Whamcloud - gitweb
shadow [Thu, 19 Nov 2009 11:22:40 +0000 (11:22 +0000)]
avoid unnecessary allocation
Branch b1_8
b=19557
i=vitaly
johann [Wed, 18 Nov 2009 19:09:31 +0000 (19:09 +0000)]
b=16774,21252
Disable lock cancelling before replay due to bug 21252.
panda [Mon, 16 Nov 2009 21:56:59 +0000 (21:56 +0000)]
b=21320
i=Mike
i=Johann
i=Yong Fan
reinitialise rq_timed_list when needed
stinson1947 [Mon, 16 Nov 2009 17:49:30 +0000 (17:49 +0000)]
Updated for 1.8.2 build 04.
zhanghc [Mon, 16 Nov 2009 10:29:37 +0000 (10:29 +0000)]
b=16818
fix a bug in previous patch
i=andrew.perepechko
zhanghc [Mon, 16 Nov 2009 01:27:06 +0000 (01:27 +0000)]
b=16818
1, the test will be skipped if the MGS and OST1 are on the same node
2, the test will be interrupted if "zconf_mount $MOUNT2" failed
i=nathan.rutman@sun.com
i=robert.read@sun.com
i=grev@sun.com
panda [Sun, 15 Nov 2009 14:34:26 +0000 (14:34 +0000)]
b=21377
use explicit type cast for find_next_bit to fix the following:
Build has failed, dumping the latest log: http://lts-head.lustre.sun.com:3000/proxy/x86-build-0/FILECACHE-LOG-build-32831/build.log
Building modules, stage 2.
Making all in ldiskfs
Making all in .
WARNING: Symbol version dump /cache/build/reused/usr/src/linux-2.6.16.60-0.42.4_lustre.1.8.1.53-obj/x86_64/smp/Module.symvers
is missing; modules will have no dependencies and modversions.
cc1: warnings being treated as errors
/cache/build/BUILD/lustre-1.8.1.53/lustre/lvfs/fsfilt-ldiskfs.c: In function ���fsfilt_ldiskfs_quotacheck���:
/cache/build/BUILD/lustre-1.8.1.53/lustre/lvfs/fsfilt-ldiskfs.c:2071: warning: passing argument 1 of ���find_next_bit��� from incompatible pointer type
make[8]: *** [/cache/build/BUILD/lustre-1.8.1.53/lustre/lvfs/fsfilt-ldiskfs.s] Error 1
Build succeeds with this patch:
http://lts-head.lustre.sun.com:3000/build_jobs/32832
johann [Sat, 14 Nov 2009 00:26:15 +0000 (00:26 +0000)]
b1_8
b=21377
define ext2_find_next_bit for find_next_bit for little endian system
on kernel that have ext2_find_next_bit yet. Similar to what mballoc does.
dzogin [Fri, 13 Nov 2009 22:47:51 +0000 (22:47 +0000)]
Branch b1_8
b=21259
i=andrew.perepechko
i=alexey.lyashkov
----------------------------------------------------------------------
Modified Files:
Tag: b1_8
lustre/ChangeLog lustre/obdclass/linux/linux-module.c
----------------------------------------------------------------------
Details : Allow non-root access for "lfs check".
grev [Fri, 13 Nov 2009 11:27:59 +0000 (11:27 +0000)]
b=21115
i=Robert.Read
t-f: mount the same config on all clients
stinson1947 [Thu, 12 Nov 2009 22:56:42 +0000 (22:56 +0000)]
Updated for 1.8.2. build 03.
brian [Thu, 12 Nov 2009 21:35:47 +0000 (21:35 +0000)]
b=19721
i=brian
o=adilger
Workaround in version_tag.pl for some (Debian?) systems where, for whatever
reason, MODULES_{TRUE|FALSE} does not get included in the {auto,}Makefile.in
files
nathan [Thu, 12 Nov 2009 19:57:31 +0000 (19:57 +0000)]
b=21306
i=manoj
insure pool additions propogate before striping
yangsheng [Thu, 12 Nov 2009 17:14:59 +0000 (17:14 +0000)]
Branch b1_8
b=20773
i=adilger, shadow, girish, brian, wangyb
Update kernel to RHEL5 U4 2.6.18-164.6.1.el5.
