Yehuda Sadeh
ddb88b575a
Merge branch 'unstable' of ssh://ceph.newdream.net/git/ceph into unstable
2009-03-09 14:32:17 -07:00
Yehuda Sadeh
2c7a38f027
kclient: sysfs, add mds name and op of current requests
2009-03-09 14:31:42 -07:00
Sage Weil
207403a5e4
kclient: simplify osd timeouts
...
Call handle_timeout N seconds after oldest request. In
handle_timeout, reschedule N seconds in future. Reschedule
with new oldest request when oldest requests completes.
Maintain a r_timeout_stamp value for each request.
2009-03-09 13:58:13 -07:00
Sage Weil
0422395507
cosd: fewer osds, with big btrfs devices
2009-03-09 13:57:22 -07:00
Sage Weil
180f416872
initscripts: allow vars with spaces
2009-03-09 13:18:56 -07:00
Sage Weil
04ca68e65b
osdmap: build default crush maps out of mostly straw buckets
2009-03-09 13:18:05 -07:00
Sage Weil
b734092c2a
confutils: make max line very large
2009-03-09 12:51:04 -07:00
Yehuda Sadeh
d9b2f7b5a1
Merge branch 'unstable' of ssh://ceph.newdream.net/git/ceph into unstable
2009-03-09 10:08:51 -07:00
Yehuda Sadeh
927926e903
kclient: add sysfs directory per mds request
2009-03-09 10:08:29 -07:00
Sage Weil
626d1db180
filepath: always prepend ino if it's nonzero
...
That means the only time we don't have an ino perfix is for fully
relative paths.
2009-03-06 16:45:55 -08:00
Sage Weil
7966f989e5
client: fix request paths
...
Use filepath, not bare string, whenever we can.
2009-03-06 16:44:37 -08:00
Sage Weil
04d6926e74
kclient: fix up replay after mds recovery
...
Do not kick unsafe requests in kick_requests (and in particular, do not
drop the session refs).
Set retry_attempt to MDS_REQUEST_REPLAY to indicate which requests are
replays.
2009-03-06 16:27:19 -08:00
Sage Weil
cd41295f53
mds: incorporate reconnect cap info after requests replay if ino originally missing
...
If the client reasserts caps from replayed requests, the inodes don't yet
exists during the reconnect (or even rejoin) stage. So, if we don't find
the inode, keep the reconnect info around. When processing a replayed
request in CInode::encode_inodestat, set wanted/issued appropriately.
This is incomplete. We really need something to ensure we deal with
replayed requests before new requests are handled.. and on a cluster-wide
basis, since requests may involve slave requests to other mds's.
We also should clean up reconnects unclaimed after all replays are
complete. And somehow inform the client when the cap is officially
nonexistent and EBADF.
2009-03-06 16:27:18 -08:00
Sage Weil
b88f450ecc
mds: include path base ino in client reconnect paths
2009-03-06 16:27:17 -08:00
Sage Weil
b857a03bf1
filepath: no more leading / weirdness
...
Get rid of weird leading '/' iff ino==1 semantic. Clean up interface a bit.
2009-03-06 16:27:16 -08:00
Sage Weil
17d716ad9d
kclient: drop unused redirty code from set_page_dirty
2009-03-06 16:27:15 -08:00
Sage Weil
5f397ce668
filestore: lower min_sync_interval
...
We mainly need something non-zero here to ensure we batch some
requests up when we get a stream of requests with 'startsync'.
2009-03-06 16:27:14 -08:00
Yehuda Sadeh
1a33f6c74b
kclient: some bookkeeper fixes
2009-03-06 10:29:21 -08:00
Sage Weil
35bb1b98d3
kclient: fix bookkeeper ifdef
2009-03-06 10:16:12 -08:00
Sage Weil
8148cb6bbc
kclient: patch series descriptions, updates
2009-03-05 21:13:22 -08:00
Sage Weil
10397b6cb7
kclient: docs
2009-03-05 20:58:01 -08:00
Sage Weil
9c4e9eef59
kclient: rename ceph_tools -> bookkeeper
2009-03-05 20:57:53 -08:00
Sage Weil
f793c7f106
kclient: some caps.c cleanup
2009-03-05 18:01:15 -08:00
Sage Weil
bc1e7ea43b
mon: tolerate stray victory message (restart the election)
...
No need to core dump.
