Project

General

Profile

Activity

From 12/19/2012 to 01/17/2013

01/17/2013

10:14 PM Feature #1236: libceph: set layout via virtual xattrs (libceph/cfuse)
Sage Weil
10:02 PM Feature #3857: mds: enforce unique mds names in mdsmap
see wip-mds-names Sage Weil
09:36 PM Feature #3857 (Resolved): mds: enforce unique mds names in mdsmap
Currently mds's are uniquely identified by their addr (i.e., a unique instance of the process). The name is useful on... Sage Weil
12:28 PM Bug #3832 (Fix Under Review): client: does not observe O_SYNC
Implemented in wip-3832. Needs review. Sam Lang
12:17 PM Bug #3845: mds: standby_for_rank not getting cleared on takeover
I dont' think it matters. It's is a fixed lifecycle from standby -> active -> dead, so the leftover standby_ just te... Sage Weil
12:13 PM Bug #3845: mds: standby_for_rank not getting cleared on takeover
This is a monitor thing; the MDS is only involved in relaying the config setting over on boot-up. Greg Farnum
11:38 AM Bug #3845 (Closed): mds: standby_for_rank not getting cleared on takeover
This is the mdsmap after mds.a was active and given rank 0, then killed, and another mds (mds.b-s-r0) that had standb... Sam Lang
11:34 AM Feature #3730: Support replication factor in Hadoop
Sage Weil wrote:
> If there are more such cases, that is a separate bug!
It was a bug I had introduced in wip-cli...
Noah Watkins
09:51 AM Feature #3730: Support replication factor in Hadoop
Noah Watkins wrote:
> In Client, osdmap is protected by client_lock? If so, new version of branch isn't broken..
...
Sage Weil
08:55 AM Feature #3730: Support replication factor in Hadoop
In Client, osdmap is protected by client_lock? If so, new version of branch isn't broken.. Noah Watkins
10:24 AM Bug #1435: mds: loss of layout policies upon mds restart
wip-mds-layout2
needs to be rebased reviewed and tested!
Sage Weil
09:08 AM Bug #3261 (Rejected): mds crashes in EMetaBlob::replay
Understood. I'm sorry we weren't able to dig in when it happened. When do you get around to retesting we should be ... Sage Weil
02:09 AM Bug #3261: mds crashes in EMetaBlob::replay
should i test the same btrfs volume with a new ceph? if so i might get to it in the next month. please close with ins... Tobias Florek

