3 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
4 for currently supported client and server kernel versions.
5 * Server known to build on patched kernels:
6 2.6.32-431.29.2.el6 (RHEL6.5)
7 2.6.32-504.30.3.el6 (RHEL6.6)
8 2.6.32-573.26.1.el6 (RHEL6.7)
9 2.6.32-642.15.1.el6 (RHEL6.8)
10 2.6.32-696.30.1.el6 (RHEL6.9)
11 3.10.0-862.14.4.el7 (RHEL7.5)
12 3.10.0-957.el7 (RHEL7.6)
13 3.0.101-0.47.71 (SLES11 SP3)
14 3.0.101-107 (SLES11 SP4)
15 3.12.74-60.64.40 (SLES12 SP1)
16 4.4.120-92.70 (SLES12 SP2)
17 4.4.155-94.50 (SLES12 SP3)
18 vanilla linux 4.6.7 (ZFS only)
19 * Client known to build on unpatched kernels:
20 2.6.32-431.29.2.el6 (RHEL6.5)
21 2.6.32-504.30.3.el6 (RHEL6.6)
22 2.6.32-573.26.1.el6 (RHEL6.7)
23 2.6.32-642.15.1.el6 (RHEL6.8)
24 2.6.32-696.30.1.el6 (RHEL6.9)
25 2.6.32-754.6.3.el6 (RHEL6.10)
26 3.10.0-862.14.4.el7 (RHEL7.5)
27 3.10.0-957.el7 (RHEL7.6)
28 3.0.101-0.47.71 (SLES11 SP3)
29 3.0.101-107 (SLES11 SP4)
30 3.12.74-60.64.40 (SLES12 SP1)
31 4.4.120-92.70 (SLES12 SP2)
32 4.4.155-94.50 (SLES12 SP3)
33 4.4.0-131 (Ubuntu 16.04)
35 * Recommended e2fsprogs version: 1.42.13.wc6 or newer
36 * Recommended ZFS / SPL version: 0.7.9
37 * Tested with ZFS / SPL version: 0.7.9
38 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
39 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
40 stack overflow. For more information, please refer to bugzilla 17630.
41 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
42 the re-exporting nodes to be mounted with "32bitapi" mount option
46 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
47 for currently supported client and server kernel versions.
48 * Server known to build on patched kernels:
49 2.6.32-431.29.2.el6 (RHEL6.5)
50 2.6.32-504.30.3.el6 (RHEL6.6)
51 2.6.32-573.26.1.el6 (RHEL6.7)
52 2.6.32-642.15.1.el6 (RHEL6.8)
53 2.6.32-696.30.1.el6 (RHEL6.9)
54 3.10.0-862.3.2.el7 (RHEL7.5)
55 3.0.101-0.47.71 (SLES11 SP3)
56 3.0.101-107 (SLES11 SP4)
57 3.12.74-60.64.40 (SLES12 SP1)
58 4.4.120-92.70 (SLES12 SP2)
59 4.4.155-94.50 (SLES12 SP3)
60 vanilla linux 4.6.7 (ZFS only)
61 * Client known to build on unpatched kernels:
62 2.6.32-431.29.2.el6 (RHEL6.5)
63 2.6.32-504.30.3.el6 (RHEL6.6)
64 2.6.32-573.26.1.el6 (RHEL6.7)
65 2.6.32-642.15.1.el6 (RHEL6.8)
66 2.6.32-696.30.1.el6 (RHEL6.9)
67 2.6.32-754.6.3.el6 (RHEL6.10)
68 3.10.0-862.3.2.el7 (RHEL7.5)
69 3.0.101-0.47.71 (SLES11 SP3)
70 3.0.101-107 (SLES11 SP4)
71 3.12.74-60.64.40 (SLES12 SP1)
72 4.4.120-92.70 (SLES12 SP2)
73 4.4.155-94.50 (SLES12 SP3)
74 4.4.0-131 (Ubuntu 16.04)
76 * Recommended e2fsprogs version: 1.42.13.wc6 or newer
77 * Recommended ZFS / SPL version: 0.7.9
78 * Tested with ZFS / SPL version: 0.7.9
79 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
80 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
81 stack overflow. For more information, please refer to bugzilla 17630.
82 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
83 the re-exporting nodes to be mounted with "32bitapi" mount option
85 05-22-2018 Intel Corporation
87 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
88 for currently supported client and server kernel versions.
89 * Server known to build on patched kernels:
90 2.6.32-431.29.2.el6 (RHEL6.5)
91 2.6.32-504.30.3.el6 (RHEL6.6)
92 2.6.32-573.26.1.el6 (RHEL6.7)
93 2.6.32-642.15.1.el6 (RHEL6.8)
94 2.6.32-696.30.1.el6 (RHEL6.9)
95 3.10.0-862.9.1.el7 (RHEL7.5)
96 3.0.101-0.47.71 (SLES11 SP3)
97 3.0.101-107 (SLES11 SP4)
98 3.12.74-60.64.40 (SLES12 SP1)
99 4.4.120-92.70 (SLES12 SP2)
100 4.4.132-94.33 (SLES12 SP3)
101 vanilla linux 4.6.7 (ZFS only)
102 * Client known to build on unpatched kernels:
103 2.6.32-431.29.2.el6 (RHEL6.5)
104 2.6.32-504.30.3.el6 (RHEL6.6)
105 2.6.32-573.26.1.el6 (RHEL6.7)
106 2.6.32-642.15.1.el6 (RHEL6.8)
107 2.6.32-696.30.1.el6 (RHEL6.9)
108 3.10.0-862.9.1.el7 (RHEL7.5)
109 3.0.101-0.47.71 (SLES11 SP3)
110 3.0.101-107 (SLES11 SP4)
111 3.12.74-60.64.40 (SLES12 SP1)
112 4.4.120-92.70 (SLES12 SP2)
113 4.4.133-94.33 (SLES12 SP3)
114 4.4.0-116 (Ubuntu 16.04)
116 * Recommended e2fsprogs version: 1.42.13.wc6 or newer
117 * Recommended ZFS / SPL version: 0.7.9
118 * Tested with ZFS / SPL version: 0.7.9
119 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
120 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
121 stack overflow. For more information, please refer to bugzilla 17630.
122 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
123 the re-exporting nodes to be mounted with "32bitapi" mount option
125 01-31-2018 Intel Corporation
127 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
128 for currently supported client and server kernel versions.
129 * Server known to build on patched kernels:
130 2.6.32-431.29.2.el6 (RHEL6.5)
131 2.6.32-504.30.3.el6 (RHEL6.6)
132 2.6.32-573.26.1.el6 (RHEL6.7)
133 2.6.32-642.15.1.el6 (RHEL6.8)
134 2.6.32-696.23.1.el6 (RHEL6.9)
135 3.10.0-693.21.1.el7 (RHEL7.4)
136 3.0.101-0.47.71 (SLES11 SP3)
137 3.0.101-107 (SLES11 SP4)
138 3.12.74-60.64.40 (SLES12 SP1)
139 4.4.120-92.70 (SLES12 SP2)
140 4.4.120-94.17 (SLES12 SP3)
141 vanilla linux 4.6.7 (ZFS only)
142 * Client known to build on unpatched kernels:
143 2.6.32-431.29.2.el6 (RHEL6.5)
144 2.6.32-504.30.3.el6 (RHEL6.6)
145 2.6.32-573.26.1.el6 (RHEL6.7)
146 2.6.32-642.15.1.el6 (RHEL6.8)
147 2.6.32-696.23.1.el6 (RHEL6.9)
148 3.10.0-693.21.1.el7 (RHEL7.4)
149 3.0.101-0.47.71 (SLES11 SP3)
150 3.0.101-107 (SLES11 SP4)
151 3.12.74-60.64.40 (SLES12 SP1)
152 4.4.120-92.70 (SLES12 SP2)
153 4.4.120-94.17 (SLES12 SP3)
154 4.4.0-116 (Ubuntu 16.04)
156 * Recommended e2fsprogs version: 1.42.13.wc6 or newer
157 * Recommended ZFS / SPL version: 0.7.5
158 * Tested with ZFS / SPL version: 0.7.5
159 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
160 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
161 stack overflow. For more information, please refer to bugzilla 17630.
162 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
163 the re-exporting nodes to be mounted with "32bitapi" mount option
165 --------------------------------------------------------------------------------
167 12-11-2017 Intel Corporation
169 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
170 for currently supported client and server kernel versions.
171 * Server known to build on patched kernels:
172 2.6.32-431.29.2.el6 (RHEL6.5)
173 2.6.32-504.30.3.el6 (RHEL6.6)
174 2.6.32-573.26.1.el6 (RHEL6.7)
175 2.6.32-642.15.1.el6 (RHEL6.8)
176 2.6.32-696.20.1.el6 (RHEL6.9)
177 3.10.0-693.17.1.el7 (RHEL7.4)
178 3.0.101-0.47.71 (SLES11 SP3)
179 3.0.101-107 (SLES11 SP4)
180 3.12.74-60.64.40 (SLES12 SP1)
181 4.4.103-92.56 (SLES12 SP2)
182 4.4.103-6.38 (SLES12 SP3)
183 vanilla linux 4.6.7 (ZFS only)
184 * Client known to build on unpatched kernels:
185 2.6.32-431.29.2.el6 (RHEL6.5)
186 2.6.32-504.30.3.el6 (RHEL6.6)
187 2.6.32-573.26.1.el6 (RHEL6.7)
188 2.6.32-642.15.1.el6 (RHEL6.8)
189 2.6.32-696.20.1.el6 (RHEL6.9)
190 3.10.0-693.17.1.el7 (RHEL7.4)
191 3.0.101-0.47.71 (SLES11 SP3)
192 3.0.101-107 (SLES11 SP4)
193 3.12.74-60.64.40 (SLES12 SP1)
194 4.4.103-92.56 (SLES12 SP2)
195 4.4.103-6.38 (SLES12 SP3)
197 * Recommended e2fsprogs version: 1.42.13.wc6 or newer
198 * Recommended ZFS / SPL version: 0.7.3
199 * Tested with ZFS / SPL version: 0.7.3
200 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
201 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
202 stack overflow. For more information, please refer to bugzilla 17630.
203 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
204 the re-exporting nodes to be mounted with "32bitapi" mount option
206 --------------------------------------------------------------------------------
208 10-02-2017 Intel Corporation
210 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
211 for currently supported client and server kernel versions.
212 * Server known to build on patched kernels:
213 2.6.32-431.29.2.el6 (RHEL6.5)
214 2.6.32-504.30.3.el6 (RHEL6.6)
215 2.6.32-573.26.1.el6 (RHEL6.7)
216 2.6.32-642.15.1.el6 (RHEL6.8)
217 2.6.32-696.10.3.el6 (RHEL6.9)
218 3.10.0-693.2.2.el7 (RHEL7.4)
219 3.0.101-0.47.71 (SLES11 SP3)
220 3.0.101-107 (SLES11 SP4)
221 3.12.74-60.64.40 (SLES12 SP1)
222 4.4.74-92.35 (SLES12 SP2)
223 vanilla linux 4.6.7 (ZFS only)
224 * Client known to build on unpatched kernels:
225 2.6.32-431.29.2.el6 (RHEL6.5)
226 2.6.32-504.30.3.el6 (RHEL6.6)
227 2.6.32-573.26.1.el6 (RHEL6.7)
228 2.6.32-642.15.1.el6 (RHEL6.8)
229 2.6.32-696.10.3.el6 (RHEL6.9)
230 3.10.0-693.2.2.el7 (RHEL7.4)
231 3.0.101-0.47.71 (SLES11 SP3)
232 3.0.101-107 (SLES11 SP4)
233 3.12.74-60.64.40 (SLES12 SP1)
234 4.4.74-92.35 (SLES12 SP2)
236 * Recommended e2fsprogs version: 1.42.13.wc4 or newer
237 * Recommended ZFS / SPL version: 0.7.1
238 * Tested with ZFS / SPL version: 0.7.1
239 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
240 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
241 stack overflow. For more information, please refer to bugzilla 17630.
242 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
243 the re-exporting nodes to be mounted with "32bitapi" mount option
245 --------------------------------------------------------------------------------
247 2017-07-05 Intel Corporation
249 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
250 for currently supported client and server kernel versions.
251 * Server known to build on patched kernels:
252 2.6.32-431.29.2.el6 (RHEL6.5)
253 2.6.32-504.30.3.el6 (RHEL6.6)
254 2.6.32-573.26.1.el6 (RHEL6.7)
255 2.6.32-642.15.1.el6 (RHEL6.8)
256 2.6.32-696.3.1.el6 (RHEL6.9)
257 3.10.0-514.21.1.el7 (RHEL7.3)
258 3.0.101-0.47.71 (SLES11 SP3)
259 3.0.101-100 (SLES11 SP4)
260 3.12.74-60.64.40 (SLES12 SP1)
261 4.4.59-92.17 (SLES12 SP2)
262 vanilla linux 4.6.7 (ZFS only)
263 * Client known to build on unpatched kernels:
264 2.6.32-431.29.2.el6 (RHEL6.5)
265 2.6.32-504.30.3.el6 (RHEL6.6)
266 2.6.32-573.26.1.el6 (RHEL6.7)
267 2.6.32-642.15.1.el6 (RHEL6.8)
268 2.6.32-696.3.1.el6 (RHEL6.9)
269 3.10.0-514.21.1.el7 (RHEL7.3)
270 3.0.101-0.47.71 (SLES11 SP3)
271 3.0.101-100 (SLES11 SP4)
272 3.12.74-60.64.40 (SLES12 SP1)
273 4.4.59-92.17 (SLES12 SP2)
275 * Recommended e2fsprogs version: 1.42.13.wc4 or newer
276 * Recommended ZFS / SPL version: 0.6.5.9
277 * Tested with ZFS / SPL version: 0.6.5.9
278 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
279 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
280 stack overflow. For more information, please refer to bugzilla 17630.
281 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
282 the re-exporting nodes to be mounted with "32bitapi" mount option
284 --------------------------------------------------------------------------------
286 12-07-2016 Intel Corporation
288 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
289 for currently supported client and server kernel versions.
290 * Server known to build on patched kernels:
291 2.6.32-431.29.2.el6 (RHEL6.5)
292 2.6.32-504.30.3.el6 (RHEL6.6)
293 2.6.32-573.26.1.el6 (RHEL6.7)
294 2.6.32-642.11.1.el6 (RHEL6.8)
295 3.10.0-514.el7 (RHEL7.3)
296 3.0.101-0.47.71 (SLES11 SP3)
297 3.0.101-80 (SLES11 SP4)
298 3.12.62-60.64.8 (SLES12 SP1)
299 vanilla linux 4.5.7 (ZFS only)
300 * Client known to build on unpatched kernels:
301 2.6.32-431.29.2.el6 (RHEL6.5)
302 2.6.32-504.30.3.el6 (RHEL6.6)
303 2.6.32-573.26.1.el6 (RHEL6.7)
304 2.6.32-642.11.1.el6 (RHEL6.8)
305 3.10.0-514.el7 (RHEL7.3)
306 3.0.101-0.47.71 (SLES11 SP3)
307 3.0.101-80 (SLES11 SP4)
308 3.12.62-60.64.8 (SLES12 SP1)
310 * Recommended e2fsprogs version: 1.42.13.wc5 or newer
311 * Recommended ZFS / SPL version: 0.6.5.8
312 * Tested with ZFS / SPL version: 0.6.5.8
313 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
314 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
315 stack overflow. For more information, please refer to bugzilla 17630.
316 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
317 the re-exporting nodes to be mounted with "32bitapi" mount option
319 --------------------------------------------------------------------------------
321 02-29-2016 Intel Corporation
323 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
324 for currently supported client and server kernel versions.
325 * Server known to build on patched kernels:
326 2.6.32-431.29.2.el6 (RHEL6.5)
327 2.6.32-504.30.3.el6 (RHEL6.6)
328 2.6.32-573.12.1.el6 (RHEL6.7)
329 3.10.0-327.3.1.el7 (RHEL7.2)
330 3.0.101-0.47.71 (SLES11 SP3)
331 3.0.101-68 (SLES11 SP4)
332 vanilla linux 4.2.1 (ZFS only)
333 * Client known to build on unpatched kernels:
334 2.6.32-431.29.2.el6 (RHEL6.5)
335 2.6.32-504.30.3.el6 (RHEL6.6)
336 2.6.32-573.12.1.el6 (RHEL6.7)
337 3.10.0-327.3.1.el7 (RHEL7.2)
338 3.0.101-0.47.71 (SLES11 SP3)
339 3.0.101-68 (SLES11 SP4)
342 * Recommended e2fsprogs version: 1.42.13.wc4 or newer
343 * Recommended ZFS / SPL version: 0.6.4.2
344 * Tested with ZFS / SPL version: 0.6.4.2
345 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
346 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
347 stack overflow. For more information, please refer to bugzilla 17630.
348 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
349 the re-exporting nodes to be mounted with "32bitapi" mount option
351 --------------------------------------------------------------------------------
353 03-10-2015 Intel Corporation
355 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
356 for currently supported client and server kernel versions.
357 * Server known to build on patched kernels:
358 2.6.32-431.29.2.el6 (RHEL6.5)
359 2.6.32-504.8.1.el6 (RHEL6.6)
360 3.0.101-0.46 (SLES11 SP3)
361 * Client known to build on unpatched kernels:
362 2.6.32-431.29.2.el6 (RHEL6.5)
363 2.6.32-504.8.1.el6 (RHEL6.6)
364 3.10.0-123.20.1.el7 (RHEL7)
365 3.0.101-0.46 (SLES11 SP3)
366 * Recommended e2fsprogs version: 1.42.9.wc1 or newer
367 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
368 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
369 stack overflow. For more information, please refer to bugzilla 17630.
370 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
371 the re-exporting nodes to be mounted with "32bitapi" mount option
373 Severity : enhancement
375 Description: control OST-index in IDIF via ROCOMPAT flag.
376 Details : Introduce new flag OBD_ROCOMPAT_IDX_IN_IDIF that is stored in the
377 last_rcvd file. For new formatted OST device, it will be auto set;
378 for the case of upgrading from old OST device, you can enable it
379 via the lproc interface osd-ldiskfs.index_in_idif. With such flag
380 enabled, for new created OST-object, its IDIF-in-LMA will contain
381 the OST-index; for the existing OST-object, the OSD will convert
382 old format IDIF as new format IDIF with OST-index stored in the
383 LMA EA when accessing such OST-object or via OI scrub. Once such
384 flag is enabled, it cannot be reverted back, so the system cannot
385 be downgraded to the orignal incompatible version.
387 --------------------------------------------------------------------------------
389 07-30-2014 Intel Corporation
391 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
392 for currently supported client and server kernel versions.
393 * Server known to build on patched kernels:
394 2.6.32-431.20.3.el6 (RHEL6.5)
395 3.0.101-0.31 (SLES11 SP3)
396 * Client known to build on unpatched kernels:
397 2.6.32-431.20.3.el6 (RHEL6.5)
398 3.0.101-0.31 (SLES11 SP3)
401 * Recommended e2fsprogs version: 1.42.9.wc1 or newer
402 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
403 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
404 stack overflow. For more information, please refer to bugzilla 17630.
405 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
406 the re-exporting nodes to be mounted with "32bitapi" mount option
408 --------------------------------------------------------------------------------
410 10-31-2013 Intel Corporation
412 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
413 * Server builds for kernels:
414 2.6.32-358.18.1.el6 (RHEL6.4)
415 * Client builds for unpatched kernels:
416 2.6.32-358.18.1.el6 (RHEL6.4)
417 3.0.92-0.8 (SLES11 SP3)
419 * Recommended e2fsprogs version: 1.42.7.wc2 or newer
420 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
421 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
422 stack overflow. For more information, please refer to bugzilla 17630.
423 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
424 the re-exporting nodes to be mounted with "32bitapi" mount option
426 --------------------------------------------------------------------------------
428 05-30-2013 Intel Corporation
430 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
431 * Server builds for kernels:
432 2.6.32-358.6.2.el6 (RHEL6.4)
433 * Client builds for unpatched kernels:
434 2.6.32-358.6.2.el6 (RHEL6.4)
435 3.0.74-0.6.6 (SLES11 SP2)
437 * Recommended e2fsprogs version: 1.42.7.wc1 or newer
438 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
439 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
440 stack overflow. For more information, please refer to bugzilla 17630.
441 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
442 the re-exporting nodes to be mounted with "32bitapi" mount option
444 --------------------------------------------------------------------------------
446 9-30-2012 Whamcloud, Inc.
448 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
449 * Server support for kernels:
450 2.6.32-279.5.1.el6 (RHEL6)
451 * Client support for unpatched kernels:
452 2.6.32-279.5.1.el6 (RHEL6)
454 2.6.18-238.19.1.el5 (RHEL5)
455 2.6.16.60-0.69.1 (SLES10)
456 * Recommended e2fsprogs version: 1.41.90.wc4
457 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
458 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
459 stack overflow. For more information, please refer to bugzilla 17630.
460 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
461 the re-exporting nodes to be mounted with "32bitapi" mount option
463 Severity : enhancement
465 Description: Update RHEL6.3 kernel to 2.6.32-279.5.1.el6.
467 Severity : enhancement
469 Description: Update RHEL6.3 kernel to 2.6.32-279.2.1.el6.
471 Severity : enhancement
473 Description: Update RHEL6.2 kernel to 2.6.32-220.17.1.el6
475 Severity : enhancement
477 Description: Update RHEL6.2 kernel to 2.6.32-220.13.1.el6
479 Severity : enhancement
481 Description: Update RHEL6.2 kernel to 2.6.32-220.7.1.el6
483 --------------------------------------------------------------------------------
485 04-30-2012 Whamcloud, Inc.
487 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
488 * Server support for kernels:
489 2.6.32-220.4.2.el6 (RHEL6)
490 * Client support for unpatched kernels:
491 2.6.32-220.4.2.el6 (RHEL6)
493 2.6.18-238.19.1.el5 (RHEL5)
494 2.6.16.60-0.69.1 (SLES10)
495 * Recommended e2fsprogs version: 1.41.90.wc4
496 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
497 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
498 stack overflow. For more information, please refer to bugzilla 17630.
