Whamcloud - gitweb
fs/lustre-release.git
11 years agob=17569
girish [Mon, 29 Jun 2009 04:34:56 +0000 (04:34 +0000)]
b=17569
i=adilger
i=johann

add check for >8TB filesystems. Don't allo mounting them until force_over_8tb mount option is specified

11 years agoAdd async create into HEAD (port from 1.x)
shadow [Sat, 27 Jun 2009 05:28:28 +0000 (05:28 +0000)]
Add async create into HEAD (port from 1.x)

Branch HEAD
b=18357
i=rread

11 years agowith AT enabled, we can have connect request timeout ~ reconnect
shadow [Sat, 27 Jun 2009 04:56:03 +0000 (04:56 +0000)]
with AT enabled, we can have connect request timeout ~ reconnect
timeout and test can't see real disconnect

Branch HEAD
b=19280
i=rread

11 years ago Updated version string for 2.0.0 build 21.
stinson1947 [Fri, 26 Jun 2009 20:22:17 +0000 (20:22 +0000)]
 Updated version string for 2.0.0 build 21.

11 years agoBranch HEAD
tappro [Fri, 26 Jun 2009 17:49:58 +0000 (17:49 +0000)]
Branch HEAD
b=19550
i=rread
i=shadow

Fix small issue with incorrect check for flags

11 years agoBranch HEAD
tappro [Fri, 26 Jun 2009 17:48:12 +0000 (17:48 +0000)]
Branch HEAD
b=19550
i=rread
i=zam

Add lock counter to export and output it in dump_exports

11 years agoBranch HEAD
robert.read [Fri, 26 Jun 2009 17:01:01 +0000 (17:01 +0000)]
Branch HEAD
b=19933
i=grev
a=johann

Add error messages to sanity test 103

11 years agob=19927
brian [Fri, 26 Jun 2009 13:20:03 +0000 (13:20 +0000)]
b=19927
i=yangsheng
i=wangyb

If the $LUSTRE_VERS environment variable is set, don't bother doing the
legwork to determine the version, and just exit early with it.

11 years agob=19086
vs [Fri, 26 Jun 2009 13:18:26 +0000 (13:18 +0000)]
b=19086
i=rread,tappro

- remove call of df in backgroup from facet_failover()
- use client_df in fail() to wait for recovery completion
- replay_single:test_70b: use fail() instead of facet_failover
- replay_single:test_74: umount/mount all clients

11 years agoBranch b1_8
yangsheng [Fri, 26 Jun 2009 11:30:49 +0000 (11:30 +0000)]
Branch b1_8
b=19848

i=brian, wangyb

Change for sles11 new kernel build.

11 years agob=19451
grev [Fri, 26 Jun 2009 11:04:59 +0000 (11:04 +0000)]
b=19451
i=Brian
IOR client load for recovery scale tests

11 years agoo=Adilger
grev [Fri, 26 Jun 2009 09:14:58 +0000 (09:14 +0000)]
o=Adilger
i=Robert
i=grev
report patchless_client for kernel version

11 years agoBranch HEAD
fanyong [Fri, 26 Jun 2009 06:31:01 +0000 (06:31 +0000)]
Branch HEAD
b=19435
i=tianzy
i=andrew.perepechko

Process quota on/off/invalid serially.

11 years agoBranch HEAD
fanyong [Fri, 26 Jun 2009 06:01:04 +0000 (06:01 +0000)]
Branch HEAD
b=18317
i=tianzy
i=andrew.perepechko

Wait until all lqs have been freed on cleanup.

11 years agoBranch HEAD
bobijam [Fri, 26 Jun 2009 01:07:52 +0000 (01:07 +0000)]
Branch HEAD
b=16317
i=adilger

add conf-sanity test_44.

11 years agob=18021
grev [Thu, 25 Jun 2009 16:58:20 +0000 (16:58 +0000)]
b=18021
i=Adilger
i=Nathan
i=Robert
t-f CLIENTONLY does not depend on *_HOST

11 years agoBranch HEAD
vitaly [Thu, 25 Jun 2009 09:58:38 +0000 (09:58 +0000)]
Branch HEAD
b=18619
i=vitaly
i=johann

max rpc in flight panda's fix

11 years agoBranch HEAD
yujian [Thu, 25 Jun 2009 01:42:57 +0000 (01:42 +0000)]
Branch HEAD
b=19413
i=eric.mei

specify $MOUNT for $LFS flushctx

11 years agoBranch HEAD
fanyong [Wed, 24 Jun 2009 23:04:49 +0000 (23:04 +0000)]
Branch HEAD
b=18574
i=robert.read

typo fix for the srcipts.

