Activity
From 02/09/2011 to 03/10/2011
03/10/2011
- 11:13 PM Bug #876 (Resolved): rbd export rbd image > 2GB, export error: Invalid argument
- Hi ,
reproduce steps:
1).ceph v0.25.
2) Imported a 10GB image into foo pool
$ rbd ls -p foo
10GBB
10GBrbd
3... - 11:11 PM Revision d1f9e5f3 (ceph): osd: fix keyring loading
- Load keyring if NOT mkfs.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 10:57 PM Revision 5c8146b5 (ceph): osdmaptool: don't require clobber to create new osdmap
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 09:17 PM Revision 95db4c5c (ceph): debian: remove so version from -dev packages
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 09:06 PM Revision 924adafe (ceph): filestore: assert on ENOTEMPTY
- ENOTEMPTY implies rmdir failed due to stray crap in the directory. We
should fail now, instead of later when we rest... - 07:34 PM Revision 00f2deec (ceph): rgw: can read old buckets format
- 07:20 PM Revision d3464af7 (ceph): filer: set RMW bit on probe
- Setting the RMW bits on the probe stat call will make the OSD wait for
pending writes on the object to flush to disk.... - 06:17 PM Revision 0abf2c9b (ceph): osd: fix peer no missing optimization
- This shortcut was broken: we need to populate peer_missing with missing
objects in terms of the master log, not the p... - 06:01 PM Revision f04b6adc (ceph): rgw: can turn off logging
- 05:54 PM Revision 99ef6f6c (ceph): osd: fix missing.rm()
- The version specifies which version of the object no longer should be
missing. We should thus remove it from the mis... - 05:44 PM Revision 62054fdb (ceph): osd: fix merge_log missing calculation at split point when logs diverge
- If primary log is something like
1'1
1'2
1'3
and remote log is
1'1
1'2
2'3
2'4
2'5
Then 1'2 is the split ... - 04:59 PM Feature #812: switch up crypto library (again)
- In branch tv-nss-crypto, commit 329fcc4dd5bb83ece520a9316c5fb1b35562e867. Will get merged after a brief code review, ...
- 04:33 PM Feature #875 (Resolved): osd: clean up old osdmaps
- This should be easier to do on the osd if the monitor keeps the full history (for now)
- 04:18 PM CephFS Bug #805 (Resolved): mds startup: _replay journaler got error -22, aborting
- The misread size got fixed by commit: d3464af7d046cfff887127710ad9656b4877c452.
The issue with lost data is an unl... - 11:25 AM CephFS Bug #805: mds startup: _replay journaler got error -22, aborting
- After discussion with Sage, it turns out that we don't need to worry about later blocks getting accepted but earlier ...
- 04:11 PM Revision d0abd58e (ceph): mkcephfs: modularize
- The goal is to support the old "ssh to everything" mode and also a
piecewise mode that lets the administrator do each... - 03:59 PM Feature #400 (Resolved): mkcephfs: work on multiple nodes without using ssh
- commit:cda125087a00f81a9363a1e7d4aa232e2e219517
- 03:50 PM Bug #836: atomic-ops error on Debian arm
- Managed to reproduce this issue locally. I'll try building without atomic-ops too, to see if that works. And will int...
- 03:49 PM Revision 8e3abf3c (ceph): Don't use -i to set id for non-daemons
- Due to conflicts, don't use -i to set id except for daemons.
Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com> - 03:29 PM Revision 33d6251a (ceph): common: replace -I with -i, rework --name
- Remove -I. All programs will take -i to set ID now.
Support setting the type and id simultaneously with --name or -n... - 03:11 PM Bug #874 (Duplicate): Fix configure checks for g++
- I'm not sure what the exact problem is, but when attempting to build Ceph on arm I had issues with the configure scri...
- 03:06 PM Revision 79605f5a (ceph): cauthtool: remove short form of --create-keyring
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 01:57 PM CephFS Bug #873 (Resolved): MDS journaler needs to truncate ahead of writes
- As discussed briefly in #805, there is a race condition due to MDS failover where an old MDS could have written log d...
- 01:34 PM Revision c56e1a92 (ceph): Merge branch 'globalfoo'
- 01:32 PM Revision a6d0a994 (ceph): librbd.cc: trim includes
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 01:20 PM Revision 9f32c500 (ceph): cram: test --conf=foo cconf syntax
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 01:20 PM Revision 7fe7a816 (ceph): common_init: Split common_init into multiple parts
- Split common_init into several different parts which can be called
individually from libraries. Eliminate startup_fla... - 01:16 PM Revision 88063e3b (ceph): config: trim deadcode
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 01:13 PM Bug #872 (Resolved): osd: crash due to missing pginfo
- 01:11 PM Bug #872: osd: crash due to missing pginfo
- Wido den Hollander wrote:
> Just thought about this, will this something a admin would run into? I ran into this due... - 12:22 PM Bug #872: osd: crash due to missing pginfo
- Just thought about this, will this something a admin would run into? I ran into this due to the recovery issue. But i...
- 11:56 AM Bug #872: osd: crash due to missing pginfo
- I do not think that crashing due to one faulty dir is what I'd do, but on the other hand, it will force a admin to ke...
- 11:43 AM Bug #872: osd: crash due to missing pginfo
- Ah, this is my fault. I made a copy of the files in 3.7c9 in a subdir called 't' (they were missing xattrs... :/) wh...
- 05:52 AM Bug #872 (Resolved): osd: crash due to missing pginfo
- I just upgraded "noisy" and saw osd1 go down after restart with:...
- 12:49 PM Bug #790 (Resolved): convert read, write, pread, pwrite to sage's safe versions
- 12:47 PM Subtask #842 (Resolved): initialization: have appropriate library-specific defaults
- library-specific defaults can be set in common_preinit or in the library itself after commit:7fe7a8165af16501c255e3a6...
- 12:44 PM Subtask #840 (Resolved): Clean separation between different components of initialization
- Done via commit:7fe7a8165af16501c255e3a6a11eedf55d5cad7b.
Libraries don't need to call common_init any more, with ... - 12:38 PM Subtask #841 (Resolved): don't call daemon_init in library code
- librados and librbd are now doing the right thing as of commit:7fe7a8165af16501c255e3a6a11eedf55d5cad7b
libceph is... - 01:06 AM Revision 2bbe83ea (ceph): rgw: escape certain strings in log dump
03/09/2011
- 09:29 PM Revision d0faba06 (ceph): rgw: fix acl checking when object does not exist
- In that case we also need to check the bucket acl so that we can
return the correct error code. - 09:28 PM Revision c6072195 (ceph): rgw: log also includes the REST operation
- 07:58 PM Revision d4d1a8c0 (ceph): rgw: get acl explicitly return success
- also don't crash if op didn't explicitly set the http return code
- 07:49 PM Revision 6a19790e (ceph): rgw: fix log when object not specified
- 07:43 PM Revision 8148d694 (ceph): ConfUtils::_parse: fix uninitialized var
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 04:06 PM Bug #868: librbd::AioCompletion::add_block_completion causes Qemu to crash
- hmm.. looking at the librbd::aio_write(), we loop through the blocks, creating a completion and triggering a write fo...
- 02:13 PM Bug #868: librbd::AioCompletion::add_block_completion causes Qemu to crash
- Finally managed to reproduce this using virtio and cache=writeback. I'm not sure if it'll occur in other configuratio...
- 03:08 PM CephFS Bug #805: mds startup: _replay journaler got error -22, aborting
- Okay, here's the problem. The previous object is ...
- 01:05 PM Revision 20c1ffb5 (ceph): config: ConfFile::open: return error code
- Return a integer result code specifying why we failed, if we fail.
Signed-off-by: Colin McCabe <colin.mccabe@dreamho... - 12:06 PM Revision 326cb5c2 (ceph): build:add common/entity_name.h to noinst_HEADERS
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 11:42 AM Revision 8f0fd62f (ceph): makefile:Suppress hash warning in include/encoding
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 10:31 AM Feature #828 (Resolved): make cpu profiler work
- oprofile, another sampling profiler, is working.
I wrote a wiki page about using oprofile with Ceph. It's at:
htt... - 12:49 AM Revision c9135519 (ceph): rgw: add logging
03/08/2011
- 11:08 PM Revision 1f120284 (ceph): Suppress gnu_cxx::hash warning a little bit better
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 10:59 PM Revision 058b3531 (ceph): config.cc: retab
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 09:25 PM Revision 0d7b9533 (ceph): rgw: cast size_t to unsigned long to fix a 32-bit warning.
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 09:24 PM Revision deca678f (ceph): librbd: cast offset values to uint64_t for unsigned comparison warning.
- It seems that size_t, off_t, and le64 have different signed/unsigned
properties on 32 and 64-bit Linux platforms, so ... - 08:30 PM Revision bb5aa7e8 (ceph): Merge branch 'stable'
- 08:30 PM Revision cf6b1de4 (ceph): uclient: Clear the CEPH_CAP_FILE_BUFFER ref on _flush, if safe.
- Previously we just returned if safe, but leaving the CEPH_CAP_FILE_BUFFER
ref around breaks _fsync horribly. The root... - 04:52 PM CephFS Bug #805: mds startup: _replay journaler got error -22, aborting
- So, the MDS that originally wrote that journal object got blacklisted shortly after writing it, and then a standby MD...
- 02:39 PM CephFS Bug #805 (In Progress): mds startup: _replay journaler got error -22, aborting
- We got this on our playground today. I have found one promising lead. Object 200.0001b88d_head is the last object in ...
- 04:43 PM rgw Feature #770: rgw: logging
- done with commit:c9135519b2eff07480b6ae81262162f0e5fac57b.
- 04:39 PM Revision dae26c55 (ceph): common_init: call to ProfilerFlush is unneeded
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 04:38 PM Revision e11bf6fe (ceph): Replace g_conf.name and g_conf.id with entityname
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 02:59 PM Revision d789d551 (ceph): makefile: Add -Wno-deprecated
- Until we can replace gnu_cxx::hash_map with std::tr1::unordered_set,
build with -Wno-deprecated.
Signed-off-by: Coli... - 02:36 PM Revision 30791abe (ceph): Merge branch 'globalfoo'
- 02:32 PM Revision d79d4e39 (ceph): Add missing includes
- Don't rely on header files being included in a certain order or
combination.
Signed-off-by: Colin McCabe <colin.mcca... - 02:32 PM Revision 37f1c172 (ceph): Create common/entity_name.h
- Create common/entity_name.h. Refactor EntityName a bit.
Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com> - 02:32 PM Revision b9da5f9a (ceph): common_init: no signal handlers for library code
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 12:21 PM Bug #859 (In Progress): Ceph does not pass fsstress
- Well, I got it to finish a run-through while testing my cfuse fix. Still need to reproduce and debug the rstat issue,...
- 12:12 PM CephFS Tasks #862 (Resolved): cap_refs[CEPH_CAP_FILE_BUFFER] isn't cleared if truncation zaps changes
- Pushed to stable in commit:cf6b1de4a692ca0f3e86a600bcf4642723ccade7, and merged stable into master.
- 08:56 AM CephFS Tasks #862: cap_refs[CEPH_CAP_FILE_BUFFER] isn't cleared if truncation zaps changes
- Ah, so it doesn't call delete on the oset. That was just me misreading the code.
However, I still believe it does ... - 12:26 AM Revision 56548b2e (ceph): mds: use projected subtree in rename anchor check
- We want to (try to) reanchor the directory on rename when our _projected_
subtree is not a leaf. If we use the norma...
03/07/2011
- 10:22 PM Revision eb31edd9 (ceph): Merge commit 'origin/master' into rgw
- Conflicts:
debian/control
src/rgw/rgw_rest.cc - 09:44 PM Revision 67b61875 (ceph): librados: add append to object
- 07:40 PM Revision 065b5189 (ceph): osd: document last_epoch_clean caveat
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 07:34 PM Revision 1cd878b1 (ceph): Bump debian/changelog timestamp so it's greater than previous one.
- This shuts up lintian, and just makes sense.
Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com> - 07:33 PM Revision 6ef3b676 (ceph): Make git ignore stamp files from debian build.
- Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com>
- 07:32 PM Revision 425570ae (ceph): osd: include all stray peers in might_have_unfound
- We should always consider any OSD that has a copy of the PG as a possible
location for missing objects. There are ca... - 07:06 PM Revision d97064e5 (ceph): Typos.
- Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com>
- 05:19 PM Revision dc8ff94e (ceph): Merge branch 'mds'
- 04:32 PM Tasks #860: mds rstat assert failure on fsstress
- Looked at this with Sage, and we're stumped. Hopefully this will become easier to reproduce once we deal with the oth...
- 04:28 PM CephFS Tasks #861 (Resolved): handle_client_rename thinks nonexistent dir is in subtree map
- After discussion, this was because the directory in question had a projected parent but not a parent, as it hadn't be...
- 03:23 PM CephFS Tasks #862: cap_refs[CEPH_CAP_FILE_BUFFER] isn't cleared if truncation zaps changes
- Okay, looks like the problem has to do with update_inode_file_bits calling objectcacher->truncate_set(). This:
1) ca... - 01:36 PM CephFS Tasks #862: cap_refs[CEPH_CAP_FILE_BUFFER] isn't cleared if truncation zaps changes
- Oh well, duh, those aren't the same inode. So for some reason the cap_refs[CEPH_CAP_FILE_BUFFER] count is off, or not...
- 11:04 AM CephFS Tasks #862 (In Progress): cap_refs[CEPH_CAP_FILE_BUFFER] isn't cleared if truncation zaps changes
- 03:23 PM Tasks #863 (Resolved): osd spinning during fsstress
- should be fixed in 67c8b91dd5a9ea6fd508b7544e397154ba330142
- 01:34 PM Revision 67c8b91d (ceph): buffer.h: push_back,push_front of an empty buffer pointer should be noop
- Also adds an assert to copy to ensure that bufferlist iterator copy
completes.
Signed-off-by: Samuel Just <samuel.ju... - 01:14 PM Revision b533727d (ceph): libceph: when calling common_init,set library flag
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 12:49 PM rbd Bug #867 (Resolved): Configure script is not updated to new librados API
- Thanks for noticing this. I just removed the CEPH_OSD_TMAP_SET test because it precedes librbd. Fixed by 3a4df898b200...
