Whamcloud - gitweb
fs/lustre-release.git
14 years agob=20101 lfs getstripe -d test for sanity 27w
Andrew Perepechko [Wed, 17 Mar 2010 07:00:49 +0000 (10:00 +0300)]
b=20101 lfs getstripe -d test for sanity 27w

o=James Simmons (ORNL)
i=Andreas Dilger
i=Robert Read

14 years agob=20878 update for commit 5b268907
Nathan Rutman [Fri, 12 Mar 2010 22:25:53 +0000 (14:25 -0800)]
b=20878 update for commit 5b268907

14 years agob=20878 change kernelcomms from netlink to pipes
Nathan Rutman [Fri, 12 Mar 2010 17:46:01 +0000 (09:46 -0800)]
b=20878 change kernelcomms from netlink to pipes

i=alexg
i=emoly.liu
for portability to Solaris, change kernel-userspace comm
mechanism from NetLink to pipes

14 years agob=16721 fix rename pdif lock, when source and target in the same directory.
Wang Di [Fri, 12 Mar 2010 17:46:01 +0000 (09:46 -0800)]
b=16721 fix rename pdif lock, when source and target in the same directory.

i=tappro
i=ericm

14 years agob=22097 mount.lustre fails to pass some options to mount()
Robert Read [Thu, 11 Mar 2010 05:49:04 +0000 (21:49 -0800)]
b=22097 mount.lustre fails to pass some options to mount()

i=yangsheng
i=dmitry

14 years agob=17952 allow relative pathnames
Brian J. Murrell [Tue, 9 Mar 2010 19:26:11 +0000 (11:26 -0800)]
b=17952 allow relative pathnames

This patch allows one to give relative pathnames to --with-linux and
friends.

i=adilger
i=dmitry.zogin

14 years agob=21669 cast the le64 to %llu in all ext*mmp patches
Andrew Perepechko [Tue, 9 Mar 2010 19:26:11 +0000 (11:26 -0800)]
b=21669 cast the le64 to %llu in all ext*mmp patches

i=Andreas Dilger
i=Andrew Perepechko

a bunch of fixes for extN_warnings in ldiskfs to avoid compilation warnings

14 years agob=22169 t-f:start_client_loads () wait the background threads to start
Elena Gryaznova [Tue, 9 Mar 2010 19:26:11 +0000 (11:26 -0800)]
b=22169 t-f:start_client_loads () wait the background threads to start

i=Robert.Read

14 years agob=22169 t-f cleanup: new do_nodev (), do_nodesv () functions
Elena Gryaznova [Tue, 9 Mar 2010 19:26:11 +0000 (11:26 -0800)]
b=22169 t-f cleanup: new do_nodev (), do_nodesv () functions

i=Robert.Read

14 years agob=22030 don't reveal unnecessary message to users
Landen [Tue, 9 Mar 2010 19:26:11 +0000 (11:26 -0800)]
b=22030 don't reveal unnecessary message to users

i=adilger
i=robert.read

14 years agob=21986 allow to apply OPEN_LOCK even when file created
Fan Yong [Tue, 9 Mar 2010 19:26:10 +0000 (11:26 -0800)]
b=21986 allow to apply OPEN_LOCK even when file created

Allow to apply OPEN_LOCK even when file created.

i=robert.read
i=oleg.drokin

14 years agob=21900 ost-pools test_25: FAIL
Dmitry Zogin [Tue, 9 Mar 2010 19:26:10 +0000 (11:26 -0800)]
b=21900 ost-pools test_25: FAIL

Make ost-pools test_25 more robust

 i=manoj.joseph

14 years agob=22127 lustre 1.8.2 lfs permissions Patch corrects cfs_curproc_euid() logic.
Dmitry Zogin [Tue, 9 Mar 2010 19:26:10 +0000 (11:26 -0800)]
b=22127 lustre 1.8.2 lfs permissions Patch corrects cfs_curproc_euid() logic.

o=bschubert@ddn.com
 i=oleg.drokin
 i=johann

14 years agob=20837 Reduce the output of test_14
Manoj Joseph [Tue, 9 Mar 2010 19:26:10 +0000 (11:26 -0800)]
b=20837 Reduce the output of test_14

