4 - limit Lustre IOVs to PTL_MD_MAX_IOV (611336)
5 * protocol change to lustre_msg: move |version| and add |flags|
6 * added replay of create, unlink, link and rename operations during
7 MDS failover; recovery should be much more robust now
8 * remove failed OSCs from LOVs
10 2002-10-23 Phil Schwan <phil@clusterfs.com>
13 - in-use dentries weren't being reused properly (617851)
14 - prevent multiple LDLM setup (599178)
15 - fix LOV size calculations for truncate (617853)
16 - fix client handling of MDS intent errors (POSIX)
17 - fix permission bug in lovstripe.c test (624321)
18 - fix MDS thread deadlock (625069)
19 - truncate past end of file could corrupt data
20 - proper cleanup after timeouts, crashes, etc (592524, 550815)
21 - a race in recovery could return ETIMEDOUT to apps (623947)
22 - building outside the source directory was fixed
23 * the howto was brought (more) up to date (582544)
24 * major progress was made on recovery functionality
26 2002-10-10 Phil Schwan <phil@clusterfs.com>
29 - recovery deadlock fix
30 - rm -rf causes LBUG fix (617817)
31 - file open by multiple tasks fix (618962)
32 - directory permissions bugs (602707 and 620007)
33 - journal_stop fixed with locking (611313)
34 - O_APPEND failures resolved (618273, perhaps 614459)
35 - lconf PATH fix (619770)
36 - IA64 build fix (621450)
37 - RPC buffer sizes scale with amount of memory
39 2002-10-01 Phil Schwan <phil@clusterfs.com>
42 - locks would be cancelled without throwing away data pages,
43 resulting in inconsistent data (605627)
44 - inode attributes were not always being refreshed (605627, 612449)
45 - lconf now continues to cleanup after lctl reports an error
46 - MDS now enforces user permissions (602707)
47 - lprocfs cleanup fixed, but not yet enabled (614157)
48 - fixed infinite server hang, should a client not respond to an AST
49 - avoid going into recovery if user calls readlink() with a buffer
50 that's too small (613941)
51 - AST RPCs no longer require replies (614867) -- this may be changed
52 - don't crash server if client sends an IOV that's too big (611336)
53 - fixed lock conversion deadlock (611892)
54 - fixed the following of symlinks (614622)
55 * recovery: the server can remove locks from a client that dies, other
56 clients can make progress
57 * more extN patch fixes
58 * compile-time configurable ptlrpc buffer allocations
60 - collaborative read cache document
61 - Lustre Lite Performance CDR document-in-progress
63 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
66 - fix typo in patch-2.4.18
68 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
71 - clear ptlrpc request each time in handle_incoming_request()
72 - unlink of files now destroys the object on the OST
74 2002-09-19 Peter Braam <braam@clusterfs.com>
76 * add hard link support
77 * change obdfile creation method
78 * kernel patch changed
80 2002-09-19 Peter Braam <braam@clusterfs.com>
83 - stack overflow bug in extN fixed
85 2002-09-18 Andreas Dilger <adilger@clusterfs.com>
87 * documentation updates
88 - add man pages for config tools
89 - update tests/README to describe testing with new config tools
90 - finish metadata API descriptions
91 * bug fixes and cleanups
92 - statfs workaround for 16TB limit
93 - LOV stripe allocation improved, can stripe on subset of OSTs
94 - LOV file size/IO offset was wrong for files > 4GB in size
95 - object EA data was being dropped, caused files to be unreadable
96 - memory overflow with non-LOV OST caused memory corruption
97 - fixed regression tests to work with new config tools, obdfilter
98 - fixed bug when directory size became larger than 1 block
99 - fixed bug (for single client case) when PWD was deleted
100 - invalidate local directory pages when doing intent-based ops
101 - avoid LDLM oops when lock callback contained bad data
103 2002-09-09 Andreas Dilger <adilger@clusterfs.com>
105 * documentation updates
106 * bug fixes and cleanups
107 - configuration tools
110 - 64-bit compile warnings
111 - 64-bit internal statfs data
113 * test_brw on persistent OST devices
117 2002-09-04 Andreas Dilger <adilger@clusterfs.com>
119 * documentation updates
120 * bug fixes and cleanups
121 * configuration tools
123 2002-08-30 Peter J. Braam <braam@clusterfs.com>
126 * many small fixes to 0_5_4
127 * io/network handling
128 * thinkos in MDS operations
130 2002-08-24 Peter J. Braam <braam@clusterfs.com>
133 * crucial basic fixes to 0.5.3
134 * IOR, Iozone work over Elan
137 2002-08-07 Phil Schwan <phil@clusterfs.com>
138 * version 0_5_3, our first alpha
139 * we use the new Portals iovs
140 * documentation updates
141 * bug fixes and cleanups
142 * small changes in the DLM wire protocol
144 2002-07-25 Peter J. Braam <braam@clusterfs.com>
145 * version 0_5_1 with some initial stability,
146 * locking on MD and file I/O.
147 * documentation updates
148 * several bug fixes since 0.5.0
149 * small changes in wire protocol
151 2002-07-18 Phil Schwan <phil@clusterfs.com>
153 * delivered as Lustre Light Alpha
154 * fixed a crash after handling invalid MDS requests
155 * fixed directory pages for architectures with non-4k pages sizes
157 2002-07-11 Andreas Dilger <adilger@clusterfs.com>
158 * release version v0_4_4
159 * Moves TCP acceptor to be on port 2432 (unused Coda port) instead
161 * Fixes a number of interruption problems with OST operations.
162 * Update documentation for portals header changes
163 * Move all wire protocol structs/defines to lustre_idl.h
164 * Fixes symlink length bug.
165 * Add tcpdump to repository.
167 2002-07-05 Andreas Dilger <adilger@clusterfs.com>
168 * release version v0_4_3
169 * Fixes statfs for inodes on extN.
170 * Fixes bug in runtests which would delete /etc/hosts.
171 * Use 64-bit object IDs wherever possible (not into VFS though)
172 Remove ost_get_info, which is unused by lustre, and out of date.
174 2002-07-03 Peter Braam <braam@clusterfs.com>
175 * release version v0_4_2 Fixes a lookup error (type not passed)
176 * move forward to head of Portals
177 * move forward to latest Lustre kernel
179 2002-06-25 Peter Braam <braam@clusterfs.com>
180 * release version v0_4_1. Hopefully stable on single node use.