Project

General

Profile

Activity

From 11/10/2016 to 12/09/2016

12/09/2016

09:06 AM Bug #11585 (Can't reproduce): kdb crash around __ceph_remove_cap() while dirfrag.sh
Zheng Yan
09:01 AM Bug #17191 (Resolved): kclient accesses "/" when mounting a subpath
Zheng Yan

12/08/2016

12:09 PM Bug #18041: periodically kernel crashes with CephFS
Thanks for testing it. The slowdown sounds like a different problem entirely and I'd suggest opening a separate bug f... Jeff Layton

12/07/2016

02:26 PM Bug #18161: kernel client failing to look up mds_namespace gives ENOENT (but it exists)
Ah, that could be it. I'll look again. John Spray
02:17 PM Bug #18161: kernel client failing to look up mds_namespace gives ENOENT (but it exists)
can't reproduce this manually. maybe the -ENOENT was from mds up check, please try adding 'norequire_active_mds' moun... Zheng Yan
12:10 AM Bug #18161 (Resolved): kernel client failing to look up mds_namespace gives ENOENT (but it exists)
See this failure:
http://pulpito.ceph.com/jspray-2016-12-06_12:37:38-kcephfs:recovery-master-testing-basic-smithi/61...
John Spray
08:28 AM Bug #18041: periodically kernel crashes with CephFS
Tested this patch (https://github.com/ceph/ceph-client/commit/06ae6a76e4570cba948084f6a1c0d206485d1706) on two server... Aurimas Lapiene

12/05/2016

12:09 PM Bug #18041 (In Progress): periodically kernel crashes with CephFS
Patch posted to mailing list and merged into ceph-client/testing branch. Jeff Layton
11:57 AM Bug #18148 (Resolved): Fix unsafe d_parent handling in kcephfs
As Zheng points out, there are several places in the kernel client that access d_parent without proper locking:
> ...
Jeff Layton

12/03/2016

12:38 AM Bug #18130: soft lockups in ceph.ko
looks like our code can't handle iov_iter of type ITER_PIPE Zheng Yan

12/02/2016

06:20 PM Bug #18130 (Resolved): soft lockups in ceph.ko
While running xfstests vs. a ceph vstart cluster, I hitting softlockups in test generic/095. Looks like a we're cycli... Jeff Layton
12:14 AM Bug #18041: periodically kernel crashes with CephFS
Jeff Layton wrote:
> Zheng Yan wrote:
>
> >
> > If we only want inode, it's better to use getattr to validate d...
Zheng Yan

12/01/2016

03:17 PM Bug #18041: periodically kernel crashes with CephFS
Zheng Yan wrote:
>
> If we only want inode, it's better to use getattr to validate dentry (getattr is basically ...
Jeff Layton
09:05 AM Bug #18041: periodically kernel crashes with CephFS
Jeff Layton wrote:
> Actually, I think we want this patch. Now that we're not manipulating the dcache with this call...
Zheng Yan

11/30/2016

09:06 PM Bug #18041: periodically kernel crashes with CephFS
Actually, I think we want this patch. Now that we're not manipulating the dcache with this call, we must handle the d... Jeff Layton
07:38 PM Bug #18041: periodically kernel crashes with CephFS
Did some basic testing (with a doctored kernel that always revalidates dentries) and this seems to do the right thing... Jeff Layton
03:58 PM Bug #18041: periodically kernel crashes with CephFS
Ok, I think this patch will probably do the right thing. Only tested for compilation so far, but I'll see if I can gi... Jeff Layton
02:49 PM Bug #18041: periodically kernel crashes with CephFS
Jeff Layton wrote:
> >
> > I think we can do this (maybe need minor modification to the code). But I don't underst...
Jeff Layton
02:37 PM Bug #18041: periodically kernel crashes with CephFS
got it. the whole point is avoid d_invalidated Zheng Yan
02:26 PM Bug #18041: periodically kernel crashes with CephFS
Zheng Yan wrote:
> Jeff Layton wrote:
> > Zheng Yan wrote:
> >
> > >
> > > I'm little worry about that d_reval...
Jeff Layton
01:51 PM Bug #18041: periodically kernel crashes with CephFS
Jeff Layton wrote:
> I think the thing we want to do here is to not set r_locked_dir, and pass a reference to the in...
Zheng Yan
01:49 PM Bug #18041: periodically kernel crashes with CephFS
Jeff Layton wrote:
> Zheng Yan wrote:
>
> >
> > I'm little worry about that d_revalidate does sleep lookup whil...
Zheng Yan
01:39 PM Bug #18041: periodically kernel crashes with CephFS
I think the thing we want to do here is to not set r_locked_dir, and pass a reference to the inode that was the resul... Jeff Layton
01:22 PM Bug #18041: periodically kernel crashes with CephFS
Aurimas, If you can compile kernel from source. The detach patch (for 4.8.x kernel) workaround this issue. Zheng Yan

