Whamcloud - gitweb
adilger [Fri, 20 Jun 2003 19:50:27 +0000 (19:50 +0000)]
The proper fix for bug 1321. We were pruning multiple dentries inside our
transaction, and that was consuming all of our transaction blocks on us.
Instead, do the dentry pruning ourselves before starting the transaction
handle and calling vfs_rmdir()->d_unhash()->shrink_dcache_parent().
adilger [Thu, 19 Jun 2003 22:25:29 +0000 (22:25 +0000)]
Add an extra reservation, because we compound the directory unlink with
the inode truncate in mds_reint_unlink(), which ext3 doesn't do. We
need enough blocks after the truncate to do the rest of the unlink.
adilger [Fri, 13 Jun 2003 07:52:13 +0000 (07:52 +0000)]
Add some more debugging to try and isolate the OST timeout bug further.
We also check the parent lock acquisition time (needs a rep ACK), the
read time, write time, and overall prepare-to-commit time.
Includes the change to the ptlrpc set timeout debugging, that dumps the
request instead of just LASSERT.
adilger [Thu, 12 Jun 2003 00:28:39 +0000 (00:28 +0000)]
Add some debugging to the journal ops in case they are taking a long time
to start/finish and leading to the mysterious OST timeouts.
pschwan [Tue, 10 Jun 2003 10:48:35 +0000 (10:48 +0000)]
b_devel -> b_orphan merge for
20030610
adilger [Wed, 4 Jun 2003 22:46:46 +0000 (22:46 +0000)]
Add matching "starting handle" message in the fsfilt_brw_start() path.
adilger [Wed, 4 Jun 2003 01:54:52 +0000 (01:54 +0000)]
Fix "used inodes" count in df by reducing the total number of inodes instead
of increasing the number of used inodes when there is a constraint somewhere.
As the constraint is eased, the total number of inodes will increase but we
always show the correct number of used inodes.
adilger [Tue, 3 Jun 2003 23:06:04 +0000 (23:06 +0000)]
Make which_patch a bit clearer.
pschwan [Mon, 2 Jun 2003 09:22:52 +0000 (09:22 +0000)]
b=1358
- update kernel patch version to 19
- remain compatible with version 18 until everyone upgrades
- new v19 code in mdc_{get,put}_rpc_lock uses the new it_int_flags member
wangdi [Sat, 31 May 2003 10:01:01 +0000 (10:01 +0000)]
fix a flaw in which_patch
adilger [Fri, 23 May 2003 07:01:37 +0000 (07:01 +0000)]
Merge b_devel -> b_orphan.
DEVEL_ORPHAN_UPDATE_PARENT_20030522 -> DEVEL_ORPHAN_UPDATE_PARENT_20030522
eeb [Thu, 22 May 2003 11:39:11 +0000 (11:39 +0000)]
* New protocol version 0x00040001
* Merged pagesize independent niobufs
* Single READ result (# bytes read or error)
* Per niobuf WRITE results
* Single XID for all RPCs types, including bulk (bulk uses a different
portal)
* XID properly 64 bit.
* 8 byte handles (lustre_handle::addr no longer used)
* Removed field order dependence of niobuf_{local,remote}
* Simplified bulk_desc usage
rread [Wed, 21 May 2003 21:23:31 +0000 (21:23 +0000)]
b=1276
r=adilger
- add the force_sync flag to fsfilt_commit
- sync to filter_disconnect, which should cause the remaining requests
on the client resend list to be purged after disconnect completes
braam [Mon, 19 May 2003 04:25:32 +0000 (04:25 +0000)]
- add portals to Lustre
- fix up makefiles
- add initial part of 2.5 in kernel make infrastructure
- cleanup tcpnal prototypes etc.
