Project

General

Profile

Activity

From 12/31/2022 to 01/29/2023

01/29/2023

10:49 AM Bug #58597 (New): The MDS crashes when deleting a specific file
... Tobias Reinhard

01/27/2023

02:22 PM Bug #58489: mds stuck in 'up:replay' and crashed.
In the meantime I rebooted my hosts for regular maintenance (rolling reboot with only one node down). Since then I ca... Thomas Widhalm
07:56 AM Bug #58434 (Rejected): Widespread metadata corruption
Venky Shankar

01/25/2023

09:06 PM Bug #58434: Widespread metadata corruption
Venky Shankar wrote:
> Nathan Fish wrote:
> > Further evidence indicates the issue may not be with Ceph; possibly a...
Nathan Fish
01:21 PM Bug #58576 (Rejected): do not allow invalid flags with cmd 'scrub start'
Currently, cmd 'scrub start' would accept any flag and looking at logs, it does actually start the scrubbing on the p... Dhairya Parmar
11:21 AM Backport #58573 (In Progress): pacific: mds: fragment directory snapshots
Venky Shankar
10:30 AM Backport #58573 (Resolved): pacific: mds: fragment directory snapshots
https://github.com/ceph/ceph/pull/49867 Backport Bot

01/24/2023

02:27 PM Bug #58564 (In Progress): workunit suites/dbench.sh fails with error code 1
This dbench test job fails with error code 1 and error message @write failed on handle 11133 (Resource temporarily un... Rishabh Dave
12:33 PM Bug #56446: Test failure: test_client_cache_size (tasks.cephfs.test_client_limits.TestClientLimits)
Rishabh - I'm taking this one since its blocking testing for https://tracker.ceph.com/issues/57985 Venky Shankar
12:10 PM Bug #57985: mds: warning `clients failing to advance oldest client/flush tid` seen with some work...
Enforce (stricter) client-id check in client limit test - https://github.com/ceph/ceph/pull/49844 Venky Shankar
08:55 AM Cleanup #58561 (New): cephfs-top: move FSTopBase to fstopbase.py
move FSTopBase to fstopbase.py and import it to cephfs-top script. This is easy to do. But the issue is when it's don... Jos Collin

01/23/2023

03:12 PM Feature #58550 (Resolved): mds: add perf counter to track (relatively) larger log events
Logging of large (in size) subtreemap over and over again in the log segment is a well know scalability limiting fact... Venky Shankar
03:07 PM Feature #58488: mds: avoid encoding srnode for each ancestor in an EMetaBlob log event
Greg mentioned that this be worked on before its actually proved that this is causing slowness in the MDS - I agree. ... Venky Shankar
01:48 PM Bug #58489 (Triaged): mds stuck in 'up:replay' and crashed.
Venky Shankar
01:18 PM Bug #58434: Widespread metadata corruption
Nathan Fish wrote:
> Further evidence indicates the issue may not be with Ceph; possibly a bug in account management...
Venky Shankar

01/20/2023

05:50 AM Bug #58376: CephFS Snapshots are accessible even when it's deleted from the other client
The client *should* receive an updated snap trace from the MDS. Using that some of the snap inodes should be invalida... Venky Shankar

01/19/2023

02:12 PM Bug #58411 (Triaged): mds: a few simple operations crash mds
Good catch! Venky Shankar
01:15 PM Bug #58489: mds stuck in 'up:replay' and crashed.
Hi,
Thanks for the help. Here are the debug logs of my two active MDS. I thought, I'll tar the whole bunch for you...
Thomas Widhalm
01:11 PM Bug #58489: mds stuck in 'up:replay' and crashed.
Xiubo Li wrote:
> There is one case could trigger IMO:
>
> For example, there are two MDLog entries:
>
> ESess...
Venky Shankar
12:51 PM Bug #58376: CephFS Snapshots are accessible even when it's deleted from the other client
The client log analysis. The lookup on *'.snap/snapshot1'* is successful from the dentry cache. The caps and the leas... Kotresh Hiremath Ravishankar

