Whamcloud - gitweb
fs/lustre-release.git
21 years ago- llanalyze first bugfixes -- use hex when needed!
braam [Sun, 10 Nov 2002 22:10:25 +0000 (22:10 +0000)]
- llanalyze  first bugfixes -- use hex when needed!

21 years ago- nodlm option to llanalyze
braam [Sun, 10 Nov 2002 21:51:19 +0000 (21:51 +0000)]
- nodlm option to llanalyze

21 years ago- a log analyzer to help us work with the Lustre logs even more easily
braam [Sun, 10 Nov 2002 21:47:38 +0000 (21:47 +0000)]
- a log analyzer to help us work with the Lustre logs even more easily
  So far:
   - indentation with spaces for call stack depth
   - select pid
   - --trace (only the call trace)
   - --nonet (leave out networking message)
Written in Perl, very easy to add more features.

21 years agoThe test for a freeing inode was reversed, oops. We want to skip inodes
adilger [Sun, 10 Nov 2002 21:02:35 +0000 (21:02 +0000)]
The test for a freeing inode was reversed, oops.  We want to skip inodes
that are marked I_FREEING.

21 years ago- change the OST / MDT thread names to also include the host pid
braam [Sun, 10 Nov 2002 08:34:42 +0000 (08:34 +0000)]
- change the OST / MDT thread names to also include the host pid
  in case of UML to aid debugging.

21 years agouse vmalloc for OBD_ALLOCs above 16k, like we do in Portals
pschwan [Sat, 9 Nov 2002 20:39:12 +0000 (20:39 +0000)]
use vmalloc for OBD_ALLOCs above 16k, like we do in Portals

21 years agothe first try at lock LRU
pschwan [Sat, 9 Nov 2002 19:40:59 +0000 (19:40 +0000)]
the first try at lock LRU

21 years agoBuild an XML config for connecting to a BA OST for echo testing.
adilger [Fri, 8 Nov 2002 22:10:27 +0000 (22:10 +0000)]
Build an XML config for connecting to a BA OST for echo testing.

21 years agoMinor code cleanups - mostly whitespace.
adilger [Fri, 8 Nov 2002 20:59:56 +0000 (20:59 +0000)]
Minor code cleanups - mostly whitespace.

21 years agoFix minor compiler warning.
adilger [Fri, 8 Nov 2002 20:58:45 +0000 (20:58 +0000)]
Fix minor compiler warning.

21 years agoAdd assertion to LL_GET_INTENT(), because we aren't very non-intent friendly
adilger [Fri, 8 Nov 2002 20:02:18 +0000 (20:02 +0000)]
Add assertion to LL_GET_INTENT(), because we aren't very non-intent friendly
right now.

21 years agoupdate file size in ll_file_read, lest the VFS decide not to call us
pschwan [Fri, 8 Nov 2002 19:33:59 +0000 (19:33 +0000)]
update file size in ll_file_read, lest the VFS decide not to call us

21 years agoFix for many-clients-and-server-on-same-node deadlock. One server task
adilger [Fri, 8 Nov 2002 19:02:41 +0000 (19:02 +0000)]
Fix for many-clients-and-server-on-same-node deadlock.  One server task
can go "in the hole" for at most one maximum-sized request, to avoid
deadlocking when the clients have consumed all of the kmapping reservations.

The server can then complete the request and both the server and client will
release their reservations, allowing further progress to be made.

21 years agoSome additional changes for kmem accounting that I missed.
adilger [Fri, 8 Nov 2002 18:29:13 +0000 (18:29 +0000)]
Some additional changes for kmem accounting that I missed.

21 years agoHighmem deadlock avoidance (server version):
adilger [Fri, 8 Nov 2002 18:25:53 +0000 (18:25 +0000)]
Highmem deadlock avoidance (server version):
- reserve highmem pages before we start doing kmaps to avoid deadlocks
  with multiple threads reserving vectors of pages

21 years agoDramatically increase the randomness of MDC UUIDs
pschwan [Fri, 8 Nov 2002 02:10:58 +0000 (02:10 +0000)]
Dramatically increase the randomness of MDC UUIDs

21 years ago- add something that prints the time every 10,000 creates
braam [Fri, 8 Nov 2002 00:11:51 +0000 (00:11 +0000)]
- add something that prints the time every 10,000 creates

21 years agoUse the fs-method statfs, so we can take into account the EA space needs
adilger [Thu, 7 Nov 2002 23:54:17 +0000 (23:54 +0000)]
Use the fs-method statfs, so we can take into account the EA space needs
for extN (1 block for each inode, excluding special files).