Fill up the OST by createing 9 files of size OST0_SIZE/10.

i=nathan.rutman
i=grev

14 years agob=15962 statahead should not alias dentry with inode, which should be done by VFS...
Fan Yong [Tue, 9 Mar 2010 19:25:59 +0000 (11:25 -0800)]
b=15962 statahead should not alias dentry with inode, which should be done by VFS layer operation with parent dir's i_mutex held

1) statahead should not alias dentry with inode, which should be done by VFS layer operation with parent dir's i_mutex held
2) code cleanup

i=tom.wang
i=jinshan.xiong

14 years agoUpdated version string for build 38. 1.10.0.38 v1_10_0_38
Terry Rutledge [Thu, 4 Mar 2010 20:42:31 +0000 (13:42 -0700)]
Updated version string for build 38.

14 years agob=20383 fix errant m4 "dnl" usage
Brian J. Murrell [Thu, 4 Mar 2010 16:58:05 +0000 (08:58 -0800)]
b=20383 fix errant m4 "dnl" usage

Some dnl() usage seems to have been causing some errors in the resulting
configure script.

o=tappro
i=brian
i=adilger

14 years agob=22074 block if grant is available or maybe available soon
Eric Mei [Thu, 4 Mar 2010 01:04:44 +0000 (17:04 -0800)]
b=22074 block if grant is available or maybe available soon

even if there's no optimal-sized RPC can be sent, don't fall back to
synchronous write if there's still grant available.

r=adilger
r=wangdi

14 years agob=21106 retry page_gang_lookup.
Eric Mei [Thu, 4 Mar 2010 01:04:36 +0000 (17:04 -0800)]
b=21106 retry page_gang_lookup.

If client cached a huge number of page, cl_lock_page_out() may hog CPU
for too long time without releasing, especially with high debug grade.
This patch break this process with voluntary resched.

r=jay
r=wangdi

14 years agob=21500 2.6.31-fc12 patchless client support
Rahul Deshmukh [Thu, 4 Mar 2010 01:03:31 +0000 (17:03 -0800)]
b=21500 2.6.31-fc12 patchless client support

i=adilger
i=brian
i=sheng.yang

14 years agob=20492 allow llverdev/llverfs to handle media errors better
Rahul Deshmukh [Thu, 4 Mar 2010 01:03:21 +0000 (17:03 -0800)]
b=20492 allow llverdev/llverfs to handle media errors better

i=adilger
i=rread

14 years agob=20738 FC11 support in Yala is needed
Rahul Deshmukh [Thu, 4 Mar 2010 01:03:09 +0000 (17:03 -0800)]
b=20738 FC11 support in Yala is needed

Added support for fc11 patchless client build on lbat

i=brain

14 years agob=21255 parallel-scale statahead test fix
Elena Gryaznova [Thu, 4 Mar 2010 01:02:49 +0000 (17:02 -0800)]
b=21255 parallel-scale statahead test fix

use mpi for create/delete files instead of createmany and rm

i=Vladimir.Saveliev
i=Andrew.Perepechko

14 years agob=21919 Do not resend quota_ctl requests
Andrew Perepechko [Thu, 4 Mar 2010 01:02:11 +0000 (17:02 -0800)]
b=21919 Do not resend quota_ctl requests

Do not resend quota_ctl requests infinitely.

i=Johann Lombardi
i=ZhiYong Tian

14 years agob=21066 ost-pools test_14 should not assert that files are from a specific OST
Manoj Joseph [Thu, 4 Mar 2010 01:01:30 +0000 (17:01 -0800)]
b=21066 ost-pools test_14 should not assert that files are from a specific OST

Round-robin allocation test should not assert that files are allocated
in strict round-robin fashion.

i=nathan.rutman
i=grev

14 years agob=22225 Pad all test descriptions with "=" characters.
Robert Read [Wed, 24 Feb 2010 03:02:00 +0000 (19:02 -0800)]
b=22225 Pad all test descriptions with "=" characters.

