Activity
From 03/05/2019 to 04/03/2019
04/03/2019
- 10:49 PM Bug #39060 (Fix Under Review): ls -S command produces AttributeError: 'str' object has no attribu...
- 07:53 PM Feature #39098 (Resolved): mds: lock caching for asynchronous unlink
- In order to allow the client to asynchronously delete files, we need Fx caps on the parent, and Lx caps on the inode ...
- 02:17 PM Bug #23262: kclient: nofail option not supported
- Hi,
Should I do something to backport it? - 01:15 PM Feature #5520 (In Progress): osdc: should handle namespaces
- 02:32 AM Bug #38832: mds: fail to resolve snapshot name contains '_'
- ...
04/02/2019
- 03:24 PM Bug #37378: truncate_seq ordering issues with object creation
- Just a quick update: I've pushed another version of my copy_from/truncate OSD fix into https://github.com/ceph/ceph/p...
- 04:18 AM Backport #38445 (In Progress): luminous: mds: drop cache does not timeout as expected
- 02:00 AM Bug #39079 (Resolved): qa: simple messenger removal causes qa build failure
- 12:07 AM Bug #39079 (Fix Under Review): qa: simple messenger removal causes qa build failure
- 12:03 AM Bug #39079 (Resolved): qa: simple messenger removal causes qa build failure
- ...
- 12:19 AM Bug #39020 (Fix Under Review): qa: qa/suites/fs/upgrade testing with upgrades from luminous no lo...
- 12:19 AM Bug #39078 (Fix Under Review): fs: we lack a feature bit for nautilus
- https://github.com/ceph/ceph/pull/27303/commits/1aaa3a50c34ed4214d7d13a87b4011a557f2f14a
- 12:19 AM Bug #39077 (Fix Under Review): fs: add note to release process that new CEPHFS_FEATURE_X bit need...
04/01/2019
- 11:50 PM Bug #39078 (Resolved): fs: we lack a feature bit for nautilus
- This prevents an operator from configuring the file system to only allow nautilus clients.
- 11:48 PM Bug #39077 (In Progress): fs: add note to release process that new CEPHFS_FEATURE_X bit needs add...
- 11:31 PM Bug #39077 (Resolved): fs: add note to release process that new CEPHFS_FEATURE_X bit needs added ...
- Currently this is done ad-hoc and we just missed Nautilus.
- 11:11 PM Bug #38944 (Won't Fix): MDSMonitor: no mechanism for seeing value of allow_new_snaps
- The value is available in `ceph fs dump --format=json` in filesystems[].mdsmap.flags. The magic bit number is 2. See ...
- 10:06 PM Bug #38844 (Pending Backport): mds: mds_cap_revoke_eviction_timeout is not used to initialize Ser...
- 10:05 PM Bug #23262 (Pending Backport): kclient: nofail option not supported
- 09:46 PM Bug #38822 (Pending Backport): mds: there is an assertion when calling Beacon::shutdown()
- 06:02 PM Feature #38951: client: implement asynchronous unlink/create
- Jeff Layton wrote:
> I've started going through the kernel client, as I figured this would be more useful there init... - 04:35 PM Feature #38951: client: implement asynchronous unlink/create
- I've started going through the kernel client, as I figured this would be more useful there initially (and because I u...
- 04:56 PM Backport #38445: luminous: mds: drop cache does not timeout as expected
- Venky, this is the backport you're looking for I think. Can you do it?
ACK -- I'll do the backport. Thanks! - 04:55 PM Backport #38877: luminous: mds: high debug logging with many subtrees is slow
- Rishabh, please do this backport. See also this guide: http://tracker.ceph.com/projects/ceph-releases/wiki/HOWTO_back...
- 04:53 PM Backport #36462 (New): luminous: ceph-fuse client can't read or write due to backward cap_gen
- 04:53 PM Backport #38735 (Resolved): luminous: qa: tolerate longer heartbeat timeouts when using valgrind
- 01:57 PM Backport #38735: luminous: qa: tolerate longer heartbeat timeouts when using valgrind
- Nathan Cutler wrote:
> https://github.com/ceph/ceph/pull/26964
merged - 04:44 PM Backport #38686 (In Progress): luminous: kcephfs TestClientLimits.test_client_pin fails with "cli...
- 04:44 PM Backport #38686 (Resolved): luminous: kcephfs TestClientLimits.test_client_pin fails with "client...
- 04:43 PM Backport #38669 (Resolved): luminous: "log [WRN] : Health check failed: 1 clients failing to resp...
- 01:56 PM Backport #38669: luminous: "log [WRN] : Health check failed: 1 clients failing to respond to capa...
- Nathan Cutler wrote:
> https://github.com/ceph/ceph/pull/27024
merged - 04:43 PM Backport #38545 (Resolved): luminous: qa: "Loading libcephfs-jni: Failure!"
- 01:58 PM Backport #38545: luminous: qa: "Loading libcephfs-jni: Failure!"
- Nathan Cutler wrote:
> https://github.com/ceph/ceph/pull/26820
merged - 04:42 PM Backport #38543 (Resolved): luminous: qa: tasks.cephfs.test_misc.TestMisc.test_fs_new hangs becau...
- 01:58 PM Backport #38543: luminous: qa: tasks.cephfs.test_misc.TestMisc.test_fs_new hangs because clients ...
