Whamcloud - gitweb
fs/lustre-release.git
16 years agob=18696
grev [Thu, 26 Mar 2009 15:26:35 +0000 (15:26 +0000)]
b=18696
i=Tappro
add recovery-random-scale.sh to rpm

16 years agob=17195
grev [Wed, 25 Mar 2009 19:39:44 +0000 (19:39 +0000)]
b=17195
i=Tappro
test_3a: recovery duration dor increasing number of clients

16 years agoBranch b_release_1_8_0
fanyong [Wed, 25 Mar 2009 03:58:52 +0000 (03:58 +0000)]
Branch b_release_1_8_0
b=18741
i=tianzy
i=huanghua

1) abort all dqacq/dqrel requests when lqc_import is invalid.
2) scripts fix.

16 years agoBranch b_release_1_8_0
fanyong [Wed, 25 Mar 2009 02:35:18 +0000 (02:35 +0000)]
Branch b_release_1_8_0
b=15691
i=huanghua
i=yep

Ignore minor performance difference and some debug adjustment.

16 years agob=18874
grev [Tue, 24 Mar 2009 17:17:01 +0000 (17:17 +0000)]
b=18874
i=Alexey.Lyashkov
test_48: cleanup properly

16 years agoBranch b_release_1_8_0
johann [Sat, 21 Mar 2009 08:47:03 +0000 (08:47 +0000)]
Branch b_release_1_8_0
b=18800
i=johann (patch from HongChao)
i=wangdi

prevent use of uninitialized llog context during unlink.

16 years agodon't leak lov_lock if adding to global ost pool is failed.
shadow [Fri, 20 Mar 2009 22:09:12 +0000 (22:09 +0000)]
don't leak lov_lock if adding to global ost pool is failed.

Branch b_release_1_8_0
b=18868
i=johann
i=tappro

16 years agob=18691
anserper [Fri, 20 Mar 2009 19:05:25 +0000 (19:05 +0000)]
b=18691
i=Vitaly Fertman
i=Johann Lombardi

fix for osc_max_rpc_in_flight

16 years agoBranch b_release_1_8_0
zam [Fri, 20 Mar 2009 16:15:54 +0000 (16:15 +0000)]
Branch b_release_1_8_0
b=18859
i=oleg.drokin
i=alexey.lyashkov

Don't do server-side locking if LL_FILE_IGNORE_LOCK is set. Otherwise we  have
a deadlock between server-side extent lock and a group lock owned by the same
client.

16 years agodon't race umount and calling llap_shrinker
shadow [Fri, 20 Mar 2009 07:31:46 +0000 (07:31 +0000)]
don't race umount and calling llap_shrinker

Branch b_release_1_8_0
b=18773
i=jay
i=tappro

16 years agob=17195
grev [Tue, 17 Mar 2009 20:37:12 +0000 (20:37 +0000)]
b=17195
i=Tappro
skip the tests if no delayed recovery

16 years agob=18623
grev [Tue, 17 Mar 2009 11:19:01 +0000 (11:19 +0000)]
b=18623
i=Andrew.Perepechko
lfs df several mount points fix

16 years agob=18589
grev [Mon, 16 Mar 2009 17:30:25 +0000 (17:30 +0000)]
b=18589
i=Adilger
do df before /proc access

16 years agob=18045
grev [Fri, 13 Mar 2009 19:05:37 +0000 (19:05 +0000)]
b=18045
i=Adilger
new mdsrate --ignore option for mdsrate tests cleanup

16 years agob=17195
grev [Fri, 13 Mar 2009 15:50:25 +0000 (15:50 +0000)]
b=17195
i=Tappro
skip the tests if no delayed recovery

16 years agob=18623
grev [Thu, 12 Mar 2009 19:12:51 +0000 (19:12 +0000)]
b=18623
i=Andrew.Perepechko
ENABLE_QUOTA acc-sm patch

16 years agob=18761
grev [Thu, 12 Mar 2009 16:56:29 +0000 (16:56 +0000)]
b=18761
i=Oleg.Drokin
declare array fix