01/18/2023

11:47 AM Bug #58489: mds stuck in 'up:replay' and crashed.
There is one case could trigger IMO:
For example, there are two MDLog entries:
ESessions entry --> {..., versio...
Xiubo Li
11:33 AM Bug #58489 (Resolved): mds stuck in 'up:replay' and crashed.
The issue is reported by upstream community user.
The cluster had two filesystems and the active mds of both the f...
Kotresh Hiremath Ravishankar
09:48 AM Feature #58488: mds: avoid encoding srnode for each ancestor in an EMetaBlob log event
More detail about this:
For example for */AAAA/BBBB/CCCC/* we create snapshots under */* and */AAAA/BBBB/*, and la...
Xiubo Li
09:39 AM Feature #58488 (New): mds: avoid encoding srnode for each ancestor in an EMetaBlob log event
This happens via MDCache::predirty_journal_parents() where MDCache::journal_dirty_inode() is called for each ancestor... Venky Shankar
02:56 AM Bug #58482 (Fix Under Review): mds: catch damage to CDentry's first member before persisting
Patrick Donnelly
02:54 AM Bug #58482 (Resolved): mds: catch damage to CDentry's first member before persisting
Patrick Donnelly

01/17/2023

12:01 PM Bug #57985: mds: warning `clients failing to advance oldest client/flush tid` seen with some work...
partial fix (debug aid): https://github.com/ceph/ceph/pull/49766 Venky Shankar

01/13/2023

06:04 PM Bug #58434: Widespread metadata corruption
Further evidence indicates the issue may not be with Ceph; possibly a bug in account management. Still investigating.... Nathan Fish

01/12/2023

07:38 PM Bug #58434 (Rejected): Widespread metadata corruption
One of our CephFS volumes has become corrupt, with user files being substituted for each other with no clear pattern;... Nathan Fish
01:21 PM Bug #57985: mds: warning `clients failing to advance oldest client/flush tid` seen with some work...
Venky Shankar wrote:
> https://bugzilla.redhat.com/show_bug.cgi?id=2134709
>
> Generally seen when the MDS is hea...
Venky Shankar
05:28 AM Bug #58082: cephfs:filesystem became read only after Quincy upgrade
Prayank Saxena wrote:
> okay i see, thanks Xiubo li
>
> i was going through the link and found reset of journal a...
Venky Shankar

01/11/2023

10:55 AM Bug #58219: Test failure: test_journal_migration (tasks.cephfs.test_journal_migration.TestJournal...
https://github.com/ceph/ceph/pull/48418#issuecomment-1378478510 Venky Shankar
09:47 AM Bug #58394 (Triaged): nofail option in fstab not supported
Venky Shankar
09:47 AM Bug #58394: nofail option in fstab not supported
Dhairya, please take this one. I think we would want to get the semantics right for `nofail` option with ceph-fuse an... Venky Shankar
07:59 AM Bug #54017 (Duplicate): Problem with ceph fs snapshot mirror and read-only folders
Duplicate of https://tracker.ceph.com/issues/55313 Milind Changire

01/10/2023

02:22 PM Bug #58411 (Resolved): mds: a few simple operations crash mds
mount a cephfs with multiple active mds daemons on '/mnt/cephfs' and then do the following operations in order:
(1...
Zhansong Gao
01:18 PM Bug #58219: Test failure: test_journal_migration (tasks.cephfs.test_journal_migration.TestJournal...
Tried with a recent build - https://pulpito.ceph.com/vshankar-2022-12-08_04:33:46-fs-wip-vshankar-testing-20221130.04... Venky Shankar
06:57 AM Bug #58219: Test failure: test_journal_migration (tasks.cephfs.test_journal_migration.TestJournal...
This is a build issue which has started to show up recently. Venky Shankar
06:09 AM Bug #58219: Test failure: test_journal_migration (tasks.cephfs.test_journal_migration.TestJournal...
Most likely, this is similar to https://bugzilla.redhat.com/show_bug.cgi?id=1569391 (memory is allocated via libc whe... Venky Shankar
11:50 AM Backport #58409 (In Progress): quincy: doc: document the relevance of mds_namespace mount option
Jos Collin
11:14 AM Backport #58409 (Resolved): quincy: doc: document the relevance of mds_namespace mount option
https://github.com/ceph/ceph/pull/49689 Backport Bot
11:45 AM Backport #58408 (In Progress): pacific: doc: document the relevance of mds_namespace mount option
Jos Collin
11:14 AM Backport #58408 (Resolved): pacific: doc: document the relevance of mds_namespace mount option
https://github.com/ceph/ceph/pull/49688 Backport Bot
11:12 AM Documentation #57673 (Pending Backport): doc: document the relevance of mds_namespace mount option
Venky Shankar
05:11 AM Bug #58392 (Rejected): mds:when want_auth is true, path_traverse should ensure the return Inode i...
Closing this in favor of https://tracker.ceph.com/issues/58395 Venky Shankar
04:32 AM Bug #58392 (Triaged): mds:when want_auth is true, path_traverse should ensure the return Inode is...
Venky Shankar
04:38 AM Feature #58133 (Resolved): qa: add test cases for fscrypt feature in kernel CephFS client
Xiubo Li