Add banner() function so we can log other messages in the same format.

i=nico

14 years agob=21380 make dist seems to exclude the "darwin" bits
Brian J. Murrell [Wed, 3 Mar 2010 01:32:48 +0000 (17:32 -0800)]
b=21380 make dist seems to exclude the "darwin" bits

Include all of the darwin bits in the distribution tarball created with
make dist.

i=adilger
i=rread

14 years agob=21485 Keep in-memory client data until export destroy
Mike Tappro [Wed, 3 Mar 2010 01:32:45 +0000 (17:32 -0800)]
b=21485 Keep in-memory client data until export destroy

Client data on disk is removed upon disconnect but in-memory data is needed
while export is still used and must be kept until its destroy.

i=adilger
i=rread

14 years agoRevert "b=21485 Keep in-memory client data until export destroy"
Robert Read [Tue, 2 Mar 2010 02:08:23 +0000 (18:08 -0800)]
Revert "b=21485 Keep in-memory client data until export destroy"

This reverts commit 64d7c036a5429a9ce5435aea96e30935b1cef44a.

Causes compile error:
.../lustre/include/lu_target.h:59: error: 'LR_CLIENT_BITMAP_SIZE' undeclared here (not in a function)

14 years agob=21485 Keep in-memory client data until export destroy
Mike Tappro [Tue, 2 Mar 2010 01:57:44 +0000 (17:57 -0800)]
b=21485 Keep in-memory client data until export destroy

Client data on disk is removed upon disconnect but in-memory data is needed
while export is still used and must be kept until its destroy.

i=adilger
i=rread

14 years agob=21988 serialize mdt and obd state bit access
Alexander.Zarochentsev [Tue, 2 Mar 2010 01:57:39 +0000 (17:57 -0800)]
b=21988 serialize mdt and obd state bit access

Use atomic bit functions to access mdt state in mdt_allow_cli(). Otherwise bit
updates can be lost in case of concurrent modifications. Protect obd_no_conn =
0 by the obd_task_processing_lock spinlock as in the rest of the code
excluding setup.

i=tappro
i=robert.read

14 years agob=21948 skip parallel grouplock test for NFSCLIENT mode
Elena Gryaznova [Tue, 2 Mar 2010 01:57:34 +0000 (17:57 -0800)]
b=21948 skip parallel grouplock test for NFSCLIENT mode

i=Johann.Lombardi

14 years agob=20814 enable frame pointers for rhel5 kernels
Vitaly Fertman [Tue, 2 Mar 2010 01:57:26 +0000 (17:57 -0800)]
b=20814 enable frame pointers for rhel5 kernels

re-land the patch, it was occasionally dropped

o=rread
i=vitaly
i=eric

14 years agob=20731 add handling of statfs ioctl in mdc.
Eric Mei [Tue, 2 Mar 2010 01:57:23 +0000 (17:57 -0800)]
b=20731 add handling of statfs ioctl in mdc.

when cluster is upgraded from 1.8, client don't have LMV up and running,
so statfs ioctl will have to handled by mdc directly.

r=adilger
r=pravin

14 years agob=21968 conf-sanity changes for config mgs and mds are not combined
Elena Gryaznova [Thu, 25 Feb 2010 08:33:27 +0000 (00:33 -0800)]
b=21968 conf-sanity changes for config mgs and mds are not combined

i=Nathan.Rutman

includes t-f:combined_mgs_mds ()

14 years agob=17660 remove unnecessary spinlock wrappings
Alexander.Zarochentev [Thu, 25 Feb 2010 08:32:58 +0000 (00:32 -0800)]
b=17660 remove unnecessary spinlock wrappings

Read access to obd_recovering is atomic and doesn't need to be protected
by a spinlock.

i=tappro
i=adilger

14 years agob=16680 diagnostic patch to verify lustre read-only device mechanism
Fan Yong [Thu, 25 Feb 2010 08:32:12 +0000 (00:32 -0800)]
b=16680 diagnostic patch to verify lustre read-only device mechanism

Diagnostic patch to check whether lustre read-only device mechanism works well or not.

i=alex.zhuravlev
i=rahul.deshmukh

14 years agob=21097 quote the passing on of it's own arguments through $@
Jian Yu [Thu, 25 Feb 2010 08:31:58 +0000 (00:31 -0800)]
b=21097 quote the passing on of it's own arguments through $@

Fix do_nodes() in acc-sm test-framework to quote the passing on
of it's own arguments through $@ to do_node().

i=jian.yu
i=edg

14 years agoRevert "b=20989 Don't search for inode in nfs get_dentry"
Robert Read [Tue, 23 Feb 2010 00:00:43 +0000 (16:00 -0800)]
Revert "b=20989 Don't search for inode in nfs get_dentry"

This reverts commit 06889c56aec37cead923dbbd636d31dba0a46950.

