Whamcloud - gitweb
fs/lustre-release.git
21 years agofor snaity.sh test_44a
jerrifer [Tue, 16 Sep 2003 07:48:15 +0000 (07:48 +0000)]
for snaity.sh test_44a

21 years agoThe easiest and least controversial part of the patch to eliminate the
phil [Tue, 16 Sep 2003 03:16:23 +0000 (03:16 +0000)]
The easiest and least controversial part of the patch to eliminate the
gymnastics associated with setting l_completion_ast on the client.

We used to set l_completion_ast when the lock is created, so that a
completion AST which arrives before the RPC reply will still be
handled correctly.  Later, we would NULL it out so that lock_enqueue's
call to grant_lock didn't needlessly execute again.  This was
race-prone and stupid.

Now ldlm_grant_lock takes a flag, so we can suppress needless calls
without this mess.

21 years ago* pruned LWT output to account for wrapped cpus
eeb [Mon, 15 Sep 2003 21:55:28 +0000 (21:55 +0000)]
*    pruned LWT output to account for wrapped cpus

*    calculate get_cycles() offsets per-cpu

21 years ago- port of extents onto 2.4.20-vanilla
alex [Mon, 15 Sep 2003 16:29:17 +0000 (16:29 +0000)]
- port of extents onto 2.4.20-vanilla

NOTE: tested by dbench 4/8 in UML only

21 years ago* Added LWT facility
eeb [Mon, 15 Sep 2003 14:02:03 +0000 (14:02 +0000)]
*   Added LWT facility

21 years ago- vanilla-2.4.20 series did't build for i386 arch w/o this patch
alex [Mon, 15 Sep 2003 06:32:05 +0000 (06:32 +0000)]
- vanilla-2.4.20 series did't build for i386 arch w/o this patch

21 years ago- raw lookup support in vfs and ext3
alex [Mon, 15 Sep 2003 06:27:10 +0000 (06:27 +0000)]
- raw lookup support in vfs and ext3

NOTE: patch could be applied for all supported 2.4 kernels

21 years ago- lstripe code for b_devel
braam [Mon, 15 Sep 2003 05:04:04 +0000 (05:04 +0000)]
- lstripe code for b_devel
  - error handling is not complete yet (sanity 27f fails)
  - lfind interaction (27g 34's) untested

21 years ago[liblustre]: pnode revalidate.
ericm [Sun, 14 Sep 2003 15:15:30 +0000 (15:15 +0000)]
[liblustre]: pnode revalidate.

21 years ago- fix against bug 1935 (OOPS in removing current working directory)
alex [Sun, 14 Sep 2003 13:38:59 +0000 (13:38 +0000)]
- fix against bug 1935 (OOPS in removing current working directory)

21 years agofix bugs for building rpm
wangdi [Sun, 14 Sep 2003 11:46:47 +0000 (11:46 +0000)]
fix bugs for building rpm

21 years agoupdate lustre_kernel.spec.in
wangdi [Sun, 14 Sep 2003 11:36:23 +0000 (11:36 +0000)]
update lustre_kernel.spec.in

21 years ago[liblustre]: get most of liblustre functions work again, except:
ericm [Sat, 13 Sep 2003 16:48:21 +0000 (16:48 +0000)]
[liblustre]: get most of liblustre functions work again, except:
  - pnode revalidation
  - file i/o

21 years ago- bug 1931 (link_path_walk bug) fix propogated into those kernels
alex [Sat, 13 Sep 2003 14:12:03 +0000 (14:12 +0000)]
- bug 1931 (link_path_walk bug) fix propogated into those kernels

21 years ago- fix agaist bug #1322 'clients looping in revalidate after recovery failure'
alex [Sat, 13 Sep 2003 11:47:31 +0000 (11:47 +0000)]
- fix agaist bug #1322 'clients looping in revalidate after recovery failure'

21 years ago- needed .pc file added
alex [Fri, 12 Sep 2003 19:54:37 +0000 (19:54 +0000)]
- needed .pc file added

