1 tbd Cluster File Systems, Inc. <info@clusterfs.com>
4 - reduce journal credits needed for BRW writes (2370)
5 - orphan handling to avoid losing space on client/server crashes
6 - ptlrpcd can be blocked, stopping ALL progress (2477)
7 - recovery for initial connections (2355)
8 - orphan recovery problems in b_eq (1934)
9 - block all file creations until orphan recovery completes (1901)
10 - client remove rq_connection from request struct (2423)
11 - conf-sanity test_5, proper cleanup in umount log not availale (2640)
12 - recovery timer race (2670)
13 - mdc_close recovey bug (2532)
15 2004-02-07 Cluster File Systems, Inc. <info@clusterfs.com>
18 - fix truncated write corruption (2366)
19 - fix for failed assertion in iopen_connect_dentry (1792,2517)
21 - don't flag the ptlrpcd thread with PF_MEMALLOC (2636)
22 - ensure len(uuid) < 37 in lmc (1171)
23 - fix ia64 OOPS in llog_test (2255)
24 - zero end of page at obdfilter for partial page writes (2648)
25 - don't leave stale dentries around after renames (bug 2428)
26 - fix timeouts when evicting a client with a single lock held (2642)
27 - set deadline for the initial HELLO message to drain (2634)
28 - print out dotted-quad IP addresses in the socknal (2302)
30 - additional debugging for MDS client eviction problem (2443)
32 2004-01-27 Cluster File Systems, Inc. <info@clusterfs.com>
35 - add series for the vanilla 2.6.0 kernel
36 - add series for the vanilla 2.4.24 kernel
37 - add series for a cray x86/64 UL kernel drop
38 - fix xattr patches for the vanilla 2.4.19 series
40 - generate true UUIDs in lmc (1171)
41 - have portals stack dumping break in UML (2466)
42 - avoid bad dchild deref; avoid inum lock w/o creation (2362)
43 - allocate with _NOFS in ldlm to avoid deadlock (1933)
44 - wake callback waiting threads on client eviction (2460)
45 - Add --ptldebug and --subsystem to lmc (1719)
46 - update assertion to allow safe interrupt allocation
47 - set rq_no_resend for cancel requests (2432)
48 - recalculate ptlrpcd timeout after resend (2494)
49 - call vfs_rmdir when removing pending directories (2368)
50 - fix renaming a file to itself (2429)
51 - lmc creates a default one-stripe lov (2454)
52 - expand procfs space to handle large clusters (2326)
53 - increase UML stack to avoid overflow
54 - update lconf's list of debug and subsystem masks
55 - fix lfs find --obd (2510)
56 - /proc tunable for disabling filter read caching (2591)
57 - stop rpm packages from altering slapd.conf (2301)
58 - disable nagle in the socknal under 0conf (2578)
59 - choose mds inode size based on stripe count (2572)
60 - fix kernel-source rpm problems (2516)
62 - add --disable-doc to avoid pdf generation (2421)
63 - update documentation, tests, type-os, comments
64 - avoid format warnings on ia64
66 - tiny code cleanups by removing unused fields
68 2004-01-07 Cluster File Systems, Inc. <info@clusterfs.com>
71 - fix obvious semaphore misuse in as-yet-unused setattr path (2348)
72 - remove the most blatant lies from BUILDING file (2371)
73 - change default debug level to reasonable production setting
74 - reduce client side cache size to reduce cache flush time
75 - reduce max RPCs in flight to avoid unnecessary file fragmentation
76 - make TCP zerocopy and pinger support enabled by default (2476)
77 - sync writes completed after process exits caused crashes (2319)
78 - maintain correct mount count on the MDS (2356)
79 - backout 1557, because 2316 wasn't really fixed
80 - better file I/O statistics gathering in /proc
81 - don't take unnecessary, deadlock-inducing bug in readpage (2383)
82 - another kernel patch to fix zero-copy TCP function export
83 - don't take duplicate lock when processing re-sent getattr (2420)
84 - lctl uses obd_self_export instead of creating new conn (2353)
85 - MDS/OST recovery case which requires object creation asserted (2425)
86 - move lfs from /usr/sbin to /usr/bin in packages
87 - fix race between mds_client_add and mds_client_free (2417)
88 - use kmalloc instead of slabs in portals (2430)
