3 - LRU counters were broken, causing constant lock purge (433, 432)
4 - garbage on read from stripes with failed OSTs (441)
5 - mark OSCs as active before reconnecting during recovery (438)
6 - lov_enqueue and lov_cancel need to handle inactive OSTs (403)
7 - lfind did not preserve OST order in output (443)
8 - symlinks cause hung clients, incorrect data (439)
9 - stop dereferencing request after dropping refcount (457)
10 - don't LASSERT(spin_is_locked) on non-SMP (455)
11 - fixes for many rename() bugs
13 2002-12-02 Andreas Dilger <adilger@clusterfs.com>
16 - fix many simultaneous client startup (392)
17 - fix dentry->d_it clobbering
18 - credentials weren't being shipped for readdir/getattr operations
19 - remove invalid assertions triggered during some concurrent MD
21 - proper Lustre versions added (336, 389)
22 - fix memory leak for create error case (398)
23 - fix LOV locking bug that would get cli/srv out of sync
24 - fix echo client over LOV (409)
25 - fix dbench 2, extN refcount problem (170, 258, 356, 418)
26 - fix double-O_EXCL intent crash (424)
27 - avoid sending multiple lock CANCELs (352)
29 - MDS can do multi-client recovery (modulo bugs in new code)
31 - many updates, edits, cleanups
33 2002-11-18 Phil Schwan <phil@clusterfs.com>
36 - fix null d_it dereference (346)
37 - fix full OST/dbench hang (333)
38 - fix permission problem with file removal (286)
39 - fix removal of OSCs from LOV when they fail
40 - fix NULL deref during bulk timeout (214)
41 - fix problems related to multiple filesystems on one MDS (241)
42 - fixed serious subtle metadata locking bugs
43 - free locks on clients when inodes are removed due to memory
45 - fix inode pointer in lock data (285)
46 - partial support for multiple MDS on a single host (241)
47 - data locks weren't cancelled at clear_inode time (290, 311)
48 - intent locks could lead to unbounded lock growth (205)
49 - added a maximum lock count, an LRU list, and a flusher
50 - fix multiple rename (365)
51 - properly abstracted the echo client
52 - OSC locked 1 byte too many; fixed
53 - rewrote brw callback code:
54 - fixed recovery bugs related to LOVs (306)
55 - fixed too-many-pages-in-one-write crash (191)
56 - fixed (again) crash in sync_io_timeout (214)
57 - probably fixed callback-related race (385)
59 - Add capability to MDS protocol
60 - LDLM cancellations and callbacks on different portals
62 2002-10-28 Andreas Dilger <adilger@clusterfs.com>
65 - limit client IOV size to PTL_MD_MAX_IOV (611336, 191)
66 - defer open object destruction to close time (601981, 138)
67 - open/close OST file handle in obdo (OBD_MD_FLHANDLE) (601981, 138)
68 - move LDLM_ENQUEUE/CONVERT back to MDS portal (625069)
69 - abstract ll_lookup2, fix ll_revalidate2 to use abstraction (256)
70 - don't call obd_setattr in ll_file_release for destroyed objects
71 * protocol change to lustre_msg: move |version| and add |flags|
72 * protocol change to osc_punch: "start" in "o_size", "end" in "o_blocks"
73 * lock replay: for LDLM_FL_REPLAY trust client to do right thing
74 * added replay of create, unlink, link and rename operations during
75 MDS failover; recovery should be much more robust now
76 * remove failed OSCs from LOVs (only lov_create uses this so far)
77 * the lustre-HOWTO was brought (more) up to date (582544)
79 2002-10-23 Phil Schwan <phil@clusterfs.com>
82 - in-use dentries weren't being reused properly (617851)
83 - prevent multiple LDLM setup (599178)
84 - fix LOV size calculations for truncate (617853)
85 - fix client handling of MDS intent errors (POSIX)
86 - fix permission bug in lovstripe.c test (624321)
87 - fix MDS thread deadlock - move LDLM handler to DLM portal (625069)
88 - truncate past end of file could corrupt data
89 - proper cleanup after timeouts, crashes, etc (592524, 550815)
90 - a race in recovery could return ETIMEDOUT to apps (623947)
91 - building outside the source directory was fixed
92 * the lustre-HOWTO was brought (more) up to date (582544)
93 * major progress was made on recovery functionality
95 2002-10-10 Phil Schwan <phil@clusterfs.com>
98 - recovery deadlock fix
99 - rm -rf causes LBUG fix (617817)
100 - file open by multiple tasks fix (618962)
101 - directory permissions bugs (602707 and 620007)
102 - journal_stop fixed with locking (611313)
103 - O_APPEND failures resolved (618273, perhaps 614459)
104 - lconf PATH fix (619770)
105 - IA64 build fix (621450)
106 - RPC buffer sizes scale with amount of memory
108 2002-10-01 Phil Schwan <phil@clusterfs.com>
111 - locks would be cancelled without throwing away data pages,
112 resulting in inconsistent data (605627)