- Nathan Cutler wrote:
> https://github.com/ceph/ceph/pull/26805
merged - 04:42 PM Backport #38541 (Resolved): luminous: qa: fsstress with valgrind may timeout
- 01:58 PM Backport #38541: luminous: qa: fsstress with valgrind may timeout
- Nathan Cutler wrote:
> https://github.com/ceph/ceph/pull/26776
merged - 04:42 PM Backport #38449 (Resolved): luminous: src/osdc/Journaler.cc: 420: FAILED ceph_assert(!r)
- 01:59 PM Backport #38449: luminous: src/osdc/Journaler.cc: 420: FAILED ceph_assert(!r)
- Nathan Cutler wrote:
> https://github.com/ceph/ceph/pull/26642
merged - 12:57 PM Bug #39061 (Duplicate): Test failure: test_drop_cache_command_timeout_asok (tasks.cephfs.test_mis...
- Seen in luminous run: http://qa-proxy.ceph.com/teuthology/yuriw-2019-03-28_20:44:19-fs-wip-yuri4-testing-2019-03-28-1...
- 12:42 PM Bug #39060 (Resolved): ls -S command produces AttributeError: 'str' object has no attribute 'decode'
- ...
03/30/2019
- 08:19 AM Backport #39051 (Resolved): nautilus: doc: add LAZYIO
- https://github.com/ceph/ceph/pull/27899
- 08:19 AM Backport #39050 (Resolved): nautilus: ceph_volume_client: Too many arguments for "WriteOpCtx"
- https://github.com/ceph/ceph/pull/27893
03/29/2019
- 09:44 PM Documentation #38729 (Pending Backport): doc: add LAZYIO
- 09:37 PM Bug #38804 (Fix Under Review): cephfs-shell: ls always lists hidden files and directories
- 04:58 PM Feature #38951: client: implement asynchronous unlink/create
- Jeff Layton wrote:
> Today, we have no support for asynchronous MDS requests. make_request always blocks on the requ... - 04:35 PM Feature #38951: client: implement asynchronous unlink/create
- Today, we have no support for asynchronous MDS requests. make_request always blocks on the request. So the first step...
- 04:18 PM Bug #39026 (Fix Under Review): mds: crash during mds restart
- 05:33 AM Bug #39026: mds: crash during mds restart
- https://github.com/ceph/ceph/pull/27256
- 03:41 AM Bug #39026 (Resolved): mds: crash during mds restart
- On version 12.2.10...
- 02:36 PM Bug #38946 (Pending Backport): ceph_volume_client: Too many arguments for "WriteOpCtx"
03/28/2019
- 10:02 PM Bug #39020 (Resolved): qa: qa/suites/fs/upgrade testing with upgrades from luminous no longer work
- ...
- 07:36 PM Bug #38996 (Duplicate): cephfs-shell: ls command produces error: no "colorize" attribute found error
- 07:19 AM Bug #38996 (Resolved): cephfs-shell: ls command produces error: no "colorize" attribute found error
- Even though the package colorize-0.3.4-15.fc29.noarch is already installed.
The following error is produced.... - 05:06 PM Feature #38951: client: implement asynchronous unlink/create
- Jeff, please link the various tracker tickets you create to sub-task the project with "related issues" so they don't ...
- 04:46 PM Bug #38822: mds: there is an assertion when calling Beacon::shutdown()
- huanwen ren wrote:
> hi Patrick @Patrick Donnelly
> Can you pull me into the development of the tracker? I can't mo... - 04:35 PM Feature #38838 (Fix Under Review): Expose CephFS snapshot creation time to clients
- 04:32 PM Bug #38994 (Fix Under Review): mds: we encountered "No space left on device" when moving huge num...
- 03:52 AM Bug #38994: mds: we encountered "No space left on device" when moving huge number of files into o...
- the same reason as adding maybe_fragment on the tail of handle_client_openc.
- 03:35 AM Bug #38994 (Resolved): mds: we encountered "No space left on device" when moving huge number of f...
- The issue was founded in version 12.2.10.
client log is listed below:
.....
mv: cannot move ‘src1325’ to ‘../destf... - 12:46 PM Bug #38835 (Fix Under Review): MDSTableServer.cc: 83: FAILED assert(version == tid)
- 10:06 AM Bug #38835: MDSTableServer.cc: 83: FAILED assert(version == tid)
- https://github.com/ceph/ceph/pull/27238
- 12:42 PM Bug #38946: ceph_volume_client: Too many arguments for "WriteOpCtx"
- I'm not sure it's worth the trouble to backport to luminous, and mimic. Maybe OK to backport to latest stable release...
- 12:29 PM Bug #38651 (Resolved): qa: powercycle suite reports MDS_SLOW_METADATA_IO
- 12:29 PM Backport #38666 (Resolved): mimic: qa: powercycle suite reports MDS_SLOW_METADATA_IO
- 12:22 PM Backport #38665 (Resolved): luminous: qa: powercycle suite reports MDS_SLOW_METADATA_IO
03/27/2019
- 07:58 PM Backport #38666: mimic: qa: powercycle suite reports MDS_SLOW_METADATA_IO
- Nathan Cutler wrote:
> https://github.com/ceph/ceph/pull/26961
merged - 07:25 PM Backport #38688 (Rejected): luminous: mds: inode filtering on 'dump cache' asok
- 07:22 PM Feature #38849: mds: proactively merge orphaned inodes into a remaining parent after deleting pri...
- Another approach may be to allow the stray directory to fragment. Are there any downsides to doing that?
- 07:10 PM Feature #38951: client: implement asynchronous unlink/create
- I'm taking the approach that if we have to contact the server at all, then we probably should just send a synchronous...
