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
7 2004-01-27 Cluster File Systems, Inc. <info@clusterfs.com>
10 - add series for the vanilla 2.6.0 kernel
11 - add series for the vanilla 2.4.24 kernel
12 - add series for a cray x86/64 UL kernel drop
13 - fix xattr patches for the vanilla 2.4.19 series
15 - generate true UUIDs in lmc (1171)
16 - have portals stack dumping break in UML (2466)
17 - avoid bad dchild deref; avoid inum lock w/o creation (2362)
18 - allocate with _NOFS in ldlm to avoid deadlock (1933)
19 - wake callback waiting threads on client eviction (2460)
20 - Add --ptldebug and --subsystem to lmc (1719)
21 - update assertion to allow safe interrupt allocation
22 - set rq_no_resend for cancel requests (2432)
23 - recalculate ptlrpcd timeout after resend (2494)
24 - call vfs_rmdir when removing pending directories (2368)
25 - fix renaming a file to itself (2429)
26 - lmc creates a default one-stripe lov (2454)
27 - expand procfs space to handle large clusters (2326)
28 - increase UML stack to avoid overflow
29 - update lconf's list of debug and subsystem masks
30 - fix lfs find --obd (2510)
31 - /proc tunable for disabling filter read caching (2591)
32 - stop rpm packages from altering slapd.conf (2301)
33 - disable nagle in the socknal under 0conf (2578)
34 - choose mds inode size based on stripe count (2572)
35 - fix kernel-source rpm problems (2516)
37 - add --disable-doc to avoid pdf generation (2421)
38 - update documentation, tests, type-os, comments
39 - avoid format warnings on ia64
41 - tiny code cleanups by removing unused fields
43 2004-01-07 Cluster File Systems, Inc. <info@clusterfs.com>
46 - fix obvious semaphore misuse in as-yet-unused setattr path (2348)
47 - remove the most blatant lies from BUILDING file (2371)
48 - change default debug level to reasonable production setting
49 - reduce client side cache size to reduce cache flush time
50 - reduce max RPCs in flight to avoid unnecessary file fragmentation
51 - make TCP zerocopy and pinger support enabled by default (2476)
52 - sync writes completed after process exits caused crashes (2319)
53 - maintain correct mount count on the MDS (2356)
54 - backout 1557, because 2316 wasn't really fixed
55 - better file I/O statistics gathering in /proc
56 - don't take unnecessary, deadlock-inducing bug in readpage (2383)
57 - another kernel patch to fix zero-copy TCP function export
58 - don't take duplicate lock when processing re-sent getattr (2420)
59 - lctl uses obd_self_export instead of creating new conn (2353)
60 - MDS/OST recovery case which requires object creation asserted (2425)
61 - move lfs from /usr/sbin to /usr/bin in packages
62 - fix race between mds_client_add and mds_client_free (2417)
63 - use kmalloc instead of slabs in portals (2430)
64 - don't create duplicate records when a failover MDS is present (2442)
65 - remove unnecessary mount age check (2332)
66 - don't remove directory inodes from locks prematurely (2451)
67 - don't break if MDS service name is the same as hostname (2103)
68 - fix races in client write RPC generation when cache full (2482)
70 2003-12-13 Cluster File Systems, Inc. <info@clusterfs.com>
73 - remove now-unused request->rq_obd (278)
74 - if an allocation fails, print out how much memory we've used (1933)
75 - use PORTAL_SLAB_ALLOC for structures, to get GFP_MEMALLOC (1933)
76 - add the "configurable stack size" patch to most series files (1256)
77 - ability to write large log records, for 100+ OST configs (2306)
78 - fix NULL deref when filter_prep fails (2314)
79 - fix operator precedence error in filter_sync
80 - dynamic allocation of socknal TX descriptors (2315)
81 - fix a missed case in the GFP_MEMALLOC patch, can cause deadlock (2310)
82 - fix gcc 2.96 compilation problem in xattr kernel patch (2294)
83 - ensure that CWARN messages in Portals always get to the syslog
