Activity
From 04/06/2012 to 05/05/2012
05/05/2012
- 11:32 PM Revision b5e9995f (ceph): objectcacher: delete unused onfinish from flush_set
- Once upon a time the caller would do this, but none of those have survived,
and this makes more sense.
Signed-off-by... - 11:32 PM Revision 38514230 (ceph): objectcacher: make cache sizes explicit
- Make ObjectCacher users specify the cache size for each ObjectCacher
instances. This avoids the confusing config nam... - 11:32 PM Revision c19f998a (ceph): common: add C_Cond
- Similar to C_SafeCond, but assume finisher already holds the relevant lock.
Signed-off-by: Sage Weil <sage@newdream.... - 11:32 PM Revision 6f3221a9 (ceph): objectcacher: explicit write-thru mode
- If the max_dirty config is 0, switch to write-thru mode, which will
explicitly flush and wait on the range we just di... - 11:31 PM Revision 38edd3bb (ceph): objectcacher: user helper to get starting point in buffer map
- A common pattern is to search for the first buffer intersecting or
following an object offset. Use a helper for that... - 11:25 PM Revision c8bd471b (ceph): objectcacher: flush range, set
- Add ability to flush a range of an object, or a vector of ObjectExtents. Flush
any buffers that intersect the specif... - 09:27 PM RADOS Subtask #2340 (Fix Under Review): crush: remove forcefeeding
- 08:21 PM Revision b50a4c92 (ceph): mon: add safety checks for 'mds rm <gid>' command
- - make sure the gid exists
- only remove it if it's inactive (state < 0)
Fixes: #2188
Signed-off-by: Sage Weil <sage... - 08:08 PM Revision 86aa07d7 (ceph): osd: check against last_peering_reset in _activate_committed
- We are checking against last_peering_reset in _activate_committed(), so we
need to pass in that value to compare agai... - 08:08 PM Revision 8ec476e5 (ceph): osd: do not mark pg clean until active is durable
- Do not mark a PG CLEAN or set last_epoch_clean until after the PG activate
is stable on all replicas.
This effective... - 08:08 PM Revision a50fbd56 (ceph): mon: move PGMap[::Incremental] to new encode/decode macros
- This is a backward-incompatible change; downgrades will no longer be
possible.
Signed-off-by: Sage Weil <sage.weil@d... - 08:08 PM Revision 9d7ec04b (ceph): osd: tweak slow request warnings
- - always include 'slow request' in the warning string
- only summarize if we warn about anything (they all may have b... - 05:13 PM Revision a4b42fc3 (ceph): keyring: clean up error output
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 05:13 PM Revision ae0ca7be (ceph): keyring: catch key decode errors
- Return EINVAL on decoding errors.
Other decode_base64() callers are already guarded.
Fixes: #2124
Signed-off-by: Sa... - 05:03 PM Revision 6812309e (ceph): debian: depend on uuid-runtime
- We use uuidgen for osd creation.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 05:01 PM Revision 3509b039 (ceph): safe_io: int -> ssize_t
- int is 32-bit on 64-bit archs, but ssize_t is 64-bits. This fixes overflow
when reading large (>2GB) extends.
Fixes... - 04:30 PM Revision 396d1fef (ceph): ceph.newdream.net -> ceph.com
- 04:30 PM Revision ce951cf4 (ceph): ceph.newdream.net -> ceph.com
- 02:11 PM Bug #2188 (Resolved): mon: mds rm should be harder to break things with
- added checks that the gid exist and that it is not active. with those in place, this is no longer dangerous... all y...
- 02:02 PM Cleanup #2344 (Fix Under Review): convert Monitor maps to use ENCODE_START
- wip-mon-encoding
- 01:57 PM Bug #2373 (Fix Under Review): osd: last_epoch_clean > last_epoch_started
- wip-osd-peering
- 11:43 AM Bug #2373: osd: last_epoch_clean > last_epoch_started
- the problem is that we set the CLEAN state bit and report that to the monitor before last_epoch_started has propagate...
- 11:10 AM Bug #2192 (Won't Fix): ceph-mon hangs consuming 100% CPU
- Yep, this sounds like the writeback sync deadlock:
- ceph-mon calls sync
- the kernel client flushes it's dirty... - 11:01 AM Bug #2124 (Resolved): crash when malformed auth key is provided
- commit:ae0ca7be3c8f1aaf5f0bf7534363ffd60b6a04e2
- 10:53 AM Bug #2275 (Resolved): osd: crash in FileJournal::wrap_read_bl
- converted int -> ssize_t in common/safe_io.c, commit:3509b039a28d41c7ae1b3d482d67a27f8e5739e8
- 03:31 AM Revision 203a7d67 (ceph): objectcacher: wait directly from writex()
- This gives us access to the original ObjectExtent (useful later), and
simplifies the callers.
Signed-off-by: Sage We... - 01:40 AM CephFS Bug #2385: max mds = 2, mds hang and crash
- Sorry..
ceph.com/debian repo wheezy binaries. hang all client task.... - 01:38 AM CephFS Bug #2385 (Can't reproduce): max mds = 2, mds hang and crash
05/04/2012
- 11:02 PM Revision 991c93ed (ceph): mon: fix call to get_uuid() on non-existant osd
- Didn't catch this with vstart.sh testing.
Signed-off-by: Sage Weil <sage@newdream.net> - 10:55 PM Revision 150adcce (ceph): debian: add rules for rest-bench
- Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
- 10:55 PM Revision 53f642e2 (ceph): rest-bench: build conditionally
- added configure --with-rest-bench, and configure --with-system-libs3
Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream... - 10:53 PM Revision d54ef1c8 (ceph): rest-bench: change thread context for libs3 calls
- Apparently S3_put_object() and S3_get_object() need to
run on the same thread as S3_runall_request_context() (at leas... - 10:53 PM Revision f60444fb (ceph): obj_bencher: changed interface
- Passing bufferlist and not const bufferlist in aio_write(). We assign
it to another object which is not const, and it... - 10:53 PM Revision ddb858c5 (ceph): rados_bencher: restructure code (initial work)
- Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
- 10:53 PM Revision a17124ef (ceph): rados_bencher: restructure code, create RadosBencher class
- Preparing for different benchmark backend.
Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> - 10:53 PM Revision 1a8eea85 (ceph): rados_bencher: fix build
- Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
- 10:53 PM Revision 0fbc3c53 (ceph): rados_bencher -> obj_bencher
- Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
- 10:53 PM Revision f9d9fb6a (ceph): rados_bencher: abstract away rados specific operations
- Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
- 10:53 PM Revision e6026fee (ceph): rest_bench: initial work
- Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
- 10:53 PM Revision 54da3e6d (ceph): rest_bench: some more implementation
- Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
- 10:53 PM Revision 13296a54 (ceph): rest_bench: cleanups, implement get and put
- Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
- 10:53 PM Revision 6e043808 (ceph): rest-bench: create workqueue for requests dispatching
- Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
- 10:53 PM Revision 4fe068ec (ceph): rest-bench: cleanups, initialization
- Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
- 10:53 PM Revision 072c316a (ceph): rest-bench: multiple fixes
- write seems to work
Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> - 10:53 PM Revision 7e96a4a8 (ceph): rest-bench: use refcount for req_state life cycle
- Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
- 10:53 PM Revision e62fd7f9 (ceph): obj_bencher: fix param order
- seq benchmark was broken, passed params in wrong order.
Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> - 10:53 PM Revision e2eb8256 (ceph): rest-bench: reuse libs3 handle
- This is necessary for keep-alive to be useful. Otherwise a new
connection will be created for each request.
Signed-o... - 10:53 PM Revision 76a5c899 (ceph): obj_bencher: use better round robin for completion slot scan
- Start where left last time, don't start from zero.
Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> - 10:53 PM Revision 99791327 (ceph): obj_bencher: fix data encoding
- There was a bug when doing a read with multiple threads, when
one of the threads was left behind; when it returned th... - 10:53 PM Revision 68322318 (ceph): rest-bench: change command line arg for seconds
- seconds should be a param, not a command.
Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> - 10:26 PM Revision 3e260aec (ceph): librados: call safe callback on read operation
- This avoids confusion for the user who isn't sure if they should wait for
complete or safe on a read aio. It also me... - 09:51 PM Revision edd73e2e (ceph): crush: note that tree bucket size is tree size, not item count
- Signed-off-by: Sage Weil <sage@newdream.net>
- 09:16 PM Revision 4061fca5 (ceph): Merge remote-tracking branch 'gh/wip-crush-forcefeed'
- Reviewed-by: Sam Just <sam.just@inktank.com>
- 09:15 PM Revision ce60e1be (ceph): OpRequest: ignore all ops while the oldest one is still young.
- Signed-off-by: Joao Eduardo Luis <jecluis@gmail.com>
Reviewed-by: Sage Weil <sage@newdream.net> - 08:12 PM Revision f3043fee (ceph): objectcacher: don't wait for write waiters; wait after dirtying
- We do three things here:
- Wait for the dirty limit to drop _after_ writing into the cache. This
means that an ac... - 07:09 PM Revision f3760da4 (ceph): crush: update_item() should pass an error back to the caller
- If you give it a nonsensical loc, it will fail check_item_loc() (false) and
then error out on insert_item().
Reporte... - 06:06 PM Revision e0a636f9 (ceph): crush: improve docs/comments for check_item_loc and insert_item semantics
- We don't adjust the internal hierarchy structure (currently). This is a
bit confusing, so describe the semantics in ... - 06:05 PM Revision 878423f9 (ceph): crush: comment and clean up checks for check_item_loc and insert_item
- - drop useless cur for check_item_loc
- comment the checks we're doing so the code is understandable
- use name_exist... - 03:48 PM Bug #2022 (Fix Under Review): osd: misdirectect request
- i just found a crush algorithm fix/change that was in ceph.git but on the kernel... i bet that is the problem. now s...
- 02:17 PM RADOS Subtask #2340 (In Progress): crush: remove forcefeeding
- just need to apply these to the kernel implementation, too
- 02:09 PM Feature #2335 (Fix Under Review): librbd: write-thru cache mode
- wip-rbd-wt
- 12:33 PM Feature #2290 (Fix Under Review): ObjectCacher: handle read/write errors
- Forgot to update this before, but it's in wip-oc-error-handling.
- 12:24 PM Linux kernel client Bug #2384 (Resolved): libceph: fix all vmalloc (buffer_new) callers to use GFP_KERNEL
- GFP_NOFS is not safe because __vmalloc may ahve to adjust page tables with GFP_KERNEL. need to fix our callers to no...
- 09:55 AM rbd Feature #2383 (New): tool for exposing block existence and location for an rbd image
- This would give users some transparency about rbd block placement and usage.
- 03:44 AM Revision 845e2aa5 (ceph): Merge branch 'wip-crush-update'
- Reviewed-by: Greg Farnum <greg@inktank.com>
- 03:43 AM Revision 720bea4a (ceph): Merge branch 'wip-osd-uuid'
- Reviewed-by: Greg Farnum <greg@inktank.com>
- 03:42 AM Revision 2629474f (ceph): global_init: do not count threads before daemonize()
- We were verifying that there was only 1 thread (the presumably main()) when
we call daemonize. However, with the new... - 03:40 AM Revision 72538c0f (ceph): Makefile: fix $shell_scripts substution
- No spaces here, apparently!
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 03:34 AM Revision 16461acf (ceph): mon: simplify 'osd create <uuid>' command
- Make the flow clearer for the three cases (exists, about to exist, new).
Signed-off-by: Sage Weil <sage@newdream.net> - 03:33 AM Revision 42f2d2fd (ceph): crushtool: another simple test for update
- If the weight doesn't change it should be a no-op.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 03:28 AM Revision 9772d132 (ceph): crush: document return values
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 03:28 AM Revision 1cd6f764 (ceph): crush: compare fixed-point weights in update_item
- This is less ugly than converting the quantized value back to a float and
comparing that.
Signed-off-by: Sage Weil <... - 03:01 AM Revision c03b8521 (ceph): thread: remove get_num_threads() static
- This looks in /proc to count threads. Kludgey and no longer needed.
Signed-off-by: Sage Weil <sage.weil@dreamhost.c... - 03:00 AM Revision e50932c2 (ceph): global_init: do not count threads before daemonize()
- We were verifying that there was only 1 thread (the presumably main()) when
we call daemonize. However, with the new... - 01:59 AM Revision 684558ac (ceph): crush: clean up check_item_loc() comments
- Thanks Greg!
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 01:25 AM CephFS Bug #2375: rrdtoll data malfuntion..
- it's still in ceph-0.46 from ceph.com/debian repository..
05/03/2012
- 10:49 PM Revision 27d98d24 (ceph): OpRequest: only show a small set of the oldest messages, instead of all.
- Signed-off-by: Joao Eduardo Luis <jecluis@gmail.com>
Reviewed-by: Greg Farnum <gregory.farnum@dreamhost.com> - 09:44 PM RADOS Subtask #2340 (Fix Under Review): crush: remove forcefeeding
- 09:41 PM Feature #2367 (Resolved): mon: osd crush add should move item if it exists elsewhere
- 09:40 PM Bug #2372 (Resolved): librbd: unit tests get different error codes
- 08:13 PM Bug #2382 (Resolved): osd: unable to start due to 1 child already started
- Sounds good to me; I never liked depending on /proc for that anyway.
Merged into master. We probably want to put it ... - 07:38 PM Bug #2382 (Fix Under Review): osd: unable to start due to 1 child already started
- wip-2382
- 07:35 PM Bug #2382: osd: unable to start due to 1 child already started
- ok, this is just a bad check. we're verifying there aren't threads because fork()/daemonize() will destroy them. th...
- 06:11 PM Bug #2382: osd: unable to start due to 1 child already started
- I just built fresh 0.46 rpms (ran 0.45 before) and now I'm seeing this too.
Notice the timestamps. I had to call thi... - 06:09 PM Bug #2382: osd: unable to start due to 1 child already started
- Also, occasionally, this also happens with ./init-ceph when starting all services, or each one individually. For inst...
- 06:04 PM Bug #2382: osd: unable to start due to 1 child already started
- I re-triggered this using 'CEPH_NUM_OSD=1 CEPH_NUM_MDS=1 CEPH_NUM_MON=1 ./vstart.sh' on my desktop (granted, it's a d...
- 04:21 PM Bug #2382: osd: unable to start due to 1 child already started
- i saw this on congress too. will reproduce on my burnupi cluster and investigate.
- 04:12 PM Bug #2382: osd: unable to start due to 1 child already started
- update: if i start the osd's with ceph-osd manually this doesn't seem to happen - just with /etc/init.d/ceph start an...
- 04:01 PM Bug #2382 (Resolved): osd: unable to start due to 1 child already started
- I had seen this bug a few days ago while setting up ceph on my desktop, but it went away by rerunning ./ceph-osd so I...
- 07:52 PM Revision 3228643f (ceph): rgw: update cache interface for put_obj_meta
- This fixes issue #2381.
The method interface was different than the one needed in order
to override the one in RGWRad... - 07:26 PM Revision b5a18ad8 (ceph): doc: fix some underscores
- Signed-off-by: Sage Weil <sage@newdream.net>
- 07:19 PM Revision 5125daa6 (ceph): osd: drop unused CEPH_OSDMAP*VERSION* #defines
- It's easier to manage/rev/grok these inline.
Signed-off-by: Sage Weil <sage@newdream.net> - 07:16 PM Revision 1e7dc441 (ceph): Merge branch 'wip-doc-rebase-2'
- 07:14 PM Revision f15c0bba (ceph): Fixed link to blog.
- Signed-off-by: John Wilkins <john.wilkins@dreamhost.com>
- 06:49 PM Revision b2100ca0 (ceph): Fixed another link to the blog.
- Signed-off-by: John Wilkins <john.wilkins@dreamhost.com>
- 06:42 PM Revision 5ac8b588 (ceph): Fixed link.
- Signed-off-by: John Wilkins <john.wilkins@dreamhost.com>
- 06:31 PM Revision 7eca0dd9 (ceph): Clean up. Changed ceph.newdream.net to ceph.com.
- Removed {ARCH} references.
Added link to Source.
Signed-off-by: John Wilkins <john.wilkins@dreamhost.com> - 06:25 PM Revision 91c30b3a (ceph): doc: more fonts
- Signed-off-by: Sage Weil <sage@newdream.net>
- 06:03 PM Revision cabe74b1 (ceph): doc: new theme
- Signed-off-by: Ross Turk <ross.turk@inktank.com>
- 05:51 PM Revision 54991fff (ceph): doc/install/debian: simplify more
- Signed-off-by: Sage Weil <sage@newdream.net>
- 05:45 PM Revision c3dc54b0 (ceph): doc/install: reorg, simplify
- Signed-off-by: Sage Weil <sage@newdream.net>
- 05:39 PM Revision dc894eee (ceph): doc: clarify path names for mkcephfs
- Signed-off-by: Sage Weil <sage@newdream.net>
- 05:38 PM Revision bd8577dd (ceph): doc: simplify quick start
- Remove dup items
Signed-off-by: Sage Weil <sage@newdream.net> - 05:15 PM Revision 5465e810 (ceph): doc: Whitespace cleanup.
- Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com>
- 05:10 PM Revision 93dcc988 (ceph): doc: Rename to use dashes not underscores in URLs.
- This makes the-separate-words in the url match as separate words in
searches, where this_way only matches an explicit... - 05:02 PM Revision 715a69c1 (ceph): doc: Remove leading whitespace, that syntax means it's a definition list.
- Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com>
- 04:42 PM Revision 3226722c (ceph): Updated URLs for the new ceph.com site.
- Signed-off-by: John Wilkins <john.wilkins@dreamhost.com>
- 12:51 PM rgw Bug #2381 (Resolved): rgw: radosgw-admin operations don't trigger notifications
- Fixed, commit:3228643f1ec34a500246ddc1e16025f05b587342
- 12:31 PM rgw Bug #2381 (Resolved): rgw: radosgw-admin operations don't trigger notifications
- radosgw-admin operations that update user info are not triggering notifications to the rgw daemons, thus rgw instance...
- 12:51 PM Bug #2377 (Resolved): watch is lost if tcp timeout reached on the connection
- Fixed, commit:a5f533a7ed210c274c44a7fa922377e1a0ab9900.
- 10:35 AM CephFS Bug #2380 (Rejected): kclient: aufs over a cephfs mount fails with Stale NFS file handle
- The aufs directory is inaccessible.
<notextile>#</notextile> mkdir /tmp/dir{1..3}
<notextile>#</notextile> mount ... - 06:01 AM Revision be855d85 (ceph): objecter: ping connections with lingering ops every tick()
- Also, make sure that we initialize linger_op.session once
we have it.
Signed-off-by: Yehuda Sadeh <yehuda@hq.newdrea... - 03:31 AM Revision d49c3d29 (ceph): Removed "Ceph Development Status" per Bryan
- Modified title syntax per Tommi
Modified paragraph width to 80-chars per Dan
Moved "Build from Source" out of Install... - 03:06 AM Bug #2316: rbd: restart of OSD leeds to stale qem-VM's with "ceph version 0.45-207-g3053e47"
- Hi Josh, Sage,
as we are now proud to be mentioned on ceph.com, we should push things a bit, cause have to breathe... - 02:28 AM Bug #2379 (Resolved): Mon crash after start
- ...
05/02/2012
- 11:50 PM Revision 606fdf47 (ceph): Merge branch 'stable'
- Conflicts:
src/common/config_opts.h - 11:49 PM Revision d14b016d (ceph): config: $cluster-whatever for admin_socket, log files
- This is more consistent with /var/lib/ceph/$type/$cluster-$id.
Signed-off-by: Sage Weil <sage@newdream.net> - 11:39 PM CephFS Cleanup #2378 (Resolved): "ceph -s" MDS output is confusing
- If you're running an RBD/RGW cluster without an MDS daemon, having output like...
- 11:13 PM Feature #685: libcephmon: interact with ceph monitors via a library
- Has there been any progress on this?
My idea is still to make some sort of nice WebGUI where you can see the state... - 10:56 PM Revision ad0330b7 (ceph): config: Admin socket path had literal "name" in it instead of $name.
- Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com>
- 10:50 PM Revision 461e62fc (ceph): config: Admin socket path had literal "name" in it instead of $name.
- Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com>
- 10:32 PM Revision 28b1fc80 (ceph): mon: make 'osd map <poolname> <objectname>' return pg and osd mapping
- This keeps coming up and is annoying to do manually.
Fixes: #2141
Signed-off-by: Sage Weil <sage@newdream.net>
Revie... - 10:14 PM Revision 4dab4acd (ceph): ceph-object-corpus: a few instances of the newly encoded types
- Signed-off-by: Sage Weil <sage@newdream.net>
- 10:13 PM Revision 72846379 (ceph): ceph-dencoder: ignore trailing goop after OSDMap and OSDMap::Incremental
- All users pass around bufferlists and avoid encoding these structures
inline, but the dencoder tests are picky. Disa... - 09:59 PM Revision 33e35478 (ceph): vstart.sh: use 'osd crush set ...'
- Signed-off-by: Sage Weil <sage@newdream.net>
- 09:59 PM Revision 69dcd288 (ceph): mon: drop 'osd crush add ...'
- 'osd crush set ...' is better, us that instead.
Signed-off-by: Sage Weil <sage@newdream.net> - 09:55 PM Revision 79edba2f (ceph): mon: 'osd crush set ...' do an add or update
- This operation will add/update/move the item to the specified location.
It is idempotent and much more useful than 'o... - 09:55 PM Revision eca7eac3 (ceph): crush: fix weights when removing items
- Reweight an item to 0 before removing it, so that the parent weights are
adjusted accordingly.