16 years agob=15524
girish [Thu, 12 Mar 2009 11:43:03 +0000 (11:43 +0000)]
b=15524
i=adilger
i=kalpak

sanity test 102 changes for lustre-tar package

16 years agob=18696
grev [Wed, 11 Mar 2009 16:06:28 +0000 (16:06 +0000)]
b=18696
i=Tappro
new RECOVERY_RANDOM_SCALE test;
zconf_umount_clients fn remote command fix;

16 years agofile recovery-random-scale.sh was added on branch b_release_1_8_0 on 2009-03-11 16...
grev [Wed, 11 Mar 2009 16:04:14 +0000 (16:04 +0000)]
file recovery-random-scale.sh was added on branch b_release_1_8_0 on 2009-03-11 16:06:28 +0000

16 years agoBranch b_release_1_8_0
yangsheng [Wed, 11 Mar 2009 10:45:08 +0000 (10:45 +0000)]
Branch b_release_1_8_0
b=17701

Extra patch for swab lov.

16 years ago- set correct thread in rq_srv_thread for requests processing in
tappro [Wed, 11 Mar 2009 09:04:38 +0000 (09:04 +0000)]
- set correct thread in rq_srv_thread for requests processing in
  process_recovery_queue()
b:18221
i:adilger,rread,johann

16 years agoBranch b_release_1_8_0
tianzy [Wed, 11 Mar 2009 08:28:16 +0000 (08:28 +0000)]
Branch b_release_1_8_0
add more At debug info and fix a problem in test_68 in replay-single.sh
b=18426
i=nathan.rutman

16 years agoBranch b_release_1_8_0
yangsheng [Wed, 11 Mar 2009 07:04:30 +0000 (07:04 +0000)]
Branch b_release_1_8_0
b=17701

i=adilger, nathan

Fixed the lov_getstripe works well in Big-Endian.

16 years agob=17846
girish [Sun, 8 Mar 2009 08:14:16 +0000 (08:14 +0000)]
b=17846
i=adilger
i=h.huang

changes to regression test after landing 17846 Lustre patch. make sure all files are not extracted on OST id 0

16 years ago- use invalidate_mapping_pages for old kernels in case of disabled ROC for simplicity
tappro [Fri, 6 Mar 2009 21:27:58 +0000 (21:27 +0000)]
- use invalidate_mapping_pages for old kernels in case of disabled ROC for simplicity
b:18718
i:adilger,johann

16 years agoBranch b_release_1_8_0
johann [Fri, 6 Mar 2009 19:47:47 +0000 (19:47 +0000)]
Branch b_release_1_8_0
b=18558

fix warning.

16 years agoBranch b_release_1_8_0
yangsheng [Fri, 6 Mar 2009 17:54:09 +0000 (17:54 +0000)]
Branch b_release_1_8_0
b=18718

i=johann, shadow

Fixed build issues for sles9.

16 years agoupdated for estimated 1.8.0 release date
stinson1947 [Thu, 5 Mar 2009 23:08:48 +0000 (23:08 +0000)]
updated for estimated 1.8.0 release date

16 years agoBranch b_release_1_8_0
johann [Thu, 5 Mar 2009 21:34:31 +0000 (21:34 +0000)]
Branch b_release_1_8_0
b=18154
i=shadow
i=johann (original patch from bobijam)

always calls ptlrpc_import_recovery_state_machine() in
invalidate_import_thread() because otherwise threads sleeping on
the import's imp_recovery_waitq could be never woken up.

16 years ago- use truncate instead of invalidate to drop referenced pages from cache. This
tappro [Thu, 5 Mar 2009 06:54:15 +0000 (06:54 +0000)]
- use truncate instead of invalidate to drop referenced pages from cache. This
  is needed to disable ROC correctly
b:18520
i:adilger,green

16 years agob=17614
green [Wed, 4 Mar 2009 19:13:48 +0000 (19:13 +0000)]
b=17614
r=adilger, shadow

Do not put cancelled locks into replay list, hold references on locks in replay list

16 years agoBranch b_release_1_8_0
zam [Wed, 4 Mar 2009 18:43:19 +0000 (18:43 +0000)]
Branch b_release_1_8_0
b=17397
i=alexey.zhuravlev
i=andrew.perepechko

To avoid a deadlock in case of concurrent punch/write requests from one
client, filter writes and filter truncates are serialized by i_alloc_sem,
allowing multiple writes or single truncate.