johann [Thu, 12 Nov 2009 09:54:21 +0000 (09:54 +0000)]
Branch b1_8
b=20892
i=shadow
i=yangsheng
Take inode semaphores instead of the bkl before calling vfs_rename.
nathan [Wed, 11 Nov 2009 21:53:27 +0000 (21:53 +0000)]
b=21304
i=manoj
compare all the uuids in the pool list
shadow [Wed, 11 Nov 2009 13:09:38 +0000 (13:09 +0000)]
remove unused macros.
b=14250
i=shadow
o=isac
panda [Wed, 11 Nov 2009 00:11:41 +0000 (00:11 +0000)]
b=21150
i=ZhiYong Tian
i=Johann Lombardi
protect dqacq processing against mds-lov cleanup
johann [Tue, 10 Nov 2009 23:00:09 +0000 (23:00 +0000)]
Branch b1_8
b=19887
Realign test_26b with the HEAD version of the test.
brian [Tue, 10 Nov 2009 21:56:06 +0000 (21:56 +0000)]
b=20451
i=yangsheng
i=johann
Evidently (Ubuntu's, at least) 2.6.28 needs to include linux/pagemap.h
to get the prototype for grab_cache_page_write_begin().
johann [Tue, 10 Nov 2009 19:57:26 +0000 (19:57 +0000)]
Branch b1_8
b=21224
i=girish
i=panda
Export ext4_get_group_desc() instead of reimplementing it
in fsfilt.
Speed up quotacheck by using find_next_bit() and stopping
scanning once we have know that no more inodes in-use remain.
dzogin [Tue, 10 Nov 2009 15:22:49 +0000 (15:22 +0000)]
Branch b1_8
b=21049
i=adilger
i=robert.read
----------------------------------------------------------------------
Modified Files:
Tag: b1_8
lustre/ChangeLog lustre/obdfilter/filter.c
lustre/ost/ost_handler.c lustre/tests/sanity.sh
lustre/tests/test-framework.sh
----------------------------------------------------------------------
johann [Tue, 10 Nov 2009 14:17:17 +0000 (14:17 +0000)]
Branch b1_8
b=19023 (via 21250)
Add replay-vbr test 0x & 0y to ALWAYS_EXCEPT until bug 19023 is fixed.
nathan [Mon, 9 Nov 2009 23:20:09 +0000 (23:20 +0000)]
b=19919
i=manoj
i=adilger
Make sure there is a valid pool when creating a new file with llapi
Fixes for 19406
Various pools test cleanups and fixes
b=19027
i=brian
i=rread
Don't be so anal about using _UUID in user interface; e.g. accept
lustre-OST0000, lustre-OST0000-osc-
4ba41334 as well as lustre-OST0000_UUID.
grev [Mon, 9 Nov 2009 12:09:10 +0000 (12:09 +0000)]
revert 21115 patch, it causes b=21308
grev [Fri, 6 Nov 2009 21:02:09 +0000 (21:02 +0000)]
b=21115
i=Robert.Read
t-f: mount the same config on all clients
grev [Fri, 6 Nov 2009 15:16:16 +0000 (15:16 +0000)]
b=21281
o=Walter (walter@routingdynamics.com)
i=Johann
i=grev
gcc-4.4.0 RHEL-5.3 x86_64 build error fix
grev [Thu, 5 Nov 2009 13:35:41 +0000 (13:35 +0000)]
b=19242
i=Robert.Read
additional mdcdev check
ericm [Wed, 4 Nov 2009 19:24:40 +0000 (19:24 +0000)]
branch: b1_8
properly prepare fake request.
b=20346
r=wangdi
r=fanyong
jxiong [Wed, 4 Nov 2009 08:31:51 +0000 (08:31 +0000)]
b=19211
Port the patch from HEAD.