pschwan [Mon, 19 May 2003 03:15:06 +0000 (03:15 +0000)]
land b_proto on b_devel:
- major rework of the rpc and bulk APIs, lots of new assertions and cleanups
- many rpc and bulk bugfixes, including bugs 937, 1057, 1072, 1162
- perform brw and getattr operations to multiple OSTs in parallel
- all bulks are tied to exactly one request; multiple requests can be
tied to a single request_set
pschwan [Sun, 18 May 2003 16:39:16 +0000 (16:39 +0000)]
landing b_open onto b_devel:
- kernel patches for bug 1203
- fixes directory open()s
- re-enables metadata caching on the MDS
- fixes bug in ext3_noread kernel patch
- fixes some symbol export problems in other kernels
- fixes compilation problems when running on Red Hat 2.4.20 kernels
If you update past this merge, you will require kernel version 18
adilger [Thu, 15 May 2003 05:21:14 +0000 (05:21 +0000)]
Fix for compile warning in cobd since preprw/commitrw now take exports.
rread [Wed, 14 May 2003 17:42:46 +0000 (17:42 +0000)]
b=1046
- add a new argument to upcalls. The first argument identifies the reason
of upcall. The upcall script path is now /proc/sys/lustre/upcall, and the
config option has changed from --recovery_upcall to --lustre_upcall.
Current lustre upcall types and arguments:
FAILED_IMPORT tgt_uuid dev_uuid net_uuid
RECOVERY_OVER dev_uuid
- The config tools also support the new --portals_upcall, and both of the
upcalls can be set to the same script with the --upcall option.
There is only one portals upcall type:
LBUG filename function line
rread [Wed, 7 May 2003 19:12:12 +0000 (19:12 +0000)]
b=1026
r=adilger@clusterfs.com
- add initial support for creating client connections on the MDS,
configs will need to be recreated. (The version has been updated
to force this.)
- create a new config object, "filesystem" which represents a
specific filesystem. In the zero-config future, the MDS will be
need to be aware of what filesystems it "owns," so this is perhaps one
tiny step in that direction.
rread [Wed, 7 May 2003 09:12:44 +0000 (09:12 +0000)]
r=phil@clusterfs.com
- protect the obd_exports list with obd_lock
- get a reference on the export from the list
- and release the reference when done.
rread [Wed, 7 May 2003 09:00:03 +0000 (09:00 +0000)]
- rename class_destroy_export to class_unlink_export
pschwan [Mon, 5 May 2003 04:29:43 +0000 (04:29 +0000)]
partial merge of b_open into b_devel:
- Lustre kernel patch version 16
- extN no longer built; ext3 patched as part of the kernel patch
- related changes to lconf/lmc, to load and mkfs ext3 instead of extN now
adilger [Sun, 4 May 2003 00:33:44 +0000 (00:33 +0000)]
Merge from b_devel to b_orphan (ORPHAN_BASE5 -> ORPHAN_BASE6).
pschwan [Fri, 2 May 2003 23:49:54 +0000 (23:49 +0000)]
- fix obdecho, which now needs to also return the read length per page
- page_debug_check: print the IDs in hex, they're more useful that way
pschwan [Tue, 29 Apr 2003 23:19:44 +0000 (23:19 +0000)]
b=1143
- forcibly_detach_exports needs to do the same import cleanup as
target_handle_disconnect
- move some class_export_put()s to _after_ the export is finished
being used
- fixup dump_lniobuf() and dump_rniobuf(), because my last checkin
added a warning instead of removing a warning
pschwan [Mon, 28 Apr 2003 15:55:11 +0000 (15:55 +0000)]
b=1154
oops, back out the changes to the debug until the patch lands
pschwan [Sun, 27 Apr 2003 17:24:35 +0000 (17:24 +0000)]
b=1178
Preparations for removing unused "addr" from lustre_handle
pschwan [Sun, 27 Apr 2003 17:23:33 +0000 (17:23 +0000)]
Some incredible madness comes over me when I see code not wrapped at
80 columns, and I am powerless to resist.