21 years ago- replace some hard-coded filter bits with a cpp constant
phil [Fri, 12 Sep 2003 19:17:22 +0000 (19:17 +0000)]
- replace some hard-coded filter bits with a cpp constant
- create 3 groups instead of 2, as a start to fixing echo_create

21 years ago[liblustre]: open(O_CREAT)/close()
ericm [Fri, 12 Sep 2003 16:56:02 +0000 (16:56 +0000)]
[liblustre]: open(O_CREAT)/close()

21 years ago[liblustre]: stat(), intent related stuff, not complete.
ericm [Fri, 12 Sep 2003 14:58:03 +0000 (14:58 +0000)]
[liblustre]: stat(), intent related stuff, not complete.

21 years agoadd spec files building lustre modules + kernel in one rpm
wangdi [Fri, 12 Sep 2003 12:01:05 +0000 (12:01 +0000)]
add spec files building lustre modules + kernel in one rpm

21 years agoUsing generation in mdc_readpage function committed by tian ying.
tianying [Fri, 12 Sep 2003 01:04:27 +0000 (01:04 +0000)]
Using generation in mdc_readpage function committed by tian ying.

21 years ago- add an LASSERTF() variant so we can get some printf style data along
zab [Thu, 11 Sep 2003 19:26:54 +0000 (19:26 +0000)]
- add an LASSERTF() variant so we can get some printf style data along
  with our assert.

21 years agob=1903
phil [Thu, 11 Sep 2003 18:28:03 +0000 (18:28 +0000)]
b=1903
Add /proc/sys/portals/console which defaults to 1; if you set it to 0,
it will suppress all console output except LASSERTs

21 years agofix annoying printf warnings in mkdirdeep
phil [Thu, 11 Sep 2003 18:25:34 +0000 (18:25 +0000)]
fix annoying printf warnings in mkdirdeep

21 years ago- i_filterdata field in struct inode
alex [Thu, 11 Sep 2003 15:58:42 +0000 (15:58 +0000)]
- i_filterdata field in struct inode

21 years ago* commit in.
niu [Thu, 11 Sep 2003 07:55:56 +0000 (07:55 +0000)]
* commit in.

21 years ago* first commit in.
niu [Thu, 11 Sep 2003 07:17:40 +0000 (07:17 +0000)]
* first commit in.

21 years agoAdd .pc files for dynamic-locks and vfs-pdirops for 2.4.20-rh kernel.
adilger [Wed, 10 Sep 2003 22:38:27 +0000 (22:38 +0000)]
Add .pc files for dynamic-locks and vfs-pdirops for 2.4.20-rh kernel.

21 years agoFix usage, error messages.
adilger [Wed, 10 Sep 2003 21:59:22 +0000 (21:59 +0000)]
Fix usage, error messages.

21 years ago- fast EA (stored in large inode body) support against 2.4.18-chaos
alex [Wed, 10 Sep 2003 19:51:07 +0000 (19:51 +0000)]
- fast EA (stored in large inode body) support against 2.4.18-chaos

21 years ago- large inode support for 2.4.18-chaos
alex [Wed, 10 Sep 2003 19:46:28 +0000 (19:46 +0000)]
- large inode support for 2.4.18-chaos

NOTE: in fact, all we need is already there, but only check

21 years ago- uml for 2.4.18-chaos and series file
alex [Wed, 10 Sep 2003 06:06:40 +0000 (06:06 +0000)]
- uml for 2.4.18-chaos and series file

21 years agoFix for missing open flags for vfs_intent-2.4.20-rh kernel patch.
adilger [Tue, 9 Sep 2003 11:19:18 +0000 (11:19 +0000)]
Fix for missing open flags for vfs_intent-2.4.20-rh kernel patch.
This change was also needed for the vfs_intent-2.4.20-hp patch.  b=1877
Also add bug 1344 revalidate loop fixes to some of the vfs_intent patches.

The diff is fairly large, so that the files in the 2.4.20-rh patch are
in the same order as in all of the other vfs_intent patches to make
comparison and maintenance easier.  Some other minor changes are made to
keep the patches consistent.

