Whamcloud - gitweb
nathan [Wed, 18 Jul 2007 19:37:28 +0000 (19:37 +0000)]
b=12860
i=adilger
i=green
fix mds_lov_synchronize race
johann [Wed, 18 Jul 2007 07:43:11 +0000 (07:43 +0000)]
Branch b1_6
b=12861
i=alex
i=adilger
i=kalpak
Severity : major
Bugzilla : 12861
Description: ldiskfs_ext_search_right: bad header in inode #3981353: unexpected
eh_depth
Details : a wrong check of extent headers in ldiskfs_ext_search_right()
can cause the filesystem to be remounted read-only.
adilger [Wed, 18 Jul 2007 05:14:58 +0000 (05:14 +0000)]
Branch b1_6
Fix ll_opcode2str() and opcode_offset() to handle LLOG request types.
Fix ChangeLog to put changes at the end of the 1.6.1 section.
b=12731
i=nathan
i=scjody
adilger [Tue, 17 Jul 2007 23:10:09 +0000 (23:10 +0000)]
Branch b1_6
sanity.sh:
Change test_78 to limit it to 1/2 of total client memory.
Document fail_loc values.
Add error messages on failure.
test-framework.sh:
Don't default to FSTYPE=ext3 since we only run 2.6 kernels now.
b=13007
i=nathan
nathan [Tue, 17 Jul 2007 21:09:49 +0000 (21:09 +0000)]
b=10968
fix entry
johann [Tue, 17 Jul 2007 19:31:50 +0000 (19:31 +0000)]
Branch b1_6
b=11039
i=adilger
i=alex
i=kalpak
attachment 10893 / fsfilt_ext3: mainline extents support
johann [Tue, 17 Jul 2007 19:01:12 +0000 (19:01 +0000)]
Branch b1_6
b=12696
i=johann
i=adilger
Severity : normal
Frequency : rare
Bugzilla : 12696
Description: ASSERTION(imp->imp_conn_current) failed
Details : an assertion failure is hit if a client node boots and attempts to
mount a lustre filesystem faster than RECONNECT_INTERVAL seconds.
scjody [Tue, 17 Jul 2007 00:14:33 +0000 (00:14 +0000)]
Branch b1_6
Revert this change since it caused build failures on RHEL 4.
b=11932
scjody [Mon, 16 Jul 2007 23:24:49 +0000 (23:24 +0000)]
Branch b1_6
Disable automatic dependencies on SLES 10.
b=11932
i=mjmac
scjody [Mon, 16 Jul 2007 23:04:09 +0000 (23:04 +0000)]
Branch b1_6
Fix format string
b=11969
jxiong [Sat, 14 Jul 2007 15:41:02 +0000 (15:41 +0000)]
Replace \t with spaces
milind [Fri, 13 Jul 2007 13:15:50 +0000 (13:15 +0000)]
b=10968
Modified lustre/ChangeLog for ltrack_stats.
jxiong [Fri, 13 Jul 2007 10:17:20 +0000 (10:17 +0000)]
LC_PAGE_CONSTANT won't work because struct page is not defined, fixed it
jxiong [Fri, 13 Jul 2007 09:12:58 +0000 (09:12 +0000)]
r=alex,nathan
b=11890
raid5 write zerocopy support for rhel4 2.6 kernels.
jxiong [Fri, 13 Jul 2007 09:11:40 +0000 (09:11 +0000)]
r=alex,nathan
b=11890
Raid5 write zerocopy support for rhel-2.6 kernels.
milind [Thu, 12 Jul 2007 14:51:53 +0000 (14:51 +0000)]
b=10968
r=nathan, komaln
Description:
track_stats runs command given and does one of the following:
1. Writes its pid to /proc/fs/lustre/llite/.../stats_track_pid
to collects stats for that process.
2. Writes its ppid to /proc/fs/lustre/llite/.../stats_track_ppid
to collect stats of that process and all its children
3. Sets gid of process to some random gid (444) and also
writes that to/proc/fs/lustre/llite/.../stats_track_gid to collect stats
of all processes in that group
It also uses llstat to generate output with interval of 1 second and duration
of run of command for plot-llstat to generate a graph
komaln [Thu, 12 Jul 2007 11:10:14 +0000 (11:10 +0000)]
This is the patch for the llanalyze man page.
Following commands are to be followed to view the man page.
#pod2man llanalyze llanalyze.1
#less llanalyze.1
deen [Thu, 12 Jul 2007 09:50:37 +0000 (09:50 +0000)]
In case of positive reply from server and failed client bulk
callback after bulk transfer shouldn't LBUG, but process this
request as erroneous.
b=11544
i=green
i=eeb
tianzy [Thu, 12 Jul 2007 08:36:47 +0000 (08:36 +0000)]
Branch b1_6
Quota slave depends qctxt->lqc_import to send its quota request.
