3 - fix many simultaneous client startup (392)
5 2002-11-18 Phil Schwan <phil@clusterfs.com>
8 - fix null d_it dereference (346)
9 - fix full OST/dbench hang (333)
10 - fix permission problem with file removal (286)
11 - fix removal of OSCs from LOV when they fail
12 - fix NULL deref during bulk timeout (214)
13 - fix problems related to multiple filesystems on one MDS (241)
14 - fixed serious subtle metadata locking bugs
15 - free locks on clients when inodes are removed due to memory
17 - fix inode pointer in lock data (285)
18 - partial support for multiple MDS on a single host (241)
19 - data locks weren't cancelled at clear_inode time (290, 311)
20 - intent locks could lead to unbounded lock growth (205)
21 - added a maximum lock count, an LRU list, and a flusher
22 - fix multiple rename (365)
23 - properly abstracted the echo client
24 - OSC locked 1 byte too many; fixed
25 - rewrote brw callback code:
26 - fixed recovery bugs related to LOVs (306)
27 - fixed too-many-pages-in-one-write crash (191)
28 - fixed (again) crash in sync_io_timeout (214)
29 - probably fixed callback-related race (385)
31 - Add capability to MDS protocol
32 - LDLM cancellations and callbacks on different portals
34 2002-10-28 Andreas Dilger <adilger@clusterfs.com>
37 - limit client IOV size to PTL_MD_MAX_IOV (611336, 191)
38 - defer open object destruction to close time (601981, 138)
39 - open/close OST file handle in obdo (OBD_MD_FLHANDLE) (601981, 138)
40 - move LDLM_ENQUEUE/CONVERT back to MDS portal (625069)
41 - abstract ll_lookup2, fix ll_revalidate2 to use abstraction (256)
42 - don't call obd_setattr in ll_file_release for destroyed objects
43 * protocol change to lustre_msg: move |version| and add |flags|
44 * protocol change to osc_punch: "start" in "o_size", "end" in "o_blocks"
45 * lock replay: for LDLM_FL_REPLAY trust client to do right thing
46 * added replay of create, unlink, link and rename operations during
47 MDS failover; recovery should be much more robust now
48 * remove failed OSCs from LOVs (only lov_create uses this so far)
49 * the lustre-HOWTO was brought (more) up to date (582544)
51 2002-10-23 Phil Schwan <phil@clusterfs.com>
54 - in-use dentries weren't being reused properly (617851)
55 - prevent multiple LDLM setup (599178)
56 - fix LOV size calculations for truncate (617853)
57 - fix client handling of MDS intent errors (POSIX)
58 - fix permission bug in lovstripe.c test (624321)
59 - fix MDS thread deadlock - move LDLM handler to DLM portal (625069)
60 - truncate past end of file could corrupt data
61 - proper cleanup after timeouts, crashes, etc (592524, 550815)
62 - a race in recovery could return ETIMEDOUT to apps (623947)
63 - building outside the source directory was fixed
64 * the lustre-HOWTO was brought (more) up to date (582544)
65 * major progress was made on recovery functionality
67 2002-10-10 Phil Schwan <phil@clusterfs.com>
70 - recovery deadlock fix
71 - rm -rf causes LBUG fix (617817)
72 - file open by multiple tasks fix (618962)
73 - directory permissions bugs (602707 and 620007)
74 - journal_stop fixed with locking (611313)
75 - O_APPEND failures resolved (618273, perhaps 614459)
76 - lconf PATH fix (619770)
77 - IA64 build fix (621450)
78 - RPC buffer sizes scale with amount of memory
80 2002-10-01 Phil Schwan <phil@clusterfs.com>
83 - locks would be cancelled without throwing away data pages,
84 resulting in inconsistent data (605627)
85 - inode attributes were not always being refreshed (605627, 612449)
86 - lconf now continues to cleanup after lctl reports an error
87 - MDS now enforces user permissions (602707)
88 - lprocfs cleanup fixed, but not yet enabled (614157)
89 - fixed infinite server hang, should a client not respond to an AST
90 - avoid going into recovery if user calls readlink() with a buffer
91 that's too small (613941)
92 - AST RPCs no longer require replies (614867) -- this may be changed
93 - don't crash server if client sends an IOV that's too big (611336)
94 - fixed lock conversion deadlock (611892)
