Whamcloud - gitweb
fs/lustre-release.git
16 years ago- repair lmv.sh - it was broken for months
alex [Fri, 5 Aug 2005 20:34:14 +0000 (20:34 +0000)]
- repair lmv.sh - it was broken for months

16 years agocomment for ll_readahead_state and assertion in osc_send_oap_request() from b1_4
nikita [Fri, 5 Aug 2005 18:12:10 +0000 (18:12 +0000)]
comment for ll_readahead_state and assertion in osc_send_oap_request() from b1_4

16 years agoBranch: HEAD
wangdi [Fri, 5 Aug 2005 14:48:14 +0000 (14:48 +0000)]
Branch: HEAD
typo fix after land

16 years agoBranch: HEAD
wangdi [Fri, 5 Aug 2005 14:45:34 +0000 (14:45 +0000)]
Branch: HEAD
land crypto api

16 years ago- warn about the race in filter_intent_policy(), just to make sure it
alex [Fri, 5 Aug 2005 09:12:57 +0000 (09:12 +0000)]
- warn about the race in filter_intent_policy(), just to make sure it
  relates to errors mds gets sometimes

16 years agob=7039
alex [Thu, 4 Aug 2005 21:49:13 +0000 (21:49 +0000)]
b=7039

 - don't skip locks on inactive imports

16 years agob=7039
alex [Thu, 4 Aug 2005 19:14:37 +0000 (19:14 +0000)]
b=7039

 - dump additional debug info

16 years ago- missing \n
alex [Thu, 4 Aug 2005 13:32:51 +0000 (13:32 +0000)]
- missing \n

16 years agob=7039
alex [Thu, 4 Aug 2005 13:04:54 +0000 (13:04 +0000)]
b=7039

 - check for llite inode magic

16 years ago- no more flooding
alex [Thu, 4 Aug 2005 12:06:55 +0000 (12:06 +0000)]
- no more flooding

16 years agob=7039
alex [Thu, 4 Aug 2005 11:40:59 +0000 (11:40 +0000)]
b=7039

 - more debugging info

16 years agob=7049
alex [Thu, 4 Aug 2005 11:33:40 +0000 (11:33 +0000)]
b=7049

 - remove needless debugging checks

16 years agogive a warnning if somebody don't pass lustre id to ost, it should be
niu [Thu, 4 Aug 2005 09:45:49 +0000 (09:45 +0000)]
give a warnning if somebody don't pass lustre id to ost, it should be
error assertion in future.

16 years agoFix of careless (miss ;;)
liangzhen [Thu, 4 Aug 2005 09:00:13 +0000 (09:00 +0000)]
Fix of careless (miss ;;)

16 years agostore lustre id in ost object's EA.
niu [Thu, 4 Aug 2005 08:46:29 +0000 (08:46 +0000)]
store lustre id in ost object's EA.

16 years agoCFLAGS for compiling in Tiger
liangzhen [Thu, 4 Aug 2005 07:56:21 +0000 (07:56 +0000)]
CFLAGS for compiling in Tiger

16 years agob=7049
alex [Wed, 3 Aug 2005 21:56:47 +0000 (21:56 +0000)]
b=7049

 - try to unmap buffers from bd-mapping

16 years agob=7049
alex [Wed, 3 Aug 2005 13:29:38 +0000 (13:29 +0000)]
b=7049

 - it's enough to check for block number

16 years agob=7049
alex [Wed, 3 Aug 2005 12:18:42 +0000 (12:18 +0000)]
b=7049

 - it's better to unmap buffers holding no truncate sem
   note: unmap_underlying_metadata() is commented out to see whether
         a race happens at all

16 years agob=7049
alex [Wed, 3 Aug 2005 12:04:15 +0000 (12:04 +0000)]
b=7049

 - check for aliases in bd-mapping for writes only

16 years agob=7049
alex [Wed, 3 Aug 2005 10:20:08 +0000 (10:20 +0000)]
b=7049

 - let's check whether dirty pages/buffers in bd-mapping

16 years agob=7237
alex [Wed, 3 Aug 2005 07:34:22 +0000 (07:34 +0000)]
b=7237

 - missed wake_up() has been returned
 - few CDEBUG() have changed their level to D_ERROR to simplify debugging