This caused a performance regression on NFS.

14 years agob=20984 cleanup md_op_data and add getstripe -M
Wang Di [Mon, 22 Feb 2010 22:21:32 +0000 (14:21 -0800)]
b=20984 cleanup md_op_data and add getstripe -M

i=adiger
i=robert
i=vitaly

14 years agob=21932 disable some tests on NFSCLIENT
Elena Gryaznova [Mon, 22 Feb 2010 22:21:27 +0000 (14:21 -0800)]
b=21932 disable some tests on NFSCLIENT

Disable simul,  write_disjoint, write_append_truncate for NFSCLIENT mode.
Setstripe on lustre client (nfsserver) instead of nfs client for ior test.

i=Oleg.Drokin

14 years agob=21853 define cfs_*_nested for CONFIG_DEBUG_LOCK_ALLOC=y
Maxim Patlasov [Mon, 22 Feb 2010 22:20:54 +0000 (14:20 -0800)]
b=21853 define cfs_*_nested for CONFIG_DEBUG_LOCK_ALLOC=y

Post-nn fix: define cfs_mutex_lock_nested and other libcfs nested lock prims as linux native nested prims when CONFIG_DEBUG_LOCK_ALLOC is defined.

i=rread
i=alex

14 years agob=21911 fix for double release of ibc_lock in o2iblnd
Maxim Patlasov [Mon, 22 Feb 2010 22:20:43 +0000 (14:20 -0800)]
b=21911 fix for double release of ibc_lock in o2iblnd

Re-acquire ibc_lock in kiblnd_post_tx_locked(). Add extra reference to conn before calling kiblnd_post_tx_locked() to avoid scenario when conn disappears inside kiblnd_post_tx_locked().

i=liang

14 years agob=21835 fix mdsrate script.
Eric Mei [Mon, 22 Feb 2010 22:20:21 +0000 (14:20 -0800)]
b=21835 fix mdsrate script.

Fix mdsrate test script to also account in inodes numbers used by
test directories.

r=robert.read

14 years agoUpdated version for buld 37. 1.10.0.37 v1_10_0_37
Terry Rutledge [Fri, 19 Feb 2010 17:30:48 +0000 (10:30 -0700)]
Updated version for buld 37.

14 years agoRevert "b=21501 flock in process hangs on eviction, does not fail"
Mikhail Pershin [Thu, 18 Feb 2010 16:26:54 +0000 (19:26 +0300)]
Revert "b=21501 flock in process hangs on eviction, does not fail"

This reverts commit 7e750a8d9f9b87b7a3a608454c194d2fe7b69704.

It produces new bug 22111.

14 years agob=21718 refer to osc_io only if lock was held.
Mikhail Pershin [Wed, 17 Feb 2010 08:07:47 +0000 (11:07 +0300)]
b=21718 refer to osc_io only if lock was held.

i=wangdi
i=ericm

14 years agob=20997 Reset imp_vbr_fail flag upon eviction
Mike Tappro [Wed, 17 Feb 2010 07:45:43 +0000 (10:45 +0300)]
b=20997 Reset imp_vbr_fail flag upon eviction

Flag was reset when recovery ends but recovery can be aborted and this flag
stands after reconnection and cause false eviction later

i=rread
i=zam

14 years agob=21501 flock in process hangs on eviction, does not fail
Oleg Drokin [Wed, 17 Feb 2010 07:01:03 +0000 (10:01 +0300)]
b=21501 flock in process hangs on eviction, does not fail

Wake up waiters when failed lock gets completion AST due to eviction.

i=adilger
i=bzzz

14 years agob=21982 long file truncate problem
Vladimir Saveliev [Tue, 16 Feb 2010 16:30:28 +0000 (19:30 +0300)]
b=21982 long file truncate problem

fix (use loff_t for truncate size instead of size_t) and unit test

i=jinshan.xiong
i=adilger

14 years agob=21147 several fixes and enhancements for quota
Andrew Perepechko [Sat, 13 Feb 2010 00:27:14 +0000 (03:27 +0300)]
b=21147 several fixes and enhancements for quota