green [Wed, 4 Nov 2009 04:16:16 +0000 (04:16 +0000)]
b=19128
r=adilger,tappro,rread
Properly add a request to the start of a replay list if there are no
requests with smaller tid in the list already.
johann [Tue, 3 Nov 2009 22:54:13 +0000 (22:54 +0000)]
Branch b1_8
b=21244
include sys/xattr.h instead of attr/xattr.h to fix build problems.
johann [Tue, 3 Nov 2009 21:59:21 +0000 (21:59 +0000)]
Branch b1_8
b=21244
add missing file.
johann [Tue, 3 Nov 2009 13:38:53 +0000 (13:38 +0000)]
Branch b1_8
b=20978
i=nathan
i=panda
Replace the global MGC lock with a per-config_llog_data semaphore.
yangsheng [Tue, 3 Nov 2009 08:50:42 +0000 (08:50 +0000)]
Branch b1_8
b=21244
i=girish, tianzy
Decode filter_fid xattr from objects.
yangsheng [Tue, 3 Nov 2009 08:34:30 +0000 (08:34 +0000)]
Branch b1_8
b=21244
i=johann, yangsheng
Move struct lu_client_seq to obd.h.
Author: adilger
manoj [Mon, 2 Nov 2009 20:27:56 +0000 (20:27 +0000)]
b=20975
r=robert.read
r=nathan.rutman
Change forced stop of MDS to a normal stop.
vitaly [Mon, 2 Nov 2009 13:49:24 +0000 (13:49 +0000)]
Branch b1_8
b=21168
i=jian.yu
ensure all the OSTs are reconnected after client's remount
panda [Mon, 2 Nov 2009 12:38:55 +0000 (12:38 +0000)]
b=18948
i=Mike Pershin
i=Andrew Perepechko
fix the compile warning
panda [Mon, 2 Nov 2009 12:33:53 +0000 (12:33 +0000)]
b=21150
a=Johann Lombardi
i=ZhiYong Tian
i=Andrew Perepechko
use lovobjid value from mds obd
panda [Mon, 2 Nov 2009 12:28:03 +0000 (12:28 +0000)]
b=21166
i=Oleg
i=Johann
reduce busy-waits by explicit set_task_state(...), cleanup
panda [Mon, 2 Nov 2009 12:23:32 +0000 (12:23 +0000)]
b=18801
i=Alexey Lyashkov
i=Alexander Zarochentsev
handle_short_read should use correct page shifts
ericm [Fri, 23 Oct 2009 22:46:27 +0000 (22:46 +0000)]
branch: b1_8
fix ost bulk read/write busy waiting.
b=20444
r=rread
r=wangdi
grev [Thu, 22 Oct 2009 15:20:20 +0000 (15:20 +0000)]
b=20308
i=Robert.Read
statahead test: N mount points on all available clients
ericm [Thu, 22 Oct 2009 14:57:19 +0000 (14:57 +0000)]
branch: b1_8
wait before test file size.
b=20801
r=rread
zhanghc [Thu, 22 Oct 2009 14:44:42 +0000 (14:44 +0000)]
b=16774
at client side, cancel unused locks before replay locks during recovery
i=johann@sun.com
i=oleg.drokin@sun.com
minh.diep [Wed, 21 Oct 2009 18:16:02 +0000 (18:16 +0000)]
b=20734
o=Robert Read
enable striping for ior
grev [Wed, 21 Oct 2009 15:15:36 +0000 (15:15 +0000)]
b=21142
i=Andrew.Perepechko
skip cascading_rw for NFSCLIENT mode
bobijam [Wed, 21 Oct 2009 09:52:22 +0000 (09:52 +0000)]
Branch b1_8
b=20592
i=johann
i=andrew.perepechko (panda)
use kthread_run() instead of kernel_thread() so that log dumper won't stuck
in do_exit().
grev [Wed, 21 Oct 2009 09:30:46 +0000 (09:30 +0000)]
b=20863
i=Brian
sanity test_200* cleanup
johann [Wed, 21 Oct 2009 09:20:28 +0000 (09:20 +0000)]
Branch b1_8
b=20318
i=adilger
i=panda
don't memcmp magic in the updgrade/downgrade case when lsm magic
has change from/to v1 to/from v3.