There is still a difference between vfs_intent-2.4.20-vanilla and -rh
in follow_down, but I don't know which one is the "keeper".

21 years agoAdd 2.4.21 jbd fix patches to rh-2.4.20 and chaos-2.4.20 kernel series now
adilger [Tue, 9 Sep 2003 04:58:14 +0000 (04:58 +0000)]
Add 2.4.21 jbd fix patches to rh-2.4.20 and chaos-2.4.20 kernel series now
that I had a chance to test them against the rh-2.4.20 kernel.
b=1871

21 years agoland b_llog on b_devel:
phil [Tue, 9 Sep 2003 04:49:16 +0000 (04:49 +0000)]
land b_llog on b_devel:
reorganization and API update of the logging infrastructure. minimal
changes to the other code.

21 years agomerge b_llpmd into b_devel. the major highlights:
phil [Tue, 9 Sep 2003 03:54:45 +0000 (03:54 +0000)]
merge b_llpmd into b_devel.  the major highlights:
- new I/O backend
- new client page cache and llite/lov/osc plumbing
- pre-creation of OST objects
- most of the OBD protocol now revolves around exports, not obd_devices

21 years ago- clear 'rc' if rc == LLOG_EMPTY in llog_init_handle, so we don't free
phil [Tue, 9 Sep 2003 00:39:06 +0000 (00:39 +0000)]
- clear 'rc' if rc == LLOG_EMPTY in llog_init_handle, so we don't free
  the header and return an error that our caller doesn't expect
- don't call llog_lvfs_pad if left == reclen, only if 0 < left < reclen
- fix the uuid magic in llog_test
- call llog_init_handle in the right places, and don't forget to close
  the log if one of the 1a checks fail

21 years ago- If the first user command isn't ENQUEUE, postsetup would be run too
phil [Mon, 8 Sep 2003 23:02:25 +0000 (23:02 +0000)]
- If the first user command isn't ENQUEUE, postsetup would be run too
  late, which causes the 'sanity' 1a hang that coop sees.  Fixed by
  running postsetup if any of the non-setup RPCs arrives.
- Remove overwriting of lgh_last_idx in llog_init_handle
- init the handle from the test code (not quite right yet)
- #if 0 out the filter catalog functions until they build

21 years agoat least get 2.6 building, more testing happens right after gdb stops 2.6 uml
zab [Mon, 8 Sep 2003 22:41:02 +0000 (22:41 +0000)]
at least get 2.6 building, more testing happens right after gdb stops 2.6 uml
from booting.  2.4 still builds and works.

- 2.4/2.6 compat macro cleaning
- add a ll wrapper for dev_t so mknod can take int in 2.4 and dev_t in 2.6
- 2.6 has to call clear_page_dirty to keep the page dirty counts accurate,
  we introduce a helper for 2.4
- tee hee, put lli_vfs_inode back.  its important, and stuff.
- add a removepage patch for 2.6
- move readpage into rw.c as both ports now use it
- llite_internal spring cleaning
- update 2.6 writepage to use the Writeback bits
- minor filter_io_26 cleanups, more to be done
- move osc_lock_contains into the right obd_ops initializer

21 years ago- new abstraction in the llog_* function wrappers
phil [Mon, 8 Sep 2003 22:18:38 +0000 (22:18 +0000)]
- new abstraction in the llog_* function wrappers
- fix double-allocation of handle headers
- added llog_read_header()
- more llog tests

21 years ago- ext3-extents-oflag-2.4.18-chaos conflicted with ext3-delete_thread-2.4.18
alex [Mon, 8 Sep 2003 20:23:49 +0000 (20:23 +0000)]
- ext3-extents-oflag-2.4.18-chaos conflicted with ext3-delete_thread-2.4.18

21 years ago- unify llog_open and llog_create
braam [Mon, 8 Sep 2003 18:31:12 +0000 (18:31 +0000)]
- unify llog_open and llog_create