Signed-off-by: Sage W... - 09:55 PM Revision bfa76351 (ceph): crush: add check_item_loc
- The check_item_loc() method will take an item and position and tell you if
it matches the items current location. Th... - 09:55 PM Revision 8220ab25 (ceph): crush: implement update_item()
- This is similar to insert_item(), except it will succeed if the item is
already there, and will move an item to the s... - 09:55 PM Revision 5aadb578 (ceph): crush: do some docs
- Signed-off-by: Sage Weil <sage@newdream.net>
- 09:55 PM Revision ce5842b6 (ceph): crushtool: add --update-item command
- Similar to --add-item, except it will move, rename, or reweight the item if
it is already present in the map.
Signed... - 09:55 PM Revision ee00095a (ceph): crushtool: extent cli test to include --remove-item and --update-item
- Signed-off-by: Sage Weil <sage@newdream.net>
- 08:59 PM Revision 76d166f9 (ceph): rbd.py: add new error for deleting an image with snapshots
- Forgot to update the python bindings in
88dda3be5aed38c8f587af9b27cb54a0fcca7530
Signed-off-by: Josh Durgin <josh.du... - 08:32 PM Revision e20cd4ba (ceph): config: no -C shorthand for --cluster
- ceph-authtool uses it!
Fixes: #2376
Signed-off-by: Sage Weil <sage@newdream.net> - 07:09 PM Revision f53c8206 (ceph): Minor edits. Still WIP.
- Signed-off-by: John Wilkins <john.wilkins@dreamhost.com>
Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com> - 07:09 PM Revision 7d501fcd (ceph): Removed this. As part of restructuring of install to include admin host.
- Signed-off-by: John Wilkins <john.wilkins@dreamhost.com>
Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com> - 07:09 PM Revision 4cd44ecc (ceph): doc: document NOIN, NOOUT, NOUP, NODOWN flags and flapping
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com> - 07:09 PM Revision ee44db4a (ceph): Significant updates to introduction, getting packages, building from so...
- Signed-off-by: John Wilkins <john.wilkins@dreamhost.com>
Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com> - 07:09 PM Revision 0fb0ef9e (ceph): Corrections.
- Signed-off-by: John Wilkins <john.wilkins@dreamhost.com>
Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com> - 07:09 PM Revision ec997750 (ceph): Edits.
- Signed-off-by: John Wilkins <john.wilkins@dreamhost.com>
Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com> - 07:09 PM Revision 541a875c (ceph): Restructuring documentation.
- Submitted by: John Wilkins <john.wilkins@dreamhost.com>
Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com> - 07:09 PM Revision b0ee6e87 (ceph): Added introduction to clustered storage and deleted older files that ha...
- Submitted by: John Wilkins <john.wilkins@dreamhost.com>
Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com> - 07:09 PM Revision bc857d86 (ceph): Remove reference to Introduction to RADOS OSDs
- Submitted by: John Wilkins <john.wilkins@dreamhost.com>
Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com> - 07:09 PM Revision 859da18e (ceph): Building out information architecture. Modified getting involved, why u...
- Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com>
- 07:09 PM Revision 9f9b7f88 (ceph): Added a plug for commercial support. ;)
- Submitted by: John Wilkins <john.wilkins@dreamhost.com>
Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com> - 07:09 PM Revision d7922e0d (ceph): Removed some files for reorg.
- Submitted by: John Wilkins <john.wilkins@dreamhost.com>
Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com> - 07:09 PM Revision f8b00f3e (ceph): Added a new landing page graphic, and made some minor edits on the land...
- Submitted by: John Wilkins <john.wilkins@dreamhost.com>
Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com> - 07:09 PM Revision 9b487571 (ceph): Minor cleanup.
- Signed off by: John Wilkins <john.wilkins@dreamhost.com>
Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com> - 07:09 PM Revision 33dc4c26 (ceph): The new files contain S3 APIs and a build from source doc.
- Signed-off-by: John Wilkins <john.wilkins@dreamhost.com>
Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com> - 07:09 PM Revision a1b31ddf (ceph): Initial cut of introduction, getting started, and installing. More to d...
- Signed-off-by: John Wilkins <john.wilkins@dreamhost.com>
Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com> - 06:46 PM Revision 31fb8f97 (ceph): Put change to rbd manpage in .rst instead, update man/ from output
- 06:21 PM Revision d3a2c565 (ceph): doc: move documentation build instructions to doc/dev section
- Signed-off-by: John Wilkins <john.wilkins@dreamhost.com>
Signed-off-by: Tommi Virtanen <tommi.virtanen@dreamhost.com> - 04:38 PM Bug #2377: watch is lost if tcp timeout reached on the connection
- Anything that relies on the underlying TCP connection to stick around forever is going to be a problem for us as we s...
- 04:23 PM Bug #2377 (Resolved): watch is lost if tcp timeout reached on the connection
- Once tcp timeout is reached on a connection that has a watch on, we don't try to reconnect immediately (should we eve...
- 04:23 PM Feature #2335 (In Progress): librbd: write-thru cache mode
- just setting the dirty limit to 0 won't work (well), because we'll end up with a single io in flight at any point in ...
- 03:32 PM Feature #2141 (Resolved): ceph: 'object map <poolname> <objectname>' or similar
- 03:22 PM Feature #2141: ceph: 'object map <poolname> <objectname>' or similar
- Looks good, assuming you actually ran it and tested something.
- 03:02 PM Feature #2141 (Fix Under Review): ceph: 'object map <poolname> <objectname>' or similar
- ceph osd map poolname objectorkeyname
- 02:47 PM Feature #2349: rados bench: Extra statistics
- Also, a header showing how the command was invoked!
- 02:19 PM Bug #2370: rbd doesn't support mv/rename across pools; tool should detect
- Trying to fix this:
commit:ee26c5d73a48b64292d16a87ebe69908c142048e - 01:59 PM Bug #2372: librbd: unit tests get different error codes
- merged into master
- 10:30 AM Bug #2372 (Fix Under Review): librbd: unit tests get different error codes
- Fixed in wip-rbdpy
- 01:55 PM Bug #2221 (Resolved): Monitor setup bugs
- I have spent way too long doing monitor additions lately, and I've gotten a few things cleaned up and checking the do...
- 01:54 PM Bug #2338 (Rejected): mon: adding new monitors simultaneously can allow a new mon to become leader
- Everything *does* work properly if you aren't located in the monmap. I would love to come up with a way of providing ...
- 01:43 PM Cleanup #2376 (Resolved): ceph-authtool -C option doesn't work
- commit:e20cd4baf912414b089adb2e2386c4dc94088d30
- 01:29 PM Cleanup #2376 (In Progress): ceph-authtool -C option doesn't work
- I broke this.. -C is short for --cluster. reverting and pushing to stable.
- 01:27 PM Cleanup #2376: ceph-authtool -C option doesn't work
- Also:...
- 01:26 PM Cleanup #2376 (Resolved): ceph-authtool -C option doesn't work
- ...
- 01:43 PM Feature #2367 (Fix Under Review): mon: osd crush add should move item if it exists elsewhere
- wip-crush-update
- 10:31 AM Feature #2336 (Resolved): qemu: wire up discard
- Accepted upstream in Kevin's block branch.
- 09:18 AM CephFS Bug #2375 (Closed): rrdtoll data malfuntion..
- Sorry for my english.
i prepared 4 debian-amd64-wheezy host cephfs. mount in /mnt
when copy rrd db data to mnt
... - 09:03 AM Linux kernel client Feature #2374: ceph-client: start laying the groundwork for Linux tracepoints
- Sounds like a great idea to me!
- 05:46 AM Linux kernel client Feature #2374 (New): ceph-client: start laying the groundwork for Linux tracepoints
- Linux supports a mechanism for very efficiently inserting trace points
in code, which allow for an arbitrary functio... - 05:26 AM Revision 715abdea (ceph): ignore syslog cron noise
- 03:08 AM Revision ee26c5d7 (ceph): Treat rename across pools as an error
- Fixes: #2370
Reviewed-by: Samuel Just <samuel.just@dreamhost.com>
05/01/2012
- 11:47 PM Revision b41f4d14 (ceph): rgw: normalize bucket/obj before updating cache
- Fixes bug #2369. The problem was that sometimes we send the
notification with the un-normalized bucket/obj pair. We
s... - 11:40 PM Revision 2a93258a (ceph): doc: add warning about multiple monitors on one machine.
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 11:20 PM Revision 695d3775 (ceph): vstart.sh: explicitly specify uuids during startup
- This exercises all the new per-osd uuid code.
Signed-off-by: Sage Weil <sage@newdream.net> - 11:16 PM Revision 09e08287 (ceph): osd: --get-{osd,journal}-uuid synonyms for --get-{osd,journal}-fsid
- Signed-off-by: Sage Weil <sage@newdream.net>
- 11:16 PM Revision 5a2fe979 (ceph): osd: allow uuid to be fed to mkfs with 'osd uuid' setting
- E.g., ceph-osd --mkfs --osd-uuid <uuid> -i 123 ...
Signed-off-by: Sage Weil <sage@newdream.net> - 11:15 PM Revision 8ec39205 (ceph): filestore: allow fsid to be fed in for mkfs
- Mkfs currently always generates a new uuid. Allow the caller to feed one
in.
Signed-off-by: Sage Weil <sage@newdrea... - 11:14 PM Revision 779914ba (ceph): mon: 'osd create <uuid>'
- Make the osd create command idempotent by providing a uuid. If you call it
multiple times with the same (or some oth... - 11:13 PM Revision 78b9ccd3 (ceph): mon: fill in osd uuid in map on boot
- We may want to make this more strict, so that if it is defined it has to
match the map, and only fill it in when the ... - 11:12 PM Revision ca4311e5 (ceph): osdmap: store a uuid for each osd
- Rev the extended section of the map to store it. Dump it when the osd
exists. Zero it out if an osd is destroyed. ... - 10:42 PM Bug #2373 (Resolved): osd: last_epoch_clean > last_epoch_started
- ...
- 10:24 PM Bug #2372 (Resolved): librbd: unit tests get different error codes
- 2012-05-01T02:28:28.447 INFO:teuthology.task.workunit.client.0.err:test_rbd.TestImage.test_remove_with_snap ... ERROR...
- 10:21 PM Linux kernel client Bug #2371: pjd chown/00.t 141, 145, 153 fail on kclient
- maybe related to toher pjd failures on ceph-fuse, #2187, #1811, #1586
- 10:19 PM Linux kernel client Bug #2371 (Can't reproduce): pjd chown/00.t 141, 145, 153 fail on kclient
- ...
- 10:12 PM Linux kernel client Bug #2260: libceph: null pointer dereference at try_write+0x638+0xfb0
- Not sure if that's a problem, need to dig deeper into that, but note that in the OPENING case there we're not jumping...
- 08:00 PM Linux kernel client Bug #2260: libceph: null pointer dereference at try_write+0x638+0xfb0
- Well, after getting my dout() lines to produce useful information
I have a new theory. Basically I think this error... - 07:20 AM Linux kernel client Bug #2260: libceph: null pointer dereference at try_write+0x638+0xfb0
- @Wish I knew about the "pahole" utility (part of the "dwarves" package)
last week.
% pahole libceph.ko -C ceph_co... - 05:20 AM Linux kernel client Bug #2260: libceph: null pointer dereference at try_write+0x638+0xfb0
- Last night I continued to decode assembly, working through making
sense of the particular instructions and the state... - 08:48 PM Revision c3fb68ee (ceph): osd: make output less ugly
- Signed-off-by: Sage Weil <sage@newdream.net>
- 08:47 PM Revision 4d095e95 (ceph): osd: create a 'ready' file on mkfs completion
- Signed-off-by: Sage Weil <sage@newdream.net>
- 08:46 PM Revision 29399eec (ceph): osd: use fsync+rename when writing meta files (during mkfs)
- It's overkill to do the dir fsync on each file, but not worth making
efficient.
Signed-off-by: Sage Weil <sage@newdr... - 08:13 PM Bug #2370 (Resolved): rbd doesn't support mv/rename across pools; tool should detect
- 08:10 PM Bug #2370: rbd doesn't support mv/rename across pools; tool should detect
- commit: ee26c5d73a48b64292d16a87ebe69908c142048e
- 07:30 PM Bug #2370 (Resolved): rbd doesn't support mv/rename across pools; tool should detect
- rbd doesn't support moving an image between RADOS pools; the rbd CLI should detect this case and issue an error. rbd...
- 08:12 PM Revision b9a54b44 (ceph): PG: check_new_interval now handles adding new maps to past intervals
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 08:12 PM Revision bb7e5da3 (ceph): PG: Add probe set to HB peers during GetInfo
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 07:37 PM Revision 06fd0b68 (ceph): filestore: allow flusher+sync_flush enable/disable via injectargs
- This only affects the decision to queue or do things inline, so it is safe
to change while the filestore is up and ru... - 06:30 PM Revision 342e76fb (ceph): filestore: fix op_queue_{len,bytes} instrumentation
- (re)set these in logger when they actually change.
Signed-off-by: Sage Weil <sage@newdream.net> - 05:49 PM Revision 81d2f888 (ceph): Merge branch 'wip-throttle'
- Reviewed-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 05:47 PM Revision 274cd0c6 (ceph): throttle: count get_or_fail success/failure explicitly
- Signed-off-by: Sage Weil <sage@newdream.net>
- 05:39 PM Revision 81f51d28 (ceph): osd: pg creation calc_priors_during() should count primary as up
- If only want to include down osds if *all* of the prior acting osds are
down. If osd->whoami is one of them, then we... - 04:49 PM Bug #2364 (Won't Fix): mon: can't specify monitor to join with -m
- I didn't have any troubles doing any of this with multiple machines so it does appear to only be a problem if you're ...
- 04:01 PM Bug #2364: mon: can't specify monitor to join with -m
- I've investigated this and it only tries to bind to existing monitor addrs if they have a local IP in the list provid...
- 04:39 PM rgw Bug #2369 (Resolved): rgw: bucket attr update does not propagate correctly to all rgw instances
- 04:10 PM Revision 75a107cb (ceph): throttle: note current value and max in perfcounters
- This exposes a snapshot of the current Throttle value and limit.
Signed-off-by: Sage Weil <sage@newdream.net> - 01:49 PM Bug #2365 (Resolved): osd: don't notice when nodes in probe set are down
- commit:bb7e5da322871bfa1ac596411c596f83083b02a6
- 01:46 PM Feature #2323 (Resolved): osd: limit 'old request' messages generated
- 12:05 PM Feature #2323 (Fix Under Review): osd: limit 'old request' messages generated
- 12:39 PM Feature #2358 (Resolved): throttle: expose via perfcounters
- 12:37 PM Bug #2306 (Resolved): objecter: accessing empty object maps to pool 0
- 12:35 PM Bug #2368 (Resolved): injectargs not working for fliestore flusher
- commit:06fd0b68f655376f1c468edacb8655f1c2b163c1
- 12:24 PM Bug #2368 (Resolved): injectargs not working for fliestore flusher
- Sage and I were trying to turn off the filestore flusher and doing a:
ceph osd tell \* injectargs '--no-filestore-... - 12:25 PM Bug #2303 (Can't reproduce): osd: failed to peer on startup
- i'm going to guess this is a dup of #2355, or fixed.. haven't seen it since.
- 10:50 AM Bug #2355 (Resolved): pgs stuck creating (with thrashing)
- commit:81f51d28d67c2a58ab621405c3da65aac726d719
- 10:42 AM Bug #2355 (Fix Under Review): pgs stuck creating (with thrashing)
- see wip-2355
- 10:31 AM Feature #2367 (Resolved): mon: osd crush add should move item if it exists elsewhere
- 09:57 AM rgw Bug #2366 (Resolved): rgw: bucket index update rely on pg state
- Since the object version number is being used to update the bucket index, if we'd backup, remove, restore the pool th...
- 09:41 AM Bug #2357: mds takes down ceph
- Are you running with both of the MDSes active, or just one of them? You'll have better luck with one active and one s...
- 09:33 AM Bug #2363 (Resolved): mon: ./ceph-mon -i b --mkfs -c ceph.conf segfaults
- Merged into master in commit:22bd5dfa25a1b93ccbacf882cea2e55dc8744004
- 03:11 AM Revision c5429bf9 (ceph): use fewer nodes for the simple singleton tasks
- 02:16 AM Revision 1acdc579 (ceph): .gitignore: ceph-kdump-copy
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 12:27 AM Revision 627761f8 (ceph): Merge remote-tracking branch 'gh/wip-ceph-kdump-copy'
- 12:24 AM Revision 18790b1f (ceph): osd: add is_unmanaged_snaps_mode() to pg_pool_t; use more consistently
- Create an is_unmanaged_snaps_mode() function to parallel
is_pool_snaps_mode(), and replace all the checks directly re... - 12:24 AM Revision 22bd5dfa (ceph): pick_address: don't bother checking struct ifaddrs which have a null if...
- I assume that's the localhost interface or similar.
Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com> - 12:22 AM Revision f4150184 (ceph): Merge remote-tracking branch 'gh/wip-2352'
- Reviewed-by: Sage Weil <sage@newdream.net>
04/30/2012
- 11:33 PM Revision 292ff7fd (ceph): msgr: include msgr name in dispatch_throttler name
- Signed-off-by: Sage Weil <sage@newdream.net>
- 11:31 PM Revision be6a9a6c (ceph): msgr: name messengers
- Give each Messenger a logical name describing its role. For instance, the
OSD will have client, cluster, and heartbe... - 11:25 PM Revision 36ac4524 (ceph): throttle: report stats via perfcounter
- Fixes: #2358
Signed-off-by: Sage Weil <sage@newdream.net> - 10:06 PM Revision ccfb6b3a (ceph): global_init: don't fail out if there is no default config.
- There are plenty of scenarios where the user doesn't need a config file.
Instead, just print a warning and let things... - 09:10 PM Revision 845718d9 (ceph): global: fix incorrect CINIT flag.
- There is nobody responding to CLOSE_STDERR, but this block sure looks
like it should be doing so. Fix that!
Signed-o... - 08:55 PM Revision f3771b0e (ceph): throttle: feed cct, name, and add logging
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
Reviewed-by: Greg Farnum <gregory.farnum@dreamhost.com> - 08:54 PM Revision 29de435d (ceph): Merge tag 'v0.46'
- v0.46
- 08:37 PM Revision 74138288 (ceph): osdmap: do no dereference NULL entity_addr_t pointer in addr accessors
- These may be NULL if we expand the addr vectors but haven't ever stored an
address yet. Check for NULL and return a ... - 06:13 PM Revision 4c597fa4 (ceph): OSD: add different config options for map bl caches
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 06:13 PM Revision dcbb8d40 (ceph): osd_recovery: test no* osdmap flags
- 06:12 PM Revision cefaa7d3 (ceph): mon: fix nion -> noin typo
- Thanks Greg!
Signed-off-by: Sage Weil <sage@newdream.net> - 06:12 PM Revision 05bbe141 (ceph): Merge branch 'wip-pi'
- Reviewed-by: Samuel Just <samuel.just@dreamhost.com>
- 05:55 PM Revision 88dda3be (ceph): librbd: use unique error code for image removal failures
- This allows the rbd tool to provide a useful error message, instead of
compounding more possible causes into one erro... - 05:55 PM Revision cb7f1c9c (ceph): v0.46
- 05:25 PM Bug #2345 (Resolved): mon: users can create both pool snapshots and self-managed snapshots on a s...
- 05:09 PM Feature #2358 (Fix Under Review): throttle: expose via perfcounters
- see wip-throttle
- 02:05 PM Feature #2358 (In Progress): throttle: expose via perfcounters
- 04:58 PM Bug #2364: mon: can't specify monitor to join with -m
- It does work if you have a monmap, though (although it's noisy for things like lack of keyrings, admin socket locatio...
- 04:39 PM Bug #2364 (In Progress): mon: can't specify monitor to join with -m
- Of course, at that point the -m is essentially ignored.
If I merge in my no-conf-necessary changes and run without... - 04:09 PM Bug #2364: mon: can't specify monitor to join with -m
- Oh, and if you do specify the fsid in the above step (apparently required, my bad) and try to start up the mon:
<pre... - 03:48 PM Bug #2364 (Won't Fix): mon: can't specify monitor to join with -m
- ...
- 04:48 PM Revision 580b5203 (ceph): run_xfstests.sh: drop #62
- Until #2359 is resolved.
Signed-off-by: Sage Weil <sage@newdream.net> - 04:47 PM Bug #2365 (Resolved): osd: don't notice when nodes in probe set are down
- we currently do
set<int> new_peers;
if (role == 0) {
for (unsigned i=0; i<acting.size(); i++)
n... - 03:46 PM Bug #2363 (Resolved): mon: ./ceph-mon -i b --mkfs -c ceph.conf segfaults
- It tries to identify if there's a local monitor based on the network interfaces, and segfaults inside of have_local_a...
- 02:04 PM Feature #2321 (Resolved): osd: investigate memory consumption from peering backlog
- 01:51 PM Bug #2360 (Resolved): osd: inconsistent use of dirty_info = true vs write_info()
- wip-pi merged into master. may still be some cleanup possible, but that's an optimization at this point.
- 01:16 PM Bug #2356 (Resolved): make dist rebuilds libcommon.la
- Fixed in a477d6be7effd26bd018e6af907ce2d8a691db85
- 11:52 AM Feature #2213 (Resolved): rbd: shouldn't need config file to get help
- Okay, just made it not care about config files in wip-2352.
- 11:52 AM Bug #2352 (Resolved): ceph -s without a conf file doesn't work when it should
- Okay, just made it not care about config files in wip-2352.
