1 tbd Cluster File Systems, Inc. <info@clusterfs.com>
4 - orphan handling to avoid losing space on client/server crashes
6 2004-01-27 Cluster File Systems, Inc. <info@clusterfs.com>
9 - add series for the vanilla 2.6.0 kernel
10 - add series for the vanilla 2.4.24 kernel
11 - add series for a cray x86/64 UL kernel drop
12 - fix xattr patches for the vanilla 2.4.19 series
14 - generate true UUIDs in lmc (1171)
15 - have portals stack dumping break in UML (2466)
16 - avoid bad dchild deref; avoid inum lock w/o creation (2362)
17 - allocate with _NOFS in ldlm to avoid deadlock (1933)
18 - wake callback waiting threads on client eviction (2460)
19 - Add --ptldebug and --subsystem to lmc (1719)
20 - update assertion to allow safe interrupt allocation
21 - set rq_no_resend for cancel requests (2432)
22 - recalculate ptlrpcd timeout after resend (2494)
23 - call vfs_rmdir when removing pending directories (2368)
24 - fix renaming a file to itself (2429)
25 - lmc creates a default one-stripe lov (2454)
26 - expand procfs space to handle large clusters (2326)
27 - increase UML stack to avoid overflow
28 - update lconf's list of debug and subsystem masks
29 - fix lfs find --obd (2510)
30 - /proc tunable for disabling filter read caching (2591)
31 - stop rpm packages from altering slapd.conf (2301)
32 - disable nagle in the socknal under 0conf (2578)
33 - choose mds inode size based on stripe count (2572)
34 - fix kernel-source rpm problems (2516)
36 - add --disable-doc to avoid pdf generation (2421)
37 - update documentation, tests, type-os, comments
38 - avoid format warnings on ia64
40 - tiny code cleanups by removing unused fields
42 2004-01-07 Cluster File Systems, Inc. <info@clusterfs.com>
45 - fix obvious semaphore misuse in as-yet-unused setattr path (2348)
46 - remove the most blatant lies from BUILDING file (2371)
47 - change default debug level to reasonable production setting
48 - reduce client side cache size to reduce cache flush time
49 - reduce max RPCs in flight to avoid unnecessary file fragmentation
50 - make TCP zerocopy and pinger support enabled by default (2476)
51 - sync writes completed after process exits caused crashes (2319)
52 - maintain correct mount count on the MDS (2356)
53 - backout 1557, because 2316 wasn't really fixed
54 - better file I/O statistics gathering in /proc
55 - don't take unnecessary, deadlock-inducing bug in readpage (2383)
56 - another kernel patch to fix zero-copy TCP function export
57 - don't take duplicate lock when processing re-sent getattr (2420)
58 - lctl uses obd_self_export instead of creating new conn (2353)
59 - MDS/OST recovery case which requires object creation asserted (2425)
60 - move lfs from /usr/sbin to /usr/bin in packages
61 - fix race between mds_client_add and mds_client_free (2417)
62 - use kmalloc instead of slabs in portals (2430)
63 - don't create duplicate records when a failover MDS is present (2442)
64 - remove unnecessary mount age check (2332)
65 - don't remove directory inodes from locks prematurely (2451)
66 - don't break if MDS service name is the same as hostname (2103)
67 - fix races in client write RPC generation when cache full (2482)
69 2003-12-13 Cluster File Systems, Inc. <info@clusterfs.com>
72 - remove now-unused request->rq_obd (278)
73 - if an allocation fails, print out how much memory we've used (1933)
74 - use PORTAL_SLAB_ALLOC for structures, to get GFP_MEMALLOC (1933)
75 - add the "configurable stack size" patch to most series files (1256)
76 - ability to write large log records, for 100+ OST configs (2306)
77 - fix NULL deref when filter_prep fails (2314)
78 - fix operator precedence error in filter_sync
79 - dynamic allocation of socknal TX descriptors (2315)
80 - fix a missed case in the GFP_MEMALLOC patch, can cause deadlock (2310)
81 - fix gcc 2.96 compilation problem in xattr kernel patch (2294)