84 - __init/__exit are not for prototype decls (ldlm_init/exit)
85 - x86-64 compile warning fixes
86 - fix gateway LMC keyword conflict (2318)
87 - fix MDS lock inversions in getattr/reint paths (1844)
88 - fix a rare lock re-ordering bug, which caused deadlock (2322)
89 - fix i_sem/journal inversion in fsfilt_ext3_write_record (2306)
90 - DLM race condition prevented some lock evictions (2328)
91 - ENOMEM detection and retry on socknal sends (2230)
92 - use GFP_NOFS throughout Lustre, to combat ENOMEM (2230)
93 - move osc_rpcd into ptlrpc, for use in MDC and others (2329)
94 - protect MDS inode fsdata with stronger locking; fixes assertion (2313)
95 - better error messages when a client is rejected during recovery (1505)
96 - avoid cancelling locks which were never granted, after failure (2330)
97 - fix i_sem/journal inversion in mds_client_add (2333)
98 - fix truncate/getattr lock cycle deadlock (2334)
99 - use rpcd to send close; allows resend after timeout, avoid leak (1897)
100 - fix two rare exit paths which could leak an l_lock() ref (2321)
101 - fencepost error in MDS/OST orphan recovery (2226)
102 - make log record alignment 8 bytes (1988)
103 - lstripe now fails when requested offset > ost_count (2237)
104 - ensure that all kernel series have a complete list.h (1607)
105 - fix crashes in special-file operations (2316)
106 - lctl create/brw OID mismatch, caused by obsolete filter loop (2339)
108 - allow configurable automake binary, for testing new versions
109 - small update to the lfs documentation
111 2003-12-03 Cluster File Systems, Inc. <info@clusterfs.com>
113 * fix negative export reference count in fsfilt_sync (2312)
115 2003-12-01 Cluster File Systems, Inc. <info@clusterfs.com>
116 * release candidate 0.9.1
118 - orphans are moved into the PENDING directory for possible recovery
119 - replayed opens now open by fid for orphan/rename safety (1042)
120 - last close of an orphan inode generates a transno (683)
121 - chdir() and mount() now pin the directory entry (1020)
122 - avoid CERROR in normal ll_setattr_raw() error case (1500)
123 - discard very old requests without processing them (1502)
124 - remove some common, well-understood CERRORs (1505)
125 - require O_DIRECT I/O to be page-sized to workaround IA64 crash (1609)
126 - clear "grant" flags in OST replies until OST grant code lands (1644)
127 - fix read performance by not clobbering i_blksize on client (1598)
128 - fix __ldlm_handle2lock oops by not dereferencing lock after PUT (1625)
129 - make LRU size a /proc tunable, clears locks when reduced (707)
130 - fix some lprocfs rot that prevented ptlbd from loading (1732)
131 - server locks take references on exports now (1558)
132 - build fixes for 2.4.20-rh trees (1663)
133 - return an error from lov_create if all OSCs are inactive (1751)
134 - fix import levels when a reconnect happens without a timeout (1597)
135 - exit early from mds_open if we get a lookup error (1749)
136 - partial page read at EOF wouldn't wait for disk before sending (1642)
137 - avoid NULL deref in obdfilter when reading page past EOF (1592)
138 - avoid LASSERT in ll_intent_lock if server failed very early (1090)
139 - fix LBUG in ll_it_open_error with rc = -2 (1861)
140 - write/truncate lock inversion (1639)
141 - Don't auto-load obdclass, portals modules during cleanup (1495)
142 - fix timestamps from jumping to "now" (1763)
143 - extra journal assertions (1648)
144 - add an extra multiunlink test (1771)
145 - fix read_record/write_record API (1776)
146 - fix leak of offset_extent, possible incorrect i_size later (1772)
147 - fix lasserts in mis-matched transnos during open-unlink testing (1541)
148 - Debugging for the kqswnal_get_idle_tx problems (1820)
149 - Allow recovery to be attempted multiple times (1536)
150 - Write out MDS last_rcvd file after it is first created (1600)
