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 - fixes for mds_cleanup_orphans (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)
31 - fix mkfsoptions support for osts (2603, 2604)
33 2004-01-27 Cluster File Systems, Inc. <info@clusterfs.com>
36 - add series for the vanilla 2.6.0 kernel
37 - add series for the vanilla 2.4.24 kernel
38 - add series for a cray x86/64 UL kernel drop
39 - fix xattr patches for the vanilla 2.4.19 series
41 - generate true UUIDs in lmc (1171)
42 - have portals stack dumping break in UML (2466)
43 - avoid bad dchild deref; avoid inum lock w/o creation (2362)
44 - allocate with _NOFS in ldlm to avoid deadlock (1933)
45 - wake callback waiting threads on client eviction (2460)
46 - Add --ptldebug and --subsystem to lmc (1719)
47 - update assertion to allow safe interrupt allocation
48 - set rq_no_resend for cancel requests (2432)
49 - recalculate ptlrpcd timeout after resend (2494)
50 - call vfs_rmdir when removing pending directories (2368)
51 - fix renaming a file to itself (2429)
52 - lmc creates a default one-stripe lov (2454)
53 - expand procfs space to handle large clusters (2326)
54 - increase UML stack to avoid overflow
55 - update lconf's list of debug and subsystem masks
56 - fix lfs find --obd (2510)
57 - /proc tunable for disabling filter read caching (2591)
58 - stop rpm packages from altering slapd.conf (2301)
59 - disable nagle in the socknal under 0conf (2578)
60 - choose mds inode size based on stripe count (2572)
61 - fix kernel-source rpm problems (2516)
63 - add --disable-doc to avoid pdf generation (2421)
64 - update documentation, tests, type-os, comments
65 - avoid format warnings on ia64
67 - tiny code cleanups by removing unused fields
69 2004-01-07 Cluster File Systems, Inc. <info@clusterfs.com>
72 - fix obvious semaphore misuse in as-yet-unused setattr path (2348)
73 - remove the most blatant lies from BUILDING file (2371)
74 - change default debug level to reasonable production setting
75 - reduce client side cache size to reduce cache flush time
76 - reduce max RPCs in flight to avoid unnecessary file fragmentation
77 - make TCP zerocopy and pinger support enabled by default (2476)
78 - sync writes completed after process exits caused crashes (2319)
79 - maintain correct mount count on the MDS (2356)
80 - backout 1557, because 2316 wasn't really fixed
81 - better file I/O statistics gathering in /proc
82 - don't take unnecessary, deadlock-inducing bug in readpage (2383)
83 - another kernel patch to fix zero-copy TCP function export
84 - don't take duplicate lock when processing re-sent getattr (2420)
85 - lctl uses obd_self_export instead of creating new conn (2353)
86 - MDS/OST recovery case which requires object creation asserted (2425)
87 - move lfs from /usr/sbin to /usr/bin in packages
88 - fix race between mds_client_add and mds_client_free (2417)
89 - use kmalloc instead of slabs in portals (2430)
90 - don't create duplicate records when a failover MDS is present (2442)
91 - remove unnecessary mount age check (2332)
92 - don't remove directory inodes from locks prematurely (2451)
93 - don't break if MDS service name is the same as hostname (2103)
94 - fix races in client write RPC generation when cache full (2482)
96 2003-12-13 Cluster File Systems, Inc. <info@clusterfs.com>
99 - remove now-unused request->rq_obd (278)
100 - if an allocation fails, print out how much memory we've used (1933)
101 - use PORTAL_SLAB_ALLOC for structures, to get GFP_MEMALLOC (1933)
102 - add the "configurable stack size" patch to most series files (1256)
103 - ability to write large log records, for 100+ OST configs (2306)
104 - fix NULL deref when filter_prep fails (2314)
105 - fix operator precedence error in filter_sync
106 - dynamic allocation of socknal TX descriptors (2315)
107 - fix a missed case in the GFP_MEMALLOC patch, can cause deadlock (2310)