89 - don't create duplicate records when a failover MDS is present (2442)
90 - remove unnecessary mount age check (2332)
91 - don't remove directory inodes from locks prematurely (2451)
92 - don't break if MDS service name is the same as hostname (2103)
93 - fix races in client write RPC generation when cache full (2482)
95 2003-12-13 Cluster File Systems, Inc. <info@clusterfs.com>
98 - remove now-unused request->rq_obd (278)
99 - if an allocation fails, print out how much memory we've used (1933)
100 - use PORTAL_SLAB_ALLOC for structures, to get GFP_MEMALLOC (1933)
101 - add the "configurable stack size" patch to most series files (1256)
102 - ability to write large log records, for 100+ OST configs (2306)
103 - fix NULL deref when filter_prep fails (2314)
104 - fix operator precedence error in filter_sync
105 - dynamic allocation of socknal TX descriptors (2315)
106 - fix a missed case in the GFP_MEMALLOC patch, can cause deadlock (2310)
107 - fix gcc 2.96 compilation problem in xattr kernel patch (2294)
108 - ensure that CWARN messages in Portals always get to the syslog
109 - __init/__exit are not for prototype decls (ldlm_init/exit)
110 - x86-64 compile warning fixes
111 - fix gateway LMC keyword conflict (2318)
112 - fix MDS lock inversions in getattr/reint paths (1844)
113 - fix a rare lock re-ordering bug, which caused deadlock (2322)
114 - fix i_sem/journal inversion in fsfilt_ext3_write_record (2306)
115 - DLM race condition prevented some lock evictions (2328)
116 - ENOMEM detection and retry on socknal sends (2230)
117 - use GFP_NOFS throughout Lustre, to combat ENOMEM (2230)
118 - move osc_rpcd into ptlrpc, for use in MDC and others (2329)
119 - protect MDS inode fsdata with stronger locking; fixes assertion (2313)
120 - better error messages when a client is rejected during recovery (1505)
121 - avoid cancelling locks which were never granted, after failure (2330)
122 - fix i_sem/journal inversion in mds_client_add (2333)
123 - fix truncate/getattr lock cycle deadlock (2334)
124 - use rpcd to send close; allows resend after timeout, avoid leak (1897)
125 - fix two rare exit paths which could leak an l_lock() ref (2321)
126 - fencepost error in MDS/OST orphan recovery (2226)
127 - make log record alignment 8 bytes (1988)
128 - lstripe now fails when requested offset > ost_count (2237)
129 - ensure that all kernel series have a complete list.h (1607)
130 - fix crashes in special-file operations (2316)
131 - lctl create/brw OID mismatch, caused by obsolete filter loop (2339)
133 - allow configurable automake binary, for testing new versions
134 - small update to the lfs documentation
136 2003-12-03 Cluster File Systems, Inc. <info@clusterfs.com>
138 * fix negative export reference count in fsfilt_sync (2312)
140 2003-12-01 Cluster File Systems, Inc. <info@clusterfs.com>
141 * release candidate 0.9.1
143 - orphans are moved into the PENDING directory for possible recovery
144 - replayed opens now open by fid for orphan/rename safety (1042)
145 - last close of an orphan inode generates a transno (683)
146 - chdir() and mount() now pin the directory entry (1020)
147 - avoid CERROR in normal ll_setattr_raw() error case (1500)
148 - discard very old requests without processing them (1502)
149 - remove some common, well-understood CERRORs (1505)
150 - require O_DIRECT I/O to be page-sized to workaround IA64 crash (1609)
151 - clear "grant" flags in OST replies until OST grant code lands (1644)
152 - fix read performance by not clobbering i_blksize on client (1598)
153 - fix __ldlm_handle2lock oops by not dereferencing lock after PUT (1625)
154 - make LRU size a /proc tunable, clears locks when reduced (707)
155 - fix some lprocfs rot that prevented ptlbd from loading (1732)
156 - server locks take references on exports now (1558)
157 - build fixes for 2.4.20-rh trees (1663)
158 - return an error from lov_create if all OSCs are inactive (1751)
159 - fix import levels when a reconnect happens without a timeout (1597)
160 - exit early from mds_open if we get a lookup error (1749)