21 years agoMake sure that we set last_xid and last_committed on replies from LDLM,
shaver [Thu, 7 Nov 2002 23:37:32 +0000 (23:37 +0000)]
Make sure that we set last_xid and last_committed on replies from LDLM,
operations, or intent-only metadata loads will result in unbounded client
memory growth.

21 years agoFix (or a workaround that will hide the real bug) for bug 322.
adilger [Thu, 7 Nov 2002 23:23:37 +0000 (23:23 +0000)]
Fix (or a workaround that will hide the real bug) for bug 322.

We are calling mdc_blocking_ast() to clean up an inode that is already
being cleaned up elsewhere.  This is the result of ldlm_cli_cancel_unused()
being called from two places at the same time:
- ll_prepare_recovery()->prepare_mdc()->ldlm_cli_cancel_unused()
- ll_clear_inode()->mdc_cancel_unused()->ldlm_cli_cancel_unused()

There is still a small race where the ll_clear_inode() RPC could have
completed and "inode" is garbage, but since we are in recovery this
is highly unlikely.

21 years agofile mkdirmany.c was initially added on branch b_symlink.
adilger [Thu, 7 Nov 2002 21:41:09 +0000 (21:41 +0000)]
file mkdirmany.c was initially added on branch b_symlink.

21 years agoFix build-breaking thinko
pschwan [Thu, 7 Nov 2002 21:26:03 +0000 (21:26 +0000)]
Fix build-breaking thinko

21 years ago- land b_symlink on HEAD
pschwan [Thu, 7 Nov 2002 21:15:07 +0000 (21:15 +0000)]
- land b_symlink on HEAD
- print peer when entering recovery, since MDC uuids are no longer descriptive

21 years agoOnly store replayable (MDC, with transno) requests on the sending_head after
shaver [Thu, 7 Nov 2002 21:03:27 +0000 (21:03 +0000)]
Only store replayable (MDC, with transno) requests on the sending_head after
they've received a reply.

21 years ago- capability fixes on head
braam [Thu, 7 Nov 2002 20:46:28 +0000 (20:46 +0000)]
- capability fixes on head
- LL_SAVE_INTENT fixes

21 years agofile createmany.c was initially added on branch b_symlink.
braam [Thu, 7 Nov 2002 18:25:19 +0000 (18:25 +0000)]
file createmany.c was initially added on branch b_symlink.

21 years ago0.5.16+ updates
pschwan [Thu, 7 Nov 2002 18:14:05 +0000 (18:14 +0000)]
0.5.16+ updates

21 years agoComment clarification.
adilger [Thu, 7 Nov 2002 09:08:26 +0000 (09:08 +0000)]
Comment clarification.

21 years agoWhitespace.
adilger [Thu, 7 Nov 2002 09:07:41 +0000 (09:07 +0000)]
Whitespace.

21 years agob=141,205,323
pschwan [Thu, 7 Nov 2002 06:22:15 +0000 (06:22 +0000)]
b=141,205,323
- When the MDS returns an intent lock to us, check to see if we already have a
  matching one.  If so, cancel the new one.
- Add a new mode to ldlm_lock_match, which allows you to pass in an existing
  lock handle.  It will look for any matches among other locks.

21 years agosmall fixups to revalidate2 lock matching
pschwan [Thu, 7 Nov 2002 04:40:21 +0000 (04:40 +0000)]
small fixups to revalidate2 lock matching

21 years ago- add another sanity test
braam [Thu, 7 Nov 2002 04:29:45 +0000 (04:29 +0000)]
- add another sanity test
- test if we have a lock already in revalidate2

21 years agob=290,311
pschwan [Thu, 7 Nov 2002 04:23:31 +0000 (04:23 +0000)]
b=290,311
- call obd_cancel_unused with the correct parameters
- only call it for regular files (those with non-NULL lli->lli_smd)

21 years agoanother shot at managing dentries
braam [Thu, 7 Nov 2002 04:03:40 +0000 (04:03 +0000)]
another shot at managing dentries

21 years agob=311
pschwan [Thu, 7 Nov 2002 03:25:43 +0000 (03:25 +0000)]
b=311
- Drop unused data locks in clear_inode
- Fix for directory and inode counts and link counts

21 years agosomehow I keep screwing this up, all over a 3-line change.
pschwan [Thu, 7 Nov 2002 00:32:27 +0000 (00:32 +0000)]
somehow I keep screwing this up, all over a 3-line change.