01/16/2013

05:59 PM Bug #3832 (Resolved): client: does not observe O_SYNC
if the file was opened with O_SYNC we need to flush the io on every write call. Sage Weil
05:34 PM Feature #3730: Support replication factor in Hadoop
Oh right, libcephfs is not built on top of librados. Never mind, that's a whole different discussion we start occasio... Greg Farnum
05:15 PM Feature #3730: Support replication factor in Hadoop
I don't think libcephfs will give up an instance of the rados client, if that's what you mean by grant access to rado... Noah Watkins
04:33 PM Feature #3730: Support replication factor in Hadoop
Sorry to back this up a little, but I can't recall — does using libcephfs automatically grant a user access to the RA... Greg Farnum
04:30 PM Feature #3730: Support replication factor in Hadoop
This interface update is up for review in wip-client-pool-api Noah Watkins
09:52 AM Feature #3730: Support replication factor in Hadoop
From stand-up, stick with int64_t for userspace, and enforce 32-bit range. Noah Watkins
09:43 AM Feature #3730: Support replication factor in Hadoop
The move from int32 -> int64 was misguided, and incomplete. At this point it's not really worth the effort to move a... Sage Weil
07:31 AM Feature #3730: Support replication factor in Hadoop
It looks like in OSDMap there is some mixed usage of int64 and int for pool id, too. In Client::_create pool id is e... Noah Watkins
06:40 AM Feature #3730: Support replication factor in Hadoop
Can we change the type in libcephfs to uint64? We're the only ones calling ceph_get_file_pool() right now as far as ... Sam Lang
04:12 PM Bug #3828 (Rejected): seeing error: fault, server, going to standby whenever I run a ceph-syn loa...
This is showing up on your MDS, about 15 minutes after a client completes accesses, right? This is associated with th... Greg Farnum
04:01 PM Bug #3828 (Rejected): seeing error: fault, server, going to standby whenever I run a ceph-syn loa...
while validating bug 520, i saw an interesting error. it may be a red herring, as I am seeing no problem with the wr... Anonymous
03:47 PM Bug #520 (Closed): mds: change ifile state mix->sync on (many) lookups?
3 Node Cluster:
ceph version 0.56.1 (e4a541624df62ef353e754391cbbb707f54b16f7)
# cat /etc/ceph/ceph.conf
[global]...
Anonymous
02:51 PM Bug #520: mds: change ifile state mix->sync on (many) lookups?
csyn is now called ceph-syn
and --debug-ms 1 to see those messages go by!
Sage Weil
03:26 PM Bug #3261: mds crashes in EMetaBlob::replay
This looks like a problem with what's in the journal, but soo much MDS code has changed since then that I don't think... Sage Weil
03:24 PM Bug #1760 (Resolved): multiple_rsync workunit cannot remove non-empty directory intermittently
this also looks like the tmap problem, commit:e52ebacb73747ef642aabdb3cc3cb2a328687a4c and preceeding 4 commits. Sage Weil
03:23 PM Bug #2380 (Rejected): kclient: aufs over a cephfs mount fails with Stale NFS file handle
this is a generic problem with lookup by ino, see #3541 and other features Sage Weil
03:23 PM Bug #2092 (Can't reproduce): BUG at fs/ceph/caps.c:999
commit:561cf283173360c39db19dc735da4a319be68ff6 fixes the multi-mds case. we haven't seen this again for single-mds..... Sage Weil
03:11 PM Feature #3826 (Resolved): uclient: Be more aggressive about checking for pools we can't write to
Right now the client will happily buffer up writes to a pool that it can't actually write to. #2753 is going to make ... Greg Farnum
03:06 PM Bug #3746 (Rejected): kclient mmap doesn't zero past EOF
Run against bad code. Greg Farnum
03:03 PM Bug #2444 (Can't reproduce): null pointer deference in ceph_d_prune inside kvm
Sage Weil
03:00 PM Bug #2071 (Can't reproduce): kclient: pjd mkfifo failures
Sage Weil
02:59 PM Bug #1770 (Can't reproduce): directory nonexistent on kernel_untar_build.sh
Sage Weil
02:58 PM Bug #1749 (Can't reproduce): nonexistent directory in kclient_workunit_kernel_untar_build
Sage Weil
02:55 PM Bug #1318 (Resolved): directories disappear across multiple rsyncs
commit:e52ebacb73747ef642aabdb3cc3cb2a328687a4c and 4 preceeding patches fix up the TMAP bug that is the likely cause... Sage Weil
02:55 PM Bug #1511: fsstress failure with 3 active mds
Sam thinks this works now! Adding to QA suite. Greg Farnum
02:50 PM Bug #3625 (Resolved): client: EEXIST error on multiple clients to create
commit:b4d3bd06d4083d780755f6ef506df1643932fa2f Sage Weil
02:49 PM Bug #3625: client: EEXIST error on multiple clients to create
Maybe you already handled this? Greg Farnum
02:11 PM Bug #3625 (Fix Under Review): client: EEXIST error on multiple clients to create
Sam Lang
06:16 AM Bug #3625: client: EEXIST error on multiple clients to create
The kernel side has been reviewed and tested, but needs to be merged. The fuse side has been tested, but I think it ... Sam Lang
02:48 PM Bug #2753: Writes to mounted Ceph FS fail silently if client has no write capability on data pool
we should return an error code on fsync().. that is the quick fix.
a more polite feature will be opened to return ...
Sage Weil
09:19 AM Bug #2753: Writes to mounted Ceph FS fail silently if client has no write capability on data pool
This is clearly a bug, bureaucracy or not. It should not be a feature. We can do new development to fix a bug. If you... Ian Colle
02:46 PM Bug #3544: ./configure checks CFLAGS for jni.h if --with-hadoop is specified but also needs to ch...
I think this can be closed. There is a bunch of autoconf changes for Java that have or will be merged. Noah Watkins
02:41 PM Bug #3544: ./configure checks CFLAGS for jni.h if --with-hadoop is specified but also needs to ch...
I just did a ./configure and using CPPFLAGS to indicate where the jni headers were and that worked just fine. Using C... Anonymous
02:45 PM Bug #3254: mds: Replica inode's parent snaprealms are not open
Multi-mds, currently low priority. Greg Farnum
02:44 PM Bug #3637 (In Progress): client: not issuing caps for with clients doing shared writes
Sage Weil
02:43 PM Bug #3637 (Fix Under Review): client: not issuing caps for with clients doing shared writes
Sage Weil
02:40 PM Bug #3498 (Resolved): mds: mds assert failure during untar_kernel
this was a msgr bug, long since fixed. commit:36c0fd220ef02b1ffd7a3ae0d98e0fdec6b55a5b or thereabouts Sage Weil
02:39 PM Bug #1666: hadoop: time-related meta-data problems
http://www.mail-archive.com/ceph-devel@vger.kernel.org/msg10334.html
Also wip-mtime-incr in the ceph repo.
Sam Lang
02:38 PM Bug #2218: CephFS "mismatch between child accounted_rstats and my rstats!"
Greg Farnum
02:32 PM Feature #3821 (New): qa: run backuppc as part of qa suite
Sage Weil
02:32 PM Bug #2494 (Can't reproduce): mds: Cannot remove directory despite it being empty.
The dupe inode suggests this is the problem fixed by Yan's tmap fixes. Greg Farnum
02:29 PM Bug #2019 (Can't reproduce): mds: CInode::filelock stuck in sync->mix
Presumably we'll see this again, but it hasn't turned up in our testing lately and we need more info to debug it. Greg Farnum
02:27 PM Bug #1811 (Duplicate): 2 pjd chown tests failed on cfuse
Ian Colle
02:22 PM Bug #1537 (Resolved): cmds 100% when copying lots of files, mds_cache_size and mds_bal_frag
This is an optimization issue, which we'll get to! Sage Weil
02:21 PM Feature #3819 (Resolved): mds: re-add snaptests to qa suite
Sage Weil
02:02 PM Bug #3818 (Duplicate): kclient: fsx fails in mapread