499 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
500 the re-exporting nodes to be mounted with "32bitapi" mount option
502 Severity : enhancement
504 Description: Update RHEL6.2 kernel to 2.6.32-220.4.2.el6
506 Severity : enhancement
508 Description: Update RHEL6.2 kernel to 2.6.32-220.4.1.el6
510 Severity : enhancement
512 Description: Update RHEL6.2 kernel to 2.6.32-220.el6
514 Severity : enhancement
516 Description: Update RHEL6.1 kernel to 2.6.32-131.17.1.el6
518 --------------------------------------------------------------------------------
520 09-30-2011 Whamcloud, Inc.
522 * Server support for kernels:
523 2.6.18-238.19.1.el5 (RHEL5)
524 2.6.32-131.6.1.el6 (RHEL6)
525 * Client support for unpatched kernels:
526 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
527 2.6.18-238.19.1.el5 (RHEL5)
528 2.6.32-131.6.1.el6 (RHEL6)
529 2.6.29.4-167.fc11 (FC11)
530 2.6.16.60-0.69.1 (SLES10)
532 * Recommended e2fsprogs version: 1.41.90.wc3
533 * NFS export disabled when stack size < 8192. Since the NFSv4 export of
534 Lustre filesystem with 4K stack may cause a stack overflow. For more
535 information, please refer to bugzilla 17630.
536 * NFSv4 reexport requires Lustre client on the re-exporting nodes to be
537 mounted with "32bitapi" mount option
539 Severity : enhancement
540 Description: Update RHEL6.1 kernel to 2.6.32-131.6.1.el6
542 Severity : enhancement
543 Description: Update RHEL5.6 kernel to 2.6.18-238.19.1.el5
545 Severity : enhancement
546 Description: Update RHEL5.6 kernel to 2.6.18-238.9.1.el5
549 Frequency : only with SLES10
551 Description: Use OFED "KMP" provided by Novell
552 Details : SLES10 SP3 ships with OFED in a separate "KMP" package.
553 Lustre is now built against this package. That means you need to
554 install the ofed-kmp package from Novell for the patchless client.
555 Note that the ofed-kmp that Novell ships may not exactly match the
556 kernel version but should still be compatible.
558 Severity : enhancement
560 Description: Update RHEL5.5 kernel to 2.6.18-194.17.1.el5
561 Update OEL5.5 kernel to 2.6.18-194.17.1.0.1.el5.
563 --------------------------------------------------------------------------------
565 2010-07-15 Oracle, Inc.
567 * Server support for kernels:
568 2.6.18-164.11.1.el5 (RHEL 5)
569 2.6.18-164.11.1.0.1.el5 (OEL 5)
570 * Client support for unpatched kernels:
571 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
572 2.6.18-164.11.1.el5 (RHEL 5),
573 2.6.18-164.11.1.0.1.el5 (OEL 5)
574 2.6.16.60-0.42.8 (SLES 10),
576 2.6.29.4-167.fc11 (FC11)
577 * Recommended e2fsprogs version: 1.41.10-sun2
579 Severity : enhancement
581 Description: Distribute lfs_migrate script with lustre
582 Details : The lfs_migrate mentioned in the manual is now included in the
583 lustre tree in lustre/scripts and will distributed in lustre RPMs.
585 Severity : enhancement
587 Description: Failover nid list added to import proc file.
589 Severity : enhancement
591 Description: "som_preview" mount option required to use SOM
592 Details : The clients will need to use som_preview mount option
596 Description: Lustre does not do 1MB IOs to HW RAID
597 Details : Bump MAX_PHYS/HW_SEGMENTS and SG_ALL to 256 in the RHEL5 kernel.
599 Severity : enhancement
601 Description: lfs quota output is not very convenient for awk/sed-parsing
602 Details : Some positions in lfs quota output table could be empty or
603 non-empty which made it hard to parse it with scripts, now
604 a dash is put instead of space where there is not supposed
609 Description: Update to OFED 1.5.1
610 Details : OFED 1.5.1 will be used in environments that don't use
611 vendor supplied OFED.
615 Description: Remove lnet. prefix from lctl params display
616 Details : Also renamed lnet.memused to lnet_memused to
621 Description: Use conf_param -d to remove permanant settings.
625 Description: Add list_param -R
626 Details : lctl list_param now supports recursive mode.
629 Bugzilla : 21829, 22312
630 Description: Fix llobdstat.
631 Details : 1. Now updating the write_bytes counter
632 2. Port script fixes from 1.8.
636 Description: decrease the usage of memory on clients.
637 Details : 1. On clients, recycle dentries and inodes unused.
638 2. Delete the code related to ll_deathrow(att 6215 in bug 1443). It
643 Description: llobdstat fix and enhancement
644 Details : add a counter to set a limit to how many samples will be returned
645 fix a wildcard in the path to limit to obdfilter stats only
649 Description: lfs check only allowed for root
650 Details : Code cleanups and added a regresion test of non-root use of
655 Description: MDS Oopses in ptlrpc_dispatch_difficult_reply()/__list_add()
656 Details : Fix hr->hr_index overload.
660 Description: Fix FIEMAP functionality.
664 Description: FC11 patchless client support added
668 Description: Update RHEL5.4 kernel to 2.6.18-164.11.1.el5 and
669 OEL5.4 kernel to 2.6.18-164.11.1.0.1.el5.
673 Description: New acc-sm script and various test-framework enhancments.
677 Description: Previously landed patch was reverted.
681 Description: Update SLES10 kernel to 2.6.16.60-0.42.8 (patchless)
685 Description: Add ldiskfs maxdirsize mount option.
689 Description: Fix mkfs.lustre support for 16TB LUNs.
693 Description: Print mmp_check_interval
694 Details : print mmp_check_interval and make it possible to abort
695 mount operation in case it takes too long.
699 Description: Add cascading_rw to lustre mpi tests.
703 Description: Store SOM data in EA.
704 Details : Also several SOM related cleanups.
708 Description: Add refcount to md_open_data
712 Description: Use refcount on lov_request_set to prevent memory corruption.
716 Description: Statahead: use parent dir's i_mutex for synchronization.
720 Description: loadgen improvements.
724 Description: Enable peer health detection in ksocklnd and ko2iblnd.
728 Description: Control DCACHE_LUSTRE_INVALID flag with MDS_INODE_LOOKCUP lock
732 Description: Fix race on obd_devcie:md_stats
736 Description: Fix SLES 11 client builds.
737 Details : Resolved uninitalized pointer gcc warnings.
741 Description: Allow non-root access for "lfs check".
742 Details : Added a check in obd_class_ioctl() for OBD_IOC_PING_TARGET.
746 Description: Clear imp_force_reconnect correctly in ptlrpc_connect_interpret()
748 Severity : enhancement
750 Description: Update kernel to RHEL5.4 2.6.18-164.6.1.el5 and
751 OEL5 2.6.18-164.6.1.0.1.el5(Both in-kernel OFED enabled).
755 Description: Sanity test_27p fails.
756 Details : OBD_FAIL_OST_ENOSPC has been moved to filter_statfs().
757 Sanity tests 27 were cleaned up and improved.
761 Description: Unify req format on client/servers
762 Details : Use new req_capsule API [almost] everywhere instead of old PTLRPC
763 buffers and swabbers approach..
768 Description: drop join file stuff.
770 Severity : enhancement
772 Description: Add 2.6.27 and SLES11 (patchless client) support
776 Description: Fix in ptlrpc_expire_one_request() to print the signed time
779 Severity : enhancement
781 Description: Build kernels (RHEL5, OEL5 and SLES10/11) using the vendor's own
784 Severity : enhancement
786 Description: Remove set_info(KEY_UNLINKED) from MDS/OSC
788 Severity : enhancement
790 Description: correctly handle big reply message.
791 Details : send LNet event if reply is bigger than buffer and adjust this buffer
796 Description: Drop unnecessary __GFP_NOMEMALLOC flag from filter_get_page()
798 Severity : enhancement
800 Description: Add support for OEL5.
804 Description: Conf-sanity.sh 50g test - deactivated OST should not cause a panic.
806 Severity : enhancement
808 Description: "weak-modules" support
809 Details : Implement "weak-modules" support which enables kernel modules
810 to be used with any kernel that implements the same kABI. In
811 order to achieve this modules are now installed in
812 /lib/modules/$(uname -r)/updates/kernel on all distributions.
816 Description: client could not reconnect to OST because of an active request.
817 Details : The client now retries to reconnect to the same server, if a
818 connect request failed with EBUSY or -EAGAIN. Test case 35b
819 conf-sanity.sh has been created.
823 Description: Don't allow make backward step on assignin osc next id.
824 Details : race between allocation next id and ll_sync thread can be cause
825 of set wrong osc next id and can be kill valid ost objects.
829 Description: Changes in raid5-large-io-rhel5.patch to calculate sectors properly
833 Description: Increase the default BLK_DEF_MAX_SECTORS value for RHEL5 and SLES11
837 Description: Error handling in osc_statfs_interpret() has been improved.
838 Details : Check in osc_statfs_interpret() for EBADR.
842 Description: Do not send statfs() requests to OSTs disabled by administrator.
843 Details : Check in lov_prep_statfs_set() for non-NULL ltd_exp.
845 Severity : enhancement
847 Description: Added sanity-quota test to for lfs(1) to work with numeric uid/gid.
849 Severity : enhancement
851 Description: Update kernel to RHEL5 2.6.18-128.7.1.el5.
853 Severity : enhancement
855 Description: Enhancement for lfs(1) command to use numeric uid/gid.
859 Description: Improved cleanup procedure for conf-sanity test_46a.
863 Description: Do not start bulk transfer, if client tried to reconnect.
864 Details : Do not start bulk transfer, if client tried to reconnect,
865 but failed due to active rpcs. Check for exp_abort_active_req
866 has been implemented.
870 Description: Prevent inconsistences between linux and lustre mount structures.
871 Details : Wait indefinitely in server_wait_finished() until mnt_count drops.
872 Make the sleep interruptible.
874 Severity : enhancement
876 Description: provide server to client comms path
877 Details : server to client communications path via new LDLM_SET_INFO rpc
881 Description: truncate starts GFP_FS allocation under transaction causing deadlock
882 Details : ldiskfs_truncate calls grab_cache_page which may start page
883 allocation under an open transaction. This may lead to
884 calling prune_icache with consequent lustre reentrance.
888 Description: send correctly lsm on open replay
889 Details : MDS is trust to LSM size on replay open, but client can set wrong size
894 Description: enable client interface failover
895 Details : When a child reconnects from another NID, properly update export
896 nid hash position and ldlm reverse import.
898 Severity : enhancement
900 Description: Communicate OST degraded/readonly state via statfs to MDS
901 Details : Flags in the statfs returned from OSTs indicate whether the
902 OST is in a degraded RAID state, or if the filesystem has
903 turned read-only after a filesystem error is detected.
908 Description: don't panic if EPROTO was hit when reading symlink
909 Details : correctly handling request reference in error cases.
911 Severity : enhancement
913 Description: Add LustreNetLink, a kernel-userspace communcation path.
915 Severity : enhancement
917 Description: Update kernel to SLES10 SP2 2.6.16.60-0.39.3.
922 Description: access to llog context before init.
923 Details : move handling CATALOGS file at osc layer and forbid access to llog
927 Frequency : rare, only when client is writing to OST on local node
929 Description: Avoid deadlock for local client writes
930 Details : Use new OBD_BRW_MEMALLOC flag to notify OST about writes in the
931 memory freeing context. This allows OST threads to set the
932 PF_MEMALLOC flag on task structures in order to allocate memory
933 from reserved pools and complete IO.
934 Use GFP_HIGHUSER for OST allocations for non-local client writes,
935 so that the OST threads generate memory pressure and allow
936 inactive pages to be reclaimed.
938 Severity : enhancement
940 Description: Update kernel to RHEL5.3 2.6.18-128.1.14.el5.
945 Description: lock ordering violation between &cli->cl_sem and _lprocfs_lock
946 Details : move ldlm namespace creation in setup phase to avoid grab
947 _lprocfs_lock with cli_sem held.
951 Description: Temporarily disable grant shrink.
952 Details : Disable the feature for debugging.
955 Frequency : only when formatting test filesystems
957 Description: Unable to run several mkfs.lustre on loop devices at the same time
958 Details : mkfs.lustre returns error 256 on the concurrent loop devices
959 formatting. The solution is to proper handle the error.
961 Severity : enhancement
963 Description: Update kernel to RHEL5.3 2.6.18-128.1.6.el5.
965 Severity : enhancement
967 Description: Update kernel to SLES10 SP2 2.6.16.60-0.37.
972 Description: resolve race between obd_disconnect and class_disconnect_exports
973 Details : if obd_disconnect will be called to already disconnected export he
974 forget release one reference and osc module can't unloaded.
976 Severity : enhancement
978 Description: Allow tuning service thread via /proc
979 Details : For each service a new
980 /proc/fs/lustre/{service}/*/thread_{min,max,started} entry is
981 created that can be used to set min/max thread counts, and get the
982 current number of running threads.
987 Descriptoin: don't return error if have partially created objects for file.
988 Details : lov_update_create_set uses set->set_success as index for created
989 objects, so if some requests will be failed, they will have hole
990 at end of array and qos_shrink_lsm for allocate correct lsm.
992 Severity : enhancement
994 Description: Update OFED support to 1.4.1
999 Description: Reduce small size read RPC
1000 Details : Set read-ahead limit for every file and only do read-ahead when
1001 available read-ahead pages are bigger than 1M to avoid small size
1004 Severity : enhancement
1006 Description: move AT tunable parameters for more consistent usage
1007 Details : add AT tunables under /proc/sys/lustre, add to conf_param parsing
1009 Severity : enhancement
1011 Description: add lazystatfs mount option to allow statfs(2) to skip down OSTs
1012 Details : allow skip disconnected ost for send statfs request and hide error
1018 Description: fix racy locking of mballoc block bitmaps causing BUG
1019 Details : The locking of the mballoc buddy bitmap and the in-memory
1020 block bitmap was using two different spin locks in some
1021 cases. This made it possible to incorrectly access the
1022 mballoc bitmap while another process was modifying it,
1023 causing a sanity assertion to fail. While no on-disk corruption
1024 was reported, there was some risk of this happening.
1027 Frequency : rare, on llog test 6
1029 Descriptoin: don't allow connect to already connected import
1030 Details : allowing connect to already connected import is hide connecting problem.
1035 Descriptoin: allow kill process which wait statahead result
1036 Details : for some reasons 'ls' can stick in waiting result from statahead,
1037 in this case need way for kill this process.
1039 Severity : enhancement
1041 Description: Add state history info file, enhance import info file
1042 Details : Track import connection state changes in a new osc/mdc proc file;
1043 add overview-type data to the osc/mdc import proc file.
1045 Severity : enhancement
1047 Description: MDS create should not wait for statfs RPC while holding DLM lock.
1049 Severity : enhancement
1051 Description: Update to RHEL5U3 kernel-2.6.18-128.1.1.el5.
1056 Descriptoin: OST grant too much space to client even there are not enough space.
1057 Details : Client will shrink its grant cache to OST if there are no write
1058 activity over 6 mins (GRANT_SHRINK_INTERVAL), and OST will retrieve
1059 this grant cache if there are already not enough avaible space
1060 (left_space < total_clients * 32M).
1063 Frequency : start MDS on uncleanly shutdowned MDS device
1065 Descriptoin: ll_sync thread stay in waiting mds<>ost recovery finished
1066 Details : stay in waiting mds<>ost recovery finished produce random bugs
1067 due race between two ll_sync thread for one lov target. send
1068 ACTIVATE event only if connect realy finished and import have
1072 Frequency : rare, connect and disconnect target at same time
1074 Descriptoin: ASSERTION(atomic_read(&imp->imp_inflight) == 0
1075 Details : don't call obd_disconnect under lov_lock. this long time
1076 operation and can block ptlrpcd which answer to connect request.
1081 Descriptoin: don't lose wakeup for imp_recovery_waitq
1082 Details : recover_import_no_retry or invalidate_import and import_close can
1083 both sleep on imp_recovery_waitq, but we was send only one wakeup
1087 Frequency : always with long access acl
1089 Descriptoin: mds can't pack reply with long acl.
1090 Details : mds don't control size of acl but they limited by reint/getattr
1093 Severity : enhancement
1095 Description: Update to SLES10 kernel-2.6.16.60-0.33.
1097 Severity : enhancement
1099 Description: Update to RHEL5 kernel-2.6.18-92.1.22.el5.
1102 Frequency : start MDS on uncleanly shutdowned MDS device
1104 Descriptoin: aborting recovery hang on MDS
1105 Details : don't throttle destroy RPCs for the MDT.
1108 Frequency : on remount
1110 Description: external journal device not working after the remount
1111 Details : clear dev_rdonly flag for external journal devices in
1117 Description: shutdown vs evict race
1118 Details : client_disconnect_export vs connect request race.
1119 if client will evicted at this time - we start invalidate
1120 thread without referece to import and import can be freed
1124 Frequency : rare, need acl's on inode.
1126 Description: client can't handle ost additional correctly
1127 Details : if ost was added after client connected to mds client can have
1128 hit lnet_try_match_md ... to big messages to wide striped files.
1129 in this case need teach client to handle config events about add
1130 lov target and update client max ea size at that event.
1132 Severity : enhancement
1134 Description: Changelogs
1135 Details : Changelogs are a lightweight mechanism to track filesystem
1136 metadata and namespace changes. The changelog is recorded
1137 permanently on the MDTs, and is periodically "consumed" / purged
1138 when records are no longer needed.
1140 Severity : enhancement
1142 Description: compact fld format with extents
1143 Details : Store range of seq rather than every seq in FLD. Seq
1144 controller update FLD rather than clients. In Case of CMD, mdt0
1145 has FLD, all other metadata server act as non persistent proxy
1146 for FLD queries and cache fld entries in fld cache.
1151 Description: don't skip ost target if they assigned to file
1152 Details : Drop slow OSCs if we can, but not for requested start idx.
1153 This means "if OSC is slow and it is not the requested
1154 start OST, then it can be skipped, otherwise skip it only
1155 if it is inactive/recovering/out-of-space.
1159 Description: more cleanup in mds_lov
1160 Details : not send LOV EA under replay, we can't know about they size at this
1161 time. Don't allow client connect to mds before any ost connected,
1162 for avoid problems with LOV EA size and returning EIO to client.
1164 Severity : enhancement
1166 Description: Interoperability at server side (Disk interoperability)
1168 Severity : enhancement
1170 Description: Update to RHEL5 kernel-2.6.18-92.1.17.el5.
1172 Severity : enhancement
1174 Description: Update to SLES10 SP2 kernel-2.6.16.60-0.31.
1176 Severity : enhancement
1178 Description: New client IO stack (CLIO).
1180 Severity : enhancement
1182 Description: Commit on sharing. Eliminate inter-client dependencies between
1183 uncommitted transactions by doing transaction commits.
1184 Thereby clients may recovery independently.
1187 Frequency : Create a symlink file with a very long name
1189 Description: ldlm_cancel_pack()) ASSERTION(max >= dlm->lock_count + count)
1190 Details : If there is no extra space in the request for early cancels,
1191 ldlm_req_handles_avail() returns 0 instead of a negative value.
1193 Severity : enhancement
1195 Description: Add /proc entry for import status
1196 Details : The mdc, osc, and mgc import directories now have
1197 an import directory that contains useful import data for debugging
1198 connection problems.
1200 Severity : enhancement
1202 Description: Re-disable certain /proc logging
1203 Details : Enable and disable client's offset_stats, extents_stats and
1204 extents_stats_per_process stats logging on the fly.
1207 Frequency : Only on FC kernels 2.6.22+
1209 Description: oops in statahead
1210 Details : Do not drop reference count for the dentry from VFS when lookup,
1211 VFS will do that by itself.
1213 Severity : enhancement
1215 Description: Generic /proc file permissions
1216 Details : Set /Proc file permissions in a more generic way to enable non-
1217 root users operate on some /proc files.
1221 Description: Hitting mdc_commit_close() ASSERTION
1222 Details : Properly handle request reference release in
1223 ll_release_openhandle().
1227 Description: quota recovery deadlock during mds failover
1228 Details : This patch includes att18982, att18236, att18237 in bz14840.
1230 1. fix osts hang when mds does failover with quotaon
1231 2. prevent watchdog storm when osts threads wait for the
1236 Frequency : only patchless client
1237 Description: add workaround for race between add/remove dentry from hash
1239 Severity : enhancement
1241 Description: Allow OST glimpses to return PW locks
1245 Description: LBUG when llog conf file is full
1246 Details : When llog bitmap is full, ENOSPC should be returned for plain log.
1250 Description: Prevent import from entering FULL state when server in recovery
1254 Description: service mount cannot take device name with ":"
1255 Details : Only when device name contains ":/" will mount treat it as
1261 Description: replace ptlrpcd with the statahead thread to interpret the async
1262 statahead RPC callback
1266 Frequency : on recovery
1267 Description: I/O failures after umount during fail back
1268 Details : if client reconnected to restarted server we need join to recovery
1269 instead of find server handler is changed and process self
1270 eviction with cancel all locks.
1272 Severity : enhancement
1274 Description: Update to RHEL5 kernel-2.6.18-92.1.10.el5.
1276 Severity : enhancement
1278 Description: Update to SLES10 SP2 kernel-2.6.16.60-0.27.
1280 Severity : enhancement
1282 Description: Upcall on Lustre log has been dumped
1283 Details : Allow for a user mode script to be called once a Lustre log has
1284 been dumped. It passes the filename of the dumped log to the
1285 script, the location of the script can be specified via
1286 /proc/sys/lnet/debug_log_upcall.
1291 Description: avoid idr_remove called for id which is not allocated.
1292 Details : Move assigment s_dev for clustered nfs to end of initialization,
1293 to avoid problem with error handling.
1298 Description: avoid Already found the key in hash [CONN_UNUSED_HASH] messages
1299 Details : When connection is reused this not moved from CONN_UNUSED_HASH
1300 into CONN_USED_HASH and this prodice warning when put connection
1301 again in unused hash.