1) allow consecutive quotaons
2) a locking fix
3) call build_lqs from a single point in the code

i=Johann Lombardi
i=ZhiYong Tian

14 years agob=16195 handle request size correctly in ptlrpc_add_rqs_to_pool()
Vitaly Fertman [Fri, 12 Feb 2010 15:45:05 +0000 (18:45 +0300)]
b=16195 handle request size correctly in ptlrpc_add_rqs_to_pool()

i=adilger
i=johann

14 years agob=21881 Fix trap in rpc.sh.
Robert Read [Fri, 12 Feb 2010 17:27:23 +0000 (09:27 -0800)]
b=21881  Fix trap in rpc.sh.

14 years agob=21881 mdt_num_threads tuning
Nicolas Williams [Thu, 11 Feb 2010 21:26:24 +0000 (15:26 -0600)]
b=21881 mdt_num_threads tuning

Fixes to mdt_handler.c so that it honors mdt_num_threads in the same manner as
ost_handler.c does for oss_num_threads.  Enhancements to the test framework and
new tests (53a and b, to check for thread tunable behavior) for conf-sanity are
included.

i=adilger@sun.com
i=robert.read@sun.com

14 years agob=19069 save nid string into buffer before creating dir
hongchao.zhang [Thu, 11 Feb 2010 20:41:06 +0000 (12:41 -0800)]
b=19069 save nid string into buffer before creating dir

in lprocfs_exp_setup, save the nid string gotten from
libcfs_nid2str into a temporary buffer before calling
lprocfs_register to create proc directory, then to
avoid the race on the internal nid string buffer.

i=johann@sun.com
i=zhen.liang@sun.com

14 years agob=1028 drop SOM dependency on quota
Vitaly Fertman [Thu, 11 Feb 2010 20:39:36 +0000 (12:39 -0800)]
b=1028 drop SOM dependency on quota

once SOM attributes are moved to EA, there is no need to disabled quota when SOM is enabled

i=fanyong
i=tappro

14 years agob=21882 handle SLV==1 on client side
Vitaly Fertman [Thu, 11 Feb 2010 20:39:33 +0000 (12:39 -0800)]
b=21882 handle SLV==1 on client side

Initialize ldlm pool SLV to 0 on client side to handle SLV==1 obtained from server correctly

i=reen
i=adilger

14 years agob=21882 lru resize SLV can get stuck
Vitaly Fertman [Thu, 11 Feb 2010 20:39:33 +0000 (12:39 -0800)]
b=21882 lru resize SLV can get stuck

calculate SLV with a greater precision to not lose small changes due to interger math truncation;
round up SLV only if the amount of granted locks less than the limit to not get stuck with this SLV

i-adilger
i=green

14 years agob=15977 analyse only consistent part of the log
Vitaly Fertman [Thu, 11 Feb 2010 20:39:25 +0000 (12:39 -0800)]
b=15977 analyse only consistent part of the log

due to debug log buffer overflow, old messages are lost;
as each CPU has its own buffers which overflows at different time,
it may happen that FREE is lost for still existing ALLOC;
make leak_finder to analyze the log since the oldest overflow time.

i=rread
i=green

14 years agob=20057 Add the timestamp back to the test headers.
Robert Read [Thu, 11 Feb 2010 05:07:17 +0000 (21:07 -0800)]
b=20057 Add the timestamp back to the test headers.

a=manoj
i=rread
i=grev

14 years agob=21411 Improvement for AT to handle timeout more accurate.
yangsheng [Tue, 9 Feb 2010 21:06:02 +0000 (05:06 +0800)]
b=21411 Improvement for AT to handle timeout more accurate.

i=nathan
i=tappro

14 years agob=16909 Simplify MDT/OST service start message
Landen [Tue, 9 Feb 2010 05:23:49 +0000 (13:23 +0800)]
b=16909 Simplify MDT/OST service start message

i=nathan
i=adilger

14 years agob=21900 ost-pools test_25: FAIL: /mnt/lustre/d0.ost-pools/d25/file1 not allocated...
Dmitry Zogin [Tue, 9 Feb 2010 02:56:16 +0000 (21:56 -0500)]
b=21900 ost-pools test_25: FAIL: /mnt/lustre/d0.ost-pools/d25/file1 not allocated from OSTs 0.

