Whamcloud - gitweb
vitaly [Thu, 29 May 2008 22:38:27 +0000 (22:38 +0000)]
Branch b1_6
b=15904
i=shadow
i=wangdi
clean log_commit_master properly at cleanup in llog_commit_thread()
shadow [Thu, 29 May 2008 05:55:02 +0000 (05:55 +0000)]
cleanup in ptlrpc code, related to ppc platform
avoid twice swabing one msg buffer
fix get LOV_GET_INFO on ppc platform.
b=14856
i=nikita
i=adilger
bobijam [Thu, 29 May 2008 02:35:27 +0000 (02:35 +0000)]
Branch b1_6
b=13310
o=eeb
i=joahnn, nikita
The bug is that target_send_reply_msg() calls ptlrpc_error() if 'rc != 0', but
"difficult" replies may not be passed to rpc_error().
vs [Thu, 29 May 2008 00:13:04 +0000 (00:13 +0000)]
b=15346
i=green,shadow,vitaly
skiplists are used to group compatible locks on granted list
that was implemented as tracking first and last lock of each lock group
the patch changes that to using doubly linked lists
grev [Wed, 28 May 2008 20:02:53 +0000 (20:02 +0000)]
b=15876
i=Adilger
skip LFSCK test for remote MDS also
grev [Wed, 28 May 2008 19:28:51 +0000 (19:28 +0000)]
b=14773
i=Adilger
rename subtests to skipping mechanism works properly
ericm [Wed, 28 May 2008 19:21:43 +0000 (19:21 +0000)]
branch: b1_6
a better way to fix the script.
b=13659
brian [Wed, 28 May 2008 19:14:21 +0000 (19:14 +0000)]
b=15908
i=johann
i=zhenyu.xu:
Small message correction.
ericm [Wed, 28 May 2008 18:33:44 +0000 (18:33 +0000)]
branch: b1_6
fix test script: at_start return correct value.
b=13659
r=grev
komaln [Wed, 28 May 2008 05:46:18 +0000 (05:46 +0000)]
b=14471
r=Andreas
to replace the sysctl lustre.* and lnet.* instances with lctl {get,set}_param
girish [Tue, 27 May 2008 17:52:15 +0000 (17:52 +0000)]
Upstream fix for memory leak when verifying checksums in the journal.
ericm [Mon, 26 May 2008 23:27:28 +0000 (23:27 +0000)]
branch: b1_6
do not drop replay according to msg flags, instead we check the
per-export recovery request queue for duplication of transno.
b=15756
r=adilger
r=rread
grev [Mon, 26 May 2008 14:17:31 +0000 (14:17 +0000)]
b=15255
i=Adilger
config for several remote clients; replay-single test_70a, test_70b
yangsheng [Mon, 26 May 2008 13:47:45 +0000 (13:47 +0000)]
Branch b1_6
b=15649
i=oleg, shadow
Enable debug options in vanilla kernel config file.
grev [Mon, 26 May 2008 10:52:48 +0000 (10:52 +0000)]
b=15874
i=Adilger
skip test_32a, test_32b for NETTYPE != tcp
anserper [Sun, 25 May 2008 19:56:47 +0000 (19:56 +0000)]
Branch b1_6
b=14443
anserper [Sun, 25 May 2008 19:42:45 +0000 (19:42 +0000)]
Branch b1_6
b=13915
adilger [Sun, 25 May 2008 05:43:55 +0000 (05:43 +0000)]
Branch b1_6
Update which_patch to indicate support for 2.6.22 patchless.
girish [Sat, 24 May 2008 10:57:26 +0000 (10:57 +0000)]
Update e2fsprogs version to 1.40.7-sun3
shadow [Sat, 24 May 2008 08:52:17 +0000 (08:52 +0000)]
fix liblustre build. use OBD_BRW macros instead of linux specific.
b=14010
i=green
bwzhou [Sat, 24 May 2008 07:51:17 +0000 (07:51 +0000)]
Branch b1_6
b=15759
r=rread, green
Don't consider -EMFILE (too many threads) a fatal startup error
robert.read [Sat, 24 May 2008 00:13:39 +0000 (00:13 +0000)]
Branch b1_6
b=15501
i=eric.mei
i=rread
Nathan's early margin debug patch.
shadow [Fri, 23 May 2008 15:51:57 +0000 (15:51 +0000)]
fix build lustre with o2ib and SLES10.