21 years agomerging all noncontroversial pieces of b_symlink into HEAD
pschwan [Thu, 7 Nov 2002 00:27:20 +0000 (00:27 +0000)]
merging all noncontroversial pieces of b_symlink into HEAD

21 years agomerging all noncontroversial pieces of b_symlink into HEAD
pschwan [Thu, 7 Nov 2002 00:21:23 +0000 (00:21 +0000)]
merging all noncontroversial pieces of b_symlink into HEAD

21 years agob=255
pschwan [Wed, 6 Nov 2002 23:52:07 +0000 (23:52 +0000)]
b=255
- Don't call cancel_unused until the last CLOSE for a given inode.

21 years ago- wholesale renaming of the brw associated callback variables,
rread [Wed, 6 Nov 2002 07:57:09 +0000 (07:57 +0000)]
- wholesale renaming of the brw associated callback variables,
  structures and functions.  Primarily distinguish between:
   - portals event handlers (no longer called callbacks)
   - brw callbacks

21 years ago* init script designed for MCR by Gord Eagle, and a useful example
rread [Wed, 6 Nov 2002 07:49:02 +0000 (07:49 +0000)]
* init script designed for MCR by Gord Eagle, and a useful example

21 years agoAdd kernel patch version checking
pschwan [Wed, 6 Nov 2002 07:28:35 +0000 (07:28 +0000)]
Add kernel patch version checking

21 years agofile stat.c was initially added on branch b_symlink.
braam [Wed, 6 Nov 2002 01:30:10 +0000 (01:30 +0000)]
file stat.c was initially added on branch b_symlink.

21 years ago- the root dentry should not be uhashed, it is always a valid dentry
braam [Wed, 6 Nov 2002 00:23:21 +0000 (00:23 +0000)]
- the root dentry should not be uhashed, it is always a valid dentry
- this is a fix for bug 341

21 years ago- sanity check for intent setting
braam [Tue, 5 Nov 2002 23:31:57 +0000 (23:31 +0000)]
- sanity check for intent setting

21 years agoPrudent check for NULL intent before dereferencing.
adilger [Tue, 5 Nov 2002 20:17:49 +0000 (20:17 +0000)]
Prudent check for NULL intent before dereferencing.

21 years agoMinor type fixup to match o_brw.
adilger [Tue, 5 Nov 2002 19:58:42 +0000 (19:58 +0000)]
Minor type fixup to match o_brw.

21 years ago- Replaces the LDLM lock counters (in /proc/ldlm/)
pschwan [Tue, 5 Nov 2002 19:18:19 +0000 (19:18 +0000)]
- Replaces the LDLM lock counters (in /proc/ldlm/)
- Removes the second LDLM import that Peter added, in favour of doing a
  last-minute fixup of the portal numbers in ldlm_cli_cancel

21 years ago- Fix up c_recovd_data list_head management, to avoid NULL dereferences in
shaver [Tue, 5 Nov 2002 18:37:28 +0000 (18:37 +0000)]
- Fix up c_recovd_data list_head management, to avoid NULL dereferences in
  recovd_conn_manage (bug 338).

21 years agoMinor fix so that the printed output is actually what we want to run.
adilger [Tue, 5 Nov 2002 08:50:56 +0000 (08:50 +0000)]
Minor fix so that the printed output is actually what we want to run.

21 years agoIssue# 330: Approver: Andreas - New utility to find files affected by loss of an...
donmilos [Tue, 5 Nov 2002 01:48:17 +0000 (01:48 +0000)]
Issue# 330: Approver: Andreas - New utility to find files affected by loss of an OBD.

21 years agoPeter tells me that the removal of "-O2" from the i386 archdep is the actual
adilger [Mon, 4 Nov 2002 21:38:46 +0000 (21:38 +0000)]
Peter tells me that the removal of "-O2" from the i386 archdep is the actual
source of the QUOTA problems.