Modify ost-pools test_25 to wait for MDS-OST connection to re-establish.

 i=johann

14 years agob=21278 fix linkea definition and unaligned access.
Eric Mei [Mon, 8 Feb 2010 23:57:43 +0000 (16:57 -0700)]
b=21278 fix linkea definition and unaligned access.

link_ea_entry->lee_reclen should be defined as unsigned char instead
of char which might get sign expansion during conversion.
fix unaligned memory access to link_ea_entry->lee_parent_fid.

i=nathan
i=rahul

14 years agob=21278 fix a memory leak in mdd_links_get().
Eric Mei [Mon, 8 Feb 2010 23:57:23 +0000 (16:57 -0700)]
b=21278 fix a memory leak in mdd_links_get().

i=nathan
i=rahul

14 years agob=17197 stride read-ahead fix
Wang Di [Mon, 8 Feb 2010 23:52:21 +0000 (18:52 -0500)]
b=17197 stride read-ahead fix

14 years agob=16526 don't send raw inode flags on the wire.
Eric Mei [Mon, 8 Feb 2010 23:24:46 +0000 (16:24 -0700)]
b=16526 don't send raw inode flags on the wire.

inode flags should be convert to platform-neutral MDS_*_FL before
being sent on the wire. Also removed unused flags MDS_BFLAG_EXT_FLAGS.

i=adilger

14 years agob=21049 Fix a missed change from original patch.
Robert Read [Wed, 10 Feb 2010 18:47:19 +0000 (10:47 -0800)]
b=21049 Fix a missed change from original patch.

14 years agob=21486 Unlink truncated reply buffer
Liang Zhen [Wed, 10 Feb 2010 13:30:31 +0000 (05:30 -0800)]
b=21486 Unlink truncated reply buffer

i=eric.mei
i=eeb

14 years agob=21396 Don't hog cpu in connd.
Liang Zhen [Wed, 10 Feb 2010 13:30:10 +0000 (05:30 -0800)]
b=21396 Don't hog cpu in connd.

i=he.huang
i=maxim.patlasov

14 years agob=21564 Print mmp_check_interval in kmmpd
Girish Shilamkar [Fri, 5 Feb 2010 15:04:15 +0000 (20:34 +0530)]
b=21564 Print mmp_check_interval in kmmpd

The patch not only prints mmp_check_interval but also makes it possible
to abort mount operation in case it takes too long.

i=adilger

14 years agob=18030 deadlock fix
Alexander.Zarochentev [Fri, 5 Feb 2010 07:25:08 +0000 (10:25 +0300)]
b=18030 deadlock fix

Start the transaction earlier in llog_lvfs_destroy to get transaction start
and inode mutex lock nested properly.

i=zam
i=adilger

14 years agob=21738 conf-sanity 42 should set conf_param on the MGS
Nathan Rutman [Wed, 3 Feb 2010 20:33:30 +0000 (12:33 -0800)]
b=21738 conf-sanity 42 should set conf_param on the MGS

i=rread

14 years agob=21551 holding ->lli_size_sem to update inode size
Jay Xiong [Thu, 4 Feb 2010 04:17:05 +0000 (12:17 +0800)]
b=21551 holding ->lli_size_sem to update inode size

- Fixed i_size_write in vvp_io_commit_write, it has to be called under ->lli_size_sem
- clenaup the code.

i=wangdi
i=green

14 years agoPrepare for Build 36. 1.10.0.36 2.0.0_alpha7 v1_10_0_36
Robert Read [Thu, 4 Feb 2010 23:55:21 +0000 (15:55 -0800)]
Prepare for Build 36.

14 years agob=21924 create test dir before starting changelogs
Nathan Rutman [Wed, 3 Feb 2010 20:09:37 +0000 (12:09 -0800)]
b=21924 create test dir before starting changelogs

i=rread
A test-level directory (d0.lustre-rsync-test) was recently added to the $tdir path.
This directory is created before the tests are run.  But lustre-rsync requires the
source and target systems to be identical before the changelog is started; the
presence of this directory in the source, missing in the target, causes the
failure seen here.  This patch simply creates $tdir on the source and target
before each test is run, insuring a coherent starting state.