11 years agoBranch b1_8
johann [Wed, 24 Jun 2009 22:48:19 +0000 (22:48 +0000)]
Branch b1_8
b=19875

backport patches to RHEL5/ext4.

11 years agoBranch b1_8
johann [Wed, 24 Jun 2009 21:33:44 +0000 (21:33 +0000)]
Branch b1_8

remove ext4-ialloc-2.6-rhel5.patch from patch series.

11 years agoBranch HEAD
tappro [Wed, 24 Jun 2009 19:12:57 +0000 (19:12 +0000)]
Branch HEAD
b=19550
i=rread
i=zam

Move unlinked exports to the new list until they become 'zombie'. Wait for these
exports during umount. Output in dump_exports() is exteneded and contain more
fields to help with debugging. The new exp_cb_count was added to track number of
callbacks keeping the export references.

11 years agob=19853
brian [Wed, 24 Jun 2009 18:45:06 +0000 (18:45 +0000)]
b=19853
i=jack
i=yangsheng

The SLES 11 kernel spec has what is IMHO a bug in that it tries to write to
a file in RPMS/i386 even though you are building i686 RPMs.

11 years agob=19850
brian [Wed, 24 Jun 2009 18:27:19 +0000 (18:27 +0000)]
b=19850
i=jack
i=wangyb

Shouldn't disable srp-target-mod after enabling it.

11 years agoBranch b1_8
johann [Wed, 24 Jun 2009 15:39:18 +0000 (15:39 +0000)]
Branch b1_8
b=19875
i=adilger
i=yansgheng

- use mb_find_next_bit() instead of find_next_bit() in
  ext4_mb_check_ondisk_bitmap() as done in the ext4 mainline
- convert ext4_lock_group to use sb_bgl_lock, patch already
  merged upstream.

11 years agoBranch b1_8
yangsheng [Wed, 24 Jun 2009 13:51:18 +0000 (13:51 +0000)]
Branch b1_8
b=19869

i=shadow, johann, tianzy

Fixes for Werror flag enabled.

11 years agob=17670
pravins [Tue, 23 Jun 2009 21:52:20 +0000 (21:52 +0000)]
b=17670
fix build failure from last commit

11 years agob=17670
pravins [Tue, 23 Jun 2009 21:03:20 +0000 (21:03 +0000)]
b=17670
i=tom.wang
i=yong.fan

osd cleanup.
        -- remove unneccessary forward declaration.
        -- reorg osd_ea_fid_get(); req for fid_in_dirent patch.
        -- remove oie_curr_pos and oie_next_pos

11 years agob=17670
pravins [Tue, 23 Jun 2009 20:59:16 +0000 (20:59 +0000)]
b=17670
i=tom.wang
i=yong.fan

build and return lu_dirent from osd iterator.
use type info available in ext3 dirent

11 years agoBranch HEAD
tappro [Tue, 23 Jun 2009 17:36:09 +0000 (17:36 +0000)]
Branch HEAD
b=19885
i=rread
i=zam

Simpified interop case. When 2.0 server is started over old 1.8 FS it evicts
clients by truncating last_rcvd file. Therefore clients will use 2.0-formatted RPC after
reconnect

11 years agoBranch HEAD
tianzy [Tue, 23 Jun 2009 06:25:18 +0000 (06:25 +0000)]
Branch HEAD
use lqs to judge which uid/gid are set quota limitaton and
which uid/gid are not, and treat them differently.
I found a few issues after landing the patch for 18574.(some are caused by it, some are easier to
be reproduced by it)
1. in setquota, quota slaves will set their quota limitation to 1 and then get more quota. But in
this time, it's lqs size is the default value. If we have set a small quota limitation, we will
meet -EDQUOT. So we should adjust qunit firstly in this situation.
2. filter_quota_ctl(), we should promise there is no qunit is calculated or sent between
qctxt_wait_pending_dqacq() and fsfilt_quotactl(). Now I did this via a "fake" qunit.
b=18574
i=johann
i=panda
aVS: ----------------------------------------------------------------------