16 years agoBranch b_release_1_8_0
johann [Tue, 3 Mar 2009 17:50:52 +0000 (17:50 +0000)]
Branch b_release_1_8_0
b=18414
i=nathan
i=shadow

fix watchdog timeout printed on the console. This was wrong since
support variable watchdogs has been landed.

16 years agob=17757
zhanghc [Tue, 3 Mar 2009 16:42:37 +0000 (16:42 +0000)]
b=17757

clear DISP_OPEN_LOCK flag in "reconstruct_open"
if the corresponding lock was canceled

i=Green
i=Johann

16 years agoBranch b_release_1_8_0
johann [Tue, 3 Mar 2009 15:08:48 +0000 (15:08 +0000)]
Branch b_release_1_8_0
b=18374
i=adilger
i=shadow

use memmove instead of memcpy when shrinking the reply since the two memory
areas can overlap.

16 years agob=17524
grev [Tue, 3 Mar 2009 13:02:31 +0000 (13:02 +0000)]
b=17524
reset_enospc fn: fix defect of previous commit

16 years agob=17524
grev [Fri, 27 Feb 2009 00:38:21 +0000 (00:38 +0000)]
b=17524
i=Adilger
exhaust_precreations, reset_enospc, test_27n, test_27s fixes

16 years agofix typo in b=17839 commit (made while ported from HEAD)
grev [Fri, 27 Feb 2009 00:16:49 +0000 (00:16 +0000)]
fix typo in b=17839 commit (made while ported from HEAD)

16 years agob=18606
nathan [Thu, 26 Feb 2009 19:51:45 +0000 (19:51 +0000)]
b=18606
i=adigler
i=johann
use pool of "all" if specified pool is empty

16 years agob=17846
girish [Thu, 26 Feb 2009 06:19:48 +0000 (06:19 +0000)]
b=17846
i=girish
i=h.huang
o=adilger

ignore stripe index 0 when setting striping via setxattr

16 years agob=18468
grev [Wed, 25 Feb 2009 16:52:48 +0000 (16:52 +0000)]
b=18468
o=Adilger
i=Oleg.Drokin
i=grev
make racer harder

16 years agoBranch b_release_1_8_0
fanyong [Wed, 25 Feb 2009 16:36:30 +0000 (16:36 +0000)]
Branch b_release_1_8_0
b=15691
i=huanghua
i=yep

Another diagnostic patch to collect more log when error.

16 years agob=17839
grev [Tue, 24 Feb 2009 22:41:36 +0000 (22:41 +0000)]
b=17839
i=Brian
cmd3-17 port to acc-sm

16 years agob=18443
grev [Fri, 20 Feb 2009 21:25:42 +0000 (21:25 +0000)]
b=18443
i=Adilger
remove obsolete tests scripts

16 years agob=18521
grev [Thu, 19 Feb 2009 22:18:01 +0000 (22:18 +0000)]
b=18521
i=Alex.Zhuravlev
sanity test_151, t-f localize var fix

16 years agob=18266
nathan [Wed, 18 Feb 2009 00:04:39 +0000 (00:04 +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

16 years agob=18285
grev [Tue, 17 Feb 2009 20:42:18 +0000 (20:42 +0000)]
b=18285
i=Rahul
i=Nathan
test_32a fix: remove mount client part

16 years agob=18428
vs [Tue, 17 Feb 2009 20:16:40 +0000 (20:16 +0000)]
b=18428
i=shadow,tappro,rread

do not send requests to inactive targets

16 years agob=18470
girish [Mon, 16 Feb 2009 07:47:42 +0000 (07:47 +0000)]
b=18470
i=johann
i=adilger