14 years agob=21558 improve test 21b COS=0 check
Alexander.Zarochentev [Wed, 3 Feb 2010 18:01:03 +0000 (21:01 +0300)]
b=21558 improve test 21b COS=0 check

In test 21b, when testing that the dependent renames are not replied if COS=0,
there is still a window when transactions may be written to disk before the
mds device is set R/O. To avoid such a rare test failure, the check is
repeated several times.

i=tappro
i=rober.read

14 years agob=21264 workaround dd bus error
Alexander.Zarochentev [Wed, 3 Feb 2010 18:01:02 +0000 (21:01 +0300)]
b=21264 workaround dd bus error

A buggy coreutils/gettext combination workaround.  Suppressing dd xfer
statistic makes dd do not call gnu gettext library and avoid crashing.

i=robert.read

14 years agob=21259 "lfs check" is only allowed for root.
Dmitry Zogin [Wed, 3 Feb 2010 14:41:27 +0000 (09:41 -0500)]
b=21259 "lfs check" is only allowed for root.

Code cleanup around obd_class_*() functions and sanity test for non-root lfs check

 i=adilger
 i=andrew.perepechko

14 years agob=21879 fix hr->hr_index overload
Alexander.Zarochentev [Wed, 3 Feb 2010 09:00:47 +0000 (12:00 +0300)]
b=21879 fix hr->hr_index overload

i=eric.barton
i=zhen.liang

14 years agob=21355 fix initiailzation of recovery thread
Jay Xiong [Wed, 3 Feb 2010 03:13:22 +0000 (11:13 +0800)]
b=21355 fix initiailzation of recovery thread

Recovery thread calls cfs_daemonize() to detach the thread context, this causes all recovery threads share a common fs_struct: init_fs. We should call cfs_daemonize_ctxt() instead.

i=wangdi
i=tappro

14 years agob=21632 Kernel update for RHEL5.4 2.6.18-164.11.1.el5.
yangsheng [Tue, 2 Feb 2010 12:40:24 +0000 (20:40 +0800)]
b=21632 Kernel update for RHEL5.4 2.6.18-164.11.1.el5.

i=johann

14 years agob=21433 trace the processing for "no quota" flag on client-side
Fan Yong [Tue, 2 Feb 2010 10:54:58 +0000 (18:54 +0800)]
b=21433 trace the processing for "no quota" flag on client-side

Trace the processing for "no quota" flag on client-side.

i=robert
i=landen

14 years agob=19742 fixed the fiemap functionality
Rahul Deshmukh [Tue, 2 Feb 2010 08:44:41 +0000 (14:14 +0530)]
b=19742 fixed the fiemap functionality

Fixed the fiemap functionality and enable sanity test_130*.

i=adilger
i=johann

14 years agob=21890 Remove bash -TE flags from acceptance-small.sh
Robert Read [Wed, 3 Feb 2010 08:15:15 +0000 (00:15 -0800)]
b=21890 Remove bash -TE flags from acceptance-small.sh

These flags is not needed since test-framework sets the error trap in all the scripts.

14 years agob=20989 Don't search for inode in nfs get_dentry
Oleg Drokin [Tue, 2 Feb 2010 05:20:23 +0000 (00:20 -0500)]
b=20989 Don't search for inode in nfs get_dentry

Getting inode separately without some sort of locking enabled a situation where
several threads tried to populate same inode with different lsm data at the same
time leading to lsm pointer change (and leak).

i=johann
i=bzzz

14 years agob=21499 check stale object before version checks
Mikhail Pershin [Mon, 1 Feb 2010 19:27:24 +0000 (11:27 -0800)]
b=21499 check stale object before version checks

i=wangdi
i=fanyong

14 years agob=21489 fix several write+utimes race conditions
Andrew Perepechko [Mon, 1 Feb 2010 19:27:18 +0000 (11:27 -0800)]
b=21489 fix several write+utimes race conditions