With the fix in #3681, fsx fails in mapread with bad data. It looks like this is unrelated to the fix, and is a se...
Sam Lang
11:09 AM Feature #3543 (In Progress): mds: new encoding
Oh, this has been in progress all week. Greg Farnum
10:35 AM Bug #3773 (Can't reproduce): mds crashed at LogEvent::decode
I have been trying to reproduce this but have not hit it yet.
will reopen the bug, when needed.
Tamilarasi muthamizhan
06:04 AM Bug #3601: client: With multiple clients, file remove doesn't free up space
Yeah its that the lru doesn't have a timeout.
The mds could send an "enable timeout" message to clients once it se...
Sam Lang

01/15/2013

08:53 PM Feature #3728 (Resolved): mds: draft design for lookup by ino
Sage Weil
08:38 PM Feature #3730: Support replication factor in Hadoop
pool ids are currently exposed via libcephfs from ceph_file_layout, which uses a 32bit integer for pool id. However, ... Noah Watkins
08:34 PM Feature #3730: Support replication factor in Hadoop
Someone could toss a 'ceph osd pool set size' Hadoop's way, so a static mapping between pg pool size and pool name co... Noah Watkins
05:35 PM Bug #3254: mds: Replica inode's parent snaprealms are not open
No. So far I'm focus on stabilize basic fs function for multiple MDS setup, completely ignore snapshot. Zheng Yan
03:28 PM Bug #3254: mds: Replica inode's parent snaprealms are not open
Hmm, did this get fixed by some of Zheng's later patches? I remember things about snaprealms and migration... Greg Farnum
04:44 PM Feature #3289: ceph-fuse: somehow exert pressure on the VFS to remove dentries from the cache
#3575 should be kept in mind while doing this/instead of this — there's a forget_multi as well. Greg Farnum
04:44 PM Bug #3601 (New): client: With multiple clients, file remove doesn't free up space
Whoops, didn't mean to change that status. Greg Farnum
04:43 PM Bug #3601 (Duplicate): client: With multiple clients, file remove doesn't free up space
The LRU actually already exists; check out Client::lru. (Unless I'm misunderstanding something?) So we might want to ... Greg Farnum
04:37 PM Bug #925: mds: update replica snaprealm on rename
De-prioritizing multi-MDS issues... Greg Farnum
04:34 PM Bug #1117: mds: rename rollback broken on slaves during replay
De-prioritizing multi-mds issues for now. Greg Farnum
04:27 PM Bug #1435: mds: loss of layout policies upon mds restart
I'm guessing we want to move this up the queue; will discuss in bug scrub tomorrow! Greg Farnum
04:23 PM Bug #1511: fsstress failure with 3 active mds
De-prioritizing multi-mds failures at this time. Greg Farnum
04:23 PM Bug #1535: concurrent creating and removing directories crashes cmds
De-prioritizing multi-MDS bugs at this time. Greg Farnum
03:51 PM Bug #2753: Writes to mounted Ceph FS fail silently if client has no write capability on data pool
Fair enough, but if I can just make a suggestion, perhaps you might want to explain these procedures somewhere in the... Florian Haas
03:45 PM Bug #2753: Writes to mounted Ceph FS fail silently if client has no write capability on data pool
I agree it's a bug, but given the procedures we have now (ack! changing procedures coming alert!) I don't think we wa... Greg Farnum
03:43 PM Bug #2753: Writes to mounted Ceph FS fail silently if client has no write capability on data pool
No, please. A write pretending to succeed while actually not writing data _is_ a bug. The filesystem _not lying to it... Florian Haas
03:33 PM Bug #2753: Writes to mounted Ceph FS fail silently if client has no write capability on data pool
This is a great suggestion but falls into feature rather than bug-fix category. My initial thought is keeping a list ... Greg Farnum
03:42 PM Bug #1675 (Can't reproduce): mds: failed rstat assert
The logs are long gone. This will presumably pop up again; it's a pretty common failure mode, but there's nothing in ... Greg Farnum
03:38 PM Bug #1938: mds: snaptest-2 doesn't pass with 3 MDS system
De-prioritizing all multi-MDS bugs for now. Greg Farnum
03:27 PM Bug #3267: Multiple active MDSes stall when listing freshly created files
Currently de-prioritizing multi-MDS bugs. Greg Farnum
03:18 PM Bug #3625: client: EEXIST error on multiple clients to create
I know you guys did a couple rounds on this one, what's the status? Greg Farnum
01:25 PM Bug #3637: client: not issuing caps for with clients doing shared writes
Sage has a different proposed fix than what's in the branch. Still needs to be tested. Sam Lang
12:50 PM Bug #3637: client: not issuing caps for with clients doing shared writes
I don't remember where this ended up. Was the proposed fix problematic, or did it never get looked at? Greg Farnum
11:39 AM Bug #3718: multi-client dbench gets stuck over NFS exported cephfs
This apparently is only a problem under re-export, which I believe we are not focusing on right now. Greg Farnum
11:35 AM Bug #3553: MDS core dumped running 0.48.2argonaut
Given what we know so far (the Op got sent to the wrong OSD) this is a bug in the Objecter, not the MDS. Or possibly ... Greg Farnum

