Whamcloud - gitweb
fs/lustre-release.git
15 years agoPutting the branch name at proper switch case to get libcfs changes
deshmukh [Tue, 1 Jul 2008 05:48:16 +0000 (05:48 +0000)]
Putting the branch name at proper switch case to get libcfs changes

15 years agob=13128
kalpak [Tue, 1 Jul 2008 04:25:18 +0000 (04:25 +0000)]
b=13128
i=adilger
i=johann

add -gid,-group,-uid-,-user options to lfs find

15 years agoBranch HEAD
bobijam [Tue, 1 Jul 2008 03:38:57 +0000 (03:38 +0000)]
Branch HEAD
b=12653
i=green, johann
Description: sanity test 65a fails if stripecount of -1 is set
Details    : handle -1 striping on filesystem in ll_dirstripe_verify

15 years agoBranch HEAD
fanyong [Tue, 1 Jul 2008 02:06:00 +0000 (02:06 +0000)]
Branch HEAD
b=15797
i=huanghua
i=vitaly

Misc fix for updating file {a/m/c}time as following policy:
1) Lustre file "{a/m/c}time" should base on client-side clock, including all
the file operations.
2) All the client-side clocks should be synchronized. It is no matter whether
the server-side clocks are synchronized with client-side ones or not.
3) File "ctime" is increased only.
4) File "{a,m}time" are increased only, except "utime()" operation can change
them backwards.

15 years agoBranch: HEAD
wangdi [Mon, 30 Jun 2008 18:52:13 +0000 (18:52 +0000)]
Branch: HEAD
Two minors fixes for stride read-ahead.
b=14483
i=adilger

15 years agobadly clear PG_Writeback bit in ll_ap_completion can produce false
shadow [Mon, 30 Jun 2008 12:29:46 +0000 (12:29 +0000)]
badly clear PG_Writeback bit in ll_ap_completion can produce false
positive assertion

Branch HEAD
b=14742
i=nikita
i=johan

15 years agofix issue with >8TB fs: group doesn't fit in 16bits anymore
alex [Mon, 30 Jun 2008 04:45:58 +0000 (04:45 +0000)]
fix issue with >8TB fs: group doesn't fit in 16bits anymore

b=16101
i=adilger
i=kalpak

15 years agoBranch HEAD
yangsheng [Sun, 29 Jun 2008 15:41:47 +0000 (15:41 +0000)]
Branch HEAD
b=16221
i=adilger

Remove 2.6.12-vanilla support from HEAD.

15 years agob=13103 (patch by adilger)
kalpak [Sun, 29 Jun 2008 10:08:39 +0000 (10:08 +0000)]
b=13103 (patch by adilger)
i=kalpak,girish

upstream e2fsprogs calls uninit_groups feature as uninit_bg. So mkfs.lustre should check for uninit_bg feature also

15 years agoIn ldlm_resource_add_lock(), call to ldlm_resource_dump()
deen [Sat, 28 Jun 2008 10:04:50 +0000 (10:04 +0000)]
In ldlm_resource_add_lock(), call to ldlm_resource_dump()
starve other threads from the resource lock for a long time in
case of long waiting queue, so change the debug level from
D_OTHER to the less frequently used D_INFO.

b=15953
i=adilger
i=nathan.rutman

15 years agoBranch: b1_8
adilger [Sat, 28 Jun 2008 06:30:54 +0000 (06:30 +0000)]
Branch: b1_8
Remove unused variable.

15 years agoBranch HEAD
adilger [Sat, 28 Jun 2008 04:54:49 +0000 (04:54 +0000)]
Branch HEAD
Ensure all module symbols from ldiskfs and OFED are available.
b=16112
i=brian
i=shadow

15 years agoBranch HEAD
adilger [Fri, 27 Jun 2008 22:04:32 +0000 (22:04 +0000)]
Branch HEAD
Make this code the same as b1_6 (mostly whitespace removal and comment updates)
for ease of comparison.  The only remaining difference is the optimization to
skip getting lq_rw_sem if we are using RR allocation.

15 years agob_HEAD_AT use the same tags of ldiskfs/libcfs as of HEAD.
ericm [Fri, 27 Jun 2008 21:53:52 +0000 (21:53 +0000)]
b_HEAD_AT use the same tags of ldiskfs/libcfs as of HEAD.

