4 - LDLM_DEBUG macro fix, for gcc 3.2 (850)
5 - failed open()s could cause deadlock; fixed (867, 869)
6 - stop cancelling OST locks when files are closed (481)
7 - overlapping XID spaces caused network corruption (851, 853)
8 - fix unsafe fsfilt counter arithmetic; change to atomic_t
9 - setattr_raw added, to do single-RPC, server-side setattrs
10 - lmc/lconf syntax change for OST UUIDs
11 - fix crashy race condition between ptlrpc_free_req and osc_close
12 - don't use request in mdc_enqueue if we hit a timeout (889)
13 - don't set the inode i_size for regular files from the MDS (896)
14 - handle out of order completion AST (842)
15 - don't LBUG if a lock request times out after receiving AST (913)
16 - avoid d_rehash race in ll_find_alias by rehashing inside dcache_lock
17 - if a bad lock AST arrives, send an error instead of dropping entirely
18 - return 0 from revalidate2 if ll_intent_lock returns -EINTR (912)
19 - fix leak in bulk IO when only partially completed (899, 900, 926)
21 - READPAGE and SETATTRs which don't take server-side locks get
24 2003-02-11 Phil Schwan <phil@clusterfs.com>
27 - Fix ldlm_lock_match on the MDS to avoid matching remote locks (592)
28 - Fix fsfilt_extN_readpage() to read a full page of directory
29 entries, or fake the remainder if PAGE_SIZE != blocksize (500)
30 - Avoid extra mdc_getattr() in ll_intent_lock when possible (534, 604)
31 - Fix imbalanced LOV object allocation and out-of-bound access (469)
32 - Most intent operations were removed, in favour of a new RPC mode
33 that does a single RPC to the server and bypasses most of the VFS
34 - All LDLM resource ID arrays were removed in favour of ldlm_res_id
35 - Aggressively cancel local locks on DLM servers
36 - mds_reint_unlink sends EA to the client if it's the last nlink.
37 client uses that EA to unlink OST objects.
38 - mds_reint_{rename,unlink,link} were rewritten to take ordered locks
39 - recursive symlinks were fixed (439)
40 - fixed NULL deref in DEBUG_REQ
41 - filter_update_lastobjid no longer calls sync, which annoyed extN
42 - fixed multi-client small-writes to a single file problem (445)
43 - fixed mtime updates during file writes (607)
44 - fixed vector writes on obdfilter causing problems when ENOSPC (670)
45 - fixed bug in obd_brw_read/write() (under guise of testing 367)
46 - fixed Linux OST size reporting problem (444, 656)
47 - OST now updates object mtime with writes or setattr (607, 619)
48 - client verifies file size before zeroing page past EOF (445)
49 - OST now writes last allocated objid to disk with allocation (108)
50 - LOV on echo now works (409)
52 - mds_reint_unlink sends a new buffer, with the EA included. this
53 buffer is only valid if body->valid & OBD_MD_FLEASIZE, which is only
54 set if a regular file was being unlinked, and it was the last link
55 - use PtlGet from the target for bulk writes (315)
56 - OST now updates object mtime with writes or setattr (607, 619)
57 - LDLM now has a grant-time callback to revalidate locked items, if
59 - Many MDS operations were reorganized to combat race conditions
61 - Merge b_intel branch (updated lprocfs code) - now at /proc/fs/lustre
62 - configure check to avoid gcc version 2.96 20000731-2.96-98) (606)
64 2003-01-06 Andreas Dilger <adilger@clusterfs.com>
67 - Fully reactivate OST imports after reconnection (512, others)
68 - Make sure client sees our -ENOTCONN from mds_handle (513 - partial)
69 - More graceful error handling for truncating on dead OST (515)
70 - Don't error out unless we're actually accessing dead stripes (474)
71 - Fix garbage sizes when stripes are missing (410)
72 - LRU counters were broken, causing constant lock purge (433, 432)
73 - garbage on read from stripes with failed OSTs (441)
74 - mark OSCs as active before reconnecting during recovery (438)
