Whamcloud - gitweb
nathan [Thu, 22 Mar 2007 17:58:13 +0000 (17:58 +0000)]
b=11973
i=green
atomic copy remote log instead of appending records -- earlier records may
have been modfied, and record count was wrong anyhow.
eeb [Thu, 22 Mar 2007 17:52:42 +0000 (17:52 +0000)]
* bug 11659 fix - finer-grained peerstamps for ptllnd connection
establishment and better credit flow checks.
eeb [Thu, 22 Mar 2007 14:36:00 +0000 (14:36 +0000)]
* Added changelog entry for previously committed ptllnd fix (11394)
eeb [Thu, 22 Mar 2007 14:33:10 +0000 (14:33 +0000)]
* Removed check for kernel TCP zero-copy patch - it's not used even if it's
there
mjmac [Thu, 22 Mar 2007 11:57:59 +0000 (11:57 +0000)]
* Add support for date-pinned co/update
r=adilger
huanghua [Thu, 22 Mar 2007 09:34:54 +0000 (09:34 +0000)]
we decide to use HEAD of lnet in b1_8.
fanyong [Thu, 22 Mar 2007 09:07:16 +0000 (09:07 +0000)]
Temporary fix for new lnet (from HEAD) test on orion.
huanghua [Thu, 22 Mar 2007 08:14:45 +0000 (08:14 +0000)]
small fixes make b1_8 compilable.
bobijam [Thu, 22 Mar 2007 06:00:31 +0000 (06:00 +0000)]
Branch HEAD
b=10997
r=adilger
Description: lfs setstripe use optional parameters instead of postional parameters.
scjody [Wed, 21 Mar 2007 20:49:17 +0000 (20:49 +0000)]
Branch HEAD
Remove "#if FOO" compiler warnings from -Wundef.
(Committed by adilger)
Define FMODE_READ and FMODE_WRITE for use in liblustre
(Committed by green)
Allow test 6 and 7 to be run in isolation.
Fix test file names.
Create test files in $DIR instead of $MOUNT.
(Committed by adilger)
openclose.c: Save errno before calling fprintf().
(Committed by adilger)
sanityN.sh: Kill multiop before exiting, if test fails.
Test is currently excluded, was retesting to see if it worked yet.
b=3192
(Committed by adilger)
r=adilger
r=nathan
r=green
scjody [Wed, 21 Mar 2007 20:44:42 +0000 (20:44 +0000)]
Branch HEAD
Merge the following changes from b1_4:
lfs quotachown <file system> deletes security relevant setuid bit
b=11463
r=Adilger
(Committed by tianzy)
Land kernel config changes for 2.4 kernels from b_release_1_4_8 to b1_4.
(Committed by adilger)
b=10736
Add getattr on lookup operation for NFS, so that it can properl revalidate
file attributes.
(Committed by green)
r=adilger
r=green
r=nathan
scjody [Tue, 20 Mar 2007 19:57:01 +0000 (19:57 +0000)]
Branch HEAD
Back out accidental commit
scjody [Tue, 20 Mar 2007 19:52:02 +0000 (19:52 +0000)]
Branch HEAD
Merge from b_release_1_4_9:
b=11643
r=adilger
Make reiserfs to compile at the expense of disabling quotas
nathan [Tue, 20 Mar 2007 18:12:11 +0000 (18:12 +0000)]
add note about bad test
adilger [Tue, 20 Mar 2007 06:39:53 +0000 (06:39 +0000)]
Branch HEAD
Untested patch for handling uninitialized block groups.
Adding to CVS to avoid loss, not currently in any patch series.
b=11269
nathan [Mon, 19 Mar 2007 23:11:10 +0000 (23:11 +0000)]
b=11013
This is a potential null pointer dereference.