01/14/2013

07:49 PM Bug #3544: ./configure checks CFLAGS for jni.h if --with-hadoop is specified but also needs to ch...
Is this still an issue? Noah Watkins
03:04 PM Documentation #3796 (Resolved): FUSE mount documentation needs some corrections for v0,56x
The FUSE instructions need to be updated for v0.56 and later
currently:
> http://ceph.com/docs/master/cephfs/fuse...
Anonymous
01:28 PM Feature #3749 (Resolved): Remove forced synchronization from Java bindings
Noah Watkins
07:00 AM Bug #2187: pjd chown/00.t failed test 97
Happened again on Friday. Time to add the delay injection to the nightlies?
2013-01-11T07:32:37.489 INFO:teutholo...
Sam Lang

01/12/2013

08:01 AM Feature #3749: Remove forced synchronization from Java bindings
In libcephfs mount/unmount race against each other, and the test of the API (e.g. unmount racing against write). In C... Noah Watkins

01/11/2013

02:45 PM Bug #3793: wrong size reported in some distributions/toolchains
That makes this sounds like a simple fix... we need to swap the frsize and bsize fields. Except that right now we ar... Sage Weil
02:39 PM Bug #3793: wrong size reported in some distributions/toolchains
I spent a bit of time with gregaf trying to find authoritative sources for what the different values denote. While `... David McBride
01:40 PM Bug #3793: wrong size reported in some distributions/toolchains
This coreutils commit may have useful data:
http://git.savannah.gnu.org/cgit/coreutils.git/commit/src?id=0863f018f0f...
Greg Farnum
01:38 PM Bug #3793 (Resolved): wrong size reported in some distributions/toolchains
In ceph_statfs we set f_bsize to be 1MB in order to report very large available spaces. However, nowadays it is appar... Greg Farnum
02:38 PM Feature #3749: Remove forced synchronization from Java bindings
This needs more thought than just removing synchronization. We'd like to be segfault free in Java, even though you co... Noah Watkins
01:39 PM Bug #3794 (Resolved): uclient: reports sizes wrong in some cases
This is the counterpart to kernel bug #3793. See Client::statfs, in which we set f_bsize to 1MB but f_frsize to 4KB. ... Greg Farnum
10:52 AM Bug #3773: mds crashed at LogEvent::decode
Sure Sage. I was running bonnie from client during upgrade.
I had debug ms=1 set, i will try to reproduce this with...
Tamilarasi muthamizhan
09:41 AM Bug #3773 (Need More Info): mds crashed at LogEvent::decode
Tamil, I wonder if you can try to reproduce this with mds logging turned up from teh start (debug mds = 20, debug ms ... Sage Weil

01/10/2013

05:06 PM Bug #3773: mds crashed at LogEvent::decode
Okay, I gathered up a core file, a high-debug MDS log, and the log with the bad event (and the bad event itself) in t... Greg Farnum
02:05 PM Bug #3773: mds crashed at LogEvent::decode
I'll at least start this off. Greg Farnum
09:55 AM Feature #3621 (Closed): qa: add knfsd reexport tests to qa suite
Ian Colle
09:52 AM Feature #3621: qa: add knfsd reexport tests to qa suite
commit:aaa03bbcd2549a38f962a61fc63be16cca3a6d90 in teuthology.git Sage Weil