11 years agoBranch HEAD
tianzy [Tue, 23 Jun 2009 06:20:26 +0000 (06:20 +0000)]
Branch HEAD
part I
1. mainly fix the lqs_key of lustre_qunit_size.
2. clean quota_compute_lqs().
3. only supply quota_search_lqs() to outside and make quota_create_lqs() internal.
b=18616
i=johann
i=panda

11 years agoi=wangdi,i=rread,b=19920:
isaac [Tue, 23 Jun 2009 04:06:17 +0000 (04:06 +0000)]
i=wangdi,i=rread,b=19920:
- LIBCFS_ALLOC_GFP should not print 'kmalloced' message at allocation failures.

11 years agoBranch HEAD
dzogin [Tue, 23 Jun 2009 01:01:29 +0000 (01:01 +0000)]
Branch HEAD
b=18624
i=zhenyu.xu
i=dmitry.zogin
Description: Unable to run several mkfs.lustre on loop devices at the same time.

11 years agob=19188
jxiong [Mon, 22 Jun 2009 23:23:51 +0000 (23:23 +0000)]
b=19188
r=rread,fanyong

cleanup the variables with `_', rename them to `unused'

11 years agobranch: HEAD
ericm [Mon, 22 Jun 2009 19:44:51 +0000 (19:44 +0000)]
branch: HEAD
server save opc in replay_state thus we don't need to access repmsg after
reply being submmitted to ptlrpc.
b=19386
r=wangdi
r=fanyong

11 years agoBranch HEAD
vitaly [Mon, 22 Jun 2009 19:21:18 +0000 (19:21 +0000)]
Branch HEAD
b=19154
i=tappro
i=vs

let the statfs cache to get old enough on replay-single 20b

11 years agob=19188
jxiong [Mon, 22 Jun 2009 18:39:46 +0000 (18:39 +0000)]
b=19188
r=pravin,rread

More debug info are added to catch bugs.

11 years agob=19188
jxiong [Mon, 22 Jun 2009 18:34:53 +0000 (18:34 +0000)]
b=19188
r=pravin,rread

It is bad to access the ols_lock if the lock is detached.

11 years agob=19398
deshmukh [Mon, 22 Jun 2009 18:04:37 +0000 (18:04 +0000)]
b=19398
i=tom.wang
i=nathan.rutman
i=jinshan.xiong

Fix for soft lockup - CPU#13 stuck for 10s! [umount:16789]

11 years agoBranch HEAD
bobijam [Mon, 22 Jun 2009 01:49:08 +0000 (01:49 +0000)]
Branch HEAD
b=19913
o=grev
i=johann

skip test_151 for remote ost && no_dsh

11 years agoBranch HEAD
bobijam [Mon, 22 Jun 2009 01:46:52 +0000 (01:46 +0000)]
Branch HEAD
b=19913
o=johann
i=grev
i=bobijam

Add support remote ost in test 151.

11 years agoBranch HEAD
fanyong [Fri, 19 Jun 2009 16:47:22 +0000 (16:47 +0000)]
Branch HEAD
b=19149
i=tappro
i=tom.wang

Cleanup pending files after client eviction.

11 years agoBranch b1_8
johann [Fri, 19 Jun 2009 12:17:37 +0000 (12:17 +0000)]
Branch b1_8
b=19851
i=girish

Remove EXT4_MAX_DIR_SIZE_NAME proc entry before the root is removed.

11 years agob=19875
girish [Fri, 19 Jun 2009 11:29:26 +0000 (11:29 +0000)]
b=19875
i=adilger
i=johann

add locking for i_cached_extent

11 years agob=19875
girish [Fri, 19 Jun 2009 11:24:46 +0000 (11:24 +0000)]
b=19875
i=johann
i=adilger

use i_fs_version in sles11

11 years agoBranch HEAD
fanyong [Thu, 18 Jun 2009 22:59:06 +0000 (22:59 +0000)]
Branch HEAD
b=19577
i=robert.read
i=jinshan.xiong

replace function name.