Caught by HEAD's conf_sanity test 29
nathan [Mon, 19 Mar 2007 19:09:48 +0000 (19:09 +0000)]
b=11778
i=adilger
wait until config file is finished before allowing MDT clients to connect
scjody [Mon, 19 Mar 2007 17:01:41 +0000 (17:01 +0000)]
Branch HEAD
Update recommended e2fsprogs version.
adilger [Sun, 18 Mar 2007 07:31:28 +0000 (07:31 +0000)]
Branch HEAD
Fixes in b1_4 missing from HEAD:
- enable test 14c (b=4035, b=7040)
- don't print an error because /proc/sys/lustre/fail_loc is missing if
the lustre is stopped when error() is called
- cleanup echo device after test_28 is done
adilger [Sat, 17 Mar 2007 18:30:03 +0000 (18:30 +0000)]
Branch HEAD
Add llapi_is_lustre_mnt() to check for client filesystems, so that we
can keep the API of llapi_is_lustre_mnttype() the same.
b=10747
r=nathan
scjody [Thu, 15 Mar 2007 22:02:05 +0000 (22:02 +0000)]
Branch HEAD
b=11801
r=adilger
r=green
Fork ext3-extents-2.6.15.patch to apply to latest SLES 10 kernel.
adilger [Wed, 14 Mar 2007 09:08:20 +0000 (09:08 +0000)]
Branch HEAD
Fix liblustre linking for LustreFUSE.
adilger [Wed, 14 Mar 2007 08:14:51 +0000 (08:14 +0000)]
Branch HEAD
Pass correct variable to obd_statfs_rqset().
Unnoticed because of missing #ifdef _HAVE_STATVFS
adilger [Wed, 14 Mar 2007 02:10:50 +0000 (02:10 +0000)]
Branch HEAD
Fix for LustreFUSE building.
adilger [Wed, 14 Mar 2007 02:00:54 +0000 (02:00 +0000)]
Branch HEAD
Add comment to ensure people keep branch list sorted.
scjody [Tue, 13 Mar 2007 19:28:16 +0000 (19:28 +0000)]
Branch HEAD
Merge SLES 10 build fix from b1_4.
shadow [Tue, 13 Mar 2007 06:34:41 +0000 (06:34 +0000)]
fix patchless client build breakage with vanila 2.6.19 kernel.
b=11804
r=adilger
scjody [Tue, 13 Mar 2007 02:29:04 +0000 (02:29 +0000)]
Branch HEAD
Update SLES 10 kernel to 2.6.16.27-0.9.
yzy [Tue, 13 Mar 2007 01:43:45 +0000 (01:43 +0000)]
Land b1_6_bug11013 onto HEAD (20070313_0924)
b=11013
r=adilger
implement a common hash table, so it will be used as a hash class.
in this patch, finished :
1. " UUID <--> export " hash
2. " NID <--> export " hash
3. "NetPeer <--> connect" hash
scjody [Tue, 13 Mar 2007 00:23:26 +0000 (00:23 +0000)]
Branch HEAD
Update RHEL 4 kernel to 2.6.9-42.0.10.EL.
scjody [Mon, 12 Mar 2007 22:52:29 +0000 (22:52 +0000)]
Branch HEAD
Merge from b_release_1_4_9: update SLES 9 kernel to 2.6.5-7.283.
adilger [Sat, 10 Mar 2007 17:54:16 +0000 (17:54 +0000)]
Branch HEAD
Use intent directly instead of file->f_it, which doesn't exist on patchless
clients.
r=shadow (original patch)
eeb [Sat, 10 Mar 2007 02:12:43 +0000 (02:12 +0000)]
* Fixed bug noticed while Nic was investigating 11394 - kptllnd didn't
propagate portals completion errors up to LNET. The fix propagates the
error and logs a D_NETERROR.
nathan [Sat, 10 Mar 2007 00:03:14 +0000 (00:03 +0000)]
clarify error message
eeb [Fri, 9 Mar 2007 21:38:32 +0000 (21:38 +0000)]
* Changed syntax of "lctl ping target [timeout]", so "target" can be
either a NID or a process ID.
eeb [Fri, 9 Mar 2007 21:08:17 +0000 (21:08 +0000)]
* removed PTL_EQ_HANDLER_NONE define now that cray portals defines it
bobijam [Fri, 9 Mar 2007 03:02:57 +0000 (03:02 +0000)]
Branch HEAD
b=11090
r=green
Description: Checking the version difference of client vs. server, report error if the gap is too big.
adilger [Fri, 9 Mar 2007 00:04:18 +0000 (00:04 +0000)]
Branch b_hd_newconfig
Add ChangeLog entry for bug 10889.
