Whamcloud - gitweb
adilger [Sat, 10 Sep 2005 09:09:48 +0000 (09:09 +0000)]
Branch: b1_4
Don't allocate zero'd memory so we write non-zero'd data to disk in test 23.
b=7279
adilger [Sat, 10 Sep 2005 08:09:58 +0000 (08:09 +0000)]
Branch: b1_4
Add lseek() tests (t23) to liblustre sanity.c for bug 7279.
Test can be run without arguments if LIBLUSTRE_MOUNT_TARGET is set.
Allow running only a single test from sanity via "-o {number}" like sanity.sh.
b=7279
adilger [Sat, 10 Sep 2005 06:31:36 +0000 (06:31 +0000)]
Branch b1_4
Added debugging for bad LOV EA detection.
Quiet spurious error for generation mismatch.
r=phil (original patch)
adilger [Fri, 9 Sep 2005 20:21:07 +0000 (20:21 +0000)]
Branch b1_4
Description: doing an ls when liblustre clients are running is slow
Details : sending a glimpse AST to a liblustre client waits for every AST
to time out, as liblustre clients will not respond while they
are processing. Since they cannot cache data anyways we refresh
the OST lock LVB from disk instead.
b=7198
r=phil, green (original patch)
adilger [Fri, 9 Sep 2005 18:32:26 +0000 (18:32 +0000)]
Branch b1_4
Remove obsolete test script (this is covered by replay-single.sh anyways).
adilger [Fri, 9 Sep 2005 18:18:48 +0000 (18:18 +0000)]
Branch b1_4
Description: specifying an (invalid) directory default stripe_size of -1
would reset the directory default striping
Details : stripe_size -1 was used internally to signal directory stripe
removal, now use "all default" to signal dir stripe removal
as a directory striping of "all default" is not useful
b=7328
r=green
adilger [Fri, 9 Sep 2005 16:15:48 +0000 (16:15 +0000)]
Branch b1_4
Description: Tuning the MDC DLM LRU size to zero triggers client LASSERT
Details : llu_lookup_finish_locks() tries to set lock data on a lock
after it has been released, only do this for referenced locks.
Tested by bogl.
b=7201 (b=7350)
r=green
adilger [Thu, 8 Sep 2005 18:16:18 +0000 (18:16 +0000)]
Branch b1_4
Description: liblustre could not open files whose last component is a symlink
Details : sysio_path_walk() would incorrectly pass the open intent to
intermediate path components.
b=6363
r=oleg, lee, devesh
adilger [Thu, 8 Sep 2005 07:49:38 +0000 (07:49 +0000)]
Branch b1_4
Description: Fix for potential infinite loop processing records in an llog.
Details : If an llog record is corrupted/zeroed, it is possible to loop
forever in llog_process(). Validate the llog record length
and skip the remainder of the block on an invalid value.
b=7359
adilger [Thu, 8 Sep 2005 07:27:59 +0000 (07:27 +0000)]
Branch b1_4
Quiet some overly noisy debug messages.
phil [Wed, 7 Sep 2005 05:45:05 +0000 (05:45 +0000)]
b=8320
r=phil (HP's patch)
Severity : minor
Frequency : rare
Bugzilla : 8320
Description: lconf incorrectly determined whether two IP networks could talk
Details : In some more complicated routing and multiple-network
configurations, lconf will avoid trying to make a network
connection to a disjoint part of the IP space. It was doing the
math incorrectly for one set of cases.
adilger [Tue, 6 Sep 2005 19:52:55 +0000 (19:52 +0000)]
Branch b1_4
Add more debugging to LASSERT.
b=5359
r=phil (original patch)
adilger [Fri, 2 Sep 2005 21:13:05 +0000 (21:13 +0000)]
Branch b1_4
Remove final vestiges of "groups_upcall". Verified that lmc with:
MDSOPT="--group_upcall=$PWD/../utils/l_getgroups" sh llmount.sh
will create a .xml file with group_upcall stanza and lconf configures
this properly. Previously only tested with "lconf --group_upcall=...".
b=9259
adilger [Fri, 2 Sep 2005 16:57:34 +0000 (16:57 +0000)]
Branch b1_4
Add documentation for the supplementary group upcall in lmc docs.
