Activity
From 09/02/2014 to 10/01/2014
09/23/2014
- 02:00 PM Bug #8568: libceph: kernel BUG at net/ceph/osd_client.c:885
- BUG_ON(!list_empty(&req->r_req_lru_item)) in __kick_osd_requests()
Can't reproduce but need to look harder into ho... - 09:39 AM Bug #9573 (New): krbd: investigate a dd-in-a-loop slowdown
- Reported at the bottom of #8818.
09/22/2014
- 04:08 AM Bug #8979 (Resolved): GPF kernel panics - auth?
- Landed in 3.17-rc5. Opened #9560 and #9561 for the issues mentioned above.
- 04:04 AM Bug #9561 (Rejected): libceph: do not crash if auth reply is not understood
- 04:02 AM Bug #9560 (Rejected): libceph: msg kmalloc failure handling on the reply path
09/20/2014
09/19/2014
- 06:58 AM Bug #9533 (Duplicate): kcephfs: fail to send requests initiated during mds restart
- this was an old bug, patch was missing from running kernel.
ceph: fix kick_requests()
09/18/2014
- 09:43 PM Bug #9533 (Duplicate): kcephfs: fail to send requests initiated during mds restart
- mds sees...
- 01:13 AM Bug #9507: calling llistxattr(2) on a symlink crashes the client
- ...
09/17/2014
- 11:57 PM Bug #9458: client wrongly fenced
- is the client using 3.16 kernel? possibly due to missing following commit...
- 02:45 PM Bug #9458: client wrongly fenced
- The kernel client is definitely doing something wrong here, but I don't know what — the userspace messenger is not in...
- 02:38 PM Bug #9458: client wrongly fenced
- The MDS went into reconnect at 4:59:50...
- 11:09 AM Bug #9458: client wrongly fenced
- Taking a look; luckily we have at least *some* of the logging...
- 08:17 AM Bug #9458: client wrongly fenced
- mds restarted and teuthology failed to reconnect again, 07:30:34.485721
- 07:18 AM Bug #9458: client wrongly fenced
- teuthology was fenced again. not sure it was during a mds restart this time, either. notably the monitors went offl...
- 09:48 AM Bug #9507 (Resolved): calling llistxattr(2) on a symlink crashes the client
- The code hits a "BUG();" line at https://github.com/ceph/ceph-client/blob/7e8a295295775ec9e05411cefc578ff4bfc94740/fs...
- 08:51 AM Bug #9505 (Duplicate): kcephfs: client gets stuck in reconnect loop?
- ...
09/16/2014
- 10:07 AM Bug #8818: IO Hang on raw rbd device - Workqueue: ceph-msgr con_work [libceph]
- Andrei,
No, I haven't, but plan to try harder. I am however seeing an extreme slowdown, will open a ticket to tak... - 02:49 AM Bug #8818: IO Hang on raw rbd device - Workqueue: ceph-msgr con_work [libceph]
- Ilya,
I was wondering if you've managed to verify my findings? Has anyone experienced similar behaviour as I am?
...
09/15/2014
09/14/2014
- 03:00 PM Bug #4614: Root cephfs does not mount at boot on Ubuntu 12.04
- Howdy. I do not know. :) It has been over a year and I moved on long ago.
- 01:22 PM Bug #8818: IO Hang on raw rbd device - Workqueue: ceph-msgr con_work [libceph]
- Done more testing. The hang tasks happen regardless of the block size (tried 4K, 1M ,4M). Without rbd cache enabled t...
- 11:24 AM Bug #8818: IO Hang on raw rbd device - Workqueue: ceph-msgr con_work [libceph]
- Okay, just tried with block size 4M and after about 20 dd runs I got a hang task.
Also, I've noticed that the spee... - 11:06 AM Bug #8818: IO Hang on raw rbd device - Workqueue: ceph-msgr con_work [libceph]
- Ilya,
Not sure as I've not left it to run for a long time. There was no progress for about 30 minutes, so I've reb... - 09:11 AM Bug #8818: IO Hang on raw rbd device - Workqueue: ceph-msgr con_work [libceph]
- Do those dds eventually complete?
- 09:04 AM Bug #8818: IO Hang on raw rbd device - Workqueue: ceph-msgr con_work [libceph]
- Ilya,
I've tried the kernel that was patched, but I am still experiencing hang tasks. Although less often than bef...