108 - fix gcc 2.96 compilation problem in xattr kernel patch (2294)
109 - ensure that CWARN messages in Portals always get to the syslog
110 - __init/__exit are not for prototype decls (ldlm_init/exit)
111 - x86-64 compile warning fixes
112 - fix gateway LMC keyword conflict (2318)
113 - fix MDS lock inversions in getattr/reint paths (1844)
114 - fix a rare lock re-ordering bug, which caused deadlock (2322)
115 - fix i_sem/journal inversion in fsfilt_ext3_write_record (2306)
116 - DLM race condition prevented some lock evictions (2328)
117 - ENOMEM detection and retry on socknal sends (2230)
118 - use GFP_NOFS throughout Lustre, to combat ENOMEM (2230)
119 - move osc_rpcd into ptlrpc, for use in MDC and others (2329)
120 - protect MDS inode fsdata with stronger locking; fixes assertion (2313)
121 - better error messages when a client is rejected during recovery (1505)
122 - avoid cancelling locks which were never granted, after failure (2330)
123 - fix i_sem/journal inversion in mds_client_add (2333)
124 - fix truncate/getattr lock cycle deadlock (2334)
125 - use rpcd to send close; allows resend after timeout, avoid leak (1897)
126 - fix two rare exit paths which could leak an l_lock() ref (2321)
127 - fencepost error in MDS/OST orphan recovery (2226)
128 - make log record alignment 8 bytes (1988)
129 - lstripe now fails when requested offset > ost_count (2237)
130 - ensure that all kernel series have a complete list.h (1607)
131 - fix crashes in special-file operations (2316)
132 - lctl create/brw OID mismatch, caused by obsolete filter loop (2339)
134 - allow configurable automake binary, for testing new versions
135 - small update to the lfs documentation
137 2003-12-03 Cluster File Systems, Inc. <info@clusterfs.com>
139 * fix negative export reference count in fsfilt_sync (2312)
141 2003-12-01 Cluster File Systems, Inc. <info@clusterfs.com>
142 * release candidate 0.9.1
144 - orphans are moved into the PENDING directory for possible recovery
145 - replayed opens now open by fid for orphan/rename safety (1042)
146 - last close of an orphan inode generates a transno (683)
147 - chdir() and mount() now pin the directory entry (1020)
148 - avoid CERROR in normal ll_setattr_raw() error case (1500)
149 - discard very old requests without processing them (1502)
150 - remove some common, well-understood CERRORs (1505)
151 - require O_DIRECT I/O to be page-sized to workaround IA64 crash (1609)
152 - clear "grant" flags in OST replies until OST grant code lands (1644)
153 - fix read performance by not clobbering i_blksize on client (1598)
154 - fix __ldlm_handle2lock oops by not dereferencing lock after PUT (1625)
155 - make LRU size a /proc tunable, clears locks when reduced (707)
156 - fix some lprocfs rot that prevented ptlbd from loading (1732)
157 - server locks take references on exports now (1558)
158 - build fixes for 2.4.20-rh trees (1663)
159 - return an error from lov_create if all OSCs are inactive (1751)
160 - fix import levels when a reconnect happens without a timeout (1597)
161 - exit early from mds_open if we get a lookup error (1749)
162 - partial page read at EOF wouldn't wait for disk before sending (1642)
163 - avoid NULL deref in obdfilter when reading page past EOF (1592)
164 - avoid LASSERT in ll_intent_lock if server failed very early (1090)
165 - fix LBUG in ll_it_open_error with rc = -2 (1861)
166 - write/truncate lock inversion (1639)
167 - Don't auto-load obdclass, portals modules during cleanup (1495)
168 - fix timestamps from jumping to "now" (1763)
169 - extra journal assertions (1648)
170 - add an extra multiunlink test (1771)
171 - fix read_record/write_record API (1776)
172 - fix leak of offset_extent, possible incorrect i_size later (1772)
173 - fix lasserts in mis-matched transnos during open-unlink testing (1541)
174 - Debugging for the kqswnal_get_idle_tx problems (1820)