1304 Severity : enhancement
1306 Description: File striping can now be set to use an arbitrary pool of OSTs.
1308 Severity : enhancement
1310 Description: Export bytes_read/bytes_write count on OSC/OST.
1314 Description: Early reply size mismatch, MGC loses connection
1315 Details : Apply the MGS_CONNECT_SUPPORTED mask at reconnect time so
1316 the connect flags are properly negotiated.
1321 Description: quotas are not honored with O_DIRECT
1322 Details : all writes with the flag O_DIRECT will use grants which leads to
1323 this problem. Now using OBD_BRW_SYNC to guard this.
1327 Description: add quota statistics
1328 Details : 1. sort out quota proc entries and proc code.
1329 2. add quota statistics
1331 Severity : enhancement
1333 Description: enable quota support for HEAD.
1337 Description: Properly propagate oinfo flags from lov to osc for statfs
1338 Details : restore missing copy oi_flags to lov requests.
1340 Severity : enhancement
1342 Description: Add man pages for llobdstat(8), llstat(8), plot-llstat(8),
1343 : l_getgroups(8), lst(8), routerstat(8)
1344 Details : included man pages for llobdstat(8), llstat(8),
1345 : plot-llstat(8), l_getgroups(8), lst(8), routerstat(8)
1347 Severity : enhancement
1349 Description: configure's --enable-quota should check the
1350 : kernel .config for CONFIG_QUOTA
1351 Details : configure is terminated if --enable-quota is passed but
1352 : no quota support is in kernel
1356 Description: Remove portals compatibility
1357 Details : Remove portals compatibility, not interoperable with releases
1362 Description: Resolve device initialization race
1363 Details : Prevent proc handler from accessing devices added to the
1364 obd_devs array but yet be intialized.
1366 Severity : enhancement
1368 Description: Update to SLES10 SP2 kernel-2.6.16.60-0.23.
1370 Severity : enhancement
1372 Description: Update to RHEL5 kernel-2.6.18-92.1.6.el5.
1377 Description: Using wrong pointer in osc_brw_prep_request
1378 Details : Access to array[-1] can produce panic if kernel compiled with
1379 CONFIG_PAGE_ALLOC enabled
1383 Description: Client runs out of low memory
1384 Details : Consider only lowmem when counting initial number of llap pages
1388 Description: *optional* service tags registration
1389 Details : if the "service tags" package is installed on a Lustre node
1390 When the filesystem is mounted, a local-node service tag will
1391 be created. See http://inventory.sun.com/ for more information
1392 about the Service Tags asset management system.
1396 Description: Kernel BUG tries to release flock
1397 Details : Lustre does not destroy flock lock before last reference goes
1398 away. So always drop flock locks when client is evicted and
1399 perform unlock regardless of successfulness of speaking to MDS.
1403 Description: add refcount protection for osc callbacks, avoid panic on shutdown
1407 Description: sanity test 65a fails if stripecount of -1 is set
1408 Details : handle -1 striping on filesystem in ll_dirstripe_verify
1413 Description: ASSERTION(CheckWriteback(page,cmd)) failed
1414 Details : badly clear PG_Writeback bit in ll_ap_completion can produce false
1417 Severity : enhancement
1419 Description: Update to RHEL5 kernel-2.6.18-53.1.21.el5.
1423 Description: do not process already freed flock
1424 Details : flock can possibly be freed by another thread before it reaches
1425 to ldlm_flock_completion_ast.
1429 Description: LBUG during stress test
1430 Details : Need properly lock accesses the flock deadlock detection list.
1434 Description: oops in page fault handler
1435 Details : kernel page fault handler can return two special 'pages' in error
1436 case, don't try dereference NOPAGE_SIGBUS and NOPAGE_OMM.
1440 Description: timeout with invalidate import.
1441 Details : ptlrpcd_check call obd_zombie_impexp_cull and wait request which
1442 should be handled by ptlrpcd. This produce long age waiting and
1443 -ETIMEOUT ptlrpc_invalidate_import and as result LASSERT.
1445 Severity : enhancement
1447 Description: Update to RHEL5 kernel-2.6.18-53.1.19.el5.
1451 Description: enable MGS and MDT services start separately
1452 Details : add a 'nomgs' option in mount.lustre to enable start a MDT with
1453 a co-located MGS without starting the MGS, which is a complement
1454 to 'nosvc' mount option.
1458 Frequency : after recovery
1459 Description: precreate to many object's after del orphan.
1460 Details : del orphan st in oscc last_id == next_id and this triger growing
1461 count of precreated objects. Set flag LOW to skip increase count
1462 of precreated objects.
1466 Frequency : rare, on clear nid stats
1467 Description: ASSERTION(client_stat->nid_exp_ref_count == 0)
1468 Details : when clean nid stats sometimes try destroy live entry,
1469 and this produce panic in free.
1473 Description: Stack overflow during MDS log replay
1474 ease stack pressure by using a thread dealing llog_process.
1478 Description: wait until IO finished before start new when do lock cancel.
1479 Details : VM protocol want old IO finished before start new, in this case
1480 need wait until PG_writeback is cleared until check dirty flag and
1481 call writepages in lock cancel callback.
1483 Severity : enhancement
1485 Description: using special macro for print time and cleanup in includes.
1489 Description: mds_mfd_close() ASSERTION(rc == 0)
1490 Details : In mds_mfd_close(), we need protect inode's writecount change
1491 within its orphan write semaphore to prevent possible races.
1495 Description: Obsolete CURRENT_SECONDS and use cfs_time_current_sec() instead.
1499 Frequency : rare, on shutdown ost
1500 Description: don't hit live lock with umount ost.
1501 Details : shrink_dcache_parent can be in long loop with destroy dentries,
1502 use shrink_dcache_sb instead.
1506 Description: don't panic with use echo client
1507 Details : echo client pass NULL as client nid pointer and this produce null
1508 pointer dereference.
1512 Description: fix build on ppc32
1513 Details : compile code with -m64 flag produce wrong object file for ppc32.
1517 Description: add message levels for liblustreapi
1521 Description: fix for occasional failure case of -ENOSPC in recovery-small tests
1522 Details : Move the 'good_osts' check before the 'total_bavail' check. This
1523 will result in an -EAGAIN and in the exit call path we call
1524 alloc_rr() which will with increasing aggressiveness attempt to
1525 aquire precreated objects on the minimum number of required OSCs.
1529 Description: Use old size assignment to avoid deadlock
1530 Details : Reverts the changes in bugs 2369 and bug 14138 that introduced
1531 the scheduling while holding a spinlock. We do not need locking
1532 for size in ll_update_inode() because size is only updated from
1533 the MDS for directories or files without objects, so there is no
1534 other place to do the update, and concurrent access to such inodes
1535 are protected by the inode lock.
1539 Description: resolve "_IOWR redefined" build error on SLES10
1543 Description: dump the memory debugging after all modules are unloaded to
1544 suppress false negative in conf_sanity test 39
1546 Severity : enhancement
1548 Description: build kernel-ib packages for OFED 1.3 in our release cycle
1553 Description: fix SLES kernel versioning
1554 Details : the kernel version for our SLES 10 kernel did not include a "-"
1555 before the "smp" at the end. while this was not a problem in
1556 general, it did mean that software trying to use the kernel
1557 version to try to detect a vendor specific kernel would fail.
1558 this was most evident by the OFED build scripts.
1562 Description: Don't update lov_desc members until making sure they are valid
1563 Details : When updating lov_desc members via proc fs, need fix their
1564 validities before doing the real update.
1568 Description: don't put request into delay list while invalidate in flight.
1569 Details : ptlrpc_delay_request sometimes put in delay list while invalidate
1570 import in flight. this produce timeout for invalidate and sometimes
1571 can cause stale data.
1575 Frequency : on ppc only
1576 Description: not convert ost objects for directory because it's not exist.
1577 Details : ll_dir_getstripe assume dirrectory has ost objects but this wrong.
1581 Description: Add FMODE_EXEC file flag for SLES10 SP1 kernel.
1583 Severity : enhancement
1585 Description: Update to support 2.6.22.14 vanilla kernel.
1589 Frequency : rare, on recovery
1590 Description: read procfs can produce deadlock in some situation
1591 Details : Holding lprocfs lock which send rpc can produce block for destroy
1592 obd objects and this also block reconnect with -EALREADY. This
1593 isn't fix all lprocfs bugs - but make it rare.
1595 Severity : enhancement
1597 Description: Update kernel to RHEL5 2.6.18-53.1.14.el5.
1600 Frequency : frequent on X2 node
1602 Description: mdc_set_open_replay_data LBUG
1603 Details : Set replay data for requests that are eligible for replay.
1607 Description: lustre_mgs: operation 101 on unconnected MGS
1608 Details : When MGC is disconnected from MGS long enough, MGS will evict the
1609 MGC, and late on MGC cannot successfully connect to MGS and a lot
1610 of the error messages complaining that MGS is not connected.
1614 Frequency : on network error
1615 Description: panic with double free request if network error
1616 Details : mdc_finish_enqueue is finish request if any network error ocuring,
1617 but it's true only for synchronus enqueue, for async enqueue
1618 (via ptlrpcd) this incorrect and ptlrpcd want finish request
1621 Severity : enhancement
1623 Description: client-side metadata stat-ahead during readdir(directory readahead)
1624 Details : perform client-side metadata stat-ahead when the client detects
1625 readdir and sequential stat of dir entries therein
1628 Frequency : on start mds
1630 Description: Implement get_info(last_id) in obdfilter.
1633 Frequency : occasional
1635 Description: Correctly check stale fid, not start epoch if ost not support SOM
1636 Details : open with flag O_CREATE need set old fid in op_fid3 because
1637 op_fid2 was overwritten with new generated fid, but mds can answer
1638 with one of these two fids and both is not stale. Setattr
1639 incorrectly started an epoch and assume will be called
1640 done_writing, but without SOM done_writing ever being called.
1643 Frequency : rare, depends on device drivers and load
1645 Description: MDS or OSS nodes crash due to stack overflow
1646 Details : Code changes in 1.8.0 increased the stack usage of some functions.
1647 In some cases, in conjunction with device drivers that use a lot
1648 of stack the MDS (or possibly OSS) service threads could overflow
1649 the stack. One change which was identified to consume additional
1650 stack has been reworked to avoid the extra stack usage.
1653 Frequency : occasional
1655 Description: Do not fail import if osc_interpret_create gets -EAGAIN
1656 Details : If osc_interpret_create got -EAGAIN it immediately exits and
1657 wakeup oscc_waitq. After wakeup oscc_wait_for_objects call
1658 oscc_has_objects and see OSC has no objests and call
1659 oscc_internal_create to resend create request.
1661 Severity : enhancement
1663 Description: Update to SLES10 SP1 latest kernel-2.6.16.54-0.2.5.
1665 Severity : enhancement
1667 Description: Update to RHEL5 latest kernel-2.6.18-53.1.13.el5.
1670 Frequency : very rare
1672 Description: Fix replay if there is an un-replied request and open
1673 Details : In some cases, older replay request will revert the
1674 mcd->mcd_last_xid on MDS which is used to record the client's
1675 latest sent request.
1677 Severity : enhancement
1679 Description: Update to RHEL5 latest kernel-2.6.18-53.1.6.el5.
1681 Severity : enhancement
1683 Description: Add rhel5 support to HEAD.
1685 Serverity : enhancement
1687 Description: Update RHEL4 kernel to 2.6.9-67.0.4.
1692 Description: Don't allow skipping OSTs if index has been specified.
1693 Details : Don't allow skipping OSTs if index has been specified, make
1694 locking in internal create lots better.
1698 Description: LBUG in ptlrpc_check_set() bad phase ebc0de00
1699 Details : access to bitfield in structure is always rounded to long
1700 and this produce problem with not atomic change any bit.
1704 Description: Lustre make rpms failed.
1705 Details : Remove ldiskfs spec file to avoids rpmbuild be confused when
1706 builds Lustre rpms from tarball.
1709 Frequency : rare on shutdown ost
1711 Description: If llog cancel was not send before clean_exports phase, this can
1712 produce deadlock in llog code.
1713 Details : If llog thread has last reference to obd and call class_import_put
1714 this produce deadlock because llog_cleanup_commit_master wait when
1715 last llog_commit_thread exited, but this never success because was
1716 called from llog_commit_thread.
1720 Description: allow userland application know is lost one of stripes.
1721 Details : fill lvb_blocks with error code on ost and return it to
1722 application if error flag found.
1726 Description: NULL lov_tgts causing MDS oops
1727 Details : more safe checks for NULL lov_tgts for avoid oops.
1729 Severity : enhancement
1731 Description: Update to RHEL4 latest kernel-2.6.9-67.0.1.EL.
1735 Descriptoin: make lov_create() will not stuck in obd_statfs_rqset()
1736 Details : If an OST is down the MDS will hang indefinitely in
1737 obd_statfs_rqset() waiting for the statfs data. While for
1738 MDS QOS usage of statfs, it should not stuck in waiting.
1740 Severity : enhancement
1742 Description: remote_acl support
1743 Details : Support ACL-based permission check for remote user.
1744 Support setfacl/getfacl for remote user with the utils
1745 "lfs {l,r}{s,g}etfacl" which follow the same parameter format as
1746 the system "{s,g}etfacl" utils.
1748 Severity : enhancement
1750 Description: Update to RHEL4 U6 kernel-2.6.9-67.EL.
1752 Severity : enhancement
1754 Description: Update to RHEL5 latest kernel-2.6.18-53.1.4.el5.
1758 Description: make mgs_setparam() handle fsname containing dash
1759 Details : fsname containing a dash does not work with lctl conf_param
1761 Severity : enhancement
1763 Description: Update to SLES10 SP1 latest kernel-2.6.16.54-0.2.3.
1765 Severity : enhancement
1767 Description: Update to RHEL5 Update-1 kernel 2.6.18-53.el5.
1771 Frequency : rare, at shutdown
1772 Description: access already free / zero obd_namespace.
1773 Details : if client_disconnect_export was called without force flag set,
1774 and exist connect request in flight, this can produce access to
1775 NULL pointer (or already free pointer) when connect_interpret
1776 store ocd flags in obd_namespace.
1780 Frequency : only at startup
1781 Description: not alloc memory with spinlock held.
1782 Details : allocation memory with GFP_KERNEL can produce sleep deadlock,
1783 if any spinlock held.
1785 Severity : enhancement
1787 Description: make lustre randomly fail allocating memory
1788 Details : Make lustre randomly failed allocating memory for testing purpose.
1790 Severity : enhancement
1792 Description: lost problems with lov objid file
1793 Details : Fixes some scability and access to not inited memory problems
1794 in work with lov objdid file.
1799 Description: lfs find does not continue on file error
1800 Details : Continue other files processing when a file/dir is absent.
1804 Description: Inconsistent usage of lustre_pack_reply()
1805 Details : Standardize the usage of lustre_pack_reply() such that it
1806 always generate a CERROR on failure.
1809 Frequency : occasional
1811 Description: MDS hang or stay in waiting lock
1812 Details : If client receive lock with CBPENDING flag ldlm need send lock
1813 cancel as separate rpc, to avoid situation when cancel request
1814 can't processed due all i/o threads stay in wait lock.
1818 Description: Update to RHEL5 kernel 2.6.18-8.1.15.el5.
1822 Description: Update to SLES10 SP1 kernel 2.6.16.53-0.16
1826 Description: Update to SLES9 kernel-2.6.5-7.287.3.
1830 Description: Update to RHEL4 latest kernel.
1832 Severity : enhancement
1834 Description: Build SLES10 patchless client fails
1835 Details : The configure was broken by run ./configure with
1836 --with-linux-obj=.... argument for patchless client. When the
1837 configure use --with-linux-obj, the LINUXINCLUDE= -Iinclude
1838 can't search header adequately. Use absolute path such as
1839 -I($LINUX)/include instead.
1843 Description: interrupt oig_wait produce painc on resend.
1844 Details : brw_redo_request can be used for resend requests from ptlrpcd and
1845 private set, and this produce situation when rq_ptlrpcd_data not
1846 copyed to new allocated request and triggered LBUG on assert
1847 req->rq_ptlrpcd_data != NULL. But this member used only for wakeup
1848 ptlrpcd set if request is changed and can be safety changed to use
1853 Description: LASSERT_{REQ,REP}SWAB macros are buggy
1854 Details : If SWAB_PARANOIA is disabled, the LASSERT_REQSWAB and
1855 LASSERT_REPSWAB macros become no-ops, which is incorrect. Drop
1856 these macros and replace them with their difinitions instead.
1860 Description: Update kernel patches for SLES10 2.6.16.53-0.8.
1861 Details : Update which_patch & target file for SLES10 latest kernel.
1864 Description: Remove client patches from SLES 10 kernel.
1865 Details : This causes SLES 10 clients to behave as patchless clients
1866 even on a Lustre-patched (server) kernel.
1868 Severity : enhancement
1870 Description: self-adjustable client's lru lists
1871 Details : use adaptive algorithm for managing client cached locks lru
1872 lists according to current server load, other client's work
1873 pattern, memory activities, etc. Both, server and client
1874 side namespaces provide number of proc tunables for controlling
1877 Severity : enhancement
1879 Description: light-weight GSS support
1880 Details : Support krb5n and krb5a mode, which keep Kerberos 5 authentication
1881 and reduce performance overhead.
1883 Severity : enhancement
1885 Description: Linux keyring support
1886 Details : Support using service of Linux keyring for Lustre GSS internal
1887 context refresh/cache mechanism.
1891 Description: Fix errors in lfs documentation
1892 Details : Fixes man pages
1896 Description: don't use GFP_* in generic Lustre code.
1897 Details : Use cfs_alloc_* functions and CFS_* flags for code portability.
1901 Description: obdclass is limited by single OBD_ALLOC(idarray)
1902 Details : replace OBD_ALLOC/OBD_FREE with OBD_VMALLOC/OBD_VFREE
1906 Description: warnings with build patchless client with vanila 2.6.19 and up
1907 Details : change old ctl_table style and replace ctl_table/ctl_table_header
1908 with cfs_sysctl_table_t/cfs_sysctl_table_header_t
1912 Frequency : Only for SLES
1913 Description: sanity_quota fail test_1
1914 Details : There are multiple occurences of $TSTUSR in SLES's /etc/group
1915 file, which makes TSTID[2] inunique.
1919 Frequency : Only for SLES9
1920 Description: Kernel patches for SLES9 2.6.5-7.286 kernel
1921 Details : Update target/ChangeLog/which_patch .
1925 Frequency : Only for test_10 in sanity-quota.sh
1926 Description: a bug in quota test code
1927 Details : checking a test flag in a wrong place
1931 Frequency : Only for test_13 in sanity-quota.sh
1932 Description: a bug in quota test script
1933 Details : shouldn't check overfull allocated quota
1937 Description: jbd statistics
1938 Details : Port older jbd statistics patch for sles10
1940 Severity : enhancement
1942 Description: Add printing inode info into message about error in writepage.
1947 Description: reply_lock_interpret crash due to race with it and lock cancel.
1948 Details : Do not replay locks that are being cancelled. Do not reference
1949 locks by their address during replay, just by their handle.
1953 Frequency : When flocks are used.
1954 Description: assertion failure in ldlm_cli_enquque_fini for non NULL lock.
1955 Details : Flock locks might destroy just granted lock if it could be merged
1956 with another existing flock, this is done in completion handler,
1957 so teach ldlm_cli_enquque_fini that this is a valid case for
1963 Description: Oops in read and write path when failing to allocate lock.
1964 Details : Check if lock allocation failed and return error back.
1968 Description: lstripe command fails for valid OST index
1969 Details : The stripe offset is compared to lov->desc.ld_tgt_count
1970 instead of lov->desc.ld_active_tgt_count.
1974 Description: sanity.sh failed test 103
1975 Details : RHEL mis-interpret setfacl "-X" param, so we won't test setfacl
1980 Description: df doesn't work properly if diskfs blocksize != 4K
1981 Details : Choose biggest blocksize of OST's as the LOV's blocksize.
1985 Description: lfs find on -1 stripe looping in lsm_lmm_verify_common()
1986 Details : Avoid lov_verify_lmm_common() on directory with -1 stripe count.
1988 Severity : enhancement
1990 Description: Adaptive timeouts
1991 Details : RPC timeouts adapt to changing server load and network
1992 conditions to reduce resend attempts and improve recovery time.
1996 Description: llapi_file_create() does not allow some changes
1997 Details : add llapi_file_open() that allows specifying the mode and
1998 open flags, and also returns an open file handle.
2002 Description: merge and cleanup kernel patches.
2003 Details : Remove mnt_lustre_list in vfs_intent-2.6-rhel4.patch.
2007 Description: Add journal checksum support.(Kernel part)
2008 Details : The journal checksum feature adds two new flags i.e
2009 JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT and
2010 JBD2_FEATURE_COMPAT_CHECKSUM. JBD2_FEATURE_CHECKSUM flag
2011 indicates that the commit block contains the checksum for
2012 the blocks described by the descriptor blocks. Now commit
2013 record can be sent to disk without waiting for descriptor
2014 blocks to be written to disk. This behavior is controlled
2015 using JBD2_FEATURE_ASYNC_COMMIT flag.
2019 Description: OSS needs mutliple precreate threads
2020 Details : Add ability to start more than one create thread per OSS.
2024 Description: Sanity.sh test_65k failed: setstripe should have succeeded
2025 Details : execute lfs setstripe on client
2029 Description: mds_obd_create error creating tmp object
2030 Details : When the user sets quota on root, llog will be affected and can't
2031 create files and write files.
2034 Frequency : Always on ia64 patchless client, and possibly others.
2036 Description: Add EXPORT_SYMBOL check for node_to_cpumask symbol.
2037 Details : This allows the patchless client to be loaded on architectures
2038 without this export.