b=15870
i=walter
anserper [Fri, 23 May 2008 10:29:04 +0000 (10:29 +0000)]
Branch b1_6
b=14010
i=shadow
i=zam
Fast-blocking read case
shadow [Fri, 23 May 2008 06:03:26 +0000 (06:03 +0000)]
don't create too many objects after del orphan.
b=14835
i=tappro
i=deen
komaln [Thu, 22 May 2008 17:56:18 +0000 (17:56 +0000)]
b=14471
r=Andreas
to remove instances of /proc/sys in sanity.sh by using either sysctl or lctl {get,set}_param
deen [Thu, 22 May 2008 09:15:05 +0000 (09:15 +0000)]
In reconstruct_* functions, LASSERTs on both the data supplied by
a client, and the data on disk are dangerous and incorrect. Change
them with client eviction.
b=15574
i=adilger
i=shadow
robert.read [Wed, 21 May 2008 17:55:32 +0000 (17:55 +0000)]
Branch b1_6
b=13285
Reverting rest of script libexedir patch to fix rpm build.
tianzy [Wed, 21 May 2008 08:08:46 +0000 (08:08 +0000)]
add entries for quota patches.
adilger [Wed, 21 May 2008 04:44:12 +0000 (04:44 +0000)]
Branch b1_6
Quiet build warnings when building MPI test programs.
fanyong [Wed, 21 May 2008 02:29:36 +0000 (02:29 +0000)]
Branch b1_6
b=15406
i=adilger
i=h.huang
replacing "ext2_xxx" in llite module.
green [Tue, 20 May 2008 19:49:12 +0000 (19:49 +0000)]
b=15776
r=shadow,vitaly
Prevent locks stalls on glimpse races
bwzhou [Tue, 20 May 2008 15:07:35 +0000 (15:07 +0000)]
Branch b1_6
b=13285
r=green, bobijam
use the libexec convention for scripts
bwzhou [Tue, 20 May 2008 09:07:38 +0000 (09:07 +0000)]
Branch b1_6
b=13285
r=adilger, deen
1. updated ext3-mballoc3-core.patch and added ext3-max-dir-size.patch to limit the maximum size of a directory;
2. updated other ldiskfs patches to resolve patch conflicts caused by the above two new patches;
3. added a test case in sanity.sh.
komaln [Tue, 20 May 2008 08:13:23 +0000 (08:13 +0000)]
b=14471
r=andrew.perepechko, johann
Patch by Adilger to replace /proc/fs/lustre access with lctl set_param in sanity-quota.sh
shadow [Tue, 20 May 2008 06:45:57 +0000 (06:45 +0000)]
fix ASSERTION(client_stat->nid_exp_ref_count == 0) during
run acc-small.
b=15139
i=tappro
i=komal
komaln [Tue, 20 May 2008 06:39:02 +0000 (06:39 +0000)]
b=14471
r=Adilger
Changed replay-single.sh to use 'lctl {get,set}_param'
fanyong [Tue, 20 May 2008 02:05:15 +0000 (02:05 +0000)]
Branch b1_6
b=15828
i=adilger
i=umka
return "-EIO" when write to read only proc interface which "write_proc" is NULL.
bobijam [Tue, 20 May 2008 01:26:29 +0000 (01:26 +0000)]
Branch b1_6
b=15575
i=wangdi, johann, green, shadow
Description: Stack overflow during MDS log replay
Details : ease stack pressure by using a thread dealing llog_process.
bwzhou [Sun, 18 May 2008 18:43:45 +0000 (18:43 +0000)]
Branch b1_6
b=13285
revert the libexec patch since it hasn't got permissions yet.
bwzhou [Sun, 18 May 2008 18:24:55 +0000 (18:24 +0000)]
Branch b1_6
b=13285
r=green, bobijam (libexec)
r=johann, deen (gcc)
1. Use the common libexec convention;
2. Quiet remaining gcc warnings.
bwzhou [Sat, 17 May 2008 05:46:52 +0000 (05:46 +0000)]
Branch b1_6
b=13380
r=green, bwzhou
Mountpoint references were being leaked during open reply reconstruction after
an MDS restart. Drop mountpoint reference in reconstruct_open() and free dentry
reference also.
shadow [Fri, 16 May 2008 06:50:00 +0000 (06:50 +0000)]
VM protocol want old IO finished before start new, in this case
need wait until PG_writeback is cleared until check dirty flag and
call writepages in lock cancel callback.