175 - Allow recovery to be attempted multiple times (1536)
176 - Write out MDS last_rcvd file after it is first created (1600)
177 - Fix tx_descriptor leak in failed transmit situations (1827)
178 - ext3 journaling fixes for assertion failure after IO error (1871)
179 - class_export_put() on freed export after completion AST error (1896)
180 - Fix revalidate looping in VFS (1322)
181 - Don't access a freed export during MDS_REINT timeout (1521)
182 - Add open-unlink recovery support on the MDS (1673,1764)
183 - Return an error if no MDS data was read from last_rcvd (1946)
184 - Fix for lookup "." or ".." crash on error (1932,1931,1935)
185 - Don't setup a disk device that doesn't match exported UUID (317)
186 - Reduce bulk RPC timeout to avoid cascading client/OST failures (1845)
187 - avoid committing NULL handle in force close
188 - local.sh is now a one-stripe LOV configuration
189 - POSIX utime.4 -EPERM on FIFO not owned by user (56)
190 - fix ext3 htree duplicate directory entry corruption (1516)
191 - POSIX creat.13, fstat.1, open.18, stat.3 new file atime/mtime (2020)
192 - update to new LOV EA format (2097)
193 - interoperability for different PAGE_SIZE/wordsize (686,1821,1343,2042)
195 2003-06-15 Phil Schwan <phil@clusterfs.com>
198 - imports and exports cleanup too early, need refcounts (349, 879, 1045)
199 - per-import/export recovery handling (958, 931, 959)
200 - multiple last-rcvd slots, for serving multiple FSes (949)
201 - connections are again shared between multiple imp/exports (963, 964)
202 - "umount -f" would hang if any requests needed to be sent (393, 978)
203 - avoid pinning large req buffer by copying for queued messages (989)
204 - add "uuid" to "lctl device" command to help upcalls (991)
205 - "open" RPCs with transnos would confuse recovery counters (1037)
206 - do proper endian conversion of all wire messages (288, 340, 891)
207 - remove OST bulk get LBUGs, fix ost_brw_write cleanup (1126)
208 - call waiting locks callback from LDLM recovery thread (1127, 1151)
209 - fix ptlrpc_connection leak in target_handle_connect (1174)
210 - fix import refcounting bug in OST and MDS cleanup (1134)
211 - if an invalid-at-open-time OSC returned before close(), LBUG (1150)
212 - fix very unlikely obd_types race condition (501)
213 - remove osc_open hack for echo_client (1187)
214 - we leaked exports/dlmimps for forcibly disconnected clients (1143)
215 - a failure in read_inode2 leads to deadlock (1139)
216 - cancel ack-locks as soon as transaction is committed (1072)
217 - fix major leaks and crashes in the bulk I/O path (937, 1057)
218 - make sure to commitrw after any preprw to avoid deadlock (1162)
219 - failing to execute a file in a lustre FS would lock inode (1203)
220 - small DEBUG_REQ fix to avoid dereferencing a NULL (1227)
221 - don't ASSERT while cleaning up an incompletely-setup obd (1248)
222 - obd_uuid2tgt would walk off the end of the list (1255)
223 - on IA64 the osc would give portals incorrect bulk size (1258)
224 - fix debug daemon ioctl interface; allows daemon on ia64 (1274)
225 - fix lock inversion caused by new llite matching code (1282)
226 - limit the number of dirty pages on a client to 10MB (1286)
227 - timed out locks were not being corrected cancelled (1289)
228 - fix O_DIRECT above 4GB on IA-32 (1292)
229 * major user-visible changes
230 - fail out/fail over policy now controlled by the upcall (993)
232 - add OBD_PING to check server availability and failure (954)
233 - lustre messages are now sent in sending host order (288, 340, 891)
234 - add eadatalen to MDS getattr reply (340)
235 - OST read replies may contain second buffer, with per-page status (593)
237 2003-03-11 Phil Schwan <phil@clusterfs.com>
240 - LDLM_DEBUG macro fix, for gcc 3.2 (850)
241 - failed open()s could cause deadlock; fixed (867, 869)
242 - stop cancelling OST locks when files are closed (481)