adilger [Thu, 8 Mar 2007 23:54:31 +0000 (23:54 +0000)]
Branch HEAD
Have lustre/ChangeLog match b1_4 as much as possible.
adilger [Thu, 8 Mar 2007 20:59:25 +0000 (20:59 +0000)]
Branch HEAD
Avoid duplicate definition of SYSIO_INTERFACE_NAME.
b=11175
r=nic (original patch)
adilger [Thu, 8 Mar 2007 20:14:45 +0000 (20:14 +0000)]
Branch HEAD
Remove old, dangerous tcp zero copy patches.
adilger [Thu, 8 Mar 2007 20:09:21 +0000 (20:09 +0000)]
Branch HEAD
Remove old, dangerous tcp-zero-copy patch from 2.6.18 series.
eeb [Thu, 8 Mar 2007 19:33:45 +0000 (19:33 +0000)]
* Fixed #if typo
nathan [Thu, 8 Mar 2007 19:32:21 +0000 (19:32 +0000)]
catch extra args without dashes
nathan [Thu, 8 Mar 2007 18:37:18 +0000 (18:37 +0000)]
check for dangling upgrade flag
nathan [Thu, 8 Mar 2007 17:13:23 +0000 (17:13 +0000)]
must reset the upgrade flag or we try to upgrade every time the server
mounts. (recently introduced bug.)
nathan [Wed, 7 Mar 2007 02:03:43 +0000 (02:03 +0000)]
add plot-llstat, move from sbin to bin
nathan [Tue, 6 Mar 2007 01:04:02 +0000 (01:04 +0000)]
Change name from --noformat to --dryrun for clarity
nathan [Tue, 6 Mar 2007 00:09:13 +0000 (00:09 +0000)]
update vers and supported kernels
scjody [Mon, 5 Mar 2007 18:05:08 +0000 (18:05 +0000)]
Branch HEAD
Revert my fix to bug 11636 since it broke the build.
b=11786
eeb [Mon, 5 Mar 2007 12:38:06 +0000 (12:38 +0000)]
* Support "lustrecvs b_self_test" to check out the lnet self-test development
branch.
adilger [Sat, 3 Mar 2007 14:23:10 +0000 (14:23 +0000)]
Branch HEAD
Remove the broken multi-mount-protection patch.
adilger [Sat, 3 Mar 2007 14:21:20 +0000 (14:21 +0000)]
Branch HEAD
Add 1.4.9 release date.
nathan [Fri, 2 Mar 2007 19:39:59 +0000 (19:39 +0000)]
update tunefs examples
nathan [Fri, 2 Mar 2007 18:08:49 +0000 (18:08 +0000)]
b=11149
wrong end condition
nathan [Fri, 2 Mar 2007 17:19:39 +0000 (17:19 +0000)]
b=11149
r=adilger
Add checks for null indicies in QOS code.
Also fix 'lfs df' and /proc...lov/target_obd.
adilger [Fri, 2 Mar 2007 14:13:07 +0000 (14:13 +0000)]
Branch HEAD
Merge readlink return type configure check from b1_4.
b=11731
r=mjmac (original patch)
adilger [Fri, 2 Mar 2007 13:56:21 +0000 (13:56 +0000)]
Branch HEAD
Enable CONFIG_E1000_DISABLE_PACKET_SPLIT to match b1_4.
Enable CONFIG_IPV6=m in UML kernel so that tcp symbols are exported.
nathan [Thu, 1 Mar 2007 19:12:57 +0000 (19:12 +0000)]
b=11752
assert possibly caused by uninitialized var
eeb [Thu, 1 Mar 2007 17:11:52 +0000 (17:11 +0000)]
* fixed copy/paste error in previous 11684 commit
scjody [Thu, 1 Mar 2007 14:58:57 +0000 (14:58 +0000)]
Branch HEAD
b=11636
r=green,adilger
Remove useless kernel configs.