16 years agoMerge b1_4_bug3389 from b1_4 (20050729_0312)
adilger [Tue, 2 Aug 2005 23:29:58 +0000 (23:29 +0000)]
Merge b1_4_bug3389 from b1_4 (20050729_0312)
Description: Mounting a Lustre file system on a node running as an OST could
             lead to deadlocks
Details    : OSTs now allocate memory needed to write out data at
             startup, instead of when needed, to avoid having to
             allocate memory in possibly low memory situations.
             Specifically, if the file system is mounted on on OST,
             memory pressure could force it to try to write out data,
             which it needed to allocate memory to do.  Due to the low
             memory, it would be unable to do so and the node would
             become unresponsive.
b=6514, b=5137

Description: Addition of lconf --service command line option
Details    : lconf now accepts a '--service <arg>' option, which is
             shorthand for 'lconf --group <arg> --select <arg>=<hostname>'
b=7015

Description: Failover mode is now the default for OSTs.
Details    : By default, OSTs will now run in failover mode.  To return to
             the old behaviour, add '--failout' to the lmc line for OSTs.
b=6101

Description: Health checks are now provided for MDS and OSTs
Details    : Additional detailed health check information on MSD and OSTs
             is now provided through the procfs health_check value.
b=1693

Description: Disk fragmentation on the OSTs could eventually cause slowdowns
             after numerous create/delete cycles
Details    : The ext3 inode allocation policy would not allocate new inodes
             very well on the OSTs because there are no new directories
             being created.  Instead we look for groups with free space if
             the parent directories are nearly full.
b=4466

Description: Network or server problems during mount may cause partially
             mounted clients instead of returning an error.
Details    : The config llog parsing code may overwrite the error return
             code during mount error handling, returning success instead
             of an error.
b=6302

16 years agob=3984
alex [Tue, 2 Aug 2005 19:46:23 +0000 (19:46 +0000)]
b=3984

 - a reference on resource must be taked under spinlock

16 years agob=7039
alex [Tue, 2 Aug 2005 18:46:35 +0000 (18:46 +0000)]
b=7039

 - concurrent lock cancelling might find inode dead while it's not yet

16 years agob=7231
alex [Tue, 2 Aug 2005 17:34:20 +0000 (17:34 +0000)]
b=7231

 - asynchronous commit in obdfilter/2.6

16 years agob=7231
alex [Tue, 2 Aug 2005 15:43:33 +0000 (15:43 +0000)]
b=7231

 - wait on bio's completion after commit. to evaluate performance impact

16 years agob=7039
alex [Tue, 2 Aug 2005 09:09:02 +0000 (09:09 +0000)]
b=7039

 - stop progress and dump a log for debugging purposes

16 years ago- b_size_on_mds landed on HEAD:
alex [Sun, 31 Jul 2005 12:43:22 +0000 (12:43 +0000)]
- b_size_on_mds landed on HEAD:

  5984 - oops in mdc_set_lock_data
         we must drop lookup lock in ll_clear_inode()
  7216 - abort due to missed clients
 replay already committed (open/close rpcs) requests
 even if some clients haven't connected in time and
 got evicted
  7057 - sanity: test_66: /mnt/lustre/f66 blocks 0 < 8
         send close rpc only when all dirty pages are written
 (otherwise i_blocks isn't valid)
  6817 - don't update attrs. on mds if the client hasn't modified file
         using openhandle being closed

16 years agoland b_hd_remote_acl: support get/set ACL from remote client.
ericm [Sun, 31 Jul 2005 04:49:34 +0000 (04:49 +0000)]
land b_hd_remote_acl: support get/set ACL from remote client.

16 years agoupdate 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

16 years agoBranch: b_hd_crypto
wangdi [Fri, 29 Jul 2005 14:06:12 +0000 (14:06 +0000)]
Branch: b_hd_crypto
minor fixes about crypto api

16 years agoland minor fixes from b_hd_sec:
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.