243 - overlapping XID spaces caused network corruption (851, 853)
244 - fix unsafe fsfilt counter arithmetic; change to atomic_t
245 - setattr_raw added, to do single-RPC, server-side setattrs
246 - lmc/lconf syntax change for OST UUIDs
247 - fix crashy race condition between ptlrpc_free_req and osc_close
248 - don't use request in mdc_enqueue if we hit a timeout (889)
249 - don't set the inode i_size for regular files from the MDS (896)
250 - handle out of order completion AST (842)
251 - don't LBUG if a lock request times out after receiving AST (913)
252 - avoid d_rehash race in ll_find_alias by rehashing inside dcache_lock
253 - if a bad lock AST arrives, send an error instead of dropping entirely
254 - return 0 from revalidate2 if ll_intent_lock returns -EINTR (912)
255 - fix leak in bulk IO when only partially completed (899, 900, 926)
256 - fix O_DIRECT for ia64 (55)
257 - (almost) eliminate Lustre-kernel-thread effects on load average (722)
258 - C-z after timeout could hang a process forever; fixed (977)
260 - client-side I/O cache (678, 924, 929, 941, 970)
262 - READPAGE and SETATTRs which don't take server-side locks get
265 2003-02-11 Phil Schwan <phil@clusterfs.com>
268 - Fix ldlm_lock_match on the MDS to avoid matching remote locks (592)
269 - Fix fsfilt_extN_readpage() to read a full page of directory
270 entries, or fake the remainder if PAGE_SIZE != blocksize (500)
271 - Avoid extra mdc_getattr() in ll_intent_lock when possible (534, 604)
272 - Fix imbalanced LOV object allocation and out-of-bound access (469)
273 - Most intent operations were removed, in favour of a new RPC mode
274 that does a single RPC to the server and bypasses most of the VFS
275 - All LDLM resource ID arrays were removed in favour of ldlm_res_id
276 - Aggressively cancel local locks on DLM servers
277 - mds_reint_unlink sends EA to the client if it's the last nlink.
278 client uses that EA to unlink OST objects.
279 - mds_reint_{rename,unlink,link} were rewritten to take ordered locks
280 - recursive symlinks were fixed (439)
281 - fixed NULL deref in DEBUG_REQ
282 - filter_update_lastobjid no longer calls sync, which annoyed extN
283 - fixed multi-client small-writes to a single file problem (445)
284 - fixed mtime updates during file writes (607)
285 - fixed vector writes on obdfilter causing problems when ENOSPC (670)
286 - fixed bug in obd_brw_read/write() (under guise of testing 367)
287 - fixed Linux OST size reporting problem (444, 656)
288 - OST now updates object mtime with writes or setattr (607, 619)
289 - client verifies file size before zeroing page past EOF (445)
290 - OST now writes last allocated objid to disk with allocation (108)
291 - LOV on echo now works (409)
293 - mds_reint_unlink sends a new buffer, with the EA included. this
294 buffer is only valid if body->valid & OBD_MD_FLEASIZE, which is only
295 set if a regular file was being unlinked, and it was the last link
296 - use PtlGet from the target for bulk writes (315)
297 - OST now updates object mtime with writes or setattr (607, 619)
298 - LDLM now has a grant-time callback to revalidate locked items, if
300 - Many MDS operations were reorganized to combat race conditions
302 - Merge b_intel branch (updated lprocfs code) - now at /proc/fs/lustre
303 - configure check to avoid gcc version 2.96 20000731-2.96-98) (606)
305 2003-01-06 Andreas Dilger <adilger@clusterfs.com>
308 - Fully reactivate OST imports after reconnection (512, others)
309 - Make sure client sees our -ENOTCONN from mds_handle (513 - partial)
310 - More graceful error handling for truncating on dead OST (515)
311 - Don't error out unless we're actually accessing dead stripes (474)
312 - Fix garbage sizes when stripes are missing (410)
313 - LRU counters were broken, causing constant lock purge (433, 432)
314 - garbage on read from stripes with failed OSTs (441)