82 - ensure that CWARN messages in Portals always get to the syslog
83 - __init/__exit are not for prototype decls (ldlm_init/exit)
84 - x86-64 compile warning fixes
85 - fix gateway LMC keyword conflict (2318)
86 - fix MDS lock inversions in getattr/reint paths (1844)
87 - fix a rare lock re-ordering bug, which caused deadlock (2322)
88 - fix i_sem/journal inversion in fsfilt_ext3_write_record (2306)
89 - DLM race condition prevented some lock evictions (2328)
90 - ENOMEM detection and retry on socknal sends (2230)
91 - use GFP_NOFS throughout Lustre, to combat ENOMEM (2230)
92 - move osc_rpcd into ptlrpc, for use in MDC and others (2329)
93 - protect MDS inode fsdata with stronger locking; fixes assertion (2313)
94 - better error messages when a client is rejected during recovery (1505)
95 - avoid cancelling locks which were never granted, after failure (2330)
96 - fix i_sem/journal inversion in mds_client_add (2333)
97 - fix truncate/getattr lock cycle deadlock (2334)
98 - use rpcd to send close; allows resend after timeout, avoid leak (1897)
99 - fix two rare exit paths which could leak an l_lock() ref (2321)
100 - fencepost error in MDS/OST orphan recovery (2226)
101 - make log record alignment 8 bytes (1988)
102 - lstripe now fails when requested offset > ost_count (2237)
103 - ensure that all kernel series have a complete list.h (1607)
104 - fix crashes in special-file operations (2316)
105 - lctl create/brw OID mismatch, caused by obsolete filter loop (2339)
107 - allow configurable automake binary, for testing new versions
108 - small update to the lfs documentation
110 2003-12-03 Cluster File Systems, Inc. <info@clusterfs.com>
112 * fix negative export reference count in fsfilt_sync (2312)
114 2003-12-01 Cluster File Systems, Inc. <info@clusterfs.com>
115 * release candidate 0.9.1
117 - orphans are moved into the PENDING directory for possible recovery
118 - replayed opens now open by fid for orphan/rename safety (1042)
119 - last close of an orphan inode generates a transno (683)
120 - chdir() and mount() now pin the directory entry (1020)
121 - avoid CERROR in normal ll_setattr_raw() error case (1500)
122 - discard very old requests without processing them (1502)
123 - remove some common, well-understood CERRORs (1505)
124 - require O_DIRECT I/O to be page-sized to workaround IA64 crash (1609)
125 - clear "grant" flags in OST replies until OST grant code lands (1644)
126 - fix read performance by not clobbering i_blksize on client (1598)
127 - fix __ldlm_handle2lock oops by not dereferencing lock after PUT (1625)
128 - make LRU size a /proc tunable, clears locks when reduced (707)
129 - fix some lprocfs rot that prevented ptlbd from loading (1732)
130 - server locks take references on exports now (1558)
131 - build fixes for 2.4.20-rh trees (1663)
132 - return an error from lov_create if all OSCs are inactive (1751)
133 - fix import levels when a reconnect happens without a timeout (1597)
134 - exit early from mds_open if we get a lookup error (1749)
135 - partial page read at EOF wouldn't wait for disk before sending (1642)
136 - avoid NULL deref in obdfilter when reading page past EOF (1592)
137 - avoid LASSERT in ll_intent_lock if server failed very early (1090)
138 - fix LBUG in ll_it_open_error with rc = -2 (1861)
139 - write/truncate lock inversion (1639)
140 - Don't auto-load obdclass, portals modules during cleanup (1495)
141 - fix timestamps from jumping to "now" (1763)
142 - extra journal assertions (1648)
143 - add an extra multiunlink test (1771)
144 - fix read_record/write_record API (1776)
145 - fix leak of offset_extent, possible incorrect i_size later (1772)
146 - fix lasserts in mis-matched transnos during open-unlink testing (1541)
147 - Debugging for the kqswnal_get_idle_tx problems (1820)
148 - Allow recovery to be attempted multiple times (1536)