- 04:49 AM rbd Bug #867: Configure script is not updated to new librados API
- We probably want to add a include for "rbd/librbd.h" here too, just to make sure librbd is present on the system.
- 12:44 PM Subtask #839: Globals cleanup. Replace g_conf.name with g_conf.entity_name.to_str(). Remove g_con...
- fixed by 86e8b161535a08c2032a77a05b93cde7e59f59b3
- 12:44 PM Subtask #839 (Resolved): Globals cleanup. Replace g_conf.name with g_conf.entity_name.to_str(). R...
- 11:40 AM Bug #865: osd: mark pg clean only after purging strays
- documented in commit:065b518996db67491fab4fe65f075185e14ee88d. not sure if this is really worth fixing.. it means th...
- 11:08 AM Cleanup #871: replace gnu_cxx::hash_map with tr1::unordered_set
- the gcc message I referred to is here http://www.mail-archive.com/gcc-bugs@gcc.gnu.org/msg99225.html
- 11:07 AM Cleanup #871 (Resolved): replace gnu_cxx::hash_map with tr1::unordered_set
- gcc has started giving me this warning message about gnu_cxx::hash_map:...
- 11:08 AM Bug #779 (Can't reproduce): mds: loses port information
- 11:04 AM Revision 468ce23e (ceph): do_autogen: add -O2 when compiling for profiling
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 10:10 AM Bug #870 (Resolved): The -dev packages are missing conflicts, could avoid SONAME in package name.
- You can't install librados1-dev.deb and librados2-dev.deb at the same time anyway, they both contain /usr/include/rad...
- 09:20 AM Bug #848 (Resolved): limited commit_partial probably broke mds recovery
- 09:20 AM Bug #848: limited commit_partial probably broke mds recovery
- Okay, after discussing this with Sage he's happy -- if we lose the journal there's not a lot we can do about partial ...
- 09:11 AM Feature #524 (Resolved): object_locator_t
- MOSDOp message updated by commit:6fd6db7ffec625d58dfe0d6e18becb6593609576
03/06/2011
- 06:08 PM Bug #793 (Resolved): osd: avoid blocking in scrub_wq
- merged in commit:0fb5ef2ce92a0c9f5cc755744e4f9ff2b0951d11
- 05:27 AM Revision 94a4e862 (ceph): Merge branch 'stable'
- 05:26 AM Revision 1b99add0 (ceph): tests/cli/osdmaptool: default pg_num changed
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 04:50 AM Revision d4e4af1b (ceph): testlibrbd: signed return types
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 04:41 AM Revision 8fe0f48b (ceph): filestore: fix signedness so we can handle errors from safe_pread
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 04:40 AM Revision 69c7e59c (ceph): version: fix const warning
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 04:40 AM Revision aa251bd6 (ceph): safe_io: fix signed/unsigned comparisons
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 04:39 AM Revision 0cbfbee3 (ceph): tests/cli/osdmaptool: default pg_num changed
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 03:44 AM phprados Tasks #869 (Resolved): Update to new librados API
- Update phprados to the new librados API
03/05/2011
- 11:27 PM Revision 7a529d58 (ceph): crush: remove misleading comment
- This is a stale comment from an earlier iteration of the implementation.
Reported-by: Alexander Lyakas <alex.bolshoy... - 11:27 PM Revision 46d63d98 (ceph): crush: fix memory leak
- Reported-by: Alexander Lyakas <alex.bolshoy@gmail.com>
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 11:27 PM Revision 10bc2d99 (ceph): Merge branch 'next'
- Conflicts:
configure.ac
debian/changelog
debian/control
src/Makefile.am
src/cmds.cc
src/cosd.cc
src/include/ra... - 09:59 PM Revision 3fa121bf (ceph): v0.25
- 09:59 PM Revision 4ac08810 (ceph): Bump librados soname to 2.
- Update Debian packaging. For human-only mentions of librados,
use just "librados"
Signed-off-by: Tommi Virtanen <tom... - 09:30 PM Revision 82282f25 (ceph): config: back to 6 pg bits for now
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 09:44 AM Bug #868: librbd::AioCompletion::add_block_completion causes Qemu to crash
- I've attached gdb to backtrace what is going on:...
- 07:04 AM Bug #868 (Can't reproduce): librbd::AioCompletion::add_block_completion causes Qemu to crash
- I've updated my Qemu to the latest code and am experiencing crashes on my system....
- 04:13 AM rbd Bug #867 (Resolved): Configure script is not updated to new librados API
- The current configure script is not updated to the new librados API, attached is a patch to fix this.
03/04/2011
- 11:46 PM Revision 3d629383 (ceph): osd: debug the hell out of heartbeat thread
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 10:25 PM Revision 08af63da (ceph): rgw: put object request returns etag
- 10:19 PM Revision 46b01f4a (ceph): Merge branch 'osd_recovery' into next
- 10:19 PM Revision c07f3578 (ceph): test_missing_unfound: asdf
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 10:11 PM Revision 072a4bd2 (ceph): rgw: rest put returns with etag on all APIs
- 10:11 PM Revision 43f6a39c (ceph): rgw: remove warnings
- 10:00 PM Revision 2494d593 (ceph): osd: requeue pg for recovery if we may have found someting
- If we get a peer log/missing and call search_for_missing, requeue the pg
for recovery so we can pull anything we may ... - 09:59 PM Revision 53428c0e (ceph): osd: include all up peers in might_have_unfound when desperate
- If our might_have_unfound calculation was off (it currently can be, see
#865) we could prematurely give up. Try any ... - 09:57 PM Revision fbc73704 (ceph): rgw: openstack HEAD on container responds with metadata
- 05:39 PM Revision 30c5091c (ceph): osd: recover_primary if recover_replicas starts no ops
- recover_replicas may fail to start anything if we see an unexpected error.
In that case, try recover_primary immediat... - 05:38 PM Revision 836f72a1 (ceph): osd: discover more missing if unfound and do_recovery can't start anything
- If we couldn't start any recovery ops and things are still
unfound, see if we can discover more missing object locati... - 05:38 PM Tasks #863: osd spinning during fsstress
- This seems to be caused by the following chain:
1) bufferlist iterator copy from a bufferlist containing one empty... - 12:55 PM Tasks #863 (Resolved): osd spinning during fsstress
- During some fsstress runs, I've managed to end up with an OSD spinning at 100% cpu, making no apparent progress, whic...
- 05:30 PM Revision 88cc244e (ceph): do_autogen.sh: add -P option
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 05:21 PM Bug #837: AuthAuthorizeHandler fails to build on s390
- Current understanding:
it's the key hashing function for an in-RAM hash table
Plan:
rewrite hash table to us... - 03:21 PM Feature #697 (Resolved): create rbd library
- 02:23 PM Bug #847 (Resolved): OSD: Recovery does not complete, PG's stay active
- phew! all pgs are now active+clean. hopefully scrub won't turn anything up :)
- 04:47 AM Bug #847: OSD: Recovery does not complete, PG's stay active
- I've searched the logs again for rb.0.1.00000000a638/head and I found:...
- 02:17 AM Bug #847: OSD: Recovery does not complete, PG's stay active
- I think it was syslog. The logger machine is not on the same location, could be that there is a bandwith issue which ...
- 02:17 PM rgw Bug #866 (Resolved): RGW PUT object does not responds with etag
- Fixed in commit:08af63da1e78c47444720f6d443421ff2b5badb1.
- 02:15 PM rgw Bug #866: RGW PUT object does not responds with etag
- This causes the s3cmd utility to spit errors when trying to upload an object.
- 02:03 PM rgw Bug #866 (Resolved): RGW PUT object does not responds with etag
- 01:40 PM Bug #865 (Won't Fix): osd: mark pg clean only after purging strays
- Right now we set last_epoch_clean prior to cleaning replicas (and actually becoming "clean"). This means we might ha...
- 01:09 PM CephFS Tasks #861: handle_client_rename thinks nonexistent dir is in subtree map
- Server.cc:4722 passes ...
- 12:51 PM CephFS Tasks #861 (Resolved): handle_client_rename thinks nonexistent dir is in subtree map
- ...
- 01:00 PM Bug #864 (Resolved): osd: pull doesn't fail when pulled-from node restarts
- Recovery should continue.
- 12:53 PM CephFS Tasks #862 (Resolved): cap_refs[CEPH_CAP_FILE_BUFFER] isn't cleared if truncation zaps changes
- ...
- 12:46 PM Tasks #860 (Resolved): mds rstat assert failure on fsstress
- This is during an unlink....
- 12:44 PM Bug #859 (Resolved): Ceph does not pass fsstress
- So far we haven't gotten Ceph to successfully run the Linux Test Project's fsstress test. A typical run for us looks ...
- 12:43 PM CephFS Bug #858 (Resolved): cfuse craps out with fsstress
- I have yet to see any cfuse issues with this. I will continue running and reopen if it does, but in the meantime I wi...
- 09:31 AM Linux kernel client Bug #852: secret key shows up in /etc/mtab after mount -o secretfile=/.../key
- > Tv, can you see if the kernel key management stuff is appropriate here?
That is what I wanted to do. They alre... - 09:06 AM Linux kernel client Bug #853: bonnieloop.sh
- First time, I think.
- 06:32 AM Revision 964f1e19 (ceph): Fix test/signals.cc
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 05:34 AM Revision ab74d498 (ceph): librados: cosmetic header changes
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 12:10 AM Revision e779a3c0 (ceph): librados, librbd: use separate IoCtxs for data and metadata
- Adds deep copy method IoCtx::dup, so that the data and metadata
contexts can have different snap_seqs and snap contex... - 12:10 AM Revision 37edd473 (ceph): librbd: fix error message and unnamed constant
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 12:08 AM Revision 49343296 (ceph): librbd: change RBD::open to take a reference to an Image instead of a p...
- This makes the API more consistent with the librados API.
Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com> - 12:08 AM Revision fdd50a15 (ceph): librados: remove unused member of IoCtx
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 12:08 AM Revision 1de0b274 (ceph): librados: IoCtx destructor should put reference only if initalized
- 12:06 AM Revision 2e3b8448 (ceph): librados: can set up object locator
- 12:05 AM Revision f45a790f (ceph): librados:rados_ioctx_stat -> rados_ioctx_pool_stat
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 12:05 AM Revision 3ebaa4c7 (ceph): object_locator: fix clear()
- 12:04 AM Revision b2ceb75c (ceph): librados: use atomic_t for reference count
- Use an use atomic_t for the reference count in IoCtxImpl.
Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com> - 12:04 AM Revision 6f797af1 (ceph): librados: make IoCtxImpl a pointer in WatchContext
- Adds get and put to IoCtxImpl for refcounting,
and uses them in WatchContext, which shouldn't
be creating a copy of t... - 12:04 AM Revision 773f0034 (ceph): librados: decrement refcount of old io_ctx_impl in assignment operator
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 12:04 AM Revision 0e32cd2f (ceph): librados: fix IoCtx::from_rados_ioctx_t
- IoCtx::from_rados_ioctx_t creates an IoCtx out of a rados_ioctx_t.
However, this IoCtx must share ownership of the Io... - 12:03 AM Revision ecab94ca (ceph): Rename radios_ioctx_{open,close} to create/destroy
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 12:03 AM Revision ae77624b (ceph): librados: remove IoCtx::close()
- We decided we don't want IoCtx::close(), since IoCtx::~IoCtx() exists.
Signed-off-by: Colin McCabe <colin.mccabe@dre... - 12:03 AM Revision 7d06b1b2 (ceph): Make git ignore core files.
- 12:03 AM Revision 77880416 (ceph): Make git ignore python generated files.
- 12:02 AM Revision 7d6a4fc6 (ceph): librados: Crashed on shutdown if connect was never called.
- Add a trivial unit test to trigger this.
- 12:00 AM Revision b7340431 (ceph): libradoshpp: put ceph stuff in namespace librados
- Try a little bit harder to avoid polluting the user's global namespace
with our stuff.
Signed-off-by: Colin McCabe <... - 12:00 AM Revision b97388f9 (ceph): librados: don't create unused SnapContext objs
- There were some unused temporary variables hanging around.
Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com> - 12:00 AM Revision 062dd5eb (ceph): librados: fix copy ctor of ObjectIterator
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
03/03/2011
- 11:59 PM Revision 46d6214b (ceph): testrados: add object stat test
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 11:59 PM Revision 60482f5a (ceph): testlibrbd: recreate test pool each time
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 11:48 PM Revision 2a4b0d1d (ceph): radosgw_admin: change usage, improve uid lookup
- commands can get also either the email or the openstack uid
- 11:24 PM Revision 75f7ea30 (ceph): librbd: fix error message and unnamed constant
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 11:24 PM Revision a6ef9077 (ceph): librados, librbd: use separate IoCtxs for data and metadata
- Adds deep copy method IoCtx::dup, so that the data and metadata
contexts can have different snap_seqs and snap contex... - 10:54 PM Revision 1ed2d8c5 (ceph): Fix test/signals.cc
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 10:15 PM Revision 9bd627d1 (ceph): rgw: can show bucket size and count
- 09:50 PM CephFS Bug #858: cfuse craps out with fsstress
- 1) Spinning the cpu. I have logs and a core file but haven't looked into them deeply. (one of my spare disks I use fo...
- 09:26 PM CephFS Bug #858: cfuse craps out with fsstress
- Greg Farnum wrote:
> What kind of issues are you expecting to crop up here? I've so far run across:
> 1) an issue w... - 05:28 PM CephFS Bug #858: cfuse craps out with fsstress
- What kind of issues are you expecting to crop up here? I've so far run across:
1) an issue with cosd spinning that I... - 09:46 PM Revision a41865e3 (ceph): messanger: shouldn't throw sigpipe on failed socket
- 09:33 PM Linux kernel client Bug #853: bonnieloop.sh
- Did it do this the first time, or after a few runs?
- 09:32 PM CephFS Bug #854: unsynchronized clocks between kernel-client/cmds cause PJD fstest failures
- The only reasonably sane idea I have here is for the client/mds to compare clocks to estimate skew and have some sort...
- 09:29 PM Linux kernel client Bug #852: secret key shows up in /etc/mtab after mount -o secretfile=/.../key
- Tv, can you see if the kernel key management stuff is appropriate here?
