3 - fix many simultaneous client startup (392)
4 - fix dentry->d_it clobbering
5 - credentials weren't being shipped for readdir/getattr operations
6 - remove invalid assertions triggered during some concurrent MD
8 - proper Lustre versions added (336, 389)
9 - fix memory leak for create error case (398)
10 - fix LOV locking bug that would get cli/srv out of sync
11 - fix echo client over LOV (409)
12 - fix dbench 2, extN refcount problem (170, 258, 356, 418)
13 - fix double-O_EXCL intent crash (424)
15 2002-11-18 Phil Schwan <phil@clusterfs.com>
18 - fix null d_it dereference (346)
19 - fix full OST/dbench hang (333)
20 - fix permission problem with file removal (286)
21 - fix removal of OSCs from LOV when they fail
22 - fix NULL deref during bulk timeout (214)
23 - fix problems related to multiple filesystems on one MDS (241)
24 - fixed serious subtle metadata locking bugs
25 - free locks on clients when inodes are removed due to memory
27 - fix inode pointer in lock data (285)
28 - partial support for multiple MDS on a single host (241)
29 - data locks weren't cancelled at clear_inode time (290, 311)
30 - intent locks could lead to unbounded lock growth (205)
31 - added a maximum lock count, an LRU list, and a flusher
32 - fix multiple rename (365)
33 - properly abstracted the echo client
34 - OSC locked 1 byte too many; fixed
35 - rewrote brw callback code:
36 - fixed recovery bugs related to LOVs (306)
37 - fixed too-many-pages-in-one-write crash (191)
38 - fixed (again) crash in sync_io_timeout (214)
39 - probably fixed callback-related race (385)
41 - Add capability to MDS protocol
42 - LDLM cancellations and callbacks on different portals
44 2002-10-28 Andreas Dilger <adilger@clusterfs.com>
47 - limit client IOV size to PTL_MD_MAX_IOV (611336, 191)
48 - defer open object destruction to close time (601981, 138)
49 - open/close OST file handle in obdo (OBD_MD_FLHANDLE) (601981, 138)
50 - move LDLM_ENQUEUE/CONVERT back to MDS portal (625069)
51 - abstract ll_lookup2, fix ll_revalidate2 to use abstraction (256)
52 - don't call obd_setattr in ll_file_release for destroyed objects
53 * protocol change to lustre_msg: move |version| and add |flags|
54 * protocol change to osc_punch: "start" in "o_size", "end" in "o_blocks"
55 * lock replay: for LDLM_FL_REPLAY trust client to do right thing
56 * added replay of create, unlink, link and rename operations during
57 MDS failover; recovery should be much more robust now
58 * remove failed OSCs from LOVs (only lov_create uses this so far)
59 * the lustre-HOWTO was brought (more) up to date (582544)
61 2002-10-23 Phil Schwan <phil@clusterfs.com>
64 - in-use dentries weren't being reused properly (617851)
65 - prevent multiple LDLM setup (599178)
66 - fix LOV size calculations for truncate (617853)
67 - fix client handling of MDS intent errors (POSIX)
68 - fix permission bug in lovstripe.c test (624321)
69 - fix MDS thread deadlock - move LDLM handler to DLM portal (625069)
70 - truncate past end of file could corrupt data
71 - proper cleanup after timeouts, crashes, etc (592524, 550815)
72 - a race in recovery could return ETIMEDOUT to apps (623947)
73 - building outside the source directory was fixed
74 * the lustre-HOWTO was brought (more) up to date (582544)
75 * major progress was made on recovery functionality
77 2002-10-10 Phil Schwan <phil@clusterfs.com>
80 - recovery deadlock fix
81 - rm -rf causes LBUG fix (617817)
82 - file open by multiple tasks fix (618962)
83 - directory permissions bugs (602707 and 620007)
84 - journal_stop fixed with locking (611313)
85 - O_APPEND failures resolved (618273, perhaps 614459)
86 - lconf PATH fix (619770)
87 - IA64 build fix (621450)
88 - RPC buffer sizes scale with amount of memory
90 2002-10-01 Phil Schwan <phil@clusterfs.com>
93 - locks would be cancelled without throwing away data pages,
94 resulting in inconsistent data (605627)
95 - inode attributes were not always being refreshed (605627, 612449)
96 - lconf now continues to cleanup after lctl reports an error
97 - MDS now enforces user permissions (602707)
98 - lprocfs cleanup fixed, but not yet enabled (614157)
99 - fixed infinite server hang, should a client not respond to an AST
100 - avoid going into recovery if user calls readlink() with a buffer