151 - Fix tx_descriptor leak in failed transmit situations (1827)
152 - ext3 journaling fixes for assertion failure after IO error (1871)
153 - class_export_put() on freed export after completion AST error (1896)
154 - Fix revalidate looping in VFS (1322)
155 - Don't access a freed export during MDS_REINT timeout (1521)
156 - Add open-unlink recovery support on the MDS (1673,1764)
157 - Return an error if no MDS data was read from last_rcvd (1946)
158 - Fix for lookup "." or ".." crash on error (1932,1931,1935)
159 - Don't setup a disk device that doesn't match exported UUID (317)
160 - Reduce bulk RPC timeout to avoid cascading client/OST failures (1845)
161 - avoid committing NULL handle in force close
162 - local.sh is now a one-stripe LOV configuration
163 - POSIX utime.4 -EPERM on FIFO not owned by user (56)
164 - fix ext3 htree duplicate directory entry corruption (1516)
165 - POSIX creat.13, fstat.1, open.18, stat.3 new file atime/mtime (2020)
166 - update to new LOV EA format (2097)
167 - interoperability for different PAGE_SIZE/wordsize (686,1821,1343,2042)
169 2003-06-15 Phil Schwan <phil@clusterfs.com>
172 - imports and exports cleanup too early, need refcounts (349, 879, 1045)
173 - per-import/export recovery handling (958, 931, 959)
174 - multiple last-rcvd slots, for serving multiple FSes (949)
175 - connections are again shared between multiple imp/exports (963, 964)
176 - "umount -f" would hang if any requests needed to be sent (393, 978)
177 - avoid pinning large req buffer by copying for queued messages (989)
178 - add "uuid" to "lctl device" command to help upcalls (991)
179 - "open" RPCs with transnos would confuse recovery counters (1037)
180 - do proper endian conversion of all wire messages (288, 340, 891)
181 - remove OST bulk get LBUGs, fix ost_brw_write cleanup (1126)
182 - call waiting locks callback from LDLM recovery thread (1127, 1151)
183 - fix ptlrpc_connection leak in target_handle_connect (1174)
184 - fix import refcounting bug in OST and MDS cleanup (1134)
185 - if an invalid-at-open-time OSC returned before close(), LBUG (1150)
186 - fix very unlikely obd_types race condition (501)
187 - remove osc_open hack for echo_client (1187)
188 - we leaked exports/dlmimps for forcibly disconnected clients (1143)
189 - a failure in read_inode2 leads to deadlock (1139)
190 - cancel ack-locks as soon as transaction is committed (1072)
191 - fix major leaks and crashes in the bulk I/O path (937, 1057)
192 - make sure to commitrw after any preprw to avoid deadlock (1162)
193 - failing to execute a file in a lustre FS would lock inode (1203)
194 - small DEBUG_REQ fix to avoid dereferencing a NULL (1227)
195 - don't ASSERT while cleaning up an incompletely-setup obd (1248)
196 - obd_uuid2tgt would walk off the end of the list (1255)
197 - on IA64 the osc would give portals incorrect bulk size (1258)
198 - fix debug daemon ioctl interface; allows daemon on ia64 (1274)
199 - fix lock inversion caused by new llite matching code (1282)
200 - limit the number of dirty pages on a client to 10MB (1286)
201 - timed out locks were not being corrected cancelled (1289)
202 - fix O_DIRECT above 4GB on IA-32 (1292)
203 * major user-visible changes
204 - fail out/fail over policy now controlled by the upcall (993)
206 - add OBD_PING to check server availability and failure (954)
207 - lustre messages are now sent in sending host order (288, 340, 891)
208 - add eadatalen to MDS getattr reply (340)
209 - OST read replies may contain second buffer, with per-page status (593)
211 2003-03-11 Phil Schwan <phil@clusterfs.com>
214 - LDLM_DEBUG macro fix, for gcc 3.2 (850)
215 - failed open()s could cause deadlock; fixed (867, 869)
216 - stop cancelling OST locks when files are closed (481)
217 - overlapping XID spaces caused network corruption (851, 853)