The client key is static.. only needs to... - 08:59 PM Linux kernel client Bug #855 (Resolved): fsstress kernel opps
- Samuel Just wrote:
> Ah, just noticed that it's fixed in commit:fe7c17ec5ec0851550e32be986788e367b1f8e00.
Yeah th... - 05:50 PM Linux kernel client Bug #855: fsstress kernel opps
- Ah, just noticed that it's fixed in fe7c17ec5ec0851550e32be986788e367b1f8e00.
- 05:44 PM Linux kernel client Bug #855: fsstress kernel opps
- Annoyingly, I am suddenly having some trouble triggering the bug, but the previous output suggests the get_user_pages...
- 04:06 PM Linux kernel client Bug #855: fsstress kernel opps
- New debug output, pages is invalid out of ceph_get_direct_page_vector on O_DIRECT.
[ 1231.684328] ceph: ... - 07:56 PM Revision 8c3133e5 (ceph): configure: fix profiler check
- 07:44 PM Revision bb42a71b (ceph): configure: fix profiler check
- 07:26 PM Revision 6a686c8f (ceph): Merge branch 'master' into rgw
- Conflicts:
src/Makefile.am - 04:43 PM Revision b37facc3 (ceph): msgr: fix typo
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 03:50 PM Revision b941cfdc (ceph): common: block SIGPIPE everywhere we can
- It's much better to get EPIPE than SIGPIPE.
Block SIGPIPE in all threads we create. In the daemon, block SIGPIPE in
... - 03:50 PM Revision 3b34e2f3 (ceph): messanger: shouldn't throw sigpipe on failed socket
- 03:25 PM Revision d1fce13f (ceph): common: block SIGPIPE everywhere we can
- It's much better to get EPIPE than SIGPIPE.
Block SIGPIPE in all threads we create. In the daemon, block SIGPIPE in
... - 03:23 PM Feature #828: make cpu profiler work
- This still isn't working for me on current master (with and without the last SIGPIPE patch)....
- 01:58 PM Bug #847: OSD: Recovery does not complete, PG's stay active
- Starting looking through your logs to debug this but it really looks like they are incomplete.. either syslog is drop...
- 11:35 AM Bug #847: OSD: Recovery does not complete, PG's stay active
- Just checked another PG (just to make sure), 3.22b in this case:...
- 10:51 AM Bug #847: OSD: Recovery does not complete, PG's stay active
- I searched for the object:...
- 08:35 AM Bug #847: OSD: Recovery does not complete, PG's stay active
- Can you look manually on the OSDs for the objects that it's showing as 'still unfound'? rb.0.1.00000000f6c0/head bec...
- 03:14 AM Bug #847: OSD: Recovery does not complete, PG's stay active
- I tried the new branch, but it isn't working as expected yet.
When following the progress I saw:... - 12:30 PM Revision ed01fa1f (ceph): dout: Log version message when (re)opening log
- Log a version message whenever we open the dout log, not just the first
time. However, only output it to log files an... - 12:24 PM Revision 82c5f3a8 (ceph): Thread: don't mask signals except in library code
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 12:15 PM Revision 5c668d2b (ceph): dout: Log version message when (re)opening log
- Log a version message whenever we open the dout log, not just the first
time. However, only output it to log files an... - 11:25 AM Feature #857 (Resolved): log open marker
- implemented by commit:5c668d2b1737ffb823b68b51bae94c89b70122ee
- 10:56 AM Linux kernel client Bug #856: busy loop osd reconnects if osd is down
- hopefully fixed by commits in for-linus, see commit:c3d5a0cd56fa1b57db04aa151a994424657cf264 among others
- 09:34 AM Bug #848: limited commit_partial probably broke mds recovery
- Hmmm. I was under the impression that inconsistencies like that would be taken care of during journal replay (ie, the...
- 08:53 AM Bug #848: limited commit_partial probably broke mds recovery
- I'm afraid this still isn't right, because the header in the CDir object includes recursive stats that should be accu...
- 02:52 AM Revision 1ad48f12 (ceph): CDir: Don't write out the header on a partial commit.
- If we write out the header as part of a partial commit, and then
fail to complete a subsequent commit (network error,... - 02:49 AM Revision a7126649 (ceph): CDir: pay attention to the max_dir_commit_size!
- Somehow it seems to have been ignoring this previously, which
doesn't make any sense at all since otherwise our tests... - 12:37 AM Revision abf46a9f (ceph): Merge branch 'tcmalloc_optimizations'
- 12:18 AM Revision f12e1a98 (ceph): tcmalloc: warn if you try and dump without the profiler running.
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 12:13 AM Revision d467fbfd (ceph): mds: rip out rename linkmerge support
- It turns out POSIX says rename(a,b) is a no-op when a and b link to the
same inode. This is super weird but good new... - 12:08 AM Revision f36e4582 (ceph): tcmalloc: create perfglue handle_command functionality.
- Switch the OSD and MDS to use it, instead of handling
the calls on their own. Note that this is a command interface c... - 12:04 AM Revision ae72c608 (ceph): Merge remote branch 'origin/master' into locator
- 12:04 AM Revision aac1e958 (ceph): locator: pool id is signed
03/02/2011
- 11:55 PM Revision 3fccc0a9 (ceph): qa/workunits: added fsstress.sh
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 11:29 PM Revision 39d70de2 (ceph): osd: drop old ceph_osd_request_head
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 11:29 PM Revision 6fd6db7f (ceph): osd: new encoding for MOSDOp when OBJECTLOCATOR feature is present
- The new encoding includes the object_locator_t. It also avoids the old
C struct.
Signed-off-by: Sage Weil <sage.wei... - 10:56 PM Revision 71e46d8f (ceph): osd: add OBJECTLOCATOR protocol feature bit
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 10:24 PM Revision 3e09e1c1 (ceph): encode locator information in osd op
- 10:23 PM Linux kernel client Bug #855: fsstress kernel opps
- Sam, are you able to reproduce this? I have a possible fix, but want to make sure it's the right thing first. If we...
- 04:22 PM Linux kernel client Bug #855: fsstress kernel opps
- somewhat more readable, from teh console:...
- 03:17 PM Linux kernel client Bug #855: fsstress kernel opps
- Forgot to mention: This is on ceph1 with sepia mounted. Also, the kernel output occurred just just after I hit ^C t...
- 03:16 PM Linux kernel client Bug #855 (Resolved): fsstress kernel opps
- /usr/lib/ltp/testcases/bin/fsstress -d . -l 1 -n 1000 -p 10
Message from syslogd@ceph1 at Wed Mar 2 15:13:50 2011... - 10:13 PM Revision 07bf9b82 (ceph): tcmalloc: switch the interface.
- Previously, we used function pointers. Fun for me to learn about, icky
to actually have!
Now we use our own wrapper f... - 10:01 PM CephFS Bug #858 (Resolved): cfuse craps out with fsstress
- ...
- 09:58 PM Revision 5edf58a3 (ceph): testradospp: don't try to read /etc/ceph/ceph.conf explicitly
- 09:39 PM Revision 75e2a077 (ceph): cmds/cosd: Fix IsHeapProfilerRunning implicit return type cast.
- G++ complains about the difference between the return type of tcmalloc's
IsHeapProfilerRunning (int) and the return t... - 09:29 PM Feature #857 (Resolved): log open marker
- We used to have a marker every time the log is (re)opened that included process name, pid, log file, and ceph version...
- 09:10 PM Revision 1a2e2a77 (ceph): msgr: fix chdir after daemonize
- We don't care of the mkdir succeeds. It has dubious value anyway, though;
if you specify a unique directory for the d... - 06:38 PM Bug #848: limited commit_partial probably broke mds recovery
- I haven't tested this yet, but I pushed a short fix to the mds branch which takes advantage of message ordering and I...
- 06:20 PM Revision f76cf9b7 (ceph): librbd: change RBD::open to take a reference to an Image instead of a p...
- This makes the API more consistent with the librados API.
Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com> - 05:50 PM Revision a6167332 (ceph): mds: drop some dead code
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 05:41 PM Revision b6bfa8c5 (ceph): mds: fix one rename dentry linkage projection case
- There are more. :(
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 03:36 PM Revision 64186f99 (ceph): dout: Reopen dout after parsing all config opts
- Reopen the dout stream only after we parse all configuration options.
Specifying --log-file on the command line now w... - 03:33 PM Revision bf1ae374 (ceph): dout: remove g_conf.log_to_file
- Remove the log_to_file configuration option. Instead, only log to a file
if either log_file or log_dir is set.
This ... - 03:29 PM Revision 5ebd4f86 (ceph): logging: default to foreground logging
- At global constructor time: default to logging everything to stderr.
During common_init: set appropriate logging def... - 03:29 PM Revision 0c87dfda (ceph): dout: Reopen dout after parsing all config opts
- Reopen the dout stream only after we parse all configuration options.
Specifying --log-file on the command line now w... - 03:26 PM Linux kernel client Bug #856 (Resolved): busy loop osd reconnects if osd is down
- This is fallout from commit:ab48fe45f3f51e54975be158a9b4ca3ec90d4aad.
There is a confusing and subtle breakdown ... - 03:16 PM Revision 07b99e85 (ceph): dout: remove g_conf.log_to_file
- Remove the log_to_file configuration option. Instead, only log to a file
if either log_file or log_dir is set.
This ... - 02:44 PM CephFS Bug #854: unsynchronized clocks between kernel-client/cmds cause PJD fstest failures
- Ah, that makes sense. This is something we're unlikely to fix -- currently a lot of operations occur "on" the MDS (re...
- 02:32 PM CephFS Bug #854 (Duplicate): unsynchronized clocks between kernel-client/cmds cause PJD fstest failures
- I'm seeing a varied number (generally 5-8) of POSIX tests within the PJD fstest suite failing when the tests are bein...
- 02:40 PM Revision 67355779 (ceph): Merge branch 'logging'
- 02:40 PM Revision b9224d81 (ceph): logging: default to foreground logging
- At global constructor time: default to logging everything to stderr.
During common_init: set appropriate logging def... - 01:52 PM Linux kernel client Bug #852: secret key shows up in /etc/mtab after mount -o secretfile=/.../key
- we should probably be using keyctl?
http://www.ibm.com/developerworks/linux/library/l-key-retention.html - 12:32 PM Linux kernel client Bug #852 (Resolved): secret key shows up in /etc/mtab after mount -o secretfile=/.../key
- mount.ceph reads secretfile in and passes mount the actual secret. It becomes <hidden> in /proc/mounts, but /etc/mta...
- 01:51 PM Revision c916905a (ceph): osd: cache map bufferlists until they are flushed to disk
- Another thread may share maps with a peer. Make sure they pull bufferlists
out of our cache if this happens prior to... - 01:29 PM Revision d5da4f07 (ceph): safe_io: fix signed/unsigned comparisons
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 01:22 PM Linux kernel client Bug #853: bonnieloop.sh
- Kernel client: 7751b01
- 01:17 PM Linux kernel client Bug #853 (Resolved): bonnieloop.sh
- On ceph1 and ceph2 I ran bonnieloop.sh against sepia (simultaneously). In both cases, I got the following output:
... - 12:49 PM Revision e55c2a94 (ceph): osd: handle osd_ping (and ack requests) while !active
- In particular, we may start getting ping requests before getting (or while
processing) our first map that makes us go... - 12:49 PM Revision 7cd8ac0b (ceph): osd: trigger discover_all_missing after replay delay
- We were calling discover_all_missing only when we went immediately active,
not after we were in the replay state (whi... - 12:49 PM Revision 49bbe0af (ceph): mds: fix one rename dentry linkage projection case
- There are more. :(
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 12:49 PM Revision 74f5bc50 (ceph): mds: drop some dead code
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 02:40 AM Bug #339: OSD crash: ReplicatedPG::sub_op_modify
- Hit this bug yesterday. The gdb output:...
- 12:29 AM Revision 0fb5ef2c (ceph): Merge remote branch 'origin/scrub_noblock'
- 12:25 AM Revision c031e5af (ceph): CDentry/CDir/CInode: Add comment on use of boost::pool.
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 12:10 AM Revision f353f596 (ceph): osd: simple test for random missing objects during recovery
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 12:04 AM Revision 0b886a8f (ceph): librados: remove unused member of IoCtx
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 12:03 AM Revision b86461dc (ceph): osd: recovery cleanups, better error messages
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 12:02 AM Revision 29cb6f86 (ceph): osd: update missing_loc when infering an empty missing set
- We infer an empty missing set, but weren't calculating object locations
based on that. Usually it was okay because w... - 12:01 AM Revision f74593ee (ceph): osd: fix unfound output
- We were printing unfound when not, and vice versa.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
03/01/2011
- 11:19 PM Revision 7c2e4bfa (ceph): librados: IoCtx destructor should put reference only if initalized
- 11:11 PM Revision f3bdfc57 (ceph): osd: add object to missing if we find it missing on disk
- If the recovery finds the object missing on disk during recovery, add it
to the local missing set so we can (hopefull... - 11:01 PM Revision 9359cb58 (ceph): Makefile.am: fix for testlibrbdpp
- 10:42 PM Revision 14592345 (ceph): osd: (semi-)handle case where primary copy isn't there
- Continue recovering, at least.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 10:25 PM Revision 5997059a (ceph): osd: continue recovery after encountering missing objects
- 1- If we try to pull an object that isn't there, send an empty push in
reply.