161 - partial page read at EOF wouldn't wait for disk before sending (1642)
162 - avoid NULL deref in obdfilter when reading page past EOF (1592)
163 - avoid LASSERT in ll_intent_lock if server failed very early (1090)
164 - fix LBUG in ll_it_open_error with rc = -2 (1861)
165 - write/truncate lock inversion (1639)
166 - Don't auto-load obdclass, portals modules during cleanup (1495)
167 - fix timestamps from jumping to "now" (1763)
168 - extra journal assertions (1648)
169 - add an extra multiunlink test (1771)
170 - fix read_record/write_record API (1776)
171 - fix leak of offset_extent, possible incorrect i_size later (1772)
172 - fix lasserts in mis-matched transnos during open-unlink testing (1541)
173 - Debugging for the kqswnal_get_idle_tx problems (1820)
174 - Allow recovery to be attempted multiple times (1536)
175 - Write out MDS last_rcvd file after it is first created (1600)
176 - Fix tx_descriptor leak in failed transmit situations (1827)
177 - ext3 journaling fixes for assertion failure after IO error (1871)
178 - class_export_put() on freed export after completion AST error (1896)
179 - Fix revalidate looping in VFS (1322)
180 - Don't access a freed export during MDS_REINT timeout (1521)
181 - Add open-unlink recovery support on the MDS (1673,1764)
182 - Return an error if no MDS data was read from last_rcvd (1946)
183 - Fix for lookup "." or ".." crash on error (1932,1931,1935)
184 - Don't setup a disk device that doesn't match exported UUID (317)
185 - Reduce bulk RPC timeout to avoid cascading client/OST failures (1845)
186 - avoid committing NULL handle in force close
187 - local.sh is now a one-stripe LOV configuration
188 - POSIX utime.4 -EPERM on FIFO not owned by user (56)
189 - fix ext3 htree duplicate directory entry corruption (1516)
190 - POSIX creat.13, fstat.1, open.18, stat.3 new file atime/mtime (2020)
191 - update to new LOV EA format (2097)
192 - interoperability for different PAGE_SIZE/wordsize (686,1821,1343,2042)
194 2003-06-15 Phil Schwan <phil@clusterfs.com>
197 - imports and exports cleanup too early, need refcounts (349, 879, 1045)
198 - per-import/export recovery handling (958, 931, 959)
199 - multiple last-rcvd slots, for serving multiple FSes (949)
200 - connections are again shared between multiple imp/exports (963, 964)
201 - "umount -f" would hang if any requests needed to be sent (393, 978)
202 - avoid pinning large req buffer by copying for queued messages (989)
203 - add "uuid" to "lctl device" command to help upcalls (991)
204 - "open" RPCs with transnos would confuse recovery counters (1037)
205 - do proper endian conversion of all wire messages (288, 340, 891)
206 - remove OST bulk get LBUGs, fix ost_brw_write cleanup (1126)
207 - call waiting locks callback from LDLM recovery thread (1127, 1151)
208 - fix ptlrpc_connection leak in target_handle_connect (1174)
209 - fix import refcounting bug in OST and MDS cleanup (1134)
210 - if an invalid-at-open-time OSC returned before close(), LBUG (1150)
211 - fix very unlikely obd_types race condition (501)
212 - remove osc_open hack for echo_client (1187)
213 - we leaked exports/dlmimps for forcibly disconnected clients (1143)
214 - a failure in read_inode2 leads to deadlock (1139)
215 - cancel ack-locks as soon as transaction is committed (1072)
216 - fix major leaks and crashes in the bulk I/O path (937, 1057)
217 - make sure to commitrw after any preprw to avoid deadlock (1162)
218 - failing to execute a file in a lustre FS would lock inode (1203)
219 - small DEBUG_REQ fix to avoid dereferencing a NULL (1227)
220 - don't ASSERT while cleaning up an incompletely-setup obd (1248)
221 - obd_uuid2tgt would walk off the end of the list (1255)
222 - on IA64 the osc would give portals incorrect bulk size (1258)
223 - fix debug daemon ioctl interface; allows daemon on ia64 (1274)
224 - fix lock inversion caused by new llite matching code (1282)
225 - limit the number of dirty pages on a client to 10MB (1286)
226 - timed out locks were not being corrected cancelled (1289)