75 - lov_enqueue and lov_cancel need to handle inactive OSTs (403)
76 - lfind did not preserve OST order in output (443)
77 - symlinks cause hung clients, incorrect data (439)
78 - stop dereferencing request after dropping refcount (457)
79 - don't LASSERT(spin_is_locked) on non-SMP (455)
80 - fixes for many rename() bugs
81 - fstat didn't correctly synchronize attributes (399)
82 - server must handle lock cancellation during blocking AST prep (487)
83 - bulk descriptors were free()d too soon (511)
84 - fix paths in lconf, which would load incorrect modules (451, 507)
85 - fix confusing lconf 'host not found' error message (386)
86 - fix lock order deadlock on OST (O/R i_sem before journal ops, 478)
87 - fix race condition in mdc_blocking_ast() for inode access (526)
88 - fix lov_unpackmd() unpacking wrong number of stripes (537)
89 - fix lov_set_osc_active() marking wrong OSC inactive (440)
90 - fix bad lstripe lov_unpackmd() assertion (fix layering too) (527)
91 - fix multiple writes of stripe MD to MDS (358, maybe 519)
92 - fix lstripe in several ways (kernel side) (527)
93 - fix request leak in ldlm_cli_enqueue (262)
94 - incorrect OSC was marked inactive after OST failure
95 - call mds_fs_cleanup before unmounting filesystem (524)
96 - fix races between taking ns_lock and ldlm_lock_change_resource
97 - fix races updating LOV export open file list
98 - fix lov_enqueue error path, avoid decref-ing bad lock handle
99 - fix recovery NULL deref in ldlm_cli_cancel_unused
100 - fix some DLM races by using new hash table for lock handles (419)
101 - permit the client to specify desired inodes, at replay
102 - duplicate requests when we queue them for replay reintegration
103 - fix last_rcvd offset calculation
104 - sync after each recovered transaction, so we always make progress
105 - never, not always, ERESTART requests without transnos
106 - store the lov_desc in the MDS, so we don't depend on getlovinfo to
108 - skip replay if the MDS says that the client is already connected
109 - don't check for a recovery-enabled export to match lctl's UUID
110 - don't INC_USE_COUNT for phantom exports
111 - don't crash when cleaning up phantom exports (567)
112 - don't double-finish or set replay data for errored mdc_open requests
113 - abort requests when they time out, so we don't get old replies
114 - send/receive replies for AST messages again
115 - if the client says that it doesn't have the lock, cancel it on the
117 - if we timeout during I/O, don't try to cancel an in-use lock; instead
118 mark it as destroyed, it will all work out when decref is called
119 - fix module use counts (22, 581)
121 - ASTs now expect a reply (server cancels lock on error reply)
123 2002-12-02 Andreas Dilger <adilger@clusterfs.com>
126 - fix many simultaneous client startup (392)
127 - fix dentry->d_it clobbering
128 - credentials weren't being shipped for readdir/getattr operations
129 - remove invalid assertions triggered during some concurrent MD
131 - proper Lustre versions added (336, 389)
132 - fix memory leak for create error case (398)
133 - fix LOV locking bug that would get cli/srv out of sync
134 - fix echo client over LOV (409)
135 - fix dbench 2, extN refcount problem (170, 258, 356, 418)
136 - fix double-O_EXCL intent crash (424)
137 - avoid sending multiple lock CANCELs (352)
139 - MDS can do multi-client recovery (modulo bugs in new code)
141 - many updates, edits, cleanups
143 2002-11-18 Phil Schwan <phil@clusterfs.com>
146 - fix null d_it dereference (346)
147 - fix full OST/dbench hang (333)
148 - fix permission problem with file removal (286)
149 - fix removal of OSCs from LOV when they fail
150 - fix NULL deref during bulk timeout (214)
151 - fix problems related to multiple filesystems on one MDS (241)
152 - fixed serious subtle metadata locking bugs
153 - free locks on clients when inodes are removed due to memory