2042 Description: RedHat Update kernel for RHEL5
2043 Details : Add the kernel config file for RHEL5.
2047 Description: RedHat Update kernel for RHEL5
2048 Details : Modify the kernel config file more closer RHEL5.
2052 Description: Build failure against Centos5 (RHEL5)
2053 Details : Define PAGE_SIZE when it isn't present.
2057 Description: test_7 on sanity-quota.sh can't be used on separate machines
2058 Details : add facet to handle it
2062 Description: "ll_intent_file_open()) lock enqueue: err: -13" with nfs
2063 Details : with NFS, the anon dentry's parent was set to itself in
2064 d_alloc_anon(), so in MDS, we use rec->ur_fid1 to find the
2065 corresponding dentry other than use rec->ur_name.
2067 Severity : enhancement
2069 Description: lfs setstripe enhancement
2070 Details : Make lfs setstripe understand 'k', 'm' and 'g' for stripe size.
2074 Description: enable data checksumming by default
2075 Details : enable checksum by default, allow --disable-checksum
2076 configure option and "-o nochecksum" mount option. Checksums
2077 can also be disabled at runtime via $LPROC/osc/*/checksum_pages.
2081 Description: lustre support for RHEL5
2082 Details : Add support for RHEL5.
2086 Description: Client eviction due to failover config
2087 Details : after a connection loss, the lustre client should attempt to
2088 reconnect to the last active server first before trying the
2089 other potential connections.
2093 Description: when mds and osts use different quota unit(32bit and 64bit),
2094 quota will be released repeatly.
2095 Details : void sending multiple quota reqs to mds, which will keep the
2096 status between the reqs.
2100 Description: osts not allocated evenly to files
2101 Details : change the condition to increase offset_idx
2104 Frequency : Always for filesystems larger than 2TB on 32-bit systems.
2106 Description: Data corruption for OSTs that are formatted larger than 2TB
2108 Details : When generating the bio request for lustre file writes the
2109 sector number would overflow a temporary variable before being
2110 used for the IO. The data reads correctly from Lustre (which
2111 will overflow in a similar manner) but other file data or
2112 filesystem metadata may be corrupted in some cases.
2116 Description: Tune the kernel for good SCSI performance.
2117 Details : Set the value of /sys/block/{dev}/queue/max_sectors_kb
2118 to the value of /sys/block/{dev}/queue/max_hw_sectors_kb
2123 Description: rewrite ext2-derived code in obdclass/uuid.c
2124 Details : rewrite inherited code (uuid parsing code from ext2 utils)
2125 from scratch preserving functionality.
2129 Description: "lfs find -obd UUID" prints directories
2130 Details : "lfs find -obd UUID" will return all directory names instead
2131 of just file names. It is incorrect because the directories
2132 do not reside on the OSTs.
2136 Description: use i_size_read and i_size_write in 2.6 port
2137 Details : replace inode->i_size access with i_size_read/write()
2140 Frequency : when removing large files
2142 Description: scheduling issue during removal of large Lustre files
2143 Details : Don't take the BKL in fsfilt_ext3_setattr() for 2.6 kernels.
2144 It causes scheduling issues when removing large files (17TB in the
2148 Frequency : only with liblustre clients on XT3
2150 Description: evictions taking too long
2151 Details : allow llrd to evict clients directly on OSTs
2154 Frequency : only on ppc
2156 Description: /proc/fs/lustre/devices broken on ppc
2157 Details : The patch as applied to 1.6.2 doesn't look correct for all arches.
2158 We should make sure the type of 'index' is loff_t and then cast
2159 explicitly as needed below. Do not assign an explicitly cast
2163 Frequency : only for rhel5
2165 Description: Kernel patches update for RHEL5 2.6.18-8.1.10.el5.
2166 Details : Modify the target file & which_kernel.
2168 Severity : enhancement
2170 Description: omit set fsid for export NFS
2171 Details : fix set/restore device id for avoid EMFILE error and mark lustre
2172 fs as FS_REQUIRES_DEV for avoid problems with generate fsid.
2176 Frequency : Always, for kernels after 2.6.16
2177 Description: Fix warning idr_remove called for id=.. which is not allocated.
2178 Details : Last kernels save old s_dev before kill super and not allow
2179 to restore from callback - restore it before call kill_anon_super.
2183 Description: buffer overruns could theoretically occur
2184 Details : llapi_semantic_traverse() modifies the "path" argument by
2185 appending values to the end of the origin string, and a
2186 overrun may occur. Adding buffer overrun check in liblustreapi.
2190 Description: Fix error on 'ls .' at the top of the Lustre mount.
2191 Details : Don't revalidate dentry if it is a root dentry.
2195 Description: Kernel patches update for RHEL4 2.6.9-55.0.6.
2196 Details : Modify vm-tunables-rhel4.patch.
2200 Description: Kernel config for 2.6.18-vanilla.
2201 Details : Modify targets/2.6-vanilla.target.in.
2202 Add config file kernel-2.6.18-2.6-vanilla-i686.config.
2203 Add config file kernel-2.6.18-2.6-vanilla-i686-smp.config.
2204 Add config file kernel-2.6.18-2.6-vanilla-x86_64.config.
2205 Add config file kernel-2.6.18-2.6-vanilla-x86_64-smp.config.
2209 Description: improve handling recoverable errors
2210 Details : if request processig with error which can be recoverable on server
2211 request should be resend, otherwise page released from cache and
2216 Description: Kernel patches update for RHEL5 2.6.18-8.1.14.el5.
2217 Details : Modify target file & which_patch.
2218 A flaw was found in the IA32 system call emulation provided
2219 on AMD64 and Intel 64 platforms. An improperly validated 64-bit
2220 value could be stored in the %RAX register, which could trigger an
2221 out-of-bounds system call table access. An untrusted local user
2222 could exploit this flaw to run code in the kernel
2223 (ie a root privilege escalation). (CVE-2007-4573).
2227 Description: O_DIRECT bypasses client statistics.
2228 Details : When running with O_DIRECT I/O, neither the client rpc_stats nor
2229 read_ahead_stats were updated. copied the stats section from
2230 osc_send_oap_rpc() into async_internal().
2234 Description: Add jbd statistics patch for RHEL5 and 2.6.18-vanilla
2238 Description: change order of libsysio includes
2239 Details : '#include sysio.h' should always come before '#include xtio.h'
2243 Description: handle "serious error: objid * already exists" more gracefully
2244 Details : If LAST_ID value on disk is smaller than the objects existing in
2245 the O/0/d* directories, it indicates disk corruption and causes an
2246 LBUG(). If the object is 0-length, then we should use the existing
2247 object. This will help to avoid a full fsck in most cases.
2249 Severity : enhancement
2251 Description: adapt the lustre_config script to support the upgrade case
2252 Details : Add "-u" option for lustre_config script to support upgrading 1.4
2253 server targets to 1.6 in parallel.
2257 Description: To avoid grant space > avaible space when the disk is almost
2258 full. Without this patch you might see the error "grant XXXX >
2259 available" or some LBUG about grant, when the disk is almost
2261 Details : In filter_check_grant, for non_grant cache write, we should
2262 check the left space by if (*left > ungranted + bytes), instead
2263 of (*left > ungranted), because only we are sure the left space
2264 is enough for another "bytes", then the ungrant space should be
2265 increase. In client, we should update cl_avail_grant only there
2266 is OBD_MD_FLGRANT in the reply.
2270 Description: Update RHEL 4 kernel to fix local root privilege escalation.
2271 Details : Update to the latest RHEL 4 kernel to fix the vulnerability
2272 described in CVE-2007-4573. This problem could allow untrusted
2273 local users to gain root access.
2276 Frequency : when using O_DIRECT and quotas
2278 Description: Incorrect file ownership on O_DIRECT output files
2279 Details : block usage reported by 'lfs quota' does not take into account
2280 files that have been written with O_DIRECT.
2285 Description: touch file failed when fs is not full
2286 Details : OST in recovery should not be discarded by MDS in alloc_qos(),
2287 otherwise we can get ENOSP while fs is not full.
2291 Description: parallel lock callbacks
2292 Details : Instead of sending blocking and completion callbacks as separated
2293 requests, adding them to a set and sending in parallel.
2296 Frequency : only for Cray XT3
2297 Bugzilla : 12829/13455
2298 Description: Changing primary group doesn't change the group lustre assigns to
2300 Details : When CRAY_XT3 is defined, the fsgid supplied by the client is
2301 overridden with the primary group provided by the group upcall,
2302 whereas the supplied fsgid can be trusted if it is in the list of
2303 supplementary groups returned by the group upcall.
2305 Severity : enhancement
2307 Description: Allow masking D_WARNING, D_ERROR messages from console
2308 Details : Console messages can now be disabled via lnet.printk.
2312 Description: User code with malformed file open parameter crashes client node
2313 Details : Before packing join_file req, all the related reference should be
2314 checked carefully in case some malformed flags cause fake
2315 join_file req on client.
2319 Description: LDLM_ENQUEUE races with LDLM_CP_CALLBACK
2320 Details : ldlm_completion_ast() assumes that a lock is granted when the req
2321 mode is equal to the granted mode. However, it should also check
2322 that LDLM_FL_CP_REQD is not set.
2326 Description: Heavy nfs access might result in deadlocks
2327 Details : After ELC code landed, it is now improper to enqueue any mds
2328 locks under och_sem, because enqueue might want to decide to
2329 cancel open locks for same inode we are holding och_sem for.
2333 Description: Client eviction while running blogbench
2334 Details : A lot of unlink operations with concurrent I/O can lead to a
2335 deadlock causing evictions. To address the problem, the number of
2336 oustanding OST_DESTROY requests is now throttled to
2337 max_rpcs_in_flight per OSC and LDLM_FL_DISCARD_DATA blocking
2338 callbacks are processed in priority.
2342 Description: enable ACLs on MDS by default
2343 Details : ACLs must be enabled on MDS by default.
2346 Frequency : PPC/PPC64 only
2348 Description: conflicts between asm-ppc64/types.h and lustre_types.h
2349 Details : fix duplicated definitions between asm-ppc64/types.h and
2350 lustre_types.h on PPC.
2353 Frequency : PPC/PPC64 only
2355 Description: asm-ppc/segment.h does not exist
2356 Details : fix compile issue on PPC.
2360 Description: better handle error messages in extents code
2363 Frequency : RHEL4 only
2365 Description: mkfs is very slow on IA64/RHEL4
2366 Details : A performance regression has been discovered in the MPT Fusion
2367 driver between versions 3.02.73rh and 3.02.99.00rh. As a
2368 consequence, we have downgraded the MPT Fusion driver in the RHEL4
2369 kernel from 3.02.99.00 to 3.02.73 until this problem is fixed.
2371 Severity : enhancement
2373 Description: SNMP support enhancement
2374 Details : Adding total number of sampled request for an MDS node in snmp
2377 Severity : enhancement
2379 Description: Optimize ldlm waiting list processing for PR extent locks
2380 Details : When processing waiting list for read extent lock and meeting read
2381 lock that is same or wider to it that is not contended, skip
2382 processing rest of the list and immediatelly return current
2383 status of conflictness, since we are guaranteed there are no
2384 conflicting locks in the rest of the list.
2388 Description: Time out and refuse to reconnect
2389 Details : When the failover node is the primary node, it is possible
2390 to have two identical connections in imp_conn_list. We must
2391 compare not conn's pointers but NIDs, otherwise we can defeat
2392 connection throttling.
2396 Description: port llog fixes from b1_6 into HEAD
2397 Details : Port llog reference couting and some llog cleanups from b1_6
2398 (bug 10800) into HEAD, for protect from panic and access to already
2399 free llog structures.
2403 Description: Detect stride IO mode in read-ahead
2404 Details : When a client does stride read, read-ahead should detect that and
2405 read-ahead pages according to the detected stride pattern.
2409 Description: data checksumming impacts single node performance
2410 Details : add support for several checksum algorithm. Currently, only CRC32
2411 and Adler-32 are supported. The checksum type can be changed on
2412 the fly via /proc/fs/lustre/osc/*/checksum_type.
2416 Description: use adler32 for page checksums
2417 Details : when available, use the Adler-32 algorithm instead of CRC32 for
2422 Description: build for x2 fails
2423 Details : fix compile issue on Cray systems.
2427 Description: Properly match for duplicate locks
2428 Details : Due to different lock order from skiplists code, we need to
2429 traverse entire list for now
2432 Frequency : only on PPC/SLES10
2434 Description: "BITS_PER_LONG is not 32 or 64" in linux/idr.h
2435 Details : On SLES10/PPC, fs.h includes idr.h which requires BITS_PER_LONG to
2436 be defined. Add a hack in mkfs_lustre.c to work around this compile
2441 Description: LASSERT on MDS when client holding flock lock dies
2442 Details : ldlm pool logic depends on number of granted locks equal to
2443 number of released locks which is not true for flock locks, so
2444 just exclude such locks from consideration.
2448 Description: MDS deadlock with many ll_sync_lov threads and I/O stalled
2449 Details : Use fsfilt_sync() for both the whole filesystem sync and
2450 individual file sync to eliminate dangerous inode locking
2451 with I_LOCK that can lead to a deadlock.
2455 Description: performance in 1.6.3
2456 Details : Force q->max_phys_segments to MAX_PHYS_SEGMENTS on SLES10 to be
2457 sure that 1MB requests are not fragmented by the block layer.
2459 Severity : enhancement
2461 Description: organize the server-side client stats on per-nid basis
2462 Details : Change the structure of stats under obdfilter and mds to
2470 The "uuid"s file would list the uuids of _active_ exports.
2471 And the clear entry is to clear all stats and stale nids.
2473 Severity : enhancement
2475 Description: eliminate client locks in face of contention
2476 Details : file contention detection and lockless i/o implementation
2477 for contended files.
2481 Description: Reinitialize optind to 0 so that interactive lfs works in all cases
2484 Frequency : very rare, if additional xattrs are used on kernels >= 2.6.12
2486 Description: MDS may lose file striping (and hence file data) in some cases
2487 Details : If there are additional extended attributes stored on the MDS,
2488 in particular ACLs, SELinux, or user attributes (if user_xattr
2489 is specified for the client mount options) then there is a risk
2490 of attribute loss. Additionally, the Lustre file striping
2491 needs to be larger than default (e.g. striped over all OSTs),
2492 and an additional attribute must be stored initially in the
2493 inode and then increase in size enough to be moved to the
2494 external attribute block (e.g. ACL growing in size) for file
2499 Description: skiplist implementation simplification
2500 Details : skiplists are used to group compatible locks on granted list
2501 that was implemented as tracking first and last lock of each
2502 lock group the patch changes that to using doubly linked lists
2506 Description: MDS LBUG: ASSERTION(!IS_ERR(dchild))
2507 Details : Change LASSERTs to client eviction (i.e. abort client's recovery)
2508 because LASSERT on both the data supplied by a client, and the
2509 data on disk is dangerous and incorrect.
2511 Severity : enhancement
2513 Description: Slow truncate/writes to huge files at high offsets.
2514 Details : Directly associate cached pages to lock that protect those pages,
2515 this allows us to quickly find what pages to write and remove
2516 once lock callback is received.
2520 Description: more ldlm soft lockups
2521 Details : In ldlm_resource_add_lock(), call to ldlm_resource_dump()
2522 starve other threads from the resource lock for a long time in
2523 case of long waiting queue, so change the debug level from
2524 D_OTHER to the less frequently used D_INFO.
2526 Severity : enhancement
2528 Description: add -gid, -group, -uid, -user options to lfs find
2532 Description: Hung threads in invalidate_inode_pages2_range
2533 Details : The direct IO path doesn't call check_rpcs to submit a new RPC
2534 once one is completed. As a result, some RPCs are stuck in the
2535 queue and are never sent.
2539 Description: filter threads hungs on waiting journal commit
2540 Details : Cleanup filter group llog code, then only filter group llog will
2541 be only created in the MDS/OST syncing process.
2545 Description: Procfs and llog threads access destoryed import sometimes.
2546 Details : Sync the import destoryed process with procfs and llog threads by
2547 the import refcount and semaphore.
2549 Severity : enhancement
2551 Description: openlock cache of b1_6 port to HEAD
2556 Description: kernel BUG at ldiskfs2_ext_new_extent_cb
2557 Details : If insertion of an extent fails, then discard the inode
2558 preallocation and free data blocks else it can lead to duplicate
2563 Description: don't always update ctime in ext3_xattr_set_handle()
2564 Details : Current xattr code updates inode ctime in ext3_xattr_set_handle.
2565 In some cases the ctime should not be updated, for example for
2566 2.0->1.8 compatibility it is necessary to delete an xattr and it
2567 should not update the ctime.
2571 Bugzilla : 15713/16362
2572 Description: Assertion in iopen_connect_dentry in 1.6.3
2573 Details : looking up an inode via iopen with the wrong generation number can
2574 populate the dcache with a disconneced dentry while the inode
2575 number is in the process of being reallocated. This causes an
2576 assertion failure in iopen since the inode's dentry list contains
2577 both a connected and disconnected dentry.
2581 Description: assertion failure in ldlm_handle2lock()
2582 Details : fix a race between class_handle_unhash() and class_handle2object()
2583 introduced in lustre 1.6.5 by bug 13622.
2588 Description: Kernel BUG: sd_iostats_bump: unexpected disk index
2589 Details : remove the limit of 256 scsi disks in the sd_iostat patch
2594 Description: oops in sd_iostats_seq_show()
2595 Details : unloading/reloading the scsi low level driver triggers a kernel
2596 bug when trying to access the sd iostat file.
2601 Description: Kernel panics during QLogic driver reload
2602 Details : REQ_BLOCK_PC requests are not handled properly in the sd iostat
2603 patch, causing memory corruption.
2608 Description: journal_dev option does not work in b1_6
2609 Details : pass mount option during pre-mount.
2611 Severity : enhancement
2613 Description: Add a FIEMAP(FIle Extent MAP) ioctl
2614 Details : FIEMAP ioctl will allow an application to efficiently fetch the
2615 extent information of a file. It can be used to map logical blocks
2616 in a file to physical blocks in the block device.
2620 Description: LDLM soft lockups - improvement
2621 Details : It is be possible to send the lock handle along with each read
2622 or write request because the client is already doing a lock match
2623 itself so there isn't any reason the OST should have to re-do that
2627 Frequency : only on Cray X2
2629 Description: X2 build failures
2630 Details : fix build failures on Cray X2.
2634 Description: xid & resent requests
2635 Details : Initialize RPC XID from clock at startup (randomly if clock is
2638 Severity : enhancement
2640 Description: Add lustre_start utility to start or stop multiple Lustre servers
2645 Description: Lustre GPF in {:ptlrpc:ptlrpc_server_free_request+373}
2646 Details : In case of memory pressure, list_del() can be called twice on
2647 req->rq_history_list, causing a kernel oops.
2651 Description: kptllnd_peer_check_sends()) ASSERTION(!in_interrupt()) failed
2652 Details : fix stack overflow in the distributed lock manager by defering
2653 export eviction after a failed AST to the elt thread instead of
2654 handling it in the dlm interpret routine.
2658 Description: Convert some comments to new format.
2659 Details : Update documenting comments to match doxygen conventions.
2663 Description: Grammar fixes.
2664 Details : A couple of trivial spelling fixes.
2668 Description: OSD_COUNTERS-mandatory
2669 Details : Make previously optional ->oti_{w,r}_locks sanity checks mandatory
2670 to simplify the code.
2674 Description: simplify cmm_device freeing logic.
2675 Details : Call cmm_device_free() in the failure path of cmm_device_alloc().
2679 Description: Add lockdep support to dt_object_operations locking interface.
2680 Details : Augment ->do_{read,write}_lock() prototypes with a `role'
2681 parameter indicating lock ordering. Update mdd code to use new
2686 Description: Introduce failloc constants for lockless IO tests.
2687 Details : Add two new failloc constants to test lockless IO. Only one of
2688 them in implemented---another is checked in yet to be landed
2693 Description: Add lockdep support for inode mutex.
2694 Details : Introduce and use new LOCK_INODE_MUTEX_PARENT() macro to be used
2695 in the situations where Lustre has to lock more than one inode
2700 Description: Add optional invariants checking support.
2701 Details : Add new LINVRNT() macro, optional on new --enable-invariants
2702 configure switch. This macro is to be used for consistency and
2703 sanity checks that are too expensive to be left in `production'
2708 Description: Zap lock->l_granted_mode with explicit LCK_MINMODE.
2709 Details : Use LCK_MINMODE rather than 0 to reset lock->l_granted_mode to
2714 Description: Add lockdep support for ldlm_lock and ldlm_resource.
2715 Details : Use spin_lock_nested() in (the only) situation where more than
2716 one ldlm_lock is locked simultaneously. Also, fix possible
2717 dead-lock in ldlm_lock_change_resource() by enforcing particular
2722 Description: Use struct ldlm_callback_suite in ldlm_lock_create().
2723 Details : Instead of specifying each ldlm_lock call-back through separate
2724 parameter, wrap them into struct ldlm_callback_suite.
2728 Description: Kill join_lru obd method and its callers.
2729 Details : CLIO uses lock weighting policy to keep locks over mmapped regions
2730 in memory---a requirement implemented through ->o_join_lru() obd
2731 method in HEAD. Remove this method and its users.
2735 Description: Add asynchronous ldlm ENQUEUE completion handler.
2736 Details : CLIO posts ENQUEUE requests asynchronously through ptlrpcd---a
2737 case that stock ldlm_completion_ast() cannot handle as it waits
2738 until lock is granted. Introduce new ldlm_completion_ast_async()
2739 for this. Also comment ldlm_completion_ast().
2743 Description: ldlm_error <-> errno conversion.
2744 Details : Add functions to map (rather arbitrary) between LDLM error codes
2745 and standard errno values. CLIO needs this to prevent LDLM specific
2746 constants from escaping ldlm and osc.
2750 Description: Kill unused ldlm_handle2lock_ns() function.
2751 Details : Kill unused ldlm_handle2lock_ns() function.
2755 Description: Add lu_ref support to ldlm_lock
2756 Details : lu_ref support for ldlm_lock and ldlm_resource. See lu_ref patch.
2757 lu_ref fields ->l_reference and ->lr_reference are added to ldlm_lock
2758 and ldlm_resource. LDLM interface has to be changed, because code that
2759 releases a reference on a lock, has to "know" what reference this is.