227 - fix O_DIRECT above 4GB on IA-32 (1292)
228 * major user-visible changes
229 - fail out/fail over policy now controlled by the upcall (993)
231 - add OBD_PING to check server availability and failure (954)
232 - lustre messages are now sent in sending host order (288, 340, 891)
233 - add eadatalen to MDS getattr reply (340)
234 - OST read replies may contain second buffer, with per-page status (593)
236 2003-03-11 Phil Schwan <phil@clusterfs.com>
239 - LDLM_DEBUG macro fix, for gcc 3.2 (850)
240 - failed open()s could cause deadlock; fixed (867, 869)
241 - stop cancelling OST locks when files are closed (481)
242 - overlapping XID spaces caused network corruption (851, 853)
243 - fix unsafe fsfilt counter arithmetic; change to atomic_t
244 - setattr_raw added, to do single-RPC, server-side setattrs
245 - lmc/lconf syntax change for OST UUIDs
246 - fix crashy race condition between ptlrpc_free_req and osc_close
247 - don't use request in mdc_enqueue if we hit a timeout (889)
248 - don't set the inode i_size for regular files from the MDS (896)
249 - handle out of order completion AST (842)
250 - don't LBUG if a lock request times out after receiving AST (913)
251 - avoid d_rehash race in ll_find_alias by rehashing inside dcache_lock
252 - if a bad lock AST arrives, send an error instead of dropping entirely
253 - return 0 from revalidate2 if ll_intent_lock returns -EINTR (912)
254 - fix leak in bulk IO when only partially completed (899, 900, 926)
255 - fix O_DIRECT for ia64 (55)
256 - (almost) eliminate Lustre-kernel-thread effects on load average (722)
257 - C-z after timeout could hang a process forever; fixed (977)
259 - client-side I/O cache (678, 924, 929, 941, 970)
261 - READPAGE and SETATTRs which don't take server-side locks get
264 2003-02-11 Phil Schwan <phil@clusterfs.com>
267 - Fix ldlm_lock_match on the MDS to avoid matching remote locks (592)
268 - Fix fsfilt_extN_readpage() to read a full page of directory
269 entries, or fake the remainder if PAGE_SIZE != blocksize (500)
270 - Avoid extra mdc_getattr() in ll_intent_lock when possible (534, 604)
271 - Fix imbalanced LOV object allocation and out-of-bound access (469)
272 - Most intent operations were removed, in favour of a new RPC mode
273 that does a single RPC to the server and bypasses most of the VFS
274 - All LDLM resource ID arrays were removed in favour of ldlm_res_id
275 - Aggressively cancel local locks on DLM servers
276 - mds_reint_unlink sends EA to the client if it's the last nlink.
277 client uses that EA to unlink OST objects.
278 - mds_reint_{rename,unlink,link} were rewritten to take ordered locks
279 - recursive symlinks were fixed (439)
280 - fixed NULL deref in DEBUG_REQ
281 - filter_update_lastobjid no longer calls sync, which annoyed extN
282 - fixed multi-client small-writes to a single file problem (445)
283 - fixed mtime updates during file writes (607)
284 - fixed vector writes on obdfilter causing problems when ENOSPC (670)
285 - fixed bug in obd_brw_read/write() (under guise of testing 367)
286 - fixed Linux OST size reporting problem (444, 656)
287 - OST now updates object mtime with writes or setattr (607, 619)
288 - client verifies file size before zeroing page past EOF (445)
289 - OST now writes last allocated objid to disk with allocation (108)
290 - LOV on echo now works (409)
292 - mds_reint_unlink sends a new buffer, with the EA included. this
293 buffer is only valid if body->valid & OBD_MD_FLEASIZE, which is only
294 set if a regular file was being unlinked, and it was the last link
295 - use PtlGet from the target for bulk writes (315)
296 - OST now updates object mtime with writes or setattr (607, 619)
297 - LDLM now has a grant-time callback to revalidate locked items, if
299 - Many MDS operations were reorganized to combat race conditions
301 - Merge b_intel branch (updated lprocfs code) - now at /proc/fs/lustre
302 - configure check to avoid gcc version 2.96 20000731-2.96-98) (606)
304 2003-01-06 Andreas Dilger <adilger@clusterfs.com>
