4 - fix removal of OSCs from LOV when they fail
5 - fix NULL deref during bulk timeout (214)
6 - fix problems related to multiple filesystems on one MDS (241)
7 - fixed serious subtle metadata locking bugs
8 - free locks on clients when inodes are removed due to memory
10 - fix inode pointer in lock data (285)
11 - partial support for multiple MDS on a single host (623279, 241)
12 - data locks weren't cancelled at clear_inode time (290, 311)
13 - intent locks could lead to unbounded lock growth (205)
15 - LDLM cancellations and callbacks on different portals
17 2002-10-28 Andreas Dilger <adilger@clusterfs.com>
20 - limit client IOV size to PTL_MD_MAX_IOV (611336, 191)
21 - defer open object destruction to close time (601981, 138)
22 - open/close OST file handle in obdo (OBD_MD_FLHANDLE) (601981, 138)
23 - move LDLM_ENQUEUE/CONVERT back to MDS portal (625069)
24 - abstract ll_lookup2, fix ll_revalidate2 to use abstraction (256)
25 - don't call obd_setattr in ll_file_release for destroyed objects
26 * protocol change to lustre_msg: move |version| and add |flags|
27 * protocol change to osc_punch: "start" in "o_size", "end" in "o_blocks"
28 * lock replay: for LDLM_FL_REPLAY trust client to do right thing
29 * added replay of create, unlink, link and rename operations during
30 MDS failover; recovery should be much more robust now
31 * remove failed OSCs from LOVs (only lov_create uses this so far)
32 * the lustre-HOWTO was brought (more) up to date (582544)
34 2002-10-23 Phil Schwan <phil@clusterfs.com>
37 - in-use dentries weren't being reused properly (617851)
38 - prevent multiple LDLM setup (599178)
39 - fix LOV size calculations for truncate (617853)
40 - fix client handling of MDS intent errors (POSIX)
41 - fix permission bug in lovstripe.c test (624321)
42 - fix MDS thread deadlock - move LDLM handler to DLM portal (625069)
43 - truncate past end of file could corrupt data
44 - proper cleanup after timeouts, crashes, etc (592524, 550815)
45 - a race in recovery could return ETIMEDOUT to apps (623947)
46 - building outside the source directory was fixed
47 * the lustre-HOWTO was brought (more) up to date (582544)
48 * major progress was made on recovery functionality
50 2002-10-10 Phil Schwan <phil@clusterfs.com>
53 - recovery deadlock fix
54 - rm -rf causes LBUG fix (617817)
55 - file open by multiple tasks fix (618962)
56 - directory permissions bugs (602707 and 620007)
57 - journal_stop fixed with locking (611313)
58 - O_APPEND failures resolved (618273, perhaps 614459)
59 - lconf PATH fix (619770)
60 - IA64 build fix (621450)
61 - RPC buffer sizes scale with amount of memory
63 2002-10-01 Phil Schwan <phil@clusterfs.com>
66 - locks would be cancelled without throwing away data pages,
67 resulting in inconsistent data (605627)
68 - inode attributes were not always being refreshed (605627, 612449)
69 - lconf now continues to cleanup after lctl reports an error
70 - MDS now enforces user permissions (602707)
71 - lprocfs cleanup fixed, but not yet enabled (614157)
72 - fixed infinite server hang, should a client not respond to an AST
73 - avoid going into recovery if user calls readlink() with a buffer
74 that's too small (613941)
75 - AST RPCs no longer require replies (614867) -- this may be changed
76 - don't crash server if client sends an IOV that's too big (611336)
77 - fixed lock conversion deadlock (611892)
78 - fixed the following of symlinks (614622)
79 * recovery: the server can remove locks from a client that dies, other
80 clients can make progress
81 * more extN patch fixes
82 * compile-time configurable ptlrpc buffer allocations
84 - collaborative read cache document
85 - Lustre Lite Performance CDR document-in-progress