panda [Wed, 21 Oct 2009 09:13:30 +0000 (09:13 +0000)]
b=16890
i=Johann Lombardi
i=ZhiYong Tian
update 14b to a real-world test case
panda [Tue, 20 Oct 2009 13:39:19 +0000 (13:39 +0000)]
b=18948
a=Mike Pershin
i=Nathan Rutman
i=Zhanghc
use request refcount instead of copying
isaac [Mon, 19 Oct 2009 18:51:35 +0000 (18:51 +0000)]
b=20775,i=wangdi:
- simple cleanup: use LASSERT_SPIN_LOCKED instead.
zam [Sat, 17 Oct 2009 17:35:15 +0000 (17:35 +0000)]
Branch b1_8
b=20136
i=johann
i=alexey.lyashkov
mds_sendpage: exit l_wait_event and abort bulk on reconnect.
grev [Fri, 16 Oct 2009 21:15:28 +0000 (21:15 +0000)]
b=20308
i=Alexey.Lyashkov
statahead test
grev [Fri, 16 Oct 2009 13:53:11 +0000 (13:53 +0000)]
b=20719
i=Alexey.Lyashkov
i=Robert.Read
module_loaded fn
panda [Fri, 16 Oct 2009 13:47:55 +0000 (13:47 +0000)]
a small style fix
panda [Fri, 16 Oct 2009 12:48:49 +0000 (12:48 +0000)]
b=16466
i=Johann Lombardi
i=ZhiYong Tian
make admin quota file updates atomic (journalled)
panda [Fri, 16 Oct 2009 12:19:32 +0000 (12:19 +0000)]
b=20201
o=Oleg Drokin
i=Andrew Perepechko
i=Alexey Lyashkov
Problem is write_sem taking. As it was before the patch, a deadlock was possible
along the lines of:
process 1:
get group lock
process 2:
write syscall
get write_sem
attempt to get write lock, block on group lock
process 1:
write syscall
block on write_sem
tests 3,4 in parallel group lock exercise this scenario.
The patch actually avoids taking write_sem if we do not do any dlm locking as pointless, main
reason
behind it was to avoid cascading timeouts with multiple ldlm locks held (introduced in bug 9332,
comment #139)
--- https://bugzilla.lustre.org/show_bug.cgi?id=20201#c7
tianzy [Fri, 16 Oct 2009 07:08:37 +0000 (07:08 +0000)]
Branch b1_8
In quota slaves, quota_chk_acq_common() shouldn't retry indefinitely if
relative clients are evicted or reconnect already. This patch fixes
this.
b=20530
i=johann
i=panda
bobijam [Fri, 16 Oct 2009 02:10:59 +0000 (02:10 +0000)]
Branch b1_8
b=19325
i=vitaly.fertman
i=oleg.drokin (green)
Description: Adjust locks' extents on their first enqueue, so that at the time
they get granted, there is no need for another pass through the
queues since they are already shaped into the proper forms.
grev [Thu, 15 Oct 2009 19:00:17 +0000 (19:00 +0000)]
b=15504
i=Brian
get rid of dash in var name
isaac [Thu, 15 Oct 2009 18:57:03 +0000 (18:57 +0000)]
b=20853,i=isaac:
- Mark portal 63 as reserved for Cray DVS.
brian [Thu, 15 Oct 2009 18:10:50 +0000 (18:10 +0000)]
This patch was obviously inspected and approved before our -Werror
requirements.