- 02:53 PM Feature #38951 (Resolved): client: implement asynchronous unlink/create
- We have an open project to teach the client how to buffer creates when it has the right caps (Fx), and a delegated se...
- 03:58 PM Feature #38838: Expose CephFS snapshot creation time to clients
- Samba changes have also been proposed: https://lists.samba.org/archive/samba-technical/2019-March/133089.html
- 01:51 PM Feature #24461: cephfs: improve file create performance buffering file unlink/create operations
- Jeff Layton wrote:
> I have a couple of starter patches that add a new delegated_inos interval_set to session_info_t... - 01:23 PM Feature #24461: cephfs: improve file create performance buffering file unlink/create operations
- I have a couple of starter patches that add a new delegated_inos interval_set to session_info_t.
Questions at this... - 01:09 PM Bug #38835: MDSTableServer.cc: 83: FAILED assert(version == tid)
- after restarting the mds, the crash will not happen again.
- 10:23 AM Bug #38835: MDSTableServer.cc: 83: FAILED assert(version == tid)
- Yes this was upgraded from luminous and yes that was the first time the MDS was running in mimic....
- 08:25 AM Bug #38835: MDSTableServer.cc: 83: FAILED assert(version == tid)
- goto frame 4 and try casting MDSIOContextBase into C_Prepare. tid is stored in C_Prepare
was the cluster upgrade... - 12:00 PM Bug #38946 (Fix Under Review): ceph_volume_client: Too many arguments for "WriteOpCtx"
- https://github.com/ceph/ceph/pull/27213
- 09:45 AM Bug #38946 (Resolved): ceph_volume_client: Too many arguments for "WriteOpCtx"
- Sebastian W hit this while linting,
pybind/ceph_volume_client.py:1480: error: Too many arguments for "WriteOpCtx"
... - 06:52 AM Bug #38944: MDSMonitor: no mechanism for seeing value of allow_new_snaps
- I also couldn't find a read setting. Would have expected it to at least be part of `ceph fs dump`.
- 03:05 AM Bug #38652: mds|kclient: MDS_CLIENT_LATE_RELEASE warning caused by inline bug on RHEL 7.5
new issue that can cause this warning (file lock become sync state while Fcb is issued)
/ceph/teuthology-archive...
03/26/2019
- 11:46 PM Bug #38944 (Won't Fix): MDSMonitor: no mechanism for seeing value of allow_new_snaps
- Although it's obvious that cephfs requires "allow_new_snaps = true" in order to enable snapshots, there appears to be...
- 04:48 PM Backport #38665: luminous: qa: powercycle suite reports MDS_SLOW_METADATA_IO
- Nathan Cutler wrote:
> https://github.com/ceph/ceph/pull/26962
merged - 02:48 PM Feature #24461: cephfs: improve file create performance buffering file unlink/create operations
- Great, so let's do a minor revision on the rules above. In order to buffer creates the client will need:
# an unus... - 11:42 AM Bug #38835: MDSTableServer.cc: 83: FAILED assert(version == tid)
- Zheng Yan wrote:
> I have trouble to check the coredump file. please use gdb to print 'tid' and 'version' if you can... - 03:07 AM Bug #38835: MDSTableServer.cc: 83: FAILED assert(version == tid)
- I have trouble to check the coredump file. please use gdb to print 'tid' and 'version' if you can.
- 12:49 AM Bug #38822: mds: there is an assertion when calling Beacon::shutdown()
- hi Patrick @Patrick Donnelly
Can you pull me into the development of the tracker? I can't modify most of the states ...
03/25/2019
- 05:31 PM Feature #38838: Expose CephFS snapshot creation time to clients
- Kernel PR: https://www.spinics.net/lists/ceph-devel/msg44894.html
Both patch-sets have been acked, but the userspa...
03/22/2019
- 01:04 PM Backport #38877 (Resolved): luminous: mds: high debug logging with many subtrees is slow
- https://github.com/ceph/ceph/pull/27679
- 01:04 PM Backport #38876 (Resolved): nautilus: mds: high debug logging with many subtrees is slow
- https://github.com/ceph/ceph/pull/27892
- 01:04 PM Backport #38875 (Resolved): mimic: mds: high debug logging with many subtrees is slow
- https://github.com/ceph/ceph/pull/29219
- 10:26 AM Feature #38851 (Rejected): mount.ceph.fuse: support secretfile option
- Hi!
As far as I can see, secretfile doesn't work with mount.ceph.fuse; Is this by intention? It would be nice to h... - 09:28 AM Feature #38849: mds: proactively merge orphaned inodes into a remaining parent after deleting pri...
- Sage suggests that forward scrub probably touches the remote hard link locations enough to force re-integrating the s...
- 09:22 AM Feature #38849 (Duplicate): mds: proactively merge orphaned inodes into a remaining parent after ...
- https://www.mail-archive.com/ceph-users@lists.ceph.com/msg53368.html
While we move an unlinked file into the stray... - 06:04 AM Backport #38736 (In Progress): mimic: qa: "[WRN] Health check failed: 1/3 mons down, quorum b,c (...
- -https://github.com/ceph/ceph/pull/27109-
03/21/2019
- 07:57 PM Bug #38844 (Resolved): mds: mds_cap_revoke_eviction_timeout is not used to initialize Server::cap...
- The config value is only used if the user changes it while the MDS is running.