- 10:50 AM Feature #2362 (New): rados: support omap (leveldb) and locator key in import/export
- Observed by Henry C Chang: http://marc.info/?l=ceph-devel&m=133575684918035&w=2
Also see #1069 for similar ticket ... - 08:43 AM Linux kernel client Bug #2260: libceph: null pointer dereference at try_write+0x638+0xfb0
- I've completed decoding all of the fields of the ceph_connection
structure, along with some other stuff dereferenced... - 08:36 AM Linux kernel client Bug #2359: xfstest 62 failing
- No I don't think that's the issue.
The problem lies in this function, defined in xfstests "common.attr":
_sort_ge... - 01:51 AM Bug #2316: rbd: restart of OSD leeds to stale qem-VM's with "ceph version 0.45-207-g3053e47"
- Hi Josh, ( sending this second time, "bad browser request..." grrr)
I had some spare-time over the weekend and dec...
04/29/2012
- 05:10 PM Revision 4465c3ab (ceph): osdmap: fix identify_osd() and find_osd_on_ip()
- In 313c1566d3b649ef81fcdc722678d77dccfa888f we switched to using the
get_addr() accessor methods, which assert that t... - 04:03 PM Revision d64e1b97 (ceph): osd: keep pgs locked during handle_osd_map dance
- Currently we drop and retake locks during handle_osd_map calls to
advance_map and activate_map. Instead, take them a... - 04:03 PM Revision 4ae857cb (ceph): mon: drop obsolete osd/PG.h #includes
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 03:11 PM Revision 36b02575 (ceph): osd: fix nested transaction in all_activated_and_committed()
- all_activated_and_committed() is called from _activate_committed(), called
from a objectstore completion, and also fr... - 03:11 PM Revision f750c3f4 (ceph): osd: use dirty flags in activate(), merge_log()
- These are all called from within the state machine, so we can simply set
the dirty flags.
Signed-off-by: Sage Weil <... - 03:11 PM Revision d48549d5 (ceph): osd: set dirty flags on rewind_divergent_log
- Make sure we record any rewind_divergent_log. In the activate case, this
will happen anyway, but mark it dirty here ... - 03:11 PM Revision 3e84ce86 (ceph): osd: use PG::write_if_dirty() helper
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 10:28 AM Bug #2361 (Resolved): osd: failed exists() assertion for identify_osd()
- 10:05 AM Bug #2361 (Resolved): osd: failed exists() assertion for identify_osd()
- 2012-04-29T01:43:43.581 INFO:teuthology.task.ceph.mds.a.err:osd/OSDMap.h: In function 'const entity_addr_t& OSDMap::g...
- 05:32 AM Revision fc8ce16a (ceph): osd: do not merge history on query
- We shouldn't modify the local notion of the history without recording it to
disk. And we (probably) also don't need ... - 05:31 AM Revision ff9bce97 (ceph): osd: dirty_info if history.merge updated anything
- In proc_replica_info and proc_primary_info, we may or may not update
the pg_info_t. If we do, set dirty_info, so tha... - 05:17 AM Revision a13672f8 (ceph): osd: write dirty info on handle info, notify, log
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 03:58 AM Revision 7fd33607 (ceph): osd: skip scrub scheduling if we aren't up
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 03:57 AM Revision 0c860c17 (ceph): osd: fix dirty_info check for advance/activate paths
- Previously we would check and write dirty_info *without the pg lock* after
doing the advance and activate map calls. ... - 01:18 AM Revision bf4966cf (ceph): Merge branch 'next'
- 01:15 AM Revision 36bfdd86 (ceph): crushtool: fix clitest now that --force is gone
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 01:12 AM Revision 8af46a4a (ceph): run_seed_to.sh: clean out merge cruft
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 12:10 AM Revision ebdf80ed (ceph): crush: remove parent maps
- These were used (poorly) for forcefeeding, but they are useless now. Which
is good, because we allow items to appear... - 12:10 AM Revision ed974b50 (ceph): crush: remove forcefeed from crush_do_rule
- Remove forcefeed functionality from CRUSH. This is an ugly misfeature that
is mostly useless. Remove it.
Signed-of...
04/28/2012
- 11:27 PM Revision c4c59a09 (ceph): log: do not set on_exit() callback for libraries
- Set this up in either global_init() or common_init_finish(), both opportune
times that occur after config parsing has... - 11:25 PM Revision c24c9e3a (ceph): Merge remote branch 'gh/wip-filestore-misc'
- Conflicts:
src/test/filestore/run_seed_to.sh - 11:00 PM Bug #2360: osd: inconsistent use of dirty_info = true vs write_info()
- void PG::all_activated_and_committed()
creates and submits its own transaction. it's called from _activate_commit... - 09:40 PM Bug #2360 (Resolved): osd: inconsistent use of dirty_info = true vs write_info()
- The dirty_info flag was added ages ago to avoid stuff done during advance*N + activate_map from rewriting the pg info...
- 10:53 PM Revision 6bb3e841 (ceph): Merge remote branch 'gh/wip-2353'
- Reviewed-by: Samuel Just <samuel.just@dreamhost.com>
- 10:49 PM Revision 254644a4 (ceph): osd: always share past_intervals
- Share past intervals when starting up new replicas. This can happen via
an MOSDPGInfo or an MOSDPGLog message.
Fix ... - 10:25 PM Revision 5047cac0 (ceph): Merge branch 'wip-osdmap'
- Conflicts:
src/mon/PGMonitor.cc
src/osd/OSDMap.h - 09:52 PM Revision 352247e1 (ceph): fix file_layout.sh layouts test
- preferred_osd is not gone.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 09:48 PM Revision c97c20de (ceph): Merge branch 'wip-mon'
- Reviewed-by: Gregory Farnum <gregory.farnum@dreamhost.com>
- 09:48 PM Revision e205e11c (ceph): mon: 'osd [un]set noin'
- Missed this one.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 09:47 PM Revision c661e66c (ceph): Merge branch 'next'
- 06:14 PM Feature #2334 (Resolved): mon: set max mark-out or mark-down
- 06:14 PM Feature #2317 (Resolved): mon: pause/unpause auto-mark-out
- 06:14 PM Feature #2318 (Resolved): mon: block osd boot
- 06:14 PM Feature #2319 (Resolved): mon: block osd mark-down
- 05:50 PM Bug #2350 (Resolved): conf: can't set subsystem settings (debug levels, logging settings) via lib...
- 05:49 PM Bug #2322 (Resolved): osd/ReplicatedPG.cc: 3832: FAILED assert(!object_contexts.size())
- commit:92becb696bde7f0aa9687b2fe7505ed1ac9f493b
- 04:39 PM Bug #2353 (Resolved): osd: current/ snap check problem
- 03:41 PM Linux kernel client Bug #2359 (Can't reproduce): xfstest 62 failing
- > 324: (3320s) collection:basic clusters:fixed-3.yaml fs:btrfs.yaml tasks:rbd_xfstests.yaml
> Command failed wit... - 02:46 PM Revision 12d1675c (ceph): osd: fix check for whether to recalculate past_intervals
- We may not recalculate all the way back to last_interval_clean due to
the oldest_map floor. Figure out what we want ... - 02:46 PM Revision db8e20b2 (ceph): osd: include past_intervals in pg debug printout
- Signed-off-by: Sage Weil <sage@newdream.net>
- 02:46 PM Revision 0c691403 (ceph): osd: only generate missing intervals in generate_past_intervals
- We can (currently) get into a situation where we don't have the full
history back to last_epoch_clean because non-pri... - 02:46 PM Revision 7e8ab0f2 (ceph): osd: share past intervals with notifies
- Send past_intervals along with pg_info_t on every notify. The reasoning
here is as follows:
- we already have the ... - 02:46 PM Revision 0c65ac6f (ceph): osd: drop useless PG::fulfill_info()
- There is a nice symmetry there with fulfill_log(), but it is a short
function with a single caller that mostly just f... - 02:46 PM Revision 944a4311 (ceph): osd: fill in past intervals during advance_map
- If ceph-osd is way behind, we will advance through past maps before we
mark ourselves up. This avoids the slow recal... - 02:46 PM Revision c971545a (ceph): osd: set dirty_info in generate_past_intervals
- This ensures that we save our work.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 02:46 PM Revision 90dae62b (ceph): osd: PG::Interval -> pg_interval_t
- Signed-off-by: Sage Weil <sage@newdream.net>
- 02:46 PM Revision 924a1251 (ceph): Merge branch 'next' into t
- 02:45 PM Revision f922dc43 (ceph): Stop rebuild of libcommon.la on "make dist"
- Fixes: 2356
Reviewed-by: Josh Durgin <josh.durgin@dreamhost.com> - 02:45 PM Revision e44b126c (ceph): mon: limit size of MOSDMap message sent as reply
- We may send an MOSDMap as a reply to various requests, including
- a failure report
- a boot message
- a pg_temp ... - 02:45 PM Revision d1df320b (ceph): ceph-object-corpus: revert rewind
- From 92becb696bde7f0aa9687b2fe7505ed1ac9f493b
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 02:29 PM Feature #2358: throttle: expose via perfcounters
- see wip-throttle
- 01:58 PM Feature #2358 (Resolved): throttle: expose via perfcounters
- it would be nice to track all throttlers for
- current utilization
- peak utilizatoin
- times they've actuall... - 11:27 AM Bug #2357 (Can't reproduce): mds takes down ceph
- Hi guys !
Really impressive DFS... nice performance... cool stuff !
But while starting to use it live I unfortunate... - 04:52 AM Revision 4274fd05 (ceph): osdmap: fix addr dedup check
- Compare *every* address for a match, or else note that it is (or might be)
different. Previously, we falsely took di... - 04:48 AM Revision 06d1bc22 (ceph): osd: fix bad map debug messages
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 01:32 AM Revision a477d6be (ceph): Stop rebuild of libcommon.la on "make dist"
- Fixes: 2356
Reviewed-by: Josh Durgin <josh.durgin@dreamhost.com>
04/27/2012
- 11:05 PM Revision 510eed0f (ceph): filestore: fix error message
- error message was misleading, fixing it.
Signed-off-by: Yehuda Sadeh <yehuda.sadeh@dreamhost.com> - 10:46 PM Revision f03dc34f (ceph): filestore: first lock osd mount point, next detect fs type
- Fixes #2353. Problem was that there were (at least) two osd processes
that were racing for the fs detection, which tr... - 09:29 PM Revision 10c616a5 (ceph): OSD: use map bl cache pinning during handle_osd_map
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 09:28 PM Revision d0d69125 (ceph): simple_cache.hpp: add pinning
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 09:23 PM Linux kernel client Bug #2260: libceph: null pointer dereference at try_write+0x638+0xfb0
- Revisiting my decoding from before, in light of my new clarity on
padding and field layout, things in the message st... - 08:14 PM Linux kernel client Bug #2260: libceph: null pointer dereference at try_write+0x638+0xfb0
- Clarifying something I described earlier, with respect to alignment
and sizing of fields, including those that have ... - 07:27 AM Linux kernel client Bug #2260: libceph: null pointer dereference at try_write+0x638+0xfb0
- The skipped 7 bytes could be explained this way.
The sender sends out over the wire the sub-components of the
str... - 07:13 AM Linux kernel client Bug #2260: libceph: null pointer dereference at try_write+0x638+0xfb0
- Here is my hunch about what's going on, and what I mentioned was a
cause for trouble interpreting the memory content... - 07:02 AM Linux kernel client Bug #2260: libceph: null pointer dereference at try_write+0x638+0xfb0
- After some false starts I decoded the beginning of the connection
structure, and I found a problem, but I have not y... - 05:51 AM Linux kernel client Bug #2260: libceph: null pointer dereference at try_write+0x638+0xfb0
- I meant to update this yesterday morning... KDB on the system definitely
works, which is really great. But since i... - 09:00 PM Revision 8ce51551 (ceph): Merge branch 'next'
- 08:58 PM Revision 92becb69 (ceph): FileJournal: simply flush by waiting for completions to empty
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 06:46 PM Revision 155700d6 (ceph): PG: in GetInfo Notify handler, fix peer_info_requested filter
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 05:47 PM Bug #2356 (Resolved): make dist rebuilds libcommon.la
- Sage noticed this and asked me to investigate; it turns out a _SOURCES macro was mistakenly including
libcommon.la r... - 03:38 PM Bug #2353: osd: current/ snap check problem
- Problem was that more than one daemon was brought up for the same osd, racing for the fs type check before mounting i...
- 03:00 PM Bug #2353: osd: current/ snap check problem
- ...
- 02:50 PM Bug #2353: osd: current/ snap check problem
- I'd say the error message is bad:...
- 09:21 AM Bug #2353 (Resolved): osd: current/ snap check problem
- 2012-04-27 12:18:08.015605 7f281c7c17a0 0 filestore(/var/lib/ceph/osd/ceph-153) mount found snaps <42996,42997,42998...
- 01:51 PM Bug #2355 (Resolved): pgs stuck creating (with thrashing)
- I was running the following teuthology config, and since test_librbd_fsx creates a pool on each run, new pgs were cre...
- 01:29 PM Bug #2354 (Resolved): osd: make watch timeout configurable
- The *notify* timeout is configurable, but the *watch* timeout is still hardcoded as 30 seconds. From ReplicatedPG.cc:...
- 11:34 AM Bug #2352 (In Progress): ceph -s without a conf file doesn't work when it should
- I've got a stupid hack that makes this work, but I'm going to see if I can generalize our common config stuff so this...
- 11:34 AM Feature #2213 (In Progress): rbd: shouldn't need config file to get help
- This is actually a little different from #2352. But keeping both in mind is better than providing a specific hack, so...
- 10:49 AM Bug #2316: rbd: restart of OSD leeds to stale qem-VM's with "ceph version 0.45-207-g3053e47"
- There was a fix to librbd's handling of short reads that span objects (commit:b94d6a6cf459478b27539eb8eccd30e19a67bbc...
- 09:06 AM Bug #2316: rbd: restart of OSD leeds to stale qem-VM's with "ceph version 0.45-207-g3053e47"
- Hi Josh,
well, just allowed myself another time-window for some investigations. Do you have some pointers to the f... - 04:57 AM Revision 05b4fb33 (ceph): Merge branch 'wip-lpg'
- Conflicts:
src/osd/OSDMap.h - 04:53 AM Revision cee218f0 (ceph): Merge branch 'next'
- 04:52 AM Revision dbd99129 (ceph): librados: test get/set of debug levels
- Also do some sanity checks on the subsystem log level settings.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 04:52 AM Revision 4e2e8794 (ceph): config: allow {get,set}_val on subsystem debug levels
- This mimics the allows you to get and set subsystem debug levels via the
normal config access methods. Among other t... - 02:41 AM Revision ec1ea6a8 (ceph): common/: added templated simple lru implementations
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 02:41 AM Revision 7f3790a9 (ceph): OSD.cc: track osdmap refs using an LRU
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 01:49 AM Revision d6359d44 (ceph): osdmap: dedup addrs and addr vectors between maps
- Compare two maps. If an addrs matches, share the reference. If all
addrs match, share the entire vector.
This lead... - 01:49 AM Revision e0436cb9 (ceph): osdmaptool: kludge to load a range of maps into memory
- Signed-off-by: Sage Weil <sage@newdream.net>
- 01:49 AM Revision 98b1d8f3 (ceph): osdmap: use shared_ptr for CrushWrapper
- Signed-off-by: Sage Weil <sage@newdream.net>
- 01:49 AM Revision 36d43825 (ceph): osdmap: dedup crush map
- If the encoded crush map is identical between two versions, share the
reference.
Signed-off-by: Sage Weil <sage@newd... - 01:49 AM Revision 2a465641 (ceph): buffer: make contents_equal() more efficient
- Iterate both lists in parallel in terms of buffers, and use memcmp() to
do the comparison.
Signed-off-by: Sage Weil ... - 01:49 AM Revision 4cfbd81c (ceph): osdmap: drop obsolete PG_ROLE_* constants
- There are cruft from the old primary/chain/splay replication code. All
current code says <0 is stray, 0 is primary, ... - 01:49 AM Revision 0188d9bb (ceph): osd: dedup osdmaps when added to the in-memory cache
- When we add an OSDMap to our in-memory cache, dedup against an existing map
at a nearby epoch.
Signed-off-by: Sage W... - 01:49 AM Revision 207eec65 (ceph): osd: make map dedup optional
- On by default. This trades CPU for memory. Some might have unlimited RAM
and not care.
Signed-off-by: Sage Weil <s... - 01:49 AM Revision ed1024fb (ceph): osdmap: use shared_ptr<> for pg_temp
- This will let us dedup later.
Signed-off-by: Sage Weil <sage@newdream.net> - 01:49 AM Revision 873e9bee (ceph): osdmap: dedup pg_temp
- We only deal with the case where the entire map is identical, since the
individual items are too small to make the po... - 12:54 AM Revision 8a3fd449 (ceph): Merge branch 'next'
- 12:42 AM Revision ee541c0f (ceph): osdmap: filter out nonexistent osds from map
- It is possible that the crush map contains device ids that do not exist as
osds. Filter them out of the CRUSH result... - 12:41 AM Revision 8f4dba62 (ceph): librbd: the length argument of aio_discard should be uint64_t
- size_t was accidentally copy-pasted.
Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com> - 12:40 AM Revision fe76c5ba (ceph): filestore: interprect any fiemap error as EOPNOTSUPP
- On 2.6.32-5-amd64 (debian) and XFS I'm getting EINVAL.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
04/26/2012
- 11:55 PM Revision 06af38ac (ceph): osdmap: uninline a bunch of stuff
- This will conflict with wip-lpg, yay.
Signed-off-by: Sage Weil <sage@newdream.net> - 11:55 PM Revision 313c1566 (ceph): osdmap: use shared_ptr for addrs, addr vectors
- We share a lot of identical addresses between map versions because they
don't tend to change very often. Instead of ... - 11:50 PM Revision 10914cab (ceph): mon: consider pending_inc in {up,in}_ratio for can_mark_{out,down}()
- Consider pending changes when calculating the current up/in ratios. Among
other things, this will make the marking o... - 11:50 PM Revision 2a22ff43 (ceph): mon: thrash pg_temp mapping, too
- Signed-off-by: Sage Weil <sage@newdream.net>
- 11:36 PM Revision 6910d838 (ceph): filestore: fix a journal replay issue with collection_add()
- Signed-off-by: Joao Eduardo Luis <jecluis@gmail.com>
- 11:31 PM Revision 96108c65 (ceph): filestore: fix a journal replay issue with collection_add()
- Signed-off-by: Joao Eduardo Luis <jecluis@gmail.com>
- 11:03 PM Revision ead5d2a8 (ceph): osd: filter osds removed from probe set from peer_info_requested
- Peef_info_requested should be a strict subset of the probe set. Filter
osds that are dropped from probe from peer_in... - 10:53 PM Revision 201bace5 (ceph): Merge branch 'next'
- 10:51 PM Feature #2314 (Resolved): remove localized pgs
- 10:45 PM Revision 3e880174 (ceph): PG: get_infos() should not post GotInfo
- The MNotifyRec handler also posts GotInfo under the same conditions
after calling get_infos().
Signed-off-by: Samuel... - 10:40 PM Bug #2350: conf: can't set subsystem settings (debug levels, logging settings) via librados or li...
- commit:4e2e87941b5f078eadf90a2c58a12765375ac66b, tested by commit:dbd99129ce58b321c0ecd641a1f8b45bd94b3b33
- 09:55 AM Bug #2350: conf: can't set subsystem settings (debug levels, logging settings) via librados or li...
- 10:38 PM Revision 7fe45fd6 (ceph): Revert "PG: whitelist MNotifyRec in started"
- This reverts commit 9579365720818125a4b15741ae65e58948b9c69f.
- 10:21 PM Bug #2351 (Resolved): osd: bad state machine event
- commit:3e880174dd233a3df88c63785186d36f9b12a137
- 09:43 AM Bug #2351: osd: bad state machine event
- i saw this on my dev copy, might be the same...
- 09:42 AM Bug #2351 (Resolved): osd: bad state machine event
- ...
- 07:41 PM Revision cbe795a7 (ceph): test_librbd: rollback when mapped to a snapshot should fail
- Rollback is effectively a write, and returns -EROFS when mapped to a
snapshot since 3ef3ab8a15b4a80a340ac6039f3957382... - 07:41 PM Feature #2323 (In Progress): osd: limit 'old request' messages generated
- 07:40 PM Feature #2334 (Fix Under Review): mon: set max mark-out or mark-down
- 07:19 PM Revision f873a771 (ceph): workload_generator: get rid of our lock.
- We don't need the lock in the WorkloadGenerator class. Everything that does
need a lock is handled by TestFileStoreSt... - 07:18 PM Revision 436f5d65 (ceph): TestFileStoreState: make 'm_in_flight' var an atomic_t.
- This allows us to increase, decrease and retrieve its value without the
need to lock the class.
Signed-off-by: Joao ... - 06:55 PM Bug #2342 (Resolved): librados: notify deadlock
- Yes. RadosClient::watch_notify() doesn't call ->notify() .. -> _notify_ack() directly anymore.
Fixed, commit:70f7... - 06:31 PM Bug #2342: librados: notify deadlock
- is this fixed with the unwatch change?
- 05:51 PM Revision 841cfcfd (ceph): Merge branch 'next'
- 05:51 PM Revision 95793657 (ceph): PG: whitelist MNotifyRec in started
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 05:38 PM Revision be9b38ec (ceph): RefCountedObject: fix constructor warning
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 04:02 PM Feature #2349: rados bench: Extra statistics
- Update: Also, please include a timestamp for each line of output. This can be in addition to or as a replacement for...
- 04:00 PM Revision 35dc2dea (ceph): workload_generator: specify number of ops to run, or 0 to run forever.