149 - Write out MDS last_rcvd file after it is first created (1600)
150 - Fix tx_descriptor leak in failed transmit situations (1827)
151 - ext3 journaling fixes for assertion failure after IO error (1871)
152 - class_export_put() on freed export after completion AST error (1896)
153 - Fix revalidate looping in VFS (1322)
154 - Don't access a freed export during MDS_REINT timeout (1521)
155 - Add open-unlink recovery support on the MDS (1673,1764)
156 - Return an error if no MDS data was read from last_rcvd (1946)
157 - Fix for lookup "." or ".." crash on error (1932,1931,1935)
158 - Don't setup a disk device that doesn't match exported UUID (317)
159 - Reduce bulk RPC timeout to avoid cascading client/OST failures (1845)
160 - avoid committing NULL handle in force close
161 - local.sh is now a one-stripe LOV configuration
162 - POSIX utime.4 -EPERM on FIFO not owned by user (56)
163 - fix ext3 htree duplicate directory entry corruption (1516)
164 - POSIX creat.13, fstat.1, open.18, stat.3 new file atime/mtime (2020)
165 - update to new LOV EA format (2097)
166 - interoperability for different PAGE_SIZE/wordsize (686,1821,1343,2042)
168 2003-06-15 Phil Schwan <phil@clusterfs.com>
171 - imports and exports cleanup too early, need refcounts (349, 879, 1045)
172 - per-import/export recovery handling (958, 931, 959)
173 - multiple last-rcvd slots, for serving multiple FSes (949)
174 - connections are again shared between multiple imp/exports (963, 964)
175 - "umount -f" would hang if any requests needed to be sent (393, 978)
176 - avoid pinning large req buffer by copying for queued messages (989)
177 - add "uuid" to "lctl device" command to help upcalls (991)
178 - "open" RPCs with transnos would confuse recovery counters (1037)
179 - do proper endian conversion of all wire messages (288, 340, 891)
180 - remove OST bulk get LBUGs, fix ost_brw_write cleanup (1126)
181 - call waiting locks callback from LDLM recovery thread (1127, 1151)
182 - fix ptlrpc_connection leak in target_handle_connect (1174)
183 - fix import refcounting bug in OST and MDS cleanup (1134)
184 - if an invalid-at-open-time OSC returned before close(), LBUG (1150)
185 - fix very unlikely obd_types race condition (501)
186 - remove osc_open hack for echo_client (1187)
187 - we leaked exports/dlmimps for forcibly disconnected clients (1143)
188 - a failure in read_inode2 leads to deadlock (1139)
189 - cancel ack-locks as soon as transaction is committed (1072)
190 - fix major leaks and crashes in the bulk I/O path (937, 1057)
191 - make sure to commitrw after any preprw to avoid deadlock (1162)
192 - failing to execute a file in a lustre FS would lock inode (1203)
193 - small DEBUG_REQ fix to avoid dereferencing a NULL (1227)
194 - don't ASSERT while cleaning up an incompletely-setup obd (1248)
195 - obd_uuid2tgt would walk off the end of the list (1255)
196 - on IA64 the osc would give portals incorrect bulk size (1258)
197 - fix debug daemon ioctl interface; allows daemon on ia64 (1274)
198 - fix lock inversion caused by new llite matching code (1282)
199 - limit the number of dirty pages on a client to 10MB (1286)
200 - timed out locks were not being corrected cancelled (1289)
201 - fix O_DIRECT above 4GB on IA-32 (1292)
202 * major user-visible changes
203 - fail out/fail over policy now controlled by the upcall (993)
205 - add OBD_PING to check server availability and failure (954)
206 - lustre messages are now sent in sending host order (288, 340, 891)
207 - add eadatalen to MDS getattr reply (340)
208 - OST read replies may contain second buffer, with per-page status (593)
210 2003-03-11 Phil Schwan <phil@clusterfs.com>
213 - LDLM_DEBUG macro fix, for gcc 3.2 (850)
214 - failed open()s could cause deadlock; fixed (867, 869)
215 - stop cancelling OST locks when files are closed (481)
216 - overlapping XID spaces caused network corruption (851, 853)