01/09/2023

02:11 PM Bug #58394: nofail option in fstab not supported
FWIW, a similar issue was attempted to be fixed for the kclient - https://github.com/ceph/ceph/pull/26992, but all th... Venky Shankar
02:57 AM Bug #58394: nofail option in fstab not supported
Results from a non-debug mount:... Brian Woods
02:56 AM Bug #58394 (Pending Backport): nofail option in fstab not supported
There are several old bug reports on this from 2019, but they should have all been resolved. However, testing on a 1... Brian Woods
02:06 PM Documentation #58393: nfs docs should mention idmap issue, refer to external tools/setup
(temp assigning to Jeff, so that its notified) Venky Shankar
02:06 PM Documentation #58393: nfs docs should mention idmap issue, refer to external tools/setup
Dan Mick wrote:
> I tried to use an NFS mount set up by others, and the files in it were owned by root:root in the u...
Venky Shankar
01:50 PM Bug #58376: CephFS Snapshots are accessible even when it's deleted from the other client
Kotresh Hiremath Ravishankar wrote:
> > Now you can still copy the snapshot file from /mnt
> > [kotresh@fedora buil...
Greg Farnum
01:48 PM Bug #58376 (Triaged): CephFS Snapshots are accessible even when it's deleted from the other client
Venky Shankar
01:41 PM Bug #58395 (Triaged): mds:in openc, if unlink is not finished we should reintegrate the dentry be...
Venky Shankar
08:58 AM Bug #58395: mds:in openc, if unlink is not finished we should reintegrate the dentry before conti...
the crash log zhikuo du
08:57 AM Bug #58395 (Triaged): mds:in openc, if unlink is not finished we should reintegrate the dentry be...
In openc, if unlink is not finished we should reintegrate the dentry before continuing further.
If not, cooperate ...
zhikuo du
01:30 PM Bug #58340 (Triaged): mds: fsstress.sh hangs with multimds (deadlock between unlink and reintegra...
Venky Shankar
10:40 AM Backport #58347 (In Progress): quincy: mds: fragment directory snapshots
Venky Shankar
10:35 AM Backport #58345 (In Progress): quincy: Thread md_log_replay is hanged for ever.
Venky Shankar
10:30 AM Backport #58346 (In Progress): pacific: Thread md_log_replay is hanged for ever.
Venky Shankar
10:23 AM Backport #58350 (In Progress): quincy: MDS: scan_stray_dir doesn't walk through all stray inode f...
Venky Shankar
10:16 AM Backport #58349 (In Progress): pacific: MDS: scan_stray_dir doesn't walk through all stray inode ...
Venky Shankar
09:05 AM Bug #58392: mds:when want_auth is true, path_traverse should ensure the return Inode is auth.
cooperate with: https://tracker.ceph.com/issues/58395 zhikuo du

01/07/2023

03:18 AM Documentation #58393 (New): nfs docs should mention idmap issue, refer to external tools/setup
I tried to use an NFS mount set up by others, and the files in it were owned by root:root in the underlying cephfs. ... Dan Mick

01/06/2023

05:27 PM Bug #58376: CephFS Snapshots are accessible even when it's deleted from the other client