09/13/2014
- 09:29 PM Bug #3087 (Can't reproduce): Hung ceph_msg_kfree
- 09:29 PM Bug #4614: Root cephfs does not mount at boot on Ubuntu 12.04
- Is this still an issue? We have not been able to replicate. my fstab line looks like:
10.214.133.104:6789,10.21... - 09:25 PM Bug #8979 (Pending Backport): GPF kernel panics - auth?
- 09:32 AM Bug #9458 (Resolved): client wrongly fenced
- ...
09/12/2014
- 01:18 PM Bug #9426: kcephfs: soft lockup in handle mds map
- Added my Reviewed-by.
- 01:18 PM Bug #9426: kcephfs: soft lockup in handle mds map
- 05:26 AM Bug #9432 (In Progress): kcephfs: null pointer deref in posix_acl_create
- it seems that ACL code can't handle traceless reply.
09/11/2014
- 09:06 AM Bug #9432 (Resolved): kcephfs: null pointer deref in posix_acl_create
- ...
- 07:08 AM Bug #9426: kcephfs: soft lockup in handle mds map
- maybe this can be fixed by: "ceph: protect kick_requests() with mdsc->mutex" and "ceph: make sure request isn't in an...
09/10/2014
- 07:21 PM Bug #9426: kcephfs: soft lockup in handle mds map
- looks like it was waiting for i_ceph_lock, did the soft lockup recover?
- 06:03 PM Bug #9426 (Resolved): kcephfs: soft lockup in handle mds map
- ...
09/09/2014
- 09:58 AM Bug #9192: krbd: poor read (about 10%) vs write performance
- If it's the queueing change that is causing this, 3.16.x with the fix will show the same results as 3.17, but it woul...
- 06:46 AM Bug #9192: krbd: poor read (about 10%) vs write performance
- I plan to test on the Ubuntu PPA 3.16.x once the fix gets into that branch. Will that be good enough?
- 03:26 AM Bug #9192: krbd: poor read (about 10%) vs write performance
- Eric,
To see if it's the new queueing regression, it'd be best to compare wip-request-fn and wip-request-fn~1. Th... - 12:41 AM Bug #8818: IO Hang on raw rbd device - Workqueue: ceph-msgr con_work [libceph]
- The patch is in the queue, just haven't been applied yet: https://lkml.org/lkml/2014/9/3/918.
09/08/2014
- 07:40 AM Bug #8806: libceph: must use new tid when watch is resent
- My tests confirmed that wip-watch-tid-8806 fixes this particular krbd bug. However, with thrashosds thrown into the ...
- 03:23 AM Bug #8818: IO Hang on raw rbd device - Workqueue: ceph-msgr con_work [libceph]
- Guillaume Subiron wrote:
> Xavier Trilla wrote:
>
> > Ok, well try 3.16.2 once it's released and we will let you ... - 03:20 AM Bug #8818: IO Hang on raw rbd device - Workqueue: ceph-msgr con_work [libceph]
- Xavier Trilla wrote:
> Ok, well try 3.16.2 once it's released and we will let you know.
Sadly, your patch has n...
09/07/2014
- 11:02 PM Feature #9345: look into using MSG_MORE and MSG_SENDPAGE_NOTLAST more
- If I understand correctly, MSG_SENDPAGE_NOLAST is an internal flag and is set by pipe_to_sendpage. Should user be set...
09/05/2014
- 06:09 AM Bug #8818: IO Hang on raw rbd device - Workqueue: ceph-msgr con_work [libceph]
- I'll try to reproduce German's fio problem.
To those who are stuck on 3.15 for some reason but can rebuild their k... - 05:36 AM Bug #8806 (Fix Under Review): libceph: must use new tid when watch is resent
- wip-watch-tid-8806
09/04/2014
- 04:37 PM Bug #9355 (Closed): rbd: map fails with EINVAL inside a container
- https://lists.linuxcontainers.org/pipermail/lxc-users/2013-October/005795.html
carmstrong on irc had the same issu... - 04:36 PM Bug #8818: IO Hang on raw rbd device - Workqueue: ceph-msgr con_work [libceph]
- We're seeing this on CoreOS, kernel 3.15.8. Upgrading the kernel ourselves isn't an option, so this'll block RBD usag...
- 10:49 AM Feature #9345 (Resolved): look into using MSG_MORE and MSG_SENDPAGE_NOTLAST more
- Mike Christie pointed out the following:...
- 10:45 AM Feature #9344 (Resolved): add a TCP_NODELAY option
- From Chaitanya Huilgol <Chaitanya.Huilgol@sandisk.com>...
Also available in: Atom