217 - fix unsafe fsfilt counter arithmetic; change to atomic_t
218 - setattr_raw added, to do single-RPC, server-side setattrs
219 - lmc/lconf syntax change for OST UUIDs
220 - fix crashy race condition between ptlrpc_free_req and osc_close
221 - don't use request in mdc_enqueue if we hit a timeout (889)
222 - don't set the inode i_size for regular files from the MDS (896)
223 - handle out of order completion AST (842)
224 - don't LBUG if a lock request times out after receiving AST (913)
225 - avoid d_rehash race in ll_find_alias by rehashing inside dcache_lock
226 - if a bad lock AST arrives, send an error instead of dropping entirely
227 - return 0 from revalidate2 if ll_intent_lock returns -EINTR (912)
228 - fix leak in bulk IO when only partially completed (899, 900, 926)
229 - fix O_DIRECT for ia64 (55)
230 - (almost) eliminate Lustre-kernel-thread effects on load average (722)
231 - C-z after timeout could hang a process forever; fixed (977)
233 - client-side I/O cache (678, 924, 929, 941, 970)
235 - READPAGE and SETATTRs which don't take server-side locks get
238 2003-02-11 Phil Schwan <phil@clusterfs.com>
241 - Fix ldlm_lock_match on the MDS to avoid matching remote locks (592)
242 - Fix fsfilt_extN_readpage() to read a full page of directory
243 entries, or fake the remainder if PAGE_SIZE != blocksize (500)
244 - Avoid extra mdc_getattr() in ll_intent_lock when possible (534, 604)
245 - Fix imbalanced LOV object allocation and out-of-bound access (469)
246 - Most intent operations were removed, in favour of a new RPC mode
247 that does a single RPC to the server and bypasses most of the VFS
248 - All LDLM resource ID arrays were removed in favour of ldlm_res_id
249 - Aggressively cancel local locks on DLM servers
250 - mds_reint_unlink sends EA to the client if it's the last nlink.
251 client uses that EA to unlink OST objects.
252 - mds_reint_{rename,unlink,link} were rewritten to take ordered locks
253 - recursive symlinks were fixed (439)
254 - fixed NULL deref in DEBUG_REQ
255 - filter_update_lastobjid no longer calls sync, which annoyed extN
256 - fixed multi-client small-writes to a single file problem (445)
257 - fixed mtime updates during file writes (607)
258 - fixed vector writes on obdfilter causing problems when ENOSPC (670)
259 - fixed bug in obd_brw_read/write() (under guise of testing 367)
260 - fixed Linux OST size reporting problem (444, 656)
261 - OST now updates object mtime with writes or setattr (607, 619)
262 - client verifies file size before zeroing page past EOF (445)
263 - OST now writes last allocated objid to disk with allocation (108)
264 - LOV on echo now works (409)
266 - mds_reint_unlink sends a new buffer, with the EA included. this
267 buffer is only valid if body->valid & OBD_MD_FLEASIZE, which is only
268 set if a regular file was being unlinked, and it was the last link
269 - use PtlGet from the target for bulk writes (315)
270 - OST now updates object mtime with writes or setattr (607, 619)
271 - LDLM now has a grant-time callback to revalidate locked items, if
273 - Many MDS operations were reorganized to combat race conditions
275 - Merge b_intel branch (updated lprocfs code) - now at /proc/fs/lustre
276 - configure check to avoid gcc version 2.96 20000731-2.96-98) (606)
278 2003-01-06 Andreas Dilger <adilger@clusterfs.com>
281 - Fully reactivate OST imports after reconnection (512, others)
282 - Make sure client sees our -ENOTCONN from mds_handle (513 - partial)
283 - More graceful error handling for truncating on dead OST (515)
284 - Don't error out unless we're actually accessing dead stripes (474)
285 - Fix garbage sizes when stripes are missing (410)
286 - LRU counters were broken, causing constant lock purge (433, 432)
287 - garbage on read from stripes with failed OSTs (441)
288 - mark OSCs as active before reconnecting during recovery (438)