11 years agob=19669
nathan [Thu, 18 Jun 2009 21:48:04 +0000 (21:48 +0000)]
b=19669
i=nrutman
i=adilger
add compat/incompat flags to lma

11 years agoBranch b1_8
yangsheng [Thu, 18 Jun 2009 09:42:02 +0000 (09:42 +0000)]
Branch b1_8
b=19574

11 years agoBranch b1_8
yangsheng [Thu, 18 Jun 2009 08:34:08 +0000 (08:34 +0000)]
Branch b1_8
b=19574

i=brian, johann

Notice unsupported modules flag for sles11.

11 years agobranch: HEAD
ericm [Thu, 18 Jun 2009 01:32:40 +0000 (01:32 +0000)]
branch: HEAD
fix a gss build error.
b=19826
r=yujian
r=fanyong

11 years agoBranch HEAD
fanyong [Thu, 18 Jun 2009 01:11:41 +0000 (01:11 +0000)]
Branch HEAD
b=19577
i=robert.read
i=jinshan.xiong

1) Reset debub level in sanity-quota test.
2) Cleanup test scripts.

11 years agoBranch b1_8
johann [Wed, 17 Jun 2009 22:06:04 +0000 (22:06 +0000)]
Branch b1_8
b=19875
i=johann (patch from girish)
i=adilger

Disable delayed allocation since this is not robust in kernels prior to 2.6.31.

11 years agoBranch HEAD
tappro [Wed, 17 Jun 2009 20:59:31 +0000 (20:59 +0000)]
Branch HEAD
 b=19550
 i=rread

Remove loop for obd refcount waiting and add asserts against possible obd
refcount leaks

11 years agob=19856
nathan [Wed, 17 Jun 2009 19:45:04 +0000 (19:45 +0000)]
b=19856
i=adilger
i=rread
move lu_fid definition into userspace

11 years ago- a trivial comment update.
isaac [Tue, 16 Jun 2009 23:56:11 +0000 (23:56 +0000)]
- a trivial comment update.

11 years agoi=liang,b=17546:
isaac [Tue, 16 Jun 2009 22:22:58 +0000 (22:22 +0000)]
i=liang,b=17546:
- fixed kptllnd HELLO protocol deadlock.

11 years agoBranch HEAD
johann [Tue, 16 Jun 2009 20:59:13 +0000 (20:59 +0000)]
Branch HEAD
b=19725
i=adilger
i=girish

ext4_iget() returns an error code now instead of NULL.

11 years agob=19867
anserper [Tue, 16 Jun 2009 19:56:50 +0000 (19:56 +0000)]
b=19867
i=Yong Fan

quota_save_version fixes

11 years agoBranch HEAD
robert.read [Tue, 16 Jun 2009 18:35:05 +0000 (18:35 +0000)]
Branch HEAD
b=16424
i=sheng.yang
i=yibin.wang
a=brian

Build fix for sles' -obj dirs.

11 years agob=19188
pravins [Tue, 16 Jun 2009 17:55:20 +0000 (17:55 +0000)]
b=19188
i=fan yong
i=rahul
add check for dead object and fix error handling in mdd_rename()

11 years agob=19756
green [Tue, 16 Jun 2009 15:42:05 +0000 (15:42 +0000)]
b=19756
r=tappro,rread

free parent entry in reconstruct_open

11 years agoBranch HEAD
vitaly [Tue, 16 Jun 2009 12:34:43 +0000 (12:34 +0000)]
Branch HEAD
b=19154
i=rread
i=tappro

an ability to wait for the completion of an MDS-OST synchronization is added

11 years agoBranch HEAD
vitaly [Tue, 16 Jun 2009 12:31:38 +0000 (12:31 +0000)]
Branch HEAD
b=19154
i=adilger
i=tappro

propagate mount count down the mdt stack so it could be properly used in llog
to send it on the connection to ost.

11 years agob=19259
zhanghc [Tue, 16 Jun 2009 11:39:30 +0000 (11:39 +0000)]
b=19259

add a NULL signal proc in osc_destroy to allow
the blocking thread to process LUSTRE_FATAL_SIGS

i=adilger@sun.com
i=johann@sun.com

11 years agob=19852
girish [Tue, 16 Jun 2009 02:43:29 +0000 (02:43 +0000)]
b=19852
update changelog to use new e2fsprogs version

11 years agoBranch HEAD
robert.read [Mon, 15 Jun 2009 21:41:44 +0000 (21:41 +0000)]
Branch HEAD

Only run compilebench 2x as long when SLOW=y, instead of 5x.