nathan [Wed, 28 Feb 2007 01:34:42 +0000 (01:34 +0000)]
add wait command to make this scriptable
eeb [Tue, 27 Feb 2007 21:04:22 +0000 (21:04 +0000)]
* fixed cut/paste error on adding darwin sysctl for "debug_peer_on_timeout"
nathan [Tue, 27 Feb 2007 20:35:56 +0000 (20:35 +0000)]
find lctl and mkfs.lustre if not local build
nathan [Mon, 26 Feb 2007 22:55:32 +0000 (22:55 +0000)]
make sure liblustre test can start
deen [Mon, 26 Feb 2007 22:06:30 +0000 (22:06 +0000)]
Fix an error where the garbage may be returned by this function because
of uninitialized pointer.
r=rmg@ (nathan)
adilger [Mon, 26 Feb 2007 20:29:26 +0000 (20:29 +0000)]
Branch HEAD
Don't try to report stats on an incoming request we can't decode
(trips LASSERT in since RPCv2 support was added).
nathan [Mon, 26 Feb 2007 20:24:48 +0000 (20:24 +0000)]
Bad incoming messages should never cause an assertion.
Assertions are for verifying logic.
isaac [Mon, 26 Feb 2007 19:57:57 +0000 (19:57 +0000)]
b=10316,r=eeb:
- fixed console chatter in case of -ETIMEDOUT.
nathan [Mon, 26 Feb 2007 19:11:33 +0000 (19:11 +0000)]
b=10894
r=adilger
Clarify watchdog timer units.
Add ability to change MDT ldlm timeout.
green [Sun, 25 Feb 2007 22:51:00 +0000 (22:51 +0000)]
b=11028
r=adilger
Fix cfs_timeval_sub to properly fill struct timeval.
Correctly assign microseconds to microsecondsand seconds to seconds.
wangdi [Sun, 25 Feb 2007 09:24:57 +0000 (09:24 +0000)]
Branch:HEAD
typo fix about join_lsm unpack, caused by 11471 landing
nathan [Sat, 24 Feb 2007 23:45:53 +0000 (23:45 +0000)]
b=11267
add echo server starter
fix pthread stderr probs
shadow [Fri, 23 Feb 2007 07:57:17 +0000 (07:57 +0000)]
fix typo.
shadow [Thu, 22 Feb 2007 18:34:56 +0000 (18:34 +0000)]
fix run test large amount free memory.
r=adilger
shadow [Thu, 22 Feb 2007 17:53:17 +0000 (17:53 +0000)]
remove ext3-wantedi-misc patch.
struct dentry_params renamed to lvfs_dentry_params and moved into
lvfs_linux.h
b=11248
r=adilger
nathan [Thu, 22 Feb 2007 16:04:26 +0000 (16:04 +0000)]
typo
nathan [Wed, 21 Feb 2007 22:17:47 +0000 (22:17 +0000)]
1.6b8
nathan [Wed, 21 Feb 2007 21:25:50 +0000 (21:25 +0000)]
b=2258
r=adilger
Insure min dynamic threads < max threads
Also make absolutely sure we don't start more than MAX
on the biggest machines, we limit the MDS auto-threads to start at 32 and go to 128, nowhere near the 512 max limit. On e.g.
ilter.c lustre/ost/ost_handler.c
on the biggest machines, we limit the MDS auto-threads to start at 32 and go to 128, nowhere near the 512 max limit. On e.g.
(8 * 1024)
on the biggest machines, we limit the MDS auto-threads to start at 32 and go to 128, nowhere near the 512 max limit. On e.g.
s/lustre-core/mds/handler.c,v
on the biggest machines, we limit the MDS auto-threads to start at 32 and go to 128, nowhere near the 512 max limit. On e.g.
threads = MDS_THREADS_AUTO_MAX;
on the biggest machines, we limit the MDS auto-threads to start at 32 and go to 128, nowhere near the 512 max limit. On e.g.
andler.c 21 Feb 2007 21:22:08 -0000
on the biggest machines, we limit the MDS auto-threads to start at 32 and go to 128, nowhere near the 512 max limit. On e.g.