2- If we get an empty push, call a new... - 10:20 PM Revision afbd7050 (ceph): librados: can set up object locator
- 09:53 PM Revision 26dfe159 (ceph): object_locator: fix clear()
- 08:06 PM Revision 412af3ec (ceph): librados: make IoCtxImpl a pointer in WatchContext
- Adds get and put to IoCtxImpl for refcounting,
and uses them in WatchContext, which shouldn't
be creating a copy of t... - 08:06 PM Revision fa8fa401 (ceph): rgw: bucket creation uses tmap for atomic update
- 08:06 PM Revision a524c093 (ceph): rgw: initialize bucket creation time
- 08:06 PM Revision 3e4cf6fe (ceph): rgw: Remove unused definition of "SERVER_NAME"
- 08:06 PM Revision 5bc1a332 (ceph): rgw: Remove unused CGI_PRINTF lines
- 07:35 PM Revision f013b8a7 (ceph): librados: decrement refcount of old io_ctx_impl in assignment operator
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 07:26 PM Revision 644bcb6c (ceph): debian: add libcurl dependency
- 07:26 PM Revision 4ed8bec6 (ceph): debian: fix libcurl dependency again
- 07:26 PM Revision 3552c29b (ceph): debian: remove extra dependencies
- 07:25 PM Revision d3285d65 (ceph): rgw: move things around, separate implementations
- 07:25 PM Revision 090a7c0b (ceph): rgw: move authorization to the specific rest handlers
- 07:25 PM Revision fc63d973 (ceph): rgw: can output data in multiple formats
- 07:25 PM Revision 543b9864 (ceph): rgw: add missing files
- 07:25 PM Revision 93e56b61 (ceph): rgw: openstack: list containers, list objects
- 07:25 PM Revision fecf37e2 (ceph): rgw: openstack: implement some more functionality
- 07:25 PM Revision 4e3ad0c6 (ceph): rgw: fix get obj
- 07:25 PM Revision be6f8489 (ceph): rgw: openstack put object fixes
- 07:25 PM Revision 0bad44a7 (ceph): rgw: multiple openstack, rgw_admin fixes
- 07:25 PM Revision e2c11254 (ceph): Makefile.am: update noinst_HEADERS with new rgw headers
- 07:25 PM Revision 0a4f4d72 (ceph): configure: add libcurl dependency for rgw
- 07:24 PM Revision 3428fe3f (ceph): rgw: verify openstack tokens
- 07:24 PM Revision fe871d98 (ceph): rgw: keep openstack user to rgw user mapping
- 07:24 PM Revision 3e62d8a2 (ceph): rgw: add a cache layer for the backend
- 07:24 PM Revision 57d0c17b (ceph): rgw: fix crash
- 07:24 PM Revision c77c753e (ceph): rgw: disable cache
- 06:36 PM Revision 57248073 (ceph): testrados: add object stat test
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 06:33 PM Revision 58d60e04 (ceph): testlibrbd: recreate test pool each time
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 04:52 PM Bug #802: osd: failing to send heartbeats (btrfs hang?)
- Fixed a few different bugs in this area, although we haven't specifically figured out why Jim was seeing that weird h...
- 04:47 PM Revision 866b161b (ceph): osd: enable cpu_profiler
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 04:38 PM Feature #609 (Resolved): osd: query pool/pg for objects with given xattr
- 04:34 PM Bug #632: init script won't stop an instance that's been removed from config
- The way to fix this would be to kill anything that has a pid file in /var/run/ceph, without looking at the ceph.conf....
- 04:33 PM Bug #645 (Closed): intermittent failure of snap ioctl
- Sam tracked this down to a incorrectly closed file descriptor in the filestore init or some such. Phew!
- 04:32 PM Bug #704 (Closed): it hang both in client and osd
- the ceph -s output shows the mds in state 'up:replay'.. that's why the client couldn't mount.
- 04:30 PM Bug #749 (Can't reproduce): test pipe queue if is empty??V0.24.2
- 04:30 PM Messengers Bug #798 (Can't reproduce): should have a empty() test on list<Messegger *>
- 04:15 PM Revision 35c4a9ff (ceph): Thread: don't mask signals except in library code
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 04:11 PM Bug #847: OSD: Recovery does not complete, PG's stay active
- Pushed an osd_recovery branch that should continuing recovering, skipping over the missing objects. If there are oth...
- 01:41 PM Bug #847: OSD: Recovery does not complete, PG's stay active
- Oh, I see the problem:
2011-03-01 22:39:59.563426 log 2011-03-01 22:27:18.519821 osd1 [2a00:f10:113:1:230:48ff... - 05:06 AM Bug #847: OSD: Recovery does not complete, PG's stay active
- I've tried your patch, but unfortunately it did not work out.
Did a little digging and out:... - 03:51 PM Bug #836: atomic-ops error on Debian arm
- __arm__ is not the problem, according to the debian arm port people. tv helped me get a qemu arm setup going, which I...
- 03:50 PM Tasks #834 (Resolved): Investigate heap usage via tcmalloc's extensions
- Ugh, so after entirely too much time looking at all kinds of things: tcmalloc had a lot of resident memory used up as...
- 03:10 PM Revision d7dea8ea (ceph): Merge branch 'profiler'
- 03:04 PM Revision f3ff51f4 (ceph): librados:rados_ioctx_stat -> rados_ioctx_pool_stat
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 02:45 PM Revision 7a429d91 (ceph): profiler: move perftools glue into perfglue/
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 02:44 PM Revision 18e9b66e (ceph): do_autogen: add -P (--with-profiler)
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 02:44 PM Revision 9f95489a (ceph): mds: cpu profiler hooks
- 02:44 PM Revision 86363131 (ceph): add --with-profiler to link in google perftools libprofiler
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 02:21 PM Revision a4668008 (ceph): librados: use atomic_t for reference count
- Use an use atomic_t for the reference count in IoCtxImpl.
Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com> - 12:20 PM Revision 0290074b (ceph): librados: fix IoCtx::from_rados_ioctx_t
- IoCtx::from_rados_ioctx_t creates an IoCtx out of a rados_ioctx_t.
However, this IoCtx must share ownership of the Io... - 11:38 AM Revision 1f5fca85 (ceph): Rename radios_ioctx_{open,close} to create/destroy
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 11:33 AM Revision d134915f (ceph): librados: remove IoCtx::close()
- We decided we don't want IoCtx::close(), since IoCtx::~IoCtx() exists.
Signed-off-by: Colin McCabe <colin.mccabe@dre... - 01:07 AM Revision f1911758 (ceph): PG: unify scrub_received_maps and peer_scrub_maps
- Previously, incoming maps were placed into peer_scrub_maps and merged
into scrub_received_maps during scrub_gather_re... - 12:05 AM Revision 19bcd747 (ceph): osd: trigger discover_all_missing after replay delay
- We were calling discover_all_missing only when we went immediately active,
not after we were in the replay state (whi... - 12:03 AM Revision 7a2bb8dc (ceph): Make git ignore core files.
- 12:03 AM Revision adb67509 (ceph): Make git ignore python generated files.
- 12:00 AM Revision d372c50f (ceph): librados: Crashed on shutdown if connect was never called.
- Add a trivial unit test to trigger this.
02/28/2011
- 11:43 PM Revision 03135a3e (ceph): OSD: activate_map should not do_peer on a replaying pg
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 11:43 PM Revision 2adb670c (ceph): OSD: add scrub_finalize_wq
- Scrub currently blocks while waiting on replica maps and for
last_update_applied==last_update. Also, the subsequent ... - 11:43 PM Revision 88ced090 (ceph): PG: make scrub non-blocking
- Previously, scrub would block using wait until
1. last_update_applied==last_update and
2. all replica scrub maps are ... - 11:43 PM Revision d03bd60c (ceph): PG: replica_scrub also should not block
- As with scrub, replica scrub wait()ed for last_update_complete to catch
up to last_update. Now, it will requeue the ... - 11:43 PM Revision dd1bd2c4 (ceph): PG: refactor scrubmap comparison and repair logic
- The previous version gave erroneous results. This version seems simpler
and can be more easily unit tested as the er... - 10:15 PM Revision ee534522 (ceph): osd: handle osd_ping (and ack requests) while !active
- In particular, we may start getting ping requests before getting (or while
processing) our first map that makes us go... - 09:11 PM Revision 399032cc (ceph): osd: share osd->osdmap with map cache
- Make OSD::osdmap just a shortcut map pointer to the most recent OSDMap, the
same one that's in the map cache.
Signed... - 09:04 PM Revision 5a18f988 (ceph): Merge branch 'next'
- 08:57 PM Revision c76a4b75 (ceph): Revert "move g_default_file_layout into ceph_fs.cc"
- This reverts commit 1dc12e3e1de1ee6aeb3ef11bb3faafa4757b1a65.
The headers and ceph_fs.cc are written such that they ... - 08:51 PM Linux kernel client Bug #851 (Resolved): make dcache readdir with I_COMPLETE work
- 08:51 PM Linux kernel client Bug #850 (Resolved): make NULL lookup using I_COMPLETE work
- 07:19 PM Revision a2ffef9d (ceph): Merge branch 'next'
- Conflicts:
src/pybind/rados.py - 05:29 PM Feature #812: switch up crypto library (again)
- I think it has what we need (mainly sha1, md5). The pkl11test.c from
the source package can be a good starting point... - 04:35 PM Bug #847: OSD: Recovery does not complete, PG's stay active
- 04:06 PM Bug #847: OSD: Recovery does not complete, PG's stay active
- I believe this will fix it: commit:19bcd747b452b79c80744620e635f036bfc1b738
- 11:58 AM Bug #847: OSD: Recovery does not complete, PG's stay active
- osd4 started printing some more lines after it's restart:...
- 11:51 AM Bug #847 (Resolved): OSD: Recovery does not complete, PG's stay active
- After I've expanded my cluster from 4 to 8 OSD's the recovery would not complete.
The state right now:... - 04:06 PM Revision 8d877abc (ceph): Add comment to ceph_fs.cc
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 03:52 PM rgw Cleanup #320 (Resolved): Remove some duplicates and unused declarations
- applied everything that was still relevant.
- 03:44 PM rgw Bug #768 (Resolved): eliminate any read/modify/write races
- bucket creation is being done now via tmap update on a rados backed gateway.
- 02:20 PM Feature #849 (Resolved): include rbd udev rules in packages
- 02:00 PM Revision 70564eb5 (ceph): libradoshpp: put ceph stuff in namespace librados
- Try a little bit harder to avoid polluting the user's global namespace
with our stuff.
Signed-off-by: Colin McCabe <... - 01:59 PM Revision a66fd3bb (ceph): librados: don't create unused SnapContext objs
- There were some unused temporary variables hanging around.
Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com> - 01:49 PM Bug #848 (Resolved): limited commit_partial probably broke mds recovery
- When I set up the MDS to limit the size of commits it sends to the OSDs, Sage points out that I probably broke safety...
- 01:32 PM Revision 13f6223f (ceph): librados: fix copy ctor of ObjectIterator
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 01:12 PM Bug #821 (Resolved): osd: fix osdmap caching
- This is fixed without shared_ptr<>. See commit:399032cca53c37dca08713818621a24b924fd4a6 and commit:96c2cb5b986d6e7fc...
- 12:31 PM Bug #833: rbd wiki documentation uses old sysfs /sys/class interface
- The way to list snaps now is to list the device directory and pick up all the entries that start with 'snap_'.
The... - 12:07 PM Bug #833: rbd wiki documentation uses old sysfs /sys/class interface
- Fixed the docs a bit, but I'm missing "list_snaps" in the current RBD sysfs tree. Is there still a away to list snaps...
- 12:03 PM Revision e7f2972d (ceph): cconf: fix clitest
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 11:50 AM Revision 3011349a (ceph): unittests: remember to use AM_LDFLAGS
- remember to use AM_LDFLAGS when setting _LDFLAGS. Otherwise, the global
flags will be lost.
Signed-off-by: Colin McC... - 11:50 AM Bug #838 (Duplicate): librados should not write to stderr or ceph logs, touch pid files, silently...
- Good observations. These are being addressed in the next release by #840 and #842, which are subtasks of #815
Comm... - 10:18 AM Bug #838: librados should not write to stderr or ceph logs, touch pid files, silently read config...
- The log to stderr part was meant to complain about logging a "greeting". Logging on an actual error is more acceptabl...
- 10:15 AM Bug #838 (Duplicate): librados should not write to stderr or ceph logs, touch pid files, silently...
- Using the Python bindings just as an easy way to init librados, the bug does not reside in the bindings.
proof of ... - 11:47 AM Subtask #846 (Resolved): De-globalize SimpleMessenger, etc.
- In librados, and other library code, we need an instance of SimpleMessenger for each active rados_cluster_t.
We pr... - 11:46 AM Subtask #845 (Resolved): g_conf should not be defined in library code.
- g_conf should live in cmon, cosd, and other daemons. Attempts to refer to it in a library should create an undefined ...
- 11:45 AM Subtask #844 (Resolved): g_conf should become a pointer
- g_conf should become a pointer. It should not be initialized or used during global constructor time.
- 11:45 AM Subtask #843 (Resolved): Decouple dout from g_conf
- Decouple dout from g_conf
Create DoutConfig to represent dout configuration state. - 11:44 AM Subtask #842 (Resolved): initialization: have appropriate library-specific defaults
- initialization: have appropriate library-specific defaults
For example, don't default to using a pid file in libra... - 11:43 AM Subtask #841 (Resolved): don't call daemon_init in library code
- Common init should not be called by libraries. We don't want to register signal handlers, call openlog(), or really d...
- 11:41 AM Subtask #840 (Resolved): Clean separation between different components of initialization
- We want to have three different components:
1. argument parsing
2. reading the ceph configuration file
3. common_i... - 11:39 AM Subtask #839 (Resolved): Globals cleanup. Replace g_conf.name with g_conf.entity_name.to_str(). R...
- Globals cleanup. Replace g_conf.name with g_conf.entity_name.to_str(). Remove g_conf.type.
In some cases, use progra... - 09:57 AM Revision f7e5e868 (ceph): Merge branch 'librados_api' into next
- 09:46 AM Feature #820 (Closed): bufferlist memory pool
- So, tried this out and it looked like maybe it helped a little bit, but it wasn't clear. After running more tests wit...
- 09:23 AM Bug #735 (Resolved): Manual drive pull testing hangs filesystem
02/27/2011
- 09:01 PM Tasks #834: Investigate heap usage via tcmalloc's extensions
- I've spent some time on this and so far I've gotten into situations where tcmalloc is keeping around a couple hundred...
- 07:42 PM Bug #698 (Resolved): cosd memory usage with large number of pools
- I'm closing this since it's become apparent that the actual memory use issues are less related to the in-memory objec...
- 07:37 PM Bug #836 (In Progress): atomic-ops error on Debian arm
- Only thing that makes any sense to me so far is the possibility that the preprocessor macros are incorrect for the sy...
- 07:04 PM Bug #836 (Rejected): atomic-ops error on Debian arm
- From Lazslo:...