- New option '--test-num-ops VAL' -- if (VAL == 0) then run forever; fi
Signed-off-by: Joao Eduardo Luis <jecluis@gmai... - 03:58 PM Revision 823afcdf (ceph): workload_generator: Delegate store tracking to TestFileStoreState.
- We had a lot of duplicate code between the WorkloadGenerator and the
TestFileStoreState classes, and the last one is ... - 03:29 PM Revision 3903b5a6 (ceph): TestFileStoreState: Fix issues affecting proper behavior when inherited.
- Fix wait_for_ready() and make the C_OnFinished class' member variables
protected instead of private (to allow proper ... - 03:26 PM Revision 22ade4ae (ceph): Makefile.am: test_filestore_workloadgen doesn't need gtests lib.
- Signed-off-by: Joao Eduardo Luis <jecluis@gmail.com>
- 02:25 PM Feature #2290 (In Progress): ObjectCacher: handle read/write errors
- 02:16 PM Bug #2316: rbd: restart of OSD leeds to stale qem-VM's with "ceph version 0.45-207-g3053e47"
- There were a couple bugs fixed that may have caused this problem. Could you try upgrading to the 'next' branch (at co...
- 10:40 AM Feature #2213 (Duplicate): rbd: shouldn't need config file to get help
- 09:58 AM Bug #2352 (Resolved): ceph -s without a conf file doesn't work when it should
- ...
04/25/2012
- 10:23 PM Revision 81a248a1 (ceph): Merge branch 'wip-2342'
- 10:22 PM Revision 39f99354 (ceph): RefCountedObject: relocate from msg/Message.h to common/RefCountedObj.h
- Following a popular request.
Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> - 09:14 PM Revision 70f70d80 (ceph): librados: call notification under different thread context
- This fixes #2342. We shouldn't call notify on the dispatcher
context. We should also make sure that we don't hold
the... - 08:43 PM Revision 8bc818c4 (ceph): mon: 'osd thrash <num epochs>'
- Thrash the osdmap for N iterations. Randomly mark OSDs up, down, in, out,
and up_thru in order to generate a difficu... - 08:07 PM Revision fa984798 (ceph): osd: filter osds removed from probe set from peer_info_requested
- Peef_info_requested should be a strict subset of the probe set. Filter
osds that are dropped from probe from peer_in... - 07:05 PM Bug #2350 (Resolved): conf: can't set subsystem settings (debug levels, logging settings) via lib...
- With the recent rework of the logging infrastructure, md_config_t::set_val() was not updated to take into account sub...
- 06:15 PM Revision f022a949 (ceph): mon: add 'mon osd min up ratio' and 'mon osd min in ratio'
- Prevent the monitor from marking osds down or out when too many are already
in that state. At this point the cluster... - 06:15 PM Revision ba1d3b1d (ceph): mon: use can_mark_*() helpers
- So we can generalize beyond NO* flags. We'll soon be adding other reasons
to not mark things up/down/in/out. This l... - 03:27 PM Revision 75ccd817 (ceph): DeterministicOpSequence: add 'ceph_asserts()' where we expect != NULL.
- Signed-off-by: Joao Eduardo Luis <jecluis@gmail.com>
- 03:26 PM Revision 19664020 (ceph): TestFileStoreState: distinguish between 'get_coll()' and 'get_coll_at()'
- get_coll_at(int pos) should return the collection at the map's position
'pos', but 'pos' was being used as a map key.... - 02:35 PM Revision 44dafc87 (ceph): run_seed_to.sh: Add valgrind support.
- Signed-off-by: Joao Eduardo Luis <jecluis@gmail.com>
- 02:34 PM Revision 4430c01c (ceph): TestFileStoreState: free memory on terminus.
- So far, it hasn't triggered any segfault, but I'm not yet convinced there
is no problem whatsoever.
Signed-off-by: J... - 01:57 PM Feature #2321 (In Progress): osd: investigate memory consumption from peering backlog
- 01:57 PM Feature #2281 (Resolved): build big burnupi cluster for testing
- 01:45 PM Feature #2349 (Resolved): rados bench: Extra statistics
- * min/max for throughput, and std deviations for both throughput and latency.
* timestamp each line of output
* hea... - 01:42 PM Feature #2111 (Fix Under Review): msgr workloads
- This has been sitting in my issue area for a while and it's distracting, so I'm reassigning for review. :)
- 01:42 PM Bug #2348 (Resolved): osd: peer_info_requested not pruned with prior_set.probe
- commit:9023aedf143aa22c14f7a73b73837080b455942d
- 01:41 PM Bug #2345 (Fix Under Review): mon: users can create both pool snapshots and self-managed snapshot...
- Pushed a fix (and tested it using rados snap create and rbd snap create) to wip-2345-snaps.
- 01:14 PM Bug #2345 (In Progress): mon: users can create both pool snapshots and self-managed snapshots on ...
- Ah, there are Monitor checks for this but they aren't quite right: they look at the contents of snaps and removed_sna...
- 11:42 AM Bug #2345 (Resolved): mon: users can create both pool snapshots and self-managed snapshots on a s...
- When the OSD gets the map it breaks, but the Monitor will happily set both. There should be guard against that.
(See... - 01:33 PM rbd Cleanup #2347 (Resolved): The rbd help text is misleading on required arguments
- The help text has a lot of arguments that look optional (being enclosed by brackets or angle brackets) that are actua...
- 01:29 PM Feature #2336 (Fix Under Review): qemu: wire up discard
- See wip-discard in qemu-kvm.git. This works with ide and scsi, but qemu doesn't have trim support for virtio right no...
- 01:02 PM Bug #2316 (In Progress): rbd: restart of OSD leeds to stale qem-VM's with "ceph version 0.45-207-...
- 12:35 PM Bug #2346 (Resolved): xfs filesystem on top of rbd volume corrupts
- Ceph version: 0.44.1-1~bpo70+1
Kernel version: 3.2.12-1
Ceph config:
[global]
auth supported = cephx
keyring =... - 10:35 AM Linux kernel client Bug #2260: libceph: null pointer dereference at try_write+0x638+0xfb0
- Yay!!! I reproduced this with KDB enabled. Now to start poking around on the
machine to see what I can learn. - 12:52 AM Revision 25114bf9 (ceph): nuke: refactor to run in parallel and add unlock option
- nuke-on-error already did this, but now teuthology-nuke does it
too. Also outputs targets that couldn't be nuked at t... - 12:48 AM Revision b32b693a (ceph): parallel: obey iterator protocol
- Once it raises StopIteration, it must continue to do so on subsequent calls to next().
04/24/2012
- 11:44 PM Revision 4bfcbe6a (ceph): mon: decode old PGMap Incrementals differently from new ones
- We need to distinguish between the old 0 (meaning undefined) and
the new 0 (meaning switch to 0 and disable the flags... - 10:47 PM Revision 2673875f (ceph): mon: do not mark booting osds in if NOIN flag is set
- If the NOIN osdmap flag is set, do not mark booting osds in. Normally
we would for a range of reasons (always, new, ... - 10:47 PM Revision 59957da2 (ceph): mon: do not mark osds out if NOOUT flag is set
- Do not mark down osds out when NOOUT flag is set. This is more or less
equivalent to setting a very long 'mon osd do... - 10:28 PM Revision 9ff535ad (ceph): mon: always remove booting osds from down_pending_out
- The down_pending_out tracks OSDs that are down that we may want to
auto-mark out. If an osd boots, it should be remo... - 09:31 PM Revision f0773863 (ceph): osd: do not attempt to boot if NOUP
- If NOUP is set, do not send the boot message.
We already send onetime subscriptions to the osdmap, so we will find o... - 09:31 PM Revision addfb2c6 (ceph): mon: prevent osd mark-down with NODOWN flag
- If the NODOWN osdmap flag is set,
- ignore osd failure reports
- do not mark osds down due to lack of osd/pg stats... - 09:16 PM Revision d84255ec (ceph): mon: prevent osd from booting if NOUP
- Do not add an osd attempting to boot to the map if NOUP is sent. Instead,
send it the latest osdmap so it knows that... - 09:10 PM Revision d3f55dd9 (ceph): mon: 'osd {set,unset} {noin,noout,noup,nodown}'
- Move the set/unset flag code into a helper, and also use that for the
pause/unpause commands.
Signed-off-by: Sage We... - 09:10 PM Revision 6003325d (ceph): osdmap: add NOUP, NODOWN, NOIN, NOOUT flags
- These prevent OSDs from being marked up, down, in, or out, respectively.
Signed-off-by: Sage Weil <sage.weil@dreamho... - 08:59 PM Revision 34ef3f37 (ceph): Merge remote branch 'origin/wip-rbd-snapid' into next
- Reviewed-by: Sage Weil <sage.weil@dreamhost.com>
- 08:39 PM Revision e51772ca (ceph): librbd: pass errors removing head back to user
- In particular, the OSD may return EBUSY if there are still watchers.
Ignore ENOENT, as that may indicate we are clean... - 07:03 PM Feature #2336: qemu: wire up discard
- Trying to get gitbuilders to work again so I can test with packages...
- 12:17 PM Feature #2336 (In Progress): qemu: wire up discard
- 06:04 PM Feature #1937 (Resolved): teuthology: --unlock option for -nuke
- Added in teuthology commit:25114bf9a4e2e237e0d3cb8fcf77b66d4ff87234.
- 12:16 PM Feature #1937 (In Progress): teuthology: --unlock option for -nuke
- 05:55 PM Revision ccf7d930 (ceph): mon: clean up handle_osd_timeouts a bit
- Signed-off-by: Sage Weil <sage@newdream.net>
- 05:55 PM Revision 36ffed45 (ceph): mon: fix pg stats timeout
- We clear out the osd entry when an osd goes up or down. Thus, if we find
it missing from an up osd, we should start ... - 05:49 PM Revision 7b832f42 (ceph): mon: fix whitespace
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 05:35 PM Revision 2b302015 (ceph): mon: fix pgmonitor ratio commands
- The indices were set incorrectly when I whipped thi sup. That's what
you get for not testing nor being careful enough... - 05:07 PM Bug #2286 (Resolved): mon: different full/near_full values on different monitors
- 03:58 PM Bug #2286 (Fix Under Review): mon: different full/near_full values on different monitors
- Actually I broke out the ENCODE_START bit into #2344.
So Sage, you want to merge the new wip-2286-map-fix into nex... - 02:21 PM Bug #2286 (In Progress): mon: different full/near_full values on different monitors
- Sage points out we need to convert the PGMap (and Incremental) encode/decode to use the auto-versioning stuff. I'll p...
- 01:37 PM Bug #2286 (Fix Under Review): mon: different full/near_full values on different monitors
- wip-2286-map-fix has the commit to fix this.
I couldn't come up with a great test for it but I did create a cluster ... - 01:08 PM Bug #2286: mon: different full/near_full values on different monitors
- Okay, pretty sure this is a result of a pre-upgrade PGMap::Incremental being committed by post-upgrade code. Sage poi...
- 09:43 AM Bug #2286 (In Progress): mon: different full/near_full values on different monitors
- 04:58 PM Feature #2319 (Fix Under Review): mon: block osd mark-down
- 10:21 AM Feature #2319 (In Progress): mon: block osd mark-down
- 04:58 PM Feature #2317 (Fix Under Review): mon: pause/unpause auto-mark-out
- 10:21 AM Feature #2317 (In Progress): mon: pause/unpause auto-mark-out
- 04:58 PM Feature #2318 (Fix Under Review): mon: block osd boot
- 10:21 AM Feature #2318 (In Progress): mon: block osd boot
- 03:57 PM Revision e17b5a85 (ceph): librbd: clarify handle_sparse_read condition
- The earlier condition is >. != means < at this point, and the nesting
is unnecessary.
Signed-off-by: Josh Durgin <jo... - 03:57 PM Revision 3ef3ab8a (ceph): librbd: clean up snapshot handling a bit
- * snapid should determine whether our mapped snapshot is gone, not snapname
* snap_set(<nonexistent_snap>) shouldn't ... - 03:57 PM Revision 7add136f (ceph): librbd: reset needs_refresh flag before re-reading header
- This way we can't miss an update if we get a notify during ictx_refresh.
Specifically, a race like this:
Thread 1 ... - 03:57 PM Revision d28f850f (ceph): test_rbd: add tests for snap_set and more complicated resizing
- * snap_set to a deleted (and recreated) snapshot
* resizing down (truncating) and back up
* resizing to non-object-al... - 03:54 PM Bug #2338: mon: adding new monitors simultaneously can allow a new mon to become leader
- Okay, looks like the new Monitor isn't joining properly because in this case it's been fed a monmap that contains its...
- 03:51 PM Bug #2267: Ceph client crashed after shutting down one mds and osd
- One more update. Maciej reports that after restoring the server again,
and restarting things (this again on the pro... - 12:52 PM Bug #2267: Ceph client crashed after shutting down one mds and osd
Staging cluster:
- running 0.45 on node n2cc: mds.n2cc, osd.1
- running 0.44 on node cc: mon.cc, mds.cc, osd.0
...- 12:29 PM Bug #2267: Ceph client crashed after shutting down one mds and osd
- Reportedly, with the ceph configuration defined herein, the problem
shows up after simply restarting one of the osd ... - 12:15 PM Bug #2267: Ceph client crashed after shutting down one mds and osd
- This could be in the same family as other problems in con_work() we've
seen lately (for example, 2260). Hoping a si... - 03:26 PM Cleanup #2344 (Resolved): convert Monitor maps to use ENCODE_START
- Looks like the Monitor stuff doesn't use the ENCODE_START idiom at all, and it should.
- 02:01 PM Bug #2339: osd: EBUSY on object delete when watchers present
- see #2343
- 01:49 PM Bug #2339: osd: EBUSY on object delete when watchers present
- After some discussion, it seems like the right path is to:
- associate an error code with notify events (e.g., 0 =... - 06:44 AM Bug #2339 (Resolved): osd: EBUSY on object delete when watchers present
- in particular, you can't delete an rbd image that has watchers. that may be okay if there is someone actively mounti...
- 01:52 PM Fix #2343 (Resolved): librados, osd: notify with special type on object deletion
- If a watched object is deleted, it would be useful to have a different type of notify sent to the watchers, letting t...
- 01:37 PM Bug #2342: librados: notify deadlock
- This bug was seen repeatedly on the 3 test clusters built for congress. After several hours of testing rgw would hang...
- 01:15 PM Bug #2342 (Resolved): librados: notify deadlock
- ...
- 10:21 AM Feature #2334 (In Progress): mon: set max mark-out or mark-down
- 10:03 AM Bug #2341 (Resolved): mon: pgstats timeout still broken
- commit:2fa90e02711100170423565302ec07b68d5e0aef
- 09:48 AM Bug #2341 (Resolved): mon: pgstats timeout still broken
- 2012-04-24 13:22:47.888291 7fa5bc587700 mon.peon5752@0(leader).osd e21633 OSDMonitor::handle_osd_timeouts: never got ...
- 07:08 AM Bug #2311: rbd: delete + create image led to EEXIST
- Hi Sage,
yeah, this time might have been from a stale VM, but the other tests should have shown, that I normally s... - 06:45 AM Bug #2311 (Resolved): rbd: delete + create image led to EEXIST
- Ah, the problem is that the rbd head object has watchers (it is mounted) and the delete request returned EBUSY, but l...
- 12:51 AM Bug #2311: rbd: delete + create image led to EEXIST
- Well, here we go with some output:
rbd create --size 2048 data/906-testdisk.rbd
create error: 2012-04-24 09:36:01... - 07:01 AM RADOS Subtask #2340 (Resolved): crush: remove forcefeeding
- 07:01 AM Feature #2314: remove localized pgs
- did another test that included broken lpgs (wouldn't create) and verified things behaved. merged the first piece of ...
- 03:31 AM Revision 2bdaba4f (ceph): run_seed_to.sh: rework the script, make it more flexible and broaden th...
- Allow for '-h' and other options such as disabling the journal sync tests,
defining it is to be run on a btrfs FS, en... - 02:43 AM Revision e65b7971 (ceph): librbd: rev version for discard addition
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
04/23/2012
- 09:48 PM Revision 198544ad (ceph): mds: remove preferred from ceph_file_layout
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 09:48 PM Revision b3cdc21a (ceph): cephfs: remove preferred osd setting
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 09:48 PM Revision f01b6dd5 (ceph): test_ioctls: remove preferred osd
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 09:48 PM Revision a5143444 (ceph): mon: ignore/remove localized pgs
- This will trigger on the next OSDMap update.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 09:48 PM Revision 771fd05b (ceph): mon: set auid for mon-created pools to 0
- Signed-off-by: Sage Weil <sage@newdream.net>
- 09:48 PM Revision 637de4d7 (ceph): mon: load CompatSet features on startup
- Signed-off-by: Sage Weil <sage@newdream.net>
- 09:48 PM Revision 19ba3475 (ceph): osdmaptool: fix clitests for lack of localized pgs
- Signed-off-by: Sage Weil <sage@newdream.net>
- 09:48 PM Revision 94adf5d9 (ceph): osd: remove localized pgs from pg_pool_t
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 09:48 PM Revision 0138a764 (ceph): osd: ignore localized pgs
- - do not load them on startup
- ignore any we hear about over the wire
Signed-off-by: Sage Weil <sage.weil@dreamhost... - 09:48 PM Revision b8f4acfb (ceph): osd: remove preferred from object_locator_t
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 09:48 PM Revision f164b877 (ceph): osdmap: do not forcefeed preferred osd to crush
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 09:48 PM Revision 6d1344c3 (ceph): libcephfs: disable ceph_set_default_preferred_pg
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 09:48 PM Revision 21ef9795 (ceph): client: rip out preferred_pg thing
- This wasn't even named properly. Blech.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 08:58 PM Revision 07776136 (ceph): Merge remote-tracking branch 'gh/wip-discard'
- 06:23 PM Bug #2338: mon: adding new monitors simultaneously can allow a new mon to become leader
- Looks like the original monitor doesn't believe in the existence of the new monitors; I'll need to check out why.
<p... - 05:46 PM Bug #2338: mon: adding new monitors simultaneously can allow a new mon to become leader
- ...
- 05:42 PM Bug #2338 (Rejected): mon: adding new monitors simultaneously can allow a new mon to become leader
- When you add two new monitors (out of 3 total) to a cluster you can end up with one of the new monitors being the lea...
- 05:26 PM Bug #2286: mon: different full/near_full values on different monitors
- Hi Guys,
After upgrading the patched-kernel btrfs test cluster from 0.45-1 to 0.45-281-g0777613, the full_ratio an... - 04:38 PM Feature #2337 (Resolved): rgw and rados performance numbers
- 04:30 PM Revision 43d1a920 (ceph): run_seed_to.sh: remove stray arg
- This crept in in commit d1740bd586db80068fc0292223cf21911de66428.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 04:21 PM Revision a11b69fd (ceph): nuke: ignore ntpdate errors
- We keep seeing a race between ntpd startup and our stop + ntpdate + start
sequence. Ignore errors here. - 04:16 PM Feature #2336 (Resolved): qemu: wire up discard
- 04:10 PM Feature #2335 (Resolved): librbd: write-thru cache mode
- 04:02 PM rgw Documentation #1813 (Fix Under Review): doc: document radosgw api diffs with s3
- 03:51 PM Feature #2334 (Resolved): mon: set max mark-out or mark-down
- 02:34 PM Feature #1044 (Resolved): librbd: discard support
- 02:34 PM Feature #2296 (Resolved): librbd: allow resizing to arbitrary sizes
- 02:29 PM Feature #1451 (Resolved): librbd: instrument via perfcounter
- 02:22 PM Feature #1888 (Rejected): log: per-thread ring buffer
- 02:20 PM Subtask #2333 (Resolved): create queueing for peering messages
- Currently, the osd dispatch calls directly into the PG peering state machine. Instead, we need to queue the events g...
- 02:10 PM Subtask #2332 (Resolved): move pg queueing into pgs
- Currently, the osd reaches into the pg to manipulate the pg queue during message receipt and during handle_osd_map. ...
- 02:07 PM Cleanup #2041 (In Progress): osd: move peering into worker threads
- 12:19 PM Revision 0112e749 (ceph): run_seed_to.sh: rework the script, make it more flexible and broaden th...
- Allow for '-h' and other options such as disabling the journal sync tests,
defining it is to be run on a btrfs FS, en... - 11:31 AM Cleanup #2331 (Resolved): Makefile.am:182: `lib/libgtest.a' is not a standard libtool library name
- Warning is still happening, despite git clean -fdx, git submodule freshening of various sorts, etc.
This should prob... - 11:27 AM Bug #2276 (Rejected): osd: eat cpu on restart
- it's up now.. i think i just didn't wait long enough.
- 11:26 AM Bug #2266 (Resolved): teuthology: nuke after failure is failing
- ignore errors caused by ntpdate vs ntpd race
- 11:25 AM Bug #2322 (Need More Info): osd/ReplicatedPG.cc: 3832: FAILED assert(!object_contexts.size())
- also going to wait until the threading refactor is complete before diving into this further.
at this point asserti... - 11:24 AM Feature #2330 (Resolved): dump open files, sockets when we run out of fds
- 11:23 AM Bug #2310 (Resolved): osd: too many open files
- this is just sockets and hitting the flusher limit. we're both increasing 'max open files' and switching to vm limit...
- 11:04 AM Bug #2329 (Resolved): fix detection of C++11 atomic header
- the C++11 atomic header is now <atomic> (I've checked gcc 4.6 and 4.7) and not <cstdatomic>
- 10:23 AM rgw Bug #1681 (Resolved): rgw: user rm with --purge doesn't remove data
- 09:56 AM Feature #2251 (Resolved): rgw long run workloads
04/22/2012
- 09:23 PM Revision e9ecd1b3 (ceph): perfcounters: tolerate multiple loggers with the same name
- Make them unique by appending -<ptr>, so that the json we dump will remain
valid.