218 - fix unsafe fsfilt counter arithmetic; change to atomic_t
219 - setattr_raw added, to do single-RPC, server-side setattrs
220 - lmc/lconf syntax change for OST UUIDs
221 - fix crashy race condition between ptlrpc_free_req and osc_close
222 - don't use request in mdc_enqueue if we hit a timeout (889)
223 - don't set the inode i_size for regular files from the MDS (896)
224 - handle out of order completion AST (842)
225 - don't LBUG if a lock request times out after receiving AST (913)
226 - avoid d_rehash race in ll_find_alias by rehashing inside dcache_lock
227 - if a bad lock AST arrives, send an error instead of dropping entirely
228 - return 0 from revalidate2 if ll_intent_lock returns -EINTR (912)
229 - fix leak in bulk IO when only partially completed (899, 900, 926)
230 - fix O_DIRECT for ia64 (55)
231 - (almost) eliminate Lustre-kernel-thread effects on load average (722)
232 - C-z after timeout could hang a process forever; fixed (977)
234 - client-side I/O cache (678, 924, 929, 941, 970)
236 - READPAGE and SETATTRs which don't take server-side locks get
239 2003-02-11 Phil Schwan <phil@clusterfs.com>
242 - Fix ldlm_lock_match on the MDS to avoid matching remote locks (592)
243 - Fix fsfilt_extN_readpage() to read a full page of directory
244 entries, or fake the remainder if PAGE_SIZE != blocksize (500)
245 - Avoid extra mdc_getattr() in ll_intent_lock when possible (534, 604)
246 - Fix imbalanced LOV object allocation and out-of-bound access (469)
247 - Most intent operations were removed, in favour of a new RPC mode
248 that does a single RPC to the server and bypasses most of the VFS
249 - All LDLM resource ID arrays were removed in favour of ldlm_res_id
250 - Aggressively cancel local locks on DLM servers
251 - mds_reint_unlink sends EA to the client if it's the last nlink.
252 client uses that EA to unlink OST objects.
253 - mds_reint_{rename,unlink,link} were rewritten to take ordered locks
254 - recursive symlinks were fixed (439)
255 - fixed NULL deref in DEBUG_REQ
256 - filter_update_lastobjid no longer calls sync, which annoyed extN
257 - fixed multi-client small-writes to a single file problem (445)
258 - fixed mtime updates during file writes (607)
259 - fixed vector writes on obdfilter causing problems when ENOSPC (670)
260 - fixed bug in obd_brw_read/write() (under guise of testing 367)
261 - fixed Linux OST size reporting problem (444, 656)
262 - OST now updates object mtime with writes or setattr (607, 619)
263 - client verifies file size before zeroing page past EOF (445)
264 - OST now writes last allocated objid to disk with allocation (108)
265 - LOV on echo now works (409)
267 - mds_reint_unlink sends a new buffer, with the EA included. this
268 buffer is only valid if body->valid & OBD_MD_FLEASIZE, which is only
269 set if a regular file was being unlinked, and it was the last link
270 - use PtlGet from the target for bulk writes (315)
271 - OST now updates object mtime with writes or setattr (607, 619)
272 - LDLM now has a grant-time callback to revalidate locked items, if
274 - Many MDS operations were reorganized to combat race conditions
276 - Merge b_intel branch (updated lprocfs code) - now at /proc/fs/lustre
277 - configure check to avoid gcc version 2.96 20000731-2.96-98) (606)
279 2003-01-06 Andreas Dilger <adilger@clusterfs.com>
282 - Fully reactivate OST imports after reconnection (512, others)
283 - Make sure client sees our -ENOTCONN from mds_handle (513 - partial)
284 - More graceful error handling for truncating on dead OST (515)
285 - Don't error out unless we're actually accessing dead stripes (474)
286 - Fix garbage sizes when stripes are missing (410)
287 - LRU counters were broken, causing constant lock purge (433, 432)
288 - garbage on read from stripes with failed OSTs (441)
289 - mark OSCs as active before reconnecting during recovery (438)