pschwan [Sun, 27 Apr 2003 17:17:48 +0000 (17:17 +0000)]
b=1178
Preparations for removing unused "addr" from lustre_handle
pschwan [Thu, 24 Apr 2003 05:02:32 +0000 (05:02 +0000)]
- convert all kmem_cache_alloc/PORTAL_SLAB_ALLOC to OBD_SLAB_ALLOC
- POISON is sometimes a no-op, so use memset() for unused handle.addr
- in target_handle_reconnect, only compare "cookie", not the whole handle
- a cleanup here, a cleanup there, while I hunted for the corruption
rread [Thu, 24 Apr 2003 01:37:46 +0000 (01:37 +0000)]
b=1075
* added --failover option to lconf and lctl cleanup
* failover flag added to obd_disconnect and obd_cleanup
* if failover ==1, then MDS and OST will not update the client
state in last_rcvd.
* Also changes to lconf to add the --group support to MDS.
rread [Tue, 15 Apr 2003 21:46:35 +0000 (21:46 +0000)]
- add verison to config, so lconf can make sure the xml/ldap was created
by a compatible lmc.
- recreate clients if remounting
pschwan [Mon, 14 Apr 2003 21:29:34 +0000 (21:29 +0000)]
print the handle when we start/commit, for some fun bug 1130 debugging
pschwan [Sat, 12 Apr 2003 17:41:41 +0000 (17:41 +0000)]
b=1111
- Pass 'force' down to all of the cleanup methods, instead of just
having special meaning for exports
- in ldlm_cleanup, if forcing, remove the ldlm hooks from ptlrpc so we
can unload the ldlm module
rread [Fri, 11 Apr 2003 19:33:01 +0000 (19:33 +0000)]
b=1084
* update dtd, ldap schema, etc to include the new group field.
adilger [Fri, 11 Apr 2003 04:55:46 +0000 (04:55 +0000)]
Fix inode->i_blksize (for fstat st_blksize and hopefully improved cp
performance) so that it is the maximum vector write size (num_stripes *
max_iov_size) as told by obd_get_info("brw_size").
Also change the obd_get_info() API to fix some minor abuses (passing part
of the key in via val, returning data via the return code). Also remove
any pretense of obd_get_info() allocating memory for the caller, following
normal portals caller-allocates behaviour. Make inputs and returns be
fixed size, so they can be network-encoded across different arches.
adilger [Wed, 9 Apr 2003 22:43:40 +0000 (22:43 +0000)]
Update fsfilt_ext3.c from fsfilt_extN.c.
nfshp [Tue, 8 Apr 2003 10:22:46 +0000 (10:22 +0000)]
land b_cray on b_devel
eeb [Tue, 8 Apr 2003 00:32:36 +0000 (00:32 +0000)]
* Landing b_swab
pschwan [Mon, 31 Mar 2003 18:32:18 +0000 (18:32 +0000)]
landing b_recovery on b_devel
* bug fixes
- imports and exports cleanup too early, need refcounts (349, 879, 1045)
- per-import/export recovery handling (958, 931, 959)
- multiple last-rcvd slots, for serving multiple FSes (949)
- connections are again shared between multiple imp/exports (963, 964)
- "umount -f" would hang if any requests needed to be sent (393, 978)
- avoid pinning large req buffer by copying for queued messages (989)
- add "uuid" to "lctl device" command to help upcalls (991)
- "open" RPCs with transnos would confuse recovery counters (1037)
* major user-visible changes
- fail out/fail over policy now controlled by the upcall (993)
* protocol changes
- add OBD_PING to check server availability and failure (954)
pschwan [Thu, 27 Mar 2003 17:59:34 +0000 (17:59 +0000)]
b=1065
- back out directory open patch, causes -ESTALE; introduces kernel v15
- source tree now compatible with v13 or v15
braam [Fri, 21 Mar 2003 03:56:07 +0000 (03:56 +0000)]
These are the kernel patches for Peter's open directory fixes. Old
code can probably work with this (we'll allow it in through a small
change in obd_class.c)
cvs2svn [Sun, 2 Mar 2003 05:24:59 +0000 (05:24 +0000)]
This commit was manufactured by cvs2svn to create branch 'unlabeled-1.4.6'.