We may also want to allow people t... - 04:01 AM Revision be438d6e (ceph): Merge branch 'master' into wip-discard
- 03:28 AM Revision 97f507ff (ceph): Makefile: disable format-security warning
- The prt() varargs function generates this warning
test/rbd/fsx.c: In function ‘prt’:
warning: test/rbd/fsx.c:203:2: ...
04/21/2012
- 09:31 PM Revision c8377e46 (ceph): filestore: verify that fiemap works
- Check for a bug present in older versions of ext4. If present, disable
FIEMAP. See #2328.
Signed-off-by: Sage Weil... - 08:36 PM Revision 6cf87673 (ceph): filestore_idempotent: url has changed
- 08:32 PM Revision 7471a9b1 (ceph): rados: fix error printout for mapext
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 07:50 PM Revision 07ddff42 (ceph): librbd: instrument with perfcounters
- Track IO operations on a per-image basis.
Implements: #1451
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 03:16 PM Feature #1451 (Fix Under Review): librbd: instrument via perfcounter
- see commit:07ddff427145e109eb820b6ed0ddb6cca74b65b6
- 03:15 PM Bug #2328 (Resolved): osd: mapext/fiemap doesn't work for small extents
- see commit:c8377e466caace018eea06c1739265111ce72c48 for a kludge that detects the bug and disabled fiemap.
- 02:35 PM Bug #2328: osd: mapext/fiemap doesn't work for small extents
- this works on a newer kernel (3.2.0-2-amd64).
should we check kernel versions in filestore and magically disable f... - 02:20 PM Bug #2328 (Resolved): osd: mapext/fiemap doesn't work for small extents
- If you query the mapping for an extent that inside a larger allocated extent, the fiemap ioctl won't tell you:
<pr... - 11:15 AM CephFS Bug #2218: CephFS "mismatch between child accounted_rstats and my rstats!"
- Logs from a clean cluster at http://matthew.royhousehold.net/cephLogs/cephLogs.mds.tar (4382MB md5 aaf9364c7e35bc6b5d...
- 12:13 AM Revision fb9fdf45 (ceph): librbd: fix ictx_check pointer weirdness by using std::string
- I was seeing failures of LibRBD.TestIOToSnapshot where we would fail to
refresh after rollback, even though the snap ... - 12:01 AM Revision 888a082f (ceph): FileJournal: don't wait flusher until completions are queued
- Fixes: #2324
Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
04/20/2012
- 11:56 PM Revision d1740bd5 (ceph): filestore: fix collection_add journal replay problem
- In collection_add we have a two-phase guard set on the linked object via
the old name. During replay, we might see t... - 11:56 PM Revision 92b299af (ceph): FileStoreDiff: flip sense of diff*() methods around
- true means diff, false means same.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 11:55 PM Revision ca369c98 (ceph): test_idempotent_sequence: Use FileStoreDiff class instead.
- Use FileStoreDiff instead of having the diff code embedded in the test,
allowing for more tests and people to use the... - 11:51 PM Revision 5466ebc7 (ceph): test_idempotent_sequence: Output missing options on "usage".
- Signed-off-by: Joao Eduardo Luis <jecluis@gmail.com>
- 11:51 PM Revision 47461291 (ceph): FileStoreDiff: check if two FileStores match.
- This code should be on a stand-alone class, instead of being embedded on
a single test, in case someone or something ... - 11:51 PM Revision b94d6a6c (ceph): librbd: fix zeroing of trailing bits on short reads that span objects
- handle_sparse_read() was taking buf_ofs and buf_len, but buf_len was being
interpreted as the total size of the buffe... - 11:51 PM Revision 81af4e86 (ceph): librbd: make discard invalidate the range in cache
- Fed this to test_librbd_fsx and it was happy.
Signed-off-by: Sage Weil <sage@newdream.net> - 11:51 PM Revision 5b603437 (ceph): librbd: fix debug output
- objects is misleading here, these are byte offsets
Signed-off-by: Sage Weil <sage@newdream.net> - 11:51 PM Revision 72323965 (ceph): objectcacher: implement Object::discard()
- Discard a range of bytes from an object.
Signed-off-by: Sage Weil <sage@newdream.net> - 11:51 PM Revision 165038d5 (ceph): objectcacher: rename truncate_set -> discard_set, and use discard
- Do not assume the object extents are at the trailing edge of objects.
Instead, discard arbitrary extents. Fix caller... - 11:51 PM Revision 4ddbbf54 (ceph): librbd: allow image resize to non-block boundaries
- The caller is still invalidating the entire cache, so we don't need to
deal with discard at this level. That might b... - 11:51 PM Revision e861e20a (ceph): revert to xfstests' fsx, which has discard support
- 11:51 PM Revision dd78c3df (ceph): test_librbd_fsx: port newer xfsprogs version
- Builds and runs... for a few ops at least.
Signed-off-by: Sage Weil <sage@newdream.net> - 11:51 PM Revision 379c5e72 (ceph): librbd: fix debug output for image resize
- Print old -> new, not new -> old.
Signed-off-by: Sage Weil <sage@newdream.net> - 11:50 PM Revision a8aaa93b (ceph): admin_socket: add a separate debug subsys/level
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 11:36 PM Revision f7699cc9 (ceph): log: prefix dump with line numbers
- This makes it easier to interpret the dump, and makes it obvious what is
dump (and potentially a dup of something tha... - 10:35 PM Revision a68b3069 (ceph): test_filestore_workloadgen: track pg log size in memory
- Avoid a stat!
Signed-off-by: Sage Weil <sage@newdream.net> - 10:30 PM Revision 6b28f55c (ceph): test_filestore_workloadgen: whitespace
- Signed-off-by: Sage Weil <sage@newdream.net>
- 10:25 PM Revision 2834f601 (ceph): organize test_filestore_idempotent*
- Signed-off-by: Sage Weil <sage@newdream.net>
- 10:23 PM Revision 8bb58bc2 (ceph): remove hopelessly flawed test_filestore_idempotent
- Signed-off-by: Sage Weil <sage@newdream.net>
- 10:22 PM Revision 94ebe8d9 (ceph): test_store -> test_filestore
- and organize source
Signed-off-by: Sage Weil <sage@newdream.net> - 10:20 PM Revision 77c645a1 (ceph): test_wrkldgen -> test_filestore_workloadgen
- Also put the code in src/test/filestore/.
Signed-off-by: Sage Weil <sage@newdream.net> - 10:13 PM Revision 27efa055 (ceph): msgr: move locking out of submit_message to its callers
- Since they're doing stuff like looking through rank_pipe, they need
to be holding the lock too.
Signed-off-by: Greg ... - 10:12 PM Revision 5a8019ec (ceph): msgr: remove useless block addition.
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 09:18 PM Feature #2327 (Resolved): mon: use external keyring for inter-mon auth
- Currently the mon. key is part of the internal mon auth database. It can't be modified without a running mon cluster...
- 09:17 PM rbd Feature #2326 (Resolved): krbd: use new class interfaces, new image format
- Update rbd_types.h to match the userspace version, and add support for opening new-format images while keeping suppor...
- 09:04 PM Feature #2325 (Resolved): setup new email/etc
- 08:32 PM Linux kernel client Tasks #2138 (Resolved): rbd: run xfstests on a local XFS filesystem over RBD
- I have two files that implement automated testing using
xfstests over rbd devices.
One is now in the ceph git tre... - 08:28 PM Linux kernel client Bug #2287: rbd: crashes with 10Gbit network and fio
- A kernel dump would likely help, but there's no guarantee because
of the delayed execution of the operation. It wou... - 07:38 PM Linux kernel client Bug #2287: rbd: crashes with 10Gbit network and fio
- Would a kernel core dump help here?
- 07:37 PM Linux kernel client Bug #2287: rbd: crashes with 10Gbit network and fio
- This is one of a family of bugs we've been trying to understand.
Here is another one:
http://tracker.newdream.n... - 08:28 PM Revision f00dab6f (ceph): ceph-authtool: rename cli test dir
- Signed-off-by: Sage Weil <sage@newdream.net>
- 08:28 PM Revision 613d8179 (ceph): ceph-conf: rename cli test dir
- Signed-off-by: Sage Weil <sage@newdream.net>
- 08:22 PM Revision 0616c5b2 (ceph): Merge branch 'stable'
- Conflicts:
src/osd/OpRequest.cc - 08:21 PM Revision ac874a62 (ceph): OpRequest: mark_event output is now at debug 5
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 08:00 PM Linux kernel client Bug #2243: btrfs: warning in orphan_commit_root
- I mentioned this somewhat informally to Chris Mason last week. I
provided him the message, and he said:
Well, ... - 07:39 PM Feature #2251 (In Progress): rgw long run workloads
- 07:37 PM Linux kernel client Bug #2260: libceph: null pointer dereference at try_write+0x638+0xfb0
- Another report, likely related:
http://tracker.newdream.net/issues/2287
I don't understand it well enough yet... - 07:37 PM rgw Documentation #1813 (In Progress): doc: document radosgw api diffs with s3
- 07:35 PM Subtask #825 (In Progress): osd: remove pg map updating from handle_osd_map
- 07:35 PM Feature #2314 (Fix Under Review): remove localized pgs
- wip-lpg
Did a basic test of a cluster with localized pgs and upgraded to this, no problems. A reasonably thorough... - 07:12 PM Linux kernel client Bug #2298 (Resolved): rbd: broken encode_op for big-endian hosts?
- This has been fixed. I have been testing it in a private branch
and will shortly be updating the ceph-client testin... - 07:10 PM Linux kernel client Bug #2242 (Resolved): rbd: spinlock on wrong cpu
- This was fixed a couple of weeks ago, and the result has been committed
both to the testing and master branches of t... - 06:32 PM Revision c9d21856 (ceph): hammer.sh: -a to archive each run
- Signed-off-by: Sage Weil <sage@newdream.net>
- 05:00 PM Bug #2324 (Resolved): osd: assert("q.empty()") failed in OpSequencer destructor
- sam fixed this in commit:888a082f23974b1f7a63f302e29a326182e7dc41
- 03:56 PM Bug #2324 (Resolved): osd: assert("q.empty()") failed in OpSequencer destructor
- This is consistently reproducible with 2 osds started by vstart on vit, but only happens intermittently with 1 osd.
... - 04:32 PM Feature #2245 (Resolved): rgw long run ceph install
- 03:09 PM Bug #2079 (Duplicate): rbd: creating a snapshot with the same name doesn't return an error
- i think this was caused by the rados class return values. in any case, it works correctly now.
- 03:07 PM Bug #2084 (Can't reproduce): segfault in tcmalloc
- 12:16 PM Feature #1618 (Resolved): libvirt: make sure migration works
- we demoed this
- 12:05 PM Feature #2323 (Resolved): osd: limit 'old request' messages generated
- If there are hundreds of old requests queued, let's say that, instead of generated gigabytes of logs.
Maybe a simp... - 10:44 AM Bug #2322 (Resolved): osd/ReplicatedPG.cc: 3832: FAILED assert(!object_contexts.size())
- ...
- 05:24 AM Revision f2d4574b (ceph): config: dump subsys log levels with --show-config
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 04:29 AM Revision 607a8a3d (ceph): osd: do not create localized pgs
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
04/19/2012
- 11:43 PM Revision 0df33e43 (ceph): Merge branch 'wip-2313'
- 11:25 PM Revision 8335eb53 (ceph): rgw: can't remove user that owns data
- This is a workaround for #1681. We don't allow the removal of
user that still has data.
Signed-off-by: Yehuda Sadeh ... - 11:11 PM Revision 9df7f6f6 (ceph): msgr: merge the send_message and lazy_send_message implementations.
- They only varied by the presence of "lazy" (in output) and a flag.
Signed-off-by: Greg Farnum <gregory.farnum@dreamh... - 10:58 PM Revision a394fb2f (ceph): msgr: do not establish a new Pipe for any server
- Servers do not establish Connections, and if there's a Connection
without a Pipe it's because the Pipe was deliberate... - 10:44 PM Revision 42a49cf1 (ceph): msgr: Create a proper Connection* lazy_send_message()
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 10:44 PM Revision 526cd9d0 (ceph): msgr: merge submit_message() functions
- Our consistency guarantees on address- and Connection-based send_message()
calls have diverged since they have differ... - 10:22 PM Revision 56fc92c0 (ceph): FileStore: don't read reference to writeq head after pop_write
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 09:49 PM Bug #2291 (Can't reproduce): objectcacher perfcounters don't work with test_librbd_fsx
- This works fine for me. Maybe it was fixed when the objectcacher naming thing was changed?...
- 09:17 PM Messengers Cleanup #2150: repair the Simple/Messenger interface
- Looks good to me, provided it makes it through the regression suite without problems!
- 07:11 PM Messengers Cleanup #2150: repair the Simple/Messenger interface
- wip-msgr-interface
- 09:08 PM Feature #2321 (Resolved): osd: investigate memory consumption from peering backlog
- 09:07 PM Feature #2320 (Duplicate): mon: detect and throttle osd flapping
- 09:07 PM Feature #2319 (Resolved): mon: block osd mark-down
- 09:07 PM Feature #2318 (Resolved): mon: block osd boot
- 09:06 PM Feature #2317 (Resolved): mon: pause/unpause auto-mark-out
- 08:33 PM Revision ff0fe372 (ceph): add rbd_fsx_[no]cache jobs to regression suite
- 08:32 PM Revision e3af0877 (ceph): rbd_fsx: show progress
- The updated fsx takes this arg.
Signed-off-by: Sage Weil <sage@newdream.net> - 08:21 PM Revision 758fa9ba (ceph): rgw: bucket HEAD operation should return 200, not 204
- Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
- 07:47 PM Revision 9cb796a0 (ceph): obsync: section 8 -> 1
- Fixes: #2263
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 07:44 PM Revision 6a58314d (ceph): fix misc checks that wait for N osds to be up
- These all cut&pasted broken code, blah!
- 06:16 PM Revision fcc44812 (ceph): msgr: minimally document the SimpleMessenger inner classes
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 06:16 PM Revision db6d467b (ceph): msgr: fix some spacing issues in Pipe
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 06:16 PM Revision c470e1a0 (ceph): msgr: start moving functions around in Pipe
- 06:16 PM Revision 5c21f7ea (ceph): msgr: remove the Pipe::disposable flag.
- It has no function as it stands. It was intended for new connections
to avoid being marked lossy, but that can't happ... - 06:16 PM Revision 146e1a46 (ceph): msgr: make mark_down_on_empty() behave more consistently.
- Previously you could get ms_handle_reset() notifications if you
used mark_down_on_empty(), but only if the Pipe's que... - 06:16 PM Revision 501553e5 (ceph): msgr: group private data members by type
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 06:16 PM Revision d3e40e00 (ceph): msgr: reorder the SimpleMessenger data members a little bit
- They make moderately more sense to me in this order.
Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com> - 04:37 PM rgw Feature #2313 (Resolved): rgw: expose extra bucket info trough S3 api
- Already pushed a test to the s3-tests functional.
- 04:33 PM rgw Feature #2313 (In Progress): rgw: expose extra bucket info trough S3 api
- wip-2313 looks sane.. let's add a test and merge for 0.46.
- 04:25 PM rgw Bug #1681 (In Progress): rgw: user rm with --purge doesn't remove data
- merged the radosgw-admin change to make it fail. let's add a test for it and then close this bug.
- 04:23 PM Revision 09100b6b (ceph): rgw: HEAD bucket, don't dump rounded size
- Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
- 04:19 PM Revision 9c2251f6 (ceph): osd: fix vector array out of bounds
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 04:17 PM Feature #2296 (Fix Under Review): librbd: allow resizing to arbitrary sizes
- see wip-discard
- 02:13 PM Feature #2296: librbd: allow resizing to arbitrary sizes
- more importantly, we need to either error out with EINVAL if it's not a block multiple, or do it... currently we sile...
- 02:37 PM Bug #2307 (Resolved): OSD & Monitor disagree on the contents of pg_temp
- Just changing the pg_num and pgp_num did fix it up, so with the osdmap workaround we should be all good now.
- 02:10 PM Feature #1044 (Fix Under Review): librbd: discard support
- 12:47 PM Bug #2263 (Resolved): obsync: move man page to section 1
- 12:45 PM Bug #2311 (Need More Info): rbd: delete + create image led to EEXIST
- 07:20 AM Bug #2311 (In Progress): rbd: delete + create image led to EEXIST
- Can you generate a log? Ideally 'debug ms = 1'?
Also, attach the output of 'ceph --show-config'?
Thanks! - 02:43 AM Bug #2311: rbd: delete + create image led to EEXIST
- Hi Sage,
uhm, not solved yet as per ceph version 0.45-207-g3053e47 (commit:3053e4773bae93cfa3158882aa4963803862f9b... - 12:44 PM Bug #2262 (Resolved): qa: osd-recovery tasks fails on flush_pg_stats
- fixed by teuthology commit:6a58314d4627d106c5fd6df186e191c19a01f64b
- 10:47 AM Bug #2192: ceph-mon hangs consuming 100% CPU
- It was some 3.0.0 Ubuntu kernel, backed by btrfs.
- 10:06 AM Bug #2192: ceph-mon hangs consuming 100% CPU
- I missed this when it came in, and I don't know where the 100% CPU usage is coming from, but the hung filesystem soun...
- 04:58 AM Revision e82c3309 (ceph): ceph-fuse: only call common_init_finish in child; fix admin socket
- If we do it in the parent and child, we break the admin socket (among
other things).
Fixes: #2293
Signed-off-by: Sag... - 04:27 AM Revision f1de0d6c (ceph): qa: drop rgw.py from test_python.sh workunit
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 04:27 AM Revision d3a57a21 (ceph): ceph.spec.in: drop rgw.py
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 04:26 AM Revision 0995d084 (ceph): Makefile: drop rgw.py
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 03:19 AM Bug #2316 (Resolved): rbd: restart of OSD leeds to stale qem-VM's with "ceph version 0.45-207-g30...
- Hi,
in my current test-setup all four VM's are started with rbd_cache parm. After all VM's are started and began t... - 12:14 AM Revision 4390c08a (ceph): mon: fix full/nearfull ratio initialization
- create_pending() is called before create_initial(), so just put these in
the first incremental.
Signed-off-by: Sage ...
04/18/2012
- 11:37 PM Revision 7fbc0fcc (ceph): msgr: add error code returns to send_keepalive.
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 11:27 PM Revision cb43eeee (ceph): msgr: add some validity checks to ~SimpleMessenger()
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 10:54 PM Revision 0c2c18c1 (ceph): msgr: convert global_seq_lock to a spinlock.
- Because nobody wants to get put to sleep waiting on a unique integer.
(I would just make this an atomic_t but we need... - 10:51 PM Bug #2263: obsync: move man page to section 1
- 10:45 PM Revision dc4c759c (ceph): msgr: convert get_policy() to return a const reference.
- The only user who didn't want a const should have been using
direct access anyway (and now is doing so).
Signed-off-... - 10:45 PM Revision aab516da (ceph): rgw: allow subuser creation on radosgw-admin user create
- This fixes issue #2312: allow the creation of both user
and a subuser in a single radosgw-admin command.
Signed-off-... - 10:44 PM Revision c9adf07b (ceph): msgr: move SimpleMessenger function declarations into good order
- We now have one declaration each of private and protected, and two
declarations of public -- one for the Messenger-re... - 10:42 PM CephFS Bug #2293 (Resolved): admin sockets don't persist with ceph-fuse
- commit:e82c33099a0efda027bc7fa991dcd2073baea539
- 10:23 PM Revision dbe779a5 (ceph): osd: dump old ops singly rather than all at once.
- Fixes #2269. Convert the OpTracker::check_ops_in_flight interface
to take a vector<string> and create a separate warn... - 10:23 PM Revision b0532aa6 (ceph): osdmap: restructure flow in OSDMap::decode
- Signed-off-by: Sage Weil <sage@newdream.net>
Reviewed-by: Greg Farnum <gregory.farnum@dreamhost.com> - 10:19 PM Revision 7ae1aefa (ceph): gather logs for cfuse dbench workload, hopefully catch #1737
- 10:13 PM Revision bc458bfd (ceph): ceph-object-corpus: prune archive
- Signed-off-by: Sage Weil <sage@newdream.net>
- 10:09 PM Revision ebb19e70 (ceph): test/encoding/import.sh: fix
- Signed-off-by: Sage Weil <sage@newdream.net>
- 09:53 PM Revision e1622e4f (ceph): rgw: HEAD bucket also dumps bucket meta info
- Shows total objects, bytes used, actual bytes used.
Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> - 09:53 PM Revision 0769efa5 (ceph): rgw: add x-rgw-* as a possible meta header prefix
- Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
- 09:46 PM rgw Bug #2027: rgw -> apache miscommunication
- Not completely unlikely. We can set it to "can't reproduce", and reopen if we see it again.
- 06:12 PM rgw Bug #2027: rgw -> apache miscommunication
- do we think this is fixed now by the rgw throttling?
- 09:36 PM Revision 2de37e01 (ceph): Merge remote-tracking branch 'gh/wip_journal'
- Reviewed-by: Sage Weil <sage@newdream.net>
- 09:10 PM Revision c3bfe456 (ceph): osdmap: remove bad cct reference
- Not in OSDMap
Signed-off-by: Sage Weil <sage@newdream.net> - 08:28 PM Revision 9ef953b5 (ceph): Merge branch 'wip-2286-ratio-a'
- Reviewed-by: Sage Weil <sage@newdream.net>
- 07:50 PM Revision eea982e5 (ceph): osdmap: workaround old bug that zeroed pool_max
- Signed-off-by: Sage Weil <sage@newdream.net>
Reviewed-by: Greg Farnum <gregory.farnum@dreamhost.com> - 06:32 PM Revision ee22c97b (ceph): python-ceph: remove rgw bindings
- We shouldn't expose these internal apis.
Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> - 06:21 PM Revision 407b2e0b (ceph): whitelist xfs_fsr syslog noise
- Ignore lines like
2012-04-17T13:44:11-07:00 plana59 fsr[5454]: DEBUG: fsize=450560 blsz_dio=450560 d_min=512 d_max=2... - 06:10 PM Revision eb8a8fdf (ceph): pgmap: allow Incrementals to specify [near]full_ratios of 0
- This commit isn't entirely safe: old monitors used 0 to mean "no change".
We can revert this (and the PGMonitor.cc po... - 06:05 PM Revision 88525eaa (ceph): pgmon: add "set_full_ratio x" and "set_nearfull_ratio x" commands
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 06:05 PM Revision 882bd5b8 (ceph): pgmon: convert PGMonitor::prepare_command to our standard infrastructure.
- So that I can add another command.
Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com> - 06:05 PM Revision f1152e67 (ceph): mon: only fill in full/nearfull sets if the ratio > 0
- This avoids putting all OSDs in both sets when the ratios are 0, as they
are with a fresh cluster and pgmap. This al... - 06:05 PM Revision 1242b754 (ceph): pgmon: remove the PGMonitor update_full_ratios stuff
- Making it a config watcher is just a huge mess in terms of consistently
updating it appropriately.
The next commit wi... - 06:05 PM Revision 841f2885 (ceph): mon: unconditionally encode PGMap full ratios in the Incremental
- This properly spreads the real value to peon monitors -- they weren't
seeing the right values at all before.
Initiali... - 06:00 PM Bug #2310: osd: too many open files
- failed to capture a full strace.. try it again (once we find a failing osd on congress) with
strace -e trace=open,... - 09:48 AM Bug #2310 (Resolved): osd: too many open files
- ...
- 04:39 PM Bug #2315 (Resolved): unrecognized admin socket command 'objecter_requests'
- From teuthology:/a/nightly_coverage_2012-04-18-a/1602/teuthology.log:...
- 03:57 PM rgw Bug #2312 (Resolved): rgw: create user and subuser in a single radosgw-admin command
- Fixed, commit:aab516da7f89310445be4e4fb61836084d2dac32.
- 02:01 PM rgw Bug #2312 (Resolved): rgw: create user and subuser in a single radosgw-admin command
- 03:41 PM Bug #2211 (Resolved): osd: entity_inst_t OSDMap::get_inst(int) const
- 03:41 PM Bug #2262 (In Progress): qa: osd-recovery tasks fails on flush_pg_stats
- 03:27 PM Feature #2314 (Resolved): remove localized pgs
- 02:47 PM rgw Feature #2313: rgw: expose extra bucket info trough S3 api
- Ok, let's just send those extra headers anyway. Otherwise we'd have some issue creating the request signature for the...
- 02:11 PM rgw Feature #2313 (Resolved): rgw: expose extra bucket info trough S3 api
- syntax:
HEAD /<bucket>
X-RGW-Params: extrainfo
extra response headers:
X-RGW-Object-Count: <object count>
X-RG... - 02:23 PM Feature #2252 (Resolved): rgw long run kernels
- 02:22 PM Feature #2250 (Resolved): rgw long run raid config
- 02:14 PM Feature #2265 (Rejected): make sure objecter/kclient error out when localized pgs don't exist
- 01:58 PM rgw Feature #2308 (Resolved): radosgw-admin: make user create idempotent
- done, commit:5a6bbd0c473e15aa7642da367e7936015d19d77a.
- 01:46 PM Bug #2307: OSD & Monitor disagree on the contents of pg_temp
- And I gave him a patched monitor so he could set pg_num, which should fix it. Waiting to hear back, and will apply th...
- 01:16 PM Bug #2307: OSD & Monitor disagree on the contents of pg_temp
- pushed workaround that will repair osdmaps that saw your corruption, commit:eea982e56739a7a91ca907ccc5c5ec1f78d9460d.
- 01:30 PM Bug #2311 (Resolved): rbd: delete + create image led to EEXIST
- this is 'rbd writeback window' at its best. long live 'rbd cache'!
- 01:06 PM Bug #2311: rbd: delete + create image led to EEXIST
- Congrats for closing the annoying ticket #2178 :-D
Fair enough, to have a new one on this issue, here my last note... - 12:46 PM Bug #2311: rbd: delete + create image led to EEXIST
- Is it possible there is some other user, or the logs are from the wrong cluster?
I see:
- client.13507 deletes 90... - 12:45 PM Bug #2311 (Resolved): rbd: delete + create image led to EEXIST
- Here is a sequence copy-n-pasted:
rbd rm data/905-testdisk.rbd
Removing image: 100% complete...done.
rbd create ... - 01:28 PM Bug #2286 (Resolved): mon: different full/near_full values on different monitors
- commit:9ef953b5e20c3d232cfe4aa90f26476a2a2f911b
- 11:18 AM Bug #2286 (Fix Under Review): mon: different full/near_full values on different monitors
- Check out wip-2286-ratio-a and see what you think. It fills in the ratios from g_conf on create_initial, only changes...
- 12:51 PM Bug #2178: rbd: corruption of first block
- Hi Sage,
sorry, was not clear enough. The logfiles provide informations for "907-testdisk.rbd..." not "906..."
Th... - 12:46 PM Bug #2178 (Resolved): rbd: corruption of first block
- moved this new issue to #2311, and resolving this bug. hooray!
- 12:45 PM Bug #2178: rbd: corruption of first block
- Oliver Francke wrote:
> Here is a sequence copy-n-pasted:
>
> rbd rm data/905-testdisk.rbd
> Removing image: 100... - 10:41 AM Bug #2178: rbd: corruption of first block
- Oliver Francke wrote:
> Hi Sage,
>
> here my notes, after almost 40 tests no bad things happened, only once a min... - 07:22 AM Bug #2178: rbd: corruption of first block
- second logfile here, sorry.
- 07:18 AM Bug #2178: rbd: corruption of first block
- Here is a sequence copy-n-pasted:
rbd rm data/905-testdisk.rbd
Removing image: 100% complete...done.
rbd create ... - 05:51 AM Bug #2178: rbd: corruption of first block
- Meanwhile continued to test...:
I noticed some negative degredation:
2012-04-18 14:43:37.282634 pg v128104: ... - 05:36 AM Bug #2178: rbd: corruption of first block
- Hi Sage,
here my notes, after almost 40 tests no bad things happened, only once a minor hickup, where the rbd-head... - 08:22 AM Linux kernel client Bug #2298 (In Progress): rbd: broken encode_op for big-endian hosts?
- I sent a note to the various lists Al Viro posted to, to confirm the
bug (wasn't sure whether Sage had or not).
I... - 05:58 AM Linux kernel client Bug #2302: xfs: warning at mutex_remove_waiter
- Interesting... The warning showed up again despite test 232 being
removed from the list. Based on the time stamp o... - 04:38 AM Revision 5a6bbd0c (ceph): rgw-admin: user create is idempotent
- Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Reviewed-by: Sage Weil <sage@newdream.net> - 04:38 AM Revision 413d8be9 (ceph): osdmap: dump pool_max
- Signed-off-by: Sage Weil <sage@newdream.net>
04/17/2012
- 10:33 PM Revision 4627b67f (ceph): msgr: Document remaining SimpleMessenger functions and data members
- This doesn't handle the inner classes (Pipe, mostly) yet. It also
raises questions about whether these functions shou... - 10:01 PM Revision 1f6b7f81 (ceph): msgr: Group the SimpleMessenger functions following the Messenger's pat...
- Ordering everything nicely will make for a cleaner interface, make
it easier to see what documentation we need, and m... - 08:47 PM Bug #2286: mon: different full/near_full values on different monitors
- Greg Farnum wrote:
> Hmm. I looked at redoing this and got stuck on the semantics we want. If we're interested in fu... - 04:52 PM Bug #2286 (In Progress): mon: different full/near_full values on different monitors
- Hmm. I looked at redoing this and got stuck on the semantics we want. If we're interested in full_ratio == 0 being an...
- 11:00 AM Bug #2286: mon: different full/near_full values on different monitors
- yeah. actually, i think the check should go in tick() inside the is_leader() block, and not update_from_paxos().
- 10:54 AM Bug #2286: mon: different full/near_full values on different monitors
- Oh, I see...I wasn't following that need_*_ratio_update stuff properly. And update_full_ratios() will be called on th...
- 10:30 AM Bug #2286: mon: different full/near_full values on different monitors
- Greg Farnum wrote:
> I'm looking at your patch and it doesn't make a lot of sense to me.
> First off, when do you t... - 09:45 AM Bug #2286: mon: different full/near_full values on different monitors
- I'm looking at your patch and it doesn't make a lot of sense to me.
First off, when do you think that peon monitors ... - 08:43 PM Bug #2307: OSD & Monitor disagree on the contents of pg_temp
- Greg Farnum wrote:
> I'm confused how you're getting that pool_max printout — I don't see it at all when I run that ... - 06:57 PM Bug #2307: OSD & Monitor disagree on the contents of pg_temp
- I'm confused how you're getting that pool_max printout — I don't see it at all when I run that command with a ceph-de...
- 04:16 PM Bug #2307: OSD & Monitor disagree on the contents of pg_temp
- at some point the osdmap pool_max got set to -1.
nine:2307 04:15 PM $ ~/src/ceph/src/ceph-dencoder type OSDMap i... - 03:56 PM Bug #2307: OSD & Monitor disagree on the contents of pg_temp
- nine:2307 03:56 PM $ osdmaptool osdmap_full/5754 -p | grep ^pool
pool 0 'data' rep size 2 crush_ruleset 0 object_has... - 03:52 PM Bug #2307: OSD & Monitor disagree on the contents of pg_temp
- It looks to me liek the 'data' pool (0) was deleted, and then a new one (vmimages) was created. but somehow that was...
- 10:34 AM Bug #2307 (Resolved): OSD & Monitor disagree on the contents of pg_temp
- See: http://marc.info/?t=133352732900001&r=1&w=2
It seems that (for example) pg 0.138 is in pg_temp, but the OSD c... - 08:22 PM Revision 3053e477 (ceph): Merge branch 'wip-2289-2'
- 08:20 PM Revision dcd34fc8 (ceph): FileJournal: let peek_write return a reference
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 08:17 PM Revision 65614847 (ceph): common: add blkdev.{h,cc}
- Signed-off-by: Sage Weil <sage@newdream.net>
- 08:12 PM Revision 82d7a681 (ceph): rbd: fix probe of source when block device
- Fixes: #2304
Signed-off-by: Sage Weil <sage@newdream.net> - 08:11 PM Revision 41cb40ad (ceph): filestore: factor block device size probe out into helper
- Signed-off-by: Sage Weil <sage@newdream.net>
- 07:56 PM Revision dea85368 (ceph): FileJournal: clarify locking in header
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 07:53 PM Revision 377f75f9 (ceph): OpRequest: mark_event output is now at debug 5
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 06:39 PM Revision 98e2369e (ceph): librados: add methods to check and wait for safe/complete -and- callback
- Allow user to check for (safe|complete) -and- callback return, and wait
for that condition.
Fixes: #2301
Signed-off-... - 06:17 PM Revision 001aa462 (ceph): msgr: add a little detail to mark_down_on_empty() and mark_disposable()
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 05:53 PM Revision 2eb2aa5a (ceph): msgr: group functions more appropriately
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 05:53 PM Revision a422c479 (ceph): msgr: remove out-of-date comments and function
- I think that destroy() is left over from a misinformed attempt
to implement refcounting on the Messenger.
Signed-off... - 05:45 PM Revision 0be9718a (ceph): librbd: 'rbd cache enabled' -> 'rbd cache'
- 'enabled' is useless verbiage. We should fix the rgw option too,
protably...
Signed-off-by: Sage Weil <sage@newdrea... - 03:59 PM Revision e875b89f (ceph): Add task for running fsx on an rbd image.
- 03:03 PM Feature #2309 (Duplicate): rados namespaces
- 01:33 PM Revision 6cbda4c3 (ceph): qa: comment out xfstest 232
- Test 232 in the xfstests suite produces an XFS error in the log
when run over an RBD device. This is most likely an ... - 01:23 PM rgw Bug #2289 (Resolved): rgw: listing a bucket hangs after removing inexisting object
- Fixes merged into master at commit:3053e4773bae93cfa3158882aa4963803862f9b2.
- 01:13 PM Bug #2304 (Resolved): rbd import fails on block device
- 11:57 AM CephFS Bug #2299 (Rejected): all MDS commit suicide on startup
- 11:54 AM Bug #2219 (Can't reproduce): OSD's commit suicide with 0.44
- Let us know if you see this again! Thanks
- 11:40 AM Bug #2301 (Resolved): librados: LibRadosMisc.AioOperatePP failure
- 11:27 AM rgw Feature #2308 (Resolved): radosgw-admin: make user create idempotent
- radosgw-admin user create should be idempotent and work similar to user modify. We would need to verify that the same...
- 08:11 AM Linux kernel client Bug #2260: libceph: null pointer dereference at try_write+0x638+0xfb0
- I believe we are seeing the same problem here. I have been able to reproduce it each time I have tried. The hardwar...
- 07:14 AM Revision e0ec2865 (ceph): objecter: recalc pgid in case it wasn't a precalculated pgid
- We had an issue where we didn't call it since we relied on the
existance of the object name and we were passed an emp... - 07:12 AM Revision f22da6c0 (ceph): cls_rgw: rgw_dir_suggest_changes move cur_disk_bl inside loop
- Otherwise we'd fail in case the cls_cxx_read_key failed, since
we rely on its size.
Signed-off-by: Yehuda Sadeh <yeh... - 06:36 AM Linux kernel client Bug #2302: xfs: warning at mutex_remove_waiter
- I have updated the run_xfstests.sh script so that it simply no longer
runs test 232. That way we can still benefit ... - 04:36 AM Revision 998498f2 (ceph): Merge branch 'wip-oc-perf'
- 04:32 AM Revision 31d0d1d7 (ceph): librados: test get_pool_name
- Signed-off-by: Sage Weil <sage@newdream.net>
- 04:32 AM Revision 2dbea128 (ceph): librados: C bindings for get_pool_name()
- Signed-off-by: Sage Weil <sage@newdream.net>
04/16/2012
- 10:02 PM Bug #2178: rbd: corruption of first block
- The most recent occurrence has been confirmed to be a replay issue with non-btrfs filesystems. The wip-guard branch ...
- 09:54 PM Bug #2255 (Resolved): osd: fix object name collisions between pools in temp collection
- 09:52 PM Bug #2286: mon: different full/near_full values on different monitors
- pushed a patch that confines the logic of when to update this into a single bit of code. look okay?
i think the b... - 12:57 PM Bug #2286: mon: different full/near_full values on different monitors
- Sage asked on irc about just setting it up on the initial create_empty. The problem with that the only data which is ...
- 11:26 AM Bug #2286 (Fix Under Review): mon: different full/near_full values on different monitors
- This got (obviously) broken by commit:b6d1c0c9b7290a237560528b6ff0d6b2b2998ee2, which put in the use of magic numbers...
- 09:37 PM Feature #2113 (Resolved): objectcacher perfcounters
- 11:24 AM Feature #2113: objectcacher perfcounters
- My bad — I'll try and do that today!
- 11:13 AM Feature #2113 (Fix Under Review): objectcacher perfcounters
- not merged yet! i wanted to get feedback first on my naming kludge...
- 10:02 AM Feature #2113 (Resolved): objectcacher perfcounters
- Sage merged this.
- 09:21 PM Revision 9f5b9ada (ceph): Merge remote branch 'upstream/wip-guard'
- 09:19 PM Revision 2f5af83f (ceph): Filestore: fix guard errors in COLL_REMOVE and COLL_MOVE
- Use object rather than collection check in COLL_REMOVE. Adjust
COLL_MOVE check for -1/1 distinction.
Signed-off-by:... - 07:32 PM Revision f87a6a0f (ceph): cls_rgw: rgw_dir_suggest_changes() don't try to decode cur_disk
- don't try to decode cur_disk if bufferlist is zero sized.
Signed-off-by: Yehuda Sadeh <yehuda.sadeh@dreamhost.com> - 07:32 PM Revision 6868ec6a (ceph): rgw: skip empty entries when listing a bucket
- Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
- 07:32 PM Revision 73badda6 (ceph): cls_rgw: don't zero out error code before reading value
- We were zeroing the return code before reading it.
Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> - 04:12 PM Bug #2306: objecter: accessing empty object maps to pool 0
- that looks right to me.
and yeah, i don't think object operations should be possible on an empty object name... - 04:03 PM Bug #2306: objecter: accessing empty object maps to pool 0
- Yep, that's pretty much exactly what I was thinking.
The only other question is if this fix is the right approach ... - 04:00 PM Bug #2306: objecter: accessing empty object maps to pool 0
- Would something like this work (not tested)?...
- 03:52 PM Bug #2306: objecter: accessing empty object maps to pool 0
- i prefer an explicit separate field for oid-vs-pg mode so that we can distinguish between pg 0.0 (really) and no pg/n...
- 03:07 PM Bug #2306: objecter: accessing empty object maps to pool 0
- Ah, nope. list_objects is broken.
- 03:06 PM Bug #2306: objecter: accessing empty object maps to pool 0
- From what I see, the pg ops call pool_op_submit() and not op_submit() so Greg's fix might be ok?
- 02:53 PM Bug #2306: objecter: accessing empty object maps to pool 0
- Ah, you're right. I missed that function when looking to see who filled in the op->pgid.
In that case we should ma... - 02:33 PM Bug #2306: objecter: accessing empty object maps to pool 0
- i think that if was there for the pg ops (PGLS) where there is no object... the list_objects code is filling in the p...
- 02:07 PM Bug #2306 (Fix Under Review): objecter: accessing empty object maps to pool 0
- Yep, the Objecter doesn't calculate pg placement for objects with a zero-length name. I'm pretty sure the if guard th...
- 01:51 PM Bug #2306: objecter: accessing empty object maps to pool 0
- Empty object <== object with empty name
- 01:51 PM Bug #2306 (Resolved): objecter: accessing empty object maps to pool 0
- Even if different pool was specified.
- 03:34 PM CephFS Bug #2299: all MDS commit suicide on startup
- this issue can be closed, there was an error in the underlying fileystem of osd.0 :)
- 02:59 PM CephFS Bug #2277: qa: flock test broken
- I was going to move this over to the kernel client project and then realized I can't — should we close this bug (reje...
- 02:46 PM CephFS Bug #2277: qa: flock test broken
- ...
- 02:15 PM Linux kernel client Bug #2298: rbd: broken encode_op for big-endian hosts?
- there are some old g5's in the closet here at aon that we can use.
in the past we've found/fixed these issues with... - 01:46 PM Linux kernel client Bug #2298: rbd: broken encode_op for big-endian hosts?
- I haven't looked at this in any detail but I presume Al is correct.
We don't have any big endian hardware anywhere, ... - 01:49 PM CephFS Bug #2288: libcephfs: setxattr returns EEXIST following removexattr
- More info:
That branch has a patch which adds a call to removexattr before the setxattr. If you run testceph twice i... - 01:38 PM Linux kernel client Bug #2302: xfs: warning at mutex_remove_waiter
- Should have waited. It have reproduced the problem by running test 232.
- 01:37 PM Linux kernel client Bug #2302: xfs: warning at mutex_remove_waiter
- After a lot of repetitions, I've narrowed it down to test 232 or 234.
- 10:12 AM Linux kernel client Bug #2302: xfs: warning at mutex_remove_waiter
- I ran subsets of that list at least three times and never
reproduced it. I tried again after a reboot, and again,
... - 06:43 AM Linux kernel client Bug #2302: xfs: warning at mutex_remove_waiter
- Looking at the list of tests that indicate they include quota testing,
the ones that are currently being run by the ... - 06:36 AM Linux kernel client Bug #2302: xfs: warning at mutex_remove_waiter
- I sent a report to the XFS mailing list about the warning. I have to try
to narrow down which test was running when... - 01:18 PM CephFS Bug #2285: libcephfs: failure with empty name components
- Yep, it's client-local; there's no request to the MDS for this either.
Guess that means we don't care right now? - 01:04 PM CephFS Bug #2285 (In Progress): libcephfs: failure with empty name components
- Oddly, this looks like it's a race. I can't reproduce it with any client debugging on...
- 11:33 AM Feature #2305: Moving rbd images between pools
- Not quite; copy works, but slowly (because of course it's duplicating all the data). I don't know if mv/rename could...
- 11:24 AM Feature #2305 (Rejected): Moving rbd images between pools
- We discovered it does work if you keep the image names the same and vary the pool names. :)
- 11:01 AM Feature #2305 (Resolved): Moving rbd images between pools
- It would be nice to have an option to move rbd's between pools with a syntax like:
rbd mv <first poolname>/<image na... - 10:02 AM Messengers Cleanup #2150 (In Progress): repair the Simple/Messenger interface
- Not really done! ;)
- 08:44 AM rbd Feature #2297: ObjectCacher: mark buffers mergeable for ksm
- I'm really not sure this is something we want to do, especially unconditionally. Let's wait until we get some idea of...
- 07:27 AM Bug #2304 (Resolved): rbd import fails on block device
- root@burnupi30:~# rbd import /dev/sda burnupi30.sda
fiemap ioctl() failed
Importing image: 100% complete...done.
... - 03:39 AM Revision 6bede298 (ceph): dump_stuck: whitelist 'wrongly marked me down'
- The test marks the osds down.. they may generate this error if they get
that faster than they get the signal via the ...