290 - lov_enqueue and lov_cancel need to handle inactive OSTs (403)
291 - lfind did not preserve OST order in output (443)
292 - symlinks cause hung clients, incorrect data (439)
293 - stop dereferencing request after dropping refcount (457)
294 - don't LASSERT(spin_is_locked) on non-SMP (455)
295 - fixes for many rename() bugs
296 - fstat didn't correctly synchronize attributes (399)
297 - server must handle lock cancellation during blocking AST prep (487)
298 - bulk descriptors were free()d too soon (511)
299 - fix paths in lconf, which would load incorrect modules (451, 507)
300 - fix confusing lconf 'host not found' error message (386)
301 - fix lock order deadlock on OST (O/R i_sem before journal ops, 478)
302 - fix race condition in mdc_blocking_ast() for inode access (526)
303 - fix lov_unpackmd() unpacking wrong number of stripes (537)
304 - fix lov_set_osc_active() marking wrong OSC inactive (440)
305 - fix bad lstripe lov_unpackmd() assertion (fix layering too) (527)
306 - fix multiple writes of stripe MD to MDS (358, maybe 519)
307 - fix lstripe in several ways (kernel side) (527)
308 - fix request leak in ldlm_cli_enqueue (262)
309 - incorrect OSC was marked inactive after OST failure
310 - call mds_fs_cleanup before unmounting filesystem (524)
311 - fix races between taking ns_lock and ldlm_lock_change_resource
312 - fix races updating LOV export open file list
313 - fix lov_enqueue error path, avoid decref-ing bad lock handle
314 - fix recovery NULL deref in ldlm_cli_cancel_unused
315 - fix some DLM races by using new hash table for lock handles (419)
316 - permit the client to specify desired inodes, at replay
317 - duplicate requests when we queue them for replay reintegration
318 - fix last_rcvd offset calculation
319 - sync after each recovered transaction, so we always make progress
320 - never, not always, ERESTART requests without transnos
321 - store the lov_desc in the MDS, so we don't depend on getlovinfo to
323 - skip replay if the MDS says that the client is already connected
324 - don't check for a recovery-enabled export to match lctl's UUID
325 - don't INC_USE_COUNT for phantom exports
326 - don't crash when cleaning up phantom exports (567)
327 - don't double-finish or set replay data for errored mdc_open requests
328 - abort requests when they time out, so we don't get old replies
329 - send/receive replies for AST messages again
330 - if the client says that it doesn't have the lock, cancel it on the
332 - if we timeout during I/O, don't try to cancel an in-use lock; instead
333 mark it as destroyed, it will all work out when decref is called
334 - fix module use counts (22, 581)
336 - ASTs now expect a reply (server cancels lock on error reply)
338 2002-12-02 Andreas Dilger <adilger@clusterfs.com>
341 - fix many simultaneous client startup (392)
342 - fix dentry->d_it clobbering
343 - credentials weren't being shipped for readdir/getattr operations
344 - remove invalid assertions triggered during some concurrent MD
346 - proper Lustre versions added (336, 389)
347 - fix memory leak for create error case (398)
348 - fix LOV locking bug that would get cli/srv out of sync
349 - fix echo client over LOV (409)
350 - fix dbench 2, extN refcount problem (170, 258, 356, 418)
351 - fix double-O_EXCL intent crash (424)
352 - avoid sending multiple lock CANCELs (352)
354 - MDS can do multi-client recovery (modulo bugs in new code)
356 - many updates, edits, cleanups
358 2002-11-18 Phil Schwan <phil@clusterfs.com>
361 - fix null d_it dereference (346)
362 - fix full OST/dbench hang (333)
363 - fix permission problem with file removal (286)
364 - fix removal of OSCs from LOV when they fail
365 - fix NULL deref during bulk timeout (214)
366 - fix problems related to multiple filesystems on one MDS (241)
367 - fixed serious subtle metadata locking bugs