87 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
90 - fix typo in patch-2.4.18
92 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
95 - clear ptlrpc request each time in handle_incoming_request()
96 - unlink of files now destroys the object on the OST
98 2002-09-19 Peter Braam <braam@clusterfs.com>
100 * add hard link support
101 * change obdfile creation method
102 * kernel patch changed
104 2002-09-19 Peter Braam <braam@clusterfs.com>
107 - stack overflow bug in extN fixed
109 2002-09-18 Andreas Dilger <adilger@clusterfs.com>
111 * documentation updates
112 - add man pages for config tools
113 - update tests/README to describe testing with new config tools
114 - finish metadata API descriptions
115 * bug fixes and cleanups
116 - statfs workaround for 16TB limit
117 - LOV stripe allocation improved, can stripe on subset of OSTs
118 - LOV file size/IO offset was wrong for files > 4GB in size
119 - object EA data was being dropped, caused files to be unreadable
120 - memory overflow with non-LOV OST caused memory corruption
121 - fixed regression tests to work with new config tools, obdfilter
122 - fixed bug when directory size became larger than 1 block
123 - fixed bug (for single client case) when PWD was deleted
124 - invalidate local directory pages when doing intent-based ops
125 - avoid LDLM oops when lock callback contained bad data
127 2002-09-09 Andreas Dilger <adilger@clusterfs.com>
129 * documentation updates
130 * bug fixes and cleanups
131 - configuration tools
134 - 64-bit compile warnings
135 - 64-bit internal statfs data
137 * test_brw on persistent OST devices
141 2002-09-04 Andreas Dilger <adilger@clusterfs.com>
143 * documentation updates
144 * bug fixes and cleanups
145 * configuration tools
147 2002-08-30 Peter J. Braam <braam@clusterfs.com>
150 * many small fixes to 0_5_4
151 * io/network handling
152 * thinkos in MDS operations
154 2002-08-24 Peter J. Braam <braam@clusterfs.com>
157 * crucial basic fixes to 0.5.3
158 * IOR, Iozone work over Elan
161 2002-08-07 Phil Schwan <phil@clusterfs.com>
162 * version 0_5_3, our first alpha
163 * we use the new Portals iovs
164 * documentation updates
165 * bug fixes and cleanups
166 * small changes in the DLM wire protocol
168 2002-07-25 Peter J. Braam <braam@clusterfs.com>
169 * version 0_5_1 with some initial stability,
170 * locking on MD and file I/O.
171 * documentation updates
172 * several bug fixes since 0.5.0
173 * small changes in wire protocol
175 2002-07-18 Phil Schwan <phil@clusterfs.com>
177 * delivered as Lustre Light Alpha
178 * fixed a crash after handling invalid MDS requests
179 * fixed directory pages for architectures with non-4k pages sizes
181 2002-07-11 Andreas Dilger <adilger@clusterfs.com>
182 * release version v0_4_4
183 * Moves TCP acceptor to be on port 2432 (unused Coda port) instead
185 * Fixes a number of interruption problems with OST operations.
186 * Update documentation for portals header changes
187 * Move all wire protocol structs/defines to lustre_idl.h
188 * Fixes symlink length bug.
189 * Add tcpdump to repository.
191 2002-07-05 Andreas Dilger <adilger@clusterfs.com>
192 * release version v0_4_3
193 * Fixes statfs for inodes on extN.
194 * Fixes bug in runtests which would delete /etc/hosts.
195 * Use 64-bit object IDs wherever possible (not into VFS though)
196 Remove ost_get_info, which is unused by lustre, and out of date.
198 2002-07-03 Peter Braam <braam@clusterfs.com>
199 * release version v0_4_2 Fixes a lookup error (type not passed)
200 * move forward to head of Portals
201 * move forward to latest Lustre kernel
203 2002-06-25 Peter Braam <braam@clusterfs.com>
204 * release version v0_4_1. Hopefully stable on single node use.