11 years agoo=Nathan
grev [Mon, 15 Jun 2009 13:07:00 +0000 (13:07 +0000)]
o=Nathan
i=Johann
minor script issue in test_17

11 years agob=19694
vs [Mon, 15 Jun 2009 00:30:14 +0000 (00:30 +0000)]
b=19694
i=tappro,zam

moo_xattr_list method is added to mdd_dot_lustre_obj_ops

11 years agob=19593
vs [Mon, 15 Jun 2009 00:24:03 +0000 (00:24 +0000)]
b=19593
i=maxim

fix memory leak in error handling of ptlrpc_init()

11 years agob=16823
green [Fri, 12 Jun 2009 18:05:55 +0000 (18:05 +0000)]
b=16823
r=adilger,shadow

Allow wide stripe size of up to 4G-64k

11 years agoBranch HEAD
zam [Fri, 12 Jun 2009 05:45:53 +0000 (05:45 +0000)]
Branch HEAD
b=19321
i=johann
i=alex.zhuravlev

jbd and jbd2 fixes: remove jiffie rounding from jbd commit timer setup code.
The rounding can cause the timer to be fired before the corresponding
transaction gets expired. That transaction can stay not committed forever if
no new transaction is created or expicit sync/umount happens.

Note: the fix for sles11 kernel is added although  there is no sles11 series
on HEAD yet.

11 years agoBranch HEAD
robert.read [Fri, 12 Jun 2009 00:20:29 +0000 (00:20 +0000)]
Branch HEAD
b=19820
a=ed
i=zhenyu.xu

Add missing stop_ost2.

11 years agoBranch HEAD
robert.read [Thu, 11 Jun 2009 18:31:21 +0000 (18:31 +0000)]
Branch HEAD

Fix typo.

11 years agob=18914
grev [Thu, 11 Jun 2009 14:46:09 +0000 (14:46 +0000)]
b=18914
i=Brian
check_client_load: try to connect several times

11 years agob=19762
brian [Thu, 11 Jun 2009 12:36:50 +0000 (12:36 +0000)]
b=19762
i=jack
i=yangsheng

Because build_kernel_with_srpm()'s caller expects it to echo a return value,
we need to make sure nothing else is on build_kernel_with_srpm()'s stdout.

11 years agoBranch HEAD
tianzy [Thu, 11 Jun 2009 06:13:29 +0000 (06:13 +0000)]
Branch HEAD
fix MDS lockup during inode quota test
b=19672
i=johann
i=yong.fan

11 years agob=19544
anserper [Thu, 11 Jun 2009 00:45:53 +0000 (00:45 +0000)]
b=19544
i=Robert Read
i=ZhiYong Tian

get rid of variable-length arrays in the quota format code

11 years agoBranch b1_8
adilger [Wed, 10 Jun 2009 21:36:22 +0000 (21:36 +0000)]
Branch b1_8
Ignore generated files.

11 years agob=16424
brian [Wed, 10 Jun 2009 21:09:56 +0000 (21:09 +0000)]
b=16424
i=rread
i=yangsheng

Allow builds from tarball (i.e. those without CVS subdirs) to properly
populate the Lustre version field.

This is done by saving the date that would have been found in the CVS files
into a status file when buiding the tarball and then using that when building
from tarball.

11 years agob=16424
brian [Wed, 10 Jun 2009 20:22:14 +0000 (20:22 +0000)]
b=16424
i=rread
i=yangsheng

Allow builds from tarball (i.e. those without CVS subdirs) to properly
populate the Lustre version field.

This is done by saving the date that would have been found in the CVS files
into a status file when buiding the tarball and then using that when building
from tarball.

11 years agob=18364
zhanghc [Wed, 10 Jun 2009 17:22:37 +0000 (17:22 +0000)]
b=18364

1, add initialized code for local obdo variables
2, move OBD_FL_* flags definition into an enum structure

i=adilger@sun.com
i=johann@sun.com

11 years agoBranch HEAD
scjody [Wed, 10 Jun 2009 13:08:37 +0000 (13:08 +0000)]
Branch HEAD
b=17986
i=adilger
i=scjody

Allow tuning of the RAID stripe cache size in mount.lustre; automatically
tune it to 2048 by default.