289 - lov_enqueue and lov_cancel need to handle inactive OSTs (403)
290 - lfind did not preserve OST order in output (443)
291 - symlinks cause hung clients, incorrect data (439)
292 - stop dereferencing request after dropping refcount (457)
293 - don't LASSERT(spin_is_locked) on non-SMP (455)
294 - fixes for many rename() bugs
295 - fstat didn't correctly synchronize attributes (399)
296 - server must handle lock cancellation during blocking AST prep (487)
297 - bulk descriptors were free()d too soon (511)
298 - fix paths in lconf, which would load incorrect modules (451, 507)
299 - fix confusing lconf 'host not found' error message (386)
300 - fix lock order deadlock on OST (O/R i_sem before journal ops, 478)
301 - fix race condition in mdc_blocking_ast() for inode access (526)
302 - fix lov_unpackmd() unpacking wrong number of stripes (537)
303 - fix lov_set_osc_active() marking wrong OSC inactive (440)
304 - fix bad lstripe lov_unpackmd() assertion (fix layering too) (527)
305 - fix multiple writes of stripe MD to MDS (358, maybe 519)
306 - fix lstripe in several ways (kernel side) (527)
307 - fix request leak in ldlm_cli_enqueue (262)
308 - incorrect OSC was marked inactive after OST failure
309 - call mds_fs_cleanup before unmounting filesystem (524)
310 - fix races between taking ns_lock and ldlm_lock_change_resource
311 - fix races updating LOV export open file list
312 - fix lov_enqueue error path, avoid decref-ing bad lock handle
313 - fix recovery NULL deref in ldlm_cli_cancel_unused
314 - fix some DLM races by using new hash table for lock handles (419)
315 - permit the client to specify desired inodes, at replay
316 - duplicate requests when we queue them for replay reintegration
317 - fix last_rcvd offset calculation
318 - sync after each recovered transaction, so we always make progress
319 - never, not always, ERESTART requests without transnos
320 - store the lov_desc in the MDS, so we don't depend on getlovinfo to
322 - skip replay if the MDS says that the client is already connected
323 - don't check for a recovery-enabled export to match lctl's UUID
324 - don't INC_USE_COUNT for phantom exports
325 - don't crash when cleaning up phantom exports (567)
326 - don't double-finish or set replay data for errored mdc_open requests
327 - abort requests when they time out, so we don't get old replies
328 - send/receive replies for AST messages again
329 - if the client says that it doesn't have the lock, cancel it on the
331 - if we timeout during I/O, don't try to cancel an in-use lock; instead
332 mark it as destroyed, it will all work out when decref is called
333 - fix module use counts (22, 581)
335 - ASTs now expect a reply (server cancels lock on error reply)
337 2002-12-02 Andreas Dilger <adilger@clusterfs.com>
340 - fix many simultaneous client startup (392)
341 - fix dentry->d_it clobbering
342 - credentials weren't being shipped for readdir/getattr operations
343 - remove invalid assertions triggered during some concurrent MD
345 - proper Lustre versions added (336, 389)
346 - fix memory leak for create error case (398)
347 - fix LOV locking bug that would get cli/srv out of sync
348 - fix echo client over LOV (409)
349 - fix dbench 2, extN refcount problem (170, 258, 356, 418)
350 - fix double-O_EXCL intent crash (424)
351 - avoid sending multiple lock CANCELs (352)
353 - MDS can do multi-client recovery (modulo bugs in new code)
355 - many updates, edits, cleanups
357 2002-11-18 Phil Schwan <phil@clusterfs.com>
360 - fix null d_it dereference (346)
361 - fix full OST/dbench hang (333)
362 - fix permission problem with file removal (286)
363 - fix removal of OSCs from LOV when they fail
364 - fix NULL deref during bulk timeout (214)
365 - fix problems related to multiple filesystems on one MDS (241)
366 - fixed serious subtle metadata locking bugs
