Whamcloud - gitweb
yangsheng [Mon, 23 Feb 2009 10:33:58 +0000 (10:33 +0000)]
Branch b1_6
b=18289
i=yangsheng, wangyb
Enable the rhel5 in-kernel OFED.
grev [Fri, 20 Feb 2009 20:10:39 +0000 (20:10 +0000)]
b=18443
i=Adilger
remove obsolete tests scripts
yangsheng [Fri, 20 Feb 2009 15:47:09 +0000 (15:47 +0000)]
Branch HEAD
b=18289
Back the changes and get another inspection.
yangsheng [Fri, 20 Feb 2009 14:33:27 +0000 (14:33 +0000)]
Branch HEAD
b=18289
i=shadow, johann, yibin.wang
Update kernel for RHEL5.3 2.6.18-128.1.1. Enable the in-kernel OFED because
out tree OFED package hasn't provide support for RHEL5.3.
yangsheng [Fri, 20 Feb 2009 10:14:54 +0000 (10:14 +0000)]
Branch b1_6
b=18289
i=shadow, johann, yibin.yang
Kernel update to rhel5.3(2.6.18.92-128.1.1). Enable the in-kernel OFED
since haven't a separate OFED release support this kernel.
jxiong [Fri, 20 Feb 2009 08:14:51 +0000 (08:14 +0000)]
b=17084
- merge md-soft-lockups.patch to raid5-merge-ios.patch
- rename md-mmp-unplug-dev-*.patch to raid5-mmp-unplug-dev-*.patch
grev [Thu, 19 Feb 2009 22:14:26 +0000 (22:14 +0000)]
b=18521
i=Alex.Zhuravlev
sanity test_151, t-f localize var fix
brian [Thu, 19 Feb 2009 20:31:06 +0000 (20:31 +0000)]
b=17952
i=wangyb
i=adilger
Given that configure won't work with a relative pathname passed to the
--with-linux option, error out if the user provides a relative path.
wangdi [Thu, 19 Feb 2009 20:12:52 +0000 (20:12 +0000)]
Branch: HEAD
Add bug reference for the race between partial page truncate and write.
isaac [Thu, 19 Feb 2009 20:03:54 +0000 (20:03 +0000)]
- removed used function lnet_nid_alive_locked.
nikita [Thu, 19 Feb 2009 13:03:49 +0000 (13:03 +0000)]
ll_put_super(): remove noisy and no longer useful debugging statement.
nikita [Thu, 19 Feb 2009 12:37:51 +0000 (12:37 +0000)]
osc_io_submit(): add a comment clarifying osc page queueing algorithm.
adilger [Thu, 19 Feb 2009 01:16:42 +0000 (01:16 +0000)]
Branch HEAD
Cleanup autoconf output.
kalpak [Wed, 18 Feb 2009 17:29:57 +0000 (17:29 +0000)]
b=18532
o=huanghua
i=nathan
i=kalpak
fix possible memory leak in mdd_dot_lustre_setup() and create_dot_lustre_dir()
tianzy [Wed, 18 Feb 2009 07:52:37 +0000 (07:52 +0000)]
Branch HEAD
Releasing quota between lquota_chkquota() and filter_direct_io()
could make false -EDQUOT
b=18491
i=johann
i=andrew.perepechko
fanyong [Wed, 18 Feb 2009 02:11:33 +0000 (02:11 +0000)]
Branch HEAD
b=18293
i=tianzy
i=jay
Diagnostic patch to dump the page cache when IO error.
nathan [Tue, 17 Feb 2009 23:05:23 +0000 (23:05 +0000)]
b=18266
allow pool_cmd to work on MGS with no MDT or client
b=18516
fix parsing to correctly handle hex OST numbers
i=jc.lafoucriere
i=johann
grev [Tue, 17 Feb 2009 20:26:16 +0000 (20:26 +0000)]
adjust test_32b to new disk1_8.tar.bz2 name
grev [Tue, 17 Feb 2009 20:21:01 +0000 (20:21 +0000)]
b=13739
i=Rahul
add disk1_8 to lustre-tests rpm
rename to tar.bz2
remove disk1_4 image
nikita [Tue, 17 Feb 2009 18:55:53 +0000 (18:55 +0000)]
OBD_SLAB_ALLOC_PTR_SAFE() is a new OBD_ macro for slab allocation that uses CFS_ALLOC_IO allocation
mask (i.e., without CFS_ALLOC_FS), thus preventing file system re-entrance.
b=18499
i=alexander.zarochentsev@sun.com
i=adilger@sun.com
nikita [Tue, 17 Feb 2009 18:50:38 +0000 (18:50 +0000)]
silence can_merge_pages() about OBD_BRW_SYNC
OBD_BRW_SYNC means that a page is a part of a synchronous transfer and hence does not consume
grant. It is fine to mix pages with and without OBD_BRW_SYNC in the same brw vector. Comment
OBD_BRW_SYNC and update can_merge_pages().