16 years agoadd missing entry for bug 6993
phil [Wed, 27 Jul 2005 23:48:35 +0000 (23:48 +0000)]
add missing entry for bug 6993

16 years agob=7200
alex [Wed, 27 Jul 2005 18:54:34 +0000 (18:54 +0000)]
b=7200

 - protect lock->l_resource from concurrent ldlm_lock_change_resource()

16 years agoBranch b1_4_bug3389
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

16 years agob=5210
alex [Wed, 27 Jul 2005 17:20:02 +0000 (17:20 +0000)]
b=5210
r=phil(actual patch)

A workaround for the crash in ll_unhash_aliases on 2.6.
The 2.6 lookup code is now lockless (thanks to the RCU machinations), which
allows it to race with the unhashing in ll_unhash_aliases. The 2.6 __d_drop
doesn't leave the list head ->pprev in a safe state, so the next pass through
this loop (which tries to re-hash it after we lose the dput race), would crash.
For the record, it looks like the kernel's own d_prune_aliases is broken in
the very same way, except that instead of crashing immediately, it just leaves
the d_hash pointers poisoned for someone else to crash on.

16 years agoBranch:b_hd_crypto
wangdi [Wed, 27 Jul 2005 09:28:29 +0000 (09:28 +0000)]
Branch:b_hd_crypto
update crypto api code

16 years agob=7017
alex [Sat, 23 Jul 2005 20:55:29 +0000 (20:55 +0000)]
b=7017

 - atomic_dec_and_test() with subsequent spin_lock() against the list
   is racy. atomic_dec_and_lock() must be used. otherwise other cpu
   can hit the race window, increment refcount, drop it again and we
   get to the situation when both cpus are freeing the structure

16 years agosome env (cnbuild) can't build without explicitly include
ericm [Fri, 22 Jul 2005 17:23:24 +0000 (17:23 +0000)]
some env (cnbuild) can't build without explicitly include
errno.h

16 years agoland b_hd_pag: rudiment support for PAG.
ericm [Thu, 21 Jul 2005 17:17:36 +0000 (17:17 +0000)]
land b_hd_pag: rudiment support for PAG.

16 years agoCopyright is now License in RPM spec files.
brian [Thu, 21 Jul 2005 14:01:53 +0000 (14:01 +0000)]
Copyright is now License in RPM spec files.

16 years agoquiet
jacob [Wed, 20 Jul 2005 21:24:18 +0000 (21:24 +0000)]
quiet

16 years agoShould allow the uml_switch to be used if wanted.
brian [Wed, 20 Jul 2005 18:06:59 +0000 (18:06 +0000)]
Should allow the uml_switch to be used if wanted.

16 years agoBranch: HEAD
wangdi [Wed, 20 Jul 2005 12:50:30 +0000 (12:50 +0000)]
Branch: HEAD
remove redundant set_md in mds_reint_create and cmobd_reint_create

16 years agoBranch: HEAD
wangdi [Wed, 20 Jul 2005 12:34:28 +0000 (12:34 +0000)]
Branch: HEAD
add setxattr kml post hook for reint set md to master mds.

16 years agoBranch: HEAD
wangdi [Tue, 19 Jul 2005 12:56:14 +0000 (12:56 +0000)]
Branch: HEAD
should reset mcd_last_close_xid to 0 when reconnect.
b:6102
r:alex

16 years ago- need couple routines to be accessible from modules for size-on-mds feature
alex [Tue, 19 Jul 2005 11:35:38 +0000 (11:35 +0000)]
- need couple routines to be accessible from modules for size-on-mds feature

16 years agoBranch: HEAD
wangdi [Sat, 16 Jul 2005 07:24:10 +0000 (07:24 +0000)]
Branch: HEAD
add kernel-2.6.10-suse-opteron.config for opteron arch

16 years agoi stand corrected
jacob [Fri, 15 Jul 2005 20:40:56 +0000 (20:40 +0000)]
i stand corrected

16 years agothis is the fix i intended
jacob [Fri, 15 Jul 2005 20:06:06 +0000 (20:06 +0000)]
this is the fix i intended

16 years agor=adilger,nic
mjmac [Fri, 15 Jul 2005 19:52:29 +0000 (19:52 +0000)]
r=adilger,nic

stage() should return 0 if $STAGEDIR is not set, not 1.