155 - fix inode pointer in lock data (285)
156 - partial support for multiple MDS on a single host (241)
157 - data locks weren't cancelled at clear_inode time (290, 311)
158 - intent locks could lead to unbounded lock growth (205)
159 - added a maximum lock count, an LRU list, and a flusher
160 - fix multiple rename (365)
161 - properly abstracted the echo client
162 - OSC locked 1 byte too many; fixed
163 - rewrote brw callback code:
164 - fixed recovery bugs related to LOVs (306)
165 - fixed too-many-pages-in-one-write crash (191)
166 - fixed (again) crash in sync_io_timeout (214)
167 - probably fixed callback-related race (385)
169 - Add capability to MDS protocol
170 - LDLM cancellations and callbacks on different portals
172 2002-10-28 Andreas Dilger <adilger@clusterfs.com>
175 - limit client IOV size to PTL_MD_MAX_IOV (611336, 191)
176 - defer open object destruction to close time (601981, 138)
177 - open/close OST file handle in obdo (OBD_MD_FLHANDLE) (601981, 138)
178 - move LDLM_ENQUEUE/CONVERT back to MDS portal (625069)
179 - abstract ll_lookup2, fix ll_revalidate2 to use abstraction (256)
180 - don't call obd_setattr in ll_file_release for destroyed objects
181 * protocol change to lustre_msg: move |version| and add |flags|
182 * protocol change to osc_punch: "start" in "o_size", "end" in "o_blocks"
183 * lock replay: for LDLM_FL_REPLAY trust client to do right thing
184 * added replay of create, unlink, link and rename operations during
185 MDS failover; recovery should be much more robust now
186 * remove failed OSCs from LOVs (only lov_create uses this so far)
187 * the lustre-HOWTO was brought (more) up to date (582544)
189 2002-10-23 Phil Schwan <phil@clusterfs.com>
192 - in-use dentries weren't being reused properly (617851)
193 - prevent multiple LDLM setup (599178)
194 - fix LOV size calculations for truncate (617853)
195 - fix client handling of MDS intent errors (POSIX)
196 - fix permission bug in lovstripe.c test (624321)
197 - fix MDS thread deadlock - move LDLM handler to DLM portal (625069)
198 - truncate past end of file could corrupt data
199 - proper cleanup after timeouts, crashes, etc (592524, 550815)
200 - a race in recovery could return ETIMEDOUT to apps (623947)
201 - building outside the source directory was fixed
202 * the lustre-HOWTO was brought (more) up to date (582544)
203 * major progress was made on recovery functionality
205 2002-10-10 Phil Schwan <phil@clusterfs.com>
208 - recovery deadlock fix
209 - rm -rf causes LBUG fix (617817)
210 - file open by multiple tasks fix (618962)
211 - directory permissions bugs (602707 and 620007)
212 - journal_stop fixed with locking (611313)
213 - O_APPEND failures resolved (618273, perhaps 614459)
214 - lconf PATH fix (619770)
215 - IA64 build fix (621450)
216 - RPC buffer sizes scale with amount of memory
218 2002-10-01 Phil Schwan <phil@clusterfs.com>
221 - locks would be cancelled without throwing away data pages,
222 resulting in inconsistent data (605627)
223 - inode attributes were not always being refreshed (605627, 612449)
224 - lconf now continues to cleanup after lctl reports an error
225 - MDS now enforces user permissions (602707)
226 - lprocfs cleanup fixed, but not yet enabled (614157)
227 - fixed infinite server hang, should a client not respond to an AST
228 - avoid going into recovery if user calls readlink() with a buffer
229 that's too small (613941)
230 - AST RPCs no longer require replies (614867) -- this may be changed
231 - don't crash server if client sends an IOV that's too big (611336)
232 - fixed lock conversion deadlock (611892)
233 - fixed the following of symlinks (614622)
234 * recovery: the server can remove locks from a client that dies, other
235 clients can make progress
236 * more extN patch fixes
237 * compile-time configurable ptlrpc buffer allocations
