Activity
From 02/12/2013 to 03/13/2013
03/13/2013
- 07:57 PM Feature #3761: kernel messenger: need to support multiple ops per request
- What I've been trying to do today is make it so a message's
header length is defined in a centralized location, base... - 06:28 PM Bug #4388 (Fix Under Review): rbd import broken
- The wip-rbd-import branch removes the fiemap usage on top of the 'next' branch. I think it should be backported to bo...
- 09:37 AM Bug #4388: rbd import broken
- Well you know the ceph sources better for sure. It's only that I'm quite surprised that such a grave bug (it causes c...
- 03:40 PM Bug #4285 (Resolved): rbd copy.sh test failing on counting images
- commit:436e5be950154fdbbd9e1cfaf4267be6159249d5
- 11:53 AM Bug #2700: blkdeviotune method at libvirt doesn`t work on RBD volumes
- Sage just pinged me about this bug report.
I'll pick this up and write patch for libvirt to fix this. - 11:52 AM Bug #2400 (Resolved): xfstest: failed #84
- we're running this one now.
- 11:45 AM Bug #4047: removing a non-existing rbd image logs error in osd logs
- 11:44 AM Bug #4012 (Won't Fix): rbd: image creation behaviour has to be uniform across bobtail and argonau...
- 11:37 AM Bug #3600 (Duplicate): rbd: assert in objectcacher destructor after flatten
- 11:36 AM Bug #3619: librbd: read_iterate sparse behavior broken
- 11:35 AM Bug #2689: qemu iozone test hangs
- 10:31 AM Subtask #4091: ObjectCacher: optionally make readx/writex calls never block
- wip-rbd-cache-aio
- 09:59 AM Bug #4392: kclient: libceph: osd ops send out of order
- On the userland side of things, we put the requests to be resent into a map<tid,Request*> so that they get sorted bef...
- 08:40 AM Bug #4392: kclient: libceph: osd ops send out of order
- All osd requests are sent in the order they sit on the
osd client req_unsent list. The tid for an osd request
is s...
03/12/2013
- 06:18 PM Fix #4429 (Fix Under Review): libceph: support a list of data items in a message
- The following patches have been posted for review:
[PATCH] libceph: collapse all data items into one
[PATCH... - 02:57 PM Fix #4429 (Duplicate): libceph: support a list of data items in a message
- This will require a number of patches to complete, but
it's a logical unit of work.
An abstracted data item is no... - 02:50 PM Feature #3761: kernel messenger: need to support multiple ops per request
- Is 4125 last thing left to do?
- 02:12 PM Feature #3761: kernel messenger: need to support multiple ops per request
- I haven't updated this in a long time.
I have now implemented the majority of what I described
in my "prototyping... - 02:03 PM Bug #4284 (Resolved): rbd: bio getting set more than once in a message?
- This has been reviewed and is committed to the "testing" branch.
c61ffd1 libceph: set response data fields earlier
- 02:01 PM Subtask #4263 (Resolved): libceph: clearly abstract message data operations
- The following have been reviewed and are now committed to the
ceph-client "testing" branch.
634a072 libceph: isol... - 01:59 PM Bug #4407 (Resolved): libceph: kill the message trail
- In order to do this, the osd client had to stop using
the trail, and doing that required enabling the use of
a page... - 01:57 PM Subtask #4406 (Resolved): libceph: data length should be sum of ops length
- This has been reviewed and is now committed to the testing branch.
58516b5 libceph: let osd ops determine request ... - 01:55 PM Cleanup #4403 (Resolved): libceph: refactor messenger for multiple data sources
- These have been reviewed and are now committed to the testing branch.
5478489 libceph: use local variables for mes... - 01:54 PM Bug #4402 (Resolved): libceph: no need to clear bio_iter field any more
- Josh reviewed this and it has been committed to the testing branch.
b0e22c9 libceph: don't clear bio_iter in prepa... - 01:54 PM Subtask #4401 (Resolved): libceph: record bytes not page count fore requests
- Josh reviewed these and they have now been committed
to the testing branch:
c9582368 libceph: record byte count n... - 01:52 PM Cleanup #4400 (Resolved): libceph: various cleanups
- Josh has reviewed these and they have now been committed
to the testing branch:
1e82218 libceph: define CEPH_MSG_... - 04:57 AM Bug #4285 (Fix Under Review): rbd copy.sh test failing on counting images
- Proposed fix in wip-4285 commit commit:0a5e92c70b85448ee16326f66cd0a7e4c7f6b1a4 has gone through a whole night of bei...
03/11/2013
- 05:58 PM Bug #4388: rbd import broken
- I'm pretty sure it's fiemap since that's almost the only difference between importing from a file and importing from ...
- 05:47 PM Subtask #4091 (In Progress): ObjectCacher: optionally make readx/writex calls never block
- 05:40 PM Bug #4364: ObjectCacher: inconsistency after flatten
- If this doesn't cause any problems, it should be backported to bobtail. Leaving in testing until then.
- 05:19 PM Bug #4285: rbd copy.sh test failing on counting images
- recetn log: ubuntu@teuthology:/a/teuthology-2013-03-10_01:00:05-regression-master-testing-gcov/20671
- 11:57 AM Bug #4285 (In Progress): rbd copy.sh test failing on counting images
- ...
- 10:29 AM Bug #4285: rbd copy.sh test failing on counting images
- Latest logs as per last Friday....