zam [Wed, 21 Feb 2007 12:49:24 +0000 (12:49 +0000)]
set CONFIG_LOCALVERISON="" to solve lbuild failures
nathan [Tue, 20 Feb 2007 22:04:00 +0000 (22:04 +0000)]
b=11692
play nicer with parameters on logs upgraded from 1.4.x
- allows updates for old logs whenever we can (i.e. ok for mdc, not for osc)
- allows updates for everyone if --writeconf is specified at tunefs upgrade
time (requires entire fs to be stopped though)
- allows updates of failover partners if --writeconf is specified
- fixes various other little bugs and error messages
- adds a couple of regression tests for upgrading from 1.4.6
zam [Tue, 20 Feb 2007 20:48:45 +0000 (20:48 +0000)]
missing linux kernel config file for the SLES10 kernel,
the config borrowed from b_release_1_4_9
zam [Tue, 20 Feb 2007 19:22:09 +0000 (19:22 +0000)]
b=11461
r=scjody,vitaly
a fix for extra suffixes for kernel version string which
confuse lbuild.
nathan [Tue, 20 Feb 2007 17:18:36 +0000 (17:18 +0000)]
b=10719
r=adilger
set external journal device read-only
nathan [Tue, 20 Feb 2007 16:23:07 +0000 (16:23 +0000)]
fail test 11 after 30 min
nathan [Tue, 20 Feb 2007 16:20:53 +0000 (16:20 +0000)]
wait a little after turning off fail_loc to make sure files show up
kalpak [Mon, 19 Feb 2007 13:48:36 +0000 (13:48 +0000)]
b=3244
r=Andreas Dilger
Description: Addition of EXT3_FEATURE_RO_COMPAT_DIR_NLINKS flag for
> 32000 subdirectories
Details : Add EXT3_FEATURE_RO_COMPAT_DIR_NLINK flag to
EXT3_FEATURE_RO_COMPAT_SUPP. This flag will be set whenever
subdirectory count crosses 32000. This will aid e2fsck to
correctly handle more than 32000 subdirectories.
nathan [Sun, 18 Feb 2007 22:26:46 +0000 (22:26 +0000)]
quota test 11 takes forever on my machine when run as part of acceptance-small.
- Add some progress indication.
- Fix "cycles" (rm -f erases test dir, so must recreate)
adilger [Sat, 17 Feb 2007 18:06:19 +0000 (18:06 +0000)]
Branch HEAD
Remove trailing whitespace.
adilger [Sat, 17 Feb 2007 09:14:44 +0000 (09:14 +0000)]
Branch HEAD
Use lprocfs_rd_atomic() because ns_locks is atomic in HEAD. Otherwise we get
bogus values from /proc/fs/lustre/ldlm/namespaces/*/lock_count.
yujian [Sat, 17 Feb 2007 07:52:44 +0000 (07:52 +0000)]
A __u64 is not always a long long. On some architectures it is only a
long. Let's declare write_bytes as unsigned long long.
adilger [Sat, 17 Feb 2007 07:22:32 +0000 (07:22 +0000)]
Branch HEAD
We don't support LUSTRE_KERNEL_VERSION < 45 anymore.
yujian [Sat, 17 Feb 2007 04:05:14 +0000 (04:05 +0000)]
Remove the 4G limitation of maximum writing bytes.
tianzy [Sat, 17 Feb 2007 01:54:57 +0000 (01:54 +0000)]
fix a typo
b=11471
nathan [Fri, 16 Feb 2007 23:14:06 +0000 (23:14 +0000)]
b=11691
r=adilger
Don't print regular error messages about the MGS when we're connecting to
an old MDT.
nathan [Fri, 16 Feb 2007 22:12:39 +0000 (22:12 +0000)]
qos test: don't fail if we write 0 bytes to smaller ost
nathan [Fri, 16 Feb 2007 20:54:26 +0000 (20:54 +0000)]
b=11667
missing symbol
eeb [Fri, 16 Feb 2007 18:01:23 +0000 (18:01 +0000)]
* landing 11667 (debug_peer_on_timeout) and 11684 (liblustre block for asynch
OST_CONNECTs to complete at startup) on HEAD
nathan [Fri, 16 Feb 2007 17:17:40 +0000 (17:17 +0000)]
fixes for running in more user environments
eeb [Fri, 16 Feb 2007 15:44:57 +0000 (15:44 +0000)]
Severity : enhancement
Bugzilla : 11684
Description: Added D_NETTRACE for recording network packet history
(initially only for ptllnd). Also a separate userspace
ptllnd facility to gather history which should really be
covered by D_NETTRACE too, if only CDEBUG recorded history in
userspace.