368 - free locks on clients when inodes are removed due to memory
370 - fix inode pointer in lock data (285)
371 - partial support for multiple MDS on a single host (241)
372 - data locks weren't cancelled at clear_inode time (290, 311)
373 - intent locks could lead to unbounded lock growth (205)
374 - added a maximum lock count, an LRU list, and a flusher
375 - fix multiple rename (365)
376 - properly abstracted the echo client
377 - OSC locked 1 byte too many; fixed
378 - rewrote brw callback code:
379 - fixed recovery bugs related to LOVs (306)
380 - fixed too-many-pages-in-one-write crash (191)
381 - fixed (again) crash in sync_io_timeout (214)
382 - probably fixed callback-related race (385)
384 - Add capability to MDS protocol
385 - LDLM cancellations and callbacks on different portals
387 2002-10-28 Andreas Dilger <adilger@clusterfs.com>
390 - limit client IOV size to PTL_MD_MAX_IOV (611336, 191)
391 - defer open object destruction to close time (601981, 138)
392 - open/close OST file handle in obdo (OBD_MD_FLHANDLE) (601981, 138)
393 - move LDLM_ENQUEUE/CONVERT back to MDS portal (625069)
394 - abstract ll_lookup2, fix ll_revalidate2 to use abstraction (256)
395 - don't call obd_setattr in ll_file_release for destroyed objects
396 * protocol change to lustre_msg: move |version| and add |flags|
397 * protocol change to osc_punch: "start" in "o_size", "end" in "o_blocks"
398 * lock replay: for LDLM_FL_REPLAY trust client to do right thing
399 * added replay of create, unlink, link and rename operations during
400 MDS failover; recovery should be much more robust now
401 * remove failed OSCs from LOVs (only lov_create uses this so far)
402 * the lustre-HOWTO was brought (more) up to date (582544)
404 2002-10-23 Phil Schwan <phil@clusterfs.com>
407 - in-use dentries weren't being reused properly (617851)
408 - prevent multiple LDLM setup (599178)
409 - fix LOV size calculations for truncate (617853)
410 - fix client handling of MDS intent errors (POSIX)
411 - fix permission bug in lovstripe.c test (624321)
412 - fix MDS thread deadlock - move LDLM handler to DLM portal (625069)
413 - truncate past end of file could corrupt data
414 - proper cleanup after timeouts, crashes, etc (592524, 550815)
415 - a race in recovery could return ETIMEDOUT to apps (623947)
416 - building outside the source directory was fixed
417 * the lustre-HOWTO was brought (more) up to date (582544)
418 * major progress was made on recovery functionality
420 2002-10-10 Phil Schwan <phil@clusterfs.com>
423 - recovery deadlock fix
424 - rm -rf causes LBUG fix (617817)
425 - file open by multiple tasks fix (618962)
426 - directory permissions bugs (602707 and 620007)
427 - journal_stop fixed with locking (611313)
428 - O_APPEND failures resolved (618273, perhaps 614459)
429 - lconf PATH fix (619770)
430 - IA64 build fix (621450)
431 - RPC buffer sizes scale with amount of memory
433 2002-10-01 Phil Schwan <phil@clusterfs.com>
436 - locks would be cancelled without throwing away data pages,
437 resulting in inconsistent data (605627)
438 - inode attributes were not always being refreshed (605627, 612449)
439 - lconf now continues to cleanup after lctl reports an error
440 - MDS now enforces user permissions (602707)
441 - lprocfs cleanup fixed, but not yet enabled (614157)
442 - fixed infinite server hang, should a client not respond to an AST
443 - avoid going into recovery if user calls readlink() with a buffer
444 that's too small (613941)
445 - AST RPCs no longer require replies (614867) -- this may be changed
446 - don't crash server if client sends an IOV that's too big (611336)
447 - fixed lock conversion deadlock (611892)
448 - fixed the following of symlinks (614622)
449 * recovery: the server can remove locks from a client that dies, other
450 clients can make progress
451 * more extN patch fixes