Fix minor inconsistency between lmc and lconf usage.
b=9259
phil [Fri, 2 Sep 2005 15:56:08 +0000 (15:56 +0000)]
b=7278
reference the bug number in the comment that was added
adilger [Thu, 1 Sep 2005 22:49:25 +0000 (22:49 +0000)]
Branch b1_4
Remove ialloc patch from fc3 kernel series, it should be (and is) in ldiskfs.
r=nathan
adilger [Thu, 1 Sep 2005 18:09:27 +0000 (18:09 +0000)]
Branch b1_4
Description: 2.6 OST async journal commit and locking fix to improve performance
Details : The filter_direct_io()+filter_commitrw_write() journal commits for
2.6 kernels are now async as they already were in 2.4 kernels so
that they can commit concurrently with the network bulk transfer.
For block-allocated files the filter allocation semaphore is held
to avoid filesystem fragmentation during allocation. BKL lock
removed for 2.6 xattr operations where it is no longer needed.
b=7116
r=alex, tested at HP
nathan [Wed, 31 Aug 2005 23:24:27 +0000 (23:24 +0000)]
Branch b1_4
b=none
r=adilger
Add/fix error messages for failing to mount
nkj [Wed, 31 Aug 2005 21:25:46 +0000 (21:25 +0000)]
commited patch submitted to bug 5649 which checks return code from 'losetup'.
adilger [Wed, 31 Aug 2005 08:59:46 +0000 (08:59 +0000)]
Land b_release_1_4_5 onto b1_4 (20050830_1747)
jacob [Tue, 30 Aug 2005 17:48:30 +0000 (17:48 +0000)]
handle running from numbered RC scripts, and exit if no configuration is present
adilger [Mon, 29 Aug 2005 19:34:56 +0000 (19:34 +0000)]
Branch b1_4
Disable test 27 (fail LOV while using OSCs) as it is constantly failing
since we enabled failover OSTs by default.
b=7288
adilger [Fri, 26 Aug 2005 22:40:31 +0000 (22:40 +0000)]
Branch b1_4
Description: Running on many-way SMP OSTs can trigger oops in llcd_send()
Details : A race between allocating a new llcd and re-getting the llcd_lock
in llcd_grab() allowed another thread to get the newly-allocated
llcd. Re-check that the list has an llcd in it before proceeding.
Make the llcd size small enough that it fits into a single page
when we are sending/receiving it.
b=7407
nikita [Wed, 24 Aug 2005 19:27:08 +0000 (19:27 +0000)]
add new testing proglet random-reads.c to benchmark bug 6252 fix.
random-reads.c randomly reads chunks of given size from the given file. See
"random-reads -h" for (ridiculously incomplete) help.
adilger [Wed, 24 Aug 2005 17:45:43 +0000 (17:45 +0000)]
Branch b1_4
Add dump_on_timeout support for client eviction.
adilger [Wed, 24 Aug 2005 05:24:14 +0000 (05:24 +0000)]
Branch b1_4
Fix "service lustre status" on OST_only hosts.
b=7396
adilger [Mon, 22 Aug 2005 10:46:22 +0000 (10:46 +0000)]
Branch b1_4
Remove request for log files, we have some now.
b=5195
adilger [Mon, 22 Aug 2005 10:31:35 +0000 (10:31 +0000)]
Branch b1_4
Fix patch names.
adilger [Fri, 19 Aug 2005 08:48:08 +0000 (08:48 +0000)]
Branch b1_4
Add some extra checks to recursive symlink test.
b=7022
adilger [Fri, 19 Aug 2005 08:40:36 +0000 (08:40 +0000)]
Branch b1_4
Update the kernel patch version for SLES9 kernels.