(Also add sanity checks to osc_page_transfer_add(), while we are here.)
b=18393
i=alexey.lyashkov@sun.com
i=adilger@sun.com
nikita [Tue, 17 Feb 2009 13:35:15 +0000 (13:35 +0000)]
rollback a bit of 18499 patch committed by mistake.
isaac [Tue, 17 Feb 2009 05:21:21 +0000 (05:21 +0000)]
b=16186,i=liangzhen,i=maxim:
- shared routing enhancements: peer health detection.
nikita [Mon, 16 Feb 2009 19:06:51 +0000 (19:06 +0000)]
osc_lock_granted(): move osc_lock_invariant() call out of lock_res_and_lock() critical section, as osc_lock_invariant() takes this lock too. This only has effect when confugured with --enable-invariants switch.
girish [Mon, 16 Feb 2009 07:47:34 +0000 (07:47 +0000)]
b=18470
i=johann
i=adilger
Correct error handling
jxiong [Sun, 15 Feb 2009 02:38:40 +0000 (02:38 +0000)]
b=15428
r=adilger
Fixed a crash issue for raid5-zerocopy
grev [Fri, 13 Feb 2009 19:56:44 +0000 (19:56 +0000)]
b=18333
i=Oleg.Drokin
add flock to default mount opts
yangsheng [Fri, 13 Feb 2009 18:32:39 +0000 (18:32 +0000)]
Branch HEAD
b=18462
i=adilger, johann
Eliminate kernel patches for linux/fs.h.
yangsheng [Fri, 13 Feb 2009 16:14:59 +0000 (16:14 +0000)]
Branch HEAD
b=16759
i=shadow, zhen.liang
Use RCU lock to enable stack dump when tasklist_lock unexport.
grev [Fri, 13 Feb 2009 13:50:00 +0000 (13:50 +0000)]
fix typo for
b=16488
i=Oleg.Drokin
RACER acc-sm test suit
shadow [Fri, 13 Feb 2009 10:22:37 +0000 (10:22 +0000)]
fix typo.
Branch HEAD
stinson1947 [Fri, 13 Feb 2009 02:21:57 +0000 (02:21 +0000)]
Updated for the next 2.0 bi-weekly build, build 12.
grev [Thu, 12 Feb 2009 18:45:48 +0000 (18:45 +0000)]
b=17839
i=Brian
recovery-mds-scale for SLOW=no
grev [Thu, 12 Feb 2009 16:17:52 +0000 (16:17 +0000)]
revert b=17839 commit
grev [Thu, 12 Feb 2009 15:34:38 +0000 (15:34 +0000)]
b=17839
i=Brian
recovery-mds-scale for SLOW=no
bobijam [Thu, 12 Feb 2009 06:39:19 +0000 (06:39 +0000)]
Branch HEAD
b=18015
sles9 doesn't define SLAB_DESTROY_BY_RCU
jxiong [Thu, 12 Feb 2009 04:24:10 +0000 (04:24 +0000)]
b=15428
r=adilger
Fix a crash issue for raid5 zerocopy patch.
wangdi [Thu, 12 Feb 2009 02:22:55 +0000 (02:22 +0000)]
Branch: HEAD
fix buildcvs for b_hd*
huanghua [Thu, 12 Feb 2009 01:38:56 +0000 (01:38 +0000)]
Branch HEAD
b=11404
i=yong.fan
i=rahul.deshmukh
more debugging code.