pschwan [Sun, 2 Mar 2003 05:23:24 +0000 (05:23 +0000)]
land 0.5.20.3 b_devel onto HEAD (b_devel will remain)
adilger [Thu, 27 Feb 2003 23:44:48 +0000 (23:44 +0000)]
file ext3-largefile.diff was initially added on branch b_devel.
zab [Thu, 27 Feb 2003 21:27:46 +0000 (21:27 +0000)]
- bring b_io up to b_devel
shaver [Tue, 25 Feb 2003 16:40:47 +0000 (16:40 +0000)]
file recovery-cleanup.sh was initially added on branch b_devel.
shaver [Mon, 24 Feb 2003 21:48:17 +0000 (21:48 +0000)]
file mlink.c was initially added on branch b_devel.
braam [Sat, 22 Feb 2003 03:15:55 +0000 (03:15 +0000)]
file .cvsignore was initially added on branch b_devel.
adilger [Fri, 21 Feb 2003 11:44:54 +0000 (11:44 +0000)]
file compile.sh was initially added on branch b_devel.
braam [Fri, 21 Feb 2003 11:05:36 +0000 (11:05 +0000)]
file Makefile.am was initially added on branch b_devel.
braam [Fri, 21 Feb 2003 11:05:35 +0000 (11:05 +0000)]
file libtest.c was initially added on branch b_devel.
amrutjoshi [Thu, 20 Feb 2003 11:51:12 +0000 (11:51 +0000)]
Fixed to make version_tag.pl work outside srcdir.
rread [Thu, 20 Feb 2003 06:52:52 +0000 (06:52 +0000)]
file lustre_peer.c was initially added on branch b_malt.
shaver [Wed, 19 Feb 2003 22:01:20 +0000 (22:01 +0000)]
file statone.c was initially added on branch b_devel.
thantry [Wed, 19 Feb 2003 19:53:19 +0000 (19:53 +0000)]
file test.txt was initially added on branch b_intel.
braam [Wed, 19 Feb 2003 04:30:16 +0000 (04:30 +0000)]
file lustre_peer.c was initially added on branch b_devel.
braam [Wed, 19 Feb 2003 04:30:15 +0000 (04:30 +0000)]
file lustre_handles.c was initially added on branch b_devel.
braam [Wed, 19 Feb 2003 04:29:17 +0000 (04:29 +0000)]
file lustre_handles.h was initially added on branch b_devel.
pschwan [Tue, 18 Feb 2003 05:45:08 +0000 (05:45 +0000)]
file acceptance-metadata-double.sh was initially added on branch b_setattr.
pschwan [Mon, 17 Feb 2003 18:43:43 +0000 (18:43 +0000)]
merge b_devel into b_setattr. highlights:
- kernel 2.5 and liblustre build changes
- fixes to Makefile.am for 7.3-era versions of automake, which is quite possibly
the worst piece of software with which I come in contact on a daily basis
- small unlink fix
braam [Mon, 17 Feb 2003 10:23:22 +0000 (10:23 +0000)]
file liblustre.h was initially added on branch b_devel.
pschwan [Sun, 16 Feb 2003 23:36:20 +0000 (23:36 +0000)]
merge b_devel into b_setattr
- sanity passes, but open(O_TRUNC) seems to leak a dentry/inode ref
zab [Sat, 15 Feb 2003 20:21:15 +0000 (20:21 +0000)]
- rebase b_io against HEAD in preparation for the latest write cache code
amrutjoshi [Fri, 14 Feb 2003 18:45:32 +0000 (18:45 +0000)]
Added support to specify the lustre obj dir.