2009-03-05 16:29:02 -08:00
Sage Weil
858df3f378
kclient: BUG instead of WARN on wanted item on rdcaps list
2009-03-05 16:29:14 -08:00
Sage Weil
3e31202885
kclient: cosmetic
2009-03-05 16:29:14 -08:00
Sage Weil
40cb55afe5
kclient: wait for requests to be safe before tearing down dcache
...
We need to drop dentry refs in kill_sb _before_ kill_anon_super
tears down the dcache. Wait for safe replies for a while, then
drop what's left on the floor.
2009-03-05 16:29:14 -08:00
Sage Weil
3af0372134
kclient: don't fall over on !dirty page in invalidatepage
...
Spit out a warning, in case we end up with writeback accounting
errors. I _think_ it's ok, though.
2009-03-05 16:29:14 -08:00
Yehuda Sadeh
a4f01bb5bf
kclient: some documentation
2009-03-05 16:18:21 -08:00
Yehuda Sadeh
243d68431d
Merge branch 'unstable' of ssh://ceph.newdream.net/git/ceph into unstable
2009-03-05 16:14:32 -08:00
Yehuda Sadeh
c2011483f3
kclient: fix osd timeout handling
2009-03-05 16:14:12 -08:00
Sage Weil
ec1f5d5479
kclient: more sanity checks in set_page_dirty
...
These don't fix anything, just make things break sooner...
2009-03-05 15:02:06 -08:00
Sage Weil
ffce6999d2
kclient: only remember leases on regular dentries
...
...not for snapdir or snap dentries, which don't have matching
d_release methods.
2009-03-05 14:26:54 -08:00
Sage Weil
61e666521a
kclient: fix snap creation/deletion, memory leak
...
This broke with the new mds request dentry pointers.
Also, clean up kmalloced snap.
2009-03-05 14:15:55 -08:00
Sage Weil
ff52f0a10b
kclient: fix snap_realm leak
...
Separate lookup/create from get (this is safe because we always
hold span_rwsem during these operations).
2009-03-05 14:03:31 -08:00
Yehuda Sadeh
1b47fef3fb
Merge branch 'unstable' of ssh://ceph.newdream.net/git/ceph into unstable
2009-03-05 13:29:58 -08:00
Yehuda Sadeh
82ffc7a329
kclient: revise osd client timeout mechanism
2009-03-05 13:28:32 -08:00
Sage Weil
4c5369d809
kclient: simplify invalidatepage; strengthen assertions
2009-03-05 13:18:45 -08:00
Sage Weil
867aa9960a
kclient: writepages cleanup
2009-03-05 11:01:39 -08:00
Sage Weil
9cd5fa4fd7
synclient: optionally do a sync on object writes; wait for sync, not ack
2009-03-05 10:40:20 -08:00
Sage Weil
bdc527eba8
filestore: enforce min and max sync interval
...
Ideally the min sync interval should be self-tuning (ala ext3, xfs),
but that can come later.
2009-03-05 10:22:03 -08:00
Sage Weil
48a53adb50
kclient: dosync in writepages if sync_mode is WB_SYNC_ALL
...
If we dosync if !pdflush, then we end up doing syncs from
balance_dirty_pages, wreaking havoc on the OSDs.
It's possible the current_is_pdflush() check isn't needed at all..
2009-03-05 10:08:31 -08:00
Sage Weil
e9ea5fe435
kclient: fix readpages zeroing bug
...
Move to the next page after zeroing the first partial page.
2009-03-04 19:54:37 -08:00
Sage Weil
0a9ed8523c
kclient: return 0 from readpage on success
2009-03-04 16:55:11 -08:00
Sage Weil
e6726b255b
kclient: addr debug, cleanup
2009-03-04 16:55:11 -08:00
Sage Weil
69e4f194ee
kclient: fix readpages zeroing
2009-03-04 16:55:10 -08:00
Yehuda Sadeh
5b52d031b5
kclient: reschedule timeout for timeout length if remaining is zero
2009-03-04 16:39:18 -08:00
Yehuda Sadeh
24df3cf067
kclient: merge conflict
2009-03-04 16:32:23 -08:00
Yehuda Sadeh
e16e9f7764
Merge branch 'unstable' of ssh://ceph.newdream.net/git/ceph into unstable
...
Conflicts:
src/kernel/caps.c
src/kernel/super.h
2009-03-04 16:31:27 -08:00
Yehuda Sadeh
5251f2e661
kclient: sparse cleanup
2009-03-04 16:30:23 -08:00