239 - collaborative read cache document
240 - Lustre Lite Performance CDR document-in-progress
242 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
245 - fix typo in patch-2.4.18
247 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
250 - clear ptlrpc request each time in handle_incoming_request()
251 - unlink of files now destroys the object on the OST
253 2002-09-19 Peter Braam <braam@clusterfs.com>
255 * add hard link support
256 * change obdfile creation method
257 * kernel patch changed
259 2002-09-19 Peter Braam <braam@clusterfs.com>
262 - stack overflow bug in extN fixed
264 2002-09-18 Andreas Dilger <adilger@clusterfs.com>
266 * documentation updates
267 - add man pages for config tools
268 - update tests/README to describe testing with new config tools
269 - finish metadata API descriptions
270 * bug fixes and cleanups
271 - statfs workaround for 16TB limit
272 - LOV stripe allocation improved, can stripe on subset of OSTs
273 - LOV file size/IO offset was wrong for files > 4GB in size
274 - object EA data was being dropped, caused files to be unreadable
275 - memory overflow with non-LOV OST caused memory corruption
276 - fixed regression tests to work with new config tools, obdfilter
277 - fixed bug when directory size became larger than 1 block
278 - fixed bug (for single client case) when PWD was deleted
279 - invalidate local directory pages when doing intent-based ops
280 - avoid LDLM oops when lock callback contained bad data
282 2002-09-09 Andreas Dilger <adilger@clusterfs.com>
284 * documentation updates
285 * bug fixes and cleanups
286 - configuration tools
289 - 64-bit compile warnings
290 - 64-bit internal statfs data
292 * test_brw on persistent OST devices
296 2002-09-04 Andreas Dilger <adilger@clusterfs.com>
298 * documentation updates
299 * bug fixes and cleanups
300 * configuration tools
302 2002-08-30 Peter J. Braam <braam@clusterfs.com>
305 * many small fixes to 0_5_4
306 * io/network handling
307 * thinkos in MDS operations
309 2002-08-24 Peter J. Braam <braam@clusterfs.com>
312 * crucial basic fixes to 0.5.3
313 * IOR, Iozone work over Elan
316 2002-08-07 Phil Schwan <phil@clusterfs.com>
317 * version 0_5_3, our first alpha
318 * we use the new Portals iovs
319 * documentation updates
320 * bug fixes and cleanups
321 * small changes in the DLM wire protocol
323 2002-07-25 Peter J. Braam <braam@clusterfs.com>
324 * version 0_5_1 with some initial stability,
325 * locking on MD and file I/O.
326 * documentation updates
327 * several bug fixes since 0.5.0
328 * small changes in wire protocol
330 2002-07-18 Phil Schwan <phil@clusterfs.com>
332 * delivered as Lustre Light Alpha
333 * fixed a crash after handling invalid MDS requests
334 * fixed directory pages for architectures with non-4k pages sizes
336 2002-07-11 Andreas Dilger <adilger@clusterfs.com>
337 * release version v0_4_4
338 * Moves TCP acceptor to be on port 2432 (unused Coda port) instead
340 * Fixes a number of interruption problems with OST operations.
341 * Update documentation for portals header changes
342 * Move all wire protocol structs/defines to lustre_idl.h
343 * Fixes symlink length bug.
344 * Add tcpdump to repository.
346 2002-07-05 Andreas Dilger <adilger@clusterfs.com>
347 * release version v0_4_3
348 * Fixes statfs for inodes on extN.
349 * Fixes bug in runtests which would delete /etc/hosts.
350 * Use 64-bit object IDs wherever possible (not into VFS though)
351 Remove ost_get_info, which is unused by lustre, and out of date.
353 2002-07-03 Peter Braam <braam@clusterfs.com>
354 * release version v0_4_2 Fixes a lookup error (type not passed)
355 * move forward to head of Portals
356 * move forward to latest Lustre kernel
358 2002-06-25 Peter Braam <braam@clusterfs.com>
359 * release version v0_4_1. Hopefully stable on single node use.