- 07:08 PM Bug #837 (Resolved): AuthAuthorizeHandler fails to build on s390
- Also from Laszlo:...
02/26/2011
- 11:51 PM Bug #835 (Resolved): osd: avoid the sync in handle_osd_map
- May not be easy (or even possible), but at the very least need to document what the dependencies are.
- 04:40 PM Feature #820 (In Progress): bufferlist memory pool
- I did a prototype of this, and in brief testing it reduced resident memory usage a little bit, but not by enough. Mor...
- 04:40 PM Tasks #834 (Resolved): Investigate heap usage via tcmalloc's extensions
- tcmalloc has a lot of knobs we can fiddle with programmatically to get information out (and to change behavior). Impl...
- 11:31 AM Bug #833 (Resolved): rbd wiki documentation uses old sysfs /sys/class interface
- http://ceph.newdream.net/wiki/Rbd
- 02:05 AM Revision 78ba46ad (ceph): testlibrbd: call rados_connect
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 01:02 AM Revision 141509c0 (ceph): Rename PoolHandle to IoContext: part 2
- The previous change changed all PoolHandle uses to IoContext. This
change also renames the variable names.
Also fix ... - 12:27 AM Revision 454854ed (ceph): testlibrbdpp: convert to new APIs
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 12:27 AM Revision 8903ea34 (ceph): testradospp: update for new librados API
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
02/25/2011
- 11:45 PM Revision b6547875 (ceph): librbd, librados: fix my last commits to use the new librados API
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 11:33 PM Revision 2641a693 (ceph): librados: add snap_get_stamp to C API
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 11:32 PM Revision 2cb86f71 (ceph): librados, librbd: remove selfmanaged_snap_rollback_object
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 11:32 PM Revision 15260473 (ceph): testlibrbdpp: initialize pointers
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 11:29 PM Revision 040e5460 (ceph): pybind/rados: Fix invalid variable references in error paths.
- Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com>
- 11:28 PM Revision 8c2d5729 (ceph): pybind/rados: Remove unused variable.
- Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com>
- 11:26 PM Revision 9591729a (ceph): pybind/rados: version call return type is void.
- Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com>
- 10:43 PM Revision 32cf5ac4 (ceph): FileStore.h: reorder queue operations in _journaled_ahead
- In writeahead mode, an op could dissappear from jq without immediately
reappearing in q. Thus, q can be empty before... - 10:43 PM Revision afe2d81b (ceph): Revert "FileStore: fix OpSequencer::flush error"
- This reverts commit c78b29a47d7211a4b8b1585112ac22b8435a82c7.
This commit introduced an error in parallel journaling... - 09:41 PM Revision 96c2cb5b (ceph): osd: simplify handle_osd_map flushing and map caching
- Extend the map_cache somewhat. Avoid one flush in handle_osd_map by
putting new maps in the cache.
There is still a... - 09:39 PM Revision 9758af7d (ceph): testlibrbdpp: use new librbd api
- 06:58 PM Revision f9fe949e (ceph): rbd: update for librbd api changes
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 06:58 PM Revision f54ba07b (ceph): librbd: tweak C++ API
- - rename image_open to open and make it return an int
- remove Image::close, replace with destructor
- make Image con... - 06:52 PM Revision 74aa6841 (ceph): librados: Rename rados_pool_t to rados_ioctx_t
- rados_pool_t -> rados_ioctx_t
class PoolCtx -> class IoCtxImpl
class PoolHandle -> class IoCtx
PoolHandle::name() ... - 05:08 PM Revision bde6dac2 (ceph): test_common.sh: should rm objects before adding
- rados_write doesn't replace the whole object, but that's what we want in
these old tests. So just rm it first.
Signe... - 04:42 PM Revision 4937854b (ceph): rbd: de-globalize rbd, rados, Image
- Use RAII for rbd, rados, and Image. Their destructors will be called
when main exits, thus doing the cleanup for us. ... - 04:17 PM Revision 1d2c5f35 (ceph): Merge branch 'librados_api_cpp' into librados_api
- Conflicts:
src/include/rbd/librbd.hpp
src/librbd.cc
src/rbd.cc - 04:06 PM Revision 06bf77e1 (ceph): librados: C++ API rework
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 12:46 PM rgw Feature #832 (Resolved): Create fastcgi repository
- created the mod_fastcgi git repository.
- 10:22 AM rgw Feature #832 (Resolved): Create fastcgi repository
- 10:39 AM rgw Bug #343: Bad 100-Continue handling
- 10:14 AM Bug #793: osd: avoid blocking in scrub_wq
- Pushed a branch (scrub_noblock) which should prevent scrub (and replica scrub) from blocking.
- 09:54 AM rgw Feature #831 (Resolved): self signed tokens for the openstack requests
- We need that so that we wouldn't need the extra step to authorize the user on the auth server.
- 09:41 AM rgw Feature #830 (Resolved): rgw: swift per-object ACLs
- We need to translate between the openstack acls to the current acls, both in requests that update the acls and in req...
- 09:37 AM rgw Feature #829 (Resolved): rgw: support swift POST
- 09:07 AM Feature #828 (Resolved): make cpu profiler work
- The google-perftools cpu profiler isn't collecting any samples. I think it's because of the signal masks on our thre...
- 09:04 AM Bug #826: autoconf: don't autodect fuse, tcmalloc, etc.
- If we do this we'll need to put some magic somewhere so that the debian packages will build happily on architectures ...
- 01:20 AM Revision e3f3177f (ceph): rbd: use new librbd C++ api
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
02/24/2011
- 09:50 PM Revision 037d6620 (ceph): Update Debian packaging for 0.26~rc.
- Make ceph x86 and x64 only because of build dependency on architecture related
google-perftools.
Put library packages... - 09:41 PM Revision 154a21f3 (ceph): librbd: make C++ api nicer
- Adds Image class and replaces aio_create_completion with a constructor.
Signed-off-by: Josh Durgin <josh.durgin@drea... - 09:40 PM Revision 9570d6e2 (ceph): Revert "FileStore: fix OpSequencer::flush error"
- This reverts commit c78b29a47d7211a4b8b1585112ac22b8435a82c7.
This commit introduced an error in parallel journaling... - 09:40 PM Revision 8dfffd00 (ceph): FileStore.h: reorder queue operations in _journaled_ahead
- In writeahead mode, an op could dissappear from jq without immediately
reappearing in q. Thus, q can be empty before... - 07:24 PM Revision fa3dee1f (ceph): librbd, rbd: fill in the rest of image_info_t
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 07:14 PM Revision d15fe7f9 (ceph): rollback rename
- 03:50 PM Revision 813ae598 (ceph): mds: add 'exit' command
- Tell a cmds process to suicide/exit immediately.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 03:50 PM Revision 0f3fe7f8 (ceph): mds: fix frag string rendering
- Was mostly gibberish from df7c7bd79237d2a8b691f4e59433b0b39a9721a2
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 03:41 PM Revision 827c950b (ceph): v0.26~rc
- Modernize configure.ac macros, set bug report address to ML.
- 03:36 PM Revision 18305fc3 (ceph): mds: strengthen assertions in rejoin ack
- The ACK only contains items we asked for with a WEAK request. Assert as
much. (The old continue bits were from ~200... - 03:35 PM Revision 7abf87a5 (ceph): mds: fix gratuitous map lookup
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 03:35 PM Revision d8446023 (ceph): mds: mark_down connections to any failed peers
- This cleans up messenger state, prevents log spam, and saves a small amount
of memory.
Signed-off-by: Sage Weil <sag... - 03:35 PM Revision 3e691d31 (ceph): mds: fix export cancellation vs nested freezes
- Prevent freezes from completing while we are canceling exports. Otherwise
if we are freezing /a/b and /a, and cancel... - 02:12 PM Revision 7c66adf1 (ceph): filejournal: fix type punning warning, drop unneeded cast
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 02:02 PM Revision 32c913a2 (ceph): Some tweaks for the librados C API
- rados_reopen_log: should take a cluster parameter.
Add rados_pool_list, rados_pool_list_free.
rados_snap_set_read -... - 01:49 PM Revision 71bbd554 (ceph): Makefile: fix libatomic_ops linking
- LDADD seems to have no effect on the final link command. Switching this
back to AM_LDFLAGS. This was changed as in ... - 08:34 AM Revision 92870d7a (ceph): mds: remove "N stopped" from short mdsmap summary
- It's confusing because it sounds like we're talking about daemons, when we
really just mean there are some ranks that... - 08:31 AM Revision f027b082 (ceph): mon: include mds gid in logs
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 08:20 AM Revision 9eb35324 (ceph): mds,osd: print 'starting ...' message to stdout
- The timestamp/threadid prefix is unnecessary, and stdout seems more
appropriate. Now matches cmon.
Signed-off-by: S... - 08:19 AM Revision 298916b1 (ceph): common: only print version to stdout for daemons
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
02/23/2011
- 11:09 PM Revision c78b29a4 (ceph): FileStore: fix OpSequencer::flush error
- In writeahead mode, an op will dissappear from jq without immediately
reappearing in q. Thus, q can be empty before ... - 09:45 PM Revision 15fa05d8 (ceph): mds: print waiter tag in hex
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 09:36 PM Revision df7c7bd7 (ceph): mds: make frag string rendering simpler
- Show actual bit prefix when rendering a frag_t. That is,
$value/$numbits -> bits*
So,
0/0 -> *
000000/1 -> 0... - 09:34 PM Revision 3b1fa97c (ceph): mon: fix dup mds takeover
- Allow a standby to take over for a single MDS only by consistently looking
at the pending_mdsmap and not mdsmap. Mix... - 09:18 PM Revision 8d370026 (ceph): mds: print msg when fragtree updates from journal
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 09:17 PM Revision 436153e6 (ceph): mds: verify frags in more approrpiate places
- Not in inner helpers, which may be called on multiple frags to get things
in sync.
Signed-off-by: Sage Weil <sage.we... - 09:01 PM Revision f13e6a4b (ceph): mds: refragment dirs when inode dirfragtree updates from journal
- Force dir fragmentation specified by dirfragtree when replayed from
the journal.
Example:
mds0 is auth for /foo, md... - 07:55 PM Revision f6a92610 (ceph): mds: fix CDir::take_waiting() on dentry waiters
- Using take_dentry_waiting() means we double-put the DNWAITER pin. It's
also way slower.
Signed-off-by: Sage Weil <s... - 06:17 PM Revision efc20c67 (ceph): Update testrados, rename functions in librados.cc
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 06:17 PM Revision 0fb83f17 (ceph): More fixes, additions for config API
- Add test of the librados configuration API to testrados.c
rados_reopen_log should return void since it can't encount... - 06:17 PM Revision f2e3d838 (ceph): config.cc: doesn't depend on ceph_ver.h
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 06:17 PM Revision 431a5bf7 (ceph): Rename config.h -> common/config.h
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 06:17 PM Revision 29b64391 (ceph): ceph_common.sh: remove get_val, get_val_bool
- get_val and get_val_bool are unused.
Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com> - 06:17 PM Revision 713f9a11 (ceph): cconf: remove second argument to cconf --lookup
- Everyone uses get_conf to get configuration values. So the logic for
defaulting to some value if we can't find the re... - 06:17 PM Revision ebc9dbc1 (ceph): common: Fix some missing includes, copyrights
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 06:17 PM Revision 21544fdc (ceph): common: more include and copyright fixes
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 06:17 PM Revision 0d54e311 (ceph): md_config_t::set_val/get_val
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 06:17 PM Revision 871fa1b3 (ceph): Add rados_conf_apply, comments
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 06:17 PM Revision 5f730868 (ceph): Splt rados_init into rados_create + rados_connect
- Splt rados_init into rados_create and rados_connect. The pattern will
be for users to call create, set configuration... - 06:17 PM Revision 0f90d156 (ceph): Fold common_set_defaults into common_init
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 06:17 PM Revision 3c6736d6 (ceph): rados_create: add id parameter
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 06:17 PM Revision 96edddea (ceph): testlibrbdpp: update for new librados and librbd APIs
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 05:31 PM Revision b2ceebf4 (ceph): librbd: stack on top of librados
- 05:31 PM Revision 15f20f5b (ceph): librados: switch to noun_verb function names
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 05:31 PM Revision b1d6e30e (ceph): librados: add constructor to allow client re-use
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 05:31 PM Revision 134f261a (ceph): librbd: implement stacking on top of librados
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 05:31 PM Revision c5bb7350 (ceph): rbd: update for new librados and librbd APIs
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 03:38 PM Feature #709: nagios plugin
- The issue for the nagios machine upgrade: https://dev.newdream.net/issues/8978
- 12:36 PM Tasks #827 (Resolved): Make i386 build cleanly in gitbuilder
- We're getting an Internal Compiler Error!
Two tasks:
1) Package this up for the gcc folks.
2) Implement a workar... - 12:01 PM Bug #826 (Closed): autoconf: don't autodect fuse, tcmalloc, etc.
- Any library that isn't present should make configure fail. This forces the user to do --without-foo and realize what...
- 09:07 AM Subtask #825 (Resolved): osd: remove pg map updating from handle_osd_map
- Currently handle_osd_map flushes worker threads before applying an update. Ideally we should advance pgs individuall...
02/22/2011
- 11:09 PM Revision 74346592 (ceph): ReplicatedPG: snap_trimmer should bail out while finalizing_scrub
- Check to make sure !finalizing_scrub when relocking.
Signed-off-by: Samuel Just <samuel.just@dreamhost.com> - 11:09 PM Revision 4ddc5db6 (ceph): OSD,PG: fix race between processing scrub and dequeueing scrub
- Previously, a second scrub could be scheduled between when the first is
dequeued and processed resulting in two scrub... - 08:45 PM Revision 89b743c4 (ceph): osd: fix recovery pointer when pulling head before snapid
- If recovery wants to pull a snapped object and needs the head first, pull()
does that, but the caller doesn't ++skipp... - 08:20 PM Revision 48f69894 (ceph): osd: verify object version during push
- Fail to push if the ondisk version doesn't match the version we want to
send.
This isn't supposed to happen. If it d... - 05:40 PM Revision e4b82742 (ceph): osd: improve up_thru request behavior
- There is some epoch the OSD wants for up_thru, based on when the PG mapping
last changed. However, once the monitor ... - 05:27 PM Revision 4df36436 (ceph): pybind: update rados python bindings for new API
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 05:11 PM Revision 158889b2 (ceph): PGMap: make osd_full and nearfull ratios configurable.