- 07:31 PM Feature #24461: cephfs: improve file create performance buffering file unlink/create operations
- Jeff Layton wrote:
> Zheng Yan wrote:
>
> > > On another note, I do have at least one minor concern about using F... - 11:59 AM Feature #24461: cephfs: improve file create performance buffering file unlink/create operations
- Zheng Yan wrote:
> > On another note, I do have at least one minor concern about using Fb to indicate that buffere... - 05:54 PM Feature #17835 (New): mds: enable killpoint tests for MDS-MDS subtree export
- 04:21 PM Feature #38838 (Resolved): Expose CephFS snapshot creation time to clients
- Samba requires the snapshot creation time for the purpose of tracking and identifying Previous Versions for files and...
- 12:50 PM Feature #24285: mgr: add module which displays current usage of file system (`fs top`)
- for completeness, linking John's performance probes blueprint here, which is really interesting: https://tracker.ceph...
- 12:42 PM Bug #38832: mds: fail to resolve snapshot name contains '_'
- Found the cause: the subdir snaps are not visible if they have a _ in the original snapshot name. Snaps without _ do ...
- 09:42 AM Bug #38835 (Resolved): MDSTableServer.cc: 83: FAILED assert(version == tid)
- We just hit this on a v13.2.5 cluster with 1 active MDS:...
03/20/2019
- 09:24 PM Bug #38822 (Fix Under Review): mds: there is an assertion when calling Beacon::shutdown()
- 06:29 AM Bug #38822: mds: there is an assertion when calling Beacon::shutdown()
- Cause:
If you run MDSDaemon::init(), an exception occurs that causes MDSDaemon::suicide()--->Beacon::shutdown()--->s... - 06:01 AM Bug #38822 (Resolved): mds: there is an assertion when calling Beacon::shutdown()
- The construction condition is:
when you use systemctl start a non-existing ceph-mds@xx object
Error info:
<p... - 04:42 PM Bug #38832 (Resolved): mds: fail to resolve snapshot name contains '_'
- We observed the following issue on latest centos 7.6 kernel && kernel v5.0.3.
We mount ceph in /cephk: ... - 01:47 PM Feature #38829 (Resolved): cephfs-shell: add a "stat" command
- I don't see a way to ask cephfs-shell to dump out all of the info available via statx for a particular inode. Somethi...
- 01:41 PM Feature #24461: cephfs: improve file create performance buffering file unlink/create operations
- Jeff Layton wrote:
> Some notes and status:
>
> I've been going over the code and playing with cephfs-shell to cr... - 12:48 PM Feature #24461: cephfs: improve file create performance buffering file unlink/create operations
- Some notes and status:
I've been going over the code and playing with cephfs-shell to create different cap handlin... - 03:41 AM Backport #38688 (Need More Info): luminous: mds: inode filtering on 'dump cache' asok
- DUMP_PATH is undefined in luminous :
[ 58%] Building CXX object src/test/CMakeFiles/unittest_admin_socket.dir/admi... - 02:51 AM Backport #38689 (In Progress): mimic: mds: inode filtering on 'dump cache' asok
- https://github.com/ceph/ceph/pull/27058
03/19/2019
- 08:50 PM Bug #37726 (Pending Backport): mds: high debug logging with many subtrees is slow
- 01:20 PM Bug #38679 (Fix Under Review): mds: behind on trimming and "[dentry] was purgeable but no longer ...
- 12:43 PM Bug #38679: mds: behind on trimming and "[dentry] was purgeable but no longer is!"
- The "behind on trimming" was caused by a frozen dirfrag. The dirfrag stayed in frozen state for a long time is becaus...
- 03:42 AM Bug #38679: mds: behind on trimming and "[dentry] was purgeable but no longer is!"
- For the "was purgeable but no longer is" errors. ...
- 09:26 AM Bug #38809 (Closed): cephfs-shell: python traceback with inexistant directory with cd
- CephFS:~/cmd2.git/.git/objects>>> cd pac
[Errno 2] chdir failed
Traceback (most recent call last):
File "../src/... - 04:17 AM Bug #38805 (Closed): cephfs-shell: put doesn't accept relative source file path of file in curren...
- $ put ./src_file .
$ put src_file .
causes traceback - 04:15 AM Bug #38804 (Resolved): cephfs-shell: ls always lists hidden files and directories
- ls should list hidden files and directories only when -a or --all argument is present
- 03:57 AM Backport #38686 (In Progress): luminous: kcephfs TestClientLimits.test_client_pin fails with "cli...
- https://github.com/ceph/ceph/pull/27040
03/18/2019
- 08:42 PM Bug #38803 (Resolved): qa: test_sessionmap assumes simple messenger
- ...
- 07:23 PM Fix #38801 (Resolved): qa: ignore "ceph.dir.pin: No such attribute" for (old) kernel client
- The issue is that getxattr support for "ceph.dir.pin" is not yet backported to RHEL. We want to ignore these spurious...
- 01:50 PM Bug #38677: qa: kclient unmount hangs after file system goes down
- I think upstream kernel is also affected. (umount(8) stat the umount root, getattr may hang if there is no shared caps)
- 01:16 PM Documentation #38729 (Fix Under Review): doc: add LAZYIO
- 03:07 AM Backport #38669 (In Progress): luminous: "log [WRN] : Health check failed: 1 clients failing to r...
- https://github.com/ceph/ceph/pull/27024
- 03:05 AM Backport #38670 (In Progress): mimic: "log [WRN] : Health check failed: 1 clients failing to resp...