21 years agoInclude quotaops.h in the headers that we munge for extN to avoid issues
adilger [Mon, 4 Nov 2002 21:36:18 +0000 (21:36 +0000)]
Include quotaops.h in the headers that we munge for extN to avoid issues
with "extern __inline__" declarations therein.

21 years agoBug# 329: Approver: Andreas - obd_ioctl_unpack() causes segmentation violation.
donmilos [Mon, 4 Nov 2002 21:25:49 +0000 (21:25 +0000)]
Bug# 329: Approver: Andreas - obd_ioctl_unpack() causes segmentation violation.

21 years agominor comment fixup
pschwan [Mon, 4 Nov 2002 20:42:05 +0000 (20:42 +0000)]
minor comment fixup

21 years agob=320
pschwan [Mon, 4 Nov 2002 00:31:03 +0000 (00:31 +0000)]
b=320
disable lprocfs, since it crashes on router nodes

21 years agoDon't include lprocfs_ll_rd() at all. lprocfs defines a no-op function
adilger [Mon, 4 Nov 2002 00:27:13 +0000 (00:27 +0000)]
Don't include lprocfs_ll_rd() at all.  lprocfs defines a no-op function
stub for this.  When lprocfs isn't available, we will get nothing from
these functions, but I don't think that is a big deal.

21 years agoThe "random" addition to the UUID were in fact not very random: dev is time-
pschwan [Sun, 3 Nov 2002 20:51:44 +0000 (20:51 +0000)]
The "random" addition to the UUID were in fact not very random: dev is time-
synchronized, and this caused big trouble.

I fixed this by seeding the PRNG from /dev/urandom at startup; my first
multi-line python hackery, so I hope I didn't pooch it.

21 years ago- minor fixes to the tests
braam [Sat, 2 Nov 2002 23:09:11 +0000 (23:09 +0000)]
- minor fixes to the tests

21 years ago- Replace client-side generation fixup with bug 304's righteous assertion of
shaver [Sat, 2 Nov 2002 09:08:58 +0000 (09:08 +0000)]
- Replace client-side generation fixup with bug 304's righteous assertion of
  generations numbers at recreate-time.  Single-client recovery should now be
  _that_much_ more robust. (Bug 299.)

21 years agonew test with lots of very basic operations
braam [Sat, 2 Nov 2002 04:35:01 +0000 (04:35 +0000)]
new test with lots of very basic operations

21 years ago1. Fixed cleanup issue where I was *indeed* accesing a zapped pointer.
thantry [Sat, 2 Nov 2002 02:41:32 +0000 (02:41 +0000)]
1. Fixed cleanup issue where I was *indeed* accesing a zapped pointer.
2. Removed typedef
3. lproc-snmp back in.

21 years ago- fixes for dentry problems from Phil and myself
braam [Sat, 2 Nov 2002 01:09:28 +0000 (01:09 +0000)]
- fixes for dentry problems from Phil and myself
- separate cancel/callback threads to avoid starvation
- fix for broken symlinks

21 years ago* add a bit of randomness to the MDC uuid
rread [Sat, 2 Nov 2002 00:51:43 +0000 (00:51 +0000)]
* add a bit of randomness to the MDC uuid

21 years agoBuild properly with lprocfs disabled.
adilger [Sat, 2 Nov 2002 00:26:06 +0000 (00:26 +0000)]
Build properly with lprocfs disabled.

21 years ago* change init_portals to use inter_module_get() instead of
rread [Fri, 1 Nov 2002 23:42:36 +0000 (23:42 +0000)]
* change init_portals to use inter_module_get() instead of
inter_module_get_request(), becuase we don't want to the kernel to load
any old nal, we want to use just the one that was loaded for us by
lconf.

* add hook for toenal

21 years ago* more bits for toenal
rread [Fri, 1 Nov 2002 22:46:56 +0000 (22:46 +0000)]
* more bits for toenal

21 years ago* rename startlevel and endlevel to minlevel and maxlevel
rread [Fri, 1 Nov 2002 22:39:30 +0000 (22:39 +0000)]
* rename startlevel and endlevel to minlevel and maxlevel

21 years agoRevert Mike's recovery changes from LOV, for the greater good of being
adilger [Fri, 1 Nov 2002 22:33:08 +0000 (22:33 +0000)]
Revert Mike's recovery changes from LOV, for the greater good of being
able to use LOV again.