Correct error handling

16 years agob=15428
jxiong [Sun, 15 Feb 2009 02:35:14 +0000 (02:35 +0000)]
b=15428
r=adilger

Fixed a crash issue for raid56-zerocopy

16 years agob=18333
grev [Fri, 13 Feb 2009 19:54:12 +0000 (19:54 +0000)]
b=18333
i=Oleg.Drokin
add flock to default mount opts

16 years agofix typo for
grev [Fri, 13 Feb 2009 13:44:13 +0000 (13:44 +0000)]
fix typo for
b=16488
i=Oleg.Drokin
RACER acc-sm test suit

16 years agob=17839
grev [Thu, 12 Feb 2009 18:51:55 +0000 (18:51 +0000)]
b=17839
i=Brian
recovery-mds-scale for SLOW=no

16 years agorevert b=17839 commit
grev [Thu, 12 Feb 2009 16:14:38 +0000 (16:14 +0000)]
revert b=17839 commit

16 years agob=17839
grev [Thu, 12 Feb 2009 15:32:45 +0000 (15:32 +0000)]
b=17839
i=Brian
recovery-mds-scale for SLOW=no

16 years agob=18417
girish [Wed, 11 Feb 2009 18:08:16 +0000 (18:08 +0000)]
b=18417
o=h.huang
i=girish

Fix typo from last commit which caused test-framework to exit wrongly

16 years agob=18417
girish [Wed, 11 Feb 2009 06:55:51 +0000 (06:55 +0000)]
b=18417
i=kalpak
i=girish
o=adilger

cleanup correctly after sanity 102 tests

16 years agob=15428
jxiong [Tue, 10 Feb 2009 07:05:18 +0000 (07:05 +0000)]
b=15428
r=adilger

Fix a crash issue for raid5 zercopy patch.

16 years agoBranch b_release_1_8_0
tianzy [Tue, 10 Feb 2009 04:29:47 +0000 (04:29 +0000)]
Branch b_release_1_8_0
add an entry under /proc to support what policy(RR or QOS) will be used.
b=18334
i=nathan
i=adilger

16 years agob=18366
grev [Mon, 9 Feb 2009 23:44:35 +0000 (23:44 +0000)]
b=18366
i=Johann
test_5 ENOSPC fix

16 years agoBranch b_release_1_8_0
johann [Mon, 9 Feb 2009 14:59:19 +0000 (14:59 +0000)]
Branch b_release_1_8_0
i=johann (patch from Tappro)
i=shadow

Return only valid cookies, empty slot for failed ones.
This fixes a memory corruption causing random oops.

16 years agoBranch b_release_1_8_0
johann [Mon, 9 Feb 2009 14:57:20 +0000 (14:57 +0000)]
Branch b_release_1_8_0
b=18374
i=shadow
i=tappro

Wait for mds_fl_synced to be set during recovery before allowing clients
to reconnect. This is done to make sure that llog is properly set up.

16 years agoBranch b_release_1_8_0
johann [Mon, 9 Feb 2009 14:56:09 +0000 (14:56 +0000)]
Branch b_release_1_8_0
b=18067
i=johann
i=adilger

Change oap magic LASSERT into LASSERTF to aid debugging.

16 years agob=16823
green [Mon, 9 Feb 2009 00:17:38 +0000 (00:17 +0000)]
b=16823
r=adilger,shadow

Update offset calculations for 32 bit platforms

16 years agoBranch b_release_1_8_0
yangsheng [Thu, 5 Feb 2009 07:49:49 +0000 (07:49 +0000)]
Branch b_release_1_8_0
b=17459

i=adilger, bobijam

Fixed ext2_{...}_bit() works well in BIGENDIAN.

16 years agoBranch b_release_1_8_0
yangsheng [Thu, 5 Feb 2009 07:43:06 +0000 (07:43 +0000)]
Branch b_release_1_8_0
b=17379
i=johann, bobijam

Fixed correct depth for test-140.