367 - free locks on clients when inodes are removed due to memory
369 - fix inode pointer in lock data (285)
370 - partial support for multiple MDS on a single host (241)
371 - data locks weren't cancelled at clear_inode time (290, 311)
372 - intent locks could lead to unbounded lock growth (205)
373 - added a maximum lock count, an LRU list, and a flusher
374 - fix multiple rename (365)
375 - properly abstracted the echo client
376 - OSC locked 1 byte too many; fixed
377 - rewrote brw callback code:
378 - fixed recovery bugs related to LOVs (306)
379 - fixed too-many-pages-in-one-write crash (191)
380 - fixed (again) crash in sync_io_timeout (214)
381 - probably fixed callback-related race (385)
383 - Add capability to MDS protocol
384 - LDLM cancellations and callbacks on different portals
386 2002-10-28 Andreas Dilger <adilger@clusterfs.com>
389 - limit client IOV size to PTL_MD_MAX_IOV (611336, 191)
390 - defer open object destruction to close time (601981, 138)
391 - open/close OST file handle in obdo (OBD_MD_FLHANDLE) (601981, 138)
392 - move LDLM_ENQUEUE/CONVERT back to MDS portal (625069)
393 - abstract ll_lookup2, fix ll_revalidate2 to use abstraction (256)
394 - don't call obd_setattr in ll_file_release for destroyed objects
395 * protocol change to lustre_msg: move |version| and add |flags|
396 * protocol change to osc_punch: "start" in "o_size", "end" in "o_blocks"
397 * lock replay: for LDLM_FL_REPLAY trust client to do right thing
398 * added replay of create, unlink, link and rename operations during
399 MDS failover; recovery should be much more robust now
400 * remove failed OSCs from LOVs (only lov_create uses this so far)
401 * the lustre-HOWTO was brought (more) up to date (582544)
403 2002-10-23 Phil Schwan <phil@clusterfs.com>
406 - in-use dentries weren't being reused properly (617851)
407 - prevent multiple LDLM setup (599178)
408 - fix LOV size calculations for truncate (617853)
409 - fix client handling of MDS intent errors (POSIX)
410 - fix permission bug in lovstripe.c test (624321)
411 - fix MDS thread deadlock - move LDLM handler to DLM portal (625069)
412 - truncate past end of file could corrupt data
413 - proper cleanup after timeouts, crashes, etc (592524, 550815)
414 - a race in recovery could return ETIMEDOUT to apps (623947)
415 - building outside the source directory was fixed
416 * the lustre-HOWTO was brought (more) up to date (582544)
417 * major progress was made on recovery functionality
419 2002-10-10 Phil Schwan <phil@clusterfs.com>
422 - recovery deadlock fix
423 - rm -rf causes LBUG fix (617817)
424 - file open by multiple tasks fix (618962)
425 - directory permissions bugs (602707 and 620007)
426 - journal_stop fixed with locking (611313)
427 - O_APPEND failures resolved (618273, perhaps 614459)
428 - lconf PATH fix (619770)
429 - IA64 build fix (621450)
430 - RPC buffer sizes scale with amount of memory
432 2002-10-01 Phil Schwan <phil@clusterfs.com>
435 - locks would be cancelled without throwing away data pages,
436 resulting in inconsistent data (605627)
437 - inode attributes were not always being refreshed (605627, 612449)
438 - lconf now continues to cleanup after lctl reports an error
439 - MDS now enforces user permissions (602707)
440 - lprocfs cleanup fixed, but not yet enabled (614157)
441 - fixed infinite server hang, should a client not respond to an AST
442 - avoid going into recovery if user calls readlink() with a buffer
443 that's too small (613941)
444 - AST RPCs no longer require replies (614867) -- this may be changed
445 - don't crash server if client sends an IOV that's too big (611336)
446 - fixed lock conversion deadlock (611892)
447 - fixed the following of symlinks (614622)
448 * recovery: the server can remove locks from a client that dies, other
449 clients can make progress
450 * more extN patch fixes