15 years agoBranch:HEAD
wangdi [Fri, 27 Jun 2008 17:32:44 +0000 (17:32 +0000)]
Branch:HEAD
missing stride start_offset in stride read-ahead.
b=16172
i=Oleg,Johann

15 years agoBranch HEAD
bwzhou [Fri, 27 Jun 2008 16:49:30 +0000 (16:49 +0000)]
Branch HEAD
b=14693
r=nathan

Don't return an error for conf param processing failures

15 years agoBranch HEAD
adilger [Fri, 27 Jun 2008 09:56:32 +0000 (09:56 +0000)]
Branch HEAD
Add comments to ldlm_enqueue_info fields.

15 years agoBranch HEAD
adilger [Fri, 27 Jun 2008 09:37:32 +0000 (09:37 +0000)]
Branch HEAD
Specify a MODULE_LICENSE for lloop module to avoid kernel tainting.
From b1_6.
b=5498
i=nathan
i=jay

15 years agob=5188
grev [Fri, 27 Jun 2008 09:34:48 +0000 (09:34 +0000)]
b=5188
add test_68 to ALWAYS_EXCEPT list

15 years agoBranch HEAD
adilger [Fri, 27 Jun 2008 09:31:19 +0000 (09:31 +0000)]
Branch HEAD
Quiet compiler warnings.

15 years agoBranch HEAD
adilger [Fri, 27 Jun 2008 09:23:48 +0000 (09:23 +0000)]
Branch HEAD
Allow both "clear" and "clear\n" to be written to ldlm.namespaces.*.lru_size.
Minor line wrapping and whitespace cleanup.
b=13868

15 years agoBranch HEAD
adilger [Fri, 27 Jun 2008 09:12:38 +0000 (09:12 +0000)]
Branch HEAD
Remove stale comment.

15 years agoBranch HEAD
huanghua [Fri, 27 Jun 2008 04:01:01 +0000 (04:01 +0000)]
Branch HEAD
b=16000
i=adilger
i=nikita.danilov

one more bit in sequence for IGIF and IDIF

15 years agoBranch HEAD
bwzhou [Fri, 27 Jun 2008 03:26:27 +0000 (03:26 +0000)]
Branch HEAD
b=15881
r=rread

rewrite sanity 129

15 years agoBranch HEAD
adilger [Thu, 26 Jun 2008 21:49:41 +0000 (21:49 +0000)]
Branch HEAD
Remove write permissions of files in the lustre-source RPM to avoid
any security risk.
b=16180
i=brian
i=terry

15 years agoBranch b1_6
adilger [Thu, 26 Jun 2008 21:09:09 +0000 (21:09 +0000)]
Branch b1_6
Fix ldiskfs configure.ac to match ChangeLog version.

15 years agoBranch b1_6
adilger [Thu, 26 Jun 2008 21:07:21 +0000 (21:07 +0000)]
Branch b1_6
Fix indenting to use tabs.

15 years agob=13934
yury [Thu, 26 Jun 2008 21:04:45 +0000 (21:04 +0000)]
b=13934
r=tappro,fanyong
- fixes unaligned access on 64 bit platforms. Thanks to Nikita for these patches.

15 years agoBranch HEAD
adilger [Thu, 26 Jun 2008 20:21:43 +0000 (20:21 +0000)]
Branch HEAD
Fix usage message to avoid referring to parameters as "files".
b=14471

15 years agoBranch HEAD
adilger [Thu, 26 Jun 2008 19:49:01 +0000 (19:49 +0000)]
Branch HEAD
Ignore generated files
b=16112

15 years agoBranch HEAD
adilger [Wed, 25 Jun 2008 22:50:07 +0000 (22:50 +0000)]
Branch HEAD
Ignore generated files
b=16112

15 years agob=15863
yury [Wed, 25 Jun 2008 22:14:03 +0000 (22:14 +0000)]
b=15863
r=shadow,wangdi
- fixes missed llcd_put() and wrong flag passed to llog_cleanup_commit_master() what again led to missing some llcd to be decrefed in cleanup. Fixes test_42 from replay-single.sh;
- some comments to understand llcd live cycle better next time.

15 years agoAdd missing .cvsignore files.
robert.read [Wed, 25 Jun 2008 21:04:08 +0000 (21:04 +0000)]
Add missing .cvsignore files.