- https://github.com/ceph/ceph/pull/27023
03/15/2019
- 04:37 PM Documentation #38728 (Resolved): doc: scrub administration docs need updated
- 04:11 PM Bug #23262: kclient: nofail option not supported
- I've an attempt to a fix here https://github.com/ceph/ceph/pull/26992
- 10:16 AM Documentation #38729 (In Progress): doc: add LAZYIO
- 03:34 AM Backport #38735 (In Progress): luminous: qa: tolerate longer heartbeat timeouts when using valgrind
- 03:31 AM Backport #38734 (In Progress): mimic: qa: tolerate longer heartbeat timeouts when using valgrind
- 02:00 AM Backport #38665 (In Progress): luminous: qa: powercycle suite reports MDS_SLOW_METADATA_IO
- https://github.com/ceph/ceph/pull/26962
- 01:54 AM Backport #38666 (In Progress): mimic: qa: powercycle suite reports MDS_SLOW_METADATA_IO
- https://github.com/ceph/ceph/pull/26961
03/14/2019
- 07:57 PM Bug #38739: cephfs-shell: python traceback with mkdir inside inexistant directory
- To be clear, you're saying that it should just return some kind of error message like "ENOENT"?
- 12:28 PM Bug #38739 (Resolved): cephfs-shell: python traceback with mkdir inside inexistant directory
- CephFS:~/>>> mkdir jack/and/jill/went/up/the/hill
[Errno 2] error in mkdir 'b'/jack/and/jill/went/up/the/hill''
Tra... - 12:58 PM Bug #38743 (Resolved): cephfs-shell: mkdir creates directory with invalid octal mode
- ...
- 12:46 PM Bug #38742 (Resolved): cephfs-shell: entering unrecognized command does not print newline after m...
- ...
- 12:34 PM Bug #38741 (Resolved): cephfs-shell: python traceback with mkdir when reattempt of mkdir
- ...
- 12:30 PM Feature #38740 (Resolved): cephfs-shell: support mkdir with non-octal mode
- ...
- 09:06 AM Backport #38737 (Rejected): luminous: qa: "[WRN] Health check failed: 1/3 mons down, quorum b,c (...
- 09:06 AM Backport #38736 (Resolved): mimic: qa: "[WRN] Health check failed: 1/3 mons down, quorum b,c (MON...
- https://github.com/ceph/ceph/pull/27906
- 09:06 AM Backport #38735 (Resolved): luminous: qa: tolerate longer heartbeat timeouts when using valgrind
- https://github.com/ceph/ceph/pull/26964
- 09:06 AM Backport #38734 (Resolved): mimic: qa: tolerate longer heartbeat timeouts when using valgrind
- https://github.com/ceph/ceph/pull/26963
- 03:51 AM Bug #38723 (Pending Backport): qa: tolerate longer heartbeat timeouts when using valgrind
- 03:44 AM Documentation #38728 (In Progress): doc: scrub administration docs need updated
- 03:32 AM Bug #38704 (Pending Backport): qa: "[WRN] Health check failed: 1/3 mons down, quorum b,c (MON_DOW...
03/13/2019
- 06:47 PM Documentation #38729 (Resolved): doc: add LAZYIO
- to doc/cephfs/{experimental-features.rst,posix.rst}. A new documentation page could also be appropriate.
- 06:29 PM Documentation #38728 (Resolved): doc: scrub administration docs need updated
- For changes from #12282.
Mostly, we should document (with examples!) how to view, abort, and pause scrubs. - 04:20 PM Bug #38723 (Fix Under Review): qa: tolerate longer heartbeat timeouts when using valgrind
- 04:11 PM Bug #38723 (Resolved): qa: tolerate longer heartbeat timeouts when using valgrind
- ...
- 11:56 AM Feature #24461: cephfs: improve file create performance buffering file unlink/create operations
- Jeff Layton wrote:
> Greg Farnum wrote:
> >
> > Well, as someone noted, you also need to make sure you aren't cre... - 11:48 AM Feature #24461: cephfs: improve file create performance buffering file unlink/create operations
- Jeff Layton wrote:
> Zheng Yan wrote:
> > For directory inode, current mds only issues Fsx (at most) caps to client... - 09:53 AM Backport #38710 (Rejected): luminous: qa: kclient unmount hangs after file system goes down
- 09:53 AM Backport #38709 (Resolved): mimic: qa: kclient unmount hangs after file system goes down
- https://github.com/ceph/ceph/pull/29218
03/12/2019
- 09:52 PM Feature #24461: cephfs: improve file create performance buffering file unlink/create operations
- Greg Farnum wrote:
>
> Well, as someone noted, you also need to make sure you aren't creating a dentry that alread... - 09:21 PM Feature #24461: cephfs: improve file create performance buffering file unlink/create operations
- Jeff Layton wrote:
> That's sort of my point here (though I didn't put it quite as succinctly). I don't think adding... - 08:25 PM Feature #24461: cephfs: improve file create performance buffering file unlink/create operations
- That's sort of my point here (though I didn't put it quite as succinctly). I don't think adding more cap flags really...
- 06:23 PM Feature #24461: cephfs: improve file create performance buffering file unlink/create operations
- Given that we already use non-cap flags, and directories are special anyway, I'm not sure extending the cap language ...