451 * compile-time configurable ptlrpc buffer allocations
453 - collaborative read cache document
454 - Lustre Lite Performance CDR document-in-progress
456 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
459 - fix typo in patch-2.4.18
461 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
464 - clear ptlrpc request each time in handle_incoming_request()
465 - unlink of files now destroys the object on the OST
467 2002-09-19 Peter Braam <braam@clusterfs.com>
469 * add hard link support
470 * change obdfile creation method
471 * kernel patch changed
473 2002-09-19 Peter Braam <braam@clusterfs.com>
476 - stack overflow bug in extN fixed
478 2002-09-18 Andreas Dilger <adilger@clusterfs.com>
480 * documentation updates
481 - add man pages for config tools
482 - update tests/README to describe testing with new config tools
483 - finish metadata API descriptions
484 * bug fixes and cleanups
485 - statfs workaround for 16TB limit
486 - LOV stripe allocation improved, can stripe on subset of OSTs
487 - LOV file size/IO offset was wrong for files > 4GB in size
488 - object EA data was being dropped, caused files to be unreadable
489 - memory overflow with non-LOV OST caused memory corruption
490 - fixed regression tests to work with new config tools, obdfilter
491 - fixed bug when directory size became larger than 1 block
492 - fixed bug (for single client case) when PWD was deleted
493 - invalidate local directory pages when doing intent-based ops
494 - avoid LDLM oops when lock callback contained bad data
496 2002-09-09 Andreas Dilger <adilger@clusterfs.com>
498 * documentation updates
499 * bug fixes and cleanups
500 - configuration tools
503 - 64-bit compile warnings
504 - 64-bit internal statfs data
506 * test_brw on persistent OST devices
510 2002-09-04 Andreas Dilger <adilger@clusterfs.com>
512 * documentation updates
513 * bug fixes and cleanups
514 * configuration tools
516 2002-08-30 Peter J. Braam <braam@clusterfs.com>
519 * many small fixes to 0_5_4
520 * io/network handling
521 * thinkos in MDS operations
523 2002-08-24 Peter J. Braam <braam@clusterfs.com>
526 * crucial basic fixes to 0.5.3
527 * IOR, Iozone work over Elan
530 2002-08-07 Phil Schwan <phil@clusterfs.com>
531 * version 0_5_3, our first alpha
532 * we use the new Portals iovs
533 * documentation updates
534 * bug fixes and cleanups
535 * small changes in the DLM wire protocol
537 2002-07-25 Peter J. Braam <braam@clusterfs.com>
538 * version 0_5_1 with some initial stability,
539 * locking on MD and file I/O.
540 * documentation updates
541 * several bug fixes since 0.5.0
542 * small changes in wire protocol
544 2002-07-18 Phil Schwan <phil@clusterfs.com>
546 * delivered as Lustre Light Alpha
547 * fixed a crash after handling invalid MDS requests
548 * fixed directory pages for architectures with non-4k pages sizes
550 2002-07-11 Andreas Dilger <adilger@clusterfs.com>
551 * release version v0_4_4
552 * Moves TCP acceptor to be on port 2432 (unused Coda port) instead
554 * Fixes a number of interruption problems with OST operations.
555 * Update documentation for portals header changes
556 * Move all wire protocol structs/defines to lustre_idl.h
557 * Fixes symlink length bug.
558 * Add tcpdump to repository.
560 2002-07-05 Andreas Dilger <adilger@clusterfs.com>
561 * release version v0_4_3
562 * Fixes statfs for inodes on extN.
563 * Fixes bug in runtests which would delete /etc/hosts.
564 * Use 64-bit object IDs wherever possible (not into VFS though)
565 Remove ost_get_info, which is unused by lustre, and out of date.
567 2002-07-03 Peter Braam <braam@clusterfs.com>
568 * release version v0_4_2 Fixes a lookup error (type not passed)
569 * move forward to head of Portals
570 * move forward to latest Lustre kernel
572 2002-06-25 Peter Braam <braam@clusterfs.com>
573 * release version v0_4_1. Hopefully stable on single node use.