kalpak [Wed, 11 Feb 2009 18:52:26 +0000 (18:52 +0000)]
b=18452
i=nathan
i=pravin
allow getstripe to work for .lustre dir
llapi_traverse should not open .lustre dir
girish [Wed, 11 Feb 2009 18:07:12 +0000 (18:07 +0000)]
b=18417
o=h.huang
i=girish
Fix typo from last commit which caused test-framework to exit wrongly
jxiong [Wed, 11 Feb 2009 15:44:09 +0000 (15:44 +0000)]
b=17914
r=nathan,h.huang
recommit the patch after fixing the problem
jxiong [Wed, 11 Feb 2009 09:27:53 +0000 (09:27 +0000)]
revert the patch for bug 17914, since it breaks sanity:65i.
girish [Wed, 11 Feb 2009 06:52:20 +0000 (06:52 +0000)]
b=18417
i=kalpak
i=girish
o=adilger
correct cleanup for sanity 102 tests
liuy [Wed, 11 Feb 2009 02:58:30 +0000 (02:58 +0000)]
Branch HEAD
b=15384
update lnet tag for b_hd_params_tree
huanghua [Wed, 11 Feb 2009 02:33:23 +0000 (02:33 +0000)]
Branch HEAD
b=17158
i=grev
i=rahul.deshmukh
exhaust_precreations fn for CMD - using get_mds_dir
huanghua [Wed, 11 Feb 2009 02:29:22 +0000 (02:29 +0000)]
Branch HEAD
b=18147
i=grev
i=yong.fan
enable this test: get new statfs data and CMD enabled.
maxim [Tue, 10 Feb 2009 18:33:35 +0000 (18:33 +0000)]
b=17087
i=matt
i=liang
Minor fix for winnt build problems introduced by landing patch making ksocklnd more os-neutral
yangsheng [Tue, 10 Feb 2009 17:21:12 +0000 (17:21 +0000)]
Branch HEAD
b=10762
Fixed for new kernel.
maxim [Tue, 10 Feb 2009 17:16:02 +0000 (17:16 +0000)]
b=14132
Few words about acceptor cleanup patch that was landed recently are added.
yangsheng [Tue, 10 Feb 2009 15:58:55 +0000 (15:58 +0000)]
Branch HEAD
b=10762
i=adilger, green
Rate limit watchdog
Author: Jim Garlick(LLNL)
fanyong [Tue, 10 Feb 2009 05:08:45 +0000 (05:08 +0000)]
Branch HEAD
b=18295
i=huanghua
i=pravin
Osd should not update object's xtime when calls ldiskfs stuff.
deshmukh [Tue, 10 Feb 2009 05:04:40 +0000 (05:04 +0000)]
b=13739
i=adilger
i=grev
conf_sanity_32b test-case is fixed and now enabled. This test-case will test very basic
interop operation between 1.8 and 2.0
fanyong [Tue, 10 Feb 2009 04:57:01 +0000 (04:57 +0000)]
Branch HEAD
b=18294
i=tianzy
i=panda
Diagnostic patch to catch what's the garbage left in the system when quota error.
tianzy [Tue, 10 Feb 2009 04:30:20 +0000 (04:30 +0000)]
Branch HEAD
add an entry under /proc to support what policy(RR or QOS) will be used.
b=18334
i=nathan
i=adilger
jxiong [Tue, 10 Feb 2009 04:01:34 +0000 (04:01 +0000)]
b=17914
r=nathan,h.huang
Parse <fsname>-MDTXXXX-mdc as mdt, in order to support CMD.
jxiong [Tue, 10 Feb 2009 02:05:08 +0000 (02:05 +0000)]
b=17025
r=adilger,nikita
If a client creates and opens some files, and then mds gets crashed, there exists a chance that the objids in OSTs will be used for multiple files.
The root cause of this issue is that we don't have the appropriate mechanism to handle the case if the replaying of open request fails.
This is a hot fix for this problem. By skipping the orphan objects, the client won't see that multiple files using the same objid any more.
huanghua [Tue, 10 Feb 2009 01:35:24 +0000 (01:35 +0000)]
Branch HEAD
b=17175
i=grev
use correct matching pattern to match the string.
grev [Tue, 10 Feb 2009 00:17:15 +0000 (00:17 +0000)]
b=18219
i=Adilger
test_70b one client fix
grev [Mon, 9 Feb 2009 23:47:31 +0000 (23:47 +0000)]
b=18366
i=Johann
test_5 ENOSPC fix
johann [Mon, 9 Feb 2009 15:31:06 +0000 (15:31 +0000)]
Branch HEAD
b=18067
i=johann
i=adilger
Change oap magic LASSERT into LASSERTF to aid debugging.