This pointer will be invalid if mds did failover or broke its
connect to osts, which leads to LBUG.
b=12374
i=green
i=adilger
grev [Wed, 11 Jul 2007 20:52:45 +0000 (20:52 +0000)]
b=12849
i=Adilger
i=Brian
cleanup_test102 fix
brian [Wed, 11 Jul 2007 20:20:09 +0000 (20:20 +0000)]
b=12607
r=tianzy,nathan
Support for remote MDT/OST.
mjmac [Tue, 10 Jul 2007 15:37:25 +0000 (15:37 +0000)]
b=7288
Removing test 27 from the ALWAYS_EXCEPT list, as it is now expected to pass.
nathan [Tue, 10 Jul 2007 15:30:40 +0000 (15:30 +0000)]
b=12910
increase conf_param time
This should be happening quickly (lock revocation signals clients to update
config), not based on any timeout.
bobijam [Tue, 10 Jul 2007 02:39:34 +0000 (02:39 +0000)]
Branch b1_6
b=12784
i=green
i=adilger
Original patch by miles@cray.com
Description: bad return value and errno from fcntl call
Details : In liblustre API, errno should be a negative value if error
happens.
vitaly [Mon, 9 Jul 2007 17:51:03 +0000 (17:51 +0000)]
Branch b1_6
early-lock-cancel string is added to obd_connect_names
yangsheng [Fri, 6 Jul 2007 10:31:06 +0000 (10:31 +0000)]
Branch b1_6
b=<12415>
r=<adilger,green>
yangsheng [Fri, 6 Jul 2007 10:22:57 +0000 (10:22 +0000)]
Branch b1_6
b=<12415>
r=<adilger,green>
nathan [Thu, 5 Jul 2007 23:42:33 +0000 (23:42 +0000)]
b=10676
i=adilger
i=wangdi
b=12461 minor cleanup
i=brian
i=adilger
adilger [Thu, 5 Jul 2007 21:43:58 +0000 (21:43 +0000)]
Branch b1_6
Quiet verbose error message on read-only filesystem. We will already have
been notified of this.
b=10595
i=behlendo
adilger [Wed, 4 Jul 2007 20:08:13 +0000 (20:08 +0000)]
Branch b1_6
Update lfs documentation to reflect current usage.
tianzy [Wed, 4 Jul 2007 07:53:53 +0000 (07:53 +0000)]
Branch b1_6
Severity : normal
Frequency : when qunit size is too small(less than 20M)
Bugzilla : 12588
Description: write is stopped by improper -EDQUOT
Details : If the master is busy and qunit size is small enough(let's say 1M),
the slave can not get quota from master on time, which will lead
slave to trigger a -EQUOTA to client.
b=12588
i=green
i=wangdi
nathan [Tue, 3 Jul 2007 21:52:17 +0000 (21:52 +0000)]
fix format
johann [Mon, 2 Jul 2007 18:48:31 +0000 (18:48 +0000)]
Branch b1_6
Severity : normal
Frequency : rare
Bugzilla : 12629
Description: Deadlock during metadata tests
Details : in prune_dir_dentries(), shrink_dcache_parent() should not be
called with the per-dentry lock held.
b=12629
i=alex
i=shadow
johann [Mon, 2 Jul 2007 12:27:05 +0000 (12:27 +0000)]
Branch b1_6
Severity : normal
Frequency : SLES9 only
Bugzilla : 12744
Description: Lustre patched kernel for SLES9 SP3 has NR_CPUS set to 8
Details : set CONFIG_NR_CPUS to 128 instead of 8.
b=12744
i=scjody
i=green
huangwei [Mon, 2 Jul 2007 10:08:30 +0000 (10:08 +0000)]
b=5491
r=alex,adilger
a minor fix for posix test suit T.fcntl 23 failure
huangwei [Mon, 2 Jul 2007 09:59:34 +0000 (09:59 +0000)]
b=11880
r=alex,oleg
landing patch from 11880.
buffalo-pull [Mon, 2 Jul 2007 06:00:19 +0000 (06:00 +0000)]
Weekly tag build -> v1_6_0_97
grev [Fri, 29 Jun 2007 15:17:25 +0000 (15:17 +0000)]
Fix: Avoid the possible problems caused by "C" option.