95 - fixed the following of symlinks (614622)
96 * recovery: the server can remove locks from a client that dies, other
97 clients can make progress
98 * more extN patch fixes
99 * compile-time configurable ptlrpc buffer allocations
101 - collaborative read cache document
102 - Lustre Lite Performance CDR document-in-progress
104 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
107 - fix typo in patch-2.4.18
109 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
112 - clear ptlrpc request each time in handle_incoming_request()
113 - unlink of files now destroys the object on the OST
115 2002-09-19 Peter Braam <braam@clusterfs.com>
117 * add hard link support
118 * change obdfile creation method
119 * kernel patch changed
121 2002-09-19 Peter Braam <braam@clusterfs.com>
124 - stack overflow bug in extN fixed
126 2002-09-18 Andreas Dilger <adilger@clusterfs.com>
128 * documentation updates
129 - add man pages for config tools
130 - update tests/README to describe testing with new config tools
131 - finish metadata API descriptions
132 * bug fixes and cleanups
133 - statfs workaround for 16TB limit
134 - LOV stripe allocation improved, can stripe on subset of OSTs
135 - LOV file size/IO offset was wrong for files > 4GB in size
136 - object EA data was being dropped, caused files to be unreadable
137 - memory overflow with non-LOV OST caused memory corruption
138 - fixed regression tests to work with new config tools, obdfilter
139 - fixed bug when directory size became larger than 1 block
140 - fixed bug (for single client case) when PWD was deleted
141 - invalidate local directory pages when doing intent-based ops
142 - avoid LDLM oops when lock callback contained bad data
144 2002-09-09 Andreas Dilger <adilger@clusterfs.com>
146 * documentation updates
147 * bug fixes and cleanups
148 - configuration tools
151 - 64-bit compile warnings
152 - 64-bit internal statfs data
154 * test_brw on persistent OST devices
158 2002-09-04 Andreas Dilger <adilger@clusterfs.com>
160 * documentation updates
161 * bug fixes and cleanups
162 * configuration tools
164 2002-08-30 Peter J. Braam <braam@clusterfs.com>
167 * many small fixes to 0_5_4
168 * io/network handling
169 * thinkos in MDS operations
171 2002-08-24 Peter J. Braam <braam@clusterfs.com>
174 * crucial basic fixes to 0.5.3
175 * IOR, Iozone work over Elan
178 2002-08-07 Phil Schwan <phil@clusterfs.com>
179 * version 0_5_3, our first alpha
180 * we use the new Portals iovs
181 * documentation updates
182 * bug fixes and cleanups
183 * small changes in the DLM wire protocol
185 2002-07-25 Peter J. Braam <braam@clusterfs.com>
186 * version 0_5_1 with some initial stability,
187 * locking on MD and file I/O.
188 * documentation updates
189 * several bug fixes since 0.5.0
190 * small changes in wire protocol
192 2002-07-18 Phil Schwan <phil@clusterfs.com>
194 * delivered as Lustre Light Alpha
195 * fixed a crash after handling invalid MDS requests
196 * fixed directory pages for architectures with non-4k pages sizes
198 2002-07-11 Andreas Dilger <adilger@clusterfs.com>
199 * release version v0_4_4
200 * Moves TCP acceptor to be on port 2432 (unused Coda port) instead
202 * Fixes a number of interruption problems with OST operations.
203 * Update documentation for portals header changes
204 * Move all wire protocol structs/defines to lustre_idl.h
205 * Fixes symlink length bug.
206 * Add tcpdump to repository.
208 2002-07-05 Andreas Dilger <adilger@clusterfs.com>
209 * release version v0_4_3
210 * Fixes statfs for inodes on extN.
211 * Fixes bug in runtests which would delete /etc/hosts.
212 * Use 64-bit object IDs wherever possible (not into VFS though)
213 Remove ost_get_info, which is unused by lustre, and out of date.
215 2002-07-03 Peter Braam <braam@clusterfs.com>
216 * release version v0_4_2 Fixes a lookup error (type not passed)
217 * move forward to head of Portals
218 * move forward to latest Lustre kernel
220 2002-06-25 Peter Braam <braam@clusterfs.com>
221 * release version v0_4_1. Hopefully stable on single node use.