To be consistent, we revert the old version and apply whatever
the current version is.
adilger [Fri, 19 Aug 2005 08:13:51 +0000 (08:13 +0000)]
Branch b1_4
Description: Racing open + rm can assert client in mdc_set_open_replay_data()
Details : If lookup is in progress on a file that is unlinked we might try
to revalidate the inode and fail in revalidate after lookup is
complete and ll_file_open() enqueues the open again but
it_open_error() was not checking DISP_OPEN_OPEN errors correctly.
b=5479 b=6816
adilger [Fri, 19 Aug 2005 07:26:48 +0000 (07:26 +0000)]
Branch b1_4
Don't override existing ALWAYS_EXCEPT on 2.6.
b=7311
adilger [Fri, 19 Aug 2005 04:13:49 +0000 (04:13 +0000)]
Branch b1_4
Description: chmod/chown may include an extra supplementary group
Details : ll{,u}_mdc_pack_op_data() does not properly initialize the
supplementary group and if none is specified this is used.
Add check to mds_fix_attr() for only ATTR_GID being set.
b=7331
adilger [Fri, 19 Aug 2005 03:48:55 +0000 (03:48 +0000)]
Branch b1_4
Quiet compiler warning for liblustre.
adilger [Thu, 18 Aug 2005 23:20:46 +0000 (23:20 +0000)]
Branch b1_4
Fix valid compiler warning of unset variables.
b=7138
adilger [Thu, 18 Aug 2005 22:00:09 +0000 (22:00 +0000)]
Branch b1_4
Frequency : during shutdown only
Description: shutdown with a failed MDS or OST can cause unmount to hang
Details : Don't resend DISCONNECT messages in ptlrpc_disconnect_import()
if server is down.
b=6827
mjmac [Wed, 17 Aug 2005 21:13:34 +0000 (21:13 +0000)]
Guard against incorrect $RUNAS_ID shenanigans.
adilger [Wed, 17 Aug 2005 18:47:02 +0000 (18:47 +0000)]
Branch b1_4
Don't add primary group to groups list twice.
b=7331
eeb [Wed, 17 Aug 2005 18:35:18 +0000 (18:35 +0000)]
* running on 2 GM nodes @ HP
mjmac [Wed, 17 Aug 2005 18:31:45 +0000 (18:31 +0000)]
Add warning to identify which UID is being used, if UID is not root.
adilger [Wed, 17 Aug 2005 10:59:56 +0000 (10:59 +0000)]
Branch b1_4
Minor message cleanups.
adilger [Wed, 17 Aug 2005 09:06:44 +0000 (09:06 +0000)]
Branch b1_4
NUMA CPU affinity patch for ext3 journal threads. Not in any series yet.
r=eeb
cvs2svn [Wed, 17 Aug 2005 09:06:40 +0000 (09:06 +0000)]
This commit was manufactured by cvs2svn to create branch 'b1_4'.
nikita [Tue, 16 Aug 2005 15:51:05 +0000 (15:51 +0000)]
fix for bug 4843
mjmac [Tue, 16 Aug 2005 14:01:29 +0000 (14:01 +0000)]
Add command-line switch to specify number of iterations. Default is still
1000 loops.
adilger [Tue, 16 Aug 2005 00:38:25 +0000 (00:38 +0000)]
Branch b1_4
Print units of jiffies in brw_stats output
b=7116
adilger [Mon, 15 Aug 2005 22:53:38 +0000 (22:53 +0000)]
Branch b1_4
Fix swabbing of the total inodes count for big-endian clients.
r=behlendorf
adilger [Mon, 15 Aug 2005 21:54:11 +0000 (21:54 +0000)]
Branch b1_4
Remove unused function, merge from HEAD.
adilger [Mon, 15 Aug 2005 21:49:14 +0000 (21:49 +0000)]
Branch b1_4
Make iozone O_DIRECT and multi-thread tests separate.
adilger [Mon, 15 Aug 2005 21:27:28 +0000 (21:27 +0000)]
Branch b1_4
Fix valid compile warning.
b=6902
adilger [Mon, 15 Aug 2005 21:16:54 +0000 (21:16 +0000)]
Branch b1_4
Bump the minimal acceptable kernel patch version a bit.