452 * compile-time configurable ptlrpc buffer allocations
454 - collaborative read cache document
455 - Lustre Lite Performance CDR document-in-progress
457 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
460 - fix typo in patch-2.4.18
462 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
465 - clear ptlrpc request each time in handle_incoming_request()
466 - unlink of files now destroys the object on the OST
468 2002-09-19 Peter Braam <braam@clusterfs.com>
470 * add hard link support
471 * change obdfile creation method
472 * kernel patch changed
474 2002-09-19 Peter Braam <braam@clusterfs.com>
477 - stack overflow bug in extN fixed
479 2002-09-18 Andreas Dilger <adilger@clusterfs.com>
481 * documentation updates
482 - add man pages for config tools
483 - update tests/README to describe testing with new config tools
484 - finish metadata API descriptions
485 * bug fixes and cleanups
486 - statfs workaround for 16TB limit
487 - LOV stripe allocation improved, can stripe on subset of OSTs
488 - LOV file size/IO offset was wrong for files > 4GB in size
489 - object EA data was being dropped, caused files to be unreadable
490 - memory overflow with non-LOV OST caused memory corruption
491 - fixed regression tests to work with new config tools, obdfilter
492 - fixed bug when directory size became larger than 1 block
493 - fixed bug (for single client case) when PWD was deleted
494 - invalidate local directory pages when doing intent-based ops
495 - avoid LDLM oops when lock callback contained bad data
497 2002-09-09 Andreas Dilger <adilger@clusterfs.com>
499 * documentation updates
500 * bug fixes and cleanups
501 - configuration tools
504 - 64-bit compile warnings
505 - 64-bit internal statfs data
507 * test_brw on persistent OST devices
511 2002-09-04 Andreas Dilger <adilger@clusterfs.com>
513 * documentation updates
514 * bug fixes and cleanups
515 * configuration tools
517 2002-08-30 Peter J. Braam <braam@clusterfs.com>
520 * many small fixes to 0_5_4
521 * io/network handling
522 * thinkos in MDS operations
524 2002-08-24 Peter J. Braam <braam@clusterfs.com>
527 * crucial basic fixes to 0.5.3
528 * IOR, Iozone work over Elan
531 2002-08-07 Phil Schwan <phil@clusterfs.com>
532 * version 0_5_3, our first alpha
533 * we use the new Portals iovs
534 * documentation updates
535 * bug fixes and cleanups
536 * small changes in the DLM wire protocol
538 2002-07-25 Peter J. Braam <braam@clusterfs.com>
539 * version 0_5_1 with some initial stability,
540 * locking on MD and file I/O.
541 * documentation updates
542 * several bug fixes since 0.5.0
543 * small changes in wire protocol
545 2002-07-18 Phil Schwan <phil@clusterfs.com>
547 * delivered as Lustre Light Alpha
548 * fixed a crash after handling invalid MDS requests
549 * fixed directory pages for architectures with non-4k pages sizes
551 2002-07-11 Andreas Dilger <adilger@clusterfs.com>
552 * release version v0_4_4
553 * Moves TCP acceptor to be on port 2432 (unused Coda port) instead
555 * Fixes a number of interruption problems with OST operations.
556 * Update documentation for portals header changes
557 * Move all wire protocol structs/defines to lustre_idl.h
558 * Fixes symlink length bug.
559 * Add tcpdump to repository.
561 2002-07-05 Andreas Dilger <adilger@clusterfs.com>
562 * release version v0_4_3
563 * Fixes statfs for inodes on extN.
564 * Fixes bug in runtests which would delete /etc/hosts.
565 * Use 64-bit object IDs wherever possible (not into VFS though)
566 Remove ost_get_info, which is unused by lustre, and out of date.
568 2002-07-03 Peter Braam <braam@clusterfs.com>
569 * release version v0_4_2 Fixes a lookup error (type not passed)
570 * move forward to head of Portals
571 * move forward to latest Lustre kernel
573 2002-06-25 Peter Braam <braam@clusterfs.com>
574 * release version v0_4_1. Hopefully stable on single node use.