315 - mark OSCs as active before reconnecting during recovery (438)
316 - lov_enqueue and lov_cancel need to handle inactive OSTs (403)
317 - lfind did not preserve OST order in output (443)
318 - symlinks cause hung clients, incorrect data (439)
319 - stop dereferencing request after dropping refcount (457)
320 - don't LASSERT(spin_is_locked) on non-SMP (455)
321 - fixes for many rename() bugs
322 - fstat didn't correctly synchronize attributes (399)
323 - server must handle lock cancellation during blocking AST prep (487)
324 - bulk descriptors were free()d too soon (511)
325 - fix paths in lconf, which would load incorrect modules (451, 507)
326 - fix confusing lconf 'host not found' error message (386)
327 - fix lock order deadlock on OST (O/R i_sem before journal ops, 478)
328 - fix race condition in mdc_blocking_ast() for inode access (526)
329 - fix lov_unpackmd() unpacking wrong number of stripes (537)
330 - fix lov_set_osc_active() marking wrong OSC inactive (440)
331 - fix bad lstripe lov_unpackmd() assertion (fix layering too) (527)
332 - fix multiple writes of stripe MD to MDS (358, maybe 519)
333 - fix lstripe in several ways (kernel side) (527)
334 - fix request leak in ldlm_cli_enqueue (262)
335 - incorrect OSC was marked inactive after OST failure
336 - call mds_fs_cleanup before unmounting filesystem (524)
337 - fix races between taking ns_lock and ldlm_lock_change_resource
338 - fix races updating LOV export open file list
339 - fix lov_enqueue error path, avoid decref-ing bad lock handle
340 - fix recovery NULL deref in ldlm_cli_cancel_unused
341 - fix some DLM races by using new hash table for lock handles (419)
342 - permit the client to specify desired inodes, at replay
343 - duplicate requests when we queue them for replay reintegration
344 - fix last_rcvd offset calculation
345 - sync after each recovered transaction, so we always make progress
346 - never, not always, ERESTART requests without transnos
347 - store the lov_desc in the MDS, so we don't depend on getlovinfo to
349 - skip replay if the MDS says that the client is already connected
350 - don't check for a recovery-enabled export to match lctl's UUID
351 - don't INC_USE_COUNT for phantom exports
352 - don't crash when cleaning up phantom exports (567)
353 - don't double-finish or set replay data for errored mdc_open requests
354 - abort requests when they time out, so we don't get old replies
355 - send/receive replies for AST messages again
356 - if the client says that it doesn't have the lock, cancel it on the
358 - if we timeout during I/O, don't try to cancel an in-use lock; instead
359 mark it as destroyed, it will all work out when decref is called
360 - fix module use counts (22, 581)
362 - ASTs now expect a reply (server cancels lock on error reply)
364 2002-12-02 Andreas Dilger <adilger@clusterfs.com>
367 - fix many simultaneous client startup (392)
368 - fix dentry->d_it clobbering
369 - credentials weren't being shipped for readdir/getattr operations
370 - remove invalid assertions triggered during some concurrent MD
372 - proper Lustre versions added (336, 389)
373 - fix memory leak for create error case (398)
374 - fix LOV locking bug that would get cli/srv out of sync
375 - fix echo client over LOV (409)
376 - fix dbench 2, extN refcount problem (170, 258, 356, 418)
377 - fix double-O_EXCL intent crash (424)
378 - avoid sending multiple lock CANCELs (352)
380 - MDS can do multi-client recovery (modulo bugs in new code)
382 - many updates, edits, cleanups
384 2002-11-18 Phil Schwan <phil@clusterfs.com>
387 - fix null d_it dereference (346)
388 - fix full OST/dbench hang (333)
389 - fix permission problem with file removal (286)
390 - fix removal of OSCs from LOV when they fail
391 - fix NULL deref during bulk timeout (214)
392 - fix problems related to multiple filesystems on one MDS (241)