b=15443
i=nikita
i=zam
walter [Fri, 16 May 2008 02:18:53 +0000 (02:18 +0000)]
b=15660
i=adilger
disable LIBLUSTRE test for NETTYPE != tcp && NETTYPE != ptl
adilger [Thu, 15 May 2008 16:33:22 +0000 (16:33 +0000)]
Branch b1_6
Ignore compiled file.
adilger [Wed, 14 May 2008 22:24:20 +0000 (22:24 +0000)]
Branch b1_6
Description: MDS may lose file striping (and hence file data) in some cases
Details : If there are additional extended attributes stored on the MDS,
in particular ACLs, SELinux, or user attributes (if user_xattr
is specified for the client mount options) then there is a risk
of attribute loss. Additionally, the Lustre file striping
needs to be larger than default (e.g. striped over all OSTs),
and an additional attribute must be stored initially in the
inode and then increase in size enough to be moved to the
external attribute block (e.g. ACL growing in size) for file
data to be lost.
i=kalpak.shah
i=alex.zhuravlev
b=15777
Sync ChangeLog from b_release_1_6_5.
deen [Wed, 14 May 2008 21:21:39 +0000 (21:21 +0000)]
Refactor start to encapsulate facet specific data.
b=12375
i=deen
i=grev
adilger [Wed, 14 May 2008 20:59:31 +0000 (20:59 +0000)]
Branch b1_6
Merge ChangeLog entries from b_release_1_6_5 to b1_6.
grev [Tue, 13 May 2008 13:20:02 +0000 (13:20 +0000)]
b=15738
i=Adilger
sanity test_118* fixes; set fail loc on all oss nodes
anserper [Tue, 13 May 2008 08:08:48 +0000 (08:08 +0000)]
Branch b1_6
b=15238
i=shadow
i=tianzy
refactoring of quota code
bwzhou [Tue, 13 May 2008 05:56:39 +0000 (05:56 +0000)]
Branch b1_6
b=14740
r=adilger, bobijam
break a single big chunk of memory allocation for lprocfs_stats into separately
allocated small chunks to prevent kmalloc failure in a huge cluster
bobijam [Tue, 13 May 2008 01:28:34 +0000 (01:28 +0000)]
Branch b1_6
b=12888
i=johann, green
Description: mds_mfd_close() ASSERTION(rc == 0)
Details : In mds_mfd_close(), we need protect inode's writecount change
within its orphan write semaphore to prevent possible races.
grev [Mon, 12 May 2008 20:14:44 +0000 (20:14 +0000)]
b=15745
i=Adilger
the minor cleanup of the previous fix:
-- after we get rid of $tdir we do not need to create $tmpdir, because
unzip -d <dir> does not need <dir> exists, it needs the parent directory
of <dir> exists; unzip creates <dir> if it does not exists.
anserper [Mon, 12 May 2008 12:18:48 +0000 (12:18 +0000)]
Branch b1_6
b=14790
i=Johann
i=ZhiYong
additional test. check if setting limits twice doesn't break oplimits.
kalpak [Mon, 12 May 2008 11:31:05 +0000 (11:31 +0000)]
Reinitialize optind to 0 on each lfs_* call, in case lfs is being used interactively. Most lfs_* functions do this already, but a few (find, setstripe, quotachown) do not.
b=15212
i=rupesh
i=johann
tianzy [Mon, 12 May 2008 05:15:36 +0000 (05:15 +0000)]
Branch b1_6
change the time of ost informing osc out of quota in order to get more
accurate quota
b=14909
i=johann, andrew
adilger [Mon, 12 May 2008 03:45:46 +0000 (03:45 +0000)]
Branch b1_6
Fix typo in error message.
bwzhou [Sun, 11 May 2008 10:39:11 +0000 (10:39 +0000)]
Branch b1_6
b=15521
r=adilger, grev
the test case for bug 15521
walter [Sat, 10 May 2008 15:24:50 +0000 (15:24 +0000)]
b=15732
i=walter
Land andreas' fix for catamount build errors
walter [Sat, 10 May 2008 15:10:59 +0000 (15:10 +0000)]
b=15732
i=adilger
fix build warnings which are treated as errors
during Cray XT3 catamount build
bwzhou [Fri, 9 May 2008 16:44:07 +0000 (16:44 +0000)]
Branch b1_6
b=14034
r=johann, bobijam
error handling for ENOMEM case when allocating a ldlm resource
shadow [Fri, 9 May 2008 10:19:15 +0000 (10:19 +0000)]
don't hit live lock with umount ost.