2760 In the most frequent case
2762 lock = ldlm_handle2lock(handle);
2764 LDLM_LOCK_PUT(lock);
2766 no changes are required. When any other reference (received _not_ from
2767 ldlm_handle2lock()) is released, LDLM_LOCK_RELEASE() has to be called
2768 instead of LDLM_LOCK_PUT().
2770 Arguably, changes are pervasive, and interface requires some discipline
2771 for proper use. On the other hand, it was very instrumental in finding
2772 a few leaked lock references.
2776 Description: Add ldlm_lock_addref_try().
2777 Details : Introduce ldlm_lock_addref_try() function (used by CLIO) that
2778 attempts to addref a lock that might be being canceled
2783 Description: Add ldlm_weigh_callback().
2784 Details : Add new ->l_weigh_ast() call-back to ldlm_lock. It is called
2785 by ldlm_cancel_shrink_policy() to estimate lock "value", instead of
2786 hard-coded `number of pages' logic.
2790 Description: Add lockdep annotations to llog code.
2791 Details : Use appropriately tagged _nested() locking calls in the places
2792 where llog takes more than one ->lgh_lock lock.
2796 Description: Add loi_kms_set().
2797 Details : Wrap kms updates into a helper function.
2801 Description: Constify instances of struct lsm_operations.
2802 Details : Constify instances of struct lsm_operations.
2806 Description: lu_conf support.
2807 Details : On a server, a file system object is uniquely identified
2808 by a fid, which is sufficient to locate and load all object
2809 state (inode). On a client, on the other hand, more data are
2810 necessary instantiate an object. Change lu_object_find() and
2811 friends to take additional `lu_conf' argument describing object.
2812 Typically this includes layout information.
2816 Description: lu_context fixes.
2817 Details : Introduce new lu_context functions that are needed on the client
2818 side, where some system threads (ptlrpcd) are shared by multiple
2819 modules, and so cannot be stopped during module shutdown.
2823 Description: Add start and stop methods to lu_device_type_operations.
2824 Details : Introduce two new methods in lu_device_type_operations, that are
2825 invoked when first instance of a given type is created and last one
2826 is destroyed respectively. This is need by CLIO.
2830 Description: Add lu_ref support to struct lu_device.
2831 Details : Add lu_ref support to lu_object and lu_device. lu_ref is used to
2832 track leaked references.
2836 Description: Introduce lu_kmem_descr.
2837 Details : lu_kmem_descr and its companion interface allow to create
2838 and destroy a number of kmem caches at once.
2842 Description: Fix lu_object finalization race.
2843 Details : Fix a race between lu_object_find() finding an object and its
2844 concurrent finalization. This race is (most likely) not possible
2845 on the server, but might happen on the client.
2849 Description: Introduce lu_ref interface.
2850 Details : lu_ref is a debugging module allowing to track references to
2851 a given object. It is quite cpu expensive, and has to be
2852 explicitly enabled with --enable-lu_ref. See usage description
2857 Description: Factor lu_site procfs stats into a separate function.
2858 Details : Separate lu_site stats printing code into a separate function
2859 that can be reused on a client.
2863 Description: Constify instances of struct {lu,dt,md}_device_operations.
2864 Details : Constify instances of struct {lu,dt,md}_device_operations.
2868 Description: Introduce struct md_site and move meta-data specific parts of
2869 struct lu_site here.
2870 Details : Move md-specific fields out of struct lu_site into special struct
2871 md_site, so that lu_site can be used on a client.
2875 Description: Kill mdd_lov_destroy().
2876 Details : Remove unused mdd code.
2880 Description: Add st_block checking to multistat.c.
2881 Details : Add st_block checking to multistat.c.
2885 Description: Add lu_ref support to struct obd_device.
2886 Details : Add lu_ref tracking to obd_device.
2890 Description: Kill obd_set_fail_loc().
2891 Details : Remove unused code.
2895 Description: Add special type for ptlrpc_request interpret functions.
2896 Details : Add lu_env parameter to ->rq_interpreter call-back. NULL is passed
2897 there. Actual usage will be in CLIO.
2901 Description: Replace RW_LOCK_UNLOCKED() macro with rwlock_init().
2902 Details : Replace RW_LOCK_UNLOCKED() with rwlock_init() as the former
2903 doesn't work with lockdep.
2907 Description: Add rwv.c test program.
2908 Details : New testing program exercising readv(2) and writev(2) (Qian).
2912 Description: Add sendfile.c test program.
2913 Details : New testing program exercising sendfile(2) (Jay).
2917 Description: Ratelimit a message that can be very frequent.
2918 Details : Ratelimit a memory allocation failure message that can
2923 Description: Use cdebug_show() in CDEBUG-style macros defined outside of libcfs.
2924 Details : Use cdebug_show() in CDEBUG-style macros defined outside of libcfs.
2928 Description: Liblustre build fixes.
2929 Details : Liblustre build fixes.
2933 Description: libcfs: add cfs_{need,cond}_resched() interface.
2934 Details : libcfs: add cfs_{need,cond}_resched() definition and
2935 implementations for Linux, NT, and liblustre.
2937 Severity : enhancement
2939 Description: More exported tunables for mballoc
2940 Details : Add support for tunable preallocation window and new tunables for
2941 large/small requests
2945 Description: Detect corruption of block bitmap and checking for preallocations
2946 Details : Checks validity of on-disk block bitmap. Also it does better
2947 checking of number of applied preallocations. When corruption is
2948 found, it turns filesystem readonly to prevent further corruptions.
2952 Description: (rw.c:1323:ll_read_ahead_pages()) ASSERTION(page_idx > ria->ria_stoff) failed
2953 Details : Once the unmatched stride IO mode is detected, shrink the stride-ahead
2954 window to 0. If it does hit cache miss, and read-pattern is still
2955 stride-io mode, does not reset the stride window, but also does not
2956 increase the stride window length in this case.
2960 Frequency : only for big-endian servers
2961 Description: Check if system is big-endian while mounting fs with extents feature
2962 Details : Mounting a filesystem with extents feature will fail on big-endian
2963 systems since ext3-based ldiskfs is not supported on big-endian
2964 systems. This can be over-riden with "bigendian_extents" mount option.
2966 Severity : enhancement
2968 Description: The root squash functionality
2969 Details : A security feature, which is to prevent users from being able
2970 to mount lustre on their desktop, run as root, and delete
2971 all of the files in the filesystem. The goal is accomplished by
2972 remapping user id (UID) and group id (GID) of the root user to
2973 a UID and GID specified by the system administartor via Lustre
2974 configuration management server (MGS). The functionality also
2975 allows to specify sets of clients for which the remapping does
2980 Description: Excessive recovery window
2981 Details : With AT enabled, the recovery window can be excessively long (6000+
2982 seconds). To address this problem, we no longer use
2983 OBD_RECOVERY_FACTOR when extending the recovery window (the connect
2984 timeout no longer depends on the service time, it is set to
2985 INITIAL_CONNECT_TIMEOUT now) and clients report the old service
2986 time via pb_service_time.
2990 Description: Watchdog triggered on MDS failover
2991 Details : enable OBD_CONNECT_MDT flag when connecting from the MDS so that
2992 the OSTs know that the MDS "UUID" can be reused for the same export
2993 from a different NID, so we do not need to wait for the export to be
2997 Frequency : rare, only if using MMP with Linux RAID
2999 Description: MMP doesn't work with Linux RAID
3000 Details : While using HA for Lustre servers with Linux RAID, it is possible
3001 that MMP will not detect multiple mounts. To make this work we
3002 need to unplug the device queue in RAID when the MMP block is being
3003 written. Also while reading the MMP block, we should read it from
3004 disk and not the cached one.
3006 Severity : enhancement
3008 Description: open file using fid
3009 Details : A file can be opened using just its fid, like
3010 <mntpt>/.lustre/fid/SEQ:OID:VER - this is needed for HSM and replication
3013 Frequency : Only in RHEL5 when mounting multiple ext3 filesystems
3016 Description: "kmem_cache_create: duplicate cache jbd_4k" error message
3017 Details : add proper locking for creation of jbd_4k slab cache
3021 Description: MMP check in ext3_remount() fails without displaying any error
3022 Details : When multiple mount protection fails during remount, proper error
3025 Severity : enhancement
3027 Description: Allow stripe size to be up to 4G-64k
3028 Details : Fix math logic to allow large stripe sizes.
3032 Description: add check for >8TB ldiskfs filesystems
3033 Details : ext3-based ldiskfs does not support greater than 8TB LUNs.
3034 Don't allow >8TB ldiskfs filesystems to be mounted without
3035 force_over_8tb mount option
3040 Descriptoin: OSS complains "too many blocks" and crashes
3041 Details : In filter_commitrw_write():
3042 1. lquota_chkquota();
3043 2. obdo_from_inode();
3044 3. lquota_pending_commit();
3045 "2." has changed uid and gid in this case, then caused
3046 lquota_chkquota() and lquota_pending_commit() to use
3047 different uid and gid. That is the root of the bug.
3051 Descriptoin: getstripe doesn't report stripe count = -1 sanely
3052 Details : Detect and print out the 'stripe count = -1' case properly.
3056 Description: Out or order replies might be lost on replay
3057 Details : In ptlrpc_retain_replayable_request if we cannot find retained
3058 request with tid smaller than one currently being added, add it
3059 to the start, not end of the list.
3061 --------------------------------------------------------------------------------
3063 2007-08-10 Cluster File Systems, Inc. <info@clusterfs.com>
3065 * Support for kernels:
3066 2.6.5-7.283 (SLES 9),
3067 2.6.9-55.EL (RHEL 4),
3068 2.6.16.46-0.14 (SLES 10),
3069 2.6.18.8 vanilla (kernel.org)
3070 * Client support for unpatched kernels:
3071 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
3072 2.6.16 - 2.6.22 vanilla (kernel.org)
3073 * Due to recently discovered recovery problems, we do not recommend
3074 using patchless RHEL 4 clients with this or any earlier release.
3075 * Recommended e2fsprogs version: 1.39.cfs8
3076 * Note that reiserfs quotas are disabled on SLES 10 in this kernel.
3077 * Starting with this release, the ldiskfs backing filesystem required
3078 by Lustre is now in its own package, lustre-ldiskfs. This package
3079 should be installed. It is versioned separately from Lustre and
3080 may be released separately in future.
3084 Description: block reactivating mgc import until all deactivates complete
3085 Details : Fix race when failing back MDT/MGS to itself (testing)
3088 Frequency : at statup only
3090 Description: mds_lov_synchronize race leads to various problems
3091 Details : simultaneous MDT->OST connections at startup can cause the
3092 sync to abort, leaving the OSC in a bad state.
3094 Severity : enhancement
3096 Description: add optional extra BUILD_VERSION info
3097 Details : add a new environment variable (namely LUSTRE_VERS) which allows
3098 to override the lustre version.
3101 Frequency : 2.6.18 servers only
3103 Description: ll_kern_mount() doesn't release the module reference
3104 Details : The ldiskfs module reference count never drops down to 0
3105 because ll_kern_mount() doesn't release the module reference.
3110 Description: server LBUG when using old ost_num_threads parameter
3111 Details : Accept the old ost_num_threads parameter but warn that it
3112 is deprecated, and fix an off-by-one error that caused an LBUG.
3117 Description: Transient SCSI error results in persistent IO issue
3118 Details : iobuf->dr_error is not reinitialized to 0 between two
3122 Frequency : sometimes when underlying device returns I/O errors
3124 Description: OSTs not going read-only during write failures
3125 Details : OSTs are not remounted read-only when the journal commit threads
3126 get I/O errors because fsfilt_ext3 calls journal_start/stop()
3127 instead of the ext3 wrappers.
3131 Description: poor connect scaling with increasing client count
3132 Details : Don't run filter_grant_sanity_check for more than 100 exports
3133 to improve scaling for large numbers of clients.
3136 Frequency : SLES10 only
3138 Description: sanity-quota.sh quotacheck failed: rc = -22
3139 Details : Quotas cannot be enabled on SLES10.
3142 Frequency : liblustre clients only
3144 Description: getdirentries does not give error when run on compute nodes
3145 Details : getdirentries does not fail when the size specified as an argument
3146 is too small to contain at least one entry
3148 Severity : enhancement
3150 Description: Add LNET router traceability for debug purposes
3151 Details : If a checksum failure occurs with a router as part of the
3152 IO path, the NID of the last router that forwarded the bulk data
3153 is printed so it can be identified.
3158 Description: OST "spontaneously" evicts client; client has imp_pingable == 0
3159 Details : Due to a race condition, liblustre clients were occasionally
3160 evicted incorrectly.
3162 Severity : enhancement
3164 Description: lfs setstripe use optional parameters instead of postional
3167 Severity : enhancement
3169 Description: Nanosecond timestamp support for ldiskfs
3170 Details : The on-disk ldiskfs filesystem has added support for nanosecond
3171 resolution timestamps. There is not yet support for this at
3172 the Lustre filesystem level.
3175 Frequency : during server recovery
3177 Description: MDS failing to send precreate requests due to OSCC_FLAG_RECOVERING
3178 Details : request with rq_no_resend flag not awake l_wait_event if they get
3182 Frequency : nfs export on patchless client
3184 Description: connectathon hang when test nfs export over patchless client
3185 Details : Disconnected dentry cannot be found with lookup, so we do not need
3186 to unhash it or make it invalid
3189 Description: fix llapi_lov_get_uuids() to allow many OSTs to be returned
3190 Details: : Change llapi_lov_get_uuids() to read the UUIDs from /proc instead
3191 of using an ioctl. This allows lfsck for > 160 OSTs to succeed.
3196 Description: open req refcounting wrong on reconnect
3197 Details : If reconnect happened between getting open reply from server and
3198 call to mdc_set_replay_data in ll_file_open, we will schedule
3199 replay for unreferenced request that we are about to free.
3200 Subsequent close will crash in variety of ways.
3201 Check that request is still eligible for replay in
3202 mdc_set_replay_data().
3207 Description: disable writes to filesystem when reading health_check file
3208 Details : the default for reading the health_check proc file has changed
3209 to NOT do a journal transaction and write to disk, because this
3210 can cause reads of the /proc file to hang and block HA state
3211 checking on a healthy but otherwise heavily loaded system. It
3212 is possible to return to the previous behaviour during configure
3213 with --enable-health-write.
3215 Severity : enhancement
3217 Description: 64-bit inode version
3218 Details: : Add a on-disk 64-bit inode version for ext3 to track changes made
3219 to the inode. This will be required for version-based recovery.
3224 Description: MDS fails to start if a duplicate client export is detected
3225 Details : in some rare cases it was possible for a client to connect to
3226 an MDS multiple times. Upon recovery the MDS would detect this
3227 and fail during startup. Handle this more gracefully.
3229 Severity : enhancement
3231 Description: Add -o localflock option to simulate old noflock
3233 Details : This will achieve local-only flock/fcntl locks
3239 Description: log_commit_thread vs filter_destroy race leads to crash
3240 Details : Take import reference before releasing llog record semaphore
3245 Description: Wrong request locking in request set processing
3246 Details : ptlrpc_check_set wrongly uses req->rq_lock for proctect add to
3247 imp_delayed_list, in this place should be used imp_lock.
3250 Frequency : when reconnection
3252 Description: Grant Leak when osc reconnect to OST
3253 Details : When osc reconnect ost, OST(filter) should check whether it
3254 should grant more space to client by comparing fed_grant and
3255 cl_avail_grant, and return the granted space to client instead
3256 of "new granted" space, because client will call osc_init_grant
3257 to update the client grant space info.
3260 Frequency : when client reconnect to OST
3262 Description: Grant Leak when osc do resend and replay bulk write
3263 Details : When osc reconnect to OST, OST(filter)should clear grant info of
3264 bulk write request, because the grant info while be sync between
3265 OSC and OST when reconnect, and we should ignore the grant info
3266 these of resend/replay write req.
3271 Description: Grant space more than available left space sometimes.
3272 Details : When then OST is about to be full, if two bulk writing from
3273 different clients came to OST. Accord the avaliable space of the
3274 OST, the first req should be permitted, and the second one
3275 should be denied by ENOSPC. But if the seconde arrived before
3276 the first one is committed. The OST might wrongly permit second
3277 writing, which will cause grant space > available space.
3280 Frequency : when client is evicted
3282 Description: Grant might be wrongly erased when osc is evicted by OST
3283 Details : when the import is evicted by server, it will fork another
3284 thread ptlrpc_invalidate_import_thread to invalidate the
3285 import, where the grant will be set to 0. While the original
3286 thread will update the grant it got when connecting. So if
3287 the former happened latter, the grant will be wrongly errased
3288 because of this race.
3293 Description: Checking Stale with correct fid
3294 Details : ll_revalidate_it should uses de_inode instead of op_data.fid2
3295 to check whether it is stale, because sometimes, we want the
3296 enqueue happened anyway, and op_data.fid2 will not be initialized.
3298 Severity : enhancement
3300 Description: update patchless client
3301 Details : Add support for patchless client with 2.6.20, 2.6.21 and RHEL 5
3304 Frequency : only with 2.4 kernel
3306 Description: random memory corruption
3307 Details : size of struct ll_inode_info is to big for union inode.u and this
3308 can be cause of random memory corruption.
3313 Description: Memory leak in recovery
3314 Details : Lov_mds_md was not free in an error handler in mds_create_object.
3315 It should also check obd_fail before fsfilt_start, otherwise if
3316 fsfilt_start return -EROFS,(failover mds during mds recovery).
3317 then the req will return with repmsg->transno = 0 and rc = EROFS.
3318 and we met hit the assert LASSERT(req->rq_reqmsg->transno ==
3319 req->rq_repmsg->transno) in ptlrpc_replay_interpret. Fcc should
3320 be freed no matter whether fsfilt_commit success or not.
3323 Frequency : only with huge count clients
3325 Description: Prevents from taking the superblock lock in llap_from_page for
3327 Details : using LL_ORIGIN_REMOVEPAGE origin flag instead of LL_ORIGIN_UNKNOW
3328 for llap_from_page call in ll_removepage prevents from taking the
3329 superblock lock for a soon died page.
3334 Description: Not check open intent error before release open handle
3335 Details : in some rare cases, the open intent error is not checked before
3336 release open handle, which may cause
3337 ASSERTION(open_req->rq_transno != 0), because it tries to release
3338 the failed open handle.
3343 Description: Set cat log bitmap only after create log success.
3344 Details : in some rare cases, the cat log bitmap is set too early. and it
3345 should be set only after create log success.
3349 Description: Accessing a block bevice can re-enable I/O when Lustre is
3350 tearing down a device.
3351 Details : dev_clear_rdonly(bdev) must be called in kill_bdev() instead of
3356 Description: service threads may hog cpus when there are a lot of requests
3357 Details : Insert cond_resched to give other threads a chance to use some CPU
3362 Description: the cat log was not initialized in recovery
3363 Details : When mds(mgs) do recovery, the tgt_count might be zero, so the
3364 unlink log on mds will not be initialized until mds post
3365 recovery. And also in mds post recovery, the unlink log will
3366 initialization will be done asynchronausly, so there will be race
3367 between add unlink log and unlink log initialization.
3371 Description: brw_stats were being printed incorrectly
3372 Details : brw_stats were being printed as log2 but all of them were not
3373 recorded as log2. Also remove some code duplication arising from
3374 filter_tally_{read,write}.
3378 Frequency : rare, only in recovery.
3379 Description: ASSERTION(req->rq_type != LI_POISON) failed
3380 Details : imp_lock should be held while iterating over imp_sending_list for
3381 prevent destroy request after get timeout in ptlrpc_queue_wait.
3385 Description: replay-single.sh test 52 fails
3386 Details : A lock's skiplist need to be cleanup when it being unlinked
3387 from its resource list.
3391 Description: Short directio read returns full requested size rather than
3393 Details : Direct I/O operations should return actual amount of bytes
3394 transferred rather than requested size.
3396 Severity : enhancement
3398 Description: metadata RPC reduction (e.g. for rm performance)
3399 Details : decrease the amount of synchronous RPC between clients and servers
3400 by canceling conflicing lock before the operation on the client
3401 and packing thier handles into the main operation RPC to server.
3403 Severity : enhancement
3405 Description: Async OSC create to avoid the blocking unnecessarily.
3406 Details : If an OST has no remaining object, system will block on the
3407 creation when it needs to create a new object on this OST. Now,
3408 ways use pre-created objects when available, instead of blocking on
3409 an empty osc while others are not empty. If we must block, we
3410 block for the shortest possible period of time.
3414 Description: improve handling recoverable errors
3415 Details : if request processig with error which can be recoverable on server
3416 request should be resend, otherwise page released from cache and
3419 Severity : enhancement
3421 Description: refine locking for avoid write wrong info into lov_objid file
3422 Details : fix possible races with add new target and write/update data in
3425 --------------------------------------------------------------------------------
3427 2007-05-03 Cluster File Systems, Inc. <info@clusterfs.com>
3432 Frequency : on some architectures
3434 Description: 1.6 client sometimes fails to mount from a 1.4 MDT
3435 Details : Uninitialized flags sometimes cause configuration commands to
3439 Frequency : patchless clients only
3441 Description: missing __iget() symbol export
3442 Details : The __iget() symbol export is missing. To avoid the need for
3443 this on patchless clients the deathrow inode reaper is turned
3444 off, and we depend on the VM to clean up old inodes. This
3445 dependency was during via the fix for bug 12181.
3447 --------------------------------------------------------------------------------
3449 2007-04-19 Cluster File Systems, Inc. <info@clusterfs.com>
3451 * CONFIGURATION CHANGE. This version of Lustre WILL NOT
3452 INTEROPERATE with older versions automatically. In many cases a
3453 special upgrade step is needed. Please read the
3454 user documentation before upgrading any part of a live system.
3455 * WIRE PROTOCOL CHANGE from previous 1.6 beta versions. This
3456 version will not interoperate with 1.6 betas before beta5 (1.5.95).