- 04:53 PM Feature #2630: teuthology: add task to run qemu-iotests against rbd
- 04:53 PM Feature #2637: teuthology: task for running a vm using libvirt
- 04:53 PM Feature #2557: QEMU support for image locking
- 04:53 PM Bug #3737: Higher ping-latency observed in qemu with rbd_cache=true during disk-write
- 12:27 PM Bug #4352 (Resolved): librbd: increased memory usage on bobtail vs argonaut (and with no caching)
- 11:34 AM Bug #4352: librbd: increased memory usage on bobtail vs argonaut (and with no caching)
- cherry-picked to bobtail.
03/10/2013
- 02:40 PM Subtask #4406 (Fix Under Review): libceph: data length should be sum of ops length
- This patch has been posted for review:
[PATCH] libceph: let osd ops determine request data length - 12:37 PM Subtask #4406 (Resolved): libceph: data length should be sum of ops length
- Currently the osd client sets the length of the outgoing
data portion of a message based on a value that's passed
i... - 02:39 PM Bug #4402 (Fix Under Review): libceph: no need to clear bio_iter field any more
- This has been posted for review:
[PATCH] libceph: don't clear bio_iter in prepare_write_message() - 01:45 PM Bug #4407 (Fix Under Review): libceph: kill the message trail
- 01:44 PM Bug #4407: libceph: kill the message trail
- I have posted for review the following patches that resolve this.
[PATCH 1/4] libceph: have osd requests support p... - 01:23 PM Bug #4407 (Resolved): libceph: kill the message trail
- The ceph messenger supports a pagelist for its data in
addition to the trail (which is also a pagelist).
The trai... - 01:42 PM Cleanup #4403: libceph: refactor messenger for multiple data sources
- Well, as I should have expected, the testing was important.
I found two bugs in one of these patches, and reposted a...
03/09/2013
- 12:47 PM Bug #4388: rbd import broken
- @Josh To make sure it's no kernel bug I just upgraded this single host to the latest mainline stable kernel (3.8.2-03...
- 03:28 AM Bug #4388: rbd import broken
- @Josh: are you sure? I mean I'm using a stable kernel versuion, not an rc or any custom build. XFS is known for it's ...
- 08:45 AM Cleanup #4403: libceph: refactor messenger for multiple data sources
- I have posted these for review. But I updated them
and I need to re-test them (I normally do that before
I post th... - 08:29 AM Cleanup #4403 (Resolved): libceph: refactor messenger for multiple data sources
- This really was work needed to make it easier to have
the messenger deal with multiple sources of data, and
to trea... - 08:25 AM Bug #4402 (Resolved): libceph: no need to clear bio_iter field any more
- This became obvious as I was refactoring some messenger
code in order to support multiple sources of data (4125).
... - 08:12 AM Subtask #4401 (Resolved): libceph: record bytes not page count fore requests
- In working through implementing a cursor to traverse
multiple hunks of data for a message I discovered that
while a... - 07:20 AM Cleanup #4400 (Resolved): libceph: various cleanups
- I was going to do this before but now I've posted the patches
for review, so I'm doing it after the fact. They're n...
03/08/2013
- 01:23 PM Feature #2232 (New): qemu: resize guest disk when rbd image is resized
- 01:23 PM Feature #2231 (New): librbd: expose header change (resize?) via api
- 01:23 PM Feature #1662 (New): libvirt: obscure qemu/rbd secrets
- 01:22 PM Feature #3205 (New): doc: pool permissions for cloning use-case
- 01:22 PM Feature #2297 (New): ObjectCacher: mark buffers mergeable for ksm
- 01:20 PM Feature #4089: rbd: improve small I/O performance
- 01:18 PM Feature #4084: rbd: incremental backups
- 12:57 PM Bug #4388: rbd import broken
- Yeah, I did that for stdin but left the file algorithm alone.
- 12:20 PM Bug #4388: rbd import broken
- we should switch to using SEEK_HOLE and SEEK_DATA.. that is supposed to work regardless of sync state.
i thought, ... - 12:16 PM Bug #4388: rbd import broken
- Yes, this probably means fiemap is broken on xfs with that kernel. The image file you're importing wasn't just writte...
- 11:24 AM Bug #4388: rbd import broken
- Import from stdin seems to work fine. I tried it twice, both times it gave the correct checksum. Just to make sure I ...
- 09:23 AM Bug #4388: rbd import broken
- Distro: Ubuntu quantal
Kernel: 3.7.10-030710-generic #201302271235 SMP Wed Feb 27 17:36:27 UTC 2013 x86_64 x86_64 x8... - 09:20 AM Bug #4388: rbd import broken
- Could you provide a few more details on your setup:
- what kernel and distro?
- what fs are the files being impor... - 03:08 AM Bug #4388 (Resolved): rbd import broken
- Im tried to import a vm image (10 GB) into a bobtail 0.56.3 (6eb7e15a4783b122e9b0c85ea9ba064145958aa5) cluster. Howev...
- 09:29 AM Bug #4392 (Resolved): kclient: libceph: osd ops send out of order
- I'm observing:...
- 03:22 AM Feature #4389 (Closed): get md5 of image
- Afaik there's no way to get a md5sum of an rbd image without exporting it first and then running md5sum. It'd be grea...
03/07/2013
- 09:36 AM Bug #4352: librbd: increased memory usage on bobtail vs argonaut (and with no caching)
- just merged the fix for this into next. still needs to be cherry-picked to bobtail branch.
03/06/2013
- 07:14 PM Bug #4364 (Resolved): ObjectCacher: inconsistency after flatten
- Using a vstart-created cluster with:...
03/05/2013
- 01:43 PM Bug #4324 (Resolved): kernel messenger: leaks in ceph_con_in_msg_alloc()
- This has been committed to the ceph-client testing branch:
d11db70 libceph: clean up skipped message logic
- 08:46 AM Bug #4324: kernel messenger: leaks in ceph_con_in_msg_alloc()
- I seem to remember having a hand in it as well. :)
- 07:35 AM Bug #4324 (Fix Under Review): kernel messenger: leaks in ceph_con_in_msg_alloc()
- The following patch has been posted for review.