Cleanup: Avoid hardcoded path.
b=12759
i=Adilger
i=Shadow
bobijam [Fri, 29 Jun 2007 05:26:09 +0000 (05:26 +0000)]
Branch b1_6
b=11815
i=adilger
i=wangdi
Description: replace obdo_alloc() with OBDO_ALLOC macro
Details : nothing special is done in obdo_alloc() function, and for
debugging purpose, it needs to be replaced with macros.
green [Fri, 29 Jun 2007 04:06:24 +0000 (04:06 +0000)]
b=11706
r=eeb,adilger
do not panic if pinger set changed reply status. The only valid action
is to check if reply was actually processed.
nathan [Thu, 28 Jun 2007 18:42:12 +0000 (18:42 +0000)]
i=adilger
* fix ldiskfs module path
* add uml debug file paths
wangdi [Thu, 28 Jun 2007 05:13:39 +0000 (05:13 +0000)]
Branch: b1_6
Checking whether lov_tgts is NULL in some lov funcations.
b=12576
i=adilger
i=shadow
bobijam [Wed, 27 Jun 2007 17:20:09 +0000 (17:20 +0000)]
Branch b1_6
b=12646
i=huanghua(original fix)
missed a typo.
jxiong [Wed, 27 Jun 2007 12:20:18 +0000 (12:20 +0000)]
Set the handle_hash size to 4096 if lustre is working under UML(user mode linux).
buffalo-pull [Wed, 27 Jun 2007 06:05:26 +0000 (06:05 +0000)]
Tag build -> v1_6_0_96
bobijam [Mon, 25 Jun 2007 02:32:02 +0000 (02:32 +0000)]
Branch b1_6
b=12646
i=adilger
Description: sanity.sh test_77h fails with "test_77h file compare failed"
Details : test_77h uses a file which was messed by other test case.
shadow [Fri, 22 Jun 2007 15:38:30 +0000 (15:38 +0000)]
improve handling recoverable errors
b=11710
i=green
i=wangdi
deen [Fri, 22 Jun 2007 09:08:44 +0000 (09:08 +0000)]
Change test to use different file sizes.
b=11737
i=adilger
i=shadow
shadow [Fri, 22 Jun 2007 07:01:50 +0000 (07:01 +0000)]
improve checks for exported symbols. This allow run check without sources, but with
Module.symvers shipped with kernel distribution.
add check for truncate_complete_page used by patchless client.
b=12415
i=adilger
i=scjody
shadow [Fri, 22 Jun 2007 05:31:36 +0000 (05:31 +0000)]
inode i_nlink equal zero is not enough for output message about disk corruption,
i_ctime and i_mode should be also checked.
b=10419
i=green
i=kalpak
nathan [Thu, 21 Jun 2007 20:19:41 +0000 (20:19 +0000)]
remove .cvsignores from old ldiskfs dirs
nathan [Thu, 21 Jun 2007 18:49:38 +0000 (18:49 +0000)]
note OBD_FAIL loc for easy referencing
deen [Thu, 21 Jun 2007 10:17:12 +0000 (10:17 +0000)]
Remove wrong chunk from the patch.
b=11971
johann [Thu, 21 Jun 2007 07:53:31 +0000 (07:53 +0000)]
Branch b1_6
b=12678
i=adilger
i=scjody
Severity : enhancement
Bugzilla : 12678
Description: remove fs_prep_san_write operation and related patches
Details : remove the ext3-san-jdike patches which are no longer useful.
johann [Thu, 21 Jun 2007 07:37:27 +0000 (07:37 +0000)]
Branch b1_6
b=11324
i=adilger
i=alex
Severity : normal
Frequency : rare
Bugzilla : 11324
Description: LDISKFS-fs error (device sdc): ldiskfs_free_blocks
Details : a disk corruption can cause the mballoc code to assert on a
double free.
shadow [Thu, 21 Jun 2007 07:20:24 +0000 (07:20 +0000)]
fix IS_ERR implementation in liblustre.h for right detect errors
b=12670
i=adilger
i=green
shadow [Thu, 21 Jun 2007 06:21:40 +0000 (06:21 +0000)]
fix some mal-formatted DEBUG_REQ and LCONSOLE_ERROR_MSG messages
b=12747
i=wangdi
i=scjody
shadow [Thu, 21 Jun 2007 05:41:50 +0000 (05:41 +0000)]
2.6.22 has only one visble change for patchless cleint, SLAB_CTOR_* constants is
removed. In this case we need drop using os depended interface to
kmem_cache and use cfs_mem_cache API.
b=12764
i=green
i=wangdi
shadow [Thu, 21 Jun 2007 05:02:55 +0000 (05:02 +0000)]
kernels from 2.6.21 not need include linux/config.h, but add include
linux/autoconf.h in commpiler command line.