3457 * WARNING: Lustre configuration and startup changes are required with
3458 this release. See https://mail.clusterfs.com/wikis/lustre/MountConf
3462 Severity : enhancement
3464 Description: MountConf
3465 Details : Lustre configuration is now managed via mkfs and mount
3466 commands instead of lmc and lconf. New obd types (MGS, MGC)
3467 are added for dynamic configuration management. See
3468 https://mail.clusterfs.com/wikis/lustre/MountConf for
3471 Severity : enhancement
3473 Description: dynamic OST addition
3474 Details : OSTs can now be added to a live filesystem
3476 Severity : enhancement
3478 Description: startup order invariance
3479 Details : MDTs and OSTs can be started in any order. Clients only
3480 require the MDT to complete startup.
3482 Severity : enhancement
3484 Description: parallel, asynchronous orphan cleanup
3485 Details : orphan cleanup is now performed in separate threads for each
3486 OST, allowing parallel non-blocking operation.
3488 Severity : enhancement
3490 Description: optimized stripe assignment
3491 Details : stripe assignments are now made based on ost space available,
3492 ost previous usage, and OSS previous usage, in order to try
3493 to optimize storage space and networking resources.
3495 Severity : enhancement
3497 Description: Permanently set tunables
3498 Details : All writable /proc/fs/lustre tunables can now be permanently
3499 set on a per-server basis, at mkfs time or on a live system.
3501 Severity : enhancement
3503 Description: Lustre message v2
3504 Details : Add lustre message format v2.
3506 Severity : enhancement
3508 Description: client OST exclusion list
3509 Details : Clients can be started with a list of OSTs that should be
3510 declared "inactive" for known non-responsive OSTs.
3514 Description: SPEC SFS validation failure on NFS v2 over lustre.
3515 Details : Changes the blocksize for regular files to be 2x RPC size,
3516 and not depend on stripe size.
3518 Severity : enhancement
3520 Description: Multiple MD RPCs in flight.
3521 Details : Further unserialise some read-only MDS RPCs - learn about intents.
3522 To avoid overly-overloading MDS, introduce a limit on number of
3523 MDS RPCs in flight for a single client and add /proc controls
3524 to adjust this limit.
3526 Severity : enhancement
3528 Description: client read/write statistics
3529 Details : Add client read/write call usage stats for performance
3530 analysis of user processes.
3531 /proc/fs/lustre/llite/*/offset_stats shows non-sequential
3532 file access. extents_stats shows chunk size distribution.
3533 extents_stats_per_process show chunk size distribution per
3536 Severity : enhancement
3538 Description: mds statistics
3539 Details : Add detailed mds operations statistics in
3540 /proc/fs/lustre/mds/*/stats.
3544 Description: Failure of copying files with lustre special EAs.
3545 Details : Client side always return success for setxattr call for lustre
3546 special xattr (currently only "trusted.lov").
3551 Description: Refcount LNET uuids
3552 Details : The global LNET uuid list grew linearly with every startup;
3553 refcount repeated list entries instead of always adding to
3556 Severity : enhancement
3558 Description: Dynamic service threads
3559 Details : Within a small range, start extra service threads
3560 automatically when the request queue builds up.
3563 Frequency : mixed-endian client/server environments
3565 Description: mixed-endian crashes
3566 Details : The new msg_v2 system had some failures in mixed-endian
3569 Severity : enhancement
3571 Description: Easy OST removal
3572 Details : OSTs can be permanently deactivated with e.g. 'lctl
3573 conf_param lustre-OST0001.osc.active=0'
3575 Severity : enhancement
3577 Description: MGS proc entries
3578 Details : Added basic proc entries for the MGS showing what filesystems
3581 Severity : enhancement
3583 Description: provide MGS failover
3584 Details : Added config lock reacquisition after MGS server failover.
3586 Severity : enhancement
3588 Description: add Linux 2.4 support
3589 Details : Added support for RHEL 2.4.21 kernel for 1.6 servers and clients
3593 Description: a large application tries to do I/O to the same resource and dies
3594 in the middle of it.
3595 Details : Check the req->rq_arrival time after the call to
3596 ost_brw_lock_get(), but before we do anything about
3597 processing it & sending the BULK transfer request. This
3598 should help move old stale pending locks off the queue as
3599 quickly as obd_timeout.
3602 Frequency : when an incorrect nid is specified during startup
3604 Description: ptlrpc connect to non-existant node causes kernel crash
3605 Details : LNET can't be re-entered from an event callback, which
3606 happened when we expire a message after the export has been
3607 cleaned up. Instead, hand the zombie cleanup off to another
3610 Severity : enhancement
3612 Description: plain/inodebits lock performance improvement
3613 Details : Grouping plain/inodebits in granted list by their request modes
3614 and bits policy, thus improving the performance of search through
3618 Frequency : only if OST filesystem is corrupted
3620 Description: client incorrectly hits assertion in ptlrpc_replay_req()
3621 Details : for a short time RPCs with bulk IO are in the replay list,
3622 but replay of bulk IOs is unimplemented. If the OST filesystem
3623 is corrupted due to disk cache incoherency and then replay is
3624 started it is possible to trip an assertion. Avoid putting
3625 committed RPCs into the replay list at all to avoid this issue.
3628 Frequency : liblustre (e.g. catamount) on a large cluster with >= 8 OSTs/OSS
3630 Description: System hang on startup
3631 Details : This bug allowed the liblustre (e.g. catamount) client to
3632 return to the app before handling all startup RPCs. This
3633 could leave the node unresponsive to lustre network traffic
3634 and manifested as a server ptllnd timeout.
3636 Severity : enhancement
3638 Description: Add "/proc/sys/lustre/debug_peer_on_timeout"
3639 Details : liblustre envirable: LIBLUSTRE_DEBUG_PEER_ON_TIMEOUT
3640 boolean to control whether to print peer debug info when a
3641 client's RPC times out.
3644 Frequency : only for kernels with patches from Lustre below 1.4.3
3646 Description: Remove old rdonly API
3647 Details : Remove old rdonly API which unused from at least lustre 1.4.3
3650 Frequency : only for devices with external journals
3652 Description: Set external device read-only also
3653 Details : During a commanded failover stop, we set the disk device
3654 read-only while the server shuts down. We now also set any
3655 external journal device read-only at the same time.
3658 Frequency : when upgrading from 1.4 while trying to change parameters
3660 Description: The wrong (new) MDC name was used when setting parameters for
3661 upgraded MDT's. Also allows changing of OSC (and MDC)
3662 parameters if --writeconf is specified at tunefs upgrade time.
3665 Frequency : when setting specific ost indicies
3667 Description: QOS code breaks on skipped indicies
3668 Details : Add checks for missing OST indicies in the QOS code, so OSTs
3669 created with --index need not be sequential.
3671 Severity : enhancement
3673 Description: Add uninit_groups feature to ldiskfs2 to speed up e2fsck
3674 Details : The uninit_groups feature works in conjunction with the kernel
3675 filesystem code (ldiskfs2 only) and e2fsprogs-1.39-cfs6 to speed
3676 up the pass1 processing of e2fsck. This is a read-only feature
3677 in ldiskfs2 only, so older kernels and current ldiskfs cannot
3678 mount filesystems that have had this feature enabled.
3680 Severity : enhancement
3682 Description: Improve multi-block allocation algorithm to avoid fragmentation
3683 Details : The mballoc3 code (ldiskfs2 only) adds new mechanisms to improve
3684 allocation locality and avoid filesystem fragmentation.
3686 ------------------------------------------------------------------------------
3688 2007-04-01 Cluster File Systems, Inc. <info@clusterfs.com>
3690 * Support for kernels:
3691 2.4.21-47.0.1.EL (RHEL 3)
3692 2.6.5-7.283 (SLES 9)
3693 2.6.9-42.0.10.EL (RHEL 4)
3694 2.6.12.6 vanilla (kernel.org)
3695 2.6.16.27-0.9 (SLES 10)
3696 * Recommended e2fsprogs version: 1.39.cfs5
3698 * Note that reiserfs quotas are disabled on SLES 10 in this kernel
3702 Frequency : occasional, depends on client load and configuration
3703 Bugzilla : 12181, 12203
3704 Description: data loss for recently-modified files
3706 Details : In some cases it is possible that recently written or created
3707 files may not be written to disk in a timely manner (this should
3708 normally be within 30s unless client IO load is very high).
3709 The problem appears as zero-length files or files that are a
3710 multiple of 1MB in size after a client crash or client eviction
3711 that are missing data at the end of the file.
3713 This problem is more likely to be hit on clients where files are
3714 repeatedly created and unlinked in the same directory, clients
3715 have a large amount of RAM, have many CPUs, the filesystem has
3716 many OSTs, the clients are rebooted frequently, and/or the files
3717 are not accessed by other nodes after being written.
3719 The presence of the problem can be detected by looking at
3720 /proc/sys/fs/inode-state. If the first number (nr_inodes) is
3721 smaller than the second (nr_unused) then dirty files will not
3722 be flushed automatically to disk. "sync; sleep 10" should be
3723 run several times on the node before unmounting it to update
3724 Lustre (this is also safe to run on nodes without this problem).
3726 There is also a related kernel bug in the RHEL4 4 2.6.9 kernel
3727 that can cause this same problem, so customers using that kernel
3728 also need to update the kernel in addition to Lustre. In order
3729 to properly fix this bug, the RHEL3 2.4.21 kernel is also updated.
3731 It is normal that files written just before a client crash (less
3732 than 30s) may not yet have been flushed to disk, even for local
3736 Frequency : frequent on thin XT3 nodes
3738 Description: UUID collision on thin XT3 Linux nodes
3739 Details : UUIDs on Compute Node Linux XT3 nodes were not generated
3740 randomly, since we relied on an insufficiently-seeded PRNG.
3745 Description: OSS hangs after "All ost request buffers busy"
3746 Details : A deadlock between quota and journal operations caused OSS
3747 hangs after printing "All ost request buffers busy."
3750 Frequency : always on liblustre builds
3752 Description: Cleanup compiler warnings on liblustre
3755 Frequency : always on liblustre builds on XT3
3757 Description: LC_CONFIG_CDEBUG don't run while build liblustre on XT3.
3761 Description: Addition of EXT3_FEATURE_RO_COMPAT_DIR_NLINKS flag for
3762 > 32000 subdirectories
3763 Details : Add EXT3_FEATURE_RO_COMPAT_DIR_NLINK flag to
3764 EXT3_FEATURE_RO_COMPAT_SUPP. This flag will be set whenever
3765 subdirectory count crosses 32000. This will aid e2fsck to
3766 correctly handle more than 32000 subdirectories.
3769 Frequency : liblustre (e.g. catamount) on a large cluster with >= 8 OSTs/OSS
3771 Description: System hang on startup
3772 Details : This bug allowed the liblustre (e.g. catamount) client to
3773 return to the app before handling all startup RPCs. This
3774 could leave the node unresponsive to lustre network traffic
3775 and manifested as a server ptllnd timeout.
3777 Severity : enhancement
3779 Description: Add "/proc/sys/lustre/debug_peer_on_timeout"
3780 (liblustre envirable: LIBLUSTRE_DEBUG_PEER_ON_TIMEOUT)
3781 boolean to control whether to print peer debug info when a
3782 client's RPC times out.
3787 Description: make O_SYNC working on 2.6 kernels
3788 Details : 2.6 kernels use different method for mark pages for write,
3789 so need add a code to lustre for O_SYNC work.
3794 Description: Failure to close file and release space on NFS
3795 Details : Put inode details into lock acquired in ll_intent_file_open.
3796 Use mdc_intent_lock in ll_intent_open to properly
3797 detect all kind of errors unhandled by mdc_enqueue.
3802 Description: proc file read during shutdown sometimes raced obd removal,
3804 Details : Add lock to prevent obd access after proc file removal.
3807 Frequency : Only for files larger than 4GB on 32-bit clients.
3809 Description: improperly doing page alignment of locks
3810 Details : Modify lustre core code to use CFS_PAGE_* defines instead of
3811 PAGE_*. Make CFS_PAGE_MASK a 64-bit mask.
3816 Description: RPCs being resent when they shouldn't be
3817 Details : Some RPCs that should not be resent are being resent. This
3818 can cause inconsistencies in the RPC state machine. Do not
3819 resend such requests.
3822 Frequency : rare, only with NFS export
3824 Description: Crash on NFS re-export node
3825 Details : under very unusual load conditions an assertion is hit in
3826 ll_intent_file_open()
3829 Frequency : only if OST filesystem is corrupted
3831 Description: client incorrectly hits assertion in ptlrpc_replay_req()
3832 Details : for a short time RPCs with bulk IO are in the replay list,
3833 but replay of bulk IOs is unimplemented. If the OST filesystem
3834 is corrupted due to disk cache incoherency and then replay is
3835 started it is possible to trip an assertion. Avoid putting
3836 committed RPCs into the replay list at all to avoid this issue.
3841 Description: large O_DIRECT requests fail under memory pressure/fragmentation
3842 Details : Large single O_DIRECT read and write calls can fail to allocate
3843 a sufficiently large buffer to process the request. In case of
3844 allocation failure the allocation is retried with a smaller
3845 buffer and broken into smaller requests.
3847 Severity : enhancement
3849 Description: Add -o localflock option to simulate old noflock behaviour.
3850 Details : This will achieve local-only flock/fcntl locks coherentness.
3855 Description: versioning check is incomplete
3856 Details : Checking the version difference of client vs. server, report
3857 error if the gap is too big.
3862 Description: add support PG_writeback bit
3863 Details : add support for PG_writeback bit for Lustre, for more carefull
3864 work with page cache in 2.6 kernel. This also fix some deadlocks
3865 and remove hack for work O_SYNC with 2.6 kernel.
3867 Severity : enhancement
3869 Description: Add uninit_groups feature to ldiskfs2 to speed up e2fsck
3870 Details : The uninit_groups feature works in conjunction with the kernel
3871 filesystem code (ldiskfs2 only) and e2fsprogs-1.39-cfs6 to speed
3872 up the pass1 processing of e2fsck. This is a read-only feature
3873 in ldiskfs2 only, so older kernels and current ldiskfs cannot
3874 mount filesystems that have had this feature enabled.
3876 Severity : enhancement
3878 Description: Improve multi-block allocation algorithm to avoid fragmentation
3879 Details : The mballoc3 code (ldiskfs2 only) adds new mechanisms to improve
3880 allocation locality and avoid filesystem fragmentation.
3882 ------------------------------------------------------------------------------
3884 2007-02-09 Cluster File Systems, Inc. <info@clusterfs.com>
3886 * Support for kernels:
3887 2.6.9-42.0.3.EL (RHEL 4)
3888 2.6.5-7.276 (SLES 9)
3889 2.4.21-47.0.1.EL (RHEL 3)
3890 2.6.12.6 vanilla (kernel.org)
3891 2.6.16.21-0.8 (SLES10)
3892 * Recommended e2fsprogs version: 1.39.cfs2-0
3894 * The backwards-compatible /proc/sys/portals symlink has been removed
3895 in this release. Before upgrading, please ensure that you change
3896 any configuration scripts or /etc/sysctl.conf files that access
3897 /proc/sys/portals/* or sysctl portals.* to use the corresponding
3898 entry in /proc/sys/lnet or sysctl lnet.*. This change can be made
3899 in advance of the upgrade on any system running Lustre 1.4.6 or
3900 newer, since /proc/sys/lnet was added in that version.
3901 * Note that reiserfs quotas are disabled on SLES 10 in this kernel
3905 Frequency : only when quota is used
3907 Description: avoid scanning export list for quota master
3908 Details : Change the algorithms to avoid scanning export list in order
3909 to improve the efficiency.
3912 Frequency : MDS failover only, very rarely
3914 Description: "went back in time" messages on mds failover
3915 Details : The greatest transno may be lost when the current operation
3916 finishes with an error (transno==0) and the client's last_rcvd
3917 record is over-written. Save the greatest transno in the
3918 mds_last_transno for this case.
3921 Frequency : always for specific kernels and striping counts
3923 Description: client may get "Matching packet too big" without ACL support
3924 Details : Clients compiled without CONFIG_FS_POSIX_ACL get an error message
3925 when trying to access files in certain configurations. The
3926 clients should in fact be denied when mounting because they do
3927 not understand ACLs.
3930 Frequency : Cray XT3 with more than 4000 clients and multiple jobs
3932 Description: many clients connecting with IO in progress causes connect timeouts
3933 Details : Avoid synchronous journal commits to avoid delays caused by many
3934 clients connecting/disconnecting when bulk IO is in progress.
3935 Queue liblustre connect requests on OST_REQUEST_PORTAL instead of
3936 OST_IO_PORTAL to avoid delays behind potentially many pending
3940 Frequency : occasionally with multiple writers to a single file
3942 Description: shared writes to file may result in wrong size reported by stat()
3943 Details : Allow growing of kms when extent lock is cancelled
3946 Frequency : always with random mmap IO to multi-striped file
3948 Description: mmap write might be lost if we are writing to a 'hole' in stripe
3949 Details : Only if the hole is at the end of OST object so that kms is too
3950 small. Fix is to increase kms accordingly in ll_nopage.
3953 Frequency : rare, only if OST filesystem is inconsistent with MDS filesystem
3955 Description: writes to a missing object would leak memory on the OST
3956 Details : If there is an inconsistency between the MDS and OST filesystems,
3957 such that the MDS references an object that doesn't exist, writes
3958 to that object will leak memory due to incorrect cleanup in the
3959 error handling path, eventually running out of memory on the OST.
3964 Description: Creating too long symlink causes lustre errors
3965 Details : Check symlink and name lengths before sending requests to MDS.
3968 Frequency : only if flock is enabled (not on by default)
3970 Description: posix locks not released on fd closure on 2.6.9+
3971 Details : We failed to add posix locks to list of inode locks on 2.6.9+
3972 kernels, this caused such locks not to be released on fd close and
3973 then assertions on fs unmount about still used locks.
3976 Frequency : MDS failover only, very rarely
3978 Description: clients may get ASSERTION(granted_lock != NULL)
3979 Details : When request was taking a long time, and a client was resending
3980 a getattr by name lock request. The were multiple lock requests
3981 with the same client lock handle and
3982 mds_getattr_name->fixup_handle_for_resent_request found one of the
3983 lock handles but later failed with ASSERTION(granted_lock != NULL).
3988 Description: handle->h_buffer_credits > 0, assertion failure
3989 Details : h_buffer_credits is zero after truncate, causing assertion
3990 failure. This patch extends the transaction or creates a new
3994 Frequency : NFS re-export or patchless client
3995 Bugzilla : 11179, 10796
3996 Description: Crash on NFS re-export node (__d_move)
3997 Details : We do not want to hash the dentry if we don't have a lock.
3998 But if this dentry is later used in d_move, we'd hit uninitialised
3999 list head d_hash, so we just do this to init d_hash field but
4000 leave dentry unhashed.
4003 Frequency : NFS re-export or patchless client
4005 Description: NFS exports has problem with symbolic link
4006 Details : lustre client didn't properly install dentry when re-exported
4007 to NFS or running patchless client.
4010 Frequency : NFS re-export or patchless client
4012 Description: Various nfs/patchless fixes.
4013 Details : fixes reuse disconected alias for lookup process - this fixes
4014 warning "find_exported_dentry: npd != pd",
4015 fix permission error with open files at nfs.
4016 fix apply umask when do revalidate.
4019 Frequency : occasional
4021 Description: Crash on NFS re-export node
4022 Details : calling clear_page() on the wrong pointer triggered oops in
4023 generic_mapping_read().
4026 Frequency : rarely, using O_DIRECT IO
4028 Description: unaligned directio crashes client with LASSERT
4029 Details : check for unaligned buffers before trying any requests.
4032 Frequency : rarely, using CFS RAID5 patches in non-standard kernel series
4034 Description: stale data returned from RAID cache
4035 Details : If only a small amount of IO is done to the RAID device before
4036 reading it again it is possible to get stale data from the RAID
4037 cache instead of reading it from disk.
4040 Frequency : always for sles10 kernel
4042 Description: sles10 support
4043 Details : ll_follow_link: compile fixes and using of nd_set_link
4044 under newer kernels.
4047 Frequency : depends on arch, kernel and compiler version, always on sles10
4050 Description: recursive or deep enough symlinks cause stack overflow
4051 Details : getting rid of large stack-allocated variable in
4055 Frequency : depends on hardware
4057 Description: lustre write performance loss in the SLES10 kernel
4058 Details : the performance loss is caused by using of write barriers in the
4059 ext3 code. The SLES10 kernel turns barrier support on by
4060 default. The fix is to undo that change for ldiskfs.
4062 ------------------------------------------------------------------------------
4064 2006-12-09 Cluster File Systems, Inc. <info@clusterfs.com>
4066 * Support for kernels:
4067 2.6.9-42.0.3EL (RHEL 4)
4068 2.6.5-7.276 (SLES 9)
4069 2.4.21-47.0.1.EL (RHEL 3)
4070 2.6.12.6 vanilla (kernel.org)
4074 Frequency : quota enabled and large files being deleted
4076 Description: releasing more than 4GB of quota at once hangs OST
4077 Details : If a user deletes more than 4GB of files on a single OST it
4078 will cause the OST to spin in an infinite loop. Release
4079 quota in < 4GB chunks, or use a 64-bit value for 1.4.7.1+.
4084 Description: statfs data retrieved from /proc may be stale or zero
4085 Details : When reading per-device statfs data from /proc, in the
4086 {kbytes,files}_{total,free,avail} files, it may appear
4087 as zero or be out of date.
4090 Frequency : systems with MD RAID1 external journal devices
4092 Description: lconf's call to blkid is confused by RAID1 journal devices
4093 Details : Use the "blkid -l" flag to locate the MD RAID device instead
4094 of returning all block devices that match the journal UUID.
4097 Frequency : always, for aggregate stripe size over 4GB
4099 Description: "lfs setstripe" fails assertion when setting 4GB+ stripe width
4100 Details : Using "lfs setstripe" to set stripe size * stripe count over 4GB
4101 will fail the kernel with "ASSERTION(lsm->lsm_xfersize != 0)"
4104 Frequency : always if "lfs find" used on a local file/directory
4106 Description: "lfs find" segfaults if used on a local file/directory
4107 Details : The case where a directory component was not specified wasn't
4108 handled correctly. Handle this properly.