- 12:49 PM Feature #24461: cephfs: improve file create performance buffering file unlink/create operations
- Patrick Donnelly wrote:
> Okay, here's how I think it should work but it may not be how it actually works:
>
> Fs... - 12:10 PM Feature #24461: cephfs: improve file create performance buffering file unlink/create operations
- Patrick Donnelly wrote:
>
> Okay, here's how I think it should work but it may not be how it actually works:
>
... - 11:10 AM Feature #24461: cephfs: improve file create performance buffering file unlink/create operations
- Zheng Yan wrote:
> For directory inode, current mds only issues Fsx (at most) caps to client. It never issues Frwcb ... - 03:49 AM Feature #24461: cephfs: improve file create performance buffering file unlink/create operations
- Jeff Layton wrote:
> Patrick Donnelly wrote:
>
> > Fx would indicate the directory contents can be cached by the ... - 05:48 PM Bug #38677 (Pending Backport): qa: kclient unmount hangs after file system goes down
- 05:40 PM Bug #38704 (Fix Under Review): qa: "[WRN] Health check failed: 1/3 mons down, quorum b,c (MON_DOW...
- 05:28 PM Bug #38704 (Resolved): qa: "[WRN] Health check failed: 1/3 mons down, quorum b,c (MON_DOWN) in cl...
- ...
- 05:26 PM Bug #38676 (Resolved): qa: src/common/Thread.cc: 157: FAILED ceph_assert(ret == 0)
- 11:58 AM Backport #38689 (Resolved): mimic: mds: inode filtering on 'dump cache' asok
- https://github.com/ceph/ceph/pull/27058
- 11:57 AM Backport #38688 (Rejected): luminous: mds: inode filtering on 'dump cache' asok
- 11:56 AM Backport #38687 (Resolved): mimic: kcephfs TestClientLimits.test_client_pin fails with "client ca...
- https://github.com/ceph/ceph/pull/29211
- 11:56 AM Backport #38686 (Resolved): luminous: kcephfs TestClientLimits.test_client_pin fails with "client...
- https://github.com/ceph/ceph/pull/27040
- 03:44 AM Support #38640 (Closed): fs: garbage in cephfs pool and tier pool after delete all files
- Fyodor Ustinov wrote:
> Patrick Donnelly wrote:
> > The garbage is just inodes left behind, probably by some MDS bu... - 03:27 AM Bug #38681 (Resolved): cephfs-shell: add commands to manipulate snapshots
- Idea being to make this a trivial one-off command to manipulate quotas/snapshots on CephFS directories without mounting.
03/11/2019
- 11:11 PM Support #38640: fs: garbage in cephfs pool and tier pool after delete all files
- Patrick Donnelly wrote:
> The garbage is just inodes left behind, probably by some MDS bug in the past. Presumably y... - 09:06 PM Support #38640: fs: garbage in cephfs pool and tier pool after delete all files
- The garbage is just inodes left behind, probably by some MDS bug in the past. Presumably you don't need these files a...
- 09:52 PM Feature #24461: cephfs: improve file create performance buffering file unlink/create operations
- Jeff Layton wrote:
> Patrick Donnelly wrote:
>
> > Fx would indicate the directory contents can be cached by the ... - 05:29 PM Feature #24461: cephfs: improve file create performance buffering file unlink/create operations
- Patrick Donnelly wrote:
> Fx would indicate the directory contents can be cached by the client, yes? Perhaps the c... - 05:02 PM Feature #24461: cephfs: improve file create performance buffering file unlink/create operations
- Jeff Layton wrote:
> > EXCL and WR. I don't think the MDS ever considers handing out WR to clients.
>
> Just so I... - 04:27 PM Feature #24461: cephfs: improve file create performance buffering file unlink/create operations
- > EXCL and WR. I don't think the MDS ever considers handing out WR to clients.
Just so I'm clear...why do we care ... - 09:24 PM Bug #11314 (New): qa: MDS crashed and the runs hung without ever timing out
- 09:22 PM Feature #12107 (Fix Under Review): mds: use versioned wire protocol; obviate CEPH_MDS_PROTOCOL
- 09:20 PM Feature #11172 (Pending Backport): mds: inode filtering on 'dump cache' asok
- 05:25 PM Bug #38679 (Resolved): mds: behind on trimming and "[dentry] was purgeable but no longer is!"
- ...
- 04:22 PM Bug #38270 (Pending Backport): kcephfs TestClientLimits.test_client_pin fails with "client caps f...
- 04:14 PM Bug #38677 (Resolved): qa: kclient unmount hangs after file system goes down
- ...
- 04:08 PM Bug #38676 (Fix Under Review): qa: src/common/Thread.cc: 157: FAILED ceph_assert(ret == 0)
- 04:00 PM Bug #38676: qa: src/common/Thread.cc: 157: FAILED ceph_assert(ret == 0)
- We should just use the async messenger instead. According to Sage, other YAML bits are necessary to use v1 protocol.
- 03:59 PM Bug #38676 (Resolved): qa: src/common/Thread.cc: 157: FAILED ceph_assert(ret == 0)
- ...
- 02:39 PM Backport #38670 (Resolved): mimic: "log [WRN] : Health check failed: 1 clients failing to respond...
- https://github.com/ceph/ceph/pull/27023
- 02:39 PM Backport #38669 (Resolved): luminous: "log [WRN] : Health check failed: 1 clients failing to resp...