(This patch is available as the top commit in branch
"review/wip-... - 06:29 AM Bug #4324 (In Progress): kernel messenger: leaks in ceph_con_in_msg_alloc()
- > FYI I mentioned this to Sage and he said the code has been
> refactored incorrectly a couple of times so it's defi... - 01:29 PM Subtask #4127 (Resolved): kernel osd client: clearly separate read and write data buffers
- These have been committed to the ceph client testing branch:
7f6346a libceph: separate read and write data
b019f5c ... - 11:51 AM Bug #4352 (Resolved): librbd: increased memory usage on bobtail vs argonaut (and with no caching)
- On irc, t0rn reports increased memory usage with caching turned off, and with bobtail compared to argonaut:...
- 06:22 AM Bug #4240 (Resolved): libceph: fix a osd request memory leak
- This was committed recently and I neglected to close this.
commit 6b3c5cdf66da552096b165a50dbd5313759a6ea6
Author... - 06:00 AM Bug #4284: rbd: bio getting set more than once in a message?
- I have posted this series of patches for review, and patch 4
resolves this issue. Patch 3 also tightens up the mds ... - 05:58 AM Subtask #4263: libceph: clearly abstract message data operations
- OK, I have posted a new series for review, and in it, patches
1, 2, and 5 resolve this issue (and the last one even ... - 05:56 AM Cleanup #4163 (Resolved): libceph: improve packing in struct ceph_osd_req_op
- This patch was committed recently and I forgot to
close this.
commit e7e319a9c51409c7effe34333ea26facf2fab9e1
Au...
03/04/2013
- 08:04 PM Bug #4285: rbd copy.sh test failing on counting images
- Ian agrees.
- 12:53 PM Bug #4285: rbd copy.sh test failing on counting images
- Okay – I can dig into this if necessary but I bet Ian will be happier if Joao does it and I work on other things. Ask...
- 12:36 PM Bug #4285: rbd copy.sh test failing on counting images
- As my previous updated noted, there are logs and a running cluster on plana13. My guesses about the causes are from c...
- 12:53 PM Bug #4324: kernel messenger: leaks in ceph_con_in_msg_alloc()
- FYI I mentioned this to Sage and he said the code has been refactored incorrectly a couple of times so it's definitel...
- 12:03 PM Bug #4324 (Resolved): kernel messenger: leaks in ceph_con_in_msg_alloc()
- I posted some code for review and Greg caught what looks
like a memory leak in error handling. I looked a bit into
... - 10:40 AM Subtask #4127 (Fix Under Review): kernel osd client: clearly separate read and write data buffers
- These patches have been posted for review. (I include
the first because it also is pretty closely connected
to the... - 08:54 AM Subtask #4127: kernel osd client: clearly separate read and write data buffers
- In addition to distinguishing incoming data from outgoing
data, the data for an osd request is either held in a bio
...
03/03/2013
- 07:58 PM Bug #4285: rbd copy.sh test failing on counting images
- Part of that AuthMonitor story sounds familiar to me — in particular I believe you'll find the max_global_id is only ...
03/02/2013
- 05:56 PM Bug #4285: rbd copy.sh test failing on counting images
- At least one cause of this is a bug in the AuthMonitor assigning the same global id more than once.
This results in ...
03/01/2013
- 05:41 PM Subtask #4127: kernel osd client: clearly separate read and write data buffers
- I have put together a series of patches that implement
this clear separation. It required work in both the
osd cli... - 05:36 PM Bug #4284: rbd: bio getting set more than once in a message?
- I think I'm going to check in the functions that
isolate/abstract message data operations without
assertions enable... - 05:30 PM Subtask #4263: libceph: clearly abstract message data operations
- This is partially done. These two have been committed:
ba1dd81 libceph: set page alignment in start_request()
46... - 11:48 AM Bug #4242: krbd: xfstest 259 failure (FS size near 4TB)
- ubuntu@teuthology:/a/teuthology-2013-02-28_20:00:04-regression-argonaut-master-basic/14358
- 11:36 AM Feature #3817: librbd: make cache write-through until a flush is encountered
- 11:35 AM Cleanup #4262: libceph: drop redundant page info assignments
- These have been committed.
5038c43 libceph: don't pass request to calc_layout()
3ed9489 libceph: format target ob... - 11:29 AM Cleanup #4262 (Resolved): libceph: drop redundant page info assignments
- 11:16 AM Bug #3664: osdc/ObjectCacher.cc: 517: FAILED assert(!i->size())
- Reproduced with client side logging after running in a loop for several days.
02/28/2013
- 03:43 PM Tasks #4210 (Fix Under Review): get ceph and all libraries versions+patches in EPEL
- Josef said earlier this week that he has submitted the epel branch. I'm not sure if that means the builds are now ha...
- 03:34 PM Tasks #4210: get ceph and all libraries versions+patches in EPEL
- Not libvirt, but librbd, librados, etc.
- 11:47 AM Bug #4285: rbd copy.sh test failing on counting images
- triggered again last night, ubuntu@teuthology:/a/teuthology-2013-02-28_01:00:03-regression-next-testing-basic$ less 1...
- 11:41 AM Bug #4285: rbd copy.sh test failing on counting images
- I haven't been able to reproduce this yet, but I suspect it's the previous remove that failed. Unfortunately there ar...