16 years agoBranch: HEAD
wangdi [Fri, 15 Jul 2005 07:11:18 +0000 (07:11 +0000)]
Branch: HEAD
add fid and group in kml for cmobd test

16 years agomove AC_SUBST into lustre-build.m4 (b=7126)
jacob [Fri, 15 Jul 2005 04:34:12 +0000 (04:34 +0000)]
move AC_SUBST into lustre-build.m4 (b=7126)

16 years agoadd --stage option to lbuild
jacob [Thu, 14 Jul 2005 23:04:14 +0000 (23:04 +0000)]
add --stage option to lbuild

16 years agorevert a broken change from an earlier fix
jacob [Thu, 14 Jul 2005 23:00:40 +0000 (23:00 +0000)]
revert a broken change from an earlier fix

16 years agoBranch b1_4
adilger [Thu, 14 Jul 2005 22:53:45 +0000 (22:53 +0000)]
Branch b1_4
Use a wildcard for /usr/share/lustre.

16 years agojust remove unused macro as i can't get it to work right anyway
jacob [Thu, 14 Jul 2005 22:25:09 +0000 (22:25 +0000)]
just remove unused macro as i can't get it to work right anyway

16 years agomy mind fell out of m4 mode for a moment
jacob [Thu, 14 Jul 2005 22:02:43 +0000 (22:02 +0000)]
my mind fell out of m4 mode for a moment

16 years agorestripe snmp to b1_4 so that andreas will add the right globs later
jacob [Thu, 14 Jul 2005 21:50:07 +0000 (21:50 +0000)]
restripe snmp to b1_4 so that andreas will add the right globs later

16 years agoremove harmless error message when snmp is not present
jacob [Thu, 14 Jul 2005 21:47:27 +0000 (21:47 +0000)]
remove harmless error message when snmp is not present

16 years agoadd support for the snmp module.
jacob [Thu, 14 Jul 2005 21:10:04 +0000 (21:10 +0000)]
add support for the snmp module.

16 years agoThis commit was generated by cvs2svn to compensate for changes in r49775,
jacob [Thu, 14 Jul 2005 21:00:40 +0000 (21:00 +0000)]
This commit was generated by cvs2svn to compensate for changes in r49775,
which included commits to RCS files with non-trunk default branches.

16 years agoInitial import of lustre-snmp module.
jacob [Thu, 14 Jul 2005 21:00:40 +0000 (21:00 +0000)]
Initial import of lustre-snmp module.

16 years agooops
jacob [Thu, 14 Jul 2005 17:11:21 +0000 (17:11 +0000)]
oops

16 years ago - skip modules lacking a tag
jacob [Thu, 14 Jul 2005 17:02:25 +0000 (17:02 +0000)]
 - skip modules lacking a tag
 - print a summary of failed updates/checkouts at the end
 - also return an error status in this case

16 years agosort correctly
jacob [Thu, 14 Jul 2005 16:23:21 +0000 (16:23 +0000)]
sort correctly

16 years agoRevert Tom's commit. It breaks recovery.
brian [Thu, 14 Jul 2005 14:25:34 +0000 (14:25 +0000)]
Revert Tom's commit.  It breaks recovery.
r=alex
r=umka

16 years agoBranch HEAD
wangdi [Thu, 14 Jul 2005 12:46:48 +0000 (12:46 +0000)]
Branch HEAD
add fid and group in kml log

16 years agospin_is_locked() always return false in UP kernel.
ericm [Thu, 14 Jul 2005 03:05:21 +0000 (03:05 +0000)]
spin_is_locked() always return false in UP kernel.

16 years agominor: verbose on sec register.
ericm [Wed, 13 Jul 2005 22:22:17 +0000 (22:22 +0000)]
minor: verbose on sec register.

16 years agoconf-sanity: don't start/stop lsvcgssd/lgssd frequently.
ericm [Wed, 13 Jul 2005 22:08:59 +0000 (22:08 +0000)]
conf-sanity: don't start/stop lsvcgssd/lgssd frequently.