16 years agob=17899
grev [Tue, 3 Feb 2009 13:32:57 +0000 (13:32 +0000)]
b=17899
i=Johann
test_53* fix: rm dirs used at start

16 years agoBranch b_release_1_8_0
tianzy [Tue, 3 Feb 2009 06:44:35 +0000 (06:44 +0000)]
Branch b_release_1_8_0
use QUOTA_REQ_RETURNED to replace -EAGAIN when a quota req is finished.
b=18307
i=johann
i=panda

16 years agoremove debug line
grev [Mon, 2 Feb 2009 13:53:14 +0000 (13:53 +0000)]
remove debug line

16 years agob=16818
grev [Mon, 2 Feb 2009 12:59:18 +0000 (12:59 +0000)]
b=16818
i=Nathan
get rid of set_obd_timeout(), improve test_26*

16 years agob=17895
kalpak [Mon, 2 Feb 2009 09:15:37 +0000 (09:15 +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.

16 years agoBranch b_release_1_8_0
huanghua [Thu, 29 Jan 2009 09:08:55 +0000 (09:08 +0000)]
Branch b_release_1_8_0

b=18318
i=rahul
i=pravin

reserve larger reply buffer for close and unlink:
2.0 mds wire protocol returns capabilities.

16 years agoReturn EXT_UNSET_BLOCK from ext3_ext_next_leaf_block.
girish [Wed, 28 Jan 2009 11:29:40 +0000 (11:29 +0000)]
Return EXT_UNSET_BLOCK from ext3_ext_next_leaf_block.
Patch was missing for 2.6.22
Added changelog entry along with newer build version.
b=16114

16 years agodon't lose wakeup for imp_recovery_waitq
shadow [Wed, 28 Jan 2009 09:09:30 +0000 (09:09 +0000)]
don't lose wakeup for imp_recovery_waitq

Branch b_release_1_8_0
b=18154
i=nathan
i=rread

16 years agob=16895
girish [Wed, 28 Jan 2009 03:52:26 +0000 (03:52 +0000)]
b=16895
i=adilger (o=johann)
i=girish

dont assert on lvfs_check_rdonly()

16 years agoBranch b_release_1_8_0
yangsheng [Sat, 24 Jan 2009 18:11:57 +0000 (18:11 +0000)]
Branch b_release_1_8_0
b=15593
i=nikita, shadow

Force to remove the suid in interop environment.

16 years agob=18045
grev [Sat, 24 Jan 2009 09:16:51 +0000 (09:16 +0000)]
b=18045
i=Adilger
-- get rid of Hendrix rates values
-- create NUM_FILES -le IFree
-- use mdsrate for cleanup instead of rm

16 years agoBranch b_release_1_8_0
huanghua [Sat, 24 Jan 2009 02:09:36 +0000 (02:09 +0000)]
Branch b_release_1_8_0
b=17777
i=robert.read

as Robert suggests, get lustre version with more precise match.

16 years agob=17777
huanghua [Fri, 23 Jan 2009 17:09:51 +0000 (17:09 +0000)]
b=17777
i=grev
i=adilger
i=robert.read

test scripts improvement to enable interop testing between 1.8 and 2.0

16 years agob=18279
grev [Fri, 23 Jan 2009 13:55:43 +0000 (13:55 +0000)]
b=18279
i=Nathan
AT tests cleanup: run setup/cleanup only if AT tests run

16 years agodon't leak openhandle in ESTALE case.
shadow [Fri, 23 Jan 2009 13:32:34 +0000 (13:32 +0000)]
don't leak openhandle in ESTALE case.

