Sage Weil
2c7cd49bd9
common: break init into set_defaults and init steps
...
This will let us tweak the default daemon/non-daemon settings as needed,
e.g. in the case of cfuse.
2010-07-01 09:25:43 -07:00
Sage Weil
8043ee277c
debug: fixup
2010-07-01 09:10:41 -07:00
Sage Weil
85e626d178
common: disable logger by default; make daemon settings explicit in (3) callers
2010-07-01 08:55:52 -07:00
Sage Weil
c2538732d2
debug: add max count for name symlinks
2010-07-01 08:47:35 -07:00
Sage Weil
633af0453d
vstart: normalize pid file names
2010-07-01 08:47:35 -07:00
Sage Weil
79991ed498
debug: revamp debug/logging
...
- By default, append to $type.$name.log.
- Get old $hostname.$pid + $type.$name symlink behavior only with
g_conf.log_per_instance
- Add new g_conf.log_file option to force a particular file.
2010-07-01 08:47:35 -07:00
Greg Farnum
1bd46c0480
client: set MetaRequest::inode to the inode or directory inode whenever possible.
...
Provides link to caps, useful when requests come back ESTALE or similar.
2010-06-30 16:10:12 -07:00
Greg Farnum
e5b60faac7
client: set inode/dentry fields where available
2010-06-30 16:10:11 -07:00
Greg Farnum
beeb08bb09
client: remove now-unused authority and pick_replica functions
2010-06-30 16:10:09 -07:00
Greg Farnum
fa62e9db4b
client: rewrite choose_target_mds to more closely mirror kclient logic and be correct.
2010-06-30 16:08:56 -07:00
Greg Farnum
39796b4794
client: make fragmap updates a little more space efficient.
...
If the auth is the same as the parent, don't add to the map (and remove if
it's already there). Absence is interpreted the same as a value of -1.
2010-06-30 16:08:07 -07:00
Greg Farnum
60f2672ee2
client: add mds_num to MDSSession struct
2010-06-30 16:08:07 -07:00
Greg Farnum
0992001c1e
client: clean up Inode::authority
2010-06-30 16:06:52 -07:00
Greg Farnum
4f2f78dd0e
client: move all mds selection code into choose_target_mds
2010-06-30 16:06:52 -07:00
Greg Farnum
060d22d043
client: remove dead code
2010-06-30 16:06:52 -07:00
Greg Farnum
194986a3a1
client: remove Inode::dir_auth data member
2010-06-30 16:06:52 -07:00
Greg Farnum
abd26eea23
client: initialize auth_caps in Inode! Prevents bad pointer derefs.
2010-06-30 16:06:52 -07:00
Greg Farnum
18c305abb8
qa: set runoncfuse.sh to run cfuse in background and kill at end of tests.
...
This is an imperfect solution at best, since it doesn't handle errors at all,
but it'll have to do for now.
2010-06-30 16:06:52 -07:00
Yehuda Sadeh
bfd335c2b1
rbd: some header cleanup
2010-06-30 16:06:57 -07:00
Yehuda Sadeh
77a1664af6
rbdtool: implement snap rollback
2010-06-30 16:06:50 -07:00
Sage Weil
2fdd43e71f
buffer: handle write_fd() on empty bufferlist
2010-06-30 15:44:38 -07:00
Sage Weil
8410e6a65a
osd: fix, cleanup ack/disk reply logic
...
There was a bug where we would get no reply if we could send ondisk but
the client didn't want it. This simplifies and cleans up the checks
to make more sense, removing the can_* helpers that were hiding which
checks were being done.
Signed-off-by: Sage Weil <sage@newdream.net>
2010-06-30 14:40:20 -07:00
Sage Weil
aa30dd18aa
buffer: make write_fd() tolerate zero length ptrs in the bufferlist
2010-06-30 14:18:06 -07:00
Sage Weil
3235abe918
msgr: clean up aborted message handling
...
No need for a dummy message. Make the (single) caller handle returning
no message as a non-error.
Signed-off-by: Sage Weil <sage@newdream.net>
2010-06-30 12:09:32 -07:00
Sage Weil
7cedafd3e3
msgr: release bytes reserved from throttlers in failure paths
...
If we don't release those bytes, the throttler count eventually fills up
with bytes we were going to read but didn't (due to socket errors, etc)
until we can't read anything.