21 years ago- set the mds's and ost's obd_logops
phil [Mon, 8 Sep 2003 17:08:33 +0000 (17:08 +0000)]
- set the mds's and ost's obd_logops
- initial oa to null in llog_lvfs_create, avoids crashing
- in llog test 1a, don't try to close if the open failed
- 'lctl modules' hook for llog_test

21 years ago- ext3-map_inode_page patch ported onto 2.6.0-test3
alex [Mon, 8 Sep 2003 13:50:54 +0000 (13:50 +0000)]
- ext3-map_inode_page patch ported onto 2.6.0-test3
- uml-2.6.0-fix patch added. it allows userspace to build with UML kernel

21 years ago- stuped typo fixed
alex [Mon, 8 Sep 2003 13:17:14 +0000 (13:17 +0000)]
- stuped typo fixed

21 years ago- bio based filter_commitrw_write() to be used with 2.6
alex [Mon, 8 Sep 2003 13:13:07 +0000 (13:13 +0000)]
- bio based filter_commitrw_write() to be used with 2.6

NOTE: to be tested yet

21 years ago- steal from the reserved bits instead of the bitmap bits, another
phil [Mon, 8 Sep 2003 07:51:03 +0000 (07:51 +0000)]
- steal from the reserved bits instead of the bitmap bits, another
  piece of sage Andreas advice
- add an llh_flags field to the header
- rename confusing lgh_index to lgh_last_idx
- if the log is marked ZAP_WHEN_EMPTY, the log is full, and all of the
  records are cancelled, call llog_destroy from llog_cancel_rec
- extend llog_open to also support reopen-by-name
- add llog_vfs_read_blob and llog_vfs_read_header
- removed a bunch of unnecessary EXPORT_SYMBOLs
- add llog_lvfs_open()

21 years ago- pad llog_* body structures correctly (thanks Andreas)
phil [Mon, 8 Sep 2003 06:25:29 +0000 (06:25 +0000)]
- pad llog_* body structures correctly (thanks Andreas)
- countless build fixes

21 years ago- remove obd_run_ctxt from obd->u.mds and obd->u.filter; put it in the
phil [Mon, 8 Sep 2003 05:10:50 +0000 (05:10 +0000)]
- remove obd_run_ctxt from obd->u.mds and obd->u.filter; put it in the
  obd_device
- add lvfs_callback_ops, and the first one, the l_fid2dentry
- add callback ops to obd_run_ctxt
- change the 100 uses of mds->mds_ctxt and filter->fo_ctxt
- add a handful of llog tests

21 years ago[liblustre]: include libsysio_sockets driver
ericm [Mon, 8 Sep 2003 04:08:59 +0000 (04:08 +0000)]
[liblustre]: include libsysio_sockets driver

21 years ago- Removes duplicate fsfilt files, which must have snuck back in during
phil [Mon, 8 Sep 2003 03:19:44 +0000 (03:19 +0000)]
- Removes duplicate fsfilt files, which must have snuck back in during
  our branch shuffle
- Adds the llog_test stub, which I'll start filling in

21 years ago- wrong branch ... backing out on b_llpmd
braam [Mon, 8 Sep 2003 02:34:26 +0000 (02:34 +0000)]
- wrong branch ... backing out on b_llpmd

21 years ago- reworking the logging api for better layering
braam [Mon, 8 Sep 2003 02:28:43 +0000 (02:28 +0000)]
- reworking the logging api for better layering
- beginning of Lustre vfs library lvfs

21 years agoFix thinko when quieting compiler warning.
adilger [Sun, 7 Sep 2003 07:07:10 +0000 (07:07 +0000)]
Fix thinko when quieting compiler warning.

21 years agoQuiet ia64 compiler warnings.
adilger [Sun, 7 Sep 2003 06:43:37 +0000 (06:43 +0000)]
Quiet ia64 compiler warnings.