01/09/2013

02:58 PM Bug #3773 (Can't reproduce): mds crashed at LogEvent::decode
ceph version: 0.56.1 (e4a541624df62ef353e754391cbbb707f54b16f7)
I had a cluster [burnupi06, burnupi07, burnupi08] ...
Tamilarasi muthamizhan
12:05 PM Feature #3570 (In Progress): teuthology: mds thrasher
Sam Lang
10:58 AM Bug #3681: kclient fsx fails nightly
Proposed fix to set i_size before the setattr request:
This will resolve the above issue, because the cap flush on...
Sam Lang

01/08/2013

04:29 PM Bug #3597: ceph-fuse: denying root access
Is root actually a member of the fuse group? If not that would be correct behavior. Greg Farnum
12:04 PM Feature #626 (Closed): qa: add IOR, rompio, or other parallel workloads suite
Added tests to the _marginal_ qa suite that run IOR, mdtest, and fsx-mpi. Sam Lang
09:39 AM Feature #3543: mds: new encoding
I'm going to get started on this (mostly just figuring out current state, probably) today. Greg Farnum

01/07/2013

04:04 PM Feature #3749 (Resolved): Remove forced synchronization from Java bindings
Remove "synchronized" keyword from native interface. This was originally added when we were seeing some pthread mutex... Noah Watkins
03:26 PM Bug #3746 (Rejected): kclient mmap doesn't zero past EOF
Error coming from fsx:
INFO:teuthology.orchestra.run.out:Mapped Write: non-zero data past EOF (0xb826) page offset...
Sam Lang
12:19 PM Cleanup #3742 (Resolved): Remove old Hadoop wrappers and configuration options
I think it's likely that the current Hadoop shim is at least at feature parity with the old wrappers. Noah Watkins
10:02 AM Bug #3726 (Resolved): Enforce Ceph's minimum stripe size in the java bindings
Noah Watkins
10:02 AM Bug #3726 (Closed): Enforce Ceph's minimum stripe size in the java bindings
Noah Watkins
09:21 AM Bug #3738 (Resolved): kclient fsx truncate/write multi-client race

This bug is similar to #3681, but occurs only in the non-exclusive case (multiple clients), where a truncate doesn'...
Sam Lang
09:09 AM Bug #3681: kclient fsx fails nightly
The race here is between a truncate down, and completion of osd write ops triggering a cap flush. The exact order th... Sam Lang