21 years agoYes, we should in fact MOD_DEC_USE_COUNT there.
adilger [Fri, 1 Nov 2002 21:19:08 +0000 (21:19 +0000)]
Yes, we should in fact MOD_DEC_USE_COUNT there.

21 years agofile sanity.sh was initially added on branch b_find_alias.
braam [Fri, 1 Nov 2002 19:51:30 +0000 (19:51 +0000)]
file sanity.sh was initially added on branch b_find_alias.

21 years agoDisable lprocfs until it can clean up.
adilger [Fri, 1 Nov 2002 19:18:17 +0000 (19:18 +0000)]
Disable lprocfs until it can clean up.

21 years ago* initial checkin
rread [Fri, 1 Nov 2002 19:11:57 +0000 (19:11 +0000)]
* initial checkin

21 years agoLproc-snmp code drop
thantry [Fri, 1 Nov 2002 17:23:18 +0000 (17:23 +0000)]
Lproc-snmp code drop

21 years ago* 1st attempt to prevent duplicate devices being started.
rread [Fri, 1 Nov 2002 02:59:48 +0000 (02:59 +0000)]
* 1st attempt to prevent duplicate devices being started.
* also don't cleanup a device if it doesn't exist.

21 years agoSimilar to local.sh, but with 2 mount points of the same filesystem.
rread [Fri, 1 Nov 2002 02:32:38 +0000 (02:32 +0000)]
Similar to local.sh, but with 2 mount points of the same filesystem.

21 years agoUse the right fstype for the kernel being used.
rread [Thu, 31 Oct 2002 22:24:34 +0000 (22:24 +0000)]
Use the right fstype for the kernel being used.

21 years agoRemoved the lprocfs_snmp.h file. Need to get rid of "snmp" usage
thantry [Thu, 31 Oct 2002 20:08:14 +0000 (20:08 +0000)]
Removed the lprocfs_snmp.h file. Need to get rid of "snmp" usage
inside lustre. New file is lprocfs_status.h

21 years agoChanging the name from lprocfs_snmp.h to lprocfs_status.h
thantry [Thu, 31 Oct 2002 20:05:51 +0000 (20:05 +0000)]
Changing the name from lprocfs_snmp.h to lprocfs_status.h

21 years ago* cleanup multiple MDC's correctly
rread [Thu, 31 Oct 2002 20:00:28 +0000 (20:00 +0000)]
* cleanup multiple MDC's correctly

21 years ago* teach lconf about ktoenal.o
rread [Thu, 31 Oct 2002 17:50:53 +0000 (17:50 +0000)]
* teach lconf about ktoenal.o

21 years agoModified the makefile.am to allow creation of only lustre-HOWTO.txt by default.
radhikav [Thu, 31 Oct 2002 17:18:47 +0000 (17:18 +0000)]
Modified the makefile.am to allow creation of only lustre-HOWTO.txt by default.
'make docs' will build the lustre.pdf document.

21 years agomany places in the reint path did not use generation numbers with their
pschwan [Thu, 31 Oct 2002 15:22:33 +0000 (15:22 +0000)]
many places in the reint path did not use generation numbers with their
locks; fixed.

21 years agoHack to force an MDS flush and also get the last_xid/last_committed from
adilger [Thu, 31 Oct 2002 07:39:18 +0000 (07:39 +0000)]
Hack to force an MDS flush and also get the last_xid/last_committed from
the MDS, so the client can flush its pending RPC requests before umount.

Also, make mdc_statfs() into a OBD method, just because we can.

21 years agoQuiet down the lprocfs warnings a bit until Hari gets new code into the tree.
adilger [Thu, 31 Oct 2002 07:12:36 +0000 (07:12 +0000)]
Quiet down the lprocfs warnings a bit until Hari gets new code into the tree.

21 years agoRobert's mutli-MDS fix.
adilger [Thu, 31 Oct 2002 06:58:34 +0000 (06:58 +0000)]
Robert's mutli-MDS fix.

21 years agoMinor update to mds_create_pack(): use "dir" instead of "inode" so that it
adilger [Thu, 31 Oct 2002 05:30:10 +0000 (05:30 +0000)]
Minor update to mds_create_pack(): use "dir" instead of "inode" so that it
is clear what we are packing into the cr_fid.