21 years agoRemove bogus include from mds_lib.c
adilger [Sun, 7 Sep 2003 05:47:33 +0000 (05:47 +0000)]
Remove bogus include from mds_lib.c

21 years ago- iopen patch against 2.6.0
alex [Sat, 6 Sep 2003 20:21:41 +0000 (20:21 +0000)]
- iopen patch against 2.6.0

21 years agoreduce the size of my locally-maintained mds-size patch:
phil [Sat, 6 Sep 2003 19:31:38 +0000 (19:31 +0000)]
reduce the size of my locally-maintained mds-size patch:
- little cleanups that I made along the way
- remove most instances of "#ifdef ENABLE_ORPHANS"; everyone needs to
  start using this code, if we're going to find the subtle bugs
- add mds_query_write_access(), which will be used to determine when
  the MDS has the authoritative file size

21 years ago- Support multiple lov OBDs per node, by tying the LOV into the same UUID group
shaver [Sat, 6 Sep 2003 04:05:25 +0000 (04:05 +0000)]
- Support multiple lov OBDs per node, by tying the LOV into the same UUID group
  as the OSCs and MDCs.

- The MDS now _always_ creates its own LOV and OSCs/MDC, and there is no longer
  a --mds_ost_conn switch.

- replay-dual should work now, but I haven't tested it.

- the mdc_lock is now per-MDC, and not global

- when a namespace is destroyed, remove the associated lprocfs directory,
  so that it's not there like a landmine with dangling namespace pointers.

- the MDS now knows its LOV by device name (which is lov_$MDSNAME)

- LOV now has an explicit 'descriptor' or 'LOV service' UUID, instead of
  reusing the the device UUID

- print the (possibly useful) mdc name instead of the largely useless and not
  really unique U"U"ID in lprocfs/lov/target_mdc

21 years agoCommitting eeb's merge of the qswnal changes into b_devel; plus one
phil [Fri, 5 Sep 2003 21:25:35 +0000 (21:25 +0000)]
Committing eeb's merge of the qswnal changes into b_devel; plus one
aesthetic change from HEAD.

21 years agob=1760
zab [Fri, 5 Sep 2003 20:05:18 +0000 (20:05 +0000)]
b=1760
Refactor the async ocp IO API to use less cpu on the client.  With these
changes streaming writes are no longer cpu-bound in llite.  The major
changes are to queue dirty pages for IO at commit_write() and reduce the
per-page book-keeping (bye otree!).

Pages are now associated with llite book-keeping for the duration of their
stay in the page cache.  A patch is needed in most kernels before 2.4.21 to
add the removepage callback to tear down this association as pages leave
the page cache.  Because of this the lustre kernel version is upped to 25.

now, in detail:

- get rid of osc page queue plugging.  instead we track pages for much longer
  in the queue and only fire rpcs when we have either enough to produce optimal
  rpcs or there are 'urgent' pages in the osc that must leave.

- add some fields to the ocp to remember its position in the lov to remove
  lov_stripe_offset() from the profiles.

- remove the nasty code for walking sb/inode dirty lists to find other pages to
  jam down the obd at writepage time.  (yay!)

- b=1755, b=1757, b=1810: make the dirty page limits and rpcs tunables really
  per-osc, getting rid of the fragile and redunant tracking that was going on
  in llite.  this fixes the way commit_write sleeps for space in the cache
  without creating an enormous bubble in the rpc pipeline.  Also cover them
  with osc spinlocks insted of using costly atomic ops.

- b=1750: add a 'read_ahead' tunable per superblock for disabling llite
  read-ahead.

- b=1837, b=1761: add some callbacks so that the osc can call back into llite
  when an rpc really fires.  it wants to know if it has raced with a truncate
  and also wants the most recent times in an obdo.

- b=1601: make commit_write's fallback to sync writes much clearer.  it also is
  more careful about extending i_size only for successful writes.

- add ll_removepage() which removes the ocp from the page and calls into the
  osc to remove it form the queues at truncate time.

- teach read-ahead to really never block looking up pages in the page cache
  by calling grab_cache_page_nowait() (duh)

- get rid of the otree that we used to track dirty offsets in objects.  an
  llite level hack for prefering its i_size that includes cache extensions
  works until we implement the proper EOC communications to track i_size.
  this adds the obd_lock_contains() call until that time.