01/04/2013

07:54 PM Bug #3666 (Resolved): Segfault running test_libcephfs
commit:3a9408742a8a6cbc870cba543a208285f1a6cec1 Sage Weil
03:25 PM Bug #3666: Segfault running test_libcephfs
I pushed a new wip-client-shutdown. This switches the clean-up order of client/messenger in libcephfs, rather than mo... Noah Watkins
01:36 PM Bug #3666: Segfault running test_libcephfs
Right, I think your fix will work, but it breaks the interface abstraction (messenger is created above the client, de... Sam Lang
01:16 PM Bug #3666: Segfault running test_libcephfs
This is what I'm running to reproduce the error. It's been running now for an hour on wip-client-shutdown without any... Noah Watkins
12:57 PM Bug #3666: Segfault running test_libcephfs
Rather than moving messenger shutdown into client shutdown? Noah Watkins
12:48 PM Bug #3666: Segfault running test_libcephfs
A similar issue was just handled in the ceph_fuse.cc code. There we just delay deleting the client till the end. Yo... Sam Lang
10:41 AM Bug #3666: Segfault running test_libcephfs
During unmount, the client is shutdown and free'd before the messenger. If any messages are delivered after the clien... Noah Watkins
03:29 PM Feature #3730 (Closed): Support replication factor in Hadoop
In order to support per-file replication values in Hadoop we need to specify that a new file should be generated in a... Noah Watkins
01:54 PM Bug #3726: Enforce Ceph's minimum stripe size in the java bindings
Also, name it something along the lines of get_stripe_granularity() and not .._min(imum)_ as that isn't entirely accu... Anonymous
01:40 PM Bug #3726: Enforce Ceph's minimum stripe size in the java bindings
After a discussion on jabber, the decision is to go with exposing a function call in libcephfs and then using that in... Anonymous
11:09 AM Bug #3726 (Resolved): Enforce Ceph's minimum stripe size in the java bindings
The Hadoop bindings are using the blocksize as the stripe size. If a block size is explicitly passed down, it ends up... Anonymous
01:00 PM Bug #3718: multi-client dbench gets stuck over NFS exported cephfs
Heads up, Zheng Yan's patches on the mds fix issues related to running multiclient dbench tests. Sam Lang
12:24 PM Feature #3626: mds: debug mode to generate traceless replies to clients
Hmm, okay. I wasn't real clear on the previous bugs so I'll need to look at it more if I end up taking this, but soun... Greg Farnum
11:46 AM Feature #3626: mds: debug mode to generate traceless replies to clients
Greg Farnum wrote:
> Hurray, it is. Nobody except the client looks at the trace_bl and setting that is the only thin...
Sage Weil
11:35 AM Feature #3626: mds: debug mode to generate traceless replies to clients
Hurray, it is. Nobody except the client looks at the trace_bl and setting that is the only thing set_trace() does. Ex... Greg Farnum
11:17 AM Feature #3626: mds: debug mode to generate traceless replies to clients
Greg Farnum wrote:
> Am I reading it correctly that this is just going to be doing the config and wrapper work to no...
Sage Weil
09:01 AM Feature #3626: mds: debug mode to generate traceless replies to clients
Am I reading it correctly that this is just going to be doing the config and wrapper work to not call set_trace() in ... Greg Farnum
12:20 PM Feature #3543: mds: new encoding
Sage Weil
12:20 PM Feature #3728: mds: draft design for lookup by ino
Sage Weil
12:14 PM Feature #3728 (Resolved): mds: draft design for lookup by ino
Sage Weil
12:20 PM Feature #3570: teuthology: mds thrasher
Sage Weil
12:06 PM Feature #3727 (Resolved): mds: refactor EMetablob encoding paths
Right now, the EMetaBlob sub-structures — for performance reasons — use an encoding pattern that doesn't match anythi... Sage Weil
11:42 AM Cleanup #89: mds: put inode dirty fields in dirty_bits_t to reduce memory footprint
Greg Farnum wrote:
> I briefly scanned the CInode and inode_t structs and it wasn't obvious to me what this should e...
Sage Weil
09:34 AM Cleanup #89: mds: put inode dirty fields in dirty_bits_t to reduce memory footprint
I briefly scanned the CInode and inode_t structs and it wasn't obvious to me what this should encompass. Are you talk... Greg Farnum
11:41 AM Subtask #547: mds: define fsck strategy, required metadata
This was a whiteboard discussion 2 years ago. Nothing was written down. We should reopen new and more detailed issu... Sage Weil
09:29 AM Subtask #547: mds: define fsck strategy, required metadata
Where are the results of this bug? It's marked resolved but I don't see any fsck references in the git tree, and ther... Greg Farnum
11:38 AM Cleanup #3677: libcephfs, mds: test creation/addition of data pools, create policy
Greg Farnum wrote:
> Do we have a separate bug for the library calls this needs?
#685, which would take the clien...
Sage Weil
09:27 AM Cleanup #3677: libcephfs, mds: test creation/addition of data pools, create policy
Do we have a separate bug for the library calls this needs? Greg Farnum
11:36 AM Feature #3244: qa: integrate Ganesha into teuthology testing to regularly exercise Ganesha CephFS...
Greg Farnum wrote:
> And for this one as well: setting up Ganesha in teuthology, run tests against it? Not using the...
Sage Weil
09:24 AM Feature #3244: qa: integrate Ganesha into teuthology testing to regularly exercise Ganesha CephFS...
And for this one as well: setting up Ganesha in teuthology, run tests against it? Not using the Ceph shim or anything... Greg Farnum
11:35 AM Feature #3243: qa: test samba reexport via libcephfs vfs plugin in teuthology
Greg Farnum wrote:
> Is this a matter of setting up (via teuthology) a Samba server which sits on top of a Ceph moun...
Sage Weil
09:24 AM Feature #3243: qa: test samba reexport via libcephfs vfs plugin in teuthology
Is this a matter of setting up (via teuthology) a Samba server which sits on top of a Ceph mount and then running tes... Greg Farnum
11:34 AM Feature #3426: ceph-fuse: build/run on os x
Greg Farnum wrote:
> Noah has done some work on this in the wip-osx branch; last I heard you could compile and get a...
Sage Weil
09:22 AM Feature #3426: ceph-fuse: build/run on os x
Noah has done some work on this in the wip-osx branch; last I heard you could compile and get a cluster going with vs... Greg Farnum
11:32 AM Feature #3542: mds: migration path for existing anchors, anchortables, etc.
Greg Farnum wrote:
> What all does this encompass? Design? Implementation? Does it need to be an online switch or ca...
Sage Weil
09:13 AM Feature #3542: mds: migration path for existing anchors, anchortables, etc.
What all does this encompass? Design? Implementation? Does it need to be an online switch or can it be an offline job? Greg Farnum
11:30 AM Feature #3541: mds: robust ino lookup using file backpointers
Greg Farnum wrote:
> Is this bug supposed to encompass the anchor table replacement work as well? I wouldn't expect ...
Sage Weil
09:12 AM Feature #3541: mds: robust ino lookup using file backpointers
Is this bug supposed to encompass the anchor table replacement work as well? I wouldn't expect so, but the presence o... Greg Farnum
11:23 AM Feature #3540: mds: maintain per-file backpointers on first file object
Greg Farnum wrote:
> Do we have any kind of design for this? We've talked about it some and it's conceptually simple...
Sage Weil
09:08 AM Feature #3540: mds: maintain per-file backpointers on first file object
Do we have any kind of design for this? We've talked about it some and it's conceptually simple, but splitting up the... Greg Farnum
11:15 AM Feature #626 (In Progress): qa: add IOR, rompio, or other parallel workloads suite
Yeah, that's what slang's working on to enable this. Assigning this to him. Sage Weil
08:57 AM Feature #626: qa: add IOR, rompio, or other parallel workloads suite
SamL has done some work on getting MPI going under teuthology, and on running some multi-client FS tests. I'm not sur... Greg Farnum
11:13 AM Feature #3621 (Resolved): qa: add knfsd reexport tests to qa suite
Sage Weil
09:43 AM Feature #3399: java: add accessor to Ceph version numbers
Oh, those are librados specific numbers, aren't they. So this bug is to create and expose a libceph version, then. Wh... Greg Farnum
09:35 AM Feature #3399: java: add accessor to Ceph version numbers
In libcephfs there is a call to get Ceph version (yes, just expose this). But, I recall Sage mentioning that it might... Noah Watkins
09:19 AM Feature #3399: java: add accessor to Ceph version numbers
This is just exposing the librados version() function to Java, right? Greg Farnum
09:41 AM Cleanup #660: mds: use helpers in mknod, mkdir, openc paths
What kind of helpers are you talking about with this? inode fetchers and lock grabbers? In a quick scan over handle_c... Greg Farnum
09:36 AM Feature #603: mds: repair directory hierarchy
This is part of #82 fsck, right? Do we have a more detailed algorithm anywhere? Greg Farnum