393 - fixed serious subtle metadata locking bugs
394 - free locks on clients when inodes are removed due to memory
396 - fix inode pointer in lock data (285)
397 - partial support for multiple MDS on a single host (241)
398 - data locks weren't cancelled at clear_inode time (290, 311)
399 - intent locks could lead to unbounded lock growth (205)
400 - added a maximum lock count, an LRU list, and a flusher
401 - fix multiple rename (365)
402 - properly abstracted the echo client
403 - OSC locked 1 byte too many; fixed
404 - rewrote brw callback code:
405 - fixed recovery bugs related to LOVs (306)
406 - fixed too-many-pages-in-one-write crash (191)
407 - fixed (again) crash in sync_io_timeout (214)
408 - probably fixed callback-related race (385)
410 - Add capability to MDS protocol
411 - LDLM cancellations and callbacks on different portals
413 2002-10-28 Andreas Dilger <adilger@clusterfs.com>
416 - limit client IOV size to PTL_MD_MAX_IOV (611336, 191)
417 - defer open object destruction to close time (601981, 138)
418 - open/close OST file handle in obdo (OBD_MD_FLHANDLE) (601981, 138)
419 - move LDLM_ENQUEUE/CONVERT back to MDS portal (625069)
420 - abstract ll_lookup2, fix ll_revalidate2 to use abstraction (256)
421 - don't call obd_setattr in ll_file_release for destroyed objects
422 * protocol change to lustre_msg: move |version| and add |flags|
423 * protocol change to osc_punch: "start" in "o_size", "end" in "o_blocks"
424 * lock replay: for LDLM_FL_REPLAY trust client to do right thing
425 * added replay of create, unlink, link and rename operations during
426 MDS failover; recovery should be much more robust now
427 * remove failed OSCs from LOVs (only lov_create uses this so far)
428 * the lustre-HOWTO was brought (more) up to date (582544)
430 2002-10-23 Phil Schwan <phil@clusterfs.com>
433 - in-use dentries weren't being reused properly (617851)
434 - prevent multiple LDLM setup (599178)
435 - fix LOV size calculations for truncate (617853)
436 - fix client handling of MDS intent errors (POSIX)
437 - fix permission bug in lovstripe.c test (624321)
438 - fix MDS thread deadlock - move LDLM handler to DLM portal (625069)
439 - truncate past end of file could corrupt data
440 - proper cleanup after timeouts, crashes, etc (592524, 550815)
441 - a race in recovery could return ETIMEDOUT to apps (623947)
442 - building outside the source directory was fixed
443 * the lustre-HOWTO was brought (more) up to date (582544)
444 * major progress was made on recovery functionality
446 2002-10-10 Phil Schwan <phil@clusterfs.com>
449 - recovery deadlock fix
450 - rm -rf causes LBUG fix (617817)
451 - file open by multiple tasks fix (618962)
452 - directory permissions bugs (602707 and 620007)
453 - journal_stop fixed with locking (611313)
454 - O_APPEND failures resolved (618273, perhaps 614459)
455 - lconf PATH fix (619770)
456 - IA64 build fix (621450)
457 - RPC buffer sizes scale with amount of memory
459 2002-10-01 Phil Schwan <phil@clusterfs.com>
462 - locks would be cancelled without throwing away data pages,
463 resulting in inconsistent data (605627)
464 - inode attributes were not always being refreshed (605627, 612449)
465 - lconf now continues to cleanup after lctl reports an error
466 - MDS now enforces user permissions (602707)
467 - lprocfs cleanup fixed, but not yet enabled (614157)
468 - fixed infinite server hang, should a client not respond to an AST
469 - avoid going into recovery if user calls readlink() with a buffer
470 that's too small (613941)
471 - AST RPCs no longer require replies (614867) -- this may be changed
472 - don't crash server if client sends an IOV that's too big (611336)
473 - fixed lock conversion deadlock (611892)
474 - fixed the following of symlinks (614622)
475 * recovery: the server can remove locks from a client that dies, other
476 clients can make progress
477 * more extN patch fixes