113 - inode attributes were not always being refreshed (605627, 612449)
114 - lconf now continues to cleanup after lctl reports an error
115 - MDS now enforces user permissions (602707)
116 - lprocfs cleanup fixed, but not yet enabled (614157)
117 - fixed infinite server hang, should a client not respond to an AST
118 - avoid going into recovery if user calls readlink() with a buffer
119 that's too small (613941)
120 - AST RPCs no longer require replies (614867) -- this may be changed
121 - don't crash server if client sends an IOV that's too big (611336)
122 - fixed lock conversion deadlock (611892)
123 - fixed the following of symlinks (614622)
124 * recovery: the server can remove locks from a client that dies, other
125 clients can make progress
126 * more extN patch fixes
127 * compile-time configurable ptlrpc buffer allocations
129 - collaborative read cache document
130 - Lustre Lite Performance CDR document-in-progress
132 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
135 - fix typo in patch-2.4.18
137 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
140 - clear ptlrpc request each time in handle_incoming_request()
141 - unlink of files now destroys the object on the OST
143 2002-09-19 Peter Braam <braam@clusterfs.com>
145 * add hard link support
146 * change obdfile creation method
147 * kernel patch changed
149 2002-09-19 Peter Braam <braam@clusterfs.com>
152 - stack overflow bug in extN fixed
154 2002-09-18 Andreas Dilger <adilger@clusterfs.com>
156 * documentation updates
157 - add man pages for config tools
158 - update tests/README to describe testing with new config tools
159 - finish metadata API descriptions
160 * bug fixes and cleanups
161 - statfs workaround for 16TB limit
162 - LOV stripe allocation improved, can stripe on subset of OSTs
163 - LOV file size/IO offset was wrong for files > 4GB in size
164 - object EA data was being dropped, caused files to be unreadable
165 - memory overflow with non-LOV OST caused memory corruption
166 - fixed regression tests to work with new config tools, obdfilter
167 - fixed bug when directory size became larger than 1 block
168 - fixed bug (for single client case) when PWD was deleted
169 - invalidate local directory pages when doing intent-based ops
170 - avoid LDLM oops when lock callback contained bad data
172 2002-09-09 Andreas Dilger <adilger@clusterfs.com>
174 * documentation updates
175 * bug fixes and cleanups
176 - configuration tools
179 - 64-bit compile warnings
180 - 64-bit internal statfs data
182 * test_brw on persistent OST devices
186 2002-09-04 Andreas Dilger <adilger@clusterfs.com>
188 * documentation updates
189 * bug fixes and cleanups
190 * configuration tools
192 2002-08-30 Peter J. Braam <braam@clusterfs.com>
195 * many small fixes to 0_5_4
196 * io/network handling
197 * thinkos in MDS operations
199 2002-08-24 Peter J. Braam <braam@clusterfs.com>
202 * crucial basic fixes to 0.5.3
203 * IOR, Iozone work over Elan
206 2002-08-07 Phil Schwan <phil@clusterfs.com>
207 * version 0_5_3, our first alpha
208 * we use the new Portals iovs
209 * documentation updates
210 * bug fixes and cleanups
211 * small changes in the DLM wire protocol
213 2002-07-25 Peter J. Braam <braam@clusterfs.com>
214 * version 0_5_1 with some initial stability,
215 * locking on MD and file I/O.
216 * documentation updates
217 * several bug fixes since 0.5.0
218 * small changes in wire protocol
220 2002-07-18 Phil Schwan <phil@clusterfs.com>
222 * delivered as Lustre Light Alpha
223 * fixed a crash after handling invalid MDS requests
224 * fixed directory pages for architectures with non-4k pages sizes
226 2002-07-11 Andreas Dilger <adilger@clusterfs.com>
227 * release version v0_4_4
228 * Moves TCP acceptor to be on port 2432 (unused Coda port) instead
230 * Fixes a number of interruption problems with OST operations.
231 * Update documentation for portals header changes
232 * Move all wire protocol structs/defines to lustre_idl.h
233 * Fixes symlink length bug.
234 * Add tcpdump to repository.
236 2002-07-05 Andreas Dilger <adilger@clusterfs.com>
237 * release version v0_4_3
238 * Fixes statfs for inodes on extN.
239 * Fixes bug in runtests which would delete /etc/hosts.
240 * Use 64-bit object IDs wherever possible (not into VFS though)
241 Remove ost_get_info, which is unused by lustre, and out of date.
243 2002-07-03 Peter Braam <braam@clusterfs.com>
244 * release version v0_4_2 Fixes a lookup error (type not passed)
245 * move forward to head of Portals
246 * move forward to latest Lustre kernel
248 2002-06-25 Peter Braam <braam@clusterfs.com>
249 * release version v0_4_1. Hopefully stable on single node use.