01/03/2013

01:59 PM Bug #3597: ceph-fuse: denying root access
I believe that we can reproduce this error. We are running Ubuntu 12.04 LTS Server on both the client and on the Cep... Graham Hemingway
12:56 PM Bug #3719 (Can't reproduce): pjd test 145 failed in the nightly runs
logs: ubuntu@teuthology:/a/teuthology-2013-01-02_19:00:03-regression-next-testing-basic/33621... Tamilarasi muthamizhan
12:48 PM Bug #3718 (Rejected): multi-client dbench gets stuck over NFS exported cephfs
When running qa/workunit dbench.sh the dbench 1 passes, but the dbench 10 gets hung up.
We should check this with ...
David Zafman
12:28 PM Feature #3621 (In Progress): qa: add knfsd reexport tests to qa suite
David Zafman
09:32 AM Bug #3681: kclient fsx fails nightly
Its most likely all the same bug, but fsx fails in different ways each time (always because of a truncate down). The... Sam Lang
09:27 AM Feature #3543: mds: new encoding
right. about 80% complete, see wip-mds-encoding. Sage Weil
09:22 AM Feature #3543: mds: new encoding
What is this task? Switching to use our versioned encoding scheme? Greg Farnum

01/02/2013

09:45 AM Bug #3700: mds: FAILED assert(!item_session_list.is_on_list())
fixed by revert of bad fix, see commit:6711a4c4038dbdf843f9dfe42c7809c5c37ae534 Sage Weil
09:37 AM Bug #3700 (Resolved): mds: FAILED assert(!item_session_list.is_on_list())
Sage Weil

12/30/2012

06:08 PM Fix #3630: mds: broken closed connection cleanup
... Sage Weil
06:06 PM Fix #3630: mds: broken closed connection cleanup
The con re-use looks like this:
- client connects
- mds ms_verify_authorizer creates a new session
- msgr see ex...
Sage Weil
06:04 PM Bug #3696 (Resolved): mds: FAILED assert(session_map.count(s->inst.name) == 0)
see #3630..let's fix this properly. Sage Weil