- 11:24 AM Subtask #4125: kernel messenger: support multiple sources of data
- > That field has no semantic meaning. All it does is
> control the alignment of the memory the OSD reads
> into, w... - 11:00 AM Subtask #4125: kernel messenger: support multiple sources of data
- That field has no semantic meaning. All it does is control the alignment of the memory the OSD reads into, which wil...
- 10:33 AM Subtask #4125: kernel messenger: support multiple sources of data
- Another thing about that field. In ceph_osdc_build_request(),
the 64-bit "off" value is subject to this:
... - 10:30 AM Subtask #4125: kernel messenger: support multiple sources of data
- I've been winding my way through code this morning and
I've discovered there's something in the ceph message
protoc... - 07:20 AM Bug #4284: rbd: bio getting set more than once in a message?
- My test to verify my explanation seems to have done so.
I had a message print when get_reply() was called. This
... - 06:59 AM Bug #4284: rbd: bio getting set more than once in a message?
- After further consideration I don't think this is going to
be good enough. My goal here is to allow multiple blocks... - 06:28 AM Bug #4284: rbd: bio getting set more than once in a message?
- Testing once more with key debug messages enabled that will I
think allow me to prove my explanation is correct.
02/27/2013
- 08:36 PM Bug #4284: rbd: bio getting set more than once in a message?
- I'm too tired to offer a guarantee, but I'm pretty sure
this is what's going on. A connection fault occurs while
r... - 05:18 PM Bug #4284: rbd: bio getting set more than once in a message?
- OK, back again after a meeting.
I was thinking about it, and I wonder whether this has to
do with a message that ... - 12:53 PM Bug #4284: rbd: bio getting set more than once in a message?
- We are in the path where an osd response message has
been received, and the messenger has called its
alloc_message ... - 12:28 PM Bug #4284: rbd: bio getting set more than once in a message?
- I have reproduced the problem--just. I haven't had a chance
to look at it but I think I will soon be able to identi... - 09:17 AM Bug #4284 (Resolved): rbd: bio getting set more than once in a message?
- I have to reproduce this again because my target machine
that hit this got rebooted. But a recent proposed change
... - 04:29 PM Feature #3417 (In Progress): krbd: read path (layering)
- 04:28 PM Subtask #2855 (Closed): krbd: copy-up on write to clone
- Tracking effort under 3419
- 04:27 PM Feature #3419 (New): krbd: copy-up on write to clone
- 04:25 PM Subtask #2854 (Closed): krbd: write path
- Tracking under 3418
- 04:25 PM Feature #3418 (In Progress): krbd: write path (layering)
- 04:23 PM Subtask #2852 (Closed): krbd: open parent on open
- Move this tracking to 3416
- 04:21 PM Feature #3416 (In Progress): krbd: open parent on open
- 04:21 PM Feature #2851: krbd: RBD layering support
- Josh created the new feature tickets so they'd show up in redmine backlog - subtasks don't.
- 11:02 AM Bug #4285 (Resolved): rbd copy.sh test failing on counting images
- ...
- 07:58 AM Bug #4242: krbd: xfstest 259 failure (FS size near 4TB)
- The issue for not completing multiple lingering requests
more than once is http://tracker.ceph.com/issues/3967.
(Ju...
02/26/2013
- 06:57 PM Bug #4255: osdc/ObjectCacher.cc: 517: FAILED assert(i->empty())
- recent log: ubuntu@teuthology:/a/teuthology-2013-02-25_01:00:05-regression-master-testing-gcov/11554
2013-02-25T03... - 02:56 PM Bug #4272 (Resolved): rbd snap rollback broken on next
- commit:f58601d681d3aee6541119ab830ba5e2a9ec3a13
- 12:23 PM Bug #4272 (Resolved): rbd snap rollback broken on next
- ...
- 11:12 AM Cleanup #4262 (Fix Under Review): libceph: drop redundant page info assignments
- 05:12 AM Bug #4242: krbd: xfstest 259 failure (FS size near 4TB)
- Of course it's not going to be easy... The problem showed up
this time at offset 2199533527040 instead of 219953839...
02/25/2013
- 07:12 PM Bug #4237 (Resolved): krbd: ignore zero-length requests
- This has been committed.
commit 4dda41d3d76747414586a4bad5615b550e0986b1
Author: Alex Elder <elder@inktank.com>
... - 07:11 PM Bug #4238 (Resolved): krbd: still more barrier fixes
- This has been committed.
commit 632b88cadece050ca925d74bda250c4a320c5cc7
Author: Alex Elder <elder@inktank.com>
... - 07:11 PM Bug #4235 (Resolved): krbd: add some dout() trace calls
- This has been committed.
commit 37206ee5bede14d59306fea3af4c0105d4712342
Author: Alex Elder <elder@inktank.com>
... - 07:10 PM Cleanup #4234 (Resolved): libceph: encapsulate flag setting
- This has been committed.
commit c9ffc77adebf9dfe3026ede6c8b3c61586b485b7
Author: Alex Elder <elder@inktank.com>
... - 07:09 PM Bug #4184 (Resolved): krbd: fix some sparse problems
- Josh reviewed these patches, and I have updated, re-tested
and committed them.
cc344fa rbd: eliminate sparse warn... - 07:01 PM Feature #3190 (Resolved): qa: test fencing
- commit:ccdafa067afed74b209537ae45ff8b49bb5fab3b
- 05:49 PM Bug #4249 (Resolved): librbd: recursive lock of snap_lock during snap_rollback
- commit:9096d70642880946b0b477e33f7debabbefec9fa
- 12:26 PM Bug #4249 (Fix Under Review): librbd: recursive lock of snap_lock during snap_rollback
- wip-4249
- 05:41 PM Bug #4252 (Resolved): rbd python bindings: error removing image on list_children
- commit:21471c1ef4cf02e068a6e6c49cd41e95efec952a
- 04:46 PM Bug #4252 (Fix Under Review): rbd python bindings: error removing image on list_children
- wip-4252, test broke with the watch-notify changes
- 04:02 PM Bug #4252 (In Progress): rbd python bindings: error removing image on list_children
- 05:08 PM Bug #4242: krbd: xfstest 259 failure (FS size near 4TB)
- The output did not indicate that rbd thought it
did something wrong.