- https://github.com/ceph/ceph/pull/27024
- 02:38 PM Backport #38666 (Resolved): mimic: qa: powercycle suite reports MDS_SLOW_METADATA_IO
- https://github.com/ceph/ceph/pull/26961
- 02:38 PM Backport #38665 (Resolved): luminous: qa: powercycle suite reports MDS_SLOW_METADATA_IO
- https://github.com/ceph/ceph/pull/26962
- 01:52 PM Bug #38597: fs: "log [WRN] : failed to reconnect caps for missing inodes"
- luminous does not need backport
- 12:31 PM Bug #38652 (Fix Under Review): mds|kclient: MDS_CLIENT_LATE_RELEASE warning caused by inline bug ...
- https://github.com/ceph/ceph/pull/26881
- 12:28 PM Bug #38636: Inline data compatibly check in Locker::issue_caps is buggy
- https://github.com/ceph/ceph/pull/26881
03/10/2019
03/09/2019
- 06:14 PM Bug #38491 (Pending Backport): "log [WRN] : Health check failed: 1 clients failing to respond to ...
- 06:13 PM Bug #38471 (Duplicate): mds does not wait for cap revoke when lock is in LOCK_LOCK_XLOCK state.
- Reversing the duplicate since the commit cites the other tracker.
- 12:24 AM Bug #14608 (Can't reproduce): snaptests.yaml failure: [WRN] open_snap_parents has:" in cluster log
- 12:19 AM Bug #1114 (Rejected): NFS export extreme slowdown
- NFS over the kernel client is not a recommended configuration.
- 12:19 AM Bug #10740 (Rejected): teuthology: nfs test getting EBUSY on umount
- knfs is dead.
- 12:18 AM Bug #3424 (Rejected): java: Add the correct JUnit package dependencies on supported platforms and...
- Java/Hadoop testing is no longer a priority.
- 12:18 AM Bug #19456 (Rejected): Hadoop suite failing on missing upstream tarball
- Java/Hadoop testing is no longer a priority.
- 12:18 AM Bug #21149 (Rejected): SubsystemMap.h: 62: FAILED assert(sub < m_subsys.size())
- Java/Hadoop testing is no longer a priority.
- 12:18 AM Bug #17621 (Rejected): Hadoop does a bad job closing files and we end up holding too many caps
- Java/Hadoop testing is no longer a priority.
- 12:18 AM Feature #3422 (Rejected): Only compile java tests if JUnit is present
- Java/Hadoop testing is no longer a priority.
- 12:17 AM Feature #3399 (Rejected): java: add accessor to Ceph version numbers
- Java/Hadoop testing is no longer a priority.
- 12:16 AM Bug #17184 (Rejected): "Segmentation fault" in samba-jewel---basic-mira
- 12:15 AM Feature #4208 (Rejected): Add more replication pool tests for Hadoop / Ceph bindings
- Java/Hadoop testing is no longer a priority.
- 12:15 AM Bug #3318 (Rejected): java: lock access to CephStat, CephStatVFS from native
- Java/Hadoop testing is no longer a priority.
- 12:15 AM Bug #3337 (Rejected): java: unit tests don't remove container directory after a failure
- Java/Hadoop testing is no longer a priority.
- 12:15 AM Feature #4576 (Rejected): java: support ByteBuffer interface for NIO and NIO.2 high-perf I/O
- Java/Hadoop testing is no longer a priority.
- 12:14 AM Fix #12296 (Rejected): cephfs-hadoop: do not stash libcephfs.jar in git repo
- Java/Hadoop testing is no longer a priority.
- 12:14 AM Feature #9287 (Rejected): qa: hadoop: add big top tests to suite
- Java/Hadoop testing is no longer a priority.
- 12:14 AM Feature #9286 (Rejected): qa: hadoop: test 2.x with teuthology
- Java/Hadoop testing is no longer a priority.
03/08/2019
- 08:42 PM Bug #38452: mds: assert crash loop while unlinking file
- I had to repeat the procedure a couple time before everything cleaned up, but now it is back to normal.
I included... - 07:58 PM Feature #24461: cephfs: improve file create performance buffering file unlink/create operations
- Jeff Layton wrote:
> To communicate this range to the client, we could use a MClientSession message. Maybe version (... - 06:44 PM Feature #24461: cephfs: improve file create performance buffering file unlink/create operations
- Some notes about the preallocation piece:
The prealloc_inos interval set is tracked per-session in session_info_t.... - 02:07 PM Feature #24461: cephfs: improve file create performance buffering file unlink/create operations
- Patrick Donnelly wrote:
>
> EXCL and WR. I don't think the MDS ever considers handing out WR to clients.
>
... - 07:13 PM Bug #38652: mds|kclient: MDS_CLIENT_LATE_RELEASE warning caused by inline bug on RHEL 7.5
- See also Zheng's analysis in the ticket he opened: #38636
- 07:03 PM Bug #38652 (Resolved): mds|kclient: MDS_CLIENT_LATE_RELEASE warning caused by inline bug on RHEL 7.5
- ...
- 07:12 PM Bug #38636 (Duplicate): Inline data compatibly check in Locker::issue_caps is buggy
- 08:00 AM Bug #38636 (Duplicate): Inline data compatibly check in Locker::issue_caps is buggy
- ...
- 06:51 PM Bug #38651 (Fix Under Review): qa: powercycle suite reports MDS_SLOW_METADATA_IO
- 06:48 PM Bug #38651 (Resolved): qa: powercycle suite reports MDS_SLOW_METADATA_IO
- http://pulpito.ceph.com/yuriw-2019-03-07_00:07:42-powercycle-wip_yuri_nautilus_3.6.19-distro-basic-smithi/
We shou... - 04:24 PM Feature #15070 (Fix Under Review): mon: client: multifs: auth caps on client->mon connections to ...
