Project

General

Profile

Activity

From 06/11/2019 to 07/10/2019

07/10/2019

06:51 PM Bug #21420: ceph_osdc_writepages(): pre-allocated osdc->msgpool_op messages vs large number of sn...
No, I don't think so. It now falls back to a regular allocation that can fail, while that code path is specifically ... Ilya Dryomov
10:28 AM Bug #21420: ceph_osdc_writepages(): pre-allocated osdc->msgpool_op messages vs large number of sn...
I was wondering if this bug could be closed, as it seems that it has already been fixed by the "libceph: preallocate ... Luis Henriques
04:38 PM Bug #40706: races in ceph_uninline_data
Patchset posted upstream:
https://marc.info/?l=ceph-devel&m=156277513921656&w=2
Jeff Layton
01:46 PM Bug #40706: races in ceph_uninline_data
Not exactly.
After the uninlining is done, the i_inline_version is not updated until much later, after the i_rwsem...
Jeff Layton
01:28 PM Bug #40706: races in ceph_uninline_data
For ceph_write_iter(), it's protected by i_rwsem. But in ceph_page_mkwrite, i_rwsem is not locked. Zheng Yan
01:09 PM Bug #40706: races in ceph_uninline_data
My initial thinking was to use the page lock of the first page in the cache to serialize the callers, since we do tha... Jeff Layton

07/09/2019

07:13 PM Bug #40706: races in ceph_uninline_data
I think we need to make ceph_uninline_data idempotent, such that if it gets called a second time it becomes a no-op. Jeff Layton
07:12 PM Bug #40706 (Won't Fix): races in ceph_uninline_data
I've been staring at ceph_uninline_data() for a while now, and I think I've convinced myself that there are potential... Jeff Layton

07/02/2019

06:47 PM Feature #16480: kernel inline data write suport
I'm making some progress on this, but it has been rather slow going. I've made a couple of draft patchsets, but they'... Jeff Layton
03:10 AM Bug #40606: the mount point of cephfs rise up permission denied
Patrick Donnelly wrote:
> This question is more appropriate for ceph-users.
>
> In short: the kernel client does ...
Hughen X

07/01/2019

09:49 PM Bug #40606 (Rejected): the mount point of cephfs rise up permission denied
This question is more appropriate for ceph-users.
In short: the kernel client does not handle being blacklisted we...
Patrick Donnelly
10:31 AM Bug #40606 (Rejected): the mount point of cephfs rise up permission denied
I mount a directory from cephfs to local filesystem.... Hughen X

06/27/2019

08:10 PM Feature #16480 (In Progress): kernel inline data write suport
Jeff Layton
08:09 PM Feature #16480: kernel inline data write suport
I spent some time looking at this today, and I think I see a path forward. The current kclient code mostly tries to u... Jeff Layton
07:34 PM Bug #39258: add statx support to kcephfs
Patches are in for this, and there are some basic statx tests in xfstests. For now, we may have to settle for that as... Jeff Layton

06/21/2019

10:45 AM Bug #40482 (Duplicate): [libceph] page allocation failure when parsing new OSD map
Ilya Dryomov
10:34 AM Bug #40482 (Duplicate): [libceph] page allocation failure when parsing new OSD map
Jun 01 02:03:49 rhnode kernel: kworker/15:1: page allocation failure: order:7, mode:0x104050
Jun 01 02:03:49 rhnode ...
Prashant D
10:43 AM Bug #40481: osdmap->osd_addr allocation is susceptible to memory fragmentation
... Ilya Dryomov
10:33 AM Bug #40481 (Resolved): osdmap->osd_addr allocation is susceptible to memory fragmentation
... Ilya Dryomov

06/18/2019

07:33 AM Bug #38224: Race handling ceph_snap_context in kernel client
fixed by "ceph: fix ci->i_head_snapc leak" in testing branch Zheng Yan