That is, the object request came back and in... - 04:08 PM Bug #4242: krbd: xfstest 259 failure (FS size near 4TB)
- Requests to write 4096 bytes came back from the osd with a
written count of 2048 bytes.
Now I'm afraid I have to ... - 01:23 PM Bug #4242: krbd: xfstest 259 failure (FS size near 4TB)
- > BTW fixed the rbd namer thing; pull latest teuthology.git.
Well, "fixed" is generous. It had a bug, which I've ... - 12:09 PM Bug #4242: krbd: xfstest 259 failure (FS size near 4TB)
- BTW fixed the rbd namer thing; pull latest teuthology.git.
- 11:12 AM Bug #4242: krbd: xfstest 259 failure (FS size near 4TB)
- I was reporting if an error code got returned from
the osd. However the loop device function that
reports the "Wri... - 11:04 AM Bug #4242: krbd: xfstest 259 failure (FS size near 4TB)
- The tests I started running this morning finally hit number 259.
It did have the problem, but I got no messages from... - 09:09 AM Bug #4242 (In Progress): krbd: xfstest 259 failure (FS size near 4TB)
- I tried to reproduce this with a smaller subset of tests
(just 258 and then 259 repeatedly) over the weekend but
th... - 04:58 PM Tasks #4210 (Rejected): get ceph and all libraries versions+patches in EPEL
- EPEL does not take packages already in RHEL.
- 03:43 PM Subtask #4263 (Fix Under Review): libceph: clearly abstract message data operations
- These patches have been posted for review.
[PATCH 0/4] libceph: abstract setting message data info
[PATCH 1/4] li... - 03:33 PM Subtask #4263 (Resolved): libceph: clearly abstract message data operations
- Currently the fields in a message structure related
what constitutes the data portion of a message are
manipulated ... - 03:15 PM Cleanup #4262: libceph: drop redundant page info assignments
- These patches have been posted for review (marking
this "Feedback").
[PATCH 0/3] libceph: focus calc_layout() on ... - 03:03 PM Cleanup #4262: libceph: drop redundant page info assignments
- Three patches implement this change. The net result is
that calc_layout() becomes focused only on filling in the
o... - 02:53 PM Cleanup #4262 (Resolved): libceph: drop redundant page info assignments
- In calc_layout(), there are some assignments to osd request
fields r_num_pages and r_page_alignment. These assignme... - 03:14 PM Bug #4240 (Fix Under Review): libceph: fix a osd request memory leak
- This has been posted for review.
[PATCH] libceph: fix a osd request memory leak - 12:27 PM Bug #4255 (Duplicate): osdc/ObjectCacher.cc: 517: FAILED assert(i->empty())
02/23/2013
- 08:43 PM Bug #4255 (Duplicate): osdc/ObjectCacher.cc: 517: FAILED assert(i->empty())
- ...
- 12:08 PM Bug #4252 (Resolved): rbd python bindings: error removing image on list_children
- ...
- 08:31 AM Bug #4249 (Resolved): librbd: recursive lock of snap_lock during snap_rollback
- ...
02/22/2013
- 06:44 PM Bug #4242: krbd: xfstest 259 failure (FS size near 4TB)
- I"m seeing this fairly consistently now.
- 01:03 PM Bug #4242: krbd: xfstest 259 failure (FS size near 4TB)
- Whoops, it was 4 TB (2^42) not 4 GB.
I'll update what I originally wrote shortly. - 01:00 PM Bug #4242 (Closed): krbd: xfstest 259 failure (FS size near 4TB)
- I have now seen this at least twice. I think it's inconsistent
but it might be important. Not marking it high yet,... - 03:33 PM Bug #3958 (Resolved): rbd fsx fails with EBUSY
- 03:14 PM Bug #1070 (Rejected): krbd: ^C doesn't work
- Yeah, marking rejected. thanks for digging in!
- 03:08 PM Bug #1070: krbd: ^C doesn't work
- I can't relate them but this is involved in at
least some of the code involved:
http://tracker.ceph.com/issues/38... - 02:24 PM Fix #3665 (Resolved): librbd: deadlock during flatten
- commit:e4fd70fcec3a9abb7f19517326e46f58031c4196
- 02:13 PM Bug #913 (Resolved): krbd: handle race between notify and rbd device shutdown
- This is very old. And--provided I understand it--it is resolved
in the current rbd code.
When a mapped rbd image... - 01:50 PM Bug #4244 (Resolved): rbd xfstests: re-enable xfstests 049 in nightly run
- Pushed to ceph-qa-suite/master....
- 01:38 PM Bug #4244 (Resolved): rbd xfstests: re-enable xfstests 049 in nightly run
- Running test 049 repeatedly is passing for me now. I'm not sure
when or how it got fixed but I'm going to re-enable... - 01:40 PM Bug #2572 (Resolved): krbd: writeback errors?
- I've run xfstests 49 a bunch of times consecutively
and I am no longer seeing this issue.
I'm about to run it 100... - 01:23 PM Bug #2700: blkdeviotune method at libvirt doesn`t work on RBD volumes
- Just checked, problem with blkdeviotune still exists in the 1.0.2.