This is low enough that a 1.2.4-patched kernel could still run 1.4 with
the caveats that any of the newer bugs wouldn't be fixed.
b=7255
adilger [Mon, 15 Aug 2005 21:08:10 +0000 (21:08 +0000)]
Branch b1_4
Don't deref NULL dirent if allocation failure, merge from HEAD.
adilger [Mon, 15 Aug 2005 20:05:24 +0000 (20:05 +0000)]
Branch b1_4
Description: service threads change working directory to that of init
Details : Starting lustre service threads may pin the working directory
of the parent thread, making that filesystem busy. Threads
now change to the working directory of init to avoid this.
b=6273
green [Mon, 15 Aug 2005 10:35:00 +0000 (10:35 +0000)]
Branch: b1_4
b=7198
r=adilger
Do not double-query file size from OSTs during stat(2). This might slow
down things considerably if somebody actively writes to stat()ed files too.
jacob [Fri, 12 Aug 2005 18:42:30 +0000 (18:42 +0000)]
fix incorrect python
green [Fri, 12 Aug 2005 12:00:25 +0000 (12:00 +0000)]
Branch: b1_4
b=6902
r=adilger
Evict clients by NID from both MDS and active OSTs
adilger [Thu, 11 Aug 2005 22:19:14 +0000 (22:19 +0000)]
Branch b1_4
Description: liblustre clients evicted if not contacting servers
Details : Don't put liblustre clients into the ping_evictor list, so
they will not be evicted by the pinger ever. Tested by Sandia.
b=7137
green [Thu, 11 Aug 2005 16:41:17 +0000 (16:41 +0000)]
b=6455
r=adilger
Correctly fill struct dirent, d_off there should hold offset to NEXT dir entry,
not current one.
green [Thu, 11 Aug 2005 10:47:53 +0000 (10:47 +0000)]
b=7278
r=adilger
Trick libsysio not to enforce O_EXCL open flag. MDS will enforce it for us.
adilger [Thu, 11 Aug 2005 07:49:55 +0000 (07:49 +0000)]
Branch b1_4
Fix now-incorrect check in test 42 since OSTs are failover and not failout.
b=7287
adilger [Thu, 11 Aug 2005 05:56:42 +0000 (05:56 +0000)]
Branch b1_4
Disable test 67 for non-local MDS - it needs to know if there is a group
upcall, and we can't tell otherwise.
adilger [Thu, 11 Aug 2005 00:41:26 +0000 (00:41 +0000)]
Branch b1_4
Allow setting ldlm_timeout smaller than 10s.
b=7201
adilger [Thu, 11 Aug 2005 00:24:19 +0000 (00:24 +0000)]
Branch b1_4
Add check for NULL imp_conn_current. It isn't clear whether this should
always be non-NULL (in which case locking is bad, though we can't really
expect to hold imp_lock over the whole operation) or whether it is OK
to handle the NULL case gracefully.
b=7269
r=jacob
jacob [Wed, 10 Aug 2005 20:44:11 +0000 (20:44 +0000)]
fix dist
adilger [Wed, 10 Aug 2005 19:36:08 +0000 (19:36 +0000)]
Land b1_4_bug3389 onto b1_4 (20050810_1124)
Description: Add support for supplementary groups on the MDS.
Details : The MDS has an upcall /proc/fs/lustre/mds/{mds}/group_upcall
(set to /usr/sbin/l_getgroups if enabled) which will do MDS-side
lookups for user supplementary groups into a cache.
b=3389, b=6253
Description: Add support for compiling against Cray portals.