Disable mtime updates on write and serialize fmd_mactime_xid checks in order to
avoid certain write(2)+utimes(2) race conditions on OSS

i=Johann Lombardi
i=Vitaly Fertman

14 years agob=20970 need add an additional barrier for write_disjoint
Landen [Fri, 29 Jan 2010 07:04:54 +0000 (15:04 +0800)]
b=20970 need add an additional barrier for write_disjoint

i=grev
i=rread

14 years agob=21846 add a hs_key entry in lqs hash operations
Landen [Fri, 29 Jan 2010 07:04:53 +0000 (15:04 +0800)]
b=21846 add a hs_key entry in lqs hash operations

When lqs rehashes, it will call its hs_key method. If it isn't
defined, an assert will be triggered. The patch just add it
for lqs hash.

a=johann
i=landen
i=adilger

14 years agob=21882 lru-resize comments
Brian Behlendorf [Fri, 29 Jan 2010 01:11:46 +0000 (17:11 -0800)]
b=21882 lru-resize comments

Unrelated to the actual bug, this commit just cleanups up various
spelling mistakes in the comments.

14 years agob=19808 2.6.29-fc11 patchless client support changes.
Rahul Deshmukh [Thu, 28 Jan 2010 11:36:39 +0000 (17:06 +0530)]
b=19808 2.6.29-fc11 patchless client support changes.

Final changes (except build related changes) for adding FC11
patchless client support.

i=alexey.lyashkov
i=oleg.drokin

14 years agob=19813 add missing lprocfs_exp_cleanup calls
Alexander.Zarochentev [Wed, 27 Jan 2010 15:01:12 +0000 (18:01 +0300)]
b=19813 add missing lprocfs_exp_cleanup calls

add missing lprocfs_exp_cleanup calls to
error handling paths in {mdt,filter,mgs}_connect.

i=tappro
i=andrew.perepechko

14 years agob=18690 Enable rehashing and increase max hash table sizes.
Robert Read [Sat, 23 Jan 2010 00:29:27 +0000 (16:29 -0800)]
b=18690 Enable rehashing and increase max hash table sizes.

i=eric.mei

14 years agob=19387 integrate LST into acc-sm
Elena Gryaznova [Tue, 26 Jan 2010 14:10:59 +0000 (17:10 +0300)]
b=19387 integrate LST into acc-sm

new acc-sm test suite: lnet-selftest

i=Maxim.Patlasov
i=He.Huang

14 years agob=21490 1.8<->2.0 interop: in 2.0 server and 1.8 client env. read 1. 8 file hung
Nicolas Williams [Mon, 25 Jan 2010 22:09:10 +0000 (16:09 -0600)]
b=21490 1.8<->2.0 interop: in 2.0 server and 1.8 client env. read 1. 8 file hung

The fix for bug 18631 introduced an interop bug for 1.8.x clients with 2.0 OSTs.
The problem was that a zero-length RMF_RCS was included in the OST_BRW replies
for READs, which caused the RPC header to have eight more bytes for storing the
length of that zero-length buffer.  The fix is to split RQF_OST_BRW into two,
one for READs and one for WRITEs, with RMF_RCS present only in the reply for
OST_BRW_WRITE.

i=robert.read@sun.com
i=tom.wang@sun.com

14 years agob=21746 compare full filename when writeconf'ing
Nathan Rutman [Mon, 25 Jan 2010 20:01:21 +0000 (12:01 -0800)]
b=21746 compare full filename when writeconf'ing

When erasing MGS logs, we could confuse fsname e.g. test with test2 and erase both sets of logs.

i=breitz
i=brian

14 years agob=21889 Rename lustre-resync-test
Robert Read [Mon, 25 Jan 2010 19:36:28 +0000 (11:36 -0800)]
b=21889 Rename lustre-resync-test

Fix Makefile.

14 years agob=20918 improve log warning
Elena Gryaznova [Mon, 25 Jan 2010 11:42:11 +0000 (14:42 +0300)]
b=20918 improve log warning

i=Brian

14 years agob=20866 DEPS assignment needs quotes
Elena Gryaznova [Mon, 25 Jan 2010 11:28:36 +0000 (14:28 +0300)]
b=20866 DEPS assignment needs quotes

o=Brian
i=grev