pschwan [Wed, 12 Feb 2003 18:00:59 +0000 (18:00 +0000)]
backport Andreas's IT_READLINK fix
amrutjoshi [Wed, 12 Feb 2003 12:24:26 +0000 (12:24 +0000)]
Added mkdirmany to lustre-lite rpm.
amrutjoshi [Wed, 12 Feb 2003 10:38:03 +0000 (10:38 +0000)]
Made changes so that tinderbox runtest doesnt fail
pschwan [Tue, 11 Feb 2003 23:51:23 +0000 (23:51 +0000)]
changelog update
pschwan [Tue, 11 Feb 2003 23:45:51 +0000 (23:45 +0000)]
b=850
Call LDLM_DEBUG0 instead of LDLM_DEBUG when there are no arguments.
This is a temporary workaround for gcc 3.2 that I really hate.
adilger [Tue, 11 Feb 2003 23:43:25 +0000 (23:43 +0000)]
file ext3-2.4-ino_t.diff was initially added on branch b_devel.
adilger [Tue, 11 Feb 2003 22:25:06 +0000 (22:25 +0000)]
Merge BA OST fixes from b_devel.
pschwan [Tue, 11 Feb 2003 20:47:08 +0000 (20:47 +0000)]
I don't want to release an already-outdated v10 kernel just for 0.5.20, so
make 0.5.20 compatible with v11
braam [Sun, 9 Feb 2003 09:10:58 +0000 (09:10 +0000)]
file vanilla-2.5 was initially added on branch b_devel.
braam [Sun, 9 Feb 2003 09:10:56 +0000 (09:10 +0000)]
file lustre-2.5.pc was initially added on branch b_devel.
cvs2svn [Fri, 7 Feb 2003 18:03:11 +0000 (18:03 +0000)]
This commit was manufactured by cvs2svn to create branch 'unlabeled-1.4.6'.
pschwan [Fri, 7 Feb 2003 18:01:04 +0000 (18:01 +0000)]
Merge b_md into HEAD
* bug fixes
- Fix ldlm_lock_match on the MDS to avoid matching remote locks (592)
- Fix fsfilt_extN_readpage() to read a full page of directory
entries, or fake the remainder if PAGE_SIZE != blocksize (500)
- Avoid extra mdc_getattr() in ll_intent_lock when possible (534, 604)
- Fix imbalanced LOV object allocation and out-of-bound access (469)
- Most intent operations were removed, in favour of a new RPC mode
that does a single RPC to the server and bypasses most of the VFS
- All LDLM resource ID arrays were removed in favour of ldlm_res_id
- Aggressively cancel local locks on DLM servers
- mds_reint_unlink sends EA to the client if it's the last nlink.
client uses that EA to unlink OST objects.
- mds_reint_{rename,unlink,link} were rewritten to take ordered locks
- recursive symlinks were fixed (440)
- fixed NULL deref in DEBUG_REQ
- filter_update_lastobjid no longer calls sync, which annoyed extN
- fixed multi-client small-writes to a single file problem (445)
- fixed mtime updates during file writes (607)
- fixed vector writes on obdfilter causing problems when ENOSPC (670)
- fixed bug in obd_brw_read/write() (under guise of testing 367)
- fixed Linux OST size reporting problem (444, 656)
- OST now updates object mtime with writes or setattr (607, 619)
- client verifies file size before zeroing page past EOF (445)
- OST now writes last allocated objid to disk with allocation (108)
- LOV on echo now works (409)
* protocol changes
- mds_reint_unlink sends a new buffer, with the EA included. this
buffer is only valid if body->valid & OBD_MD_FLEASIZE, which is only
set if a regular file was being unlinked, and it was the last link
- use PtlGet from the target for bulk writes (315)
- OST now updates object mtime with writes or setattr (607, 619)
- LDLM now has a grant-time callback to revalidate locked items, if
necessary (604)
- Many MDS operations were reorganized to combat race conditions
* other changes
- Merge b_intel branch (updated lprocfs code) - now at /proc/fs/lustre
- configure check to avoid gcc version 2.96
20000731-2.96-98 (606)
eeb [Fri, 7 Feb 2003 15:59:13 +0000 (15:59 +0000)]
file obdiolib.h was initially added on branch b_md.