16 years agosupply nettype to conf-sanity 5d
ericm [Wed, 13 Jul 2005 21:53:00 +0000 (21:53 +0000)]
supply nettype to conf-sanity 5d
r=6327

16 years ago- changes in cmobd.sh - there should not be cache LMV
yury [Wed, 13 Jul 2005 16:23:33 +0000 (16:23 +0000)]
- changes in cmobd.sh - there should not be cache LMV
- fixes in lmc

16 years ago- remove out of date warning
alex [Wed, 13 Jul 2005 11:54:30 +0000 (11:54 +0000)]
- remove out of date warning

16 years agoSpinlock twice.
liangzhen [Wed, 13 Jul 2005 10:20:05 +0000 (10:20 +0000)]
Spinlock twice.

16 years agoland b_hd_sec: perm/acl authorization for remote users.
ericm [Wed, 13 Jul 2005 00:59:04 +0000 (00:59 +0000)]
land b_hd_sec: perm/acl authorization for remote users.

16 years ago- many cleanups with case like foo (). Replaced by foo()
yury [Tue, 12 Jul 2005 13:05:52 +0000 (13:05 +0000)]
- many cleanups with case like foo (). Replaced by foo()

- in cmobd fixed init of EA size for LOV with multiple OSTs as master. As there is no another way MD layer is getting asked for correct lovdesc, as it is needed in some cases. Thus, added handler for "lovdesc" nfo key to all MD devices.For instance, cmobd asks LMV for lov desc, it asks first MDS and MDS has it from LOV anyway.

- fixed possible LMV obd corruption in cmobd_init_ea_size() where it unconditionaly accessed u.cli things abd modified them whereas here was LMV.

- removed obd_getready() as it is not needed anymore. It was needed earlier for CMOBD to get LMV ready before asking rootid, etc. Currently alternative way is found. lmv_get_info() is just taking care of LMV targets readiness it self.

- fixed layering violations and possible oopses in cmobd data code where master export was unconditionaly expected as LOV and accessed as LOV what could be wrong in the case of using OSC. OSC using with no LOV is deprecated, but in principle is possible.

- in LMV module many functions are added by static specificator as they are not used outside of their object files.

16 years ago- fixed possible access to of bounds in cmobd
yury [Tue, 12 Jul 2005 06:14:49 +0000 (06:14 +0000)]
- fixed possible access to of bounds in cmobd

16 years ago- cleanups in cmobd and others:
yury [Mon, 11 Jul 2005 17:18:36 +0000 (17:18 +0000)]
- cleanups in cmobd and others:
  - change LUSTRE_MDC_NAME (and others) to form like OBD_MDC_DEVICENAME to be coherent with
    OBD_LMV_DEVICENAME and others. Because some OBDs used one form and others used another one.

  - removed different guesses of OBD types from name of some functions in cmobd. Say
    cmobd_find_master() now is called cmobd_find_obd() as all it does is finds obd
    by uuid and may be used not only for finding master. Also cmobd_lmv_obd() is renamed to
    cmobd_md_obd() etc. Improved function (a bit smaller of code) wich looks for master.

- in memory leaks detector do not remove memory track in case of freeing memory chunk different
  of size. marking it instead as such to show in final log of leaks with special hint.

- fixes in lconf. COBD and CMOBD should use LMV and LOV devices with diffrent names to avoid
  issues in the case when both they run on the same node (like in test 45) and try to setup
  and clanup the same device twise (from lconf).

16 years ago- plugin handler should pass EXIT call in any case.
tappro [Mon, 11 Jul 2005 16:41:50 +0000 (16:41 +0000)]
- plugin handler should pass EXIT call in any case.