- get rid of the old dead code for passing extents from ll_file_read to
  readpage.  also old 2.6 lli experiments.

21 years agomerge b_devel into b_llpmd
phil [Fri, 5 Sep 2003 18:23:33 +0000 (18:23 +0000)]
merge b_devel into b_llpmd

21 years agoThe precreate process was leaking ea's that it never used anyway.
rread [Fri, 5 Sep 2003 05:57:54 +0000 (05:57 +0000)]
The precreate process was leaking ea's that it never used anyway.

21 years agoAdd jbd-* fixes to some of the series files where I know they apply:
adilger [Fri, 5 Sep 2003 00:16:37 +0000 (00:16 +0000)]
Add jbd-* fixes to some of the series files where I know they apply:
- vanilla-2.4.20 and hp-pnnl-2.4.20 need all three
- chaos-2.4.18 doesn't need the flushtime patch (it is in their tree already)

I'm not sure whether chaos-2.4.20 and rh-2.4.20 need them or not (patches
were originally from RH kernels, so they may already be in rh-2.4.20).

21 years agoFix up the jbd patches slightly.
adilger [Fri, 5 Sep 2003 00:04:42 +0000 (00:04 +0000)]
Fix up the jbd patches slightly.
Add .pc files for the jbd patches.

21 years agoExt3 journaling patches from 2.4.21:
adilger [Thu, 4 Sep 2003 21:17:55 +0000 (21:17 +0000)]
Ext3 journaling patches from 2.4.21:
- jbd-flushtime: fixes dirty buffer starvation during IO (may help perf!)
- jbd-ctx_switch: avoids thousands of spurious context switches on busy buffers
- jbd-get_write_access: avoids assertion failure after IO error (bug 1871)

21 years ago- added page_private patch to all non-2.6 series files that were missing it
ccooper [Thu, 4 Sep 2003 19:55:17 +0000 (19:55 +0000)]
- added page_private patch to all non-2.6 series files that were missing it

21 years agoMove inode_update_time() to filter_io_24.c, as this is a VFS function in 2.6.
adilger [Thu, 4 Sep 2003 18:47:11 +0000 (18:47 +0000)]
Move inode_update_time() to filter_io_24.c, as this is a VFS function in 2.6.

21 years ago- those files contain 2.4- and 2.6-specific routines of filter I/O part
alex [Thu, 4 Sep 2003 17:01:15 +0000 (17:01 +0000)]
- those files contain 2.4- and 2.6-specific routines of filter I/O part

21 years agoadd some patches for 2.4.20-rh
wangdi [Thu, 4 Sep 2003 15:01:58 +0000 (15:01 +0000)]
add some patches for 2.4.20-rh

21 years ago[liblustre]: supress an error message.
ericm [Thu, 4 Sep 2003 10:03:20 +0000 (10:03 +0000)]
[liblustre]: supress an error message.

21 years ago[liblustre]: reorganize & fixes of test.
ericm [Thu, 4 Sep 2003 10:02:16 +0000 (10:02 +0000)]
[liblustre]: reorganize & fixes of test.

21 years ago[liblustre]: add test_common
ericm [Thu, 4 Sep 2003 09:58:57 +0000 (09:58 +0000)]
[liblustre]: add test_common

21 years agore-formatting as per CFS code guidelines
girishc [Thu, 4 Sep 2003 08:52:50 +0000 (08:52 +0000)]
re-formatting as per CFS code guidelines

21 years agorefactor the replay testing framework into test-framework.sh. I'll
rread [Thu, 4 Sep 2003 07:51:41 +0000 (07:51 +0000)]
refactor the replay testing framework into test-framework.sh. I'll
bet sanity*.sh can use this too, but I haven't checked.  Also adding an
embryonic replay-ost-single.sh.