15 years agoBranch b_hd_dmu
rcorreia [Wed, 25 Jun 2008 20:38:29 +0000 (20:38 +0000)]
Branch b_hd_dmu
Updated library names and paths due to zfs-lustre changes.
Updated lustrecvs to show helpful message if mercurial is missing.
Changed default zfs-lustre path from zfs-lustre/ to lustre/zfs-lustre/.
Added paths of zfs-lustre tools to test-framework.sh.

b=14778
i=alex.zhuravlev
i=manoj.joseph

15 years agob=16165
grev [Wed, 25 Jun 2008 19:56:57 +0000 (19:56 +0000)]
b=16165
i=Jian.Yu
localize TYPE

15 years agoBranch: HEAD
wangdi [Wed, 25 Jun 2008 19:34:50 +0000 (19:34 +0000)]
Branch: HEAD
Set fed_group in filter reconnect process.
b=16174
i=adilger,umka

15 years agoBranch HEAD
bwzhou [Wed, 25 Jun 2008 00:24:12 +0000 (00:24 +0000)]
Branch HEAD
b=13285

initialize save_errno

15 years agoBranch HEAD
adilger [Tue, 24 Jun 2008 17:52:01 +0000 (17:52 +0000)]
Branch HEAD
Fix typo in comment.

15 years agofix typo intoduced by grev 2008-02-12 rev. 1.48
grev [Tue, 24 Jun 2008 17:39:42 +0000 (17:39 +0000)]
fix typo intoduced by grev 2008-02-12 rev. 1.48

15 years agoBranch HEAD
bwzhou [Tue, 24 Jun 2008 16:15:36 +0000 (16:15 +0000)]
Branch HEAD
b=13285
r=adilger, bobijam (cid-1579)
  johann,  deen    (cid-1613)
  johann,  bobijam (cid-1685)

Coverity bugs

15 years agoBranch b1_6
bwzhou [Tue, 24 Jun 2008 15:26:25 +0000 (15:26 +0000)]
Branch b1_6
b=13285
r=adilger, deen

separate the limit for dir size into per-dev tunables.

15 years agoUpdate an obsolete wirecheck.c generator.
deen [Tue, 24 Jun 2008 13:47:58 +0000 (13:47 +0000)]
Update an obsolete wirecheck.c generator.

b=15566
i=adilger
i=shadow

15 years agodon't skip invalidate for already invalidated import (mgc reconnect).
shadow [Tue, 24 Jun 2008 09:49:57 +0000 (09:49 +0000)]
don't skip invalidate for already invalidated import (mgc reconnect).

Branch HEAD
b=15716
i=rread
i=tappro

15 years agoBranch HEAD
johann [Tue, 24 Jun 2008 08:12:44 +0000 (08:12 +0000)]
Branch HEAD
b=15677
i=zhenyu.xu

Tiny fix when the MDS is also the client node. Don't set the checksum
type for MDS's OSCs.

15 years agoUpdated the buildcvs file
deshmukh [Tue, 24 Jun 2008 07:00:55 +0000 (07:00 +0000)]
Updated the buildcvs file

15 years agoBranch HEAD
yangsheng [Tue, 24 Jun 2008 02:46:57 +0000 (02:46 +0000)]
Branch HEAD
b=15865
i=johann, zhenyu.xu

Update to RHEL5 latest kernel-2.6.18-53.1.21.el5.

15 years agoBranch HEAD
bobijam [Tue, 24 Jun 2008 01:56:04 +0000 (01:56 +0000)]
Branch HEAD
b=15924
i=green, johann

Severity   : major
Bugzilla   : 15924
Description: do not process already freed flock
Details    : flock can possibly be freed by another thread before it reaches
             to ldlm_flock_completion_ast.

15 years agob=15316
brian [Fri, 20 Jun 2008 16:35:49 +0000 (16:35 +0000)]
b=15316

Forgotten kernel config file in the previous landing.

15 years agoupdate buildcvs to use HEAD lnet and libcfs
nikita [Fri, 20 Jun 2008 10:55:48 +0000 (10:55 +0000)]
update buildcvs to use HEAD lnet and libcfs

15 years agoBranch HEAD
yangsheng [Fri, 20 Jun 2008 05:46:48 +0000 (05:46 +0000)]
Branch HEAD
b=15593

i=shadow, johann

Force to remove SUID when the do_truncate() be called in 2.6.22 kernel.

15 years agoBranch HEAD
robert.read [Fri, 20 Jun 2008 00:36:04 +0000 (00:36 +0000)]
Branch HEAD
b=15969

Fixup the a global autoconf macro that is shared by libcfs and lnet.
Fix Makefile.in to use LIBCFS_SUBDIR parameter.