21 years agoMore ignores.
adilger [Wed, 30 Oct 2002 22:39:44 +0000 (22:39 +0000)]
More ignores.

21 years agoIgnore more generated files.
adilger [Wed, 30 Oct 2002 22:38:39 +0000 (22:38 +0000)]
Ignore more generated files.

21 years agoDemote the 'sending list' CERRORs to CDEBUGs, so that it's easier to spot bugs
pschwan [Wed, 30 Oct 2002 22:09:41 +0000 (22:09 +0000)]
Demote the 'sending list' CERRORs to CDEBUGs, so that it's easier to spot bugs
in console output.

21 years agob=300
pschwan [Wed, 30 Oct 2002 21:59:59 +0000 (21:59 +0000)]
b=300
MDT wasn't calling 'unregister_type'; fixed

21 years agoWhitespace and a small comment.
adilger [Wed, 30 Oct 2002 21:52:44 +0000 (21:52 +0000)]
Whitespace and a small comment.

21 years ago* change device_list to use the command line interface instead of stdin
rread [Wed, 30 Oct 2002 21:03:43 +0000 (21:03 +0000)]
* change device_list to use the command line interface instead of stdin

21 years ago- Fix OBD_IOC_OPEN to return the obdo, so user-space has a file handle,
adilger [Wed, 30 Oct 2002 20:56:32 +0000 (20:56 +0000)]
- Fix OBD_IOC_OPEN to return the obdo, so user-space has a file handle,
  and runregression-brw.sh can run against obdfilter again.
- Destroy the IPC shmem segment so we can run more than 160 runtests in a row.
- Allow COUNT to be specified, so you can run smaller runregression-brw.sh
  against obdfilter or BA OSTs and not run out of disk space.

21 years ago- Bug 296: remove LDLM_MDSINTENT to fix mismatches with LDLM_PLAIN locks.
shaver [Wed, 30 Oct 2002 20:38:28 +0000 (20:38 +0000)]
- Bug 296: remove LDLM_MDSINTENT to fix mismatches with LDLM_PLAIN locks.
  We now use LDLM_FL_HAS_INTENT in flags, instead.
- Remove some over-verbose debugging in lookup2_finish.
- Automatically reactivate OSC when recovery reconnects successfully.

21 years ago- minor fixes to get mount to roll in 2.5
braam [Wed, 30 Oct 2002 20:25:03 +0000 (20:25 +0000)]
- minor fixes to get mount to roll in 2.5

21 years ago"mds" was in SUBDIRS twice--this is bad.
pschwan [Wed, 30 Oct 2002 18:58:41 +0000 (18:58 +0000)]
"mds" was in SUBDIRS twice--this is bad.

21 years ago* add --fstype arg to both --mds and --ost
rread [Wed, 30 Oct 2002 18:38:36 +0000 (18:38 +0000)]
* add --fstype arg to both --mds and --ost
eg:
   lmc ... --node node --fstype ext3 --mds mds1 $MDSDEV $MDSSIZE
   lmc ... --node node --fstype ext3 --ost $OSTDEV $OSTSIZE

21 years agoUpdate changelog
pschwan [Wed, 30 Oct 2002 08:29:45 +0000 (08:29 +0000)]
Update changelog

21 years agob=201
pschwan [Wed, 30 Oct 2002 08:20:49 +0000 (08:20 +0000)]
b=201
b=285
- Added generation numbers to MDS locks (still need to for OSC locks, bug 294)
- fixup the inode pointer in the ldlm_lock after intent gives us a lock on a
  different resource (bug 285)
- cancel locks when the inodes are flushed from the cache (bug 201)
- fix serious cache coherancy bugs by really tying locks to dentries:
  - always return 0 from revalidate2 for a negative dentry, forcing lookup2
  - when MDS locks are cancelled, unhash the associated dentries (but do NOT
    remove them from the aliases list)
    - avoid calling the cancellation callback at shutdown and clear_inode time,
      by adding yet another flag, this time to cancel_unused
  - lookup2 searches for old dentries that may satisfy a new request, and may
    reuse a dentry that we previously unhashed

We broke hard links even more than they were before--you've been warned!