- 02:46 PM Backport #38643 (Resolved): mimic: fs: "log [WRN] : failed to reconnect caps for missing inodes"
- https://github.com/ceph/ceph/pull/31282
- 01:07 PM Support #38640 (Closed): fs: garbage in cephfs pool and tier pool after delete all files
- Hi!
# ceph version
ceph version 13.2.4 (b10be4d44915a4d78a8e06aa31919e74927b142e) mimic (stable)
After remov...
03/07/2019
- 11:56 PM Bug #38270 (Fix Under Review): kcephfs TestClientLimits.test_client_pin fails with "client caps f...
- 11:23 PM Bug #36730 (In Progress): mds: should apply policy to throttle client messages
- 11:23 PM Bug #36171 (In Progress): mds: ctime should not use client provided ctime/mtime
- 11:04 PM Feature #24461: cephfs: improve file create performance buffering file unlink/create operations
- Jeff Layton wrote:
> I think this is not really a single project, but a set of them. At a high level:
> ensure ... - 01:48 PM Feature #24461: cephfs: improve file create performance buffering file unlink/create operations
- I think this is not really a single project, but a set of them. At a high level:
* ensure that the MDS can hand ou... - 10:40 PM Bug #38597 (Pending Backport): fs: "log [WRN] : failed to reconnect caps for missing inodes"
- 02:47 PM Feature #36663 (In Progress): mds: adjust cache memory limit automatically via target that tracks...
- 10:18 AM Feature #17309 (In Progress): qa: mon_thrash test for CephFS
- 03:50 AM Backport #38545 (In Progress): luminous: qa: "Loading libcephfs-jni: Failure!"
- https://github.com/ceph/ceph/pull/26808
- 03:49 AM Backport #38544 (In Progress): mimic: qa: "Loading libcephfs-jni: Failure!"
- https://github.com/ceph/ceph/pull/26807
- 01:18 AM Backport #38543 (In Progress): luminous: qa: tasks.cephfs.test_misc.TestMisc.test_fs_new hangs be...
- https://github.com/ceph/ceph/pull/26805
- 01:15 AM Backport #38542 (In Progress): mimic: qa: tasks.cephfs.test_misc.TestMisc.test_fs_new hangs becau...
- https://github.com/ceph/ceph/pull/26804
03/06/2019
- 04:44 PM Backport #38132 (Resolved): luminous: mds: stopping MDS with a large cache (40+GB) causes it to m...
- 03:52 PM Backport #38132: luminous: mds: stopping MDS with a large cache (40+GB) causes it to miss heartbeats
- Patrick Donnelly wrote:
> https://github.com/ceph/ceph/pull/26232
merged - 04:44 PM Backport #38130 (Resolved): luminous: mds: provide a limit for the maximum number of caps a clien...
- 03:52 PM Backport #38130: luminous: mds: provide a limit for the maximum number of caps a client may have
- Patrick Donnelly wrote:
> https://github.com/ceph/ceph/pull/26232
merged - 04:43 PM Bug #38488 (Resolved): luminous: mds: message invalid access
- 03:52 PM Bug #38488: luminous: mds: message invalid access
- https://github.com/ceph/ceph/pull/26661 merged
- 08:25 AM Bug #38597 (Fix Under Review): fs: "log [WRN] : failed to reconnect caps for missing inodes"
- https://github.com/ceph/ceph/pull/26781
- 12:54 AM Bug #38597: fs: "log [WRN] : failed to reconnect caps for missing inodes"
- Adding tag to #18461 in case they're related. Feel free to disconnect if they're not Zheng.
- 12:53 AM Bug #38597 (Resolved): fs: "log [WRN] : failed to reconnect caps for missing inodes"
- ...
- 03:03 AM Bug #38452: mds: assert crash loop while unlinking file
- there is no clue in the log. If you want to fix the issue, you can:...
- 02:19 AM Backport #38541 (In Progress): luminous: qa: fsstress with valgrind may timeout
- https://github.com/ceph/ceph/pull/26776
03/05/2019
- 11:16 PM Bug #23429 (Can't reproduce): File corrupt after writing to cephfs
- 07:35 PM Bug #17594: cephfs: permission checking not working (MDS should enforce POSIX permissions)
- > Why would there be more than one uid/gid pair for a client request?
There wouldn't be, but a client like ganesha... - 06:51 PM Bug #17594: cephfs: permission checking not working (MDS should enforce POSIX permissions)
- Jeff Layton wrote:
> No. This problem is not specific to rename -- I just used that as an example.
Sorry can you ... - 06:47 PM Bug #17594: cephfs: permission checking not working (MDS should enforce POSIX permissions)
- Jeff Layton wrote:
> To elaborate, today the rule looks something like this:
>
> "If there is a uid= qualifier in... - 03:38 PM Bug #17594: cephfs: permission checking not working (MDS should enforce POSIX permissions)
- No. This problem is not specific to rename -- I just used that as an example. The upshot here is that you cannot sepa...
- 03:25 PM Bug #17594: cephfs: permission checking not working (MDS should enforce POSIX permissions)
- Jeff Layton wrote:
> No, the order has already been established, because each operation _really_ starts with the per... - 12:31 PM Backport #37906 (New): mimic: make cephfs-data-scan reconstruct snaptable
- first attempted backport PR https://github.com/ceph/ceph/pull/26298 was closed
- 09:23 AM Bug #38452: mds: assert crash loop while unlinking file
- ...
Also available in: Atom