11/29/2016

02:18 PM Bug #18041: periodically kernel crashes with CephFS
Zheng Yan wrote:
>
> I'm little worry about that d_revalidate does sleep lookup while parent is not locked. I th...
Jeff Layton
01:04 PM Bug #18041: periodically kernel crashes with CephFS
With kernel v4.5.4 there's no kernel panics, but CPUs started to soft lockup on one node every few hours (one node so... Aurimas Lapiene
02:38 AM Bug #18041: periodically kernel crashes with CephFS
Jeff Layton wrote:
> Ok, thanks. I see that you had written that in the original description too. If you have the ti...
Zheng Yan
02:17 AM Bug #18041: periodically kernel crashes with CephFS
Jeff Layton wrote:
> From ceph_d_revalidate:
>
> [...]
>
> The result handling after ceph_mdsc_do_request look...
Zheng Yan

11/28/2016

07:19 PM Bug #18041: periodically kernel crashes with CephFS
From ceph_d_revalidate:... Jeff Layton
05:07 PM Bug #18041: periodically kernel crashes with CephFS
Ok, thanks. I see that you had written that in the original description too. If you have the time, then it might be g... Jeff Layton
03:51 PM Bug #18041: periodically kernel crashes with CephFS
There are 6 nodes that have mounted the same cephfs directory and 4 of them doing active R/W requests on same paths (... Aurimas Lapiene
03:44 PM Bug #18041: periodically kernel crashes with CephFS
That BUG_ON also looks suspicious. We may need to hold the rcu_read_lock there to ensure that dn->d_parent doesn't va... Jeff Layton
03:31 PM Bug #18041: periodically kernel crashes with CephFS
Sure...
First, does this kernel have any patches beyond what went into v4.8.4 ? There are some other BUG_ONs in th...
Jeff Layton
12:59 PM Bug #18041: periodically kernel crashes with CephFS
fs/ceph/inode.c:1272 is BUG_ON(d_inode(dn->d_parent) != dir); For cephfs case, I think only thing that can change d_p... Zheng Yan
11:52 AM Bug #18041: periodically kernel crashes with CephFS
could you please check if this issue exists in 4.5 kernel Zheng Yan
09:03 AM Bug #18041: periodically kernel crashes with CephFS
It was quite regular load. Graphs attached. Aurimas Lapiene
02:46 AM Bug #18041: periodically kernel crashes with CephFS
what workloads trigger this bug? Zheng Yan
02:37 PM Feature #17524: krbd: support disabling auto-exclusive lock transition logic
Reference: https://github.com/kubernetes/kubernetes/issues/33013 Ilya Dryomov

11/27/2016

12:46 PM Bug #18041: periodically kernel crashes with CephFS
... Donatas Abraitis
12:42 PM Bug #18045 (Duplicate): BUG_ON(d_inode(dn->d_parent) != dir) in fs/ceph/inode.c
Dup of #18041, moved from the fs project. Ilya Dryomov
12:21 PM Bug #18045 (Duplicate): BUG_ON(d_inode(dn->d_parent) != dir) in fs/ceph/inode.c
On IRC:
< donatas> kernel panics ~3-5 every day...
Ilya Dryomov

11/25/2016

02:30 PM Bug #18041: periodically kernel crashes with CephFS
More info about the incident.
Topology:
2 osd nodes with 24 total osds
3 monitor/mds nodes
6 cephfs kernel clie...
Aurimas Lapiene
01:56 PM Bug #18041: periodically kernel crashes with CephFS
Any more clues about what else might have changed in the last few days? John Spray
01:11 PM Bug #18041 (Resolved): periodically kernel crashes with CephFS
Hi,
last few days we are having ~5-15 kernel crashes according to:...
Donatas Abraitis

11/19/2016

11:56 AM Bug #17825 (Resolved): 4.8.6's cephfs.ko can't read any files from old fs running ceph-10.2.3, bu...
in 4.8.9. Ilya Dryomov

11/14/2016

09:55 PM Bug #17825 (Pending Backport): 4.8.6's cephfs.ko can't read any files from old fs running ceph-10...
https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=3890dce1d3a8b9fe3bc36de99496792e468cd079 i... Ilya Dryomov
01:45 PM Feature #17805: Match fuse_require_active_mds behaviour in kernel client
Can you test it with the branch here? https://github.com/ceph/ceph-qa-suite/pull/1156
You will need to remove the...
John Spray

11/10/2016

08:54 AM Feature #17805: Match fuse_require_active_mds behaviour in kernel client
https://github.com/ceph/ceph-client/commit/8d663773cafe6815da239bdbdba4c8d34e0793bb Zheng Yan
 

Also available in: Atom