478 * compile-time configurable ptlrpc buffer allocations
480 - collaborative read cache document
481 - Lustre Lite Performance CDR document-in-progress
483 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
486 - fix typo in patch-2.4.18
488 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
491 - clear ptlrpc request each time in handle_incoming_request()
492 - unlink of files now destroys the object on the OST
494 2002-09-19 Peter Braam <braam@clusterfs.com>
496 * add hard link support
497 * change obdfile creation method
498 * kernel patch changed
500 2002-09-19 Peter Braam <braam@clusterfs.com>
503 - stack overflow bug in extN fixed
505 2002-09-18 Andreas Dilger <adilger@clusterfs.com>
507 * documentation updates
508 - add man pages for config tools
509 - update tests/README to describe testing with new config tools
510 - finish metadata API descriptions
511 * bug fixes and cleanups
512 - statfs workaround for 16TB limit
513 - LOV stripe allocation improved, can stripe on subset of OSTs
514 - LOV file size/IO offset was wrong for files > 4GB in size
515 - object EA data was being dropped, caused files to be unreadable
516 - memory overflow with non-LOV OST caused memory corruption
517 - fixed regression tests to work with new config tools, obdfilter
518 - fixed bug when directory size became larger than 1 block
519 - fixed bug (for single client case) when PWD was deleted
520 - invalidate local directory pages when doing intent-based ops
521 - avoid LDLM oops when lock callback contained bad data
523 2002-09-09 Andreas Dilger <adilger@clusterfs.com>
525 * documentation updates
526 * bug fixes and cleanups
527 - configuration tools
530 - 64-bit compile warnings
531 - 64-bit internal statfs data
533 * test_brw on persistent OST devices
537 2002-09-04 Andreas Dilger <adilger@clusterfs.com>
539 * documentation updates
540 * bug fixes and cleanups
541 * configuration tools
543 2002-08-30 Peter J. Braam <braam@clusterfs.com>
546 * many small fixes to 0_5_4
547 * io/network handling
548 * thinkos in MDS operations
550 2002-08-24 Peter J. Braam <braam@clusterfs.com>
553 * crucial basic fixes to 0.5.3
554 * IOR, Iozone work over Elan
557 2002-08-07 Phil Schwan <phil@clusterfs.com>
558 * version 0_5_3, our first alpha
559 * we use the new Portals iovs
560 * documentation updates
561 * bug fixes and cleanups
562 * small changes in the DLM wire protocol
564 2002-07-25 Peter J. Braam <braam@clusterfs.com>
565 * version 0_5_1 with some initial stability,
566 * locking on MD and file I/O.
567 * documentation updates
568 * several bug fixes since 0.5.0
569 * small changes in wire protocol
571 2002-07-18 Phil Schwan <phil@clusterfs.com>
573 * delivered as Lustre Light Alpha
574 * fixed a crash after handling invalid MDS requests
575 * fixed directory pages for architectures with non-4k pages sizes
577 2002-07-11 Andreas Dilger <adilger@clusterfs.com>
578 * release version v0_4_4
579 * Moves TCP acceptor to be on port 2432 (unused Coda port) instead
581 * Fixes a number of interruption problems with OST operations.
582 * Update documentation for portals header changes
583 * Move all wire protocol structs/defines to lustre_idl.h
584 * Fixes symlink length bug.
585 * Add tcpdump to repository.
587 2002-07-05 Andreas Dilger <adilger@clusterfs.com>
588 * release version v0_4_3
589 * Fixes statfs for inodes on extN.
590 * Fixes bug in runtests which would delete /etc/hosts.
591 * Use 64-bit object IDs wherever possible (not into VFS though)
592 Remove ost_get_info, which is unused by lustre, and out of date.
594 2002-07-03 Peter Braam <braam@clusterfs.com>
595 * release version v0_4_2 Fixes a lookup error (type not passed)
596 * move forward to head of Portals
597 * move forward to latest Lustre kernel
599 2002-06-25 Peter Braam <braam@clusterfs.com>
600 * release version v0_4_1. Hopefully stable on single node use.