04/15/2012
- 08:30 PM Bug #2303 (Can't reproduce): osd: failed to peer on startup
- ubuntu@teuthology:/a/nightly_coverage_2012-04-14-b/994
- 08:24 PM Linux kernel client Bug #2302 (Can't reproduce): xfs: warning at mutex_remove_waiter
- ...
- 03:33 PM Feature #1044 (In Progress): librbd: discard support
- 03:33 PM Feature #2163 (Resolved): qa: full xfstests on rbd
- 03:33 PM Subtask #2249 (Resolved): teuthology task (3)
- 03:33 PM Feature #2226 (Resolved): osd: better filestore idempotency test
- 05:39 AM Linux kernel client Bug #2287: rbd: crashes with 10Gbit network and fio
- Here some more info from the crash:
@[58113.180039] libceph: tid 387083 timed out on osd92, will reset osd
[5818...
04/14/2012
- 10:38 PM Revision 1bc01289 (ceph): Merge branch 'stable'
- 09:06 PM Revision 19e673cc (ceph): filestore_idempotent: use new sequence-based tester
- random seed, inject at 50-300.
- 08:53 PM Revision 20ab74ea (ceph): test_librbd: fix write sizes
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 08:07 PM CephFS Bug #2299: all MDS commit suicide on startup
- after i told osd.0 to get lost and reformatted it, the cluster started resyncing.
then (magically) mds.0 started up ... - 09:39 AM CephFS Bug #2299 (Rejected): all MDS commit suicide on startup
- my setup is: 1 MON, 2 MDS and 4 OSD.
ceph version is commit:1e76a8713feac6883c648512dcdc28c83f7ff69e.
after copyi... - 05:26 PM Revision 81c8a7b9 (ceph): test_idempotent_sequence: no config file
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 04:43 PM Revision a40fe5bd (ceph): run_xfstests.sh: ensure cleanup on errors
- Because we exit on any error (due to 'set -e'), the cleanup call was
never getting made in the event of an error. Th... - 04:41 PM Bug #2301: librados: LibRadosMisc.AioOperatePP failure
- the problem is that the completion callback is now async, but wait_for_complete() is not.
do we think that is ok?
- 02:59 PM Bug #2301 (Resolved): librados: LibRadosMisc.AioOperatePP failure
- 2012-04-14T00:11:00.763 INFO:teuthology.task.workunit.client.0.out:[ RUN ] LibRadosMisc.AioOperatePP
2012-04-14... - 04:26 PM Revision 3d187db5 (ceph): run_xfstests.sh: pass test result via exit status
- It turns out that xfstests *does* exit with non-zero status
when a test fails. Its exit status is the number of test... - 01:58 PM Bug #2300 (Rejected): objecter: not sending stat request
- 01:50 PM Bug #2300: objecter: not sending stat request
- Ah, actually we try to access an object with empty oid, which is obviously wrong. Probably due to #2289 issues.
- 12:34 PM Bug #2300 (Rejected): objecter: not sending stat request
- Happens in rgw (can only see it on congress). Following a rgw.bucket_list call response, we call librados io_ctx->sta...
- 12:38 PM rgw Bug #2289: rgw: listing a bucket hangs after removing inexisting object
- Pushed several fixes to wip-2289. The scenario was:
creating bucket
trying to remove object that does not exist
... - 06:53 AM Revision a4dd9204 (ceph): objclass: fix cls_cxx_map_remove_key()
- didn't set the correct osd op, didn't do anything.
Signed-off-by: Yehuda Sadeh <yehuda.sadeh@dreamhost.com> - 06:10 AM Revision b352a3e1 (ceph): rgw: call dir_suggest_changes with correct bufferlist
- Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
- 05:28 AM Revision 6ba4efcd (ceph): rbd.py: add xfstests functionality
- Add tasks for running xfstests over a pair of rbd volumes. The main
one is called xfstests, and it sets up rbd volum... - 05:27 AM Revision 4498825a (ceph): add rbd_xfstests to regression suite
- 05:19 AM Revision a31392cf (ceph): objectcacher: name them
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 05:18 AM Revision 28e61811 (ceph): librados: add IoCtx::get_pool_name() to c++ api
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 04:51 AM Revision 446b5e24 (ceph): ObjectCacher: move functions out of the header.
- These all set off my "big-functions-in-the-header" alarm.
Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com> - 04:51 AM Revision 6257a94e (ceph): ObjectCacher: add a bunch of perfcounters
- Hurray, we can collect cache statistics!
Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com> - 04:51 AM Revision f355875b (ceph): ObjectCacher: group all the function definitions properly
- Some of the ObjectCacher functions were separate, which works poorly
if you're trying to do things like use ldout.
S... - 04:51 AM Revision d80c126a (ceph): ObjectCacher: use "objectcacher" instead of the CephContext::name for p...
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 04:51 AM Revision 72a75730 (ceph): objectcacher: perf skeleton
- Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Conflicts:
src/osdc/ObjectCacher.cc
src/osdc/ObjectCacher.h - 04:41 AM Revision 2cc854f7 (ceph): msgr: Document all the functions in the Messenger interface.
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
- 04:40 AM Revision b8207dd2 (ceph): msgr: remove several functions from the interface.
- The suicide() function is unused and its current SimpleMessenger
implementation just calls shutdown(). And I don't se... - 04:02 AM Revision bf8847e7 (ceph): qa: add run_xfsests.sh script
- Add a script that runs xfstests over a pair of devices that are
specified using command line arguments. The tests ar... - 04:01 AM Revision ee69f271 (ceph): test/rbd/fsx: fix signed/unsigned warnings
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 03:52 AM Revision 812c7c7a (ceph): librbd: implement discard
- Implement sync and async discard. Embed an ObjectWriteOperation in the
BlockCompletion struct.
The sync version doe... - 03:51 AM Revision 589c6269 (ceph): Merge branch 'wip-librbd-caching'
- Reviewed-by: Sage Weil <sage.weil@dreamhost.com>
- 03:46 AM Revision 5c2d5b18 (ceph): test: copy fsx to rbd directory
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 03:46 AM Revision b26d7024 (ceph): test: port fsx to run on an rbd image instead of a file
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 03:46 AM Revision 78d14da8 (ceph): librbd: flush pending writes when a new snapshot is created
- This makes sure the state is as consistent as librbd can make it
before the snapshot is actually created.
Signed-off... - 03:46 AM Revision 30805bfa (ceph): test_librbd_fsx: clean up properly
- Delete and close everything on exit.
Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com> - 03:46 AM Revision 68e2e004 (ceph): librbd: flush cache before creating a snapshot
- This is a temporary workaround until the ObjectCacher
is smarter about snapshots.
Signed-off-by: Josh Durgin <josh.d... - 03:46 AM Revision d0b1f72c (ceph): ObjectCacher: add some asserts about when the lock is held
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 03:46 AM Revision 7197825d (ceph): Remove unused Completion.h header
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 03:46 AM Revision 1dd9c303 (ceph): ObjectCacher: remove redundant 'public:'
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 03:46 AM Revision 7868dbe0 (ceph): ObjectCacher: remove unused/unimplemented methods
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 03:46 AM Revision ee9987b2 (ceph): librados: debug logging for aio writes
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 03:46 AM Revision 20be92ec (ceph): ObjectCacher: always delete callback in commit_set
- The GatherBuilder used later always deletes it. To make
the interface sane, delete it in every return path.
Signed-o... - 03:46 AM Revision 6cf22655 (ceph): ObjectCacher: remove useless conditional
- We assert this condition at the beginning of the function.
Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com> - 03:46 AM Revision 55b65544 (ceph): test_librbd: flush before opening another handle to the image
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 03:46 AM Revision d9aa81b0 (ceph): finisher: add indent settings header
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 03:46 AM Revision f464e6c3 (ceph): ObjectCacher: include object when outputting a BufferHead
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 03:46 AM Revision 9c8b3f70 (ceph): .gitignore: clean up to not match so many things recursively
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
- 03:46 AM Revision a0fc3336 (ceph): librados: unconditonally associate IoCtxImpl with AioCompletionImpl
- We were only setting the pointer if we were on the aio_write_list. Set
this pointer for any aio.
Signed-off-by: Sag... - 03:46 AM Revision 34a10fae (ceph): librados: do aio callbacks in async thread
- Call user completions in an async thread. This allows callers to call back
into librados from the callback, and allo... - 03:46 AM Revision cbbd5f97 (ceph): ObjectCacher: check for read/write errors
- The objecter doesn't fix every error for us.
Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com> - 03:46 AM Revision abbfd937 (ceph): Filer: make file_to_extents static
- Now the ObjectCacher has no need for a Filer instance (not that
it had a good reason for one before).
Signed-off-by:... - 03:46 AM Revision d6dbff01 (ceph): ObjectCacher: remove dependency on Objecter
- Abstract out how writeback is done with a WritebackHandler object.
For RBD caching, this will be done by librados, bu... - 03:46 AM Revision 9546c158 (ceph): librbd: remove writeback window
- This is superseded by a full-fledged writeback cache.
Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com> - 03:46 AM Revision c0073770 (ceph): librbd: allow writeback caching
- This uses the existing infrastructure of ObjectCacher for
buffer management and expiry.
Signed-off-by: Josh Durgin <... - 03:46 AM Revision 579d30b6 (ceph): librbd: check for writes to snapshots
- librados does this for us normally, but caching does not check for this.
We might as well check early to avoid schedu... - 03:46 AM Revision bb35ecf0 (ceph): librbd: fix bytes read accounting in read_iterate
- ObjectCacher will never do short reads, and always returns 0.
librados may do short reads at the end of an object.
S... - 03:41 AM Revision cc1ae672 (ceph): filestore: flush log on failure injection
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 03:41 AM Revision ad2eba43 (ceph): Merge branch 'wip-2226b'
- Conflicts:
src/common/config_opts.h
src/os/FileStore.cc
src/os/FileStore.h - 03:38 AM Revision 4fc48e5b (ceph): run_seed_to_range.sh: run filestore idempotent sequence tests over a range
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 03:38 AM Revision f052c82b (ceph): run_seed_to: assume tool is in path, not cwd
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 03:38 AM Revision 2cb0032e (ceph): DeterministicOpSequence: writing to object being cloned in the same tx.
- We write to the object being cloned prior to the clone to ensure we are
cloning a valid range of bytes.
The write an... - 03:38 AM Revision c7729847 (ceph): test_idempotent_sequence: simple test script
- Takes seed and kill point, then verifies the result.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 03:38 AM Revision 95bb4239 (ceph): filestore: use our assert
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 03:38 AM Revision a8b9b98c (ceph): run_seed_to.sh: set -e
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 03:38 AM Revision 56fb027f (ceph): test_idempotent_sequence: make get-last-op return 0 if no txns applied
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 03:38 AM Revision 2d24eb33 (ceph): FileStore: don't return ENOENT from object_map getters
- ENOENT in those contexts means that the object_map does not know
about the object, not that the object does not exist... - 03:38 AM Revision 21be4ad9 (ceph): test_idempotent_sequence: Add more info on diff's dout's.
- Signed-off-by: Joao Eduardo Luis <jecluis@gmail.com>
- 03:38 AM Revision bb0499a4 (ceph): test_idempotent_sequence: outputting collection being diff'ed.
- Signed-off-by: Joao Eduardo Luis <jecluis@gmail.com>
- 03:38 AM Revision 27cb82e5 (ceph): test_idempotent_sequence: use get_obj_at() for clone source
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 03:38 AM Revision 1b451294 (ceph): test_idempotent_sequence: fix/enable do_coll_add
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 03:38 AM Revision a5f533a7 (ceph): test_idempotent_sequence: fix diff return value, mount behavior
- No mkfs!
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 03:38 AM Revision 08cf4d28 (ceph): test_idempotent_sequence: include object name in diff output
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 03:38 AM Revision 2d20b4c2 (ceph): test_idempotent_sequence: return true from diff if different
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 03:38 AM Revision 5d30581e (ceph): test_idempotent_sequence: add remove op
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 03:38 AM Revision e270e0b5 (ceph): test_idempotent_sequence: diff two stores.
- Signed-off-by: Joao Eduardo Luis <jecluis@gmail.com>
- 03:38 AM Revision 2c940a4f (ceph): test_idempotent_sequence: Remove references to VerifyFileStore
- Signed-off-by: Joao Eduardo Luis <jecluis@gmail.com>
- 03:38 AM Revision 42899110 (ceph): Makefile: remove VerifyFileStore from test_idempotent_sequence build
- Signed-off-by: Joao Eduardo Luis <jecluis@gmail.com>
- 03:38 AM Revision 84175c7d (ceph): VerifyFileStore: delete from tree
- Signed-off-by: Joao Eduardo Luis <jecluis@gmail.com>
- 03:38 AM Revision 794fc0b6 (ceph): test_idempotent_sequence: fix collection_move arg order
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 03:38 AM Revision a8c32057 (ceph): test_idempotent_sequence: whitesepace
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 03:38 AM Revision bc5dffe9 (ceph): test_idempotent_sequence: replace on coll_move
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 03:38 AM Revision c056f603 (ceph): test_idempotent_sequence: simplify clone
- - pick existing source
- any target is ok, as long as it != source
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 03:38 AM Revision 51bc6dfd (ceph): test_idempotent_sequence: make object names overlap
- This way when we move objects between collections they will replace
each other in interesting ways.
Signed-off-by: S... - 03:38 AM Revision db3d14b9 (ceph): test_idempotent_sequence: fewer objects per collection by default
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 03:38 AM Revision bec1e50e (ceph): test_idempotent_sequence: remove existing target for collection_add
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 03:38 AM Revision 975e59ad (ceph): test_idempotent_sequence: Add commands and lose a couple of optional args.
- Signed-off-by: Joao Eduardo Luis <jecluis@gmail.com>
- 03:38 AM Revision 5ced86be (ceph): filestore: init filestore_kill_at in ctor
- Otherwise we don't get the option for FileStore instances created after
common_init_finish() (which does md_config_t:... - 03:38 AM Revision 9dba3172 (ceph): test_idempotent_sequence: require nonexistent filestore dir
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 03:38 AM Revision 795274bc (ceph): test_idempotent_sequence: include tx number in each transaction
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 03:38 AM Revision 57866946 (ceph): test_idempotent_sequence: implement get-last-op
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 03:38 AM Revision 6573690a (ceph): VerifyFileStore: Check if two FileStore's match after applying a set of...
- With DeterministicOpSequence we are able to reproduce exactly the same
sequence of operations, over and over. However... - 03:38 AM Revision dee47bb2 (ceph): test_idempotent_sequence: Generate a reproducible sequence of txs.
- With this test we aim at reproducing the same sequence of transactions
as long as we are provided with the same seed ... - 03:38 AM Revision 4a6cfcb3 (ceph): deterministicopseq: add collection_rename() support
- 03:38 AM Revision 01daa88d (ceph): test_idempotent_sequence: no need to reinject value that is already there
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 03:38 AM Revision a5a83950 (ceph): filestore: name internally
- We need to allow the perfcounter name to be controlled so that we can have
two instances of FileStore in the same pro... - 03:38 AM Revision 83227550 (ceph): TestFileStoreState: Represent a FileStore's state to be used by tests.
- Instead of having each test creating the same representation of a
FileStore's state, with a map/set of collections an... - 03:38 AM Revision 2f02a8a5 (ceph): DeterministicOpSequence: Generate a reproducible sequence of operations.
- Signed-off-by: Joao Eduardo Luis <jecluis@gmail.com>
- 03:33 AM Revision 583c79af (ceph): filestore: fix collection_add argument names
- No functional changes, just fixing and clarifying argument names so that it
is less confusing/wrong.
Signed-off-by: ... - 03:33 AM Revision dfcadc94 (ceph): filestore: simple failure injections via --filestore-kill-at <n>
- This will make filestore suicide (_exit(1)) on the n'th potential failure
call site. We can potentially fail:
... - 03:33 AM Revision f4001790 (ceph): filestore: two-phase guard
- For certain operations (collection_add) we need a two-phase guard, and an
"in-progress" state.
- before exposing an... - 03:33 AM Revision 8d3acb5a (ceph): filestore: replay collection_move using add+remove
- This approximates the buggy collection_move. It is still buggy. It is
only there to replay old journals.
Rip out b... - 01:14 AM Revision c6551fb8 (ceph): test_rados.py: add basic tests for the Object class
- Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
Reviewed-by: Greg Farnum <gregory.farnum@dreamhost.com> - 01:13 AM Revision 76799680 (ceph): rados.py: fix Object.read argument order
- 57cfd7673da66d191cc397689da35d0a1f3dc61a switched the argument order but didn't change this caller.
Signed-off-by: T...
04/13/2012
- 11:12 PM Revision 5ce8d71f (ceph): FileJournal: allow write_thread and write_finisher to run in parallel
- Previously, the write_thread and the write_finisher serialized on
write_lock limiting performance. aio_lock has been... - 11:03 PM Feature #1044 (Resolved): librbd: discard support
- 11:03 PM Feature #2163: qa: full xfstests on rbd
- 11:02 PM Feature #2052 (Resolved): librbd: caching
- 06:37 PM Feature #2052: librbd: caching
- This is passing long-running fsx with osd thrashing consistently, and all the other rbd tests. I think the branch (wi...
- 10:40 PM Linux kernel client Bug #2298 (Resolved): rbd: broken encode_op for big-endian hosts?
- ...
- 10:17 PM Subtask #2249: teuthology task (3)
- 09:26 PM Subtask #2237 (Resolved): failure+replay tester (8)
- 06:39 PM Bug #2278 (Resolved): librados: python read has arguments swapped
- Fixed by 76799680546a79fc73ad7bbc58960a31ae2290ad.
- 10:10 AM Bug #2278: librados: python read has arguments swapped
- 07:56 AM Bug #2278 (Resolved): librados: python read has arguments swapped
- Object.read from rados.py is passing arguments to ioctx.read in a wrong order.
--- rados.py.dist 2012-04-13 16:5... - 06:38 PM rbd Feature #2297 (New): ObjectCacher: mark buffers mergeable for ksm
- This is done with a simple madvise call, but we should test that it works with ksm and verify that all the buffers ar...
- 06:29 PM Feature #2296 (Resolved): librbd: allow resizing to arbitrary sizes
- Right now resizing to a non-object-size multiple will round down the remainder. With discard support, we support this...
- 06:25 PM Feature #2295 (Resolved): make qemu cache=writeback,writethrough option turn on librbd caching
- This will enable more familiar use of caching with qemu/rbd, and let people configure it with libvirt's existing xml.
- 05:51 PM rbd Feature #2294 (New): librbd: optionally cache entire objects, instead of only requesting the part...
- This may save many round trips for small read sizes (common to vms).
- 05:49 PM Feature #2113 (Fix Under Review): objectcacher perfcounters
- Okay, I checked and these work — if you run ceph-fuse -f and play around you can do a dump_perfcounters and see the v...
- 05:46 PM CephFS Bug #2293 (Resolved): admin sockets don't persist with ceph-fuse
- It looks like the admin socket is associated with the launching process, rather than the background process that cont...
- 05:46 PM rbd Feature #2292 (New): ObjectCacher: support sparse objects
- The ObjectCacher doesn't store which objects or parts of objects don't exist. This info could improve read performance.
- 05:46 PM Revision dc44950b (ceph): filestore: implement collection_move() as add + remove
- This ensures we get add and remove steps with different spos values, which
makes the guard work. The collection_move... - 05:46 PM Bug #2291 (Can't reproduce): objectcacher perfcounters don't work with test_librbd_fsx
- The admin socket perfcounters_dump command only outputs objecter data. I'm speculating that it has to do with the obj...
- 05:30 PM Feature #2290 (Resolved): ObjectCacher: handle read/write errors
- Currently the return value of the underlying read/write calls is ignored (I left TODO notes there). We should figure ...
- 05:02 PM rgw Bug #2289 (Resolved): rgw: listing a bucket hangs after removing inexisting object
- 04:59 PM Revision dc10d04e (ceph): OSD: improve information and format of OSDTracker messages
- Also, Message now has a timestamp indicating when the message
was fully recieved for use by OSDTracker.
Signed-off-b... - 03:12 PM Subtask #2235 (Resolved): generate deterministic sequence of transactions (5)
- 02:01 PM CephFS Bug #2288 (Resolved): libcephfs: setxattr returns EEXIST following removexattr
- running cephtest a couple of time (out of wip-testlibcephfs):...
- 01:48 PM Linux kernel client Bug #2287 (Resolved): rbd: crashes with 10Gbit network and fio
- From http://permalink.gmane.org/gmane.comp.file-systems.ceph.devel/5968:...
- 01:47 PM Bug #2286 (Resolved): mon: different full/near_full values on different monitors
- If you run vstart, you get...
- 01:39 PM CephFS Bug #2285 (Resolved): libcephfs: failure with empty name components
- the following in client/testceph.cc fails:
// test empty name components
my_fd = ret = ceph_open(cmount, "rea... - 11:19 AM rgw Feature #2284 (Resolved): rgw: bench based on rados_bench
- 11:17 AM rgw Feature #2171 (Rejected): rgw: asynchronously calculate md5
- 11:16 AM Feature #2283: The ceph command should time out
- 10:21 AM Feature #2283 (New): The ceph command should time out
- When using ceph to query certain parts of the cluster, there should be an option to time out after a certain set numb...
- 09:44 AM Subtask #2282 (Resolved): Handle map updates on a per-pg basis
- Currently, we advance all pgs to the next map at once. This requires us to flush the filestore queue and basically h...