4111 Frequency : always on ppc64
4113 Description: the write to an ext3 filesystem mounted with mballoc got stuck
4114 Details : ext3_mb_generate_buddy() uses find_next_bit() which does not
4115 perform endianness conversion.
4118 Frequency : rarely (truncate to non-zero file size after write under load)
4119 Bugzilla : 10730, 10687
4120 Description: Files padded with zeros to next 4K multiple
4121 Details : With filesystems mounted using the "extents" option (2.6 kernels)
4122 it is possible that files that are truncated to a non-zero size
4123 immediately after being written are filled with zero bytes beyond
4124 the truncated size. No file data is lost.
4126 Severity : enhancement
4128 Description: Allow recovery/failover for liblustre clients.
4129 Details : liblustre clients were unaware of failover configurations until
4132 Severity : enhancement
4134 Description: user file locks should fail when not mounting with flock option
4135 Details : Set up an error-returning stub in ll_file_operations.lock field
4136 to prevent incorrect behaviour when client is mounted without
4137 flock option. Also, set up properly f_op->flock field for
4141 Frequency : always on ia64
4143 Description: "lfs df" loops on printing out MDS statfs information
4144 Details : The obd_ioctl_data was not initialized and in some systems
4145 this caused a failure during the ioctl that did not return
4146 an error. Initialize the struct and return an error on failure.
4149 Frequency : SLES 9 only
4151 Description: Error of copying files with lustre special EAs as root
4152 Details : Client side always return success for setxattr call for lustre
4153 special xattr (currently only "trusted.lov").
4156 Frequency : rarely on clusters with both ia64+i386 clients
4158 Description: ia64+i686 clients doing shared IO on the same file may LBUG
4159 Details : In rare cases when both ia64+i686 (or other mixed-PAGE_SIZE)
4160 clients are doing concurrent writes to the same file it is
4161 possible that the ia64 clients may LASSERT because the OST
4162 extent locks are not PAGE_SIZE aligned. Ensure that grown
4163 locks are always aligned on the request boundary.
4166 Frequency : specific use, occasional
4168 Description: Overwriting in use executable truncates on-disk binary image
4169 Details : If one node attempts to overwrite an executable in use by
4170 another node, we now correctly return ETXTBSY instead of
4171 truncating the file.
4173 Severity : enhancement
4175 Description: Async OSC create to avoid the blocking unnecessarily.
4176 Details : If an OST has no remaining object, system will block on the creating
4177 when need to create a new object on this OST. Now, ways use
4178 pre-created objects when available, instead of blocking on an
4179 empty osc while others are not empty. If we must block, we block
4180 for the shortest possible period of time.
4185 Description: chmod on Lustre root is propagated to other clients
4186 Details : Re-validate root's dentry in ll_lookup_it to avoid having it
4187 invalid by the follow_mount time.
4190 Frequency : liblustre clients only
4192 Description: Race in 'instant cancel' lock handling could lead to such locks
4193 never to be granted in case of SMP MDS
4194 Details : Do not destroy not yet granted but cbpending locks in
4198 Frequency : replay/resend of open
4200 Description: non null lock assetion failure in mds_intent_policy
4201 Details : Trying to replay/resend lockless open requests resulted in
4202 mds_open() returning 0 with no lock. Now it sets a flag if
4203 a lock is going to be returned.
4205 Severity : enhancement
4207 Description: Checksum enhancements
4208 Details : New checksum enhancements allow for resending RPCs that failed
4211 Severity : enhancement
4213 Description: Tunables on number of dirty pages in cacche
4214 Details : Allow to set limit on number of dirty pages cached.
4219 Description: client crash on unmount - lock still has references
4220 Details : In some error handling cases it was possible to leak a lock
4221 reference on a client while accessing a file. This was not
4222 harmful to the client during operation, but would cause the
4223 client to crash when the filesystem is unmounted.
4226 Frequency : specific case, rare
4228 Description: ETXTBSY on mds though file not in use by client
4229 Details : ETXTBSY is no longer incorrectly returned when attempting to
4230 chmod or chown a directory that the user previously tried to
4231 execute or a currently-executing binary.
4234 Frequency : extremely rare except on liblustre-based clients
4236 Description: Lustre space not freed when files are deleted
4237 Details : Clean up open-unlinked files after client eviction. Previously
4238 the unlink was skipped and the files remained as orphans.
4243 Description: OST failure "would be an LBUG" in waiting_locks_callback()
4244 Details : In some cases it was possible to send a blocking callback to
4245 a client doing a glimpse, even though that client didn't get
4246 a lock granted. When the glimpse lock is cancelled on the OST
4247 the freed lock is left on the waiting list and corrupted the list.
4250 Frequency : all core dumps
4252 Description: Broke core dumps to lustre
4253 Details : Negative dentry may be unhashed if parent does not have UPDATE
4254 lock, but some callers, e.g. do_coredump, expect dentry to be
4255 hashed after successful create, hash it in ll_create_it.
4257 ------------------------------------------------------------------------------
4259 2006-09-13 Cluster File Systems, Inc. <info@clusterfs.com>
4261 * Support for kernels:
4262 2.6.9-42.0.2.EL (RHEL 4)
4263 2.6.5-7.276 (SLES 9)
4264 2.4.21-40.EL (RHEL 3)
4265 2.6.12.6 vanilla (kernel.org)
4269 Frequency : always on RHEL 3
4271 Description: Number of open files grows over time
4272 Details : The number of open files grows over time, whether or not
4273 Lustre is started. This was due to a filp leak introduced
4274 by one of our kernel patches.
4276 ------------------------------------------------------------------------------
4278 08-20-2006 Cluster File Systems, Inc. <info@clusterfs.com>
4280 * Support for kernels:
4281 2.6.9-42.EL (RHEL 4)
4282 2.6.5-7.276 (SLES 9)
4283 2.4.21-40.EL (RHEL 3)
4284 2.6.12.6 vanilla (kernel.org)
4289 Bugzilla : 5719, 9635, 9792, 9684
4290 Description: OST (or MDS) trips assertions in (re)connection under heavy load
4291 Details : If a server is under heavy load and cannot reply to new
4292 connection requests before the client resends the (re)connect,
4293 the connection handling code can behave badly if two service
4294 threads are concurrently handing separate (re)connections from
4295 the same client. Add better locking to the connection handling
4296 code, and ensure that only a single connection will be processed
4297 for a given client UUID, even if the lock is dropped.
4299 Severity : enhancement
4301 Description: add TCP zero-copy support to kernel
4302 Details : Add support to the kernel TCP stack to allow zero-copy bulk
4303 sends if the hardware supports scatter-gather and checksumming.
4304 This allows socklnd to do client-write and server-read more
4305 efficiently and reduce CPU utilization from skbuf copying.
4308 Frequency : only if NFS exporting from client
4310 Description: NULL pointer deref in ll_iocontrol() if chattr mknod file
4311 Details : If setting attributes on a file created under NFS that had
4312 never been opened it would be possible to oops the client
4313 if the file had no objects.
4316 Frequency : always for liblustre
4318 Description: liblustre client does MDS+OSTs setattr RPC for each write
4319 Details : When doing a write from a liblustre client, the client
4320 incorrectly issued an RPC to the MDS and each OST the file was
4321 striped over in order to update the timestamps. When writing
4322 with small chunks and many clients this could overwhelm the MDS
4323 with RPCs. In all cases it would slow down the write because
4324 these RPCs are unnecessary.
4326 Severity : enhancement
4328 Description: allow number of MDS service threads to be changed at module load
4329 Details : It is now possible to change the number of MDS service threads
4330 running. Adding "options mds mds_num_threads={N}" to the MDS's
4331 /etc/modprobe.conf will set the number of threads for the next
4332 time Lustre is restarted (assuming the "mds" module is also
4333 reloaded at that time). The default number of threads will
4334 stay the same, 32 for most systems.
4339 Description: OST crash if filesystem is unformatted or corrupt
4340 Details : If an OST is started on a device that has never been formatted
4341 or if the filesystem is corrupt and cannot even mount then the
4342 error handling cleanup routines would dereference a NULL pointer.
4347 Description: NULL pointer deref in llap_from_page.
4348 Details : get_cache_page_nowait can return a page with NULL (or otherwise
4349 incorrect) mapping if the page was truncated/reclaimed while it was
4350 searched for. Check for this condition and skip such pages when
4351 doing readahead. Introduce extra check to llap_from_page() to
4352 verify page->mapping->host is non-NULL (so page is not anonymous).
4355 Frequency : Sometimes when using sys_sendfile
4357 Description: "page not covered by a lock" warnings from ll_readpage
4358 Details : sendfile called ll_readpage without right page locks present.
4359 Now we introduced ll_file_sendfile that does necessary locking
4360 around call to generic_file_sendfile() much like we do in
4364 Frequency : with certain MDS communication failures at client mount time
4366 Description: NULL pointer deref after failed client mount
4367 Details : a client connection request may delayed by the network layer
4368 and not be sent until after the PTLRPC layer has timed out the
4369 request. If the client fails the mount immediately it will try
4370 to clean up before the network times out the request. Add a
4371 reference from the request import to the obd device and delay
4372 the cleanup until the network drops the request.
4375 Frequency : occasionally during client (re)connect
4377 Description: assertion failure during client (re)connect
4378 Details : processing a client connection request may be delayed by the
4379 client or server longer than the client connect timeout. This
4380 causes the client to resend the connection request. If the
4381 original connection request is replied in this interval, the
4382 client may trip an assertion failure in ptlrpc_connect_interpret()
4383 which thought it would be the only running connect process.
4386 Frequency : only with obd_echo servers and clients that are rebooted
4388 Description: kernel BUG accessing uninitialized data structure
4389 Details : When running an obd_echo server it did not start the ping_evictor
4390 thread, and when a client was evicted an uninitialized data
4391 structure was accessed. Start the ping_evictor in the RPC
4392 service startup instead of the OBD startup.
4394 Severity : enhancement
4395 Bugzilla : 10193 (patchless)
4396 Description: Remove dependency on various unexported kernel interfaces.
4397 Details : No longer need reparent_to_init, exit_mm, exit_files,
4398 sock_getsockopt, filemap_populate, FMODE_EXEC, put_filp.
4401 Frequency : rare (only users of deprecated and unsupported LDAP config)
4403 Description: write_conf for zeroconf mount queried LDAP incorrectly for client
4404 Details : LDAP apparently contains 'lustreName' attributes instead of
4405 'name'. A simple remapping of the name is sufficient.
4408 Frequency : rare (only with non-default dump_on_timeout debug enabled)
4410 Description: waiting_locks_callback trips kernel BUG if client is evicted
4411 Details : Running with the dump_on_timeout debug flag turned on makes
4412 it possible that the waiting_locks_callback() can try to dump
4413 the Lustre kernel debug logs from an interrupt handler. Defer
4414 this log dumping to the expired_lock_main() thread.
4416 Severity : enhancement
4418 Description: Support NFS exporting on 2.6 kernels.
4419 Details : Implement non-rawops metadata methods for NFS server to use without
4420 changing NFS server code.
4423 Frequency : very rare (synthetic metadata workload only)
4425 Description: two racing renames might cause an MDS thread to deadlock
4426 Details : Running the "racer" program may cause one MDS thread to rename
4427 a file from being the source of a rename to being the target of
4428 a rename at exactly the same time that another thread is doing
4429 so, and the second thread has already enqueued these locks after
4430 doing a lookup of the target and is trying to relock them in
4431 order. Ensure that we don't try to re-lock the same resource.
4434 Frequency : only very large systems with liblustre clients
4436 Description: slow eviction of liblustre clients with the "evict_by_nid" RPC
4437 Details : Use asynchronous set_info RPCs to send the "evict_by_nid" to
4438 all OSTs in parallel. This allows the eviction of stale liblustre
4439 clients to proceed much faster than if they were done in series,
4440 and also offers similar improvements for other set_info RPCs.
4445 Description: excessive CPU usage during initial read phase on client
4446 Details : During the initial read phase on a client, it would agressively
4447 retry readahead on the file, consuming too much CPU and impacting
4448 performance (since 1.4.5.8). Improve the readahead algorithm
4449 to avoid this, and also improve some other common cases (read
4450 of small files in particular, where "small" is files smaller than
4451 /proc/fs/lustre/llite/*/max_read_ahead_whole_mb, 2MB by default).
4456 Description: MDS crash when receiving packet with unknown intent.
4457 Details : Do not LBUG in unknown intent case, just return -EFAULT
4459 Severity : enhancement
4460 Bugzilla : 9293, 9385
4461 Description: MDS RPCs are serialised on client. This is unnecessary for some.
4462 Details : Do not serialize getattr (non-intent version) and statfs.
4465 Frequency : occasional, when OST network is overloaded/intermittent
4467 Description: client evicted by OST after bulk IO timeout
4468 Details : If a client sends a bulk IO request (read or write) the OST
4469 may evict the client if it is unresposive to its data GET/PUT
4470 request. This is incorrect if the network is overloaded (takes
4471 too long to transfer the RPC data) or dropped the OST GET/PUT
4472 request. There is no need to evict the client at all, since
4473 the pinger and/or lock callbacks will handle this, and the
4474 client can restart the bulk request.
4477 Frequency : Always when mmapping file with no objects
4479 Description: client crashes when mmapping file with no objects
4480 Details : Check that we actually have objects in a file before doing any
4481 operations on objects in ll_vm_open, ll_vm_close and
4487 Description: Request leak when working with deleted CWD
4488 Details : Introduce advanced request refcount tracking for requests
4489 referenced from lustre intent.
4491 Severity : Enhancement
4493 Description: Cache open file handles on client.
4494 Details : MDS now will return special lock along with openhandle, if
4495 requested and client is allowed to hold openhandle, even if unused,
4496 until such a lock is revoked. Helps NFS a lot, since NFS is opening
4497 closing files for every read/write openration.
4499 Severity : Enhancement
4501 Description: Cache open negative dentries on client when possible.
4502 Details : Guard negative dentries with UPDATE lock on parent dir, drop
4503 negative dentries on lock revocation.
4508 Description: Remounting a client read-only wasn't possible with a zconf mount
4509 Details : It wasn't possible to remount a client read-only with llmount.
4511 Severity : enhancement
4512 Description: Include MPICH 1.2.6 Lustre ADIO interface patch
4513 Details : In lustre/contrib/ or /usr/share/lustre in RPM a patch for
4514 MPICH is included to add Lustre-specific ADIO interfaces.
4515 This is based closely on the UFS ADIO layer and only differs
4516 in file creation, in order to allow the OST striping to be set.
4517 This is user-contributed code and not supported by CFS.
4522 Description: extended inode attributes (immutable, append-only) work improperly
4523 when 2.4 and 2.6 kernels are used on client/server or vice versa
4524 Details : Introduce kernel-independent values for these flags.
4526 Severity : enhancement
4529 Description: Allow fractional MB tunings for lustre in /proc/ filesystem.
4530 Details : Many of the /proc/ tunables can only be tuned at a megabyte
4531 granularity. Now, Fractional MB granularity is be supported,
4532 this is very useful for low memory system.
4534 Severity : enhancement
4536 Description: Getattr by fid
4537 Details : Getting a file attributes by its fid, obtaining UPDATE|LOOKUP
4538 locks, avoids extra getattr rpc requests to MDS, allows '/' to
4539 have locks and avoids getattr rpc requests for it on every stat.
4542 Frequency : Always, for filesystems larger than 2TB
4544 Description: ldiskfs crash at mount for filesystem larger than 2TB with mballoc
4545 Details : Kenrel kmalloc limits allocations to 128kB and this prevents
4546 filesystems larger than 2TB to be mounted with mballoc enabled.
4549 Frequency : Always, for 32-bit kernel without CONFIG_LBD and filesystem > 2TB
4551 Description: ldiskfs crash at mount for filesystem larger than 2TB with mballoc
4552 Details : If a 32-bit kernel is compiled without CONFIG_LBD enabled and a
4553 filesystems larger than 2TB is mounted then the kernel will
4554 silently corrupt the start of the filesystem. CONFIG_LBD is
4555 enabled for all CFS-supported kernels, but the possibility of
4556 this happening with a modified kernel config exists.
4558 Severity : enhancement
4560 Description: add client O_DIRECT support for 2.6 kernels
4561 Details : It is now possible to do O_DIRECT reads and writes to files
4562 in the Lustre client mountpoint on 2.6 kernel clients.
4564 Severity : enhancement
4566 Description: parallel glimpse, setattr, statfs, punch, destroy requests
4567 Details : Sends glimpse, setattr, statfs, punch, destroy requests to OSTs in
4568 parallel, not waiting for response from every OST before sending
4569 a rpc to the next OST.
4574 Description: setattr vs write race when updating file timestamps
4575 Details : Client processes that update a file timestamp into the past
4576 right after writing to the file (e.g. tar) it is possible that
4577 the updated file modification time can be reset to the current
4578 time due to a race between processing the setattr and write RPC.
4580 Severity : enhancement
4582 Description: Bring 'lfs find' closer in line with regular Linux find.
4583 Details : lfs find util supports -atime, -mtime, -ctime, -maxdepth, -print,
4584 -print0 options and obtains all the needed info through the lustre
4587 Severity : enhancement
4589 Description: support up to 1024 configured devices on one node
4590 Details : change obd_dev array from statically allocated to dynamically
4591 allocated structs as they are first used to reduce memory usage
4596 Description: Flush dirty partially truncated pages during truncate
4597 Details : Immediatelly flush partially truncated pages in filter_setattr,
4598 this way we completely avoid having any pages in page cache on OST
4599 and can retire ugly workarounds during writes to flush such pages.
4604 Description: i_sem vs transaction deadlock in mds_obd_destroy during unlink.
4605 Details : protect inode from truncation within vfs_unlink() context
4606 just take a reference before calling vfs_unlink() and release it
4607 when parent's i_sem is free.
4612 Description: last_id value checked outside lock on OST caused LASSERT failure
4613 Details : If there were multiple MDS->OST object precreate requests in
4614 flight, it was possible that the OST's last object id was checked
4615 outside a lock and incorrectly tripped an assertion. Move checks
4616 inside locks, and discard old precreate requests.
4619 Frequency : always, if extents are used on OSTs
4621 Description: index ei_leaf_hi (48-bit extension) is not zeroed in extent index
4622 Details : OSTs using the extents format would not zero the high 16 bits of
4623 the index physical block number. This is not a problem for any
4624 OST filesystems smaller than 16TB, and no kernels support ext3
4625 filesystems larger than 16TB yet. This is fixed in 1.4.7 (all
4626 new/modified files) and can be fixed for existing filesystems
4627 with e2fsprogs-1.39-cfs1.
4632 Description: import connection selection may be incorrect if timer wraps
4633 Details : Using a 32-bit jiffies timer with HZ=1000 may cause backup
4634 import connections to be ignored if the 32-bit jiffies counter
4635 wraps. Use a 64-bit jiffies counter.
4638 Frequency : very large clusters immediately after boot
4640 Description: LNET request buffers exhausted under heavy short-term load
4641 Details : If a large number of client requests are generated on a service
4642 that has previously never seen so many requests it is possible
4643 that the request buffer growth cannot keep up with the spike in
4644 demand. Instead of dropping incoming requests, they are held in
4645 the LND until the RPC service can accept more requests.
4648 Frequency : Sometimes during replay
4650 Description: Assertion failure in ll_local_open after replay.
4651 Details : If replay happened on an open request reply before we were able
4652 to set replay handler, reply will become not swabbed tripping the
4653 assertion in ll_local_open. Now we set the handler right after
4654 recognising of open request
4657 Frequency : very rare
4659 Description: kernel reports "badness in vsnprintf"
4660 Details : Reading from the "recovery_status" /proc file in small chunks
4661 may cause a negative length in lprocfs_obd_rd_recovery_status()
4662 call to vsnprintf() (which is otherwise harmless). Exit early
4663 if there is no more space in the output buffer.
4665 Severity : enhancement
4667 Description: clear OBD RPC statistics by writing to them
4668 Details : It is now possible to clear the OBD RPC statistics by writing
4669 to the "stats" file.
4674 Description: Client mtime is not the same on different clients after utimes
4675 Details : In some cases, the client was using the utimes() syscall on
4676 a file cached on another node. The clients now validate the
4677 ctime from the MDS + OSTs to determine which one is right.
4682 Description: Inability to activate failout mode
4683 Details : lconf script incorrectly assumed that in pythong string's numeric
4684 value is used in comparisons.
4687 Frequency : always with multiple stripes per file
4689 Description: Inefficient object allocation for mutli-stripe files
4690 Details : When selecting which OSTs to stripe files over, for files with
4691 a stripe count that divides evenly into the number of OSTs,
4692 the MDS is always picking the same starting OST for each file.
4693 Return the OST selection heuristic to the original design.
4698 Description: mount failures may take full timeout to return an error
4699 Details : Under some heavy load conditions it is possible that a
4700 failed mount can wait for the full obd_timeout interval,
4701 possibly several minutes, before reporting an error.
4702 Instead return an error as soon as the status is known.
4704 Frequency : quota enabled and large files being deleted
4706 Description: releasing more than 4GB of quota at once hangs OST
4707 Details : If a user deletes more than 4GB of files on a single OST it
4708 will cause the OST to spin in an infinite loop. Release
4709 quota in < 4GB chunks, or use a 64-bit value for 1.4.7.1+.
4714 Description: statfs data retrieved from /proc may be stale or zero
4715 Details : When reading per-device statfs data from /proc, in the
4716 {kbytes,files}_{total,free,avail} files, it may appear
4717 as zero or be out of date.
4720 Frequency : systems with MD RAID1 external journal devices
4722 Description: lconf's call to blkid is confused by RAID1 journal devices
4723 Details : Use the "blkid -l" flag to locate the MD RAID device instead
4724 of returning all block devices that match the journal UUID.
4727 Frequency : always, for aggregate stripe size over 4GB
4729 Description: assertion fails when trying to use 4GB stripe size
4730 Details : Use "setstripe" to set stripe size over 4GB will fail the kernel,
4731 complaining "ASSERTION(lsm->lsm_xfersize != 0)"
4734 Frequency : always on ppc64
4736 Description: the first write on an ext3 filesystem with mballoc got stuck
4737 Details : ext3_mb_generate_buddy() uses find_next_bit() which does not
4738 perform endianness conversion.