101 that's too small (613941)
102 - AST RPCs no longer require replies (614867) -- this may be changed
103 - don't crash server if client sends an IOV that's too big (611336)
104 - fixed lock conversion deadlock (611892)
105 - fixed the following of symlinks (614622)
106 * recovery: the server can remove locks from a client that dies, other
107 clients can make progress
108 * more extN patch fixes
109 * compile-time configurable ptlrpc buffer allocations
111 - collaborative read cache document
112 - Lustre Lite Performance CDR document-in-progress
114 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
117 - fix typo in patch-2.4.18
119 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
122 - clear ptlrpc request each time in handle_incoming_request()
123 - unlink of files now destroys the object on the OST
125 2002-09-19 Peter Braam <braam@clusterfs.com>
127 * add hard link support
128 * change obdfile creation method
129 * kernel patch changed
131 2002-09-19 Peter Braam <braam@clusterfs.com>
134 - stack overflow bug in extN fixed
136 2002-09-18 Andreas Dilger <adilger@clusterfs.com>
138 * documentation updates
139 - add man pages for config tools
140 - update tests/README to describe testing with new config tools
141 - finish metadata API descriptions
142 * bug fixes and cleanups
143 - statfs workaround for 16TB limit
144 - LOV stripe allocation improved, can stripe on subset of OSTs
145 - LOV file size/IO offset was wrong for files > 4GB in size
146 - object EA data was being dropped, caused files to be unreadable
147 - memory overflow with non-LOV OST caused memory corruption
148 - fixed regression tests to work with new config tools, obdfilter
149 - fixed bug when directory size became larger than 1 block
150 - fixed bug (for single client case) when PWD was deleted
151 - invalidate local directory pages when doing intent-based ops
152 - avoid LDLM oops when lock callback contained bad data
154 2002-09-09 Andreas Dilger <adilger@clusterfs.com>
156 * documentation updates
157 * bug fixes and cleanups
158 - configuration tools
161 - 64-bit compile warnings
162 - 64-bit internal statfs data
164 * test_brw on persistent OST devices
168 2002-09-04 Andreas Dilger <adilger@clusterfs.com>
170 * documentation updates
171 * bug fixes and cleanups
172 * configuration tools
174 2002-08-30 Peter J. Braam <braam@clusterfs.com>
177 * many small fixes to 0_5_4
178 * io/network handling
179 * thinkos in MDS operations
181 2002-08-24 Peter J. Braam <braam@clusterfs.com>
184 * crucial basic fixes to 0.5.3
185 * IOR, Iozone work over Elan
188 2002-08-07 Phil Schwan <phil@clusterfs.com>
189 * version 0_5_3, our first alpha
190 * we use the new Portals iovs
191 * documentation updates
192 * bug fixes and cleanups
193 * small changes in the DLM wire protocol
195 2002-07-25 Peter J. Braam <braam@clusterfs.com>
196 * version 0_5_1 with some initial stability,
197 * locking on MD and file I/O.
198 * documentation updates
199 * several bug fixes since 0.5.0
200 * small changes in wire protocol
202 2002-07-18 Phil Schwan <phil@clusterfs.com>
204 * delivered as Lustre Light Alpha
205 * fixed a crash after handling invalid MDS requests
206 * fixed directory pages for architectures with non-4k pages sizes
208 2002-07-11 Andreas Dilger <adilger@clusterfs.com>
209 * release version v0_4_4
210 * Moves TCP acceptor to be on port 2432 (unused Coda port) instead
212 * Fixes a number of interruption problems with OST operations.
213 * Update documentation for portals header changes
214 * Move all wire protocol structs/defines to lustre_idl.h
215 * Fixes symlink length bug.
216 * Add tcpdump to repository.
218 2002-07-05 Andreas Dilger <adilger@clusterfs.com>
219 * release version v0_4_3
220 * Fixes statfs for inodes on extN.
221 * Fixes bug in runtests which would delete /etc/hosts.
222 * Use 64-bit object IDs wherever possible (not into VFS though)
223 Remove ost_get_info, which is unused by lustre, and out of date.
225 2002-07-03 Peter Braam <braam@clusterfs.com>
226 * release version v0_4_2 Fixes a lookup error (type not passed)
227 * move forward to head of Portals
228 * move forward to latest Lustre kernel
230 2002-06-25 Peter Braam <braam@clusterfs.com>
231 * release version v0_4_1. Hopefully stable on single node use.