Details : Conditional compiling for some areas that are different
on Cray Portals.
adilger [Wed, 10 Aug 2005 08:35:42 +0000 (08:35 +0000)]
Land b_release_1_4_4 onto b1_4 (20050810_0211)
Description: Heavy concurrent write and delete load may cause data corruption.
Details : It was possible under high-load situations to have an extent
metadata block in the block device cache from a just-unlinked
file overwrite a newly-allocated data block. We now unmap any
metadata buffers that alias just-allocated data blocks.
b=6998
Description: lconf+mke2fs fail when formatting filesystem with > 77 stripes
Details : lconf specifies an inode size of 4096 bytes when the default
stripe_count is larger than 77. This conflicts with the default
inode density of 1 per 4096 bytes. Allocate smaller inodes in
this case to avoid pinning too much memory for large EAs.
b=7241
adilger [Wed, 10 Aug 2005 08:02:40 +0000 (08:02 +0000)]
Branch b1_4
Make the "return grant" constant dependent upon RPC size, not hard coded.
Currently the values are the same.
b=7138
adilger [Wed, 10 Aug 2005 08:00:27 +0000 (08:00 +0000)]
Branch b1_4
Description: Client will block write RPCs if not enough grant
Details : If a client has max_dirty_mb smaller than max_rpcs_in_flight,
then the client will block writes while waiting for another RPC
to complete instead of consuming its dirty limit. With change
we get improved performance when max_dirty_mb is small.
adilger [Wed, 10 Aug 2005 07:36:45 +0000 (07:36 +0000)]
Branch b1_4
Description: If an OST is inactive, its locks might reference stale inodes.
Details : lov_change_cbdata() must iterate over all namespaces, even if
they are inactive to clear inode references from the lock.
Otherwise we get stale locks referencing random memory (possibly
not even an inode) when the locks drop off the LRU.
b=7039
r=alex
mjmac [Tue, 9 Aug 2005 20:29:24 +0000 (20:29 +0000)]
Added an info() sub for non-warning, non-error messages, and modified
warning() as well. Both now use stderr for output.
adilger [Tue, 9 Aug 2005 20:24:25 +0000 (20:24 +0000)]
Branch b1_4
Description: Mounting an ldiskfs file system with mballoc may crash OST node.
Details : ldiskfs mballoc code may reference an uninitialized buddy struct
at startup during orphan unlinking. Instead, skip buddy update
before setup, as it will be regenerated after recovery is complete.
b=7264
r=alex
nikita [Mon, 8 Aug 2005 22:05:56 +0000 (22:05 +0000)]
liblustre doesn't have PageLocked
cliffw [Mon, 8 Aug 2005 16:36:16 +0000 (16:36 +0000)]
Added warning message for change in stripe_cnt 0 policy
nikita [Mon, 8 Aug 2005 16:05:50 +0000 (16:05 +0000)]
make it compilable when CONFIG_PTRACK is not defined (like on UML)
adilger [Sun, 7 Aug 2005 05:01:41 +0000 (05:01 +0000)]
Branch b1_4
Update lmc documntation to match recent changes.
b=7140
devesh [Fri, 5 Aug 2005 21:03:00 +0000 (21:03 +0000)]
Added 26f for testing removal of directory which contains recursive symlink.
nikita [Fri, 5 Aug 2005 17:21:13 +0000 (17:21 +0000)]
assert that page is locked when IO is started by the osc layer
nikita [Fri, 5 Aug 2005 16:18:36 +0000 (16:18 +0000)]
add description of struct ll_readahead_state fields
adilger [Thu, 4 Aug 2005 22:39:55 +0000 (22:39 +0000)]
Branch b1_4
Add ext3-ialloc patch to Cray patch series, avoids poor inode placement and
filesystem fragmentation on the OSTs.
b=4466
r=alex
adilger [Thu, 4 Aug 2005 19:06:17 +0000 (19:06 +0000)]
Branch b1_4
Allow the "created" parameter to be NULL. We don't actually use this data
for anything, and the current obdfilter code just passes in a dummy array
to make the API happy. This leads to confusion when modifying the code.
b=7049, b=6998
adilger [Thu, 4 Aug 2005 18:42:46 +0000 (18:42 +0000)]
Branch b1_4
Description: Unusual file access patterns on the MDS may result in inode
data being lost in very rare circumstances.