15 years agoBranch HEAD
robert.read [Fri, 20 Jun 2008 00:12:55 +0000 (00:12 +0000)]
Branch HEAD
b=15969

Set flag for libcfs module.

15 years agoBranch HEAD
robert.read [Fri, 20 Jun 2008 00:08:17 +0000 (00:08 +0000)]
Branch HEAD
b=15969

Use @LIBCFS_SUBDIR@ param.

15 years agoBranch HEAD
robert.read [Fri, 20 Jun 2008 00:06:45 +0000 (00:06 +0000)]
Branch HEAD
b=15969

Fix build to parameterize location of libcfs.

15 years agob=16090
grev [Thu, 19 Jun 2008 23:58:20 +0000 (23:58 +0000)]
b=16090
i-Deen
move export ${facet}_svc=${label} from start fn to mount_facet fn

15 years agoBranch HEAD
robert.read [Thu, 19 Jun 2008 23:22:47 +0000 (23:22 +0000)]
Branch HEAD

b=15969

Instead of fighting with a special case of having libcfs only exist on HEAD, I
created an empty branch for b1_x in libcfs.

15 years agoadd .empty file
robert.read [Thu, 19 Jun 2008 23:19:06 +0000 (23:19 +0000)]
add .empty file

15 years agoadd empty file.
robert.read [Thu, 19 Jun 2008 23:12:46 +0000 (23:12 +0000)]
add empty file.

15 years agoBranch HEAD
robert.read [Thu, 19 Jun 2008 23:08:19 +0000 (23:08 +0000)]
Branch HEAD
b=15969

Don't require libcfs in autogen.sh - this breaks non-HEAD branches.

15 years agoBranch HEAD
robert.read [Thu, 19 Jun 2008 22:14:59 +0000 (22:14 +0000)]
Branch HEAD
b=15969
i=isaac
i=adilger

Move libcfs to its own module.

This removes libcfs from lnet and changes the build system
to use the new libcfs module.

15 years agoBranch HEAD
robert.read [Thu, 19 Jun 2008 22:04:35 +0000 (22:04 +0000)]
Branch HEAD
b=15969
i=isaac
i=adilger

Move libcfs to a module.

15 years agoBranch HEAD
robert.read [Thu, 19 Jun 2008 19:46:44 +0000 (19:46 +0000)]
Branch HEAD
b=16069

Make sure the wildcard cases are actually last.

15 years agoBranch HEAD
robert.read [Thu, 19 Jun 2008 18:29:42 +0000 (18:29 +0000)]
Branch HEAD
b=16069

Checkout libcfs for HEAD.
Also move the b_hd_* wildcard case to bottom, so it doesn't override specific setting for other branches.

15 years ago- b=15936. Unified last_rcvd format for MDS and OSS
tappro [Thu, 19 Jun 2008 10:48:11 +0000 (10:48 +0000)]
- b=15936. Unified last_rcvd format for MDS and OSS
  i:nathan, adilger

15 years agob=16073
jxiong [Thu, 19 Jun 2008 10:41:05 +0000 (10:41 +0000)]
b=16073
r=nikita

Fixup list_for_each_entry_safe_from since prior kernels don't have this macro.

15 years agoBranch HEAD
adilger [Wed, 18 Jun 2008 22:55:05 +0000 (22:55 +0000)]
Branch HEAD
Specify tag reference correctly.
b=16008
i=rread

15 years agoBranch v1_6_5
adilger [Wed, 18 Jun 2008 22:32:22 +0000 (22:32 +0000)]
Branch v1_6_5
Update the top-level "lustre" module to the specified version/release tag
so that it is not affected by later changes to the files (esp. Makefile.in).
b=16008
i=rread

15 years agob=15967
grev [Wed, 18 Jun 2008 21:01:03 +0000 (21:01 +0000)]
b=15967
i=Robert.Read
skip sanity-quota for HEAD

15 years agobranch: HEAD
ericm [Wed, 18 Jun 2008 19:07:38 +0000 (19:07 +0000)]
branch: HEAD
fix sanity check in mdt_req_handle.
b=16076
r=wangdi
r=umka

15 years agob=15870
brian [Wed, 18 Jun 2008 16:47:18 +0000 (16:47 +0000)]
b=15870
i=adilger

Forgot to remove the ChangeLog entry when I backed this patch out.