307 - Fully reactivate OST imports after reconnection (512, others)
308 - Make sure client sees our -ENOTCONN from mds_handle (513 - partial)
309 - More graceful error handling for truncating on dead OST (515)
310 - Don't error out unless we're actually accessing dead stripes (474)
311 - Fix garbage sizes when stripes are missing (410)
312 - LRU counters were broken, causing constant lock purge (433, 432)
313 - garbage on read from stripes with failed OSTs (441)
314 - mark OSCs as active before reconnecting during recovery (438)
315 - lov_enqueue and lov_cancel need to handle inactive OSTs (403)
316 - lfind did not preserve OST order in output (443)
317 - symlinks cause hung clients, incorrect data (439)
318 - stop dereferencing request after dropping refcount (457)
319 - don't LASSERT(spin_is_locked) on non-SMP (455)
320 - fixes for many rename() bugs
321 - fstat didn't correctly synchronize attributes (399)
322 - server must handle lock cancellation during blocking AST prep (487)
323 - bulk descriptors were free()d too soon (511)
324 - fix paths in lconf, which would load incorrect modules (451, 507)
325 - fix confusing lconf 'host not found' error message (386)
326 - fix lock order deadlock on OST (O/R i_sem before journal ops, 478)
327 - fix race condition in mdc_blocking_ast() for inode access (526)
328 - fix lov_unpackmd() unpacking wrong number of stripes (537)
329 - fix lov_set_osc_active() marking wrong OSC inactive (440)
330 - fix bad lstripe lov_unpackmd() assertion (fix layering too) (527)
331 - fix multiple writes of stripe MD to MDS (358, maybe 519)
332 - fix lstripe in several ways (kernel side) (527)
333 - fix request leak in ldlm_cli_enqueue (262)
334 - incorrect OSC was marked inactive after OST failure
335 - call mds_fs_cleanup before unmounting filesystem (524)
336 - fix races between taking ns_lock and ldlm_lock_change_resource
337 - fix races updating LOV export open file list
338 - fix lov_enqueue error path, avoid decref-ing bad lock handle
339 - fix recovery NULL deref in ldlm_cli_cancel_unused
340 - fix some DLM races by using new hash table for lock handles (419)
341 - permit the client to specify desired inodes, at replay
342 - duplicate requests when we queue them for replay reintegration
343 - fix last_rcvd offset calculation
344 - sync after each recovered transaction, so we always make progress
345 - never, not always, ERESTART requests without transnos
346 - store the lov_desc in the MDS, so we don't depend on getlovinfo to
348 - skip replay if the MDS says that the client is already connected
349 - don't check for a recovery-enabled export to match lctl's UUID
350 - don't INC_USE_COUNT for phantom exports
351 - don't crash when cleaning up phantom exports (567)
352 - don't double-finish or set replay data for errored mdc_open requests
353 - abort requests when they time out, so we don't get old replies
354 - send/receive replies for AST messages again
355 - if the client says that it doesn't have the lock, cancel it on the
357 - if we timeout during I/O, don't try to cancel an in-use lock; instead
358 mark it as destroyed, it will all work out when decref is called
359 - fix module use counts (22, 581)
361 - ASTs now expect a reply (server cancels lock on error reply)
363 2002-12-02 Andreas Dilger <adilger@clusterfs.com>
366 - fix many simultaneous client startup (392)
367 - fix dentry->d_it clobbering
368 - credentials weren't being shipped for readdir/getattr operations
369 - remove invalid assertions triggered during some concurrent MD
371 - proper Lustre versions added (336, 389)
372 - fix memory leak for create error case (398)
373 - fix LOV locking bug that would get cli/srv out of sync
374 - fix echo client over LOV (409)
375 - fix dbench 2, extN refcount problem (170, 258, 356, 418)
376 - fix double-O_EXCL intent crash (424)
377 - avoid sending multiple lock CANCELs (352)
379 - MDS can do multi-client recovery (modulo bugs in new code)
381 - many updates, edits, cleanups
383 2002-11-18 Phil Schwan <phil@clusterfs.com>