- 09:27 AM Feature #2281 (Resolved): build big burnupi cluster for testing
- 09:23 AM Feature #2280 (Resolved): improve gitbuilder infrastructure
- * do not fill up local disk; sync results out immediately
* resolve branches immediately, not after each full pass
? - 09:20 AM rbd Feature #2279 (Resolved): rbd: trivial layering design doc
- - how parent images are marked read-only
- how parent/child relationship is represented
- possibly how this allow... - 09:16 AM Bug #2192 (Need More Info): ceph-mon hangs consuming 100% CPU
- 09:14 AM Feature #2246 (Resolved): force10s on sepia
- 09:13 AM Messengers Cleanup #2150 (Resolved): repair the Simple/Messenger interface
- 09:13 AM Feature #2240 (Resolved): osd: new default locations
- 06:16 AM Revision 3d11455b (ceph): osdmaptool: fix clitest for lpg_num 0
- Broke this in 11b93d3af2c0f1bca3e48fe773a14f07f479ccc3
Signed-off-by: Sage Weil <sage@newdream.net> - 05:56 AM Revision 55535d04 (ceph): move tasks:cfuse_workunit_suites_dbench.yaml to stress pending #1737 fix
- 05:36 AM Revision 1e76a871 (ceph): Merge remote branch 'gh/wip-defaults'
- 04:47 AM Revision 8a6c3c17 (ceph): msgr: dispatcher: Documentation of security functions
- Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
04/12/2012
- 11:24 PM Revision 9b2aff20 (ceph): FileStore: don't return ENOENT from object_map getters
- ENOENT in those contexts means that the object_map does not know
about the object, not that the object does not exist... - 11:17 PM Subtask #2237 (In Progress): failure+replay tester (8)
- 11:17 PM Subtask #2235: generate deterministic sequence of transactions (5)
- 11:15 PM Feature #2240: osd: new default locations
- 10:58 PM CephFS Bug #2277 (New): qa: flock test broken
- ubuntu@teuthology:/a/nightly_coverage_2012-04-12-b/687
ubuntu@teuthology:/a/nightly_coverage_2012-04-11-b/525
thi... - 10:50 PM Revision 241c5e5e (ceph): msgr: Reorder functions to reduce the number of privacy tags.
- No functional changes.
Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com> - 10:48 PM CephFS Bug #1737: ceph-fuse crash in xlist::remove
- ubuntu@teuthology:/a/nightly_coverage_2012-04-12-b/717
- chef: null
- ceph: null
- ceph-fuse: null
- workunit:
... - 10:45 PM CephFS Bug #2187: pjd chown/00.t failed test 97
- 2012-04-12T13:09:27.496 INFO:teuthology.task.workunit.client.0.out:../pjd-fstest-20080816/tests/chown/00.t (Wstat: ...
- 10:45 PM Revision 5cf7628f (ceph): msgr: miscellaneous fixes.
- Indentation on a public label; various function documentation; remove
an inappropriate definition of a virtual functi... - 10:35 PM Bug #2276 (Rejected): osd: eat cpu on restart
- osd.856 on congress.
- 10:17 PM Revision 8e5f5625 (ceph): DBObjectMap: check in clone and link that source is not destination
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 09:35 PM Bug #2275 (Resolved): osd: crash in FileJournal::wrap_read_bl
- ...
- 07:16 PM Revision a5730c85 (ceph): Merge branch 'wip-2223'
- Reviewed-by: Sage Weil <sage@newdream.net>
- 05:59 PM Revision 9565a8ba (ceph): ObjectStore: Remove code duplication when dumping transactions.
- By using OStreamFormatter, we can have a single function responsible for
dumping a transaction. We keep the same old ... - 05:59 PM Revision ddbaf584 (ceph): filestore: remove commented code, fix whitespace
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 05:59 PM Revision 7223f00a (ceph): filestore: dump transactions sets, include seq and trans_num
- Lists of transactions commit as a unit; reflect that in the dump. Also
include seq and trans_num.
Signed-off-by: Sa... - 05:59 PM Revision 816a5128 (ceph): objectstore: tweak dump() a bit
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 05:59 PM Revision 308c1fc6 (ceph): filestore: use config observer to start/stop dumping/tracing
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 05:59 PM Revision 6ee0b718 (ceph): filejournal: dump injson
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 05:59 PM Revision 6a7bbe86 (ceph): filestore: dump transaction in json on error
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 04:29 PM Documentation #2274 (Closed): Basic Availability Model
- (1) Construct a continuous-time markov availability model for a basic cluster (3 mons, 4 osds, 2 copy)
(Petri ne... - 04:19 PM Documentation #2273 (Closed): basic reliability models
- 1. construct a probabilistic model for data loss in 1, 2, and 3 copy systems, assuming independent failures
2. plug ... - 04:13 PM RADOS Documentation #2272 (Closed): FAQs: RADOS reliability and availability
- I expect others to improve this, but this is just to capture the ideas.
It is probably more of a white paper than an... - 04:06 PM Documentation #2271 (Resolved): FAQ: BTRFS vs XFS
- I expect others to improve this list, but to start it out ...
what file systems we run on (and test on)
how you... - 12:15 PM Feature #2223 (Resolved): Tracing facility on FileStore
- 09:05 AM RADOS Feature #2268 (Resolved): crush: update item's position in crush map
- via crushtool and 'ceph osd crush ...'
- 03:55 AM Bug #2267 (Closed): Ceph client crashed after shutting down one mds and osd
- Ceph version: 0.44.1-1~bpo70+1
Kernel version: 3.2.12-1
Ceph config:
[global]
auth supported = cephx
keyri... - 12:38 AM Revision cfac4a97 (ceph): pgmon: clear out osd reports after the OSD has gone down
- Previously we never removed report times from last_osd_report. Do
so, in check_osd_map (which, on the leader, is call...
04/11/2012
- 11:00 PM Revision 119dd5ae (ceph): mkcephfs: update man page
- Signed-off-by: Sage Weil <sage@newdream.net>
- 11:00 PM Revision 4a4b7994 (ceph): ceph-authtool: update man page
- Signed-off-by: Sage Weil <sage@newdream.net>
- 10:53 PM Revision ab08fb8b (ceph): mkcephfs: note that btrfs (and --mkbtrfs) are optional and experimental
- And that --mkbtrfs will be deprecated soon.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 10:53 PM Revision ee39291a (ceph): ceph-authtool: add warning to man page
- - data is not encrypted over the wire
- intended for trusted environments
Signed-off-by: Sage Weil <sage.weil@dreamh... - 10:40 PM Revision 11b93d3a (ceph): osd: disable localized pgs by default
- Signed-off-by: Sage Weil <sage@newdream.net>
- 10:40 PM Revision 8836b81f (ceph): mon: alloc pgp_num adjustment up and down
- Signed-off-by: Sage Weil <sage@newdream.net>
- 10:39 PM Revision 83e1260b (ceph): mon: set pgp_num == pg_num (by default) for new pools
- For when pg_num is specified but not pgp_num. Thanks Greg!
Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com> - 10:39 PM Revision 58671a4c (ceph): mon: command to disable localized pgs for a pool
- ceph osd pool disable_lpgs <poolname> --yes-i-really-mean-it
Grr, these should be off by default. We can't adjust t... - 08:35 PM Revision 7fdf25bc (ceph): debian: python-support -> dh_python2
- I followed the instructions on
http://wiki.debian.org/Python/TransitionToDHPython2
Signed-off-by: Sage Weil <sage@... - 07:35 PM Revision ed0653b4 (ceph): COPYING: doc/ CC BY-SA
- Signed-off-by: Sage Weil <sage@newdream.net>
- 07:33 PM Revision 6e83e119 (ceph): README: update
- - refer to COPYING, SubmittingPatches
- a word about dependencies
- building packages
- drop the list of built binari... - 06:42 PM Revision 838a7618 (ceph): ceph-rbdnamer: include in dist tarball and debs/rpms
- Signed-off-by: Sage Weil <sage@newdream.net>
- 06:42 PM Revision af502735 (ceph): obsync: include man page in tarball, packages
- Signed-off-by: Sage Weil <sage@newdream.net>
- 06:42 PM Revision 9678c097 (ceph): init-radosgw: start in runlevel 4
- Fixes lintian error
W: radosgw: init.d-script-missing-start etc/init.d/radosgw 4
Signed-off-by: Sage Weil <sage@new... - 06:42 PM Revision 84efc554 (ceph): debian: drop unnecessary conflicts on librgw
- Cut and paste baggage from libcephfs, it looks like.
Signed-off-by: Sage Weil <sage@newdream.net> - 06:18 PM Messengers Cleanup #2150 (In Progress): repair the Simple/Messenger interface
- I haven't done it, but I had enough time to glance over it and see at least a couple things that need fixing before t...
- 05:49 PM Feature #2113: objectcacher perfcounters
- Sage asked me to run it under an rbd mount and look at it. Need to get tests from Josh and then figure out how to do ...
- 04:30 PM Feature #2113 (Fix Under Review): objectcacher perfcounters
- Compile-tested.
- 10:51 AM Feature #2113 (In Progress): objectcacher perfcounters
- Yoink.
- 05:40 PM Revision 292898a8 (ceph): init-ceph: start at all runlevels
- This fixes lintian error:
W: ceph: init.d-script-missing-start etc/init.d/ceph 4
Signed-off-by: Sage Weil <sage@new... - 05:03 PM Revision b1946290 (ceph): Merge branch 'stable'
- 04:30 PM Bug #2266 (Resolved): teuthology: nuke after failure is failing
- it fails, and then fails to unlock, and eats up machines.
for example, ubuntu@teuthology:/a/nightly_coverage_2012-... - 03:08 PM Feature #2265 (Rejected): make sure objecter/kclient error out when localized pgs don't exist
- 11:02 AM Bug #2264 (Can't reproduce): mon: failed assert in bump_epoch
- During startup of a teuthology run on commit 1775301bb46379648f3f88914ef56aa1982db020 (before the cluster was healthy...
- 10:48 AM Bug #2263 (Resolved): obsync: move man page to section 1
- 09:25 AM Bug #2262 (Resolved): qa: osd-recovery tasks fails on flush_pg_stats
- consistently
- 08:09 AM Linux kernel client Bug #2260: libceph: null pointer dereference at try_write+0x638+0xfb0
- Looks like the problem arose while running fsstress on the xfs loop
mount on top of a file on the ext2 filesystem.
... - 07:56 AM Linux kernel client Bug #2260: libceph: null pointer dereference at try_write+0x638+0xfb0
- FYI, xfstests 49 tests running XFS on a loop device. I have to wait for a
reboot in order to see if I can tell at w... - 07:49 AM Linux kernel client Bug #2260: libceph: null pointer dereference at try_write+0x638+0xfb0
- Looks like xfstests #49 is a reproducer for this problem, at least
after running the tests that lead up to it first ... - 05:47 AM Revision be5b25b6 (ceph): filestore: fix collection_move guard
- We had a sequence like:
1- write A block 1
2- write A block 2
3- write A block 3
4- write A block 4
5- move A -... - 05:47 AM Revision 4bd9d1bb (ceph): filestore: fix collection_add guard
- If we crash between the link() and setting the guard, we will get
EEXIST. Tolerate that.
Signed-off-by: Sage Weil <... - 05:47 AM Revision df4d7a47 (ceph): filestore: fix collection_rename guard
- If we crash between the rename and setting the guard, we can get EEXIST
or ENOTEMPTY on rename. Tolerate that.
Sign... - 05:47 AM Revision 85db25e8 (ceph): filestore: fix fd leak on collection_rename
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 05:47 AM Revision c3e4c5b7 (ceph): filestore: cleanup: flip sense of replay guard check
- The other are all if (_check_replay_guard(..)) do_it;. Make this one
match.
Signed-off-by: Sage Weil <sage.weil@dre... - 05:43 AM Revision 43de5e4f (ceph): FileStore: dumping transactions to a file
- Dump each queued transaction to a predefined file, specified with
--filestore-dump-file, in JSON format.
Signed-off... - 05:43 AM Revision cd4a760e (ceph): osd: fix heartbeat set_port()
- set_port() fails an assert if it isn't an in4 or in6 address, which a
default entity_addr_t is not.
Signed-off-by: S... - 05:29 AM Linux kernel client Bug #2261 (In Progress): paging error in libceph after crashed osd comes back online
- 05:22 AM Linux kernel client Bug #2261 (Can't reproduce): paging error in libceph after crashed osd comes back online
- ...
- 04:43 AM Revision 1775301b (ceph): osd: reenable clone on recovery
- This hasn't turned up problems in QA.
Fixes: #2002
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 02:25 AM Bug #2178: rbd: corruption of first block
- Well Sage,
I have a torture-test already :-D
OK, so it's independent from yours and that's good. It sounds, we ar...
04/10/2012
- 11:24 PM Feature #2223: Tracing facility on FileStore
- did some cleanup, changed the way the output is structured wrt the transaction lists, and tweaked a few other things....
- 11:23 PM Revision ddb98f77 (ceph): ceph_manager: don't try to start greenlet twice
- spawn already scheduled it. Trying to start it again hits an assert.
- 11:11 PM Revision 6fbac10d (ceph): osd: allow users to specify the osd heartbeat server address.
- Reported-by: Nick Bartos <nick@pistoncloud.com>
Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
Reviewed-by... - 10:23 PM Bug #2002 (Resolved): osd: racy push/pull for clones
- 10:19 PM Bug #2161 (Resolved): nonlinear scaling for PGMap::pg_stat encode
- commit:bd518e998c0ff12d611db19a8cff6da3622597cb
- 10:18 PM Bug #1953 (Resolved): teuthology: core files aren't archived when using valgrind
- it works!
- 10:10 PM Bug #2225 (Resolved): gitbuilder.ceph.com returning 503: Service Temporarily Unavailable.
- Yehuda found the bad apache option.. override it in the domain_service (maxconnperip=1000 param)
- 09:56 PM Revision 4f030e1b (ceph): osd_types: fix off by one error in is_temp
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
Reviewed-by: Gregory Farnum <gregory.farnum@dreamhost.com> - 09:49 PM Messengers Cleanup #2150 (Resolved): repair the Simple/Messenger interface
- 09:49 PM Feature #1044 (Fix Under Review): librbd: discard support
- 09:48 PM Revision 31f16a4c (ceph): rgw: list multipart response fix
- LastModified was formatted outside of the Part block.
Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> - 09:06 PM Revision 89fecda6 (ceph): Makefile.am: remove some clutter
- Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
- 09:04 PM Linux kernel client Bug #2260: libceph: null pointer dereference at try_write+0x638+0xfb0
- I'm going to have to look at this again in the morning, but I think
we're in this block of code:
#ifdef CONFIG_BL... - 08:37 PM Linux kernel client Bug #2260: libceph: null pointer dereference at try_write+0x638+0xfb0
- Here's a disassembled block of the code where the fault occurred.
The address listed corresponds to offset 3468 belo... - 08:10 PM Linux kernel client Bug #2260 (Resolved): libceph: null pointer dereference at try_write+0x638+0xfb0
- It's not an exact match but it's close enough that I wanted to reopen
bug 1793 or 1866, but found myself unable to. ... - 08:41 PM Revision 1ac5554d (ceph): kernel: kludge around mysterious 0-byte .git/HEAD files
- No idea where these are coming from, but they break nodes with behavior
like
ubuntu@plana08:~$ sudo install -d -m075... - 05:42 PM Revision 0aea1cb1 (ceph): v0.45
- 04:17 PM Revision 0d5918f8 (ceph): kernel: reset to remote firmware branch; don't pull
- Pull might merge if upstream rebases. Just make our branch match the
remote one. - 04:12 PM Revision 9b755fd6 (ceph): kernel: change git incantation for firmware pull
- The 'git pull <uri>' seemed to consistently fail on some nodes. Can't be
sure this was really the problem with them ... - 03:59 PM Revision 22b1f17f (ceph): ls: another newline
- 03:57 PM Revision 7757fbb9 (ceph): ls: remote stray newline
- 03:27 PM Feature #2246: force10s on sepia
- Fabric brought up by Networking group. Interfaces up, configured, and working (nuttcp shows 9.5GB/s or so with
defa... - 01:26 PM Feature #2111: msgr workloads
- I think the messenger tester may be at a point where we can call this bug satisfied.
- 01:18 PM Bug #2178: rbd: corruption of first block
- the good news is i see the problem. the bad news is its the exact bug we thought we fixed. the other good news is w...
- 07:38 AM Bug #2178: rbd: corruption of first block
- Hi Sage,
just in case, the reply from yesterday did not reach you:
--- 8-< ---
Good morning,
it's already... - 12:27 PM Feature #2258 (Resolved): use external leveldb package
- autoconf lets you use the installed library. not doing so by default to avoid the pain of building on older distros.
- 04:22 AM Revision 965f83d4 (ceph): Merge branch 'next'
- 04:20 AM Revision d348e1ab (ceph): configure: --with-system-leveldb
- Default to bundled leveldb. Optionally check.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com> - 04:20 AM Revision 34cc308e (ceph): filestore: fix leveldb includes
- Signed-off-by: Sage Weil <sage@newdream.net>
- 03:23 AM Revision 0b2e1cd2 (ceph): cephfs: fix uninit var warning
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
04/09/2012
- 11:58 PM Revision 9906d5ed (ceph): Change to local mirror of linux-firmware repo to try to stop failures
- 11:17 PM Revision f79b95e5 (ceph): Makefile: add missing .h to tarball
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 10:56 PM Revision 8d5c87a8 (ceph): rgw: fix object name with slashes when vhost style bucket used
- Fixes issue #2259. The problem was that we were initializing the
object name, then in the case of a virtual host buck... - 09:02 PM Revision 853b0458 (ceph): OSD: use per-pg temp collections, bug #2255
- Signed-off-by: Samuel Just <samuel.just@dreamhost.com>
- 07:08 PM Revision 36d42dea (ceph): buffer: allow advance() to move an iterator backward
- Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
- 07:08 PM Revision bd518e99 (ceph): encoding: fix iterator use for struct_len copy_in
- The end() iterator position does not record an offset when the list is
modified.
Signed-off-by: Sage Weil <sage.weil... - 04:30 PM rgw Bug #2259 (Resolved): rgw: object name cut after slash when virtual host style is used
- Fixed, commit:8d5c87a86e070b4e95ef0d58a469bdbbef4a826c.
- 03:42 PM rgw Bug #2259 (Resolved): rgw: object name cut after slash when virtual host style is used
- 09:32 AM Bug #2178: rbd: corruption of first block
- The missing piece of information is mapping the file offset to a block device offset. Can you, inside the VM,...
- 03:59 AM Revision 7951d7e4 (ceph): Merge remote branch 'gh/stable' into next
- 03:58 AM Revision dd8fd168 (ceph): configure: HAVE_FALLOCATE -> CEPH_HAVE_FALLOCATE
- /usr/include/linux/fs.h defines this on CentOS 5, even though it does not
in fact compile. This stupid workaround av...
04/08/2012
- 09:53 PM Feature #2258 (Resolved): use external leveldb package
- - make our configure take/require a --with-system-leveldb or similar to not use the bundled leveldb
- update the deb... - 08:31 AM Bug #2178: rbd: corruption of first block
- Hi Sage and *Happy easter*,
yesterday I had some "luck" after 10 tries....
Here is what I have for you:
first ...
04/06/2012
- 09:27 PM Feature #1692 (Duplicate): librbd: Support TRIM (hole punching) (userspace client)
- dup of #1044
- 09:07 PM Revision 8e1cc8ab (ceph): init-ceph: manage pid_file from init script
- With upstart the daemon shouldn't manage the pid file itself. Move this
out of the default config and into the legac... - 08:48 PM Revision 81d2cbeb (ceph): config: move /var/run and /var/log defaults to config_opts.h
- This flips the sense of the common_init defaults. Before, the alternate
defaults were filled in if it was a daemon. ... - 08:39 PM Revision dfa043df (ceph): config: {osd,mon}_data default to /var/lib/ceph/$type/$cluster-$id
- Signed-off-by: Sage Weil <sage@newdream.net>
- 07:26 PM Revision 2ceda946 (ceph): Merge branch 'stable'
- 06:44 PM Revision 7680cdad (ceph): dencoder, rgw: make ceph-dencoder load much faster
- by avoiding linking with unneeded shared objects.
Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> - 05:00 PM Revision 98326968 (ceph): encoding: use iterator to copy_in encoded length
- This gives us a pointer to the position into the list where the final
length value will be copied. Previously we use... - 03:47 PM rgw Feature #2257 (Rejected): rgw: detect fastcgi module 100-continue support automatically
- The current default that is used doesn't work with vanilla fastcgi module. It'd be great if that could be set automat...
- 02:46 PM rbd Feature #2256 (Resolved): rbd: parallelize deletions
- There are a few places where we delete things one at a time: resizing to a smaller size, deleting all snapshots, and ...
- 02:04 PM Feature #2240 (Fix Under Review): osd: new default locations
- wip-defaults
- 12:05 PM Bug #2161: nonlinear scaling for PGMap::pg_stat encode
- wip-encoding
- 09:18 AM Bug #2161: nonlinear scaling for PGMap::pg_stat encode
- Ake van der Meer wrote:
> My ceph-osd processes run at 100% CPU for many minutes at a time doing this: http://pasteb... - 08:25 AM Bug #2161: nonlinear scaling for PGMap::pg_stat encode
- My ceph-osd processes run at 100% CPU for many minutes at a time doing this: http://pastebin.com/wYnPKWeJ
In src/i... - 10:05 AM Feature #2246 (In Progress): force10s on sepia
- Ports being mapped yesterday and today in preparation for switch config review.
- 09:21 AM Bug #2255 (Resolved): osd: fix object name collisions between pools in temp collection
- 08:28 AM Feature #2223: Tracing facility on FileStore
- Made some changes to the ObjectStore.cc, regarding code duplication of the transaction's dump methods. Feedback would...
Also available in: Atom