Signed-off-by: Sage Weil <sage@newdream.net>
2010-06-30 12:09:31 -07:00
Sage Weil
3c80b9dafe
uclient: rename make_path; cleanup/move make_request_from_Meta
...
Document, make it clear what kind of "path" it's really building.
Signed-off-by: Sage Weil <sage@newdream.net>
2010-06-30 11:14:50 -07:00
Yehuda Sadeh
e76ce44131
man: add man pages for cclass and cclsinfo
2010-06-30 10:32:21 -07:00
Sage Weil
2f9ea76fba
filejournal: fix buffer realignment condition; useful debug output on error
...
We don't care about contiguous, only about page alignment.
2010-06-30 08:34:26 -07:00
Yehuda Sadeh
134a0d0f72
rbdtool: fix uninitialized variable
2010-06-29 22:58:52 -07:00
Yehuda Sadeh
0627965198
rbdtool: implement rename
2010-06-29 17:23:18 -07:00
Yehuda Sadeh
015faa8853
config: update __isarg before reading config param
2010-06-29 17:23:18 -07:00
Sage Weil
4d9f49a725
debian: include cclsinfo
2010-06-29 16:25:27 -07:00
Sage Weil
013f9e52e3
Makefile: include cclsinfo; lose the old .cc
2010-06-29 16:23:11 -07:00
Sage Weil
0812b8f4d9
Makefile: include cclass.in in dist
2010-06-29 16:08:33 -07:00
Yehuda Sadeh
f9dc4b9091
buffer: write_fd should skip empty buffers
2010-06-29 16:00:05 -07:00
Yehuda Sadeh
96b1db80d0
rbdtool: init rbd block id, later will be used for rename
2010-06-29 16:00:05 -07:00
Yehuda Sadeh
c9930900e6
cclass.in: LIBDIR=.libs in certain cases
2010-06-29 16:00:05 -07:00
Yehuda Sadeh
0f4ddbac3b
cls: cls_read, cls_cxx_read return the number of bytes read
2010-06-29 16:00:05 -07:00
Sage Weil
2ec729d1fe
config: use <<20 for MB
2010-06-29 14:40:30 -07:00
Sage Weil
2df8b9fd49
script/plot.pl: don't pause
...
Run like so:
$ script/plot.pl path/to/log osd c_wrb [smooth bezier] | gnuplot -persist
2010-06-29 14:40:23 -07:00
Sage Weil
fcc39c8113
msgr: use dedicated reaper thread
...
We were calling the reaper from the wait() loop. The problem is that
the OSD has two messengers, and only the first was in wait().. the second
wait() was only called after the first terminated (i.e, when the OSD was
shutting down).
Instead, launch a separate reaper thread when we bind, and close it out
on shutdown right after the accepter.
2010-06-29 14:40:23 -07:00
Sage Weil
3e334024f4
osd: removed unused RepGather::indata
...
Signed-off-by: Sage Weil <sage@newdream.net>
2010-06-29 14:38:39 -07:00
Sage Weil
852ab94a22
osd: change write osd logging timing
2010-06-29 14:38:28 -07:00
Sage Weil
419bd914dc
osd: always use original Connection when replying
...
...even when the op came from another OSD. Not that that should happen
anyway, since we don't forward messages currently. (And can't, since the
OSD doesn't initiate connections to the client!)
2010-06-29 14:32:28 -07:00
Sage Weil
def4b40e96
osd: always include osd op result, result code in the first reply
2010-06-29 14:31:12 -07:00
Sage Weil
e85d98ba62
osd: track open repops in logger
...
Signed-off-by: Sage Weil <sage@newdream.net>
2010-06-29 14:21:06 -07:00
Sage Weil
ff8df7dcca
osd: add 'logger reset' command
2010-06-29 14:21:06 -07:00
Sage Weil
fc1d1665d9
journal: set max journal write to 10MB
...
If we take too big a bite of data to write in a single writev(2), we can
end up making performance worse, because everyone waits for the full write
to complete. Bigger writes mean better throughput but higher latency.
So, balance the two by placing some upper limit.
2010-06-29 14:21:05 -07:00
Yehuda Sadeh
1d517986a8
conf: fix parsing when there's no eol at the end of file
2010-06-29 09:59:26 -07:00
Sage Weil
d9dc7cb581
msg: fix entity_addr_t::parse() to return false on failure
2010-06-29 08:34:53 -07:00