06/17/2019

01:30 PM Bug #40339: kernel BUG at fs/ceph/mds_client.c:600! invalid opcode: 0000 [#1] SMP
Zheng Yan
01:08 PM Bug #40339: kernel BUG at fs/ceph/mds_client.c:600! invalid opcode: 0000 [#1] SMP
fixed by "ceph: remove request from waiting list before unregister" in testing branch Zheng Yan

06/16/2019

01:51 PM Bug #40340: kernel client stuck at opening forever after network outage.
hmm, re-read the code seems we will backoff and retry every 5mins....not sure if I read it correctly and how it hit i... Xiaoxi Chen

06/14/2019

01:35 PM Bug #40362: amending caps with "ceph auth caps" vs using an existing client
Note that these irregular I/O errors can occur today (i.e. with no invalidation in the kernel client) if enough time ... Ilya Dryomov
12:36 PM Bug #40362: amending caps with "ceph auth caps" vs using an existing client
If we simply invalidate, things become super confusing. If the OSD that pool2/img header resides on is open, "rbd ma... Ilya Dryomov
12:14 PM Bug #40362 (New): amending caps with "ceph auth caps" vs using an existing client
... Ilya Dryomov
07:40 AM Bug #40339: kernel BUG at fs/ceph/mds_client.c:600! invalid opcode: 0000 [#1] SMP
request was unregistered twice. one is from wait_requests, another is from __wake_requests Zheng Yan
03:35 AM Bug #40339: kernel BUG at fs/ceph/mds_client.c:600! invalid opcode: 0000 [#1] SMP
Thanks zheng, can you explain more backgorund? Xiaoxi Chen
12:23 AM Bug #40339: kernel BUG at fs/ceph/mds_client.c:600! invalid opcode: 0000 [#1] SMP
maybe this one... Zheng Yan
02:49 AM Bug #39523: small write in ab mode does not change mtime
can't reproduce this issue on 5.0.16 kernel + ceph v12.2.12 Zheng Yan
02:34 AM Bug #37713 (Resolved): Centos 7 kernel client overwriting files
Fixed in 7.5 (kernel-3.10.0-862.33.1.el7) and 7.6 (kernel-3.10.0-957.16.1.el7). Zheng Yan
02:08 AM Bug #38824 (Resolved): kclient: ceph: remove duplicated filelock ref increase
Zheng Yan
02:05 AM Bug #40340: kernel client stuck at opening forever after network outage.
maybe calling peer_reset after tcp timeout is a better fix Zheng Yan
01:12 AM Bug #40340: kernel client stuck at opening forever after network outage.
140 06/13/2019 09:21:22 AM umount /import/ceph/
141 06/13/2019 09:23:05 AM mount -a
The time of umount
Xiaoxi Chen
01:09 AM Bug #40340: kernel client stuck at opening forever after network outage.
Negative.
The peer_reset get called during eviction but not after connection establish timeout.
2264570:Jun 13 ...
Xiaoxi Chen
12:01 AM Bug #40340: kernel client stuck at opening forever after network outage.
does peer_reset() get called after tcp timeout Zheng Yan

06/13/2019

05:08 PM Bug #40340: kernel client stuck at opening forever after network outage.
Step to reproduce
1. [MDS]ceph --daemon /path-to-admin-socket/ session evict <client_id>
2. [Client] After 2M...
Xiaoxi Chen
05:02 PM Bug #40340 (Can't reproduce): kernel client stuck at opening forever after network outage.
We dont retry(in ceph code) for sending create_session_open_msg. Thus once a network outage happen and mds evict the ... Xiaoxi Chen
04:48 PM Bug #40339 (Resolved): kernel BUG at fs/ceph/mds_client.c:600! invalid opcode: 0000 [#1] SMP
It happens on 3.10.0-957.1.3.el7.x86_64
Back to that time,
1. there is a mds session stucking in "opening" due ...
Xiaoxi Chen
 

Also available in: Atom