Original patch from James Simmons <simmonsja@ornl.gov>

11 years agoBranch HEAD
vitaly [Wed, 10 Jun 2009 10:17:01 +0000 (10:17 +0000)]
Branch HEAD
b=19802
i=grev
i=vs

drop the wrong output redirection

11 years agob=18881
jxiong [Wed, 10 Jun 2009 06:06:26 +0000 (06:06 +0000)]
b=18881
r=rread

a diagnose patch to help print the needed info.

11 years agofix up some cvsignores
nathan [Tue, 9 Jun 2009 23:10:41 +0000 (23:10 +0000)]
fix up some cvsignores

11 years agob=18618
nathan [Tue, 9 Jun 2009 22:56:57 +0000 (22:56 +0000)]
b=18618
i=adilger
i=johann
don't increase ldlm timeout if previous client was evicted

11 years agoBranch HEAD
robert.read [Tue, 9 Jun 2009 22:34:54 +0000 (22:34 +0000)]
Branch HEAD
b=19577

Change error_exit calls to error so the rest of the tests can run.

11 years agoBranch b1_8
johann [Tue, 9 Jun 2009 09:54:11 +0000 (09:54 +0000)]
Branch b1_8
b=19686
i=adilger
i=girish

Add ext4-alloc-policy-2.6-sles11.patch to the SLES11 series.
Fixed wantedi patch to use ext4_claim_inode().

11 years agoBranch b1_8
adilger [Mon, 8 Jun 2009 22:56:08 +0000 (22:56 +0000)]
Branch b1_8
Update "make checkstack" target to find all modules in the tree.  Because
we no longer need to support 2.4 kernels this can just find all "*.ko" files.

Update the "checkstack.pl" program to work better with newer compilers.

b=19769
i=zhenyu.xu
i=brian

11 years agob=18787
zhanghc [Mon, 8 Jun 2009 16:00:37 +0000 (16:00 +0000)]
b=18787

disable per-CPU stats data structures for clients stats
to shrink memory usage

i=oleg.drokin@sun.com
i=hongchao.zhang@sun.com

11 years agoBranch b_hd_kdmu
rcorreia [Mon, 8 Jun 2009 15:13:32 +0000 (15:13 +0000)]
Branch b_hd_kdmu
b=17825

Make b_hd_kdmu branch use the new ZFS repo.

11 years agob=19702
grev [Mon, 8 Jun 2009 14:13:50 +0000 (14:13 +0000)]
b=19702
i=Johann
i=Vladimir.Saveliev
use NULL instead of newline as filename terminator

11 years agoBranch HEAD
vitaly [Mon, 8 Jun 2009 11:13:07 +0000 (11:13 +0000)]
Branch HEAD
b=19773
i=vs
i=grev

declare arrays properly

11 years agoBranch HEAD
bobijam [Sat, 6 Jun 2009 05:57:05 +0000 (05:57 +0000)]
Branch HEAD
b=18918
i=nathan.rutman

fix service name on (un)mount for MGS.

11 years agob=17560
pravins [Sat, 6 Jun 2009 05:55:42 +0000 (05:55 +0000)]
b=17560
i=yong.fan
i=grev

test case for hash collision handling.

11 years agoBranch HEAD
bobijam [Sat, 6 Jun 2009 05:53:03 +0000 (05:53 +0000)]
Branch HEAD
b=18638
i=jinshan.xiong (jay)
i=andrew.perepachko (panda)

ldlm_lock_put() use RCU to call ldlm_lock_free, so need call synchronize_rcu()
to wait a grace period elapsed, so that ldlm_lock_free() get a chance to be
called.

11 years agob=19784
nathan [Fri, 5 Jun 2009 21:15:00 +0000 (21:15 +0000)]
b=19784
i=rread
i=manoj
remove unneeded lu_env_init from fid2path

11 years agob=19407
manoj [Fri, 5 Jun 2009 21:14:15 +0000 (21:14 +0000)]
b=19407
r=grev
r=manoj.joseph
o=robert.read

Use do_nodes instead of looping through the nodes sequentially

11 years agoUpdated version for 2.0 Build 20.
stinson1947 [Fri, 5 Jun 2009 01:21:30 +0000 (01:21 +0000)]
Updated version for 2.0 Build 20.