Small fix thanks to panda.
panda [Thu, 15 Oct 2009 16:06:55 +0000 (16:06 +0000)]
b=20585
i=Johann Lombardi
at_max changes take too long to take effect, enforce them
dzogin [Thu, 15 Oct 2009 15:16:08 +0000 (15:16 +0000)]
Branch b1_8
b=20200
i=alexey.lyashkov
i=adilger
i=dmitry.zogin
----------------------------------------------------------------------
Modified Files:
Tag: b1_8
lustre/ChangeLog lustre/mds/handler.c lustre/mds/mds_reint.c
----------------------------------------------------------------------
Description: Fix mds_shrink_intent_reply()/mds_intent_policy() to pass correct
arguments and prevent LBUG() in lustre_shrink_reply_v2().
brian [Thu, 15 Oct 2009 13:21:48 +0000 (13:21 +0000)]
b=19689
i=adilger
i=bobijam
i=nathan
o=Jim Garlick
Change tunefs.lustre and mkfs.lustre --mountfsoptions so that
exactly the specified mount options are used.
Leaving off any "mandatory" mount options is an error.
Leaving off any default mount options causes a warning, but is allowed.
Change errors=remount-ro from mandatory to default.
Sanitize the mount string before storing it.
Update man pages accordingly.
This hopefully makes two awkward situations less so:
- making errors=panic the default (before we had to append errors=remount-ro
and hope ldiskfs parsing caused the last option to override the first)
- setting other mount options on the OST's dropped mballoc,extents
without warning.
johann [Thu, 15 Oct 2009 13:15:18 +0000 (13:15 +0000)]
Branch b1_8
b=19763
fix build issue with ext3-based ldiskfs which uses
EXT4_FEATURE_RO_COMPAT_GDT_CSUM instead of
EXT3_FEATURE_RO_COMPAT_GDT_CSUM.
johann [Thu, 15 Oct 2009 12:40:25 +0000 (12:40 +0000)]
Branch b1_8
b=16890
i=adilger
i=tianzy
i=panda
CERROR cleanups in the quota code.
tianzy [Thu, 15 Oct 2009 08:25:01 +0000 (08:25 +0000)]
Branch b1_8
clean the code relative to lqs_refcount
b=18782
i=johann
i=panda
shadow [Thu, 15 Oct 2009 02:54:14 +0000 (02:54 +0000)]
lost part of patch.
Branch b1_8
b=19557
shadow [Thu, 15 Oct 2009 01:57:58 +0000 (01:57 +0000)]
simplify client export disconnect on server side.
Branch b1_8
b=19557
i=tappro
i=shadow
grev [Wed, 14 Oct 2009 22:03:34 +0000 (22:03 +0000)]
b=20237
i=Robert.Read
more tests gather logs changes
brian [Wed, 14 Oct 2009 20:58:10 +0000 (20:58 +0000)]
Add missing changelog entry.
johann [Wed, 14 Oct 2009 20:48:45 +0000 (20:48 +0000)]
Branch b1_8
b=18630
fix build issue when quotas are disabled.
dzogin [Wed, 14 Oct 2009 17:36:56 +0000 (17:36 +0000)]
Branch b1_8
b=20302
----------------------------------------------------------------------
Modified Files:
Tag: b1_8
lustre/ptlrpc/client.c
----------------------------------------------------------------------
Description : Change debug message in ptlrpc_expire_one_request() to D_INFO,
if rq_fake is set.
dzogin [Wed, 14 Oct 2009 16:03:14 +0000 (16:03 +0000)]
Branch b1_8
b=20302
i=andrew.perepechko
i=adilger
----------------------------------------------------------------------
Modified Files:
Tag: b1_8
lustre/ChangeLog lustre/mds/handler.c lustre/ptlrpc/client.c
----------------------------------------------------------------------
Description: mds_getattr() should return 0, even if mds_fid2entry() fails with
-ENOENT. Also fix in ptlrpc_expire_one_request() to print signed
time difference.
johann [Wed, 14 Oct 2009 11:44:05 +0000 (11:44 +0000)]
Branch b1_8
Changelog update:
1) update 1.8.1.1 section;
2) remove duplicated changelog entries.
johann [Wed, 14 Oct 2009 11:39:09 +0000 (11:39 +0000)]
Branch b1_8
update to build version 1.8.1.52.
johann [Wed, 14 Oct 2009 09:18:57 +0000 (09:18 +0000)]
Branch b1_8
b=20816
Revert patch from bug 13520 because of 20816.