b=12605
i=wangdi
i=scjody
bobijam [Thu, 21 Jun 2007 03:37:37 +0000 (03:37 +0000)]
b=11686
i=green, adilger
improve some error messages.
walter [Thu, 21 Jun 2007 01:33:12 +0000 (01:33 +0000)]
Branch b1_6
b=12348
i=adilger
i=mjmac
fix build errors seen when compiling on cray xt3 systems
vitaly [Thu, 21 Jun 2007 00:29:07 +0000 (00:29 +0000)]
Land b1_6_elc onto b1_6 (20070621_0218)
land Early Lock Cancel code
mjmac [Wed, 20 Jun 2007 20:14:42 +0000 (20:14 +0000)]
b=12348
i=adilger
i=scjody
* Include llog_test.ko in lustre-tests RPM
* Include liblustre tests, when present, in lustre-tests RPM
deen [Wed, 20 Jun 2007 18:39:38 +0000 (18:39 +0000)]
Direct I/O operations should return actual amount of bytes transferred
rather than requested size.
b=11737
i=adilger
i=shadow
buffalo-pull [Wed, 20 Jun 2007 02:33:39 +0000 (02:33 +0000)]
tag version -> 1.6.0.95
bobijam [Mon, 18 Jun 2007 03:33:55 +0000 (03:33 +0000)]
Branch b1_6
b=12689
i=green, vitaly
Description: replay-single.sh test 52 fails
Details : A lock's skiplist need to be cleanup when it being unlinked
from its resource list.
jxiong [Sat, 16 Jun 2007 07:23:18 +0000 (07:23 +0000)]
b=10706
r=adilger,johann
use RCU to improve the scalability of class handle object search, reloaded.
jxiong [Sat, 16 Jun 2007 05:06:46 +0000 (05:06 +0000)]
Broke SLES build, restored.
jxiong [Sat, 16 Jun 2007 03:37:23 +0000 (03:37 +0000)]
b=10706
r=adilger,johann
Use RCU to improve the scalability of class handle hash.
vitaly [Fri, 15 Jun 2007 16:47:06 +0000 (16:47 +0000)]
Branch b1_6
b=12731
i=adilger
i=green
ASSERTION(offset < LUSTRE_MAX_OPCODES) fix
shadow [Fri, 15 Jun 2007 14:11:34 +0000 (14:11 +0000)]
imp_lock should be held while iterating over imp_sending_list for
prevent destroy request after get timeout in ptlrpc_queue_wait.
b=11674
i=green
i=wangdi
johann [Fri, 15 Jun 2007 13:43:44 +0000 (13:43 +0000)]
Branch b1_6
b=12194
i=adilger
i=scjody
scjody [Thu, 14 Jun 2007 20:26:49 +0000 (20:26 +0000)]
Branch b1_6
Modify test-framework tests to run on real hardware - based on a patch
by Nic Henke <nic@cray.com>
b=11217
i=mjmac
i=nathan
shadow [Thu, 14 Jun 2007 08:56:42 +0000 (08:56 +0000)]
fix NULL pointer dereference
b=12731
i=vitaly (original author)
i=adilger
shadow [Thu, 14 Jun 2007 08:33:38 +0000 (08:33 +0000)]
add macro LCONSOLE_ERROR_MSG with extra parameter and map
LCONSOLE_ERROR(fmt, arg) to LCONSOLE_ERROR_MSG(0x000, fmt, ## arg).
b=10595
i=adilger
vitaly [Wed, 13 Jun 2007 12:06:21 +0000 (12:06 +0000)]
Branch b1_6
the build is broken with last commits. fixing
jitendra [Wed, 13 Jun 2007 07:28:47 +0000 (07:28 +0000)]
This update includes Hex error ID's and checksum calculation for console error messages.
It also includes code fixes for first five improved error messages.
b=10595
i=adilger
i=braam
scjody [Tue, 12 Jun 2007 22:53:48 +0000 (22:53 +0000)]
Branch b1_6
Remove ldiskfs and ldiskfs2 from lustre/
b=12068
scjody [Tue, 12 Jun 2007 22:52:05 +0000 (22:52 +0000)]
Branch b1_6
Remove ldiskfs patches and series
b=12068
scjody [Tue, 12 Jun 2007 21:58:46 +0000 (21:58 +0000)]
Branch b1_6
Move ldiskfs to its own package and CVS module.
b=12068
i=mjmac
i=brian
adilger [Sat, 9 Jun 2007 15:55:31 +0000 (15:55 +0000)]
Branch b1_6
Add MPICH2 1.0.3 ADIO_Lustre patch to user-contrib area.