- These were previously set by #defines. Pretty stupid
when we have a nice config system already!
Signed-off-by: Greg ... - 05:10 PM Revision 0e4e5a5b (ceph): osd: set correct epoch for up_thru osd->mon request
- Put the epoch we need for up_thru in the request. Putting the most recent
epoch causes incorrect osdmap churn.
Fixe... - 04:09 PM Revision 2de8b7b8 (ceph): librados: add cluster handle to C API
- Had to add a layer of indirection to the list context handles.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 04:09 PM Revision 1dc12e3e (ceph): move g_default_file_layout into ceph_fs.cc
- It's defined in ceph_fs.h.
Fix a bunch of headers that use types without including the headers that
define those typ... - 04:09 PM Revision 950e8c3e (ceph): librados: more API cleanup; rados_conf_ stubs
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 09:14 AM Bug #817: Implement custom propose logic for OSDMap/PGMap
- This is less important since Sage figured out #824.
- 09:07 AM Bug #824 (Resolved): Map churn with lots of PGs
- commit:0e4e5a5b26c1027cfb99edce869fecc4a5a2d487
- 07:46 AM Bug #824 (Resolved): Map churn with lots of PGs
- Like the title says. This may be related to #817 but Sage seems to think there's also an issue with the OSD code prom...
- 08:53 AM Bug #810 (Duplicate): 1). PG bits don't get recognized and 2). Takes too long for OSDs to boot up.
- Setting pg bits is working properly, but you're using the wrong config name. :) Underscores should only be used on th...
- 08:51 AM Feature #823 (Resolved): osd: make FULL and NEAR_FULL configurable
- Made these configurable via mon_osd_full_ratio and mon_osd_nearfull_ratio in commit:158889b2e4089db2361b7f61d6e2e2c13...
- 08:41 AM Bug #816: fs size underflowed
- First step here is to figure out how to reproduce, and/or find/generate full osd logs of it happening.
- 08:34 AM Bug #752 (Resolved): High I/O wait when writing data
- That's good news. I'm not really sure what was going wrong here before either. Let's see if this comes back.
- 06:09 AM Bug #752: High I/O wait when writing data
- I've spent a lot of time testing and finding out where this could come from, but it seems to be done now for no good ...
02/21/2011
- 05:00 AM Revision 0a96e6a7 (ceph): Makefile: include ceph_argsparse.h in dist tarball
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 04:58 AM Revision 4ffe0698 (ceph): filestore: fix clone_range
- This was broken by the safe_write() switchover; the success return value
is now 0, not the number of bytes written.
... - 04:36 AM Feature #823 (Resolved): osd: make FULL and NEAR_FULL configurable
- FULL and NEAR_FULL are set in osd/osd_types.h:...
02/20/2011
- 09:54 PM Revision 349cb128 (ceph): keyring_init: don't print error when explicit key/keyfile is specified
- e.g. when I am non-root and specify a key explicitly, no need to complain
about not being able to read root's /etc/ce... - 09:52 PM Revision 859aaf78 (ceph): Revert "keyring_init: g_conf.keyring is not a list"
- This reverts commit 2fb6036aa53f5eb3173b80fd17b7240bd3daf156.
- 06:17 PM Revision 2a59b213 (ceph): common: Split argument parsing into ceph_argparse
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
02/19/2011
- 12:01 AM Revision b96f1642 (ceph): Revert "Makefile.am: remove unused libs from linking with librbd tests ...
- Same problem as 38f38a99149e88f18072fcbdbee316ac21f6f30f.
This reverts commit e5db46cea0997f3f959b2ae896c980585f079ac0.
02/18/2011
- 11:42 PM Revision a7929c5e (ceph): Merge branch 'pool_memory'
- 11:41 PM Revision f49909ba (ceph): PG: convert hash_maps to maps, remove unused.
- waiting_for_[missing|degraded]_object don't need to be
hash_maps, and we don't use stat_object_temp_rd at all.
Swap t... - 11:41 PM Revision 935eaf3a (ceph): PG: remove the object locking stubs and some dead code.
- These are unused (#if 0'd, so no way to use them!) and require
a memory-hogging hash_map. Goodbye!
Signed-off-by: Gr... - 11:41 PM Revision eea1f098 (ceph): OSD: convert waiting_for_pg from hash_map to map.
- This doesn't need to be a hash_map; there will only be an entry
for each PG that gets a message request while it's no... - 11:41 PM Revision 654dc79c (ceph): vstart: remove directories, too.
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 11:41 PM Revision a2f0cd0d (ceph): test: Add new memory tests, move to own subdir.
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 09:11 PM Revision 38f38a99 (ceph): Revert "Makefile.am: remove unused libs from linking with librbd"
- librados doesn't export ceph::buffer_total_alloc
This reverts commit 9bbd6c32a59ce0a2e4cc21a498e0b04bcd4781ed. - 08:22 PM Feature #400: mkcephfs: work on multiple nodes without using ssh
- After talking with TV about this, I think this should look more like:
- run node-specific steps on every host. ge... - 07:17 PM Revision eadc3c87 (ceph): testlibrbdpp: fix off by one error in read test
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 07:12 PM Revision e5db46ce (ceph): Makefile.am: remove unused libs from linking with librbd tests and rbd
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 06:50 PM Revision 9bbd6c32 (ceph): Makefile.am: remove unused libs from linking with librbd
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 06:02 PM Revision 2fb6036a (ceph): keyring_init: g_conf.keyring is not a list
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 04:40 PM Revision f51040cc (ceph): Clock: remove unused mutex
- We don't use a mutex in g_clock any more, so let's not construct one any
more.
Signed-off-by: Colin McCabe <colin.mc... - 04:26 PM Feature #801: librados: allow access to multiple clusters
- > the read and write functions have some strange parameters ordering:
> offset, then buffer, then length
It may ... - 04:25 PM Feature #801: librados: allow access to multiple clusters
- > BTW if we do end up changing this, we should be consistent with naming...
> some functions are rados_verb_noun an... - 03:40 PM CephFS Cleanup #814: hadoop: refactor hadoop shim in terms of java libceph bindings
- Sage,
I see the general benefits of a Java-based libceph interface, but what are your long-term plans for Hadoop o... - 03:36 PM Revision 5db1e50d (ceph): debug.h: cleanup includes
- Shouldn't need to include DoutStreambuf.h; that's all implementation.
Don't include Mutex.h, since we don't use it.
*... - 03:25 PM Bug #698: cosd memory usage with large number of pools
- I merged code which removes some hash_maps in commit:a7929c5e265d1b6502733ee9525fd93bbcfc739e; this takes per-PG memo...
- 03:18 PM Feature #820: bufferlist memory pool
- To elaborate: it appears when I compare the heap memory use reported by massif* and the resident memory usage reporte...
- 03:05 PM Feature #820 (Closed): bufferlist memory pool
- Investigate putting buffers in a separate memory pool.
- 03:16 PM Bug #822 (Resolved): fix i386 build of radostestpp
- Hitting a gcc error on gitbuilder-i386 (ubuntu 10.10, gcc 4.4.5):
testradospp.cc: In function ‘int main(int, con... - 03:08 PM Bug #821 (Resolved): osd: fix osdmap caching
- two issues (at least):
- we end up regenerating the osdmaps from disk data frequently
- we throw out cached copies ... - 02:40 PM rbd Bug #391: snap create/delete caused corruption
- oh.. missed the external rbd tool call. That might have caused the problem, as doing it while there's a running vm is...
- 02:04 PM rbd Bug #391: snap create/delete caused corruption
- > I haven't been able to reproduce this with the latest ceph and qemu-rbd. I'd like to upgrade the kvmtest cluster an...
- 01:22 PM rbd Bug #391: snap create/delete caused corruption
- I haven't been able to reproduce this with the latest ceph and qemu-rbd. I'd like to upgrade the kvmtest cluster and ...
- 02:35 PM Revision d362874e (ceph): common: Move hex dump functions into hex.h
- Move hex dump functions into hex.h. Remove unecessary includes from
debug.cc
Signed-off-by: Colin McCabe <colin.mcca... - 02:28 PM Revision 61b661b2 (ceph): Makefile: version.cc should depend on ceph_ver
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 02:25 PM Revision ef0e07d9 (ceph): debug.h: move Ceph version stuff into version.h
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 02:07 PM rbd Bug #809 (Resolved): listing snapshots when none exist causes qemu to crash
- 01:47 PM Messengers Bug #798: should have a empty() test on list<Messegger *>
- This was before 940d58d94ef6dec16affea88c5c79b1f5eba16db? Can you try to reproduce with the latest?
Thanks! - 12:54 PM Linux kernel client Feature #819 (Closed): robust nfs re-export
- 12:54 PM CephFS Feature #818 (Resolved): mds: robust lookuphash
- Make the lookuphash be thorough about locating the directory inode.
- 11:06 AM Bug #817 (Won't Fix): Implement custom propose logic for OSDMap/PGMap
- It's way too easy to generate map spam when peering. If I run "test/memuse/test_pool_memuse.sh 99 2" then in one exam...
- 07:06 AM Revision 4231cef6 (ceph): Merge remote branch 'origin/max_commit_size'
- 01:32 AM Revision 7171cf68 (ceph): librbd: hold image context lock minimally
- Holding the image context lock during snapshot removal prevented the
client from responding to a notify, causing a de...
02/17/2011
- 07:53 PM Revision da416004 (ceph): mkcephfs: fix premature tmp directory deletion
- Previously, the temp directory would be deleted after the first daemon
on a host was started leaving the second one t... - 07:24 PM Revision 04066325 (ceph): .gitignore: ignore testsnaps
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 07:02 PM Revision dfafa9f3 (ceph): .gitignore: ignore debian packaging outputs
- 06:53 PM Revision b56cafe4 (ceph): pybind/rados: write_full: remove silly extra param
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 06:48 PM Revision 3e4aaf95 (ceph): pybind/rados: implement Pool.write_full
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 06:24 PM Revision b93be8c6 (ceph): pybind/rados: implement Pool::change_auid
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 06:05 PM Revision 23df5ba3 (ceph): pybind/rados: add rados.version
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 05:58 PM Revision 8be3ad12 (ceph): pybind/rados: Add Rados.pool_exists
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 05:46 PM Revision 09e01fa4 (ceph): pybind/rados: Snap.name should be a py string
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 05:38 PM Revision aeac6a63 (ceph): pybind/rados: add snapshots
- Add snapshot lookup, iteration, creation, destruction interface.
Add test.
Signed-off-by: Colin McCabe <colin.mccab... - 04:46 PM Bug #816 (Can't reproduce): fs size underflowed
- ceph version 0.25~rc (commit:0c97d79056ba982f571ef8e720c9d488e3982f81)...
- 02:46 PM Revision ae702b3e (ceph): Add Pool::list_objects
- Add a Pool::list_objects method. Add a test for this to pybind-test.py
Signed-off-by: Colin McCabe <colin.mccabe@dre... - 02:21 PM Subtask #815: Remove globals & partition g_conf
- Good point. Some things probably should be global. However, we have a lot of things that library clients *don't* want...
- 09:11 AM Subtask #815: Remove globals & partition g_conf
- So one issue that occurred to me with setting this up is that we have a number of things that really *ought* to be gl...
- 09:08 AM Subtask #815 (Resolved): Remove globals & partition g_conf
- This bug is to track the progress of removing globals and setting up g_conf so it can be used as a parameter rather t...
- 12:41 PM Revision 0a3798ef (ceph): Add pybind-test
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 12:40 PM Revision f06604cd (ceph): rados-python bindings: Fix pool deletion a bit
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 11:10 AM CephFS Cleanup #814: hadoop: refactor hadoop shim in terms of java libceph bindings
- I mean that, at the end of the day, we should probably have:
- a java Ceph interface binding that is identical to t... - 09:00 AM CephFS Cleanup #814: hadoop: refactor hadoop shim in terms of java libceph bindings
- I'm not sure what you're after here -- you mean you want Java bindings for librados, and then the Hadoop patches shou...
- 07:14 AM CephFS Cleanup #814 (Resolved): hadoop: refactor hadoop shim in terms of java libceph bindings
- Refactor the hadoop code in terms of generic Java bindings for libceph, instead of mixing the SWIG crap in with the a...
- 09:24 AM Bug #698: cosd memory usage with large number of pools
- So, yeah. All the heapdump commands were sent *after* ceph -w reported the PGs were all active+clean, although I didn...
- 09:13 AM Feature #801: librados: allow access to multiple clusters
- Do we want to implement support for multiple clusters in SimpleMessenger, or are we going to have basically full inca...
- 07:13 AM Cleanup #813 (Resolved): libceph interface cleanup
- A few things:
- clean up the API wrt to setting up a client instance (see #801)
- clean up packaging (/usr/includ... - 07:09 AM Feature #812 (Closed): switch up crypto library (again)
- RHEL doesn't want to include crypto++ because it is not FIPS 140-2 certified (although they got a kick of out a DOE l...
- 05:36 AM Linux kernel client Bug #811 (Can't reproduce): rbd warning and crash
- ceph-client
commit 9c01177349b435186025a088f612a6f5ce2f3de9
After execute command: echo "192.168.250.70,192.168.2... - 12:47 AM Bug #810 (Duplicate): 1). PG bits don't get recognized and 2). Takes too long for OSDs to boot up.
- 1.
By default, (0.24.3) there are 264 number of placement groups per osd.
I've tried to change to something else...
02/16/2011
- 11:07 PM Revision 29b5ae0c (ceph): simplemessenger: Fix num_threads bug printout.
- Also add documentation to get_num_threads since its contract
changed significantly.
Signed-off-by: Greg Farnum <greg... - 09:22 PM Bug #698: cosd memory usage with large number of pools
- Hmm, I just tried to run one of my tests using 2 OSDs instead of one for the first time. It looks like it's still pos...
- 08:16 PM CephFS Feature #601: mds: order directory commits after rename
- Err, I guess actually that should read:
8) a commits
9) hold on b/bar gets removed
10) commit b comes in
11) b co... - 07:53 PM CephFS Feature #601: mds: order directory commits after rename
- I haven't gotten too far into it yet, but my current line of attack is to see how feasible it is to place "holds" on ...