386 - fix null d_it dereference (346)
387 - fix full OST/dbench hang (333)
388 - fix permission problem with file removal (286)
389 - fix removal of OSCs from LOV when they fail
390 - fix NULL deref during bulk timeout (214)
391 - fix problems related to multiple filesystems on one MDS (241)
392 - fixed serious subtle metadata locking bugs
393 - free locks on clients when inodes are removed due to memory
395 - fix inode pointer in lock data (285)
396 - partial support for multiple MDS on a single host (241)
397 - data locks weren't cancelled at clear_inode time (290, 311)
398 - intent locks could lead to unbounded lock growth (205)
399 - added a maximum lock count, an LRU list, and a flusher
400 - fix multiple rename (365)
401 - properly abstracted the echo client
402 - OSC locked 1 byte too many; fixed
403 - rewrote brw callback code:
404 - fixed recovery bugs related to LOVs (306)
405 - fixed too-many-pages-in-one-write crash (191)
406 - fixed (again) crash in sync_io_timeout (214)
407 - probably fixed callback-related race (385)
409 - Add capability to MDS protocol
410 - LDLM cancellations and callbacks on different portals
412 2002-10-28 Andreas Dilger <adilger@clusterfs.com>
415 - limit client IOV size to PTL_MD_MAX_IOV (611336, 191)
416 - defer open object destruction to close time (601981, 138)
417 - open/close OST file handle in obdo (OBD_MD_FLHANDLE) (601981, 138)
418 - move LDLM_ENQUEUE/CONVERT back to MDS portal (625069)
419 - abstract ll_lookup2, fix ll_revalidate2 to use abstraction (256)
420 - don't call obd_setattr in ll_file_release for destroyed objects
421 * protocol change to lustre_msg: move |version| and add |flags|
422 * protocol change to osc_punch: "start" in "o_size", "end" in "o_blocks"
423 * lock replay: for LDLM_FL_REPLAY trust client to do right thing
424 * added replay of create, unlink, link and rename operations during
425 MDS failover; recovery should be much more robust now
426 * remove failed OSCs from LOVs (only lov_create uses this so far)
427 * the lustre-HOWTO was brought (more) up to date (582544)
429 2002-10-23 Phil Schwan <phil@clusterfs.com>
432 - in-use dentries weren't being reused properly (617851)
433 - prevent multiple LDLM setup (599178)
434 - fix LOV size calculations for truncate (617853)
435 - fix client handling of MDS intent errors (POSIX)
436 - fix permission bug in lovstripe.c test (624321)
437 - fix MDS thread deadlock - move LDLM handler to DLM portal (625069)
438 - truncate past end of file could corrupt data
439 - proper cleanup after timeouts, crashes, etc (592524, 550815)
440 - a race in recovery could return ETIMEDOUT to apps (623947)
441 - building outside the source directory was fixed
442 * the lustre-HOWTO was brought (more) up to date (582544)
443 * major progress was made on recovery functionality
445 2002-10-10 Phil Schwan <phil@clusterfs.com>
448 - recovery deadlock fix
449 - rm -rf causes LBUG fix (617817)
450 - file open by multiple tasks fix (618962)
451 - directory permissions bugs (602707 and 620007)
452 - journal_stop fixed with locking (611313)
453 - O_APPEND failures resolved (618273, perhaps 614459)
454 - lconf PATH fix (619770)
455 - IA64 build fix (621450)
456 - RPC buffer sizes scale with amount of memory
458 2002-10-01 Phil Schwan <phil@clusterfs.com>
461 - locks would be cancelled without throwing away data pages,
462 resulting in inconsistent data (605627)
463 - inode attributes were not always being refreshed (605627, 612449)
464 - lconf now continues to cleanup after lctl reports an error
465 - MDS now enforces user permissions (602707)
466 - lprocfs cleanup fixed, but not yet enabled (614157)
467 - fixed infinite server hang, should a client not respond to an AST
468 - avoid going into recovery if user calls readlink() with a buffer
469 that's too small (613941)
470 - AST RPCs no longer require replies (614867) -- this may be changed
471 - don't crash server if client sends an IOV that's too big (611336)
472 - fixed lock conversion deadlock (611892)