Details : Bad interaction between the ea-in-inode patch and the "no-read"
code in the 2.6 kernel caused the inode and/or EA data not to
be read from disk, causing single-file corruption.
b=6866, b=6869
r=alex
jacob [Thu, 4 Aug 2005 18:38:49 +0000 (18:38 +0000)]
remove unneeded argument
jacob [Thu, 4 Aug 2005 18:37:52 +0000 (18:37 +0000)]
name the cluster using a sort list of node names
jacob [Thu, 4 Aug 2005 18:11:50 +0000 (18:11 +0000)]
remove stale series
jacob [Thu, 4 Aug 2005 18:06:06 +0000 (18:06 +0000)]
b=7165
b=7224
- initial support for creating clumanager xml files from lustre xml
- support for using lustre init script from clumanager
- more changes to init scripts from scripps
cliffw [Wed, 3 Aug 2005 19:18:30 +0000 (19:18 +0000)]
added test of /proc/fs/lustre/health_check
jacob [Tue, 2 Aug 2005 20:29:58 +0000 (20:29 +0000)]
update for current errata
jacob [Tue, 2 Aug 2005 19:05:40 +0000 (19:05 +0000)]
properly update kernel configs for sles 9 sp 2
adilger [Tue, 2 Aug 2005 05:52:43 +0000 (05:52 +0000)]
Branch b1_4
Fix indenting problem in lconf causing a use-before-assignment error.
Gotta love python.
b=7221
adilger [Sat, 30 Jul 2005 00:28:01 +0000 (00:28 +0000)]
Branch b1_4
Add ialloc patch to fc3 series.
b=4644
phil [Sat, 30 Jul 2005 00:05:48 +0000 (00:05 +0000)]
b=6266 (oleg's patch)
Severity : minor
Bugzilla : 6266
Frequency : rare (liblustre)
Description: fsx running with liblustre complained that using truncate() to
extend the file doesn't work. This patch corrects that issue.
Details : This is the liblustre equivalent of the fix for bug 6196. Fixes
ATTR_SIZE and lsm use in llu_setattr_raw.
adilger [Fri, 29 Jul 2005 23:58:00 +0000 (23:58 +0000)]
Branch b1_4
Add ialloc patch to 2.4.29 series.
b=4644
jacob [Fri, 29 Jul 2005 23:54:55 +0000 (23:54 +0000)]
update patches and scripts for SLES 9 SP2
jacob [Fri, 29 Jul 2005 23:54:32 +0000 (23:54 +0000)]
update patches and scripts for SLES 9 SP2
wangdi [Fri, 29 Jul 2005 14:06:12 +0000 (14:06 +0000)]
Branch: b_hd_crypto
minor fixes about crypto api
ericm [Thu, 28 Jul 2005 18:31:21 +0000 (18:31 +0000)]
land minor fixes from b_hd_sec:
* "EA" is used for lustre internal metadata ea. so change user ea's name to
"XATTR", including ACLs.
* remove a unnecessary param from md_getattr().
* various more fixes.
phil [Thu, 28 Jul 2005 00:03:16 +0000 (00:03 +0000)]
Make the ChangeLog crystal clear that although flock has landed, it's not yet
supported
phil [Wed, 27 Jul 2005 23:48:35 +0000 (23:48 +0000)]
add missing entry for bug 6993
alex [Wed, 27 Jul 2005 18:54:34 +0000 (18:54 +0000)]
b=7200
- protect lock->l_resource from concurrent ldlm_lock_change_resource()
adilger [Wed, 27 Jul 2005 18:18:31 +0000 (18:18 +0000)]
Branch b1_4_bug3389
Fix portals_nid2str() to print something reasonable for Cray NALs.
b=6902
r=bogl