernel mount. To make ceph-fuse keeps the kernel pagecache, you need to set ...
04/15/2016
- 09:32 PM Bug #15169 (Rejected): have trouble in mount.ceph in precise
- This is an old ticket so you've probably already worked this out, but this was failing because your kernel was too old.
- 09:31 PM Bug #15168 (Duplicate): have trouble in mount.ceph
- http://tracker.ceph.com/issues/15169#change-69275
- 02:30 PM Bug #15502: files read or written with cephfs (fuse or kernel) on client drop all their page cach...
- Kernel is 3.10.0-327.el7.x86_64 ie the GA kernel for RHEL7.2
- 01:59 PM Bug #15502: files read or written with cephfs (fuse or kernel) on client drop all their page cach...
- This was on both ceph-fuse and a recent-ish rhel (7.2? 7.3-prerelease?) kernel. Unless there's some extra thing in th...
- 01:33 PM Bug #15502: files read or written with cephfs (fuse or kernel) on client drop all their page cach...
This should be fixed in upstream, Barry, which version of RHEL kernel do you use?- 12:29 PM Bug #15502: files read or written with cephfs (fuse or kernel) on client drop all their page cach...
- Similar to http://tracker.ceph.com/issues/13640#change-66216
- 12:43 PM Feature #15507: MDS: support "watching" an inode/dentry
- Not a million miles away from http://tracker.ceph.com/projects/ceph/wiki/Live_Performance_Probes
- 07:12 AM Backport #15512 (Resolved): hammer: Double decreased the count to trim caps which will cause fail...
- https://github.com/ceph/ceph/pull/8804
- 03:19 AM Bug #15508: client: simultaneous readdirs are very racy
- Hmm, I think the end result would be pretty much the same, although just having an array might be simpler. A pointer ...
- 03:12 AM Bug #15508: client: simultaneous readdirs are very racy
- Another option is assign dentry a cache index and use array to track the dentry list. If the shared_gen hasn't change...
04/14/2016
- 11:06 PM Bug #13271: Missing dentry in cache when doing readdirs under cache pressure (?????s in ls-l)
- This may get handled alongside http://tracker.ceph.com/issues/15508.
- 11:06 PM Bug #15508: client: simultaneous readdirs are very racy
- Some obvious solutions are disqualified, both because we can't really track what directory listing's are in progress ...
- 11:01 PM Bug #15508 (Resolved): client: simultaneous readdirs are very racy
- Imagine we have a ceph-fuse user doing readdirs a and b on a very large directory (which requires multiple MDS round-...
- 10:49 PM Bug #11792: mds: recursive statistics are either inaccurate or too "chunky"
- This is going to pop up for users. Let's be proactive about checking and fixing it if possible.
- 10:45 PM Bug #13906 (Closed): pjd failure on hammer
- I think maybe this got backported after all? If not, it hasn't come up again, so closing it.
- 10:44 PM Bug #13107 (Resolved): ceph-fuse: handle multiple mounts on the same host (don't set nonce with o...
- This got fixed when #13032 did!
- 10:41 PM Bug #14319 (Pending Backport): Double decreased the count to trim caps which will cause failing t...
- Whoops, this got merged in way back then in January!
Looks like no backport got scheduled or anything; we'll see i... - 10:28 PM Bug #14716 (Won't Fix): "Thread.cc: 143: FAILED assert(status == 0)" in fs-hammer---basic-smithi
- This was a result of the OSD full handling changes, which got partly backported. I think the resolution we ended up a...
- 10:25 PM Bug #14759 (Closed): ovh: failed snaptest-git-ceph.sh test
- It hasn't popped up again.
- 10:24 PM Bug #14685: dbench hang on native cifs mount
- Apparently still happening? http://qa-proxy.ceph.com/teuthology/teuthology-2016-04-02_23:14:02-samba-master---basic-s...
- 10:19 PM Bug #15235 (Closed): MDS : erroneous error message about reading config file
- There's a conversation on that PR.
- 10:03 PM Feature #15507 (New): MDS: support "watching" an inode/dentry
- It would be great if we could monitor all client access to a specific file. Obviously we can't check for each access ...
- 09:58 PM Bug #15270 (Resolved): "mount error 5 = Input/output error" in smoke-master-distro-basic-openstack
- 09:58 PM Bug #15270: "mount error 5 = Input/output error" in smoke-master-distro-basic-openstack
- changed @smoke@ suite to use kernel @testing@
- 09:48 PM Bug #15270: "mount error 5 = Input/output error" in smoke-master-distro-basic-openstack
- I think anything running kernel tests needs to run with our testing kernel, not the distro kernel that happens to be ...
- 09:51 PM Bug #15045 (Fix Under Review): CephFSVolumeClient.evict should be limited by path, not just auth ID
- https://github.com/ceph/ceph/pull/8602
- 09:47 PM Feature #15506 (Resolved): qa: run at least one upgrade test in the FS suite
- We've broken upgrading at least twice during the Jewel freeze/RC period. They would all have been detected if the PRs...
- 09:36 PM Bug #15399 (In Progress): MDS incarnation get lost after remove filesystem
- For Jewel: https://github.com/ceph/ceph/pull/8484
But a more comprehensive one (that works with pools shared betwe... - 09:28 PM Bug #11314: qa: MDS crashed and the runs hung without ever timing out
- We clearly aren't treating this as very important, and I think we've had more trouble with OSDs doing this than MDSes...
- 09:27 PM Feature #14642: Validate layouts everywhere we load them
- Okay, let's put this in as a feature.
- 09:26 PM Bug #15502: files read or written with cephfs (fuse or kernel) on client drop all their page cach...
- Zheng, can you look at this? Hopefully we just have a bad cap transition on the server or something.
- 05:36 PM Bug #15502 (Resolved): files read or written with cephfs (fuse or kernel) on client drop all thei...
- Testing cephfs file system I/O with early jewel bits (ceph-10.0.4-1.el7cp.x86_64) on:
RHEL72 client mounting a cep... - 09:23 PM Bug #9994 (Resolved): ceph-qa-suite: nfs mount timeouts
- Well, this hasn't been updated in a while, and the tests I really remember failing were all on OVH, which was resolve...
- 09:22 PM Bug #14807: MDS crashes repeatedly after upgrade to Infernalis from Hammer
- Haven't seen this elsewhere, and no logs.
- 09:20 PM Bug #15050: deleting striped file in cephfs doesn't free up file's space
- Just waiting on backports, FS team doesn't need to worry about those.
- 09:19 PM Bug #15402 (Resolved): Failure in TestMultiFilesystems.test_standby_for_fscid
- 09:18 PM Bug #14608: snaptests.yaml failure: [WRN] open_snap_parents has:" in cluster log
- Still waiting to reproduce with logs. :(
- 09:18 PM Bug #15303 (Resolved): Client holds incorrect complete flag on dir after losing caps
- 09:28 AM Bug #15379: ceph mds continiously crashes and going into laggy state (stray purging problems)
- I send a RP that make MDS not pin strays in memory
could you take a try:
https://github.com/ukernel/ceph/tree/inf...
04/13/2016
- 03:37 PM Bug #15485 (Duplicate): drop /usr/bin/cephfs
- Greg tells me that /usr/bin/cephfs is deprecated. We should rip it out post-Jewel.
Also available in: Atom