473 - fixed the following of symlinks (614622)
474 * recovery: the server can remove locks from a client that dies, other
475 clients can make progress
476 * more extN patch fixes
477 * compile-time configurable ptlrpc buffer allocations
479 - collaborative read cache document
480 - Lustre Lite Performance CDR document-in-progress
482 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
485 - fix typo in patch-2.4.18
487 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
490 - clear ptlrpc request each time in handle_incoming_request()
491 - unlink of files now destroys the object on the OST
493 2002-09-19 Peter Braam <braam@clusterfs.com>
495 * add hard link support
496 * change obdfile creation method
497 * kernel patch changed
499 2002-09-19 Peter Braam <braam@clusterfs.com>
502 - stack overflow bug in extN fixed
504 2002-09-18 Andreas Dilger <adilger@clusterfs.com>
506 * documentation updates
507 - add man pages for config tools
508 - update tests/README to describe testing with new config tools
509 - finish metadata API descriptions
510 * bug fixes and cleanups
511 - statfs workaround for 16TB limit
512 - LOV stripe allocation improved, can stripe on subset of OSTs
513 - LOV file size/IO offset was wrong for files > 4GB in size
514 - object EA data was being dropped, caused files to be unreadable
515 - memory overflow with non-LOV OST caused memory corruption
516 - fixed regression tests to work with new config tools, obdfilter
517 - fixed bug when directory size became larger than 1 block
518 - fixed bug (for single client case) when PWD was deleted
519 - invalidate local directory pages when doing intent-based ops
520 - avoid LDLM oops when lock callback contained bad data
522 2002-09-09 Andreas Dilger <adilger@clusterfs.com>
524 * documentation updates
525 * bug fixes and cleanups
526 - configuration tools
529 - 64-bit compile warnings
530 - 64-bit internal statfs data
532 * test_brw on persistent OST devices
536 2002-09-04 Andreas Dilger <adilger@clusterfs.com>
538 * documentation updates
539 * bug fixes and cleanups
540 * configuration tools
542 2002-08-30 Peter J. Braam <braam@clusterfs.com>
545 * many small fixes to 0_5_4
546 * io/network handling
547 * thinkos in MDS operations
549 2002-08-24 Peter J. Braam <braam@clusterfs.com>
552 * crucial basic fixes to 0.5.3
553 * IOR, Iozone work over Elan
556 2002-08-07 Phil Schwan <phil@clusterfs.com>
557 * version 0_5_3, our first alpha
558 * we use the new Portals iovs
559 * documentation updates
560 * bug fixes and cleanups
561 * small changes in the DLM wire protocol
563 2002-07-25 Peter J. Braam <braam@clusterfs.com>
564 * version 0_5_1 with some initial stability,
565 * locking on MD and file I/O.
566 * documentation updates
567 * several bug fixes since 0.5.0
568 * small changes in wire protocol
570 2002-07-18 Phil Schwan <phil@clusterfs.com>
572 * delivered as Lustre Light Alpha
573 * fixed a crash after handling invalid MDS requests
574 * fixed directory pages for architectures with non-4k pages sizes
576 2002-07-11 Andreas Dilger <adilger@clusterfs.com>
577 * release version v0_4_4
578 * Moves TCP acceptor to be on port 2432 (unused Coda port) instead
580 * Fixes a number of interruption problems with OST operations.
581 * Update documentation for portals header changes
582 * Move all wire protocol structs/defines to lustre_idl.h
583 * Fixes symlink length bug.
584 * Add tcpdump to repository.
586 2002-07-05 Andreas Dilger <adilger@clusterfs.com>
587 * release version v0_4_3
588 * Fixes statfs for inodes on extN.
589 * Fixes bug in runtests which would delete /etc/hosts.
590 * Use 64-bit object IDs wherever possible (not into VFS though)
591 Remove ost_get_info, which is unused by lustre, and out of date.
593 2002-07-03 Peter Braam <braam@clusterfs.com>
594 * release version v0_4_2 Fixes a lookup error (type not passed)
595 * move forward to head of Portals
596 * move forward to latest Lustre kernel
598 2002-06-25 Peter Braam <braam@clusterfs.com>
599 * release version v0_4_1. Hopefully stable on single node use.