yujian [Sat, 9 Jun 2007 07:09:09 +0000 (07:09 +0000)]
Branch b1_6
b=10760
i=brian
i=nathan
Improve the operation to get the corresponding NID(s) of the MGS node from
the "mgs nids" field of one lustre target line for checking lnet connectivity
between that lustre node and the MGS node.
adilger [Fri, 8 Jun 2007 20:24:32 +0000 (20:24 +0000)]
Branch b1_6
Remove the extra checkstat call, which should have been removed with the
landing of the original patch.
b=10818 b=12623
i=scjody
i=nathan
mjmac [Fri, 8 Jun 2007 15:49:32 +0000 (15:49 +0000)]
b=12348
i=scjody
i=adilger
* Create a lustre-tests RPM that includes stuff from lustre/tests
johann [Fri, 8 Jun 2007 08:39:30 +0000 (08:39 +0000)]
Branch b1_6
remove fs_prep_san_write operation
b=11039
i=adilger
i=nathan
johann [Fri, 8 Jun 2007 06:21:23 +0000 (06:21 +0000)]
Branch b1_6
use PF_SYNCWRITE only with kernels having this flag
attachment 10735
b=11039
i=adilger
i=shadow
scjody [Thu, 7 Jun 2007 22:08:09 +0000 (22:08 +0000)]
version -> 1.6.0.92
nathan [Thu, 7 Jun 2007 19:20:45 +0000 (19:20 +0000)]
i=adilger
fix 2.6.10 uml compile
komaln [Thu, 7 Jun 2007 11:49:33 +0000 (11:49 +0000)]
b=10968
i= Nathan, Kalpak
1) merged vfs_stats to stats file.
2) changed stats file to be used under differebt modes (all process, specific GID, PID or PPID)
3) replaced ll_lock by new locks for ll_rw_stats_tally.
4) Turn on collection of statistics ll_rw_stats_tally only when a special
character is echoed into any of the extents_stats, extents_stats_per_process,
offset_stats files.
kalpak [Thu, 7 Jun 2007 08:02:42 +0000 (08:02 +0000)]
b=12597
i=nathan, komal
brw_stats were being printed as log2 but all of them were not recorded as log2. Also remove some code duplication arising from filter_tally_{read,write}.
wangdi [Thu, 7 Jun 2007 07:22:37 +0000 (07:22 +0000)]
Branch:b1_6
The catlog should be initialized anyway even in recovery.
b=12086
i=nathan
i=adilger
green [Wed, 6 Jun 2007 17:02:29 +0000 (17:02 +0000)]
b=11706
r=eeb,adilger,nic(original author)
Insert cond_resched in ptlrpc_main to give other threads a chance to use some cp
u.
scjody [Tue, 5 Jun 2007 15:24:13 +0000 (15:24 +0000)]
Branch b1_6
Update to match reality
johann [Tue, 5 Jun 2007 07:22:51 +0000 (07:22 +0000)]
Branch b1_6
Release the module reference in ll_kern_mount()
b=12546
i=green
i=shadow
adilger [Sat, 2 Jun 2007 08:00:52 +0000 (08:00 +0000)]
Branch b1_6
Fix up error message for the case where a client hits a bogus checksum
error. This only happens during mmap IO, and can be seen when the
client-side checksum matches the new checksum found on the server,
or has changed again since the message was originally sent.
In preparation for turning checksumming on by default.
b=11742
i=scjody
i=adilger (behlendorf original patch)
adilger [Fri, 1 Jun 2007 20:52:25 +0000 (20:52 +0000)]
Branch b1_6
Update supported patchless client list.
b=11547
komaln [Fri, 1 Jun 2007 07:18:13 +0000 (07:18 +0000)]
b=12475
r=Adilger, Nathan.
Prevent machine freeze when echo into proc/fs/lustre/llite/*/extents_stats
adilger [Fri, 1 Jun 2007 06:13:24 +0000 (06:13 +0000)]
Branch b1_6
Don't actually need a cast for this.
nathan [Thu, 31 May 2007 22:41:47 +0000 (22:41 +0000)]
accidentally lost adilger's line from 1.9.34.1
adilger [Thu, 31 May 2007 21:54:48 +0000 (21:54 +0000)]
Branch b1_6
Always send IO to OST_IO_PORTAL regardless of what the default portal is.
Was previously broken for echo_client on catamount, but would also break
on 1.8 once the default request portal is REQPORTAL.
Detect if OST_WRITE or OST_READ requests go to non-IO portals, as that
would LBUG the server due to lack of IO buffers.
b=11813
i=green
i=shadow