15 years agoBranch HEAD
bobijam [Wed, 18 Jun 2008 01:48:38 +0000 (01:48 +0000)]
Branch HEAD
b=14480
o=green
i=shadow, bobijam

Description: LBUG during stress test
Details    : Need properly lock accesses the flock deadlock detection list.

15 years agob=16046
jxiong [Tue, 17 Jun 2008 15:46:56 +0000 (15:46 +0000)]
b=16046
r=nikita,isaac

cl_lock_page_list optimization.

For lnet: just copied the macro list_for_each_entry_safe_from from linux kernel.

15 years agoBranch HEAD
shadow [Tue, 17 Jun 2008 06:50:36 +0000 (06:50 +0000)]
Branch HEAD
b=15837
i=nikita
i=green

15 years agoi=adilger
nathan [Mon, 16 Jun 2008 22:49:20 +0000 (22:49 +0000)]
i=adilger
i=rread
use HEAD versions of ldiskfs and lnet on generic b_hd_* derived branches

15 years agob=14384
grev [Mon, 16 Jun 2008 20:24:07 +0000 (20:24 +0000)]
b=14384
i=Adilger
remove tdir creation from run_one: the test has to create tdir by
itself if it needs this dir

15 years agoBranch HEAD
bwzhou [Sun, 15 Jun 2008 16:48:51 +0000 (16:48 +0000)]
Branch HEAD
b=13285
r=adilger, deen

not to fail the import when -EFBIG is returned

15 years agoi=15968:
isaac [Fri, 13 Jun 2008 16:10:49 +0000 (16:10 +0000)]
i=15968:
-   use lnet HEAD for HEAD, b_HEAD_AT, b_head_capa, and b_head_interop_disk.

15 years agob=15968:
isaac [Fri, 13 Jun 2008 15:13:11 +0000 (15:13 +0000)]
b=15968:
-   use lnet branch b1_x for lustre 1.x releases.

15 years agob=15226
yury [Fri, 13 Jun 2008 15:03:39 +0000 (15:03 +0000)]
b=15226
r=nikita,shadow
- committed missed part of 15226. Cleanups in comments, use doxygen style.

15 years ago- reverting part of 15716 as it causes constant assert and stops all testing. Shadow...
yury [Fri, 13 Jun 2008 10:46:30 +0000 (10:46 +0000)]
- reverting part of 15716 as it causes constant assert and stops all testing. Shadow is informed and will clean this up shortly.

15 years agob=15932
kalpak [Fri, 13 Jun 2008 10:09:37 +0000 (10:09 +0000)]
b=15932
i=adilger,alex

- if blocks per group is less than blocksize*8, set rest of the bitmap to 1.
- check return of ext2_find_next_le_bit for overflow.

15 years agob=15230
yury [Fri, 13 Jun 2008 08:53:45 +0000 (08:53 +0000)]
b=15230
r=nikita,shadow

- fixed handling for OBD_FAIL_$PREF_$OPC_NET fail_ids in mdt. Former code did not
check it correctly (due to typo with && instead of &) in mdt_req_handle() and
they all did not work. In same time, some handlers like mdt_close() and
mdt_enqueue() tried to check them again (result of some wrong fix) but again, did
it not correctly. They returned 0 error without doing anything. This should
have to emulate network failure. But as they did not allocate reply buffer and
returned 0 error, they caused rs != NULL assert in ptlrpc. Fxing this also fixed
replay-single.sh test_53* and replay_dual.sh test_12 and possibly others;

- removed checking for NET fail_id in mdt_close() and mdt_enqueue() - sources
of recent assert;

- added sanity check in mdt_req_handle() for any other invalid situation about
returning 0 error and not allocating reply buffers;

- removed mdt_reply(), move its one line call into mdt_req_handle(). This was
needed to simplify handling NET fail_ids in which case we should just return 0
and make sure that no reply is sent;

- comments and cleanups;

- in reply-dual.sh - remove test 8 from ALWAYS_EXCEPT. It passes in HEAD.
Originally for placed into ALWAYS_EXCEPT for old mds code and later moved to
HEAD test scripts but as mds in HEAD is completely new this bug is making any
sense there;

- in reply-single.sh - remove tests 0b 39 56 from ALWAYS_EXCEPT. They are
passing in HEAD. Also they are obsolete and related to closed bugs.

15 years agoBranch HEAD
bobijam [Fri, 13 Jun 2008 01:29:20 +0000 (01:29 +0000)]
Branch HEAD
b=14134
i=johann

minor fix handling '/etc/mtab' issue.