b=14645
i=johan
i=alex
grev [Fri, 9 May 2008 08:36:05 +0000 (08:36 +0000)]
b=15745
i=Adilger
conf-sanity test_32a, test_32b: create tmp dir correctly
grev [Fri, 9 May 2008 08:01:41 +0000 (08:01 +0000)]
b=15660
i=Adilger
disable LIBLUSTRE test for NETTYPE != tcp
shadow [Thu, 8 May 2008 11:54:07 +0000 (11:54 +0000)]
rollback bug 15575, because at any run acc-sm i hit
LustreError: 29518:0:(lvfs_linux.c:194:pop_ctxt()) ASSERTION(current->fs->pwd == new_ctx->pwd)
after run-tests.
shadow [Thu, 8 May 2008 10:18:09 +0000 (10:18 +0000)]
don't panic with use echo client
b=14949
i=johan
i=deen
zam [Thu, 8 May 2008 05:46:55 +0000 (05:46 +0000)]
b=11270
i=vitaly.vertman
i=oleg.drokin
Lockless i/o code fixes and improvements:
(1) lockless truncate checks for OBD_CONNECT_TRUNCLOCK and ll_file_punch
sets OBD_FL_TRUNCLOCK correctly.
(2) an lproc control for lockless truncate,
lproc statistics for lockless truncate.
(3) sanityN tests for lockless code
adilger [Wed, 7 May 2008 22:23:22 +0000 (22:23 +0000)]
Branch b_release_1_6_5
Update client side RPC messages to match those on the server, as changed
previously on the server in bug 13537, allowing the same :pid:xid: string
to be searched in both client and server logs.
b=13537
bobijam [Wed, 7 May 2008 01:50:50 +0000 (01:50 +0000)]
Branch b1_6
b=15575
i=wangdi, johann
Description: Stack overflow during MDS log replay
Details : ease stack pressure by using a thread dealing llog_process.
shadow [Tue, 6 May 2008 18:16:09 +0000 (18:16 +0000)]
compile with -m64 flag produce wrong object file for ppc32.
b=15278
i=johann
i=deen
bwzhou [Mon, 5 May 2008 15:15:28 +0000 (15:15 +0000)]
Branch b1_6
b=14219
r=adilger, shadow
use punch instead of setattr to change file size
johann [Fri, 2 May 2008 20:26:35 +0000 (20:26 +0000)]
Branch b1_6
b=15517
i=johann
i=umka
Fix compile warnings on i686.
johann [Fri, 2 May 2008 19:41:01 +0000 (19:41 +0000)]
Branch b1_6
b=15069
i=rread
i=wangdi
remove LASSERT in ptlrpc_invalidate_import() since ping req can be queued even
imp_invalid is set. See also bug 15716.
johann [Fri, 2 May 2008 19:18:39 +0000 (19:18 +0000)]
Branch b1_6
b=15069
i=rread
i=wangdi
Skip the "sleep until imp_inflight == 0" part when invalidating an mgc import
which is already invalid.
johann [Fri, 2 May 2008 19:01:34 +0000 (19:01 +0000)]
Branch b1_6
b=13904
i=johann
i=tianzy
Quota data should not be stored in a 32-bit variable. Use __u64 instead.
deen [Wed, 30 Apr 2008 20:55:46 +0000 (20:55 +0000)]
Don't disable LIBLUSTRE_POSIX_ACL if --disable-liblustre given and
don't build wiretest with build utils.
b=15642
i=adilger
i=shadow
fanyong [Wed, 30 Apr 2008 17:23:56 +0000 (17:23 +0000)]
Branch b1_6
b=15691
i=huanghua
i=tappro
1) Prevent "ll_sa_wrong" from being increased repeatedly when hit ratio is low.