- 11:53 AM Bug #2700: blkdeviotune method at libvirt doesn`t work on RBD volumes
- This is something that we should fix. Is it still an issue in current libvirt?
- 01:20 PM Bug #2654 (Resolved): Stale rbd volume cannot be unmaped
- Somewhere along the way, this got fixed....
- 01:15 PM Bug #4243 (Resolved): rbd cli: usage confusing for snapshot operations
- The correct usage for creating a snapshot seems to be:...
- 12:05 PM Bug #3939 (Duplicate): krbd: circular locking report in sysfs code
- Duplicate of 3925. I *did* write it up before.
- 12:02 PM Bug #3950 (Can't reproduce): krbd: new assertion failure running concurrent rbd test
- Not seeing this in current code.
- 11:59 AM Bug #3871 (In Progress): krbd: initial header read may be out of date
- 11:58 AM Bug #3743 (Won't Fix): krbd: errors on submitted requests are ignored
- This was found in old request code and therefore will not be fixed.
- 11:38 AM Bug #3936 (Rejected): rbd: Strange dd speed behaviour (server side issue?)
- Network issue.
- 11:37 AM Bug #3664: osdc/ObjectCacher.cc: 517: FAILED assert(!i->size())
- Need to reproduce with:...
- 11:35 AM Bug #3697 (Duplicate): rbd copy.sh test failing in nightly
- #3958
- 11:34 AM Bug #3980 (Won't Fix): rbd image created with size zero on a mixed cluster crashes rbd
- Problem not in Bobtail and future releases. Won't fix.
- 11:32 AM Bug #3524 (Duplicate): test_librbd_fsx: crash after flatten
- Same as #3665
- 11:26 AM Bug #3703 (Can't reproduce): osd: crash while encrypting
- 11:12 AM Bug #4240 (Resolved): libceph: fix a osd request memory leak
- If an invalid layout is provided to ceph_osdc_new_request(), its
call to calc_layout() might return an error. At th... - 09:39 AM Bug #4184: krbd: fix some sparse problems
- Forgot to set target version.
- 09:38 AM Bug #4184 (Fix Under Review): krbd: fix some sparse problems
- Re-posted for review. The patches have been re-based on
top of fixes for 4234, 4235, 4237, and 4238. - 09:37 AM Cleanup #4234: libceph: encapsulate flag setting
- Posted for review. (Marked this as "Feedback" because
I can't mark a cleanup issue "need review")... - 06:22 AM Cleanup #4234 (In Progress): libceph: encapsulate flag setting
- Forgot to indicate this was underway.
- 06:02 AM Cleanup #4234 (Resolved): libceph: encapsulate flag setting
- While working on http://tracker.ceph.com/issues/4079 I created
some helper routines to wrap around the code that tes... - 09:36 AM Bug #4235 (Fix Under Review): krbd: add some dout() trace calls
- Posted for review....
- 06:23 AM Bug #4235 (Resolved): krbd: add some dout() trace calls
- While working on http://tracker.ceph.com/issues/4079 I added
some calls to dout() to some of the new request code. ... - 09:36 AM Bug #4238 (Fix Under Review): krbd: still more barrier fixes
- Posted for review....
- 07:21 AM Bug #4238 (Resolved): krbd: still more barrier fixes
- While looking at http://tracker.ceph.com/issues/4079 I
concluded that the fix I put in for this might still
not be ... - 09:35 AM Bug #4237 (Fix Under Review): krbd: ignore zero-length requests
- Posted for review....
- 06:40 AM Bug #4237 (Resolved): krbd: ignore zero-length requests
- The old request code simply ignored zero-length requests
as soon as they were seen. The new request code currently
... - 06:41 AM Feature #4236: krbd: properly handle flush commands
- Note that http://tracker.ceph.com/issues/4237 describes
making the new request code ignore all zero-length commands.... - 06:38 AM Feature #4236 (Duplicate): krbd: properly handle flush commands
- REQ_FLUSH and REQ_FUA are intended to implement drive
cache flushes and barrier semantics. I don't know
precisely ... - 02:45 AM Feature #4231 (Closed): librbd: Java bindings
- As it seems right now the new CloudStack code will require more complex RBD operations which can not be accomplished ...
02/21/2013
- 06:16 PM Fix #3665 (Fix Under Review): librbd: deadlock during flatten
- wip-rbd-flatten-deadlock
- 03:50 PM Bug #4186: rbd: add another "udevadm settle" in the CLI
- awesome, I fixed this without even knowing it existed! :)
- 02:50 PM Bug #3737: Higher ping-latency observed in qemu with rbd_cache=true during disk-write
- I've looked at the logs, and I think #4091 should fix this. The high ping times tend to occur around when the cache f...
02/20/2013
- 11:20 PM Bug #3737: Higher ping-latency observed in qemu with rbd_cache=true during disk-write
- Sigh. The attachment might help...
- 11:18 PM Bug #3737: Higher ping-latency observed in qemu with rbd_cache=true during disk-write
- Confirmed here, with ceph-0.56.3 and qemu-1.3.1.
See attached test output.
A summary is, the average ping time,... - 09:15 PM Bug #3958 (In Progress): rbd fsx fails with EBUSY
- reproducing with logs, vapre f.yaml.out, to confirm this is just the osd's EBUSY. which we should fix.
- 12:50 PM Cleanup #4053 (Resolved): ceph: cleanup ceph page vector functions
- Committed to the testing branch.