15 years agob=15870
brian [Thu, 12 Jun 2008 20:10:18 +0000 (20:10 +0000)]
b=15870
i=adilger

Backing this out as it tries to solve a problem that doesn't really exist
and causes build breakage in doing so.

I have advised the customer that opened this bug as to how they can fix
thier OFED source pool to match the interface provided by the
kernel-ib-devel RPM, which is our reference implementation.

15 years agofix for
grev [Thu, 12 Jun 2008 08:34:41 +0000 (08:34 +0000)]
fix for
b=15711
i=Tappro
set mds${num}failover_HOST properly

fix for
b=15255
i=Adilger
umount $MOUNT2 only if it is set

15 years agoBranch HEAD
bobijam [Thu, 12 Jun 2008 02:51:11 +0000 (02:51 +0000)]
Branch HEAD
b=15345
o=johann
i=nathan, bobijam

check --writeconf flag when forcing the index.

15 years agoCleanup wrong mount_facet message
grev [Wed, 11 Jun 2008 20:11:09 +0000 (20:11 +0000)]
Cleanup wrong mount_facet message
(introduced by deen rev 1.102)

15 years agob=14384
grev [Wed, 11 Jun 2008 17:43:30 +0000 (17:43 +0000)]
b=14384
i=Adilger
test_9, test_10a fixes: do not break DIR value

15 years ago- remove trailing spaces, expand tabs.
isaac [Wed, 11 Jun 2008 16:48:54 +0000 (16:48 +0000)]
-   remove trailing spaces, expand tabs.

15 years agoBranch HEAD
yangsheng [Wed, 11 Jun 2008 09:03:10 +0000 (09:03 +0000)]
Branch HEAD
b=15667
i=andrew, johann

Remove the useless function echo_client_ubrw().

15 years agotimeout and assert in invalidate import.
shadow [Wed, 11 Jun 2008 06:48:53 +0000 (06:48 +0000)]
timeout and assert in invalidate import.
b=15716
i=rread
i=johann

15 years agoBranch b1_6
adilger [Tue, 10 Jun 2008 21:43:18 +0000 (21:43 +0000)]
Branch b1_6
Ignore both Module.symvers and Modules.symvers, for SLES.

15 years agob=15316
brian [Tue, 10 Jun 2008 20:10:42 +0000 (20:10 +0000)]
b=15316
i=yibin.wang
i=cliff.white:

Add the kernel-ib* RPMs to the reuse strategy.

15 years agoBranch b1_6
adilger [Tue, 10 Jun 2008 17:51:07 +0000 (17:51 +0000)]
Branch b1_6
Remove obsolete tips from cvsdiffclient header.  Move the "editable" comments
together, after the instructions, to be more consistent.

15 years agob=14471
komaln [Tue, 10 Jun 2008 14:27:20 +0000 (14:27 +0000)]
b=14471
i=adilger
 replace /proc/fs/lustre access with lctl {get,set}_param in test-framework.sh

15 years agoBranch HEAD
yangsheng [Tue, 10 Jun 2008 08:04:17 +0000 (08:04 +0000)]
Branch HEAD
b=15742
i=johann, zhenyu.xu

Update to RHEL4 kernel-2.6.9-67.0.15.

15 years agoBranch HEAD
bwzhou [Tue, 10 Jun 2008 07:42:25 +0000 (07:42 +0000)]
Branch HEAD
b=15759
r=rread, green

Description: MDS or OSS service threads fail startup with -24 (-EMFILE)
Details    : During startup under recovery, it is possible for service
             thread startup to fail in ptlrpc_start_threads() if one
             of the threads begins processing a request and then starts
             an additional thread.  This causes ptlrpc_start_threads()
             to try and start 1 too many threads and get an error.

15 years agoBranch HEAD
bwzhou [Tue, 10 Jun 2008 06:59:39 +0000 (06:59 +0000)]
Branch HEAD
b=15776
r=shadow, vitaly

Prevent locks stalls on glimpse races

15 years agoBranch HEAD
bwzhou [Tue, 10 Jun 2008 06:26:59 +0000 (06:26 +0000)]
Branch HEAD
b=13380

remove test 2 and 4 form ALWAYS_EXCEPT since 13380(13129) is fixed

15 years agoBranch HEAD
bwzhou [Tue, 10 Jun 2008 05:50:35 +0000 (05:50 +0000)]
Branch HEAD
b=15436
r=nathan, bobijam

remove failover mgs setting from mkfs_lustre