12/29/2012

02:39 PM Bug #3700 (Resolved): mds: FAILED assert(!item_session_list.is_on_list())
logs: ubuntu@teuthology:/a/teuthology-2012-12-29_03:00:03-regression-master-testing-gcov/30039... Tamilarasi muthamizhan
02:32 PM Bug #3696: mds: FAILED assert(session_map.count(s->inst.name) == 0)
ubuntu@teuthology:/a/teuthology-2012-12-29_03:00:03-regression-master-testing-gcov/30036 Tamilarasi muthamizhan
09:43 AM Bug #3696: mds: FAILED assert(session_map.count(s->inst.name) == 0)
reverted the broken fix, reproducing the original problem again. Sage Weil

12/28/2012

09:11 PM Bug #3696: mds: FAILED assert(session_map.count(s->inst.name) == 0)
Sage Weil
06:42 PM Bug #3696 (Resolved): mds: FAILED assert(session_map.count(s->inst.name) == 0)
This occurred shortly after startup when trying to reproduce another bug on the master branch:... Josh Durgin
06:21 PM Fix #3630: mds: broken closed connection cleanup
Sage Weil

12/26/2012

09:59 AM Bug #3681 (Resolved): kclient fsx fails nightly
... Sage Weil
08:39 AM Feature #3679 (Closed): Any API to get metadata?
Yep! See libcephfs. There is... Sage Weil
01:08 AM Feature #3679 (Closed): Any API to get metadata?
hello,there.
I am wondering if there is any API to get the metadata of a file .
I have the ceph file system run by ...
lollipop king
01:10 AM Tasks #3680 (Rejected): deduplication in ceph
I am wondering how to do deduplication in ceph...the big problem is how to get the metadata of a file
and how to mod...
lollipop king

12/24/2012

02:58 PM Feature #1448 (In Progress): test hadoop on sepia
Sage Weil
02:58 PM Cleanup #814 (Resolved): hadoop: refactor hadoop shim in terms of java libceph bindings
Sage Weil

12/23/2012

09:12 PM Cleanup #3677 (Closed): libcephfs, mds: test creation/addition of data pools, create policy
the create data pool argument is tested only with the default pools. once an lib is in place for the unit/functional... Sage Weil
09:06 PM Bug #3663 (Rejected): ceph kernel client is getting stuck on xstat* operations
No worries. Let us know if you do come across behavior that looks like a bug! Sage Weil
08:59 PM Bug #3663: ceph kernel client is getting stuck on xstat* operations
Hi Sage,
i am very sorry for taking your time with this issue, I feel like an idiot :(
The buggy client is runnin...
Roman Hlynovskiy

12/21/2012

02:39 PM Documentation #3672 (Resolved): doc: how to mount ceph-fuse from fstab
There's a new mount helper in bobtail for this. It contains these comments:... Josh Durgin
10:20 AM Bug #3666 (Resolved): Segfault running test_libcephfs
... Noah Watkins
08:36 AM Bug #3655 (Can't reproduce): client: hang in fsstress
I ran this test throughout the day yesterday and couldn't reproduce it, with message delays enabled. Marking as can'... Sam Lang
07:52 AM Bug #3663: ceph kernel client is getting stuck on xstat* operations
Hi Roman-
The logging levels are right, but in both mds logs neither mds was ever active; both were in the up:stan...
Sage Weil

12/20/2012

10:19 PM Bug #3663: ceph kernel client is getting stuck on xstat* operations
Hello Sage,
added 4 logs:
screen output from console of the laggy client. it ends up on 'jroger@pr02:~/data$ cp...
Roman Hlynovskiy
09:07 PM Bug #3663 (Need More Info): ceph kernel client is getting stuck on xstat* operations
Hmm. It's actually just saying its the oldest client; it's not actually too old (yet). The looping connect attempts... Sage Weil
08:48 PM Bug #3663 (Rejected): ceph kernel client is getting stuck on xstat* operations
there are 2 kernel clients happily working with ceph. as soon as I try mounting ceph from the third client, it's gett... Roman Hlynovskiy

12/19/2012

11:19 PM Bug #3655 (Can't reproduce): client: hang in fsstress
fsstress stuck in _read_sync()
#0 pthread_cond_wait@@GLIBC_2.3.2 ()
at ../nptl/sysdeps/unix/sysv/linux/x86_6...
Sam Lang
04:03 PM Bug #3637: client: not issuing caps for with clients doing shared writes
Proposed fix in wip-3637. The client's max size request in MClientCaps gets dropped if the file lock is in a non-sta... Sam Lang
12:30 PM Bug #3625: client: EEXIST error on multiple clients to create
Pushed fixes to wip-3625 (ceph and ceph-client repos) that implement #3 (mds sends back the created flag in reply to ... Sam Lang
12:29 PM Bug #3625: client: EEXIST error on multiple clients to create
David and I have posted comments on github about the fix to allow multiple
clients opening the same file to get a va...
Sam Lang
 

Also available in: Atom