1 2002-10-28 Andreas Dilger <adilger@clusterfs.com>
4 - limit client IOV size to PTL_MD_MAX_IOV (611336, 191)
5 - defer open object destruction to close time (601981, 138)
6 - open/close OST file handle in obdo (OBD_MD_FLHANDLE) (601981, 138)
7 - move LDLM_ENQUEUE/CONVERT back to MDS portal (625069)
8 - abstract ll_lookup2, fix ll_revalidate2 to use abstraction (256)
9 - don't call obd_setattr in ll_file_release for destroyed objects
10 * protocol change to lustre_msg: move |version| and add |flags|
11 * protocol change to osc_punch: "start" in "o_size", "end" in "o_blocks"
12 * lock replay: for LDLM_FL_REPLAY trust client to do right thing
13 * added replay of create, unlink, link and rename operations during
14 MDS failover; recovery should be much more robust now
15 * remove failed OSCs from LOVs (only lov_create uses this so far)
16 * the lustre-HOWTO was brought (more) up to date (582544)
18 2002-10-23 Phil Schwan <phil@clusterfs.com>
21 - in-use dentries weren't being reused properly (617851)
22 - prevent multiple LDLM setup (599178)
23 - fix LOV size calculations for truncate (617853)
24 - fix client handling of MDS intent errors (POSIX)
25 - fix permission bug in lovstripe.c test (624321)
26 - fix MDS thread deadlock - move LDLM handler to DLM portal (625069)
27 - truncate past end of file could corrupt data
28 - proper cleanup after timeouts, crashes, etc (592524, 550815)
29 - a race in recovery could return ETIMEDOUT to apps (623947)
30 - building outside the source directory was fixed
31 * the lustre-HOWTO was brought (more) up to date (582544)
32 * major progress was made on recovery functionality
34 2002-10-10 Phil Schwan <phil@clusterfs.com>
37 - recovery deadlock fix
38 - rm -rf causes LBUG fix (617817)
39 - file open by multiple tasks fix (618962)
40 - directory permissions bugs (602707 and 620007)
41 - journal_stop fixed with locking (611313)
42 - O_APPEND failures resolved (618273, perhaps 614459)
43 - lconf PATH fix (619770)
44 - IA64 build fix (621450)
45 - RPC buffer sizes scale with amount of memory
47 2002-10-01 Phil Schwan <phil@clusterfs.com>
50 - locks would be cancelled without throwing away data pages,
51 resulting in inconsistent data (605627)
52 - inode attributes were not always being refreshed (605627, 612449)
53 - lconf now continues to cleanup after lctl reports an error
54 - MDS now enforces user permissions (602707)
55 - lprocfs cleanup fixed, but not yet enabled (614157)
56 - fixed infinite server hang, should a client not respond to an AST
57 - avoid going into recovery if user calls readlink() with a buffer
58 that's too small (613941)
59 - AST RPCs no longer require replies (614867) -- this may be changed
60 - don't crash server if client sends an IOV that's too big (611336)
61 - fixed lock conversion deadlock (611892)
62 - fixed the following of symlinks (614622)
63 * recovery: the server can remove locks from a client that dies, other
64 clients can make progress
65 * more extN patch fixes
66 * compile-time configurable ptlrpc buffer allocations
68 - collaborative read cache document
69 - Lustre Lite Performance CDR document-in-progress
71 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
74 - fix typo in patch-2.4.18
76 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
79 - clear ptlrpc request each time in handle_incoming_request()
80 - unlink of files now destroys the object on the OST
82 2002-09-19 Peter Braam <braam@clusterfs.com>
84 * add hard link support
85 * change obdfile creation method
86 * kernel patch changed
88 2002-09-19 Peter Braam <braam@clusterfs.com>
91 - stack overflow bug in extN fixed
93 2002-09-18 Andreas Dilger <adilger@clusterfs.com>
95 * documentation updates
96 - add man pages for config tools
97 - update tests/README to describe testing with new config tools
98 - finish metadata API descriptions
99 * bug fixes and cleanups
100 - statfs workaround for 16TB limit
101 - LOV stripe allocation improved, can stripe on subset of OSTs
102 - LOV file size/IO offset was wrong for files > 4GB in size
103 - object EA data was being dropped, caused files to be unreadable
104 - memory overflow with non-LOV OST caused memory corruption
105 - fixed regression tests to work with new config tools, obdfilter
106 - fixed bug when directory size became larger than 1 block
107 - fixed bug (for single client case) when PWD was deleted
108 - invalidate local directory pages when doing intent-based ops
109 - avoid LDLM oops when lock callback contained bad data
111 2002-09-09 Andreas Dilger <adilger@clusterfs.com>
113 * documentation updates
114 * bug fixes and cleanups
115 - configuration tools
118 - 64-bit compile warnings
119 - 64-bit internal statfs data
121 * test_brw on persistent OST devices
125 2002-09-04 Andreas Dilger <adilger@clusterfs.com>
127 * documentation updates
128 * bug fixes and cleanups
129 * configuration tools
131 2002-08-30 Peter J. Braam <braam@clusterfs.com>
134 * many small fixes to 0_5_4
135 * io/network handling
136 * thinkos in MDS operations
138 2002-08-24 Peter J. Braam <braam@clusterfs.com>
141 * crucial basic fixes to 0.5.3
142 * IOR, Iozone work over Elan
145 2002-08-07 Phil Schwan <phil@clusterfs.com>
146 * version 0_5_3, our first alpha
147 * we use the new Portals iovs
148 * documentation updates
149 * bug fixes and cleanups
150 * small changes in the DLM wire protocol
152 2002-07-25 Peter J. Braam <braam@clusterfs.com>
153 * version 0_5_1 with some initial stability,
154 * locking on MD and file I/O.
155 * documentation updates
156 * several bug fixes since 0.5.0
157 * small changes in wire protocol
159 2002-07-18 Phil Schwan <phil@clusterfs.com>
161 * delivered as Lustre Light Alpha
162 * fixed a crash after handling invalid MDS requests
163 * fixed directory pages for architectures with non-4k pages sizes
165 2002-07-11 Andreas Dilger <adilger@clusterfs.com>
166 * release version v0_4_4
167 * Moves TCP acceptor to be on port 2432 (unused Coda port) instead
169 * Fixes a number of interruption problems with OST operations.
170 * Update documentation for portals header changes
171 * Move all wire protocol structs/defines to lustre_idl.h
172 * Fixes symlink length bug.
173 * Add tcpdump to repository.
175 2002-07-05 Andreas Dilger <adilger@clusterfs.com>
176 * release version v0_4_3
177 * Fixes statfs for inodes on extN.
178 * Fixes bug in runtests which would delete /etc/hosts.
179 * Use 64-bit object IDs wherever possible (not into VFS though)
180 Remove ost_get_info, which is unused by lustre, and out of date.
182 2002-07-03 Peter Braam <braam@clusterfs.com>
183 * release version v0_4_2 Fixes a lookup error (type not passed)
184 * move forward to head of Portals
185 * move forward to latest Lustre kernel
187 2002-06-25 Peter Braam <braam@clusterfs.com>
188 * release version v0_4_1. Hopefully stable on single node use.