2) Remount client before sanity 123a to eliminate previous test effect.
ericm [Wed, 30 Apr 2008 17:18:03 +0000 (17:18 +0000)]
branch: b1_6
set msghdr flags right before rpc sent out.
b=15487
r=rread
r=shadow
tianzy [Wed, 30 Apr 2008 09:37:06 +0000 (09:37 +0000)]
Branch b1_6
b=15643
i=johann
Prolong the waiting time in test_12 of sanity-quota.sh
bwzhou [Wed, 30 Apr 2008 05:32:57 +0000 (05:32 +0000)]
Branch b1_6
Add ChangeLog entries.
bobijam [Wed, 30 Apr 2008 01:44:14 +0000 (01:44 +0000)]
Branch b1_6
b=14995
i=wangdi, rread
send activate event only for evicted imports.
walter [Tue, 29 Apr 2008 22:49:51 +0000 (22:49 +0000)]
b=13539
i=johann
i=adilger
fix for configure: WARNING: linux/random.h: present but cannot be compiled
walter [Tue, 29 Apr 2008 16:25:58 +0000 (16:25 +0000)]
b=15689
i=johann
i=alexey.lyashkov
fix error building lustre/liblustre/tests for Cray XT3 catamount
fanyong [Tue, 29 Apr 2008 02:44:39 +0000 (02:44 +0000)]
Branch b1_6
b=15406
Add ChangeLog entry.
anserper [Mon, 28 Apr 2008 20:51:51 +0000 (20:51 +0000)]
Branch b1_6
b=15441
i=johann
i=tianzy
quota might appear on quota files causing deadlock, this is fixed with this patch
brian [Mon, 28 Apr 2008 19:21:02 +0000 (19:21 +0000)]
b=15316,13969
ChangeLog entries.
anserper [Mon, 28 Apr 2008 14:04:32 +0000 (14:04 +0000)]
Branch b1_6
b=15036
updated changelog with respect to 15036 updates
bwzhou [Mon, 28 Apr 2008 04:21:59 +0000 (04:21 +0000)]
Branch b1_6
b=11777
r=green, shadow
invalidate the import to halt request handling so the namespace can be safely
cleaned up
bobijam [Mon, 28 Apr 2008 02:17:23 +0000 (02:17 +0000)]
Branch b1_6
b=14803
i=johann, adilger
Description: Don't update lov_desc members until making sure they are valid
Details : When updating lov_desc members via proc fs, need fix their
validities before doing the real update.
stinson1947 [Sun, 27 Apr 2008 02:48:57 +0000 (02:48 +0000)]
Updated for future beta release.
stinson1947 [Sun, 27 Apr 2008 02:47:15 +0000 (02:47 +0000)]
Added header for the next release 1.6.6.
stinson1947 [Sun, 27 Apr 2008 02:45:41 +0000 (02:45 +0000)]
*** empty log message ***
stinson1947 [Sun, 27 Apr 2008 01:51:13 +0000 (01:51 +0000)]
Changed version numbers for 1.6.5.
stinson1947 [Sun, 27 Apr 2008 01:48:47 +0000 (01:48 +0000)]
Added date for the 1.6.5 release.
fanyong [Sat, 26 Apr 2008 14:30:54 +0000 (14:30 +0000)]
Branch b1_6
b=15658
i=huanghua
i=tappro
Support directory read ahead under patchless client mode.
When "ls" a dentry, the system trigger more than once "revalidate" or
"lookup", for "getattr", for "getxattr", and maybe for others.
Under patchless client mode, the operation intent is not accurate,
it maybe misguide the statahead thread. For example:
The "revalidate" call for "getattr" and "getxattr" of a dentry maybe
have the same operation intent -- "IT_GETATTR".
In fact, one dentry should has only one chance to interact with the
statahead thread, otherwise the statahead windows will be confused.
deen [Sat, 26 Apr 2008 12:39:55 +0000 (12:39 +0000)]
It is be possible to send the lock handle along with each read
or write request because the client is already doing a lock match
itself so there isn't any reason the OST should have to re-do that
search.
b=15198
i=adilger
i=shadow
walter [Sat, 26 Apr 2008 02:23:43 +0000 (02:23 +0000)]
b=15277
i=adilger
i=rread
fixes error linking lustre/liblustre/tests/echo_test
during catamount build of b1_6 for Cray XT3 systems
adilger [Sat, 26 Apr 2008 00:10:33 +0000 (00:10 +0000)]
Branch b1_6
Clarify instructions.
adilger [Sat, 26 Apr 2008 00:08:57 +0000 (00:08 +0000)]
Branch b1_6
Fix line wrap.
adilger [Fri, 25 Apr 2008 23:22:02 +0000 (23:22 +0000)]
Branch b1_6
Minor cleanup.