- 05:07 PM CephFS Feature #601: mds: order directory commits after rename
- I think this is going to be much ahrder than I originally imagined. For example:
- mv /a/foo /b/foo
- mv /b/bar ... - 03:51 PM CephFS Feature #601: mds: order directory commits after rename
- I'm going to take a look at this while waiting for data collection to run as I work on #698.
- 06:28 PM Revision fc1ce083 (ceph): monmaptool: fix command-line output
- Don't check errno if it isn't set.
Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com> - 05:52 PM Bug #799: mkcephfs error
- cconf output many messages in debug mode,but some messages is useless,and affect cconf.
finally, i try to change ce... - 12:13 PM Bug #799: mkcephfs error
- can you attach your ceph.conf?
- 05:21 PM Revision 651fac6d (ceph): dout: don't print version when forcing fg logging.
- dout: don't print version when forcing fg logging.
Fix tests that were assuming us to spew errors about /var/log, wh... - 05:19 PM Revision e86e4cf8 (ceph): init-ceph: use do_cmd for pid_file dir creation
- Signed-off-by: Sage Weil <sage@newdream.net>
- 05:03 PM Bug #802: osd: failing to send heartbeats (btrfs hang?)
- I think this is the same thing Jim is seeing.
- 04:57 PM Revision e4ecddd4 (ceph): dout initialization: remove unecessary flush()
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 04:22 PM Revision e4054d2f (ceph): dout: properly output ceph version on opening dout
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 04:05 PM Revision 5676253c (ceph): monmaptool: set_foreground_logging
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 04:02 PM rbd Bug #809 (Resolved): listing snapshots when none exist causes qemu to crash
- I'm guessing this is just a null pointer dereference. To reproduce:
* Create an rbd-backed vm.
* Run it in qemu. ... - 03:16 PM Revision 03e56b77 (ceph): msgr: complain if there are > 1 threads, not 1
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 01:11 PM rbd Feature #808 (Resolved): rbd: store snapshot timestamp and vmclock
- In the qemu monitor, 'info snapshots' will list the date and vmclock of each, in addition to snapshot id and tag.
- 09:16 AM Bug #806 (Resolved): init-ceph should use do_cmd consistently, or not at all
- this bug fixed by commit:e86e4cf
the ssh stuff kicks in only with -a. it's weird, yes, but incredibly convenient.... - 03:32 AM Revision 3b03c5a3 (ceph): osd: fix population on unconnected_watchers on obc load
- Fixes: #807
Signed-off-by: Sage Weil <sage@newdream.net>
02/15/2011
- 11:24 PM Revision b0862e32 (ceph): test/osd: Fix indentation on RadosModel.h and TestSnaps.cc
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 09:57 PM Revision 28bb6fb5 (ceph): testsnaps: add snapshot test
- Uses RadosModel.h to check the results of a randomized sequence of
writes, reads, snapshots, snapshot removals, and r... - 07:28 PM Bug #807 (Resolved): unconnected_watchers assert fail
- The (violated) invariant is that any item in oi.watchers must be either in obc->watchers or obc->unconnected_watchers...
- 04:47 PM Bug #807: unconnected_watchers assert fail
- Here's roughly what I did that lead to this (I'm not sure exactly when cosd crashed):
- imported two images using rb... - 04:18 PM Bug #807 (Resolved): unconnected_watchers assert fail
- cosd crashed when running qemu off an rbd image calles kvm213_root:...
- 07:02 PM Revision 6fb416b0 (ceph): common: thread: get number of threads from /proc
- The kernel knows how many threads we have; just ask it. One less atomic
variable to carry around.
We will eventually... - 05:52 PM Revision fec24834 (ceph): Remove ExportControl (we have better auth now)
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 05:16 PM Revision 26504682 (ceph): init-ceph: status: use daemon_is_running
- daemon_is_running does some nice things like check /proc/$pid/cmdline.
Signed-off-by: Colin McCabe <colin.mccabe@dre... - 05:09 PM Revision a9c330e5 (ceph): OSD: ignore osd_max_write_size if it's set to 0.
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 05:01 PM Revision d496b651 (ceph): Journaler: add some checks for expire_pos.
- I don't think these are necessary checks, but the expire_pos >= trim_pos
invariant got broken somehow by johnl, and t... - 05:01 PM Revision bfe2c3cf (ceph): Journaler: call set_layout after init_headers.
- set_layout modifies last_committed, but then init_headers
uses operator= and overwrites those changes. In this case
i... - 04:39 PM Revision 76667c41 (ceph): init-ceph: fix status for multi-node clusters
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 03:58 PM Bug #806 (Resolved): init-ceph should use do_cmd consistently, or not at all
- init-ceph sometimes uses do_cmd, and sometimes not.
Example:... - 03:39 PM Bug #795 (Resolved): init-ceph assumes only one daemon per node
- fixed by commit:76667c41e4bc242d089656485d02ba9c4f9b4fec
- 01:49 PM Revision 8cd5b6a5 (ceph): Merge branch 'dout2'
- 01:33 PM Revision 31a17c6a (ceph): common/lockdep.cc: don't use dout unlocked
- Lockdep should use the regular dout() interfaces, rather than going
around them. In particular, we shouldn't output t... - 09:38 AM CephFS Bug #805 (Can't reproduce): mds startup: _replay journaler got error -22, aborting
- Well, I can't figure out how the header could have gotten corrupted like that, but I've put in a few more checks for ...
02/14/2011
- 04:26 PM CephFS Bug #805: mds startup: _replay journaler got error -22, aborting
- Okay, somehow expire < trim. That shouldn't happen.
- 02:29 PM CephFS Bug #805: mds startup: _replay journaler got error -22, aborting
- Dumped and bzipped the journal here: http://johnleach.co.uk/downloads/ceph/805-journal.dump.tbz2
- 01:21 PM CephFS Bug #805: mds startup: _replay journaler got error -22, aborting
- You don't have any other logs from before, do you? At this point there's a problem with the way the MDS journal ended...
- 01:02 PM CephFS Bug #805 (Resolved): mds startup: _replay journaler got error -22, aborting
- As per #803, the assert is now fixed but starting up the cluster I now get:...
- 03:20 PM Revision 16cd9173 (ceph): assert: allow assertions inside calls to dout()
- We should handle the situation where we assert() while already holding
the dout() lock. At the same time, we want to ... - 01:56 PM Revision 37361926 (ceph): dout: Convert _dout_lock to plain pthread_mutex_t
- Convert _dout_lock to plain pthread_mutex_t. This way, we don't have to
depend on the order of global constructor ini... - 01:52 PM Revision 1543d364 (ceph): assert.cc: some cleanup
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 01:52 PM Revision dd21803a (ceph): common/debug.h: use std::string rather than string
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 01:51 PM Revision 65bb14f0 (ceph): common: Remove common/tls.cc
- Using ELF TLS via the __thread keyword is much faster than using
pthread_getspecific and pthread_setspecific. It's al... - 12:17 PM Revision db188f0c (ceph): dout: use DoutLocker rather than Mutex::Locker
- Use DoutLocker rather than Mutex::Locker, in preparation for making the
dout_lock a plain old pthread_mutex_t.
Signe... - 12:17 PM Revision a7b8e9c6 (ceph): os/FileStore: use derr/dendl for dout locking
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 12:05 PM Revision e7b584f0 (ceph): os/FileStore: use ceph_abort rather than abort
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 11:30 AM Revision 97809ee3 (ceph): lockdep: balance dout and dendl, fix whitespace
- Make lockdep use dendl the same way as the other code. This is in
preparation for making lockdep use normal dout() ra... - 10:41 AM Bug #777 (Resolved): mount hung, tid timed out messages in log
- All right, looks like this is good!
- 09:23 AM Bug #698: cosd memory usage with large number of pools
- It seems the messages stacking up are just a result of massif slowing down operations too much.
According to massi... - 09:16 AM Linux kernel client Bug #804: read performance slow
- In general read performance is a little bit slower than write performance, for reasons we haven't managed to diagnose...
- 03:50 AM Linux kernel client Bug #804 (Resolved): read performance slow
- Ceph cluster is configured and running with 64 OSD’s daemon( 64 hard disk ), 1 Metadata servers(MDS) an...
02/13/2011
- 10:16 PM Revision 4ee35b03 (ceph): Journaler: fix bad assert.
- We can call reread_head during normal replay under
certain circumstances. So add the REREAD_HEAD state
as allowed.
S... - 02:17 PM Bug #803 (Resolved): mds assert failed replaying journal after respawn
- This was just a bad assert missing an allowed case. Looks like this got hit while going through error-handling code, ...
- 12:49 PM Bug #803 (Resolved): mds assert failed replaying journal after respawn
- ...
- 08:57 AM Bug #777: mount hung, tid timed out messages in log
- Tested with a build of master and it looks good. I'm writing files to a dir atm and am up to 430,000 entries and stil...
- 04:52 AM Messengers Bug #798: should have a empty() test on list<Messegger *>
- a exception is throwed out when pop_front()
------gdb msg--------------------------
-----------------
[root@mods2 ...
02/12/2011
- 09:23 PM Revision da696695 (ceph): testlibrbd: fix printf args
- Stupid me!
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 07:02 PM Revision 4f4306e2 (ceph): common: create dout_emergency interface and use it
- Create the dout_emergency interface, which is safe to call from a signal
handler or from inside dout itself.
Signed-... - 07:06 AM Revision 81beac55 (ceph): testlibrbd: check return values
- Stupid printf!
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 07:06 AM Revision 6a800643 (ceph): cfuse: use safe_read and check return value
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 06:47 AM Revision a930c7af (ceph): debian: add python, python-dev build-deps
- Might be overkill? The error I see from pbuilder is
checking for a Python interpreter with version >= 2.4... none
e... - 06:41 AM Revision 940d58d9 (ceph): msgr: clean up Pipe::queue_received locking
- Ensure we maintain the invariant that a pipe has a non-empty queue IFF
the pipe is queued.
Prompted by #798.
Signed... - 01:03 AM Revision 12e14f70 (ceph): MDCache: add max_dir_commit_size.
- Configured by setting mds_dir_max_commit_size in conf, or else
by looking at osd_max_write_size. This should lead to ... - 01:03 AM Revision 3129581e (ceph): MDCache: switch CDir::_commit so that it can limit max write size.
- This should fix #777.
Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com> - 12:58 AM Revision 25c71ea2 (ceph): MDCache: switch CDir::_commit so that it can limit max write size.
- This should fix #777.
Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com> - 12:16 AM Revision fa4a9230 (ceph): MDS: Don't always _commit_full just because we have a complete dir.
- Instead, commit if a certain percentage of the dentries are dirty.
Configurable via mds_dir_commit_ratio!
Signed-off...
02/11/2011
- 11:54 PM Revision 4328a550 (ceph): MDCache: add max_dir_commit_size.
- Configured by setting mds_dir_max_commit_size in conf, or else
by looking at osd_max_write_size. This should lead to ... - 11:38 PM Revision b7de2fd0 (ceph): .gitignore: py-compile
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 10:36 PM Revision cd859923 (ceph): librados: include headers in dist tarball
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 10:09 PM Bug #799: mkcephfs error
- 1. g++ with CXXFLAGS='-g -pg' FFLASGS='-g' CFLAGS='-g'
2. ceph.conf ,open all debug options at 20 level
so cconf w... - 09:09 PM Bug #799: mkcephfs error
- Which version are you running? What is your ceph.conf? What arguments did you pass to mkcephfs?
- 09:56 PM Revision 74c4d867 (ceph): MDS: Don't always _commit_full just because we have a complete dir.
- Instead, commit if a certain percentage of the dentries are dirty.
Configurable via mds_dir_commit_ratio!
Signed-off... - 09:50 PM Revision e21034e1 (ceph): Make git ignore librbd tests.
- 09:48 PM Feature #801: librados: allow access to multiple clusters
- And while we're breaking backward compatibility, the read and write functions have some strange parameters ordering: ...
- 09:08 PM Feature #801: librados: allow access to multiple clusters
- See librados_api branch.
BTW if we do end up changing this, we should be consistent with naming... some functions ... - 02:46 PM Feature #801 (Resolved): librados: allow access to multiple clusters
- the current api assumes you are only talking to a single cluster. we need to have a 'cluster handle' returned by the...
- 09:48 PM Revision 0c97d790 (ceph): ceph.spec.in: add librbd
- 09:42 PM Revision 27a7b5a7 (ceph): ceph: add --concise flag
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 09:41 PM Revision 1c320ebe (ceph): Merge branch 'stable'
- Conflicts:
configure.ac - 09:29 PM Revision 7fea444b (ceph): Merge remote branch 'origin/librbd'
- 09:21 PM Revision b7d1d052 (ceph): objecter: set linger op target pg when a linger is resent
- send_linger always creates a new Op, but op_submit does not fill in
the target pg if an existing session is passed in... - 09:05 PM Messengers Bug #798 (In Progress): should have a empty() test on list<Messegger *>
- Hi Longguang,
I took a look at this and think it's okay. Basically the invariant is that the pipe should be in the ... - 07:00 PM Messengers Bug #798: should have a empty() test on list<Messegger *>
- void SimpleMessenger::dispatch_entry(){
...
pipe->pipe_lock.Lock();
list<Message *>& m_queue = pipe->in_q[priority... - 08:58 PM Revision 718c42b9 (ceph): Make git ignore qa/workunits/direct_io_test.
- 08:55 PM Revision 09c019bb (ceph): monclient: fix weirdness when we explicitly start with non-master mon
- Fix
$ ./ceph -m 10.0.1.252:6790 health
2011-02-11 12:30:38.132728 mon <- [health]
2011-02-11 12:30:38.133243 mon <- ... - 07:10 PM Revision 7e1ebd7a (ceph): Disable lockdep for ExportControl, ConfFile locks
- Currently, we haven't read the configuration at the time we initialize
these locks. So we can't know whether lockdep ... - 04:51 PM Bug #777: mount hung, tid timed out messages in log
- Sage Weil wrote:
> And after thinking about it a bit, Greg, _commit_full should be pretty easily fixed up to behave ... - 04:49 PM Tasks #796 (Resolved): Let CDir::_commit_full write in pieces
- Pushed into master in commit: 3129581e89cb38f9ad2ff85673e50a033ec77d40
- 04:12 PM Tasks #796: Let CDir::_commit_full write in pieces
- Pushed it into the max_commit_size branch. Testing it now.