21 years agob=1825
girishc [Thu, 4 Sep 2003 06:07:22 +0000 (06:07 +0000)]
b=1825
Addresses review comments from peter

21 years agomerge HEAD into b_devel. Fixes for:
phil [Wed, 3 Sep 2003 18:19:46 +0000 (18:19 +0000)]
merge HEAD into b_devel.  Fixes for:

- bug 1698: lconf errors during Elan cleanup
- bug 1843: NULL deref in dcache_notify
- bug 1861: LBUG in ll_it_open_error with rc = -2

This commit does NOT merge the qswnal changes from HEAD!  eeb will
help me with that.

21 years agob=1825
girishc [Wed, 3 Sep 2003 17:13:44 +0000 (17:13 +0000)]
b=1825
new file added to lustre related to nfs export

21 years ago[liblustre]: add replay_single.c
ericm [Wed, 3 Sep 2003 16:24:12 +0000 (16:24 +0000)]
[liblustre]: add replay_single.c

21 years agoRemove the now-unused 'data' field from lock_match
phil [Wed, 3 Sep 2003 07:13:07 +0000 (07:13 +0000)]
Remove the now-unused 'data' field from lock_match

21 years agoremove LDLM_FL_MATCH_DATA workaround, which was obsoleted by the
phil [Wed, 3 Sep 2003 01:33:12 +0000 (01:33 +0000)]
remove LDLM_FL_MATCH_DATA workaround, which was obsoleted by the
proper fixes to the code which takes locks and clears inodes

21 years agomissing file being added now
braam [Tue, 2 Sep 2003 14:35:01 +0000 (14:35 +0000)]
missing file being added now

21 years agoadd kgdb_eth patch
wangdi [Tue, 2 Sep 2003 13:47:23 +0000 (13:47 +0000)]
add kgdb_eth patch

21 years ago- notification message is shown if fs is mounted with pdirops option
alex [Tue, 2 Sep 2003 08:57:45 +0000 (08:57 +0000)]
- notification message is shown if fs is mounted with pdirops option

21 years ago- ext3 mount option 'pdirops' added
alex [Tue, 2 Sep 2003 07:45:41 +0000 (07:45 +0000)]
- ext3 mount option 'pdirops' added
- uninitialized i_rename_sem bug fixed

21 years ago- parallel directory operations patches have been ported onto 2.4.18-chaos
alex [Mon, 1 Sep 2003 14:09:59 +0000 (14:09 +0000)]
- parallel directory operations patches have been ported onto 2.4.18-chaos
- new chaos-2.4.18-pdirops series have been created

NOTE: not for production! I'm still testing this

21 years ago[liblustre]: add 'lrun' into utils.
ericm [Mon, 1 Sep 2003 07:38:57 +0000 (07:38 +0000)]
[liblustre]: add 'lrun' into utils.

21 years ago[liblustre]:
ericm [Mon, 1 Sep 2003 07:23:20 +0000 (07:23 +0000)]
[liblustre]:
 - add test for service side cancel client lock
 - reorganize lltest

21 years agoChanged mds_lov_set_nextid mds_lov_set_growth to use obd_set_info
rread [Fri, 29 Aug 2003 23:33:29 +0000 (23:33 +0000)]
Changed mds_lov_set_nextid mds_lov_set_growth to use obd_set_info
instead of obd_setattr, and added this code to lov_ and osc_set_info.
It is now ok if stripe_cnt  != target_cnt.

Removed the mds_lov_lsm, lsmsize fields, as they are no longer
needed now that we use obd_set_info.

Renamed lov_create_orphans to lov_clear_orphans, as that is what it does.

Fixed filter_next_id() and filter_last_id() to really be next and
last. (I take "last" to mean last value that filter_next_id()
returned.)

Added some tests to replay_single.sh.

21 years ago[liblustre]: add C version of recovery-small for liblustre.
ericm [Fri, 29 Aug 2003 13:17:57 +0000 (13:17 +0000)]
[liblustre]: add C version of recovery-small for liblustre.