4740 ------------------------------------------------------------------------------
4742 02-14-2006 Cluster File Systems, Inc. <info@clusterfs.com>
4744 * WIRE PROTOCOL CHANGE. This version of Lustre networking WILL NOT
4745 INTEROPERATE with older versions automatically. Please read the
4746 user documentation before upgrading any part of a live system.
4747 * WARNING: Lustre networking configuration changes are required with
4748 this release. See https://bugzilla.clusterfs.com/show_bug.cgi?id=10052
4751 * Support for kernels:
4752 2.6.9-22.0.2.EL (RHEL 4)
4753 2.6.5-7.244 (SLES 9)
4754 2.6.12.6 vanilla (kernel.org)
4757 Severity : enhancement
4758 Bugzilla : 7981/8208
4759 Description: Introduced Lustre Networking (LNET)
4760 Details : LNET is new networking infrastructure for Lustre, it includes
4761 a reorganized network configuration mode (see the user
4762 documentation for full details) as well as support for routing
4763 between different network fabrics. Lustre Networking Devices
4764 (LNDs) for the supported network fabrics have also been
4765 created for this new infrastructure.
4767 Severity : enhancement
4768 Description: Introduced Access control lists
4769 Details : clients can set ACLs on files and directories in order to have
4770 more fine-grained permissions than the standard Unix UGO+RWX.
4771 The MDS must be started with the "-o acl" mount option.
4773 Severity : enhancement
4774 Description: Introduced filesystem quotas
4775 Details : Administrators may now establish per-user quotas on the
4778 Severity : enhancement
4780 Description: Configuration change for the XT3
4781 The PTLLND is now used to run Lustre over Portals on the XT3
4782 The configure option(s) --with-cray-portals are no longer used.
4783 Rather --with-portals=<path-to-portals-includes> is used to
4784 enable building on the XT3. In addition to enable XT3 specific
4785 features the option --enable-cray-xt3 must be used.
4790 Description: Running on many-way SMP OSTs can trigger oops in llcd_send()
4791 Details : A race between allocating a new llcd and re-getting the llcd_lock
4792 allowed another thread to grab newly-allocated llcd.
4794 Severity : enhancement
4796 Description: 2.6 OST async journal commit and locking fix to improve performance
4797 Details : The filter_direct_io()+filter_commitrw_write() journal commits for
4798 2.6 kernels are now async as they already were in 2.4 kernels so
4799 that they can commit concurrently with the network bulk transfer.
4800 For block-allocated files the filter allocation semaphore is held
4801 to avoid filesystem fragmentation during allocation. BKL lock
4802 removed for 2.6 xattr operations where it is no longer needed.
4807 Description: lconf incorrectly determined whether two IP networks could talk
4808 Details : In some more complicated routing and multiple-network
4809 configurations, lconf will avoid trying to make a network
4810 connection to a disjoint part of the IP space. It was doing the
4811 math incorrectly for one set of cases.
4816 Description: Fix for potential infinite loop processing records in an llog.
4817 Details : If an llog record is corrupted/zeroed, it is possible to loop
4818 forever in llog_process(). Validate the llog record length
4819 and skip the remainder of the block on error.
4822 Frequency : occasional (liblustre only)
4824 Description: liblustre could not open files whose last component is a symlink
4825 Details : sysio_path_walk() would incorrectly pass the open intent to
4826 intermediate path components.
4829 Frequency : rare (liblustre only with non-standard tuning)
4830 Bugzilla : 7201 (7350)
4831 Description: Tuning the MDC DLM LRU size to zero triggers client LASSERT
4832 Details : llu_lookup_finish_locks() tries to set lock data on a lock
4833 after it has been released, only do this for referenced locks
4835 Severity : enhancement
4837 Description: specifying an (invalid) directory default stripe_size of -1
4838 would reset the directory default striping
4839 Details : stripe_size -1 was used internally to signal directory stripe
4840 removal, now use "all default" to signal dir stripe removal
4841 as a directory striping of "all default" is not useful
4844 Frequency : common for large clusters running liblustre clients
4846 Description: doing an ls when liblustre clients are running is slow
4847 Details : sending a glimpse AST to a liblustre client waits for every AST
4848 to time out, as liblustre clients will not respond. Since they
4849 cannot cache data we refresh the OST lock LVB from disk instead.
4851 Severity : enhancement
4853 Description: doing an ls at the same time as file IO can be slow
4854 Details : enqueue and other "small" requests can be blocked behind many
4855 large IO requests. Create a new OST IO portal for non-IO
4856 requests so they can be processed faster.
4859 Frequency : rare (only HPUX clients mounting unsupported re-exported NFS vol)
4861 Description: an HPUX NFS client would get -EACCESS when ftruncate()ing a newly
4862 created file with mode 000
4863 Details : the Linux NFS server relies on an MDS_OPEN_OWNEROVERRIDE hack to
4864 allow an ftruncate() as a non-root user to a file with mode 000.
4865 Lustre now respects this flag to disable mode checks when
4866 truncating a file owned by the user
4869 Frequency : liblustre-only, when liblustre client dies unexpectedly or becomes
4872 Description: Revoking locks from clients that went dead or catatonic might take
4874 Details : New lock flags FL_CANCEL_ON_BLOCK used by liblustre makes
4875 cancellation of such locks instant on servers without waiting for
4876 any reply from clients. Clients drops these locks when cancel
4877 notification from server is received without replying.
4880 Frequency : liblustre-only, when liblustre client dies or becomes busy
4882 Description: Doing ls on Linux clients can take a long time with active
4884 Details : Liblustre client cannot handle ASTs in timely manner, so avoid
4885 granting such locks to it in the first place if possible. Locks
4886 are taken by proxy on the OST during the read or write and
4887 dropped immediately afterward. Add connect flags handling, do
4888 not grant locks to liblustre clients for glimpse ASTs.
4890 Severity : enhancement
4892 Description: Improve read-ahead algorithm to avoid excessive IO for random reads
4893 Details : Existing read-ahead algorithm is tuned for the case of streamlined
4894 sequential reads and behaves badly with applications doing random
4895 reads. Improve it by reading ahead at least read region, and
4896 avoiding excessive large RPC for small reads.
4898 Severity : enhancement
4900 Description: Creating more than 1000 files for a single job may cause a load
4901 imbalance on the OSTs if there are also a large number of OSTs.
4902 Details : qos_prep_create() uses an OST index reseed value that is an
4903 even multiple of the number of available OSTs so that if the
4904 reseed happens in the middle of the object allocation it will
4905 still utilize the OSTs as uniformly as possible.
4910 Description: OST or MDS may oops in ping_evictor_main()
4911 Details : ping_evictor_main() drops obd_dev_lock if deleting a stale export
4912 but doesn't restart at beginning of obd_exports_timed list
4915 Severity : enhancement
4917 Description: improve by-nid export eviction on the MDS and OST
4918 Details : allow multiple exports with the same NID to be evicted at one
4919 time without re-searching the exports list.
4922 Frequency : rare, only with supplementary groups enabled on SMP 2.6 kernels
4924 Description: MDS may oops in groups_free()
4925 Details : in rare race conditions a newly allocated group_info struct is
4926 freed again, and this can be NULL. The 2.4 compatibility code
4927 for groups_free() checked for a NULL pointer, but 2.6 did not.
4930 Frequency : common for liblustre clients doing little filesystem IO
4931 Bugzilla : 9352, 7313
4932 Description: server may evict liblustre clients accessing contended locks
4933 Details : if a client is granted a lock or receives a completion AST
4934 with a blocking AST already set it would not reply to the AST
4935 for LDLM_FL_CANCEL_ON_BLOCK locks. It now replies to such ASTs.
4938 Frequency : lfs setstripe, only systems with more than 160 OSTs
4940 Description: unable to set striping with a starting offset beyond OST 160
4941 Details : llapi_create_file() incorrectly limited the starting stripe
4942 index to the maximum single-file stripe count.
4945 Frequency : LDAP users only
4947 Description: lconf did not handle in-kernel recovery with LDAP properly
4948 Details : lconf/LustreDB get_refs() is searching the wrong namespace
4950 Severity : enhancement
4952 Description: bind OST threads to NUMA nodes to improve performance
4953 Details : all OST threads are uniformly bound to CPUs on a single NUMA
4954 node and do their allocations there to localize memory access
4956 Severity : enhancement
4958 Description: llmount can determine client NID directly from Myrinet (GM)
4959 Details : the client NID code from gmnalnid was moved directly into
4960 llmount, removing the need to use this or specifying the
4961 client NID explicitly when mounting GM clients with zeroconf
4964 Frequency : if client is started with down MDS
4966 Description: if client is started with down MDS mount hangs in ptlrpc_queue_wait
4967 Details : Having an LWI_INTR() wait event (interruptible, but no timeout)
4968 will wait indefinitely in ptlrpc_queue_wait->l_wait_event() after
4969 ptlrpc_import_delayed_req() because we didn't check if the
4970 request was interrupted, and we also didn't break out of the
4971 event loop if there was no timeout
4976 Description: data loss during non-page-aligned writes to a single file from
4977 both multiple nodes and multiple threads on one node at same time
4978 Details : updates to KMS and lsm weren't protected by common lock. Resulting
4979 inconsistency led to false short-reads, that were cached and later
4980 used by ->prepare_write() to fill in partially written page,
4981 leading to data loss.
4984 Frequency : always, if lconf --abort_recovery used
4986 Description: lconf --abort_recovery fails with 'Operation not supported'
4987 Details : lconf was attempting to abort recovery on the MDT device and not
4990 Severity : enhancement
4992 Description: remove cleanup logs
4993 Details : replace lconf-generated cleanup logs with lustre internal
4994 cleanup routines. Eliminates the need for client-cleanup and
4997 Severity : enhancement
4999 Description: add support for EAs (user and system) on lustre filesystems
5000 Details : it is now possible to store extended attributes in the Lustre
5001 client filesystem, and with the user_xattr mount option it
5002 is possible to allow users to store EAs on their files also
5004 Severity : enhancement
5006 Description: Add possibility (config option) to show minimal available OST free
5008 Details : When compiled with --enable-mindf configure option, statfs(2)
5009 (and so, df) will return least minimal free space available from
5010 all OSTs as amount of free space on FS, instead of summary of
5011 free spaces of all OSTs.
5013 Severity : enhancement
5015 Description: do not expand extent locks acquired on OST-side
5016 Details : Modify ldlm_extent_policy() to not expand local locks, acquired
5017 by server: they are not cached anyway.
5020 Frequency : when mmap is used/binaries executed from Lustre
5022 Description: Unmmap pages before throwing them away from read cache.
5023 Details : llap_shrink cache now attempts to unmap pages before discarding
5024 them (if unmapping failed - do not discard). SLES9 kernel has
5025 extra checks that trigger if this unmapping is not done first.
5030 Description: lconf didn't resolve symlinks before checking to see whether a
5031 given mountpoint was already in use
5034 Frequency : when migrating failover services
5035 Bugzilla : 6395, 9514
5036 Description: When migrating a subset of services from a node (e.g. failback
5037 from a failover service node) the remaining services would
5038 time out and evict clients.
5039 Details : lconf --force (implied by --failover) sets the global obd_timeout
5040 to 5 seconds in order to quickly disconnect, but this caused
5041 other RPCs to time out too quickly. Do not change the global
5042 obd_timeout for force cleanup, only set it for DISCONNECT RPCs.
5044 Severity : enhancement
5045 Frequency : if MDS is started with down OST
5046 Bugzilla : 9439,5706
5047 Description: Allow startup/shutdown of an MDS without depending on the
5048 availability of the OSTs.
5049 Details : Asynchronously call mds_lov_synchronize during MDS startup.
5050 Add appropriate locking and lov-osc refcounts for safe
5051 cleaning. Add osc abort_inflight calls in case the
5052 synchronize never started.
5055 Frequency : occasional (Cray XT3 only)
5057 Description: root not authorized to access files in CRAY_PORTALS environment
5058 Details : The client process capabilities were not honoured on the MDS in
5059 a CRAY_PORTALS/CRAY_XT3 environment. If the file had previously
5060 been accessed by an authorized user then root was able to access
5061 the file on the local client also. The root user capabilities
5062 are now allowed on the MDS, as this environment has secure UID.
5065 Frequency : occasional
5067 Description: ldiskfs "too long searching" message happens too often
5068 Details : A debugging message (otherwise harmless) prints too often on
5069 the OST console. This has been reduced to only happen when
5070 there are fragmentation problems on the filesystem.
5075 Description: Division by zero in statfs when all OSCs are inactive
5076 Details : lov_get_stripecnt() returns zero due to incorrect order of checks,
5077 lov_statfs divides by value returned by lov_get_stripecnt().
5081 Bugzilla : 9489, 3273
5082 Description: First write from each client to each OST was only 4kB in size,
5083 to initialize client writeback cache, which caused sub-optimal
5084 RPCs and poor layout on disk for the first written file.
5085 Details : Clients now request an initial cache grant at (re)connect time
5086 and so that they can start streaming writes to the cache right
5087 away and always do full-sized RPCs if there is enough data.
5088 If the OST is rebooted the client also re-establishes its grant
5089 so that client cached writes will be honoured under the grant.
5094 Description: Slow ls (and stat(2) syscall) on files residing on IO-loaded OSTs
5095 Details : Now I/O RPCs go to different portal number and (presumably) fast
5096 lock requests (and glimses) and other RPCs get their own service
5097 threads pool that should be able to service those RPCs
5100 Severity : enhancement
5102 Description: Ability to exchange lustre version between client and servers and
5103 issue warnings at client side if client is too old. Also for
5104 liblustre clients there is ability to refuse connection of too old
5106 Details : New 'version' field is added to connect data structure that is
5107 filled with version info. That info is later checked by server and
5111 Frequency : rare, liblustre only.
5112 Bugzilla : 9296, 9581
5113 Description: Two simultaneous writes from liblustre at offset within same page
5114 might proceed at the same time overwriting eachother with stale
5116 Details : I/O lock withing llu_file_prwv was released too early, before data
5117 actually was hitting the wire. Extended lock-holding time until
5118 server acknowledges receiving data.
5121 Frequency : extremely rare. Never observed in practice.
5123 Description: avoid generating lustre_handle cookie of 0.
5124 Details : class_handle_hash() generates handle cookies by incrementing
5125 global counter, and can hit 0 occasionaly (this is unlikely, but
5126 not impossible, because initial value of cookie counter is
5127 selected randonly). Value of 0 is used as a sentinel meaning
5128 "unassigned handle" --- avoid it. Also coalesce two critical
5129 sections in this function into one.
5131 Severity : enhancement
5133 Description: allow liblustre clients to delegate truncate locking to OST
5134 Details : To avoid overhead of locking, liblustre client instructs OST to
5135 take extent lock in ost_punch() on client's behalf. New connection
5136 flag is added to handle backward compatibility.
5138 Severity : enhancement
5139 Bugzilla : 4928, 7341, 9758
5140 Description: allow number of OST service threads to be specified
5141 Details : a module parameter allows the number of OST service threads
5142 to be specified via "options ost ost_num_threads={N}" in the
5143 OSS's /etc/modules.conf or /etc/modprobe.conf.
5147 Bugzilla : 6146, 9635, 9895
5148 Description: servers crash with bad pointer in target_handle_connect()
5149 Details : In rare cases when a client is reconnecting it was possible that
5150 the connection request was the last reference for that export.
5151 We would temporarily drop the export reference and get a new
5152 one, but this may have been the last reference and the export
5153 was just destroyed. Get new reference before dropping old one.
5155 Severity : enhancement
5156 Frequency : if client is started with failover MDS
5158 Description: Allow multiple MDS hostnames in the mount command
5159 Details : Try to read the configuration from all specified MDS
5160 hostnames during a client mount in case the "primary"
5163 Severity : enhancement
5165 Description: Stop sending data to evicted clients as soon as possible.
5166 Details : Check if the client we are about to send or are sending data to
5167 was evicted already. (Check is done every second of waiting,
5168 for which l_wait_event interface was extended to allow checking
5169 of exit condition at specified intervals).
5172 Frequency : rare, normally only when NFS exporting is done from client
5174 Description: 'bad disk LOV MAGIC: 0x00000000' error when chown'ing files
5176 Details : Make mds_get_md() recognise empty md case and set lmm size to 0.
5179 Frequency : always, if srand() is called before liblustre initialization
5181 Description: Liblustre uses system PRNG disturbing its usage by user application
5182 Details : Introduce internal to lustre fast and high-quality PRNG for
5183 lustre usage and make liblustre and some other places in generic
5184 lustre code to use it.
5186 Severity : enhancement
5187 Bugzilla : 9477, 9557, 9870
5188 Description: Verify that the MDS configuration logs are updated when xml is
5189 Details : Check if the .xml configuration logs are newer than the config
5190 logs stored on the MDS and report an error if this is the case.
5191 Request --write-conf, or allow starting with --old_conf.
5193 Severity : enhancement
5195 Description: Handle symlinks in the path when checking if Lustre is mounted.
5196 Details : Resolve intermediate symlinks when checking if a client has
5197 mounted a filesystem to avoid duplicate client mounts.
5202 Description: lconf can hit an error exception but still return success.
5203 Details : The lconf command catches the Command error exception at the top
5204 level script context and will exit with the associated exit
5205 status, but doesn't ensure that this exit status is non-zero.
5210 Description: failure of ptlrpc thread startup can cause oops
5211 Details : Starting a ptlrpc service thread can fail if there are a large
5212 number of threads or the server memory is very fragmented.
5213 Handle this without oopsing.
5216 Frequency : always, only if liblustre and non-default acceptor port was used
5218 Description: liblustre cannot connect to servers with non-default acceptor port
5219 Details : tcpnal_set_default_params() was not called and was therefore
5220 ignoring the environment varaible TCPNAL_PORT, as well as other
5221 TCPNAL_ environment variables
5226 Description: two objects could be created on the same OST for a single file
5227 Details : If an OST is down, in some cases it was possible to create two
5228 objects on a single OST for a single file. No problems other
5229 than potential performance impact and spurious error messages.
5233 Bugzilla : 5681, 9562
5234 Description: Client may oops in ll_unhash_aliases
5235 Details : Client dcache may become inconsistent in race condition.
5236 In some cases "getcwd" can fail if the current directory is
5242 Description: Inode refcounting problems in NFS export code
5243 Details : link_raw functions used to call d_instantiate without obtaining
5244 extra inode reference first.
5248 Bugzilla : 9942, 9903
5249 Description: Referencing freed requests leading to crash, memleaks with NFS.
5250 Details : We used to require that call to ll_revalidate_it was always
5251 followed by ll_lookup_it. Also with revalidate_special() it is
5252 possible to call ll_revalidate_it() twice for the same dentry
5253 even if first occurence returned success. This fix changes semantic
5254 between DISP_ENQ_COMPLETE disposition flag to mean there is extra
5255 reference on a request referred from the intent.
5256 ll_intent_release() then releases such a request.
5259 Frequency : rare, normally benchmark loads only
5261 Description: unlinked inodes were kept in memory on the client
5262 Details : If a client is repeatedly creating and unlinking files it
5263 can accumulate a lot of stale inodes in the inode slab cache.
5264 If there is no other client load running this can cause the
5265 client node to run out of memory. Instead flush old inodes
5266 from client cache that have the same inode number as a new inode.
5269 Frequency : SLES9 2.6.5 kernel and long filenames only
5270 Bugzilla : 9969, 10379
5271 Description: utime reports stale NFS file handle
5272 Details : SLES9 uses out-of-dentry names in some cases, which confused
5273 the lustre dentry revalidation. Change it to always use the
5277 Frequency : rare, unless heavy write-truncate concurrency is continuous
5278 Bugzilla : 4180, 6984, 7171, 9963, 9331
5279 Description: OST becomes very slow and/or deadlocked during object unlink
5280 Details : filter_destroy() was holding onto the parent directory lock
5281 while truncating+unlinking objects. For very large objects this
5282 may block other threads for a long time and slow overall OST
5283 responsiveness. It may also be possible to get a lock ordering
5284 deadlock in this case, or run out of journal credits because of
5285 the combined truncate+unlink. Solution is to do object truncate
5286 first in one transaction without parent lock, and then do the
5287 final unlink in a new transaction with the parent lock. This
5288 reduces the lock hold time dramatically.
5291 Frequency : rare, 2.4 kernels only
5293 Description: MDS or OST cleanup may trip kernel BUG when dropping kernel lock
5294 Details : mds_cleanup() and filter_cleanup() need to drop the kernel lock
5295 before unmounting their filesystem in order to avoid deadlock.
5296 The kernel_locked() function in 2.4 kernels only checks whether
5297 the kernel lock is held, not whether it is this process that is
5298 holding it as 2.6 kernels do.
5303 Description: MDS or OST may oops/LBUG if a client is connecting multiple times
5304 Details : The client ptlrpc code may be trying to reconnect to a down
5305 server before a previous connection attempt has timed out.
5306 Increase the reconnect interval to be longer than the connection
5307 timeout interval to avoid sending duplicate connections to
5311 Frequency : echo_client brw_test command
5313 Description: fix echo_client to work with OST preallocated code
5314 Details : OST preallocation code (5137) didn't take echo_client IO path
5315 into account: echo_client calls filter methods outside of any
5316 OST thread and, hence, there is no per-thread preallocated
5317 pages and buffers to use. Solution: hijack pga pages for IO. As
5318 a byproduct, this avoids unnecessary data copying.
5322 Bugzilla : 3555, 5962, 6025, 6155, 6296, 9574
5323 Description: Client can oops in mdc_commit_close() after open replay
5324 Details : It was possible for the MDS to return an open request with no
5325 transaction number in mds_finish_transno() if the client was
5326 evicted, but without actually returning an error. Clients
5327 would later try to replay that open and may trip an assertion
5328 Simplify the client close codepath, and always return an error
5329 from the MDS in case the open is not successful.
5332 Frequency : rare, 2.6 OSTs only
5334 Description: OST may deadlock under high load on fragmