Activity
From 07/03/2011 to 08/01/2011
08/01/2011
- 11:18 PM Revision 68a62911 (ceph): testradospp: add version tests
- get_last_version
assert_version
assert_src_version
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 11:18 PM Revision a6d697e1 (ceph): osd: add ASSERT_SRC_VERSION operation
- Assert a src object has a particular version. This is analogous to the
ASSERT_VER operation, but operations on a src... - 11:18 PM Revision 3a41aa55 (ceph): objecter: add assert_src_version
- as an ObjectOperation
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 11:18 PM Revision 0df20a05 (ceph): librados: add assert_src_version
- Like set_assert_version, this is an IoCtx operation that affects the next
(and only the next) operation we perform.
... - 11:16 PM Revision 09059b73 (ceph): Merge branch 'stable'
- 11:16 PM Revision dcaca3e3 (ceph): osd: set reply_version for read operations
- This was probably broken by the OSD prepare_transaction refactor a few
months ago. Or it never worked. Adding test ... - 08:35 PM Revision a42a2481 (ceph): Merge remote branch 'origin/objecter-latest-map'
- 08:25 PM Revision c06260f7 (ceph): rgw: quiet down some log messages
- 08:24 PM CephFS Bug #1350 (Closed): mds: dup lookups on scp workload
- jojy's dup lookup issue on ceph-devel.
- 06:42 PM Revision e5dfa3dd (ceph): escape_json_attr: don't escape single quotes
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 06:24 PM Revision 1d3acd5d (ceph): Formatter.cc: use common/escape.h
- * Rename rgw/rgw_escape.h to common/escape.h
* Use escape.h in common/Formatter.cc
Signed-off-by: Colin McCabe <col... - 06:13 PM Revision 95d49de0 (ceph): Makefile.am: always #define __STDC_FORMAT_MACROS
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 06:02 PM CephFS Bug #1349: mds: standby-replay leaks memory
- For instance: (the one on top is the standby; the workload finished)...
- 05:57 PM CephFS Bug #1349: mds: standby-replay leaks memory
- I don't think this is actually a memory leak. On kai I ran a 1-MDS system with standby and did snaptest-2. The standb...
- 03:27 PM CephFS Bug #1349 (Resolved): mds: standby-replay leaks memory
- When i do 'CEPH_NUM_MDS=4 ./vstart.sh -d -n -x -s' and cfuse and 'fsstress -d foo -l 1 -n 1000 -p 10 -v' i see the st...
- 04:50 PM Revision a25fd826 (ceph): Merge branch 'stable'
- 04:49 PM Revision 0dc235a3 (ceph): mds: request attempt comes from fwd count, not retry flag
- 03:29 PM CephFS Bug #1166 (Resolved): mds: standby-replay bound asserts
- Runs as close to completion (about 70%) as I can get before filling my disk with logs or running out of swap (see #13...
- 01:46 PM Feature #1348 (Resolved): osd: categorize object stats
- Current the pg/pool stats sum all objects. We need the ability to categorize objects into different totals to track ...
- 10:59 AM CephFS Bug #1284 (Can't reproduce): client: leaked inode ref with cfuse_workunit_snaps.yaml
- I don't know how many times I've run this, but it's measured in the dozens. Haven't seen it once. I suspect we got wh...
07/30/2011
- 05:11 AM Revision 15c344c1 (ceph): mds: fix create_subtree_map for new dirs
- Currently mkdir foo ; rmdir foo fails because we can't get_subtree_map()
on a new directory that isn't linked in the ... - 05:11 AM Revision 534afabb (ceph): vstart: static mapping of names to ranks
- a always 0, b always 1, etc. makes multi-mds debugging much easier.
Signed-off-by: Sage Weil <sage@newdream.net> - 04:44 AM Revision c08d08ba (ceph): v0.32
07/29/2011
- 11:27 PM Revision 360ab7f4 (ceph): rgw: don't silently ignore bad user/group when setting acl
- 10:38 PM Revision 178c463b (ceph): monclient: add method to retrieve the latest version of a map
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 10:38 PM Revision 6833f2f7 (ceph): objecter: fix error check - error return code is negative
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 10:38 PM Revision 9447ce64 (ceph): objecter: check for updated osdmap when requesting a non-existent pool
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 10:38 PM Revision 94239eca (ceph): objecter: rename POOL_DISAPPEARED to POOL_DNE
- The pool may never have existed.
Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com> - 10:38 PM Revision 9efd3a54 (ceph): mon: add GetVersion message
- This allows clients to determine whether they have the latest
mds, mon, or osd map. This is useful for figuring out i... - 10:14 PM Revision 29158d7d (ceph): mds: fix validation of (slave) request attempts
- Verify that slave requests received are not stale.
Verify that slave replies match the currently processing request.... - 10:14 PM Revision 5ba1dfe2 (ceph): Makefile: include HeartbeatMap.h in dist
- Signed-off-by: Sage Weil <sage@newdream.net>
- 10:14 PM Revision 6ad7dfbc (ceph): mds: identify slave requests with reqid + attempt number
- We need to distinguish between different attempts to process a request, or
else we can get annoying races in the slav... - 09:58 PM Revision 97c3bcb7 (ceph): scatterlock: fix flag assignments.
- Want |= to set a flag, not &=!
Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com> - 09:49 PM Revision 7b574ffc (ceph): osdmap: in json dump, dump out/in, up/down status
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 09:37 PM Revision c775c03d (ceph): rgw: get current utc epoch differently
- beforehand tm.tm_isdst was returning random results which happened
to work correctly most of the time since we're cur... - 08:17 PM Revision aba88f52 (ceph): rgw: init correctly req_state->{bucket, object}
- 06:48 PM Revision 5c194f5d (ceph): rgw: tweak content-md5 handling
- 06:48 PM Revision a4e4c083 (ceph): rgw: fix total time reporting in rgw_admin
- 05:53 PM Linux kernel client Bug #1347 (Can't reproduce): forced unmount kernel bug
- When trying unmounting (umount -f), sometimes I get a kernel trace:
BUG: Dentry ffff8800043c98c0{i=1000001a5aa,n... - 05:35 PM Revision 6ac6f7ab (ceph): teuthology: convert from bzip2 to gzip.
- gzip is much, much faster on large log files. With a 7.7GB client log, gzip
took 2:45 to compress it to 624MB. bzip2 ... - 04:38 PM rgw Feature #1313 (Resolved): rgw: multiple concurrent requests per process
- Seems to be working well.
- 04:37 PM rgw Bug #1315 (Resolved): rgw: Bucket ACL Grant on a Nonexistant User silently fails
- Fixed, commit:360ab7f4edb4ba45085102e4945572048b8bdbae.
- 04:23 PM Revision 86c7260b (ceph): heartbeatmap: fix/clarify the commenting
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 04:17 PM Revision acca584b (ceph): scatterlock: rename scatter_flags -> state_flags
- We want to use this for all the bools, not just the scatter ones.
Signed-off-by: Greg Farnum <gregory.farnum@dreamho... - 04:17 PM Revision 925cb462 (ceph): scatterlock: compress boolean flags into a set of state flags
- While we're at it, unify the naming structure a bit and remove
the unused stale flag.
Signed-off-by: Greg Farnum <gr... - 03:33 PM CephFS Bug #1346 (Resolved): failed Locker.cc assert
- My bad! Fixed by commit:97c3bcb7fd22159ecc450f830d75129c0a44a5e4
- 01:27 PM CephFS Bug #1346 (Resolved): failed Locker.cc assert
- ...
- 02:55 PM Feature #1269 (Resolved): collectd: monitor cluster stats
- I implemented the proposed metrics, as well as these three additional ones:...
- 12:55 PM CephFS Bug #1318: directories disappear across multiple rsyncs
- That's great to hear, that makes a lot more sense. Let me know when you have a log and I'll jump on it.
Thanks! - 12:07 PM CephFS Bug #1318: directories disappear across multiple rsyncs
- Unintentional investigation (still messing with my cluster big-time, couldn't turn much of mds logging on), looks lik...
- 11:04 AM Bug #1345 (Resolved): rgw: removing a bucket logs bad bucket name
- root@peon5752:~# rados -p .log ls | grep 5864
2011-07-28-5864-?j?3?
2011-07-28-5864-yehuda-cvhr8dfmdf9l6sx53focvr-... - 10:59 AM Bug #1344 (Rejected): qa: move all fetched files to a central machine
- Right now our teuthology tasks are fetching stuff from two different github repositories, and I'm about to add a thir...
- 10:43 AM Bug #1343 (Resolved): teuthology: use gzip instead of bzip2
- Done in commit:6ac6f7ab386ba70fef7814572a36b630a6a84f52.
07/28/2011
- 11:42 PM Revision 52194024 (ceph): Makefile: remove from libglobal
- Signed-off-by: Sage Weil <sage@newdream.net>
- 11:31 PM Revision 90ce2f7d (ceph): Add -ltr to libcommon
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 11:28 PM Revision 8524ed5d (ceph): Makefile: -lrt for libglobal.la only
- Debugging linking is a pita.
Signed-off-by: Sage Weil <sage@newdream.net> - 11:26 PM Revision fadb5ae7 (ceph): unittest_bufferlist: change include order
- fixes a build error (int type conflicts) for me on fatty.
Signed-off-by: Sage Weil <sage@newdream.net> - 11:01 PM Revision 61a50128 (ceph): mon: 'ceph mon dump [--format=json]'
- Signed-off-by: Sage Weil <sage@newdream.net>
- 11:01 PM Revision 8fe50b84 (ceph): mds: separate type for gratuitous debug ESubtreeMaps
- Give these a different type so they are not interpreted as subtree
boundaries during replay. Otherwise we break the ... - 11:01 PM Revision f82b6298 (ceph): mds: fix log trimming races
- trim() would iterate over segments. It would take the *p segment, ++p,
then call try_expire(). But the _expired() f... - 11:01 PM Revision 7815237e (ceph): client: open session with all mds targets
- If we have an open session with an mds, we need to have an open session.
The problem is if we, say,
- client has ol... - 11:01 PM Revision 33366653 (ceph): heartbeatmap: we don't care about pthread_t
- Workers don't have to be threads.
Signed-off-by: Sage Weil <sage@newdream.net> - 11:01 PM Revision 70dee896 (ceph): heartbeatmap: unit test
- Signed-off-by: Sage Weil <sage@newdream.net>
- 10:55 PM Revision a3ed402b (ceph): Makefile: fix unittest_ceph_argparse build
- Signed-off-by: Sage Weil <sage@newdream.net>
- 10:17 PM Revision c304c2c6 (ceph): injectargs: complain about unparsed args
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 09:49 PM Revision 498dd537 (ceph): injectargs: print out what is changing
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 09:40 PM Revision 0b66ed2c (ceph): rgw: fix base64 check
- 09:29 PM Revision ef509b31 (ceph): rgw: check content md5 validity when doing auth
- 09:03 PM Revision 07e60616 (ceph): rgw: fix authentication
- 09:03 PM Revision 4a8d8f0e (ceph): rgw: fix date checks
- 07:55 PM Revision f5f6b120 (ceph): mds: Split the CInode::scatter_wanted field in two
- We use this field to indicate we want a scatter or an unscatter. Make
that distinction explicit.
Also, clear the unsc... - 07:55 PM Revision 579f2e92 (ceph): mds: Handle unscatter_wanted in try_eval(lock, need_issue)
- commit:dac1dc83ee5598ca97c29cd5d0b12150685cd05b added handling
for scatter_wanted, but we need to handle unscatter_wa... - 07:55 PM Revision dc4834b6 (ceph): scatterlock: convert [un]scatter_wanted to a bitfield
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 05:43 PM CephFS Bug #1206: NFS reexport file creation lags 1-3 seconds
- actual commit is:
3e9a982a88ed16e5b47136f617193d6056b32ab4
I still need to get the kernel commit.. but they were ... - 02:58 PM CephFS Bug #1206: NFS reexport file creation lags 1-3 seconds
- Looks like that commit-id may be an internal reference (we host a local repo)
This is a recent master.. 0.31 is the ... - 02:51 PM CephFS Bug #1206: NFS reexport file creation lags 1-3 seconds
- I've reproduced this half-second-NFS-create bug (f3de1a506d6c2debb399a1ae71f8f50714a31c8a), and it only appears throu...
- 05:25 PM Revision 277c4ff7 (ceph): set max_mds based on non-standbys
- 05:11 PM Revision 3d9621fe (ceph): heartbeatmap: fix mode
- Signed-off-by: Sage Weil <sage@newdream.net>
- 05:10 PM Revision 8e4a3588 (ceph): heartbeatmap: warn if previous deadline is missed
- This will generate missed deadline noise in the log that may otherwise be
missed by an infrequent heartbeat_interval.... - 04:53 PM Bug #1343 (Resolved): teuthology: use gzip instead of bzip2
- I tested it briefly, and it looks to me like the time savings of gzip are probably worth the loss of space efficiency...
- 04:50 PM Revision a9813336 (ceph): ceph_context: only wake up periodically if heartbeat_interval is set
- Signed-off-by: Sage Weil <sage@newdream.net>
- 04:49 PM Revision 6eb213d3 (ceph): osd: no need to explicitly check health
- The service thread does it now.
Signed-off-by: Sage Weil <sage@newdream.net> - 04:49 PM Revision 06ebba7f (ceph): mds: pin inodes on LogSegment::truncating_inodes list
- For active MDS, pin when we add to the list, unpin when we finish
truncating.
For replay, pin when we replay a trunc... - 04:49 PM Revision cc82707f (ceph): mds: mark ambig imports in ESubtreeMap during resolve
- During resolve we may journal EImportFinish(true/false) as we resolve our
imports/exports. And as a side-effect we m... - 04:49 PM Revision bdfccb09 (ceph): heartbeatmap: introduce heartbeat_map
- Each thread registered and gets a private structure it can write a timeout
value to. The timeout is time_t and alway... - 04:49 PM Revision d7b45882 (ceph): workqueue: provide op timeout to workqueue constructor
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 04:49 PM Revision 7c5f3bc1 (ceph): workqueue: register and time out worker threads
- Register and unregister worker threads. Periodically touch heartbeat
when idle. Set heartbeat timeout before proces... - 04:49 PM Revision f5db9afb (ceph): heartbeatmap: use a list<> instead of map<>
- Don't need a map<> here.
Signed-off-by: Sage Weil <sage@newdream.net> - 04:49 PM Revision 7aad8f03 (ceph): heartbeatmap: fix stupid race
- atomic_t is probably better here, actually... :/
Signed-off-by: Sage Weil <sage@newdream.net> - 04:49 PM Revision 058647f9 (ceph): heartbeatmap: simplify api
- reset_timeout(), clear_timeout() makes more sense than "touch".
Signed-off-by: Sage Weil <sage@newdream.net> - 04:49 PM Revision e68e4f33 (ceph): heartbeatmap: put in ceph namespace
- Signed-off-by: Sage Weil <sage@newdream.net>
- 04:49 PM Revision 3dfe830e (ceph): heartbeatmap: use atomic_t
- Signed-off-by: Sage Weil <sage@newdream.net>
- 04:49 PM Revision 31d5cbbb (ceph): heartbeatmap: config options, method to touch a file if healthy
- Signed-off-by: Sage Weil <sage@newdream.net>
- 04:49 PM Revision 7265e5cc (ceph): ceph_context: check internal heartbeat in cct service thread
- Signed-off-by: Sage Weil <sage@newdream.net>
- 04:49 PM Revision 9e8bb84e (ceph): vstart: set heartbeat file
- Signed-off-by: Sage Weil <sage@newdream.net>
- 04:49 PM Revision 3bf413dc (ceph): mds: better debugging for scatter_wanted flag
- Signed-off-by: Sage Weil <sage@newdream.net>
- 04:49 PM Revision 1531e88f (ceph): mds: implement try_eval() on a single lock
- We frequently call eval() on locks, usually after dropping an rd/wr/xlock.
At that point the eval() may do nothing be... - 04:49 PM Revision 1bb12c77 (ceph): mds: try_eval in many places
- These are the obvious places where we drop locks and may need to defer the
eval until after unfreeze. There are prob... - 04:49 PM Revision dac1dc83 (ceph): mds: honor scatter_wanted while freezing
- - mds A authpins item on mds B
- mds B starts to freeze tree containing item
- mds A tries wrlock_start on A, sends R... - 04:49 PM Revision a20b15cf (ceph): mds: handle aborted slave rename while waiting for second prep
- When we get the first prep, we may respond to the master with an expanded
list of witnesses for the rename before mak... - 02:52 PM Bug #1316: HAVE_SYNCFS, acconfig.h, #define HAVE_SYNCFS 1
- Sounds like it to me. If you install with /usr prefix does everything behave?
- 01:42 PM Bug #1316: HAVE_SYNCFS, acconfig.h, #define HAVE_SYNCFS 1
- grep does not find anything:
localhost ~ # grep -r syncfs /usr/include
localhost ~ #
OS: Gentoo Base System re... - 01:03 PM Bug #1316 (In Progress): HAVE_SYNCFS, acconfig.h, #define HAVE_SYNCFS 1
- weird. it looks like configure successfully links w/ syncfs, but there's no prototype. maybe your libc and -dev pac...
- 12:27 PM Bug #1316: HAVE_SYNCFS, acconfig.h, #define HAVE_SYNCFS 1
- I added #define _GNU_SOURCE to ./src/common/sync_filesystem.h before #include <unistd.h> on line 35 - but this wont w...
- 10:33 AM Bug #1316 (Resolved): HAVE_SYNCFS, acconfig.h, #define HAVE_SYNCFS 1
- going to assume this is fixed unless i hear otherwise!
- 10:35 AM Feature #1308 (Resolved): teuthology: run standby-replay mds's, and verify they don't crash
- 10:32 AM Feature #1321 (Resolved): osd: touch file if alive
- 09:17 AM Bug #1342 (Resolved): Adjust/fix standby-replay configuration options, handling
- Fixed in commit:9c6ec619f8ad3f88de788cf79de557c1c398274f. There was a minor bug in prepare_beacon.
- 12:42 AM Revision 4b70d583 (ceph): osd: turn down debug level on repop commit message
- We really don't need that to be the only thing sitting in logs.
Signed-off-by: Greg Farnum <gregory.farnum@dreamhost... - 12:42 AM Revision 50c42bcb (ceph): osd: label ReplicatedPG::_scrub as virtual.
- It is virtual in the parent class PG, and the style guide says to
label them in all classes so people don't forget.
... - 12:42 AM Revision 5c37057d (ceph): PG: add an assert for negative entries in the scrub map
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 12:42 AM Revision c4a5380d (ceph): vstart: use paired MDSes with a specified standby.
- I think this is a bit cleaner than specifying ranks manually.
Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com> - 12:42 AM Revision 9c6ec619 (ceph): mdsmon: Fix handling of follow-by-name MDSes.
- We were accidentally setting them to standby-for-rank -1 if their
leader MDS wasn't active on startup. Things worked ... - 12:02 AM Revision 9d4f7414 (ceph): rgw: parse date from http header
07/27/2011
- 11:05 PM Revision 29e07736 (ceph): rgw: return required error when conent length missing on PUT
- 07:43 PM Revision 407467e4 (ceph): Merge branch 'next'
- 07:42 PM Revision 5089cf70 (ceph): mds: make two passes on scatter_nudge
- It's possible for scatter_nudge on a scatterlock in LOCK with dirty set to
go to MIX immediately and remain stable. ... - 07:40 PM Revision d22e48f6 (ceph): mds: honor scatter_wanted flag in scatter_eval()
- We do in file_eval, but not here.
Signed-off-by: Sage Weil <sage@newdream.net> - 07:26 PM Revision 447d4f60 (ceph): testrados_delete_pool_while_open: remove from make
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 06:45 PM Revision 07745f8a (ceph): no ++ in python
- 06:45 PM Revision c220a4f3 (ceph): roles/3-simple: include a standby mds
- 05:57 PM Revision 77dd3b77 (ceph): remove testrados_delete_pool_while_open
- This test duplicates the functionality of
testrados_delete_pools_parallel, but not as well.
Signed-off-by: Colin McC... - 05:04 PM Revision 573c9ff2 (ceph): configure mds's with -s suffix as standby
- 04:16 PM Bug #1342: Adjust/fix standby-replay configuration options, handling
- Okay, after more discussion, we just want it to work properly when you set mds.as to follow mds.a. Right now mds.as w...
- 03:41 PM Bug #1342: Adjust/fix standby-replay configuration options, handling
- Hmm, looking at what happens right now, I think it actually might be adequate for those who read the documentation (s...
- 11:37 AM Bug #1342 (Resolved): Adjust/fix standby-replay configuration options, handling
- What we'd like to see:
rename standby_replay config option to standby_for_any
Make it so that setting mds.a to stan... - 11:26 AM Feature #1269: collectd: monitor cluster stats
- number of osds
number of osds up/down
number of osds in/out
df info : total disk available, used, free
number of ... - 11:06 AM Bug #1339 (Resolved): test threaded operation with librados
- 11:06 AM Bug #1339: test threaded operation with librados
- Although I'm a little surprised, this works fine:...
- 10:42 AM Feature #1218: perfcounter collectd plugin
- testing. See the collectd plugin in ssh://ceph.newdream.net/git/collectd.git
- 09:17 AM Bug #1340 (Rejected): OSD: Strange scrub behavior
- Not A Bug!
Fyodor wrote:
>"Hmmm" I said to myself.
>
>stop osd5
>fsck -fy /dev/sdb1
>
>And got:
>
>Inode 2... - 08:25 AM CephFS Bug #1108: Large number of files in a directory makes things grind to a halt
- Unfortunately I was unable to get any successful results from the test, the cluster crashed after I'd left it copying...
- 05:07 AM Revision 5b092449 (ceph): tolerate named (not numbered) mons
- 05:07 AM Revision 5015b906 (ceph): roles: use letters for mon, mds names
- 05:00 AM Revision 7c0f7c23 (ceph): specify and clean up admin socket
- 03:44 AM Revision 79b8374d (ceph): client: reencode cap releases for each request
- I think commit f7170f9 was based on some of my bad advice. Every time the
client sends a request, it should look at ... - 03:44 AM Revision 699bbe18 (ceph): mds: only create up renamed diri frag subtrees if they differ from parent
- Commit 00ec86a2041 opens up subtrees with CDIR_AUTH_UNDEF blindly for any
renamed dir inode. This is correct on the ... - 03:44 AM Revision 7952beae (ceph): mds: minor cap cleanups
- Signed-off-by: Sage Weil <sage@newdream.net>
- 03:44 AM Revision 854ab3b0 (ceph): mds: track projected rename effect on subtree map
- Renames can effect the subtree map. We don't actually update it until the
rename commits, but while it is in flight ... - 03:44 AM Revision 208e8e34 (ceph): mds: simplify subtree map after adjusting for rename
- Merge the subtree with the parent if appropriate.
Signed-off-by: Sage Weil <sage@newdream.net> - 03:44 AM Revision c7236d9d (ceph): mds: clean out rename subtree cruft
- We used to force these subtrees for rename. We don't anymore.. this is
old weirdness.
Signed-off-by: Sage Weil <sag... - 03:44 AM Revision 56f955b3 (ceph): mds: fix projected rename adjustment
- - we may journal one (or _maybe_ both, probably not) of the subtree root
addition OR the bound addition, depending ...
07/26/2011
- 11:41 PM Revision 2458c896 (ceph): admin_socket: implement schema request
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 11:27 PM Revision 28fa2509 (ceph): perf counters: add back inc type for collectd
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 10:48 PM Revision 68e4d68d (ceph): client/Client.cc: remove superfluous check
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 10:48 PM Revision c9fbef44 (ceph): rgw: get object with range should return 206
- 09:59 PM Revision 03953df2 (ceph): Makefile: sysconfdir for bash_completion install
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 09:54 PM Revision 6fab382a (ceph): rgw: fix bad initialization
- 09:53 PM Revision c08876a1 (ceph): rgw: fix warning
- 09:34 PM Revision 857f18e1 (ceph): rgw: modify print macro to log using default log api
- so that different output fds being used are the same and the log
would actually make sense - 09:34 PM Revision 35bebee8 (ceph): rgw: return error on partial object put
- 09:17 PM Revision dec6ef1d (ceph): bash_completion: no more injectargs command
- Use 'ceph [osd|mds|mon] tell injectargs "foo bar"'.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 09:03 PM Revision f82c254a (ceph): Makefile: add bash_completion
- Signed-off-by: Wido den Hollander <wido@widodh.nl>
- 09:02 PM CephFS Bug #1326 (Resolved): mds: project subtree_map changes due to rename
- 09:02 PM Revision 440382fb (ceph): bash_completion: Add the bash_completion files to the Debian packages
- Signed-off-by: Wido den Hollander <wido@widodh.nl>
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 09:02 PM Revision 0678e51f (ceph): bash_completion: Add bash completion for ceph, rados, radosgw_admin and...
- Signed-off-by: Wido den Hollander <wido@widodh.nl>
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 07:56 PM Revision 3e9a982a (ceph): Add ObsyncException
- This class has subclasses representing temporary and permanent
exceptions, as well as argument parsing errors. An ins... - 07:23 PM Revision b16fa971 (ceph): rgw: return Bad Request on PUT requests with empty content length
- 06:57 PM Revision be8b8264 (ceph): rgw: fix error code returned for bad digest
- 05:09 PM CephFS Bug #1284: client: leaked inode ref with cfuse_workunit_snaps.yaml
- I guess I'll see if I can reproduce this.
- 05:04 PM CephFS Bug #1341 (Won't Fix): Ceph mount hangs when OSD and Ceph Client on the same node
- Unless you're using btrfs or a very new kernel and glibc (which support the new syncfs syscall), this is never going ...
- 04:20 PM CephFS Bug #1341 (Won't Fix): Ceph mount hangs when OSD and Ceph Client on the same node
- Hi guys.
When I tried to run 2 OSD nodes (1-6 OSD, 7-12 OSD) on node1 and node2, (MDS/MON on node3), and run the C... - 04:40 PM rgw Bug #1332: Object creation with invalid Expect non-empty results in 417
- Not sure why we should support this. We can modify apache to ignore that, or tweak it in the proxy.
- 04:24 PM rgw Bug #1333 (Resolved): Object creation with invalid Expect empty results in 500
- This was a bug in mod_fastcgi, commit:f9e7321d1c5a874e7551a735a3b30bc6f84a8ebd fixes that.
- 03:01 PM Bug #1340: OSD: Strange scrub behavior
- Okay, this is just bizarre. There are actually 4 different PGs that report a stat mismatch error. 3 of them (includin...
- 12:19 PM Bug #1340: OSD: Strange scrub behavior
- OSD5 logs.
- 11:06 AM Bug #1340 (Rejected): OSD: Strange scrub behavior
- See http://marc.info/?l=ceph-devel&m=131107339412524&w=2
- 02:41 PM rgw Bug #1331: Object creation with invalid Content-Length above actual results in BotoClientError
- I modified the code to handle this correctly, but alas, we're hitting an apache bug:
https://issues.apache.org/bug... - 01:10 PM Feature #1325 (Resolved): obsync: summary error on exit
- commit:3e9a982a88ed16e5b47136f617193d6056b32ab4 should give us enough to get started with.
Right now, I just made... - 12:58 PM rgw Bug #1329: Object creation with invalid empty Content-Length results in wrong reason
- btw, the actual test still fails since we return also the error as xml, but I think that's not a real error.
- 12:52 PM rgw Bug #1329 (Resolved): Object creation with invalid empty Content-Length results in wrong reason
- 12:52 PM rgw Bug #1329: Object creation with invalid empty Content-Length results in wrong reason
- Fixed, commit:b16fa971a021682370a8af726c574bee5aaa74db.
- 12:54 PM rgw Bug #1330 (Resolved): Object creation with invalid Content-Length below actual results in 500
- This was fixed by commit:be8b82646e3d616827086e4851b46aae53d9550d.
- 12:51 PM rgw Bug #1328 (Resolved): Object creation with invalid zero Content-Length results in 500
- Fixed, commit:be8b82646e3d616827086e4851b46aae53d9550d
- 09:10 AM Bug #1338 (Rejected): osd: scrub should account for clone_range
- Looks like it already does this, actually:...
07/25/2011
- 11:31 PM Revision 7a5e7ff7 (ceph): admin socket: handle schema request
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 08:53 PM Revision c9f4f906 (ceph): sync_filesystem: er, fix #define
- Signed-off-by: Sage Weil <sage@newdream.net>
- 08:49 PM Revision a7520e1f (ceph): sync_files: #include _GNU_SOURCE for syncfs()
- Hopefully fixes: #1316
Signed-off-by: Sage Weil <sage@newdream.net> - 08:48 PM Revision 4a48e7a0 (ceph): admin_socket: no sock for unprivileged daemons; rename default
- Signed-off-by: Sage Weil <sage@newdream.net>
- 08:16 PM Revision a6d9d633 (ceph): objecter: treat RESETSESSION like a reset
- Commit 065cdf5 rewrote ms_handle_reset but didn't adjust
ms_handle_remote_reset (they used to be identical). The res... - 08:16 PM Revision 74fdf705 (ceph): osd: verify src range exists for CLONERANGE
- Make sure the source object extent does not extend past EOF when doing a
CLONERANGE.
Signed-off-by: Sage Weil <sage.... - 07:53 PM Revision 5f437511 (ceph): admin_socket: change request 0 to get version
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 06:19 PM Revision 105a9914 (ceph): Add more debugging for pthread_create failures
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 05:08 PM Revision 6dd5f187 (ceph): use glibc d_type macros, helpers
- Fixes: #1135
Signed-off-by: Sage Weil <sage@newdream.net> - 04:14 PM Revision 2bce766c (ceph): perfcounters: no more stat/ dir, yay!
- Signed-off-by: Sage Weil <sage@newdream.net>
- 03:54 PM Feature #457 (Rejected): osd: alphanumeric names
- 02:06 PM Bug #1339 (Resolved): test threaded operation with librados
- test threaded operation with librados
- 01:54 PM Bug #1338 (Rejected): osd: scrub should account for clone_range
- Right now scrub doesn't account for clone_range. This means that if any of the objects have been cloned, the size sum...
- 01:43 PM Bug #1316: HAVE_SYNCFS, acconfig.h, #define HAVE_SYNCFS 1
- Does it work if you
#define _GNU_SOURCE
#include <unistd.h>
? - 01:33 PM Bug #1298 (Duplicate): osd: memory leak in 3f708ee
- 01:32 PM Bug #1298 (Can't reproduce): osd: memory leak in 3f708ee
- 01:32 PM CephFS Bug #1318: directories disappear across multiple rsyncs
- 09:50 AM CephFS Bug #1318: directories disappear across multiple rsyncs
- Hmm, this sounds like an MDS issue, probably unrelated to the OSD degradation. Is this something you can reproduce?
... - 01:30 PM Bug #1191 (Can't reproduce): FAILED assert(!missing.is_missing(soid))
- 01:26 PM Bug #1334 (Resolved): osd: oi.size mismatch
- fixed by commit:74fdf7053b69a69cb47f33c21398ecd33ec8ad56
- 10:47 AM CephFS Bug #1310 (Can't reproduce): mds crash 'T interval_set<T>::range_start() const [with T = inodeno_t]'
- Yep; the inode table was somehow not being read off disk properly, but we've no idea what's going on, exactly.
- 10:10 AM CephFS Bug #1310: mds crash 'T interval_set<T>::range_start() const [with T = inodeno_t]'
- I looked at this some Friday; I'm afraid that without the entire log or the on-disk state there's little chance of fi...
- 10:14 AM CephFS Bug #1108: Large number of files in a directory makes things grind to a halt
- Did enabling mds frags help?
- 10:12 AM Cleanup #1135 (Resolved): d_type cleanup
- 10:12 AM Cleanup #1135: d_type cleanup
- commit:6dd5f187a9924b2bcd053b5d81bcf9237205887b
- 10:02 AM Feature #1337 (Rejected): osd: support alphanumerically sorted list-objects
- 09:59 AM rgw Feature #769 (Closed): cache pool acls, pool list, etc.
- dup of #1251
- 09:54 AM Feature #1303 (Closed): rgw: make it multithreaded
- dup of #1313
- 09:53 AM Bug #1304 (Can't reproduce): MDS crashes / marked laggy due to getting unknow request op 4361
- 09:33 AM Feature #1327 (Closed): collectd plugin to suck in monitor stats
- duplicates #1269
- 09:32 AM rgw Feature #1336 (Resolved): rgw: precreate pools
- something along the lines of:
- maintain an object with a list of pre-created pools
- user bucket creation should... - 09:31 AM CephFS Bug #1158 (Can't reproduce): Unfinished freeze hangs fsstress
- FWIW I've hit several of these over the past two weeks and they've all boiled down to unstable locks, usually due to ...
- 09:27 AM Feature #1269: collectd: monitor cluster stats
- Wido den Hollander wrote:
> Isn't this related to #685 ?
Yes. There are no libceph bindings, though (at least no... - 05:08 AM Revision d32e5d33 (ceph): mds: be careful about calls to try_subtree_merge
- try_subtree_merge will, on occasion, journal something. And anytime we
journal something we may open a new segment a... - 05:06 AM Revision 1f077a01 (ceph): mds: fix subtree_map simplify, again
- Erasing from the vector<> seems to be clobbering the iterator, even when it
preceeds the deletion point. Use integer...
07/24/2011
- 10:46 PM Revision 0c09377d (ceph): mds: journal EImportFinish after updating subtree/import state
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 10:40 PM Revision 0e6b0dc3 (ceph): mds: journal copious ESubtreeMaps if mds_debug_subtrees
- Journal a new subtree map after every event if subtree debugging is
enabled.
Signed-off-by: Sage Weil <sage.weil@dre... - 10:30 PM Revision f8c7bfab (ceph): mds: fix stl iterator misuse in create_subtree_map
- q might be the first element, so we can't blindly --q; ++q.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 10:29 PM Revision 0b7e9fb6 (ceph): perfcounters: remove stray #includes
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 10:04 PM Revision 937976fc (ceph): admin_socket: fix EINTR check
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 10:04 PM Revision b98d7650 (ceph): admin_socket: fix busy loop
- Passing timeout==0 will return immediately. Pass -1 to block until
interesting happens.
Signed-off-by: Sage Weil <s... - 09:45 PM Revision da689c06 (ceph): mds: kill useless MDLog::writing_subtree_map
- No reason we can't have multiple maps in flight to the journal. This looks
like baggage from when subtree_map_expire... - 09:22 PM Revision 4a40c1ac (ceph): mds: submit_entry last
- MDLog::submit_entry() may journal an ESubtreeMap as a side-effect, so make
sure we have updated our state correctly _... - 09:02 PM Revision c8490aae (ceph): mon: simplify 'mon tell ...'
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 05:13 AM Revision 04e52958 (ceph): mon: rip out injectargs command(s)
- Use 'ceph [osd|mds|mon] tell <who> injectargs ...'
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 05:06 AM Revision 0586d4db (ceph): don't include all the type headers in Client.h
- 04:55 AM Revision 8f360e66 (ceph): client: uninline link/unlink helpers
- finally
- 04:54 AM Revision 757ee66b (ceph): move some types
- 04:47 AM Revision b27fa8a5 (ceph): client: MetaRequest.cc
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 04:46 AM Revision dc1dbdff (ceph): add Inode.cc
- 04:40 AM Revision fad4a633 (ceph): client: InodeCap -> Cap
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 04:39 AM Revision 59098fd3 (ceph): client: move (most) types into separate headers
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
07/23/2011
- 09:10 PM Revision 39453709 (ceph): osd: fix OSDOp printer for CLONERANGE
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 07:30 AM Feature #1335 (Rejected): osd: optionally tolerate per-object attr corruption
- if e.g. oi.size is wrong (#1334) we want to complain, but we don't need to assert on a production box. weaken any si...
- 12:29 AM Bug #1334 (Resolved): osd: oi.size mismatch
- ...
07/22/2011
- 11:25 PM Revision ea5d016f (ceph): mon: 'dump [--format=json]'
- Signed-off-by: Sage Weil <sage@newdream.net>
- 10:10 PM Revision d3887b84 (ceph): auth: fix compile error when using nss libs
- auth/Crypto.cc:218: error: expected primary-expression before ?}? token
auth/Crypto.cc:218: error: expected ?;? befor... - 09:00 PM Revision 94eb4046 (ceph): mds: remove broken delay of cap releases from a replica.
- This hasn't worked in a very long time and serves little purpose
since the clients will have their own cap delay rele... - 08:50 PM Revision c81a8ead (ceph): Revert "mds: Only change in->replica_caps_wanted when actually messaging"
- This reverts commit a2c761e62acdb3cff941867c224ae295cf6337b3. We actually
want to change this whenever we try to send... - 06:45 PM Revision 440de005 (ceph): mds: reorder timing checks in request_inode_file_caps
- We do want to hold onto caps for a few seconds after the client
closes it, just in case it decides to re-open again! ... - 06:36 PM Revision 9e4e6603 (ceph): update documentation on cache expires during exports
- Add an assert to check conditions.
Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com> - 06:36 PM Revision 18314fab (ceph): mds: rename cache expire variables for clarity
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 06:36 PM Revision 92031843 (ceph): mds: fix keep caps test.
- I think this has just been broken since it was written: previously it
dropped the caps if the keep time was after cur... - 06:36 PM Revision a2c761e6 (ceph): mds: Only change in->replica_caps_wanted when actually messaging
- Also, only send the message if the auth is past the REJOIN state.
The auth will just drop it if it gets this request ... - 05:46 PM Revision 3c7318ee (ceph): mon: 'pg dump [--format=json] [all|sum|pools|pgs|osds]'
- Signed-off-by: Sage Weil <sage@newdream.net>
- 05:25 PM Revision 13d68ef4 (ceph): librados: async pool creation
- 05:11 PM Revision f7f9a483 (ceph): ceph: understand -- argument
- Signed-off-by: Sage Weil <sage@newdream.net>
- 04:28 PM rgw Bug #1333 (Resolved): Object creation with invalid Expect empty results in 500
- Amazon creates the object successfully.
- 04:27 PM rgw Bug #1332 (Resolved): Object creation with invalid Expect non-empty results in 417
- Amazon creates the object successfully. Note the 417 is also in html rather than xml.
- 04:25 PM rgw Bug #1331 (Resolved): Object creation with invalid Content-Length above actual results in BotoCli...
- Amazon gives the following:...
- 04:22 PM rgw Bug #1330 (Resolved): Object creation with invalid Content-Length below actual results in 500
- Amazon gives the following:...
- 04:20 PM rgw Bug #1329 (Resolved): Object creation with invalid empty Content-Length results in wrong reason
- Amazon gives the following...
- 04:19 PM rgw Bug #1328 (Resolved): Object creation with invalid zero Content-Length results in 500
- Amazon will give...
- 04:02 PM Feature #1327 (Closed): collectd plugin to suck in monitor stats
- some or all of
ceph pg dump pools --format=json
ceph pg dump osds --format=json
ceph pg dump sum --format=jso... - 03:51 PM Revision ee20a8d9 (ceph): mds: cleanup-let: use is_leased() for lease revocation checks
- And add const-ness.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 03:50 PM Revision 1f479146 (ceph): mds: revoke client leases on replica dn lock
- If a replica dentry gets a LOCK message, revoke client leases. Don't just
wait for them to expire!
Signed-off-by: S... - 03:41 PM Revision 9f8ab9ad (ceph): mds: fix ambiguous check when journaling subtree map
- We journal the EImportStart--and become ambiguous--when we set the state
to IMPORT_LOGGINGSTART; the subtree auth bec... - 03:19 PM Revision 11eecce5 (ceph): client: fix stalls on racing revocations can cap migrations
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 12:48 PM CephFS Bug #1310 (In Progress): mds crash 'T interval_set<T>::range_start() const [with T = inodeno_t]'
- 11:43 AM CephFS Bug #1302 (Resolved): mds: mds_caps_wanted vs migration
- Okay, so this is actually already implemented: The replica will put the message on a waiter if the inode has an ambig...
- 10:29 AM rgw Feature #1251 (Resolved): rgw: consistent caching of bucket objects (acls)
07/21/2011
- 09:57 PM Feature #1271 (Resolved): proflogger: query interface
- 09:28 PM Revision 02b20682 (ceph): osdmaptool: fix clitest
- flags now unconditionally followed by space.
Signed-off-by: Sage Weil <sage@newdream.net> - 08:46 PM Revision 38822401 (ceph): formatter: escape quoted strings
- Signed-off-by: Sage Weil <sage@newdream.net>
- 08:33 PM Revision d6bb5718 (ceph): common: add Formatter class
- This is based on the RGW class, but
- uses a stringstream
- has an additional dump_stream() method that gives you ... - 08:33 PM Revision d4cf3123 (ceph): osd: dump osdmap in json
- Signed-off-by: Sage Weil <sage@newdream.net>
- 08:33 PM Revision b3f7a478 (ceph): formatter: dump float, simplify pending_string stuff
- Signed-off-by: Sage Weil <sage@newdream.net>
- 08:33 PM Revision 5059eb28 (ceph): mon: rewrite pgmap json dump using Formatter
- Signed-off-by: Sage Weil <sage@newdream.net>
- 08:15 PM Revision 56283eb6 (ceph): admin_socket srv:set F_CLOEXEC on listening socket
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 08:10 PM Revision 555714cd (ceph): admin_socket: fix uninitialized accept() param
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 07:31 PM Revision c663d1db (ceph): admin_socket.cc: Use older sizeof syntax
- Fix compilation for compilers that don't support the
sizeof(type::field_name) syntax.
Signed-off-by: Colin McCabe <c... - 07:19 PM Revision 7abaf20c (ceph): rgw: socket path includes full socket name
- 06:05 PM Revision 72fff26e (ceph): perfcounters: don't output trailing commas in JSON
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 04:20 PM Revision 81ef3b5d (ceph): mds: uninline a few things
- Signed-off-by: Sage Weil <sage@newdream.net>
- 04:20 PM Revision 8f5e6646 (ceph): mds: more CInode un-inlining
- Signed-off-by: Sage Weil <sage@newdream.net>
- 04:20 PM Revision e138f23d (ceph): Add line break after add_key() failure message.
- Signed-off-by: Alexandre Oliva <oliva@lsd.ic.unicamp.br>
Signed-off-by: Sage Weil <sage@newdream.net>
07/20/2011
- 11:43 PM Revision 0c6b2c84 (ceph): test/perf_counters: more renaming
- * more renaming
* common/admin_socket_client.cc: use old-fashioned sizeof construct in
case some compilers can't han... - 11:38 PM Revision 7cb1edd9 (ceph): ceph_context: remove set_module_type
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 11:26 PM Revision 15e3cb78 (ceph): Merge branch 'wip-proflogger'
- Conflicts:
src/common/config.cc
src/common/config.h - 11:23 PM Revision a2627f4a (ceph): Client.cc: rename client_logger -> client_counters
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 11:18 PM Revision 1b473329 (ceph): Rename ProfLogger to Perfcounters: part 2
- Part 2: class names, variable names
Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com> - 10:58 PM Revision 387e46b4 (ceph): Filer: return error codes from probing up to the calling layer.
- This is pretty limited; if you get multiple errors in one batch
of probes it'll only return the last one to get sent ... - 10:58 PM Revision 3bb63130 (ceph): mds: handle EBLACKLISTED in MDCache::do_file_recover; assert on errors
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 10:58 PM Revision 1212ceb0 (ceph): Journaler: return errors from probing back to callers.
- Adjust Dumper to handle the error codes. The other callers already do
so!
Signed-off-by: Greg Farnum <gregory.farnum... - 10:58 PM Revision 3d0c314f (ceph): C_Gather: fix comment.
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 10:58 PM Revision 1b0d9ff2 (ceph): Journaler: weaken a few asserts to let it pass errors back up.
- Adjust MDLog to handle the possibility of errors appearing.
Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com> - 10:58 PM Revision 1cf5211e (ceph): Rename ProfLogger to PerfCounters
- Part 1: file names
Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com> - 10:48 PM Revision 9b1ad7b9 (ceph): re-enable proflogger test
- Re-enable the proflogger test. This time, it doesn't test the basic
socket functionality (that is what test/admin_soc... - 10:41 PM Revision ffd92d20 (ceph): rgw: dump is_truncated when listing bucket
- 10:33 PM Revision 8d1193b9 (ceph): test/admin_socket: test message 0
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 09:07 PM Revision b2a38996 (ceph): Move ProfLogger socket code into admin_socket
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 08:22 PM Revision 022c5261 (ceph): uclient: correctly initialize mseq in flush_snaps.
- Previously we set mseq=0 unconditionally; this was a mistake that
creeped in via bitrot. Instead, set mseq from the a... - 08:22 PM Revision b2b44d47 (ceph): mds: Drop locks and auth pins when waiting for freezingtree
- In most cases we don't end up in this branch because there's an escape
if you already have an auth_pin on the ref in ... - 08:22 PM Revision 73d66f67 (ceph): mds: cinode: Add a little debugging and fix an output
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 08:17 PM Revision 4adfd466 (ceph): mds: mark rmdir with slaves as a master in journal
- Signed-off-by: Sage Weil <sage@newdream.net>
- 08:06 PM Revision 573e9720 (ceph): mds: fix rmdir slave commit, rollback
- We need to set the slave commit callback to journal a commit or rollback
as needed.
Signed-off-by: Sage Weil <sage@n... - 07:49 PM Revision 4fe09c1c (ceph): mds: set LogEvent _start_off, drop unused _end_off
- Signed-off-by: Sage Weil <sage@newdream.net>
- 07:49 PM Revision a3e47311 (ceph): mds: add mds_debug_subtrees option
- Verify that replayed ESubtreeMap events match what we have.
Signed-off-by: Sage Weil <sage@newdream.net> - 07:49 PM Revision 8692ed7d (ceph): mds: simplify journaled subtree_map
- We may have subtrees split locally due to migrations that are just getting
started or stopped. Simplify the map we j... - 07:46 PM Revision a25374e4 (ceph): mds: witness rmdir when subtrees are on other hosts
- If there is an rmdir with an empty subtree on another mds, we need to witness/
journal that on the dirfrag's auth mds... - 05:35 PM Revision 0c913a87 (ceph): Merge branch 'wip-rgw-multithreaded'
- 05:35 PM Revision 820b3f36 (ceph): rgw: setup as external server is configurable
- 05:01 PM Revision 9eed12f5 (ceph): rgw: clearer logging context
- 04:31 PM Revision 41f7acf5 (ceph): config.cc: fix comments
- Fix comments that were reverted to an older version by commit
a51b983a8bbafc455bcd2acd993981ece00abcc7.
Signed-off-b... - 04:07 PM Bug #1256: mds dies on ESHUTDOWN under too-high mon load (time-outs?)
- Although I did make some adjustments so hopefully it won't produce a core dump anymore in the latest unstable.
- 01:40 PM Bug #1256 (Won't Fix): mds dies on ESHUTDOWN under too-high mon load (time-outs?)
- ESHUTDOWN is also EBLACKLISTED for the Ceph project. So the MDS is timing out on its heartbeats and the mon is killin...
- 03:56 PM rgw Feature #1313: rgw: multiple concurrent requests per process
- 03:53 PM rgw Feature #1313: rgw: multiple concurrent requests per process
- Implemented as of commit:0c913a87218611a9fc31c55188d63bbd615cd617. Not closing this one yet, as still need to test it...
- 03:56 PM rgw Feature #1251: rgw: consistent caching of bucket objects (acls)
- 03:55 PM rgw Bug #1309 (Resolved): rgw fails to send IsTruncated on 'GET /bucket/' (list keys)
- Fixed, commit:ffd92d20caefb50412f9e3ec0fe708ec880164be.
- 01:29 PM CephFS Bug #1323 (Resolved): cfuse: 2-MDS snaptest2 doesn't complete
- Fixed in commit:b2b44d47c9177fb3d1454a5f973f4f05e37d0e38
It turns out we had an improper set of checks when waiting ... - 01:28 PM CephFS Bug #1324 (Resolved): cfuse: 2-MDS snaptest2 doesn't let client unmount
- Pushed in commit:022c5261800ba6f9642b1f8dc7e44c91b12956eb
- 09:49 AM CephFS Bug #1324: cfuse: 2-MDS snaptest2 doesn't let client unmount
- And it turns out the client is unconditionally setting mseq to 0 and then the MDS is rejecting it if the auth has cha...
- 09:21 AM CephFS Bug #1324: cfuse: 2-MDS snaptest2 doesn't let client unmount
- Okay, that's not correct. Sage says this put() should be happening in handle_cap_flushsnap_ack, which is just never b...
- 08:28 AM CephFS Bug #1324: cfuse: 2-MDS snaptest2 doesn't let client unmount
- queue_cap_snap takes a reference to the inode that's never put by anything as best I can tell. I've added one to the ...
- 01:22 PM CephFS Bug #1326 (Resolved): mds: project subtree_map changes due to rename
- Currently we adjust the subtree map when we commit a rename. That's fine, except that we could
- journal rename o... - 01:21 PM Feature #1277 (Resolved): teuthology: manage known_hosts
- 01:21 PM CephFS Bug #1295 (Resolved): mds: handle rmdir of subtree root
- 01:18 PM Feature #1281 (Resolved): s3-test should test removal of owner's bucket grants behavior
- 12:37 PM Feature #1281: s3-test should test removal of owner's bucket grants behavior
- test_bucket_acl_no_grants written. Amazon allows the owner to still read the bucket, and we match that behavior.
- 10:40 AM Bug #1185 (Can't reproduce): rados: export caught in loop on 'buck' bucket (1.5M objects)
- no luck.
- 09:59 AM Bug #1312 (Closed): testsnaps: incorrect reads
- Bug in test.
- 01:37 AM Revision 4630a0a1 (ceph): lock server: configure for apache with mod_wsgi
- 01:34 AM Revision 629187f3 (ceph): Set content-type with PUT.
- 12:59 AM Revision 019955a1 (ceph): PGMonitor.cc: add force_create_pg command to monitor
- ceph pg force_create_pg <pgid> will now reset that pg's status to
creating to handle corner cases where no osd has th... - 12:25 AM Revision 02d0efad (ceph): schedule: make default owner different from that of a normal run
- This way the machines locked by scheduled jobs aren't confused
with those locked by manual runs, so they're harder to... - 12:13 AM Revision 160174cb (ceph): Update lock db schema.
- 12:13 AM Revision 77174bda (ceph): lock server: allow sshpubkey to be updated
- 12:13 AM Revision 81bebfca (ceph): lock server: return host pubkeys with locked machine names
- 12:13 AM Revision e5f33ca5 (ceph): Add command to update ssh hostkeys.
- 12:13 AM Revision 8d196b00 (ceph): Make targets a dictionary mapping hosts to ssh host keys.
- 12:13 AM Revision 271e066d (ceph): Connect without using any known_hosts files.
- 12:13 AM Revision 731fe410 (ceph): Remove print that clutters the worker logs.
- 12:13 AM Revision 99afde14 (ceph): Update example targets in readme.
- 12:13 AM Revision 00861097 (ceph): Make targets a dictionary mapping hosts to ssh host keys.
07/19/2011
- 09:42 PM Feature #1270 (Closed): qa: tests for splitting pgs vs read/write workload
- this will be an ObjectStore test program that's part of #488
- 09:40 PM CephFS Bug #1195 (Resolved): ceph mds crash on version upgrade
- closing this out
- 09:02 PM Revision 43575c7a (ceph): rgw: fix concurrency issue
- 07:08 PM Revision bd89bd9e (ceph): ProfLogger: add request codes
- The client now starts by sending a 4-byte request code.
Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com> - 06:48 PM Revision 77fc5619 (ceph): ProfLogger: use double quotes, add commas
- Technically, JSON always uses double quotes, even though Javascript
often uses single quotes for strings. Also add a ... - 06:26 PM Revision b3b06d02 (ceph): Merge branch 'stable'
- 06:25 PM Revision 62e6f36a (ceph): mds: clean up file flags to file mode translation
- There was some seriously wrong and ancient cruft in there. open(2)
specifies that one of O_RDONLY, O_WRONLY, and O_R... - 06:08 PM Revision f412942f (ceph): ProfLogger: enable by default for daemons
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 06:04 PM Revision c3fa22fa (ceph): Put each ProfLogger's data into its own object
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 06:03 PM Revision 82faf3d6 (ceph): client: check for invalid file mode
- Signed-off-by: Sage Weil <sage@newdream.net>
- 06:03 PM Revision ff82b60b (ceph): mds: check for invalid file mode
- Signed-off-by: Sage Weil <sage@newdream.net>
- 05:28 PM Revision 5e988b8e (ceph): rgw: fix i386 compilation
- 04:43 PM Revision a51b983a (ceph): config: fix compilation in certain environments
- 03:41 PM Revision 626f0b1f (ceph): mon: fix dump_json quoting
- Signed-off-by: Sage Weil <sage@newdream.net>
- 03:33 PM Revision 650f9f31 (ceph): mds: fix O_APPEND file mode calculation
- Currently
fd = open("aaa",O_RDWR|O_CREAT|O_APPEND,0666);
gives us fmode of 1 instead of 3. We should only infer ... - 03:13 PM Bug #1298: osd: memory leak in 3f708ee
- It might be interesting to attach with gdb and run .capacity() on some of the std::vector elements of class PG.
- 08:31 AM Bug #1298: osd: memory leak in 3f708ee
- If you dump the heap stats on eg osd1 (...
- 02:39 PM CephFS Bug #1323: cfuse: 2-MDS snaptest2 doesn't complete
- The parent directory is being frozen (...on an import, it looks like? Is that possible?) and never finishes its freez...
- 10:37 AM CephFS Bug #1323 (Resolved): cfuse: 2-MDS snaptest2 doesn't complete
- The client sends a request off to the MDS which is never responded to. Logs in kai:~gregf/logs/snaptest2hang
- 11:52 AM Feature #1325 (Resolved): obsync: summary error on exit
- Print a final error message that classifies errors with source vs destination and transient vs permanent. This shoul...
- 11:34 AM CephFS Bug #1324 (Resolved): cfuse: 2-MDS snaptest2 doesn't let client unmount
- On occasions when running snaptests-2 doesn't hit #1323, the client will happily unmount from the filesystem but the ...
- 10:22 AM Feature #1265 (Resolved): osd: test osd recovery clone handling
- 10:21 AM rgw Bug #1287 (Resolved): Setting metadata with unreadable characters is not consistent with amazon S3
- fixed by commit:5cb98c95c001b2a0658a219c717a717bc37e444d
- 09:06 AM Bug #1185: rados: export caught in loop on 'buck' bucket (1.5M objects)
- Still having trouble hitting this. Running in a loop without any debugging to see if I can trigger it.
- 09:03 AM Feature #1320 (Resolved): mon: dump pg (and other?) stats in json
- 12:38 AM Revision 80c20642 (ceph): ProfLogger: don't start threads until it's safe
- Introduce a new configuration variable, internal_safe_to_start_threads.
This will be set by common_init_finish once i... - 12:25 AM Revision 3286048b (ceph): vstart.sh: use out/proflog.$name
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 12:21 AM Revision daeffbd5 (ceph): proflogger: Fix compiler warnings
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 12:18 AM Revision 64d33142 (ceph): proflogger: test multiple profloggers
- Test multiple profloggers, and removing profloggers.
Add a clear function which removes all profloggers.
Make the o... - 12:14 AM Revision 51039da1 (ceph): rgw: use thread pool for rgw work
07/18/2011
- 11:53 PM Revision 48769e22 (ceph): test/profloger: test fl_avg variables
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 11:49 PM Revision 6454ecc9 (ceph): test/proflogger: test JSON formatting a bit
- Test JSON formatting a bit. Use single quotes in our JSON because that
seems to make things easier.
Signed-off-by: C... - 11:33 PM Revision cbeedeb1 (ceph): proflogger: Unlink our UNIX domain sockets on exit
- When we're exiting normally, we ought to unlink our UNIX domain sockets.
Also fix a bug in ProfLogThread, and enable... - 11:31 PM Revision 59a8e65e (ceph): mon: pg dump_json
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 11:23 PM Revision 431df83e (ceph): rgw: radosgw_admin log show dumps parsable json
- 10:52 PM Revision adafec4a (ceph): test/proflogger.cc: read length of message first
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 10:51 PM Revision f8b4aa3c (ceph): ProfLogger: write out length of message first
- Write out the length of the message first, so that it's easier to write
clients. Also, serialize ProfLogger instances... - 10:02 PM Revision 325951dd (ceph): test/proflogger: Add TeardownSetup and SimpleTest
- common/ProfLogger.cc: check the sock_path length.
Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com> - 09:13 PM Revision 134a680a (ceph): Add test/proflogger.cc, fix ProfLogger::init()
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 08:53 PM Revision 5517b8f5 (ceph): Rework ProfLogger
- * Replace existing proflogger config options with "profiling_logger_uri".
This option controls profiling logger sinks... - 08:24 PM Revision 442b5d5b (ceph): clien: rados: bench: remove unused variables
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 08:24 PM Revision 7db65d57 (ceph): tools: remove unused variables
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 08:24 PM Revision 1dedbf63 (ceph): mds: remove unused variables
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 08:24 PM Revision d07c480e (ceph): mon: remove unused variables
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 08:24 PM Revision 64241497 (ceph): osd: remove unused variables
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 08:13 PM Revision 73cf6b89 (ceph): ProfLoggerBuilder: new API
- There are now three different types of variables that can be created:
u64, float, float average.
The distinction bet... - 07:45 PM Revision c54aa7db (ceph): testsnaps: use self managed snaps and other fixes
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 07:45 PM Revision 09c8d12a (ceph): ReplicatedPG: _rollback_to
- _delete_head modifies clone_overlap and stats incorrectly for
_rollback_to
Signed-off-by: Samuel Just <samuel.just@d... - 07:18 PM Revision 35b60c69 (ceph): gitignore: ignore Eclipse files/folders
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 07:18 PM Revision 94c37639 (ceph): Makefile: librados-config apparently depends on libcrypto++
- In newer versions of gcc/ld, librados-config breaks without
the CRYPTO_LIBS.
Signed-off-by: Greg Farnum <gregory.far... - 05:54 PM Revision f71ab2aa (ceph): ProfLogger: don't return val from inc/set/etc.
- Nobody was using those return values.
Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com> - 05:54 PM Revision c767b613 (ceph): Introduce ProfLoggerBuilder
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 05:37 PM Revision 77bcc069 (ceph): Remove lingering references to g_conf
- These were harmless, but silly.
Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com> - 05:07 PM Revision 92df6e32 (ceph): rgw: fix range dump
- 04:33 PM Bug #1322 (Resolved): rgw log show json format should be parsable
- Fixed, commit:140c05fb51abc68bef85331021a7302fdeddd3ef.
- 03:27 PM Bug #1322 (Resolved): rgw log show json format should be parsable
- 02:42 PM Feature #1321 (Resolved): osd: touch file if alive
- 02:41 PM Feature #1320 (Resolved): mon: dump pg (and other?) stats in json
- 12:24 PM Bug #1319 (Resolved): link error with libcrypto++
- I think it's because I'm running gcc 4.4.6 instead of 4.4.4 -- some difference in the behavior of --as-needed that's ...
- 11:29 AM Bug #1319 (Resolved): link error with libcrypto++
- ...
- 09:15 AM Bug #1317 (Won't Fix): deadlocks modifying ceph.ko-mounted filesystem on osd (and mon?)
- Circular syncs are definitely one problem, which syncfs should solve. If you have a sufficiently new libc and kernel ...
- 06:58 AM Bug #1317 (Won't Fix): deadlocks modifying ceph.ko-mounted filesystem on osd (and mon?)
- I've observed frequent system deadlocks while writing lots of data to the kernel-mounted ceph filesystem, which do no...
- 07:07 AM CephFS Bug #1318 (Resolved): directories disappear across multiple rsyncs
- Because of bug 1317, I upload about 1TB of files to my home cluster with one of the 3 osds disabled, so all 3-plicate...
07/17/2011
- 11:49 PM Revision aef29caf (ceph): mds: fix protocol versions
- I screwed this up in a928f767b974cd36f4be6ea618c5cec366d28362.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 05:17 AM Bug #1316 (Rejected): HAVE_SYNCFS, acconfig.h, #define HAVE_SYNCFS 1
- While compile from source, i ran into the folowing Problem:...
07/16/2011
- 12:15 AM Revision ac5c1c41 (ceph): Add an overrides section for the ceph task.
- This lets you run a suite against a particular version of ceph, or
with special debug settings.
07/15/2011
- 11:46 PM Revision a7d791de (ceph): rgw: multithreaded handling
- 10:01 PM Revision d4ed22ff (ceph): recalc_op_target must give an error if pool gone
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 09:44 PM Revision 172dc20e (ceph): Makefile: remove dup testrados_list_parallel_SOURCES
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 09:02 PM Feature #1271: proflogger: query interface
- We've got to be able to come up with a better name than 'profiling logger', especially now that this isn't actually l...
- 08:58 PM Bug #1194: kclient: NFS reexport does not survive ceph fs remount
- I don't have things set up to reproduce/test this easily, but it looks like this is the problem. Can you give it a g...
- 08:03 PM Revision ccf5e266 (ceph): Merge remote branch 'origin/stable'
- 08:01 PM Revision 41c9680a (ceph): Merge remote branch 'origin/wip-1287'
- 07:54 PM Revision 02abdca1 (ceph): rgw: multiple fixes, prepare main loop for multithreading
- 07:45 PM Revision e056686f (ceph): Unit test that connection.connect actually calls create_key.
- 07:40 PM Revision 7587f79c (ceph): Optionally override system host keys.
- 05:58 PM Revision 4f08881d (ceph): rgw: fix compilation for certain architectures
- 05:09 PM Revision b4b3672d (ceph): boto_tool: add --get-obj-xattr
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 05:09 PM Revision 29557834 (ceph): botol_tool: add --xattr argument for PUT
- boto_tool can now set extended attributes when performing a PUT
operation.
Signed-off-by: Colin McCabe <colin.mccabe... - 05:09 PM Revision 5cb98c95 (ceph): rgw_op: mime-encode ugly xattrs in PUT
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 04:47 PM Feature #1300 (Resolved): teuthology: merge config bits
- Done in ac5c1c41dd0819cf487bbc6e4953b0c652b83757 of teuthology.git.
- 10:15 AM Feature #1300: teuthology: merge config bits
- talked with tv and durgin, we decided:
- top level section of overrides for the ceph task specifically
this wi... - 03:38 PM Feature #1277: teuthology: manage known_hosts
- This is all working in the hostkey branch. I'm going to wait to merge it until we have a better machine to run the lo...
- 02:59 PM rgw Bug #1315 (Resolved): rgw: Bucket ACL Grant on a Nonexistant User silently fails
- Look at test_buck_acl_grant_nonexist_user in the @s3tests.git:s3tests/functional/test_s3.py@. Doing an ACL grant for ...
- 10:53 AM rgw Feature #1251: rgw: consistent caching of bucket objects (acls)
- This should include a (stress?) test that gives us some confidence it is working properly. Something like:
- star... - 10:35 AM Feature #1273 (Resolved): test pool creation and pool destruction races
- 10:15 AM Feature #1214 (Resolved): qa: flock/fnctl lock tests
- The extra bits will be tracked in #1314
- 09:08 AM Feature #1214 (In Progress): qa: flock/fnctl lock tests
- Not right now -- I guess I can whip up an executable and a teuthology task for it.
- 08:55 AM Feature #1214: qa: flock/fnctl lock tests
- Does this include something like:
A: start process to take lock
B: start process to try to take lock, block
B: k... - 10:15 AM Bug #1314 (Resolved): teuthology: tasks for file lock behavior on process exit
- In the past we've had bugs where dead processes didn't result in the MDS removing waiting locks. Write a teuthology t...
- 10:12 AM Bug #1257 (Resolved): test/fix concurrency issues with RADOS
- 09:00 AM Bug #1122: kclient: async readahead
- see #804
- 08:57 AM rgw Feature #1313 (Resolved): rgw: multiple concurrent requests per process
- multiple threads, locks
- 08:52 AM CephFS Bug #1168 (Closed): fsstress 2-mds Standby-replay adjust_subtree asserts
- 08:52 AM CephFS Bug #1302: mds: mds_caps_wanted vs migration
- When a client opens a file via an MDS replica, the replica sends the auth a message letting them know which caps are ...
- 08:17 AM Bug #1298: osd: memory leak in 3f708ee
- It occurs to me that there were many degraded PGs, and currently that means the pg logs aren't trimmed and are kept i...
- 01:24 AM Revision ac07e766 (ceph): rgw_common: remove unused variable
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
07/14/2011
- 11:25 PM Revision 91343a83 (ceph): Add control character detection
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 11:25 PM Revision 5f893991 (ceph): mime encoding: encode control chars
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 10:55 PM Revision 5d606c22 (ceph): rgw: fix log
- 10:29 PM Revision 0593001d (ceph): rgw: cleanup, remove globals and static craft
- 09:39 PM Revision 9721e334 (ceph): Better interface for running functions in parallel.
- 08:37 PM Revision 8b1119f7 (ceph): bloom filter: update to latest version
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 08:35 PM Revision 585239af (ceph): filepath: remove asserts
- These were included as a safety check and haven't been triggered, so
there's no point slowing anything down with them... - 07:55 PM Revision 56abfd89 (ceph): honor CINIT_FLAG_NO_DEFAULT_CONFIG_FILE
- Don't use CEPH_CONF_FILE_DEFAULT when CINIT_FLAG_NO_DEFAULT_CONFIG_FILE
is set.
Signed-off-by: Colin McCabe <colin.m... - 07:54 PM Revision d607be8c (ceph): crushtool: don't parse any generic args
- Notably, crushtool uses -c, which elsewhere means parse this config file.
Signed-off-by: Sage Weil <sage@newdream.net> - 07:45 PM Revision 0cfaee69 (ceph): Merge branch 'rgw-sync-cache'
- Conflicts:
src/rgw/rgw_op.cc - 07:43 PM Revision ba31462a (ceph): rgw: configurable way to enable/disable cache
- 06:22 PM Revision 8a8c1e81 (ceph): ReplicatedPG: project changes to clone_overlap
- Previously, changes to clone_overlap were incorrect since make_writeable
is called after do_osd_ops. Now, ctx->modif... - 06:15 PM Revision 9f06f5f4 (ceph): Merge branch 'wip-parallel'
- 05:58 PM Revision 279a6ff6 (ceph): Merge branch 'wip-codingstyle'
- 05:50 PM Revision a9f815b5 (ceph): CodingStyle: whitespace
- Signed-off-by: Sage Weil <sage@newdream.net>
- 05:49 PM Revision bc6eb105 (ceph): CodingStyle: final decisions?
- - _d suffix for naked struct/class types (not _t!)
- m_ prefix for class members
- prefer braces for single line if... - 04:41 PM Feature #1205 (Resolved): qa: run teuthology clients in parallel
- 610f5ab7900e9d9d969b6875b7a0f21b4e9f6ff2 in teuthology.git.
- 04:39 PM Feature #1214 (Resolved): qa: flock/fnctl lock tests
- 04:39 PM Feature #1214: qa: flock/fnctl lock tests
- Okay, having looked into my previous modifications, they just needed a bit of clean up. I gave them some, and wrote a...
- 04:22 PM Bug #1312 (Closed): testsnaps: incorrect reads
- testsnaps reports incorrect reads if it is run while toggling osds in and out. Removing rollback from the op sequenc...
- 04:19 PM Bug #1311 (Closed): qa: TestSnaps: stuck in active
- While running testsnaps and toggling osds in and out, several pgs got stuck in active.
Logs are available at samue... - 11:01 AM Feature #1190 (Resolved): codingstyle document
- 10:06 AM Bug #1304: MDS crashes / marked laggy due to getting unknow request op 4361
- This looks like you're just running an old copy of the server. Is that possible? That opcode has been defined since ...
- 09:19 AM CephFS Bug #1310: mds crash 'T interval_set<T>::range_start() const [with T = inodeno_t]'
- End of log:...
- 01:11 AM CephFS Bug #1310 (Can't reproduce): mds crash 'T interval_set<T>::range_start() const [with T = inodeno_t]'
- I have a reproducible mds crash, I've attached the last 200 lines of the log with debug mds = 20.
- 09:08 AM CephFS Cleanup #1307: client cleanup
- More detail, please?
- 09:08 AM CephFS Bug #1302: mds: mds_caps_wanted vs migration
- More detail, please?
- 04:54 AM Revision 0000b637 (ceph): ceph.conf: remove other random bits
- obsolete sections, mds tuning. stick with defaults.
- 12:07 AM Revision 357db1d0 (ceph): ReplicatedPG: sub_op_push fix
- We need to iterate over the intersection of data_subset and data_needed,
not the original data_needed.
Signed-off-by...
07/13/2011
- 11:47 PM Revision 3d145790 (ceph): Merge remote branch 'origin/wip-1257'
- 11:12 PM Revision 143e2dd5 (ceph): Don't build build tests unless requested
- Build tests (that check if there are unresolved symbols in libraries)
can slow down the build a lot. We should only e... - 11:01 PM Revision 2de0156c (ceph): rados parallel tests: prepend "test"
- Just to avoid any confusion, all these test apps should begin with the
word "test".
Signed-off-by: Colin McCabe <col... - 10:58 PM Revision f4ea3312 (ceph): rados_list_parallel: add test of list-while-modify
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 10:56 PM Revision 2d757ac0 (ceph): rgw: unwatch cache sync object before exiting
- 10:43 PM Revision 8e8bd67c (ceph): testrados: fix
- Change testrados back to a simple single-threaded test. The
multi-threaded tests need to be properly designed to run ... - 10:42 PM Revision b7d12e5e (ceph): systest_runnable: print line when joining runnables
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 10:25 PM Revision a32bc894 (ceph): Create rados_delete_pools_parallel test
- * add rados_delete_pools_parallel test
* change RETURN_IF_NOT_VAL -> RETURN1_IF_NOT_VAL.
We want to return a non-zer... - 10:12 PM Revision a8dc9a01 (ceph): test/osd: TestSnaps now writes large, random objects with holes
- In order to test clone recovery pathways, TestSnaps now writes to random
subsets of an object between snapshots. Obj... - 10:12 PM Revision 5869591b (ceph): ReplicatedPG: fix snaps decode in recover_primary
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 10:12 PM Revision 44ea232a (ceph): encoding.h: need to pass by reference
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 10:12 PM Revision 585fab7c (ceph): ReplicatedPG: data_subset may not contain data_needed
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 09:57 PM Revision f0409c45 (ceph): mds: clear [want_]loner_cap on cap export
- These fields are meaningless on the replica.
Signed-off-by: Sage Weil <sage@newdream.net> - 09:57 PM Revision a928f767 (ceph): mds: migrate loner_cap state
- It is tedious to infer what the old loner_cap was pre-migration. Just send
it over the wire and set it explicitly. Us... - 09:57 PM Revision 314be340 (ceph): mds: improve print() for MInodeFileCaps
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 09:57 PM Revision 7fe5bb53 (ceph): mds: print mds_caps_wanted on inode
- along with the client caps info
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 09:57 PM Revision a078aee1 (ceph): mds: move to MIX state if writer wanted and no wanted loner
- We can just look at the target loner here, which also takes any caps wanted
by other replicas on other MDSs into acco... - 09:57 PM Revision 2539d16a (ceph): mds: print cur+nested auth_pins on inode
- And clean up a bit.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 09:57 PM Revision 1e319612 (ceph): mds: drop useless from field from MInodeFileCaps
- While we're updating the protocol, clean this up too.
Signed-off-by: Sage Weil <sage@newdream.net> - 09:57 PM Revision aec77962 (ceph): mds: fix handle_simple_lock on LOCK replica
- We were ignoring caps. To avoid this kind of mistake, use eval_gather(),
like handle_file_lock() does.
Signed-off-b... - 09:40 PM Revision b9c0a728 (ceph): rados_list_parallel: make num_objects selectable
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 09:40 PM Revision a360b7dd (ceph): Add rerun.sh
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 09:40 PM Revision 6d7c86a4 (ceph): rados_list_parallel: test adding objects + listing
- Test one runnable adding objects while another runnable lists them.
Signed-off-by: Colin McCabe <colin.mccabe@dreamh... - 09:40 PM Revision 0c4f937e (ceph): rados_list_parallel: add test4
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 09:40 PM Revision 9b83072f (ceph): rados_list_parallel: fix rados_ioctx_destroy uses
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 09:40 PM Revision 16bedd2f (ceph): Move list objects logic to st_rados_list_objects
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 09:40 PM Revision 27fbe8be (ceph): Now we can put runnable logs into separate files
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 09:40 PM Revision 2b95a614 (ceph): add CrossProcessSem::reinit, fix rados_list_parallel
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 09:40 PM Revision cccdd43b (ceph): Add rados_open_pools_parallel test
- Also add rados_open_pools_parallel to .gitignore
Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com> - 09:16 PM Revision aaa1a022 (ceph): rgw: clean up temp object if failed to PUT
- Conflicts:
src/rgw/rgw_op.cc - 09:08 PM Revision 4089e4d1 (ceph): rgw: clean up temp object if failed to PUT
- 09:02 PM Revision 3e2c6c6c (ceph): Run workunits on different clients in parallel.
- 09:02 PM Revision b99e33f0 (ceph): Download ceph binaries in parallel.
- 09:02 PM Revision 176b304c (ceph): fusermount runs on a single mount point.
- 09:02 PM Revision a282991b (ceph): Download and run autotests on multiple clients in parallel.
- These clients must still be on different machines,
or they'll clobber each other's results. - 08:52 PM Revision 4b245fce (ceph): Add a utility for running functions in parallel.
- 07:38 PM Revision c88ef9e7 (ceph): Merge branch 'localdir'
- Conflicts:
teuthology/task/ceph.py - 07:34 PM Revision 127ef686 (ceph): Feed locally-created binary tarball to remotes in parallel.
- This should be faster as long as we have the bandwidth for it.
- 07:23 PM Revision 59836044 (ceph): rgw: remove include of curl/types.h
- 07:18 PM Revision aee94422 (ceph): Use a nameless tempfile for local tarball, avoids cleanup.
- 07:07 PM Revision f52df63b (ceph): More careful error checking, avoid need for shell quoting.
- 06:47 PM Revision cb869782 (ceph): test/rgw_escape: add JSON + utf-8 test
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 06:46 PM Revision e8b61257 (ceph): test/rgw_escape.cc: add utf-8 test
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 06:32 PM Revision e7618a67 (ceph): Clean up tarball tmpdir in all cases.
- Prefer shutil.rmtree over os.system('rm -rf ...').
- 06:25 PM Revision f29b9bd7 (ceph): CodingStyle: Typos, whitespace.
- 05:58 PM Revision c701fe50 (ceph): Use tempfile instead of ad hoc temp dir creation.
- 05:46 PM Revision 6bca619f (ceph): rgw: fix escaping clobbering utf8
- 05:45 PM rgw Bug #1309 (Resolved): rgw fails to send IsTruncated on 'GET /bucket/' (list keys)
- Per the S3 API documentation, when the total number of items in the bucket exceeds max-keys (default: 1000), the resp...
- 05:44 PM Revision 69a6b04b (ceph): Remove TODO note covered by teuthology-nuke.
- 05:17 PM Revision 3547eba7 (ceph): Avoid identifier clash with builtin "dir".
- 04:55 PM Revision da1f92c8 (ceph): crushtool: Add call to global_init() to avoid dout segfault
- Signed-off-by: Jim Schutt <jaschut@sandia.gov>
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 04:26 PM CephFS Bug #1170 (Closed): 2-MDS fsstress: SR fails subtree assert
- 04:26 PM CephFS Bug #1169 (Closed): 2-MDS fsstress: Active fails adjust_subtree_auth
- 04:26 PM Revision 8af8ec85 (ceph): Revert "mds: handle importer failure on exporter"
- This reverts commit bf768cb146b23d345efcc51badfe668bf37d4046.
This won't actually happen with f44cffdb8ccc8cf258b418... - 04:21 PM Feature #1308 (Resolved): teuthology: run standby-replay mds's, and verify they don't crash
- 04:20 PM CephFS Bug #1166: mds: standby-replay bound asserts
- yay, the active mds's now pass
fsstress -d foo -l 1 -n 1000 -p 10 -v
with a 4x cluster and 'mds thrash export... - 04:12 PM CephFS Cleanup #1307 (Closed): client cleanup
- 04:10 PM Feature #1272 (Resolved): test rados list-objects in parallel with object modifications
- implemented
- 04:05 PM Revision fa488672 (ceph): obsync: Depend on python-pyxattr instead of python-xattr
- Signed-off-by: Wido den Hollander <wido@widodh.nl>
Signed-off-by: Sage Weil <sage@newdream.net> - 03:53 PM Bug #1275 (Resolved): fix testrados
- fixed by 8e8bd67cddd093a7de690bfd869f40cd7d5b48ee
- 02:24 PM Bug #1306 (Resolved): rgw leaks temp objects if error occurs when doing PUT
- Fixed, commit:aaa1a022756fd01a6f38d7ce3818c6f1db822885.
- 02:06 PM Bug #1306 (Resolved): rgw leaks temp objects if error occurs when doing PUT
- 01:41 PM Feature #1246 (Resolved): teuthology: bundle/install ceph from a local source tree
- 01:37 PM Feature #1305 (Resolved): qa: add a bunch of rbd+ext34 tests to suite
- 01:16 PM Bug #1304: MDS crashes / marked laggy due to getting unknow request op 4361
- what version of cmds are you running? (cmds -v output)
- 01:14 PM Bug #1304 (Can't reproduce): MDS crashes / marked laggy due to getting unknow request op 4361
- I was running some test where ceph is underlying filsystem and slave nodes have corresponding directories under moun...
- 12:43 PM Feature #1303 (Closed): rgw: make it multithreaded
- 12:40 PM rgw Feature #1251: rgw: consistent caching of bucket objects (acls)
- This is done, using watch/notify to sync between the different daemons. However, we should probably make a few change...
- 12:02 PM rgw Bug #1301 (Resolved): rgw doesn't dump unicode characters in xml correctly
- 12:02 PM rgw Bug #1301: rgw doesn't dump unicode characters in xml correctly
- Fixed, commit:6bca619f4bb4b45da3db928a251b462f2fd3626f.
- 09:59 AM rgw Bug #1301 (Resolved): rgw doesn't dump unicode characters in xml correctly
- 10:41 AM CephFS Bug #1302 (Resolved): mds: mds_caps_wanted vs migration
- 09:08 AM Bug #1245 (Resolved): obsync: Issues with python-xattr (AttributeError: 'module' object has no at...
- applied, thanks!
- 06:24 AM Bug #1245: obsync: Issues with python-xattr (AttributeError: 'module' object has no attribute 'set')
- I found out that obsync depends on python-xattr, but it should actually depend on python-pyxattr.
Attached is a pa... - 06:28 AM Feature #1269: collectd: monitor cluster stats
- Isn't this related to #685 ?
I've also written a backend for collectd some time ago where it uses RADOS as a backe...
07/12/2011
- 11:49 PM Revision f44cffdb (ceph): mds: verify deferred messages aren't stale
- We may defer processing of some messages because we are laggy (in hearing
from the monitor). When we eventually get ... - 11:47 PM Revision bf768cb1 (ceph): mds: handle importer failure on exporter
- If the export is reversed but we still receive the ack, behave.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 08:43 PM Feature #1300 (Resolved): teuthology: merge config bits
- Teuthology (the suite in particular) should be able to merge bits of yaml together. For instance, we should be able ...
- 05:13 PM Linux kernel client Bug #912: rbd: remove watch when closing rbd device
- librbd calls unwatch when the image is closed.
- 05:08 PM Linux kernel client Bug #912: rbd: remove watch when closing rbd device
- awesome, thanks! we should verify librbd is doing the same...
- 05:05 PM Linux kernel client Bug #912 (Resolved): rbd: remove watch when closing rbd device
- done, commit:ffdae98cb37f27b9da638145026a0f432faa4826.
- 04:33 PM Revision c9d6907e (ceph): rados: fix warning
- warning: rados.cc:319: large integer implicitly truncated to unsigned type
Signed-off-by: Sage Weil <sage.weil@dream... - 09:30 AM Linux kernel client Bug #69 (Can't reproduce): ceph: ffff88001976ba50 auth cap (null) not mds0 ???
- 09:29 AM Linux kernel client Tasks #1112 (Resolved): check all igrab at ceph-client,remove deadlock : spin_lock(&inode->i_lock...
- 09:03 AM Feature #1299 (Resolved): osd: interface to query/manage missing, unfound, lost objects
- add way to query unfound objects, and manually mark one or more of them lost.
- 03:58 AM Revision 36985229 (ceph): client: print ino along with &fh for ll_ file ops
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 03:47 AM Revision ad20e710 (ceph): rados: fix warning
- warning: rados.cc:319: large integer implicitly truncated to unsigned type
Signed-off-by: Sage Weil <sage.weil@dream... - 03:32 AM Revision 479af4bb (ceph): ls: mention directories without summary.yaml
- 03:32 AM Revision 8f3d56ad (ceph): include sha1 in summary
- Redundant (there's also a ceph-sha1 file), but convenient.
- 03:32 AM Revision 49e0e3f5 (ceph): ceph.conf: clean out random debug level changes
- keep it simple!
- 01:07 AM Revision 1edd435e (ceph): Clean up from pyflakes.
- 01:07 AM Revision 5fadb1c1 (ceph): Whitespace and style cleanup.
- 01:00 AM Revision e69cf0b1 (ceph): Success of test may not have been set yet.
- 01:00 AM Revision c53615ce (ceph): Remove unused variable.
07/11/2011
- 11:55 PM Revision 41451a02 (ceph): rgw: bucket deletion updates caches
- 11:45 PM Revision 38ad4d1b (ceph): task ceph: distribute monmap to all nodes, not just mons.
- And clean up the monmap, too!
Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com> - 11:45 PM Revision f92b3a1c (ceph): add locktest task
- This will retrieve xfstests' locktest and run it on two clients.
I still need to tweak this so the logging output we... - 11:23 PM Revision 28f19a41 (ceph): Add an option to keep machines locked if a test fails.
- 10:29 PM Feature #1214: qa: flock/fnctl lock tests
- Have initial commit of locktest task in teuthology. Still remaining: figuring out how to expand it to test stuff like...
- 10:28 PM Revision 2f35eddb (ceph): clean up locked machine list
- 10:28 PM Revision 1767b21e (ceph): print --lock-many result as yaml targets: stanza
- 10:28 PM Revision c47dc179 (ceph): lock: specify machines as input yaml targets: clause
- 10:12 PM Revision d5c8b965 (ceph): mime.c: avoid compiler warning
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 10:12 PM Revision 1fdf7212 (ceph): test/mime: test invalid decodes
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 10:11 PM Revision 1bfe9b86 (ceph): mime.h: better comments
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 09:53 PM Revision 4f1d6acd (ceph): Add mime encoding
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 09:39 PM Revision 91c6f351 (ceph): tell user which machines you locked
- 09:39 PM Revision 6cf9633a (ceph): nuke: use default owner
- 09:26 PM Revision cc40f115 (ceph): rgw: fix lru
- 09:23 PM Revision a8d4901f (ceph): make connect work if no roles are specified
- This is useful for -nuke.
- 08:49 PM Revision d1eb9cef (ceph): suite: schedule jobs instead of executing each configuration serially.
- 08:49 PM Revision 85c24bda (ceph): Add teuthology-schedule and teuthology-worker.
- schedule puts jobs in a beanstalk queue, worker takes them out and runs them.
- 08:41 PM Revision 059019c9 (ceph): rgw: restructure runtime config env
- 07:15 PM Revision 48837e3c (ceph): mds: rely on master to do anchor locks for slave_link_prep
- The replica can't take all these locks without confusing things, since it
maybe need to unlock/relock, may screw up a... - 07:15 PM Revision 844186f8 (ceph): mds: only issue xlocker_caps if we are also the loner
- We cannot issue caps to a client purely because they have something
xlocked, because we do not revoke caps when we dr... - 07:15 PM Revision cd5c7f48 (ceph): client: set I_COMPLETE after adding cap
- This avoids immediately clearing it because we are newly (re)issued the
Fs cap (a generic check in the cap add code).... - 07:15 PM Revision 84ff4393 (ceph): client: reenable local ENOENT
- This was accidentally disabled by commit:c7e248b.
Signed-off-by: Sage Weil <sage@newdream.net> - 04:57 PM Revision 24faa3d1 (ceph): CodingStyle
- 02:44 PM rgw Bug #1287: Setting metadata with unreadable characters is not consistent with amazon S3
- Also, python 2.X DOES mangle your strings, but only if you prepend 'U', making it a "unicode string"
So u'\04a' ==... - 12:47 PM rgw Bug #1287: Setting metadata with unreadable characters is not consistent with amazon S3
- I wanted to be sure about this, so I verified using tcpdump that we were really sending the data over the wire not en...
- 01:51 PM Feature #1210 (Resolved): qa suite: schedule nodes and run suite in parallel
- Done in 85c24bda7f67768dcbd6a75a7ac92ea38b88e1e1 and d1eb9cef38455db149751d0d5b399be36868fdc7 of teuthology.git.
- 10:56 AM Bug #1297: write a manypools teuthology task
- Nevermind the "track_memory_use" part, that can be done with collectd, and collectd buys us much much more.
- 10:42 AM Bug #1297 (Resolved): write a manypools teuthology task
- Write a teuthology task that creates lots of pools:...
- 10:50 AM Bug #1298 (Duplicate): osd: memory leak in 3f708ee
- dho osds running commit:3f708ee are leaking memory
- 09:30 AM Linux kernel client Feature #1296 (New): Support fanotify/inotify
- Having Ceph support distributed filesystem notification events would be desirable.
As requested by Damien Churchil...
07/10/2011
- 09:10 PM Revision 6a793f2b (ceph): mds: journal renamed dirino on rmdir
- This makes replay work on the auth by updating the subtrees accordingly
(since rmdir is really just renaming into the... - 09:05 PM Revision d2cdbe75 (ceph): mds: defer lock eval if freezing or frozen
- We were only deferring if frozen. But if freezing we need to too, because
of the way cap messages are deferred. We ... - 02:22 PM CephFS Bug #1295 (Resolved): mds: handle rmdir of subtree root
- If we rmdir an empty directory whose dirfrag(s) are exports/subtree roots,
- it gets journaled locally, replay is... - 05:44 AM Bug #1150: kclient: ERESTARTSYS from flock/fcntl locks
- Ceph: 0.31
Kernel 3.0.0-rc6
The same.
ps: In mds.0.log:
2011-07-10 15:38:45.094709 2011-07-10 15:38:45.094750...
07/09/2011
- 09:37 PM Revision 44ca5b3f (ceph): Merge remote branch 'origin/stable'
- 02:52 PM Tasks #1182 (Resolved): qa: lock machines
- 12:17 AM Revision 9bdde934 (ceph): rgw: cache lru
07/08/2011
- 11:30 PM Revision dc41c8d8 (ceph): client: remove unused variable
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 10:34 PM Revision 4f731219 (ceph): rgw: fix put of zero sized objects
- 09:17 PM Revision 9019c6ce (ceph): v0.31
- 09:16 PM Revision bae40eaa (ceph): filejournal: parse kernel version 3.0+
- Signed-off-by: Sage Weil <sage@newdream.net>
- 06:09 PM Revision 3f708ee5 (ceph): rgw: fix copy operation (clean up acls)
- 06:09 PM Revision 44e76fb8 (ceph): rgw: when listing objects, set locator key only when needed
- 06:09 PM Revision 3eb1ba86 (ceph): qa: make dir tests run by default in rename all.sh
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 05:33 PM Revision 73d98a05 (ceph): Merge branch 'mds_wrlock'
- 05:24 PM Revision f200f0dd (ceph): mds: fix warning
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 05:23 PM Revision 2dd514ab (ceph): osd: use shared helper for WRITE, CLONERANGE stats updates
- Signed-off-by: Sage Weil <sage@newdream.net>
- 05:23 PM Revision da89ef05 (ceph): osd: clean up a few things in do_osd_ops
- don't set head_exists
remove dup touch on setxattr
set maybe_created for clonerange
Signed-off-by: Sage Weil <sage@n... - 04:58 PM rgw Bug #1294 (Resolved): rgw: can't put zero sized objects
- 03:24 PM rgw Bug #1294 (Resolved): rgw: can't put zero sized objects
- We fail on trying to clone_range the source object as it wasn't created.
- 04:48 PM Revision 025748a6 (ceph): mds: implement remote_wrlock
- For the rename code to behave, we need to hold a wrlock on the slave node
to ensure that any racing gather (mix->lock... - 04:48 PM Revision d72bdab7 (ceph): mds: take a remote_wrlock on srcdir for cross-mds rename
- This ensures that we hold a wrlock on the srcdn auth when the slave
makes it's changes to the src directory, and prev... - 04:48 PM Revision 054fbc46 (ceph): mds: refactor drop_*_locks() to use helpers
- This slightly changes the unlock order for drop_locks() (rdlocks now
last instead of after xlocks, before [remote_]wr... - 04:32 PM Revision c3a40829 (ceph): mds: add mix->lock(2) state
- There is a problem with the wrlocks and cross-mds renames:
- master (dest auth, srci auth, srcdir replica) takes wr... - 04:32 PM Revision 4d5b0538 (ceph): client: clean up debug output
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 02:26 PM CephFS Bug #1167: Fsstress: bad rstat assert
- merged into master in commit:73d98a057bfeaf6f9a87c38f85b199c6fdbbbf04
- 02:21 PM CephFS Bug #1167 (Resolved): Fsstress: bad rstat assert
- 02:20 PM Bug #1293 (Resolved): get_kernel_version fails with Linux 3.0
- 04:14 AM Bug #1293 (Resolved): get_kernel_version fails with Linux 3.0
- As the title says, @get_kernel_version@ in @FileJournal.cc@ fails due to expecting 3 parts to the version number and ...
- 02:02 PM Bug #1283 (Resolved): rados df produces negative numbers and other incorrect bucket sizes
- 10:32 AM Feature #1271: proflogger: query interface
- i like json for output formatting. partly we shouldn't need any library here since the syntax is trivial, basically
... - 12:06 AM Revision 4218d706 (ceph): Add httplib2 to setup.py.
07/07/2011
- 11:46 PM Revision 5981ffbd (ceph): readme
- 11:19 PM Revision a260da12 (ceph): teuthology-suite: pass --lock and --block to teuthology
- 11:15 PM Revision fd30ed76 (ceph): Add --block option to retry until machines are locked.
- If there are not enough machines up, fail immediately.
- 11:05 PM Revision d6f35b5d (ceph): rados: fix usage()
- Exit after dumping usage to stderr on error.
Print to stdout for -h.
Signed-off-by: Sage Weil <sage@newdream.net> - 10:28 PM Revision 84f75c58 (ceph): Merge branch 'next'
- 10:28 PM Revision fbdb0a3a (ceph): mds: check that dnl exists before dereferencing it.
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 10:04 PM Revision a20d1101 (ceph): rgw: cache cleanups
- 09:56 PM Revision f14b4e19 (ceph): Check more invalid argument combinations for teuthology-lock.
- 09:24 PM Revision aece153a (ceph): Add systest framework. Create rados_list_parallel
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 09:24 PM Revision fbadee0a (ceph): Add rados_list_parallel to .gitignore
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 09:24 PM Revision 5a51afa3 (ceph): rados_list_parallel: debug hacks
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 09:24 PM Revision dbb122d1 (ceph): systest_runnable: implement processes
- Implement process-based tests as an alternative to threads.
Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com> - 09:24 PM Revision 8845749b (ceph): systest: add cross process semaphores
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 09:24 PM Revision 75950ac1 (ceph): rados_list_parallel: try to handle consistency issues
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 09:24 PM Revision 54ca67bb (ceph): handle_pool_op_reply: update osdmap even on error
- Even when a PoolOp returns an error, we still want to update the OSDMap
of our client. If not, we can get into scenar... - 09:24 PM Revision ffdbcb5d (ceph): systest: support passing in argc, argv
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 09:24 PM Revision 61fdbbf6 (ceph): librados: add conf_parse_argv, use in tests
- Signed-off-by: Colin McCabe <colin.mccabe@dreamhost.com>
- 09:21 PM Revision af93d811 (ceph): mon: fix up pending_inc pool op mess
- You can't look at pending_inc in preprocess methods. Or return an error
based on pending_inc before it commits. Fix... - 09:21 PM Revision 4f87d843 (ceph): mon: don't use pending_inc in poolop preprocess_*
- Signed-off-by: Sage Weil <sage@newdream.net>
- 09:21 PM Revision 12a5bb82 (ceph): mon: fix pool vs unmanaged snaps check
- Signed-off-by: Sage Weil <sage@newdream.net>
- 08:35 PM Revision 5dc09dd6 (ceph): mds: set old and new dentry lease bits
- Recent kernels got the new CEPH_LOCK_DN definition but we were still
setting the old bit. Set both so we work with b... - 08:12 PM Revision acd256cc (ceph): Merge branch 'next'
- 08:10 PM Revision a91b1bff (ceph): Revert "mds: fail out of path_traverse if we have a null dentry."
- This reverts commit 39d50c1362db1d86782a60a5714e088d9ef7deaa.
- 08:10 PM Revision 616ff502 (ceph): mds: conclude ENOENT during travrse if we have a readable null dentry
- Only if the lock is readable. For example, replicas with dn lock state
of LOCK aren't readable.
Signed-off-by: Sage... - 07:50 PM rgw Bug #1287: Setting metadata with unreadable characters is not consistent with amazon S3
- Confirmed through s3-tests. Amazon gives it back to you in mime-encoded format rather than giving you x-amz-missing.
- 07:14 PM rgw Bug #1287: Setting metadata with unreadable characters is not consistent with amazon S3
- Amazon says (in the developers' guide):...
- 07:35 PM Revision 9bfca879 (ceph): Check that all machines are locked, and add an option to lock machines ...
- 07:35 PM Revision 9158c831 (ceph): Verify that machines are locked before nuking them.
- 07:35 PM Revision a55d2eb5 (ceph): Read lock server from ~/teuthology.yaml.
- 07:35 PM Revision f6efcd39 (ceph): Update readme for locking.
- 07:35 PM Revision e3275155 (ceph): Remove locking from TODO.
- 07:35 PM Revision 4d62dd3b (ceph): Add command line tool for locking machines.
- 07:32 PM Revision 09bee435 (ceph): Move username to a utility method.
- 07:32 PM Revision dbf5e543 (ceph): Add simple lock server HTTP interface.
- 04:55 PM Bug #1292 (Closed): qa: bench & thrashosd PG won't go clean
- I ran a 360 second rados bench while thrashing in teuthology, and it never exited because one of the PGs never went c...
- 04:48 PM Documentation #1266 (Resolved): qa: document ceph-qa-suite
- 04:36 PM Feature #1214 (In Progress): qa: flock/fnctl lock tests
- I'll do this since I wrote most of the code, and the xfstests changes I made when writing it might be useful here.
- 04:36 PM Feature #1229 (Resolved): rados load generator
- 04:08 PM Bug #1282 (Resolved): rados CLI segfault
- fixed by commit:d6f35b5d628d73d43d2aa4061c1653f60e363472, will be pushed to cluster later.
rados was just failing ... - 03:12 PM Bug #1282: rados CLI segfault
- what machine is this on?
- 03:49 PM Bug #1248 (Resolved): teuthology: setup fails horribly if you don't have a mon on each node
- It was only copying the monmap to nodes that had a mon; simple enough to fix.
- 01:29 PM Bug #1248 (In Progress): teuthology: setup fails horribly if you don't have a mon on each node
- Lucky me!
- 03:30 PM Feature #1291 (Closed): teuthology: optionally pause on error
- For now, this can be dropping into interactive mode when a task fails.
- 03:14 PM Feature #1271: proflogger: query interface
- we can simplify this to only dump the current values and time differential.
maybe a separate command to reset th... - 01:41 PM Linux kernel client Bug #1288 (Resolved): kclient does not pass mds rename workunits
- This is fallout from an old change of the lock bit fields, and an incomplete removal of the lease bit mask code. Onl...
- 09:24 AM Linux kernel client Bug #1288 (Resolved): kclient does not pass mds rename workunits
- ...
- 01:18 PM Feature #1264 (Resolved): teuthology: allocate+lock ad/or verify nodes locked
- 01:17 PM Feature #1209 (Resolved): qa: node locking/reservation
- 01:11 PM CephFS Bug #1289 (Resolved): misdirected mds request results in improper ENOENT during mds rename workunits
- Resolved by commit:3ef1e656e95eacb1d0d8acd530e2cd9cbe08d586.
We had some problems with false ENOENTs on lookups th... - 11:24 AM CephFS Bug #1289 (Resolved): misdirected mds request results in improper ENOENT during mds rename workunits
- Detected as #1288, the kclient is doing a lookup on mds0 for "srci2" immediately following a create "srci2" request. ...
- 12:27 PM rgw Bug #1286 (Rejected): atomic write not working for large files
- The test wasn't really doing what it was supposed to do, and effectively tested the atomic reads, which is a differen...
- 12:24 PM rgw Feature #1290 (Resolved): rgw atomic GETs
- 12:02 AM Revision 4761317b (ceph): rgw: fix copy operation (clean up acls)
07/06/2011
- 11:45 PM Revision 850f337f (ceph): workunits task: clean up properly if there's an error.
- Previously it would fail out and leave the workunits directory, causing
final cleanup to fail.
Signed-off-by: Greg F... - 11:45 PM Revision 4fcecf01 (ceph): task ceph: set_max_mds so multiple MDS nodes are used
- The current check will be insufficient when we handle standby-replays,
standbys, etc, but it's a lot better than the ... - 10:14 PM rgw Bug #1286: atomic write not working for large files
- Looking at the logs, atomic writes are ok, however, the reads are not atomic.
- 06:16 PM rgw Bug #1286 (Rejected): atomic write not working for large files
- steps to reproduce:
* write a large file (4+ MB) of all A's
* start overwriting the file with all B's
* before ove... - 10:02 PM Revision 03b6c40b (ceph): qa: mds rename: account for not being in the ceph root dir
- We need to know the Ceph absolute path. We can't actually
derive that for sure (if we aren't mounted into the root), ... - 09:20 PM Revision e483bb6b (ceph): qa: mds rename: Rework so it will function in teuthology as a workunit:
- work in current directory, not hardcoded mnt path
use CEPH_TOOL variable rather than hardcoded local executable
pass ... - 09:20 PM Revision b5e4a319 (ceph): move mds rename tests to workunits directory
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 09:17 PM Revision 9655042d (ceph): Skip s3-tests marked fails_on_rgw, they will fail anyway.
- 09:17 PM Revision 1b2c9641 (ceph): Skip s3-tests marked fails_on_rgw, they will fail anyway.
- 09:15 PM Revision 3665f753 (ceph): The shell exits after the command, hence there is no need for pushd/popd.
- 09:15 PM Revision 06fb9b95 (ceph): The shell exits after the command, hence there is no need for pushd/popd.
- 08:49 PM Revision 088013b8 (ceph): mds: cleanup: use enum for lock states
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 08:49 PM Revision 528b6151 (ceph): Merge branch 'next'
- 08:34 PM Revision 8f9eaf0d (ceph): rgw: when listing objects, set locator key only when needed
- 07:36 PM rgw Bug #1287 (Resolved): Setting metadata with unreadable characters is not consistent with amazon S3
- If you have a string like '\x04world', Amazon will encode it using MIME encoded-word syntax. Currently, our S3 implem...
- 06:40 PM Revision 7ea8ecdc (ceph): Don't set unix env vars in fastcgi env.
- 05:47 PM Revision 0c6d2e6f (ceph): rgw: async rados delete should be called with the correct object
- 05:19 PM Bug #1285 (Resolved): rgw: copy object sets bad policy
- Fixed by commit:4761317b754c6a903862d1913b53c8ab5e06b8eb.
- 04:52 PM Feature #1219 (Closed): qa: migrate mds rename tests to teuthology
- Success. Had to fix a few bugs in teuthology too.
- 09:28 AM Feature #1219: qa: migrate mds rename tests to teuthology
- I've got this working I think, but it's exposed some bugs in cfuse and the kclient :( so I want to get those patched ...
- 04:19 PM CephFS Bug #1284 (Can't reproduce): client: leaked inode ref with cfuse_workunit_snaps.yaml
- Seems to be pretty rare. ...
- 04:13 PM Bug #1283 (Resolved): rados df produces negative numbers and other incorrect bucket sizes
- rados df produces negative numbers and other incorrect bucket sizes! this means we have no way of easily knowing the...
- 03:59 PM Revision 1d7fbed6 (ceph): rados: rename load-gen options
- No abbreviations, update usage().
Signed-off-by: Sage Weil <sage@newdream.net> - 02:30 PM Bug #1282 (Resolved): rados CLI segfault
- Horrible subject -- sorry, couldn't think of a better summary off hand. :(...
- 01:39 PM Bug #1279 (Resolved): osd: [WRN] bad locator
- Ok, the issue happened actually when listing objects inside a bucket. From what I understand this warning in this sit...
- 12:11 PM Bug #1279: osd: [WRN] bad locator
- ...
- 11:09 AM Bug #1279: osd: [WRN] bad locator
- I can't reproduce (this specific problem), what's your rgw commit version?
- 05:56 AM Bug #1279 (Resolved): osd: [WRN] bad locator
- Like I mentioned yesterday, I'm seeing bad locator messages on my OSD's after uploading a file through the RADOS gate...
- 11:28 AM Feature #1281 (Resolved): s3-test should test removal of owner's bucket grants behavior
- When owner doesn't have any grant to a bucket, it shouldn't be able to read or write to that bucket, however, it shou...
- 11:25 AM rgw Bug #1241 (Rejected): rgw denies owner access to bucket when acl is empty
- That's the expected behavior.
Will open a different issue for adding a test in s3-tests to verify that we do that. - 10:30 AM Bug #1280 (Closed): gceph: missing dependencies
07/05/2011
- 11:48 PM Revision f80a2f69 (ceph): Remove quotes from apache conf.
- 11:47 PM Revision f164dd79 (ceph): nuke: sudo for the final rm -rf
- 11:47 PM Revision 7e603ceb (ceph): adjust ceph.conf from yaml input
- 11:38 PM Revision a3b42e65 (ceph): Let apache kill rgw.
- It seems to be getting killed correctly now.
- 11:32 PM Revision ba6545f8 (ceph): Set LD_LIBRARY_PATH for rgw.
- 10:14 PM Revision d04256aa (ceph): set flavor to 'local' if pushing from local src dir
- always record in summary
- 09:57 PM Revision 5b2de2b9 (ceph): mkcephfs: Only create OSD journal dir if we have a journal.
- Thanks to huang jun <hjwsm1989@gmail.com> for finding the bug.
Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamho... - 09:56 PM Revision 1da8f817 (ceph): honor CINIT_FLAG_NO_DEFAULT_CONFIG_FILE
- Don't use CEPH_CONF_FILE_DEFAULT when CINIT_FLAG_NO_DEFAULT_CONFIG_FILE
is set.
Signed-off-by: Colin McCabe <colin.m... - 09:22 PM Revision 2aa146a7 (ceph): mds: always clear_flushed() after finish_flush()
- The scatter_writebehind_finish() is always followed up by an eval_gather(),
which does the clear_flushed(). For ever... - 08:58 PM Revision fb7696f3 (ceph): client: fix num_flushing_caps accounting
- This only affects debug output, fwiw.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 08:48 PM Revision 003492a2 (ceph): Retry connecting based on more error codes.
- 08:43 PM Revision e9e3883d (ceph): client: don't call flush_snaps when nothing to flush
- Otherwise we fail an assert.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 08:43 PM Revision 933e3495 (ceph): mds: kill stray break
- This broke with the gatherbuilder addition.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 08:33 PM Revision 7e1f09ff (ceph): context: implement complete()
- finish() requires the caller to delete. complete() does that for you by
calling finish() and then doing delete this.... - 05:07 PM Revision b11b5826 (ceph): Merge branch 'stable'
- 05:00 PM Revision 531f46c3 (ceph): logrotate.conf: Mark stat/*.log as "missingok"; it's not always there.
- Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com>
- 05:00 PM Revision 8459f80b (ceph): rgw: remove temp objects asynchronously
- 04:59 PM rgw Bug #1249 (Can't reproduce): radosgw does not exit when apache is killed
- 04:49 PM Feature #1254 (Resolved): teuthology: ceph.conf additions in ceph stanza of config.yaml
- 04:29 PM Bug #1278 (Resolved): teuthology: make nuke nuke kernel mounts
- usually the daemons are already down and the umount hangs. instead, umount -f, or just reboot (and then wait for the...
- 04:18 PM Revision 529df5db (ceph): Merge branch 'stable'
- 04:18 PM Revision b670f31d (ceph): Move stat/*.log to the end of logrotate.conf
- Logrotate ignores entries after a rule that doesn't match any files.
Signed-off-by: Alexandre Oliva <oliva@lsd.ic.un... - 04:01 PM Revision 2b168b03 (ceph): nuke: do not escape fusermount .../mnt.*
- 03:58 PM Revision 6feab3cb (ceph): mds: fix file_excl assert
- If we are in XSYN state and want to move to anything else, we must go via
EXCL, but we may not be loner anymore. Wea... - 03:30 PM Feature #1277 (Resolved): teuthology: manage known_hosts
- (02:21:43 PM) tv: host locking db gets an extra column, sshpubkey
(02:22:12 PM) tv: reinstalling script does ssh-key... - 02:40 PM CephFS Feature #1276 (Resolved): client: expose mds partition via virtual xattrs
- our options are basically and ioctl or virtual xattrs. virtual xattrs are easier to use and portable across fuse and...
- 02:36 PM CephFS Bug #1252 (Resolved): regression: cfuse fails snaptest-2 with 2-MDS system
- commit:e9e3883d0db4cc03183c0432773825f84170c42b
- 01:26 PM Feature #1271: proflogger: query interface
- - rip out current logging
- keep two copies of the counter vectors
- query via unix domain socket
- on each query,... - 12:40 PM Feature #1271 (Resolved): proflogger: query interface
- query via unix domain socket (instead of logging to a file and scraping that)
- 12:42 PM Bug #1275 (Resolved): fix testrados
- Fix testrados to always succeed, never fail or segfault. I probably will split the multi-threaded stuff into a separa...
- 12:41 PM Feature #1274 (Resolved): qa: test watch/notify races
- What happens if we put a watch on an object, and then another user destroys the pool it's in? Do we handle this case ...
- 12:41 PM Feature #1273 (Resolved): test pool creation and pool destruction races
- What happens if one guy destroys a pool while the other guy is using it? We should give an error code back to the guy...
- 12:40 PM Feature #1272 (Resolved): test rados list-objects in parallel with object modifications
- We need to test modifying a pool while its objects are being listed. Do we handle all the cases correctly? (Objects g...
- 11:57 AM Feature #1270 (Closed): qa: tests for splitting pgs vs read/write workload
- 11:33 AM Feature #1269 (Resolved): collectd: monitor cluster stats
- 10:55 AM rgw Feature #1268 (Rejected): rgw: optimistically fetch acl and first chunk
- 10:50 AM Feature #1267 (Rejected): osd: rgw class to do acl check
- 10:38 AM Documentation #1266 (Resolved): qa: document ceph-qa-suite
- 10:36 AM Feature #1265 (Resolved): osd: test osd recovery clone handling
- 10:33 AM Feature #1264 (Resolved): teuthology: allocate+lock ad/or verify nodes locked
- 10:32 AM Feature #1263 (Resolved): teuthology: scan cluster log for badness
- error out on any ERR (and possibly any WARN)? This will catch the errors the system notices but handles/ignores.
- 10:32 AM Feature #1262 (Closed): teuthology: monitor health during run
- 09:45 AM Revision d4fdaba9 (ceph): a few basic kclient workunits, reorg
- 09:45 AM Revision 3dd95f6c (ceph): .gitignore: ignore emacs backups
- 09:35 AM Feature #801 (Resolved): librados: allow access to multiple clusters
- 09:34 AM Bug #1185: rados: export caught in loop on 'buck' bucket (1.5M objects)
- See #1258
- 09:30 AM Subtask #815 (Resolved): Remove globals & partition g_conf
- This bug was really all about refactoring stuff to be thread-safe. Now, that task has been completed.
It's time to... - 09:28 AM Subtask #1227 (Rejected): write tests of libceph, librgw, librados library thread-safety
- this isn't really a subtask of #815. Will create more detailed bugs for all these.
- 09:26 AM Subtask #1231 (Rejected): NUM_THREADS=3 testrados segfaults
- Moving this into issue #1261
- 09:23 AM Linux kernel client Bug #1253 (Resolved): null point deref in __dentry_open
- 09:13 AM Bug #1257 (Resolved): test/fix concurrency issues with RADOS
- In general, we need to make sure that concurrent operation works properly with RADOS / librados / DHO.
We have never...
07/04/2011
- 10:18 PM Bug #1256 (Won't Fix): mds dies on ESHUTDOWN under too-high mon load (time-outs?)
- It's relatively common for the active mds to die while I run backups from/to the filesystems that hold the mon data i...
07/03/2011
- 10:29 PM Bug #1255 (Duplicate): configure does not detect missing g++
- When g++ is missing, it fails on some other check (boost). Should verify that g++ exists first.
- 03:57 AM Revision e1db8e9e (ceph): new workunits
Also available in: Atom