pravins [Sun, 8 Feb 2009 17:02:20 +0000 (17:02 +0000)]
b=15134
i=nikita.danilov
i=h.huang
use unsigned long index for radix tree lookup
pravins [Sun, 8 Feb 2009 16:56:20 +0000 (16:56 +0000)]
b=15134
i=adilger
i=yury.umanets
add file-type attribute to lu_dirent.
jxiong [Sat, 7 Feb 2009 06:16:36 +0000 (06:16 +0000)]
b=17508
r=nikita,h.huang
cmm_mdc device is not stacked into lu devices tree right now, which causes it
is not initialized correctly.
rcorreia [Fri, 6 Feb 2009 22:22:06 +0000 (22:22 +0000)]
Branch b_hd_kdmu
b=17584
Trivial autoconf fixes for 0.4.x versions of the SPL and ZFS repos.
shadow [Fri, 6 Feb 2009 12:31:22 +0000 (12:31 +0000)]
resolve lots races and use after free.
Branch b1_8
b=HEAD
i=jc.lafoucriere
i=panda
bobijam [Fri, 6 Feb 2009 02:36:01 +0000 (02:36 +0000)]
Branch HEAD
b=17011
i=tom.wang
i=joahnn
remove redundant brw_stats on obdfilter.
shadow [Thu, 5 Feb 2009 20:05:26 +0000 (20:05 +0000)]
fix typo
Branch HEAD
isaac [Thu, 5 Feb 2009 19:25:13 +0000 (19:25 +0000)]
- corrected a typo in ChangeLog.
stinson1947 [Thu, 5 Feb 2009 17:42:32 +0000 (17:42 +0000)]
sync previous release entries (1.6.6, 1.6.7, 1.8.0, & 1.8.1) into the file.
shadow [Thu, 5 Feb 2009 11:05:00 +0000 (11:05 +0000)]
handle MDS connect in failover as real connect.
Branch HEAD
b=18304
i=johann
i=green
shadow [Thu, 5 Feb 2009 10:35:55 +0000 (10:35 +0000)]
send activate event only when connect finished and import state is FULL.
Branch HEAD
b=16839
i=nikita
i=umka
nikita [Wed, 4 Feb 2009 16:33:20 +0000 (16:33 +0000)]
Enqueue glimpse locks with CEF_MUST flag so that they are not accidentaly
converted into a lockless mode.
b=18104
i=oleg.drokin@sun.com
i=jinshan.xiong@sun.com
bobijam [Wed, 4 Feb 2009 01:45:47 +0000 (01:45 +0000)]
Branch HEAD
b=18015
i=johann
i=sheng.yang
add SLAB_DESTROY_BY_RCU to ldlm_lock_slab's flags for ldlm_locks are freed
with RCU mechanism.
kalpak [Tue, 3 Feb 2009 19:17:29 +0000 (19:17 +0000)]
b=17187
i=nikita
i=nathan
open file using fid
girish [Tue, 3 Feb 2009 19:10:11 +0000 (19:10 +0000)]
b=18417
i=adilger
i=kalpak
remove star tests from sanity.sh
zhanghc [Tue, 3 Feb 2009 14:16:52 +0000 (14:16 +0000)]
b=17887
1, print the actual time to wait for lock
cancellation in "waiting_locks_callback"
2, compare resource in "ost_rw_hpreq_lock_match"
3, replace "LDLM_ERROR" with "LDLM_WARN" if a lock
is completed after waiting more than odb_timeout
i=vitaly.fertman
i=johann
grev [Tue, 3 Feb 2009 13:44:24 +0000 (13:44 +0000)]
b=17899
i=Johann
test_53* fix: rm dirs used at start
maxim [Tue, 3 Feb 2009 13:43:21 +0000 (13:43 +0000)]
b=14132
i=isaac
acceptor.c cleanup:
- removed code duplication in acceptor.c for the cases of kernel
and user-space code
- uniformed user-space libcfs tcpip primitives to have prototypes
similar to kernel ones
- removed parse_int_tunable() from libcfs to lnet because it's not
generic enough to reside in libcfs
- untabified and removed trailing spaces in files touched
- minor cosmetic changes in usocklnd becasue now a socket is
represented with cfs_socket_t structure and should be released
properly (just close(fd) is not ehough).