21 years ago2.6 do not have smp_num_cpus
wangdi [Fri, 29 Aug 2003 06:36:48 +0000 (06:36 +0000)]
2.6 do not have smp_num_cpus

21 years ago* Make sure we don't delete the orphans until after recovery has
rread [Fri, 29 Aug 2003 00:24:23 +0000 (00:24 +0000)]
* Make sure we don't delete the orphans until after recovery has
  finished. This is done in mds_postrecovery, which should now be
  called instead of mds_lov_set_nextid.

* When we delete orphans, make sure the MDS's and OST's last_id get
  dropped back, so subsequent creates will actually create objects.

21 years agoA slightly hacked up version of Zach's patch to change the OBD API to
phil [Thu, 28 Aug 2003 22:55:08 +0000 (22:55 +0000)]
A slightly hacked up version of Zach's patch to change the OBD API to
use exports intead of handles.

It runs and cleans up normally -- as normally as llpmd ever did -- but
probably still contains some bugs.  These bugs will likely happen in
unused (cobd, ptlbd) or rarely taken (startup/cleanup error cases),
and will be of the type that leak an export or do an extra export_put.

I also reduced the obd_refcount sleep at cleanup to 1 second from 30.
I have never seen the count ever abort before the full 30 seconds were
up, and because of the unique configuration situation on b_llpmd, this
path is always taken during the cleanup of the LOV (because the MDS is
still holding an export reference)

21 years agocpu affinity are not depended on arch, fix bug 1307
wangdi [Thu, 28 Aug 2003 14:34:51 +0000 (14:34 +0000)]
cpu affinity are not depended on arch, fix bug 1307

21 years agoFixes from Martin to allow write_append_truncate to run on multiple nodes
adilger [Wed, 27 Aug 2003 22:39:27 +0000 (22:39 +0000)]
Fixes from Martin to allow write_append_truncate to run on multiple nodes
(albeit only two at a time, in rotation).

21 years ago- open replay should NOT go to the OSC which is unreachable during
braam [Wed, 27 Aug 2003 05:53:38 +0000 (05:53 +0000)]
- open replay should NOT go to the OSC which is unreachable during
  recovery
- minor fix to the test
- remove the wrong code that did osc create replay

21 years ago- extents support for ext3 added
alex [Tue, 26 Aug 2003 15:57:56 +0000 (15:57 +0000)]
- extents support for ext3 added
- O_EXTENTS flag support (one must pass the flag to vfs_create() in order
  to force ext3 to use extents)

NOTE: extents support is disabled by default. to enable use mount option
      'extents'

21 years ago- basic replay code for open; passes sanity as well as without it I think
braam [Tue, 26 Aug 2003 13:17:56 +0000 (13:17 +0000)]
- basic replay code for open; passes sanity as well as without it I think

21 years agoScript to set MDS, OST, client to wildly different dates for mtime testing.
adilger [Mon, 25 Aug 2003 23:09:20 +0000 (23:09 +0000)]
Script to set MDS, OST, client to wildly different dates for mtime testing.
Then, dates in the filesystem that are 1973 are MDS, 1976 are OST, and those
in Aug (1979) are the client.  These also conveniently appear as ~101010101,
~202020202, and ~303030303 in unix time format (seconds since epoch).

21 years agofix many bugs that prevented previously-mounted filesystems from
phil [Sun, 24 Aug 2003 21:05:02 +0000 (21:05 +0000)]
fix many bugs that prevented previously-mounted filesystems from
mounting a second time:
 - set the oscc_next_id (via obd_setattr) before we use that value to
   recover orphans
 - when recovering orphans, send the actual last-used ID (oscc_next_id - 1),
   instead of oscc_next_id
 - in filter_destroy_precreated, call filter_destroy with a NULL oti
   to avoid saving locks for the reply-ack
 - remove the boot count component of the filter objid; it is no
   longer required and makes orphan recovery harder
 - start the filter objid count at 1
 - orphan recovery is a dangerous process which if done incorrectly
   will delete the wrong objects; with that in mind, we now call a
   dramatically simpler lov_create_orphans() instead of the normal
   out-of-control lov_create