9e0eb85 ceph: remove a few bogus declarations
b324814 libceph: use void pointers... - 12:49 PM Subtask #4007 (Resolved): libceph: support STAT osd operation
commit fbfab53966b279f9cdb36b96ffa1e22f042c96ff
Author: Alex Elder <elder@inktank.com>
Date: Fri Feb 8 09:55:48...- 12:42 PM Fix #3665 (In Progress): librbd: deadlock during flatten
- 11:26 AM Feature #4211 (Rejected): get good qemu, libvirt versions+patches in CentOS+
- 11:26 AM Tasks #4210 (Resolved): get ceph and all libraries versions+patches in EPEL
02/19/2013
- 02:34 PM Feature #2770 (In Progress): krbd: define tasks to add osd_client compound class op support
- 02:23 PM Feature #2770 (Resolved): krbd: define tasks to add osd_client compound class op support
- 02:22 PM Feature #4021 (Resolved): rbd: openstack: add ability to copy volume to image for rbd
- 02:20 PM Feature #4195 (Duplicate): rbd: migrate an active image between pools
- 11:40 AM Subtask #4007: libceph: support STAT osd operation
- Josh has reviewed this. Testing again before committing
to be included in the testing branch. - 11:39 AM Cleanup #4053: ceph: cleanup ceph page vector functions
- Josh has reviewed these, and I've rebased them onto
the top of the testing branch. I'm going to test some
before u... - 10:36 AM Bug #4184 (In Progress): krbd: fix some sparse problems
- I have resolved all of the reports that remain in the
code currently in the testing branch. There are a
few report... - 08:57 AM Bug #4183 (Resolved): rbd: EBUSY on unmap right after io
- commit:45a4fe09151bcdeaaa2c3b5dbd089a91deaa055f
- 08:45 AM Bug #4186 (Resolved): rbd: add another "udevadm settle" in the CLI
- commit:8ca2274cc09c6a9925cbe86ccaf68db7898956e8
- 05:06 AM Bug #4186: rbd: add another "udevadm settle" in the CLI
- The change looks good to me.
Reviewed-by: Alex Elder <elder@inktank.com>
02/18/2013
- 08:37 PM Bug #4186 (Fix Under Review): rbd: add another "udevadm settle" in the CLI
- see wip-4186
- 06:04 PM Bug #4186 (Resolved): rbd: add another "udevadm settle" in the CLI
- In http://tracker.ceph.com/issues/4183 it was reported that
unmapping an image was failing because it was busy.
F... - 06:10 PM Bug #4183 (In Progress): rbd: EBUSY on unmap right after io
- Reassigning to Sage.
Sage, I see you've been working on ceph qa/rbd/map-snapshot-io.sh
and I don't want to interf... - 06:05 PM Bug #4183: rbd: EBUSY on unmap right after io
- ...
- 06:04 PM Bug #4183: rbd: EBUSY on unmap right after io
- OK, I think I understand now.
It turns out that "blkid" is running again on the
device after its final close. Th... - 05:51 PM Bug #4183: rbd: EBUSY on unmap right after io
- We now are very careful about allowing an image to be unmapped
if it's still open, whereas previously we would allow... - 11:24 AM Bug #4183: rbd: EBUSY on unmap right after io
- NOTE: I adjusted the test to add the sleep 1 until this is fixed.
- 11:23 AM Bug #4183 (Resolved): rbd: EBUSY on unmap right after io
- the map-snapshot-io.sh does...
- 05:13 PM Bug #4184 (Resolved): krbd: fix some sparse problems
- The Linux kernel has a tool called "sparse" that scans
the source tree and reports various code problems that
the c... - 05:07 PM Cleanup #4164 (Resolved): libceph: update rados.h and related code
- commit 0315a7770983bbe69211efed1aaee08324acd54c
Author: Alex Elder <elder@inktank.com>
Date: Fri Feb 15 11:42:30 ... - 05:05 PM Subtask #4126 (Resolved): kernel osd client: kill off some dead code
- 87f979d ceph: kill ceph_osdc_writepages() "nofail" parameter
fbf8685 ceph: kill ceph_osdc_writepages() "dosync" para... - 05:01 PM Cleanup #4165 (Resolved): libceph: update ceph_fs.h and related code
- commit dd6f5e105d85e02bc41db0891eb07152b1746ad9
Author: Alex Elder <elder@inktank.com>
Date: Fri Feb 15 11:42:30 ...
02/16/2013
- 09:11 AM Cleanup #4165: libceph: update ceph_fs.h and related code
- The following have been posted for review to resolve
this issue.
[PATCH 0/2] libceph: update ceph_fs.h and relate... - 08:11 AM Cleanup #4165: libceph: update ceph_fs.h and related code
- (Another related commit)
libceph: update ceph_mds_state_name() and ceph_mds_op_name()
Update ceph_mds_state_nam... - 08:10 AM Cleanup #4165 (Resolved): libceph: update ceph_fs.h and related code
- (This work is done. I'm just documenting it because it's really
distinct from http://tracker.ceph.com/issues/4126.)... - 09:10 AM Subtask #4126 (Fix Under Review): kernel osd client: kill off some dead code
- (Oops, I forgot to mark this in progress yesterday.)
The following have been posted for review to resolve this
is... - 08:58 AM Cleanup #4164: libceph: update rados.h and related code
- The following have been posted for review to resolve this:
[PATCH 0/3] libceph: update rados.h and related code (4... - 08:07 AM Cleanup #4164: libceph: update rados.h and related code
- (Another commit related to this.)
libceph: update ceph_osd_op_name()
Update ceph_osd_op_name() to include the n... - 08:06 AM Cleanup #4164: libceph: update rados.h and related code
- (Another commit related to this.)
libceph: add ceph_osd_state_name()
Add the definition of ceph_osd_state_name(... - 08:05 AM Cleanup #4164 (Resolved): libceph: update rados.h and related code
- (This work is done, I'm just documenting it because it's really
something distinct from http://tracker.ceph.com/issu... - 08:29 AM Cleanup #4163: libceph: improve packing in struct ceph_osd_req_op
- This has been posted for review.