zhanghc [Wed, 14 Oct 2009 09:00:47 +0000 (09:00 +0000)]
b=20584
set "IT_CREAT" in the default intent(oit) in function
ll_file_open(llite/file.c) if "it_flags" contains O_CREAT
i=johann@sun.com
i=alexey.lyashkov@sun.com
johann [Wed, 14 Oct 2009 08:39:29 +0000 (08:39 +0000)]
Branch b1_8
b=18630
i=johann
Don't LBUG on release quota error. Just a workaround until the problem
is understood (and it is now).
johann [Wed, 14 Oct 2009 08:37:44 +0000 (08:37 +0000)]
Branch b1_8
b=21044
i=shadow
i=bobijam
Change LASSERT from bug 19934 to CWARN.
bobijam [Wed, 14 Oct 2009 08:30:11 +0000 (08:30 +0000)]
Branch b1_8
b=20149
i=johann
add more debug info.
johann [Wed, 14 Oct 2009 00:03:56 +0000 (00:03 +0000)]
Branch b1_8
b=19763
remove EXT3_FEATURE_RO_COMPAT_GDT_CSUM check because it causes
build failures. Will get this fixed asap.
johann [Tue, 13 Oct 2009 23:15:05 +0000 (23:15 +0000)]
Branch b1_8
b=19763
fix compile issue on non ext4-based ldiskfs.
johann [Tue, 13 Oct 2009 21:02:02 +0000 (21:02 +0000)]
Branch b1_8
b=19763
i=adilger
i=girish
speed up quotacheck. take into account itable_unused_count.
vitaly [Mon, 12 Oct 2009 19:35:38 +0000 (19:35 +0000)]
Branch b1_8
b=18629
i=rread
reconnect the client back
bobijam [Mon, 12 Oct 2009 13:37:12 +0000 (13:37 +0000)]
Branch b1_8
b=19720
i=joahnn
i=andrew.perepechko (panda)
Fix some compile warnings.
adilger [Fri, 9 Oct 2009 19:19:57 +0000 (19:19 +0000)]
Branch b1_8
Update to build version 1.8.1.51.
grev [Fri, 9 Oct 2009 16:03:57 +0000 (16:03 +0000)]
b=18150
i=Johann
tets_22: correct check of mount client without ost
panda [Wed, 7 Oct 2009 17:47:24 +0000 (17:47 +0000)]
b=19113
o=Johann Lombardi
i=Nathan Rutman
i=Andrew Perepechko
fix for ptlrpc stopping race
johann [Wed, 7 Oct 2009 12:31:13 +0000 (12:31 +0000)]
Branch b1_8
b=19345
i=shadow
i=wangdi
use shrink_dcache_sb() instead of shrink_dcache_parent().
grev [Tue, 6 Oct 2009 11:27:56 +0000 (11:27 +0000)]
b=20698
i=Robert.Read
new base for new test_[a-z] tests
johann [Tue, 6 Oct 2009 08:22:21 +0000 (08:22 +0000)]
Branch b1_8
b=20903
i=rread
i=dmitry
Make obd_zombie_barrier() no-op if called from the zombie thread context.
panda [Mon, 5 Oct 2009 15:39:55 +0000 (15:39 +0000)]
b=20575
o=Tom Wang
i=Johann Lombardi
i=Andrew Perepechko
avoid overflow in ll_readahead
dzogin [Fri, 2 Oct 2009 19:14:34 +0000 (19:14 +0000)]
Branch b1_8
b=20902
i=grev
----------------------------------------------------------------------
Modified Files:
Tag: b1_8
lustre/tests/conf-sanity.sh
----------------------------------------------------------------------
Details: Conf-sanity.sh tests cleanup
anserper [Fri, 2 Oct 2009 16:21:45 +0000 (16:21 +0000)]
b=20953
i=Elena Gryaznova
additional debug for test 30
anserper [Fri, 2 Oct 2009 16:18:44 +0000 (16:18 +0000)]
b=16890
i=Johann Lombardi
i=ZhiYong Tian
i=Elena Gryaznova
get rid of noisy config messages