eeb [Fri, 7 Feb 2003 15:59:12 +0000 (15:59 +0000)]
file obdiolib.c was initially added on branch b_md.
eeb [Fri, 7 Feb 2003 15:59:11 +0000 (15:59 +0000)]
file obdbarrier.c was initially added on branch b_md.
adilger [Wed, 5 Feb 2003 21:45:40 +0000 (21:45 +0000)]
file createtest.c was initially added on branch b_md.
adilger [Mon, 3 Feb 2003 18:31:08 +0000 (18:31 +0000)]
file exports_hp.txt was initially added on branch b_md.
adilger [Mon, 3 Feb 2003 18:31:06 +0000 (18:31 +0000)]
file exports_hp.pc was initially added on branch b_md.
adilger [Mon, 3 Feb 2003 18:31:03 +0000 (18:31 +0000)]
file exports_hp.patch was initially added on branch b_md.
braam [Sat, 1 Feb 2003 21:58:01 +0000 (21:58 +0000)]
file busy.sh was initially added on branch b_md.
braam [Sat, 1 Feb 2003 21:58:00 +0000 (21:58 +0000)]
file extN-iget-debug.diff was initially added on branch b_md.
adilger [Sat, 1 Feb 2003 07:42:40 +0000 (07:42 +0000)]
file invalidate_show.txt was initially added on branch b_md.
adilger [Sat, 1 Feb 2003 07:42:37 +0000 (07:42 +0000)]
file invalidate_show.pc was initially added on branch b_md.
pschwan [Fri, 31 Jan 2003 21:15:07 +0000 (21:15 +0000)]
file invalidate_show.patch was initially added on branch b_intent.
adilger [Fri, 31 Jan 2003 01:22:41 +0000 (01:22 +0000)]
Skip portals-related allocations in leak_finder.pl, because we cannot see the
free message in the kernel log (unless we called portals_do_debug_dumplog()
from portals_exit() or something).
pschwan [Thu, 30 Jan 2003 21:09:07 +0000 (21:09 +0000)]
file open_delay.c was initially added on branch b_intent.
shaver [Thu, 30 Jan 2003 19:24:56 +0000 (19:24 +0000)]
file recovery-small.sh was initially added on branch b_md.
braam [Thu, 30 Jan 2003 05:16:59 +0000 (05:16 +0000)]
file iod-rmap-exports.pc was initially added on branch b_intent.
braam [Thu, 30 Jan 2003 05:16:58 +0000 (05:16 +0000)]
file iod-rmap-exports.patch was initially added on branch b_intent.
adilger [Wed, 29 Jan 2003 22:13:04 +0000 (22:13 +0000)]
file invalidate-show.diff was initially added on branch b_intent.
eeb [Wed, 29 Jan 2003 21:17:31 +0000 (21:17 +0000)]
file obdstat.c was initially added on branch b_md.
eeb [Wed, 29 Jan 2003 21:17:30 +0000 (21:17 +0000)]
file obdio.c was initially added on branch b_md.
adilger [Wed, 29 Jan 2003 01:08:21 +0000 (01:08 +0000)]
file ext3-use-after-free.diff was initially added on branch b_intent.
adilger [Wed, 29 Jan 2003 01:08:20 +0000 (01:08 +0000)]
file ext3-unmount_sync.diff was initially added on branch b_intent.
adilger [Fri, 24 Jan 2003 07:00:33 +0000 (07:00 +0000)]
Improve openunlink so it works with mount2 setups.