- 04:48 PM Feature #709: nagios plugin
- The nagios machines are still etch so I think I'm going to push to get them updated to lenny before proceeding with t...
- 02:40 PM Feature #709: nagios plugin
- fwiw i've been creating a 'logical cluster' machine in teh dh db. see ceph-playground-fs, for example. although the ...
- 02:37 PM Feature #709 (In Progress): nagios plugin
- I've started on this. I think it will look something like this...
A new "cephmonitor" service entry on a dummy ma... - 04:04 PM CephFS Tasks #797 (Resolved): Don't _commit_full just because dir is_complete()
- This got pushed in commit: fa4a9230c6ba68b9b66d1560abb89114caedf74b
- 03:57 PM Bug #802 (Closed): osd: failing to send heartbeats (btrfs hang?)
- mkcephfs -c /etc/ceph/ceph.conf --allhosts -k /etc/ceph/keyring.bin --mkbtrfs
service ceph -a start
watch osds flap... - 03:50 PM Revision cde53e95 (ceph): Merge branch 'pybind'
- 02:35 PM rbd Bug #391: snap create/delete caused corruption
- Using ext3freezer on the guest during the snapshotting doesn't help either — as far as I can tell, simply taking/remo...
- 11:48 AM rbd Bug #391: snap create/delete caused corruption
- Managed to reproduce this by running this script (to create and delete snapshots) on the host:...
- 11:33 AM rbd Bug #391 (In Progress): snap create/delete caused corruption
- reopening
- 02:00 PM Revision f70d9047 (ceph): config: Remove g_conf.num_osd, num_mds, num_mon
- These configuration options don't serve any use and were confusing.
Signed-off-by: Colin McCabe <colin.mccabe@dreamh... - 12:23 PM rgw Tasks #774 (Rejected): move to radosgw.git
- 12:03 PM Revision 8a1906a0 (ceph): MonitorStore: check for ENOENT, not EEXIST
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 11:51 AM Revision a350296d (ceph): common: bufferlist::read_file: use safe_read
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 11:31 AM Revision d4312952 (ceph): auth/Crypto.cc: use safe_read_exact
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 10:55 AM Revision 22dece17 (ceph): os/FileStore: remove unused variable
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
02/10/2011
- 11:43 PM Revision 12a47f56 (ceph): s3.sh: simple bug fix
- 11:41 PM Revision 1021f5e2 (ceph): rados: add create op
- 08:44 PM rgw Bug #768: eliminate any read/modify/write races
- That should be accomplished using a new class: the 'rgw' class. This, however, won't fix the problem when using a dif...
- 08:42 PM rgw Bug #766: slow bucket creation times
- It's confirmed. Bucket creation time can be slow. There is a problem with osd having to digest the new osdmap after e...
- 08:22 PM Feature #800 (Rejected): pool namespace
- Pools should have namespaces.
- 06:24 PM Bug #799 (Closed): mkcephfs error
- [root@localhost ceph]# mkcephfs -c /etc/ceph/ceph.conf --allhosts -k /etc/ceph/adminkeyring.bin
usage: cconf <-c fi... - 06:02 PM Messengers Bug #798 (Can't reproduce): should have a empty() test on list<Messegger *>
- void SimpleMessenger::dispatch_entry(){
...
pipe->pipe_lock.Lock();
list<Message *>& m_queue = pipe->in_q[pr... - 05:49 PM Revision b60444b5 (ceph): make:add messages/MOSDRepScrub.h to NOINST_HEADERS
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 05:49 PM Revision 2cd2c56d (ceph): v0.24.3
- 05:16 PM Revision e5fbf5ec (ceph): os/FileStore: fix error handling
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 05:16 PM Revision c7886ba5 (ceph): common/common_init: use safe_read
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 05:16 PM Revision 42b735c4 (ceph): rados.cc: use fwrite to output data to stdout
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 05:16 PM Revision 378ba0a1 (ceph): mon/MonitorStore.cc: use safe_read
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 05:16 PM Revision c695a61f (ceph): cephtool: get_indata should use safe_read_exact
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 05:01 PM Revision 500edf7a (ceph): rbd: fix signed/unsigned comp on i386
- We -D_FILE_OFFSET_BITS=64, so off_t is off64_t.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 04:53 PM Revision a476cceb (ceph): Removing files that look completely unused.
- 04:52 PM Revision 8f43d4e7 (ceph): osd: avoid message leak in error path
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 04:24 PM Revision 5d3a129f (ceph): config: fix -f option settings.
- We don't want to force foreground logging just because we run in the
foreground -- use no-daemon for that!
Signed-of... - 04:00 PM Revision ff85407b (ceph): common/DoutStreambuf: use safe_io.h
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 03:09 PM Revision 2aabb909 (ceph): tools/common: use fwrite to write stdout
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 03:08 PM Revision 70a8b625 (ceph): osd: write_meta/read_meta: fix error handling
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 02:49 PM Revision 59874008 (ceph): msgr: better error checking for write_pid_file
- Use safe_write. Check error code of close.
Fix what looks like a "lack of braces around multi-statement block"
error... - 02:47 PM Revision c3f0f09d (ceph): common/buffer.cc: kill deadcode
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 02:47 PM Revision 635c535c (ceph): os/FileStore.cc: use safe_write
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 02:47 PM Revision 0d4244d8 (ceph): common/ConfUtils.cc: use safe_write
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 02:21 PM Revision d2491941 (ceph): MonitorStore::put_int: handle I/O errors
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 02:02 PM Revision 56943deb (ceph): common/safe_io.h: warn about unchecked returns
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 01:55 PM Revision 4a4c2eeb (ceph): mds/Dumper: use safe_write
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 01:55 PM Revision e8ac5aa2 (ceph): cfuse.cc: use safe_write
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 01:55 PM Revision ea80a066 (ceph): btrfs_ioc_test.c: use safe_write
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 01:49 PM Revision cd2a2d73 (ceph): test_disk_bw.cc: use safe_write
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 01:47 PM Revision 4de7fd49 (ceph): common: safe_io: allow C files to include safe_io
- Don't use extern C unless we're compiling with C++.
This allows C files to include safe_io.h
Signed-off-by: Colin Mc... - 01:14 PM Revision 84b90892 (ceph): client/fuse_ll: use safe_write
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 12:58 PM Revision 380ccfd7 (ceph): common: safe_io: handle EOF on read
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 12:58 PM Revision a743e314 (ceph): common:safe_io: create exact and non-exact variant
- Create safe_read_exact, which fails with -EDOM unless it reads exactly
the requested number of bytes. More often, wha... - 12:58 PM Revision b161e6a3 (ceph): os: convert some read/write uses to safe variants
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 11:57 AM CephFS Tasks #797 (Resolved): Don't _commit_full just because dir is_complete()
- We should use some kind of metric to see if we should commit_partial or commit_full!
- 11:57 AM Tasks #796 (Resolved): Let CDir::_commit_full write in pieces
- CDir::_commit_full attempts to write out the whole directory to the PG, no matter how large. It doesn't need to do th...
- 09:10 AM Bug #462 (Resolved): cephx: verify_authorizer_reply exception in decode_decrypt
- 08:38 AM Bug #698: cosd memory usage with large number of pools
- I ran the OSD through massif and in one of my tests, creating 2000 pools (with default 8 PGs), 1/3 of the ending memo...
- 08:09 AM Bug #794 (Resolved): massif is unusable on current code
- This seems to have been because of a bad setting in config for the foreground logging option. Simple enough to fix!
... - 12:19 AM Revision af755be9 (ceph): filestore: fix fiemap
- 12:19 AM Revision 65560eff (ceph): testlibrbd: change strcmp to memcmp, other stuff
02/09/2011
- 10:30 PM Revision 82f5989d (ceph): librbd: use watch/notify internally to keep ImageCtx updated
- 09:45 PM Bug #777 (In Progress): mount hung, tid timed out messages in log
- Greg Farnum wrote:
> It's not exactly a duplicate, but #594 is the latest open bug on fragment split/merge. When tha... - 04:50 PM Bug #777 (Duplicate): mount hung, tid timed out messages in log
- It's not exactly a duplicate, but #594 is the latest open bug on fragment split/merge. When that's done and fragmenta...
- 04:02 PM Bug #777: mount hung, tid timed out messages in log
- .....oh. So you're trying to create a million files in a single directory?
Ceph stores inodes in the directory the... - 03:45 PM Bug #777: mount hung, tid timed out messages in log
- no ooms (the boxes are 6gig ram) and now I'm relatively sure the mds is not crashing, it seems to be an orderly respa...
- 03:41 PM Bug #777: mount hung, tid timed out messages in log
- to be clear, this is not resulting in the "tid timed out" messages now, just the journal hangs, so it might be a sepa...
- 03:38 PM Bug #777: mount hung, tid timed out messages in log
- got a debug log from the mds now. Looks like the mds decides to commit it's data to the osds and something goes wrong...
- 03:36 PM Bug #777: mount hung, tid timed out messages in log
- Do you have any core dumps from the MDS? Or is this running in a memory-constrained system so it's getting OOM-killed?
- 03:32 PM Bug #777: mount hung, tid timed out messages in log
- I can reproduce this:
* format new cluster (in my case: 3 mons, 2 mds, 4 osds)
* mount filesystem using kernel cl... - 03:08 PM Bug #777 (In Progress): mount hung, tid timed out messages in log
- I don't think the MDS can do that (and if it can, we need to put blocks in place to prevent that). I guess I'll need ...
- 03:04 PM Bug #777: mount hung, tid timed out messages in log
- Only the kernel client was in use, so am assuming it was the mds that made a write bigger than the journal size (120m...
- 08:39 PM Revision e26b194a (ceph): librbd: fully specify std::string in list function
- 08:39 PM Revision ec01b300 (ceph): librbd: add C++ interface tests
- 08:05 PM Linux kernel client Bug #750 (Won't Fix): run "dd", printk " libceph: osd1 172.16.10.68:6805 socket closed"
- I imagine this has the same cause as #742. Please re-open if you manage to reproduce while using btrfs for the cosd s...
- 07:49 PM Linux kernel client Bug #750: run "dd", printk " libceph: osd1 172.16.10.68:6805 socket closed"
hi ,
1.
i git ceph-client ,checkout unstable (commit 9c01177349b435186025a088f612a6f5ce2f3de9)
make menuconfig...- 08:04 PM Linux kernel client Bug #742 (Won't Fix): ceph-client.git unstable , tiotest ,printk "timeout, ceph: mds0 caps stale"...
- This is an unfortunate result of running the kernel client on the same machine as the OSD:
The cosd process needs to... - 07:56 PM Linux kernel client Bug #742: ceph-client.git unstable , tiotest ,printk "timeout, ceph: mds0 caps stale",then oops
- ceph-client unstable:
kernel hacking:
select:
[] RCU debugging: sparse-based checks for pointer usage
< > tor... - 07:55 PM Linux kernel client Bug #742: ceph-client.git unstable , tiotest ,printk "timeout, ceph: mds0 caps stale",then oops
- Hi samuel,
osds are running on ext4 .
the steps:
1.
git ceph-client ,checkout unstable (commit 9c01177349b435... - 07:28 PM Revision f7475de2 (ceph): dout: don't print time to syslog
- Syslog itself prepends a time to messages, so don't duplicate that.
Use a faster snprintf-based method to generate th... - 06:28 PM Bug #792 (Resolved): dout: don't output the date to syslog
- implemented in f7475de2f672a39afda430af04a6f8557859abf5
- 10:01 AM Bug #792 (Resolved): dout: don't output the date to syslog
- Dout shouldn't output the date to syslog. Syslog takes care of that itself, and outputting the date twice is a waste....
- 06:07 PM Revision 820d1a78 (ceph): On i386, pthread ids are 8 hexdigits. Adjust cli tests.
- 05:49 PM Revision 061b8d97 (ceph): pybind/rados: expand on error handling a bit
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 05:40 PM Revision 8284b6b0 (ceph): Add python bindings for rados
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 02:48 PM Revision e45fd9f2 (ceph): Merge branch 'fix32'
- 02:39 PM Revision fe298fc3 (ceph): msg/SimpleMessenger.c: fix signed/unsigned compare
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 02:32 PM Revision 76fd853f (ceph): os/FileStore: fix signed/unsigned comparison warn
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 02:28 PM Revision af77fa4f (ceph): os/FileJournal: compare like types in MAX()
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 02:20 PM Bug #795 (Resolved): init-ceph assumes only one daemon per node
- The init-ceph script (usually installed as /etc/init.d/ceph) assumes that there is only one daemon running on the nod...
- 01:33 PM Bug #794 (Resolved): massif is unusable on current code
- When starting up valgrind stuff via vstart, it now logs everything to the console. I presume this is because of the c...
- 01:32 PM Revision 2ca337d0 (ceph): tools/gui.cc: fix signed/unsigned comparison
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 10:49 AM Bug #698 (In Progress): cosd memory usage with large number of pools
- Taking a look now!
- 10:21 AM Bug #759 (Resolved): osd: pgs spend a long time peering when marking osds out
- 10:18 AM Bug #759: osd: pgs spend a long time peering when marking osds out
- 1a01e5ee1b88a217547873296e0371858be13f37 merged in a branch moving replica scrubbing to rep_scrub_wq with a new non-o...
- 10:21 AM Bug #793 (Resolved): osd: avoid blocking in scrub_wq
- 10:05 AM Tasks #584: do throughput scaling tests on sepia
- I don't remember: did we actually finish this and just not update the bug? If not we should decide what to do with it!
- 10:01 AM Bug #644: rsync can be sloooow
- Well, Sage's above patch makes it some better, but there are still issues with losing the I_COMPLETE flag due to pree...
- 12:22 AM Revision 1a01e5ee (ceph): Merge branch 'stable'
- Conflicts:
src/osd/OSD.cc - 12:22 AM Revision ec9d14c1 (ceph): Merge remote branch 'origin/rep_scrub_wq' into stable
Also available in: Atom