16 years agob=6817
alex [Mon, 11 Jul 2005 11:43:43 +0000 (11:43 +0000)]
b=6817
r=niu

 - memleak in lov_match() fixed: it didn't handle LDLM_FL_TEST_LOCK well

16 years ago- memleak in mds_create_objects():
alex [Mon, 11 Jul 2005 11:28:09 +0000 (11:28 +0000)]
- memleak in mds_create_objects():
  we should free ids array if an error is returned

16 years ago- dump log if memory leak observed
alex [Mon, 11 Jul 2005 11:05:19 +0000 (11:05 +0000)]
- dump log if memory leak observed

16 years agomove the initiation of fo_denylist/fo_denylist_lock from filter_setup to filter_commo...
wangchao [Mon, 11 Jul 2005 04:50:14 +0000 (04:50 +0000)]
move the initiation of fo_denylist/fo_denylist_lock from filter_setup to filter_common_setup

16 years agob=3984
alex [Sun, 10 Jul 2005 23:10:17 +0000 (23:10 +0000)]
b=3984

 - b_ldlm_newlocking landed. we need it to pass some CMD2 performance tests
   NOTE: the new locking rules introduced by the patch are still experimental!

16 years ago- drop reference for open replay data in mdc_commit_open() instead of
alex [Sun, 10 Jul 2005 23:05:06 +0000 (23:05 +0000)]
- drop reference for open replay data in mdc_commit_open() instead of
  mdc_commit_close() as in some cases we may have no close at all

16 years ago- out of date warning has been removed
alex [Sun, 10 Jul 2005 22:12:32 +0000 (22:12 +0000)]
- out of date warning has been removed

16 years ago- make 512-bytes inode size by default: with new EAs (fid, parent) we just
alex [Sun, 10 Jul 2005 22:07:43 +0000 (22:07 +0000)]
- make 512-bytes inode size by default: with new EAs (fid, parent) we just
  can't fit 256-bytes inodes

16 years ago- convert CWARN() to CDEBUG()
alex [Sun, 10 Jul 2005 20:38:42 +0000 (20:38 +0000)]
- convert CWARN() to CDEBUG()

16 years ago- we don't need deadlock-monitor in general series
alex [Sun, 10 Jul 2005 19:53:09 +0000 (19:53 +0000)]
- we don't need deadlock-monitor in general series

16 years agob=6817
alex [Sun, 10 Jul 2005 19:52:22 +0000 (19:52 +0000)]
b=6817

 - b_size_on_mds landed (prototype for CMD2 w/o recovery support on OST)

16 years ago- minor debug level change to have only DEBUG_REQ with D_RPCTRACE
alex [Sun, 10 Jul 2005 18:36:24 +0000 (18:36 +0000)]
- minor debug level change to have only DEBUG_REQ with D_RPCTRACE

16 years ago- update enqueue time upon lock conversion otherwise we get too long
alex [Sun, 10 Jul 2005 18:31:06 +0000 (18:31 +0000)]
- update enqueue time upon lock conversion otherwise we get too long
  enqueue wait time warning

16 years agomds_open() better to check permission even it's open request for device
ericm [Sun, 10 Jul 2005 05:17:51 +0000 (05:17 +0000)]
mds_open() better to check permission even it's open request for device
files, otherwise client vfs will check that and failed siliently which
cause request leak. (client's interaction with vfs seems not perfect --
can't handle all error cases)

16 years agotry to be a little bit more nice to remote use: as long as it be mapped
ericm [Sat, 9 Jul 2005 19:33:55 +0000 (19:33 +0000)]
try to be a little bit more nice to remote use: as long as it be mapped
to an local user, the supp groups of target usr will take part in fs
access checking.

16 years agoadjust gss timeout valud, don't use fixed value.
ericm [Sat, 9 Jul 2005 16:04:16 +0000 (16:04 +0000)]
adjust gss timeout valud, don't use fixed value.

16 years agoBranch HEAD
wangdi [Sat, 9 Jul 2005 03:19:27 +0000 (03:19 +0000)]
Branch HEAD
fix a typo in sanity-cmobd.sh

16 years ago- putting file fid into o_inline for cases when crow object may be created. It will...
yury [Fri, 8 Jul 2005 15:23:42 +0000 (15:23 +0000)]
- putting file fid into o_inline for cases when crow object may be created. It will be used later for storing into OSS object EA what is needed for quota.

16 years agob=6285
tappro [Fri, 8 Jul 2005 13:57:25 +0000 (13:57 +0000)]
b=6285
- remove few debug and LBUG()