Branch b_release_1_8_0
b=16417
i=green
i=johann
i=bobijam

16 years agoBranch b_release_1_8_0
tianzy [Fri, 23 Jan 2009 09:50:27 +0000 (09:50 +0000)]
Branch b_release_1_8_0
when there is any inactive ost, refuse to do quotacheck
b=18151
i=johann
i=andrew.perepechko

16 years agoBranch b_release_1_8_0
tianzy [Fri, 23 Jan 2009 06:21:14 +0000 (06:21 +0000)]
Branch b_release_1_8_0
fix the test issue of test_12 in sanity-quota.sh
b=18151
i=johann
i=panda

16 years agob=18173
girish [Thu, 22 Jan 2009 07:37:33 +0000 (07:37 +0000)]
b=18173
i=adilger
i=kalpak

If there is error during mount, MMP should error out correctly

16 years agob=16895
girish [Wed, 21 Jan 2009 18:38:51 +0000 (18:38 +0000)]
b=16895
i=adilger (o=johann)
i=girish

don't LASSERT on read-only device

16 years agoBranch b_release_1_8_0
huanghua [Wed, 21 Jan 2009 18:22:18 +0000 (18:22 +0000)]
Branch b_release_1_8_0
b=18267

i=yong.fan
i=rahul.deshmukh

pass default group value, that is 0,  to ost:
  this is for such interop: 1.8 client + 1.8 mds + 2.0 ost.

16 years agob=17310
yury [Wed, 21 Jan 2009 16:48:03 +0000 (16:48 +0000)]
b=17310
r=johann

- tmp fix for long connect rpc interpreyt which causes the assert

16 years agoBranch b_release_1_8_0
tianzy [Wed, 21 Jan 2009 02:56:48 +0000 (02:56 +0000)]
Branch b_release_1_8_0
take refcount of export when quotacheck is going on
b=18126
i=shadow
i=panda

16 years agoalways handle mds connect as reconnect.
shadow [Tue, 20 Jan 2009 12:08:22 +0000 (12:08 +0000)]
always handle mds connect as reconnect.

Branch b_release_1_8_0
b=18224
i=johhan
i=green
i=adilger
i=wangdi

16 years agodon't confuse mds_finish_transno() with PTR_ERR(-ENOENT)
shadow [Tue, 20 Jan 2009 09:02:21 +0000 (09:02 +0000)]
don't confuse mds_finish_transno() with PTR_ERR(-ENOENT)

Branch b_release_1_8_0
b=18238
i=green
i=umka

16 years agob=17942
kalpak [Tue, 20 Jan 2009 04:43:58 +0000 (04:43 +0000)]
b=17942
i=adilger
i=girish

update fiemap structures and API to match upstream

16 years agorelease lock refernce only after add page to pages list.
shadow [Mon, 19 Jan 2009 21:21:30 +0000 (21:21 +0000)]
release lock refernce only after add page to pages list.

Branch b_release_1_8_0
b=17972
i=shadow
i=johann

16 years agob=17787
zhanghc [Tue, 13 Jan 2009 23:54:06 +0000 (23:54 +0000)]
b=17787

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 obd_timeout

i=vitaly.fertman
i=johann

16 years ago- replace assert with cwarn when gap is found in objids. This is allowed
tappro [Tue, 13 Jan 2009 07:21:42 +0000 (07:21 +0000)]
- replace assert with cwarn when gap is found in objids. This is allowed
situation due to inode reuse
b:18168
i:adilger, shadow

16 years agowait until mds<>ost connect finished.
shadow [Sun, 11 Jan 2009 08:29:34 +0000 (08:29 +0000)]
wait until mds<>ost connect finished.
Branch b_release_1_8_0
b=18150
i=grev

16 years agoresolve compile issue after 16919 landing.
shadow [Sun, 11 Jan 2009 07:16:58 +0000 (07:16 +0000)]
resolve compile issue after 16919 landing.
b=16919
Branch b_release_1_8_0

16 years agob=18016
green [Sat, 10 Jan 2009 06:45:17 +0000 (06:45 +0000)]
b=18016
r=deen,wangdi

page index type in ll_read_ahead_page should be pgoff_t, not int

16 years agob=18114
yury [Fri, 9 Jan 2009 12:29:36 +0000 (12:29 +0000)]
b=18114
r=umka,bzzz

- patch from shadow to fix use after free in lov_cleanup()

16 years agob=16919
green [Fri, 9 Jan 2009 04:35:08 +0000 (04:35 +0000)]
b=16919
r=adilger,shadow

Implement replay of bulk write RPCs, allow servers to reply to write RPCs before
all the metadata was finally committed to disk