tianzy [Tue, 3 Feb 2009 06:45:38 +0000 (06:45 +0000)]
Branch HEAD
use QUOTA_REQ_RETURNED to replace -EAGAIN when a quota req is finished.
b=18307
i=johann
i=panda
fanyong [Tue, 3 Feb 2009 02:10:04 +0000 (02:10 +0000)]
Branch HEAD
b=18135
i=tianzy
i=andrew.perepechko
Wake up all wait threads on the "lq_waitq" when "lqc_import" is invalid.
nikita [Mon, 2 Feb 2009 19:49:38 +0000 (19:49 +0000)]
Implement ll_glimpse_ioctl() in HEAD client.
b=18040
i=adilger@sun.com
i=robert.read@sun.com
shadow [Mon, 2 Feb 2009 17:30:44 +0000 (17:30 +0000)]
don't call obd_disconnect under lov_lock.
Branch HEAD
b=17310
i=johann
i=rread
grev [Mon, 2 Feb 2009 14:18:08 +0000 (14:18 +0000)]
b=16818
i=Nathan
get rid of set_obd_timeout(), improve test_26*
kalpak [Mon, 2 Feb 2009 08:50:37 +0000 (08:50 +0000)]
b=17895
i=adilger
i=alex
While using HA for Lustre servers with Linux RAID, it is possible that MMP will not detect multiple mounts. To make this work we need to unplug the device queue in RAID when the MMP block is being written. Also while reading the MMP block, we should read it from disk and not the cached one.
kalpak [Mon, 2 Feb 2009 07:17:42 +0000 (07:17 +0000)]
b=17895
i=adilger
i=alex
While using HA for Lustre servers with Linux RAID, it is possible that MMP will not detect multiple mounts. To make this work we need to unplug the device queue in RAID when the MMP block is being written. Also while reading the MMP block, we should read it from disk and not the cached one.
isaac [Sun, 1 Feb 2009 20:53:27 +0000 (20:53 +0000)]
b=11245,i=liangzhen:
- IB path MTU mistakenly set to 1st path MTU when ib_mtu is off.
vs [Sat, 31 Jan 2009 14:36:51 +0000 (14:36 +0000)]
b=17807
i=nikita
i=huanghua
ll_revalidate_nd: do not return 0 in case of error
adilger [Fri, 30 Jan 2009 21:05:43 +0000 (21:05 +0000)]
Branch HEAD
Change incorrect error message about file join.
ericm [Fri, 30 Jan 2009 17:35:56 +0000 (17:35 +0000)]
branch: HEAD
rewrite the bulk i/o security, use same gssapi as on rpc data instead of
invent our own mechanism.
b=16830
r=fanyong
r=vitaly
stinson1947 [Fri, 30 Jan 2009 02:34:11 +0000 (02:34 +0000)]
updated version for the nex 2.0 bi-weekly build
nikita [Thu, 29 Jan 2009 20:01:59 +0000 (20:01 +0000)]
fix obsolete module name.
isaac [Thu, 29 Jan 2009 15:23:34 +0000 (15:23 +0000)]
b=15984, i=maxim:
- uptllnd credit overflow fix.
huanghua [Thu, 29 Jan 2009 09:30:45 +0000 (09:30 +0000)]
Branch HEAD
b=18318
i=rahul
i=pravin
shrink close reply in resend case.
pravins [Thu, 29 Jan 2009 09:03:37 +0000 (09:03 +0000)]
b=18165
i=rahul
i=huanghua
fix mem leak by releasing readdir private data calling f_op release().
shadow [Wed, 28 Jan 2009 09:40:51 +0000 (09:40 +0000)]
fix typo
shadow [Wed, 28 Jan 2009 09:20:21 +0000 (09:20 +0000)]
don't lose wakeup fro imp_recovery_waitq
Branch HEAD
b=18154
i=nathan
i=rread
nikita [Tue, 27 Jan 2009 21:40:01 +0000 (21:40 +0000)]
fix a typo and an obsolete name in documentation.
yury [Tue, 27 Jan 2009 14:22:47 +0000 (14:22 +0000)]
- added comment about LDLM_FL_* flags at Adilger's request.
yury [Tue, 27 Jan 2009 14:07:07 +0000 (14:07 +0000)]
b=17282
r=johann,shadow
- simplifies a lot lru resize memory presssure handling. No estimation pass is needed right now as it was too CPU consuming.