> Now you can still copy the snapshot file from /mnt
> [kotresh@fedora build]$ cp -p /mnt/dir1/.snap/snapshot1/fil...
Kotresh Hiremath Ravishankar
01:22 PM Backport #58254 (In Progress): pacific: mds/PurgeQueue: don't consider filer_max_purge_ops when _...
https://github.com/ceph/ceph/pull/49656 Dhairya Parmar
01:19 PM Backport #58253 (In Progress): quincy: mds/PurgeQueue: don't consider filer_max_purge_ops when _c...
https://github.com/ceph/ceph/pull/49655 Dhairya Parmar
01:11 PM Backport #58348 (In Progress): quincy: "ceph nfs cluster info" shows junk data for non-existent c...
Dhairya Parmar
01:11 PM Backport #58348: quincy: "ceph nfs cluster info" shows junk data for non-existent cluster
https://github.com/ceph/ceph/pull/49654 Dhairya Parmar
10:58 AM Bug #58392 (Rejected): mds:when want_auth is true, path_traverse should ensure the return Inode i...
When want_auth is true(want_dentry is true or false), we should ensure the return Inode is auth.
If not, it can cras...
zhikuo du

01/05/2023

01:28 PM Bug #58376: CephFS Snapshots are accessible even when it's deleted from the other client
Venky Shankar wrote:
> Kotresh Hiremath Ravishankar wrote:
> > The issue is seen by upstream user. The snapshot is ...
Kotresh Hiremath Ravishankar
09:32 AM Bug #58376: CephFS Snapshots are accessible even when it's deleted from the other client
Kotresh Hiremath Ravishankar wrote:
> The issue is seen by upstream user. The snapshot is still accessible from a cl...
Venky Shankar
07:36 AM Bug #58082: cephfs:filesystem became read only after Quincy upgrade
okay i see, thanks Xiubo li
i was going through the link and found reset of journal and session resolved the issue...
Prayank Saxena
05:21 AM Bug #58082: cephfs:filesystem became read only after Quincy upgrade
Prayank Saxena wrote:
> Thanks Xiubo Li for the update
>
> We are facing similar issue currently where client I/O...
Xiubo Li
05:03 AM Bug #58082: cephfs:filesystem became read only after Quincy upgrade
Thanks Xiubo Li for the update
We are facing similar issue currently where client I/O is not visible in ceph statu...
Prayank Saxena
04:09 AM Backport #58254: pacific: mds/PurgeQueue: don't consider filer_max_purge_ops when _calculate_ops
Dhairya, please take this one. Venky Shankar
04:05 AM Backport #58344 (In Progress): quincy: mds: switch submit_mutex to fair mutex for MDLog
Xiubo Li
04:04 AM Backport #58343 (In Progress): pacific: mds: switch submit_mutex to fair mutex for MDLog
Xiubo Li

01/04/2023

10:47 AM Documentation #57673 (In Progress): doc: document the relevance of mds_namespace mount option
Jos Collin
10:04 AM Bug #58376 (Triaged): CephFS Snapshots are accessible even when it's deleted from the other client
The issue is seen by upstream user. The snapshot is still accessible from a client which was copying it even when it'... Kotresh Hiremath Ravishankar
06:19 AM Bug #58082: cephfs:filesystem became read only after Quincy upgrade
Prayank Saxena wrote:
> Prayank Saxena wrote:
> > Hello Team,
> >
> > We got the issue similar to 'mds read-only...
Xiubo Li
05:04 AM Bug #58082: cephfs:filesystem became read only after Quincy upgrade
Prayank Saxena wrote:
> Hello Team,
>
> We got the issue similar to 'mds read-only' in Pacific 16.2.9 where one w...
Prayank Saxena
04:56 AM Bug #58082: cephfs:filesystem became read only after Quincy upgrade
Hello Team,
We got the issue similar to 'mds read-only' in Pacific 16.2.9 where one write commit failed and made t...
Prayank Saxena
04:53 AM Bug #52260: 1 MDSs are read only | pacific 16.2.5
Hello Team,
We got the issue similar to 'mds read-only' in Pacific 16.2.9 where one write commit failed and made t...
Prayank Saxena
 

Also available in: Atom