[PATCH] libceph: improve packing in struct ceph_osd_req_op - 08:00 AM Cleanup #4163 (Resolved): libceph: improve packing in struct ceph_osd_req_op
- (This is done, just documenting it because it's not directly
related to http://tracker.ceph.com/issues/4126)
The ...
02/15/2013
- 01:59 PM Cleanup #4156 (Resolved): qa: don't chown rbd stuff in rbd/* workunits
- Once we are doing proper installs, we can call sudo as needed and not worry about $PATH etc.
02/14/2013
- 05:03 PM Subtask #3272: send/receive rbd snapshots
- From Sage - "I think this breaks down into a few different pieces:
1) Decide what output format to use. We want t... - 05:02 PM Subtask #3387: librbd: expose changed objects since a given snapshot
- 8 points
- 04:25 PM Feature #4021 (Fix Under Review): rbd: openstack: add ability to copy volume to image for rbd
- 04:25 PM Subtask #4019 (In Progress): rbd: openstack: simplify volume booting with new api: add boot optio...
- 04:12 PM Subtask #4019: rbd: openstack: simplify volume booting with new api: add boot option to volumes
- 3 points
- 04:25 PM Subtask #4020 (In Progress): rbd: openstack: simplify volume booting with new api: make image boo...
- 04:12 PM Subtask #4020: rbd: openstack: simplify volume booting with new api: make image boot button add s...
- 2 points
- 04:24 PM Subtask #4018 (In Progress): rbd: openstack: simplify volume booting with new api: modify boot pa...
- 04:24 PM Feature #4017 (In Progress): rbd: openstack: simplify volume booting with new api
- 04:02 PM Bug #3947 (Resolved): krbd: read zeroing freed bio?
- This got resolved in version 2 of the "new request code,"
which is now present in the testing branch. - 03:12 PM Feature #4136 (Duplicate): rbd xfstests: update list of tests run
- The latest version of xfstests includes tests numbered
up to 296. The last test we run is 278. Some of the
update... - 07:06 AM Subtask #4125: kernel messenger: support multiple sources of data
- Here's a short list of smaller tasks I can identify right
now in order to implement this. I may create a new issue
... - 05:57 AM Subtask #4125 (Resolved): kernel messenger: support multiple sources of data
- We need to allow the osd client to have an array of osd ops
such that each can supply its own buffer for sending and... - 06:25 AM Subtask #4127 (Resolved): kernel osd client: clearly separate read and write data buffers
- The osd client (and messenger) have generally speaking one
field representing *either* the data to be sent *or* the
... - 06:14 AM Subtask #4126 (Resolved): kernel osd client: kill off some dead code
- While looking through how to implement multiple ops I have
noticed that there's some dead code in the osd client, or... - 05:40 AM Subtask #2854: krbd: write path
- This work actually did begin last week, after getting STAT
calls implemented (see http://tracker.ceph.com/issues/400... - 05:35 AM Tasks #2853: krbd: read path
- This work is close to complete, in that I have shown
that a lot of functionality seems to work correctly.
I was n... - 05:24 AM Feature #2770: krbd: define tasks to add osd_client compound class op support
- After the walk through described above, I spent a little more
time thinking about the messenger aspect of allowing m...
02/13/2013
- 06:10 AM Feature #3761: kernel messenger: need to support multiple ops per request
- Here's what I'm prototyping.
I'm going to try to expand the definition of the trail portion of
a message so inste... - 05:12 AM Feature #3761 (In Progress): kernel messenger: need to support multiple ops per request
- In order to support multiple ops for osd requests we're
going to have to do some work on the messenger as well.
I'm...
02/12/2013
- 05:34 PM Bug #4100 (Resolved): rbd: unprotecting a snapshot in the "UNPROTECTING" state fails with EINVAL
- commit:fe283813b44a7c45def6768ea0788a3a0635957e and commit:bfb4482c4596759b464caf45f8f30368898519d8 in bobtail.
- 09:37 AM Bug #4100: rbd: unprotecting a snapshot in the "UNPROTECTING" state fails with EINVAL
- Dan - please review the wip branch.
- 03:52 PM Bug #3958: rbd fsx fails with EBUSY
- ubuntu@teuthology:/a/teuthology-2013-02-11_20:00:06-regression-bobtail-master-basic/4941
- 02:11 PM Feature #2770: krbd: define tasks to add osd_client compound class op support
- The way the osd client handles an object class method right now
assumes that outbound data (headed from the client t... - 10:18 AM Feature #2770: krbd: define tasks to add osd_client compound class op support
- ...
- 10:16 AM Feature #2770: krbd: define tasks to add osd_client compound class op support
- ...
- 11:42 AM Feature #4104: osd_client: support passing page array as data for CALL op
- I guess this needs to be considered an rbd task if it is
to show up as a subtask for 2770. - 11:41 AM Feature #4104 (Resolved): osd_client: support passing page array as data for CALL op
- The rbd object "copyup" operation is defined as a class method
operation. Currently when a class method needs to su... - 12:49 AM Bug #3936: rbd: Strange dd speed behaviour (server side issue?)
- Journal is 1G in RAM. Since there are no writers at all yet, I don't think that journal is filled and causes device s...
- 12:47 AM Bug #3936: rbd: Strange dd speed behaviour (server side issue?)
- Tried on regular 1G ethernet (other client server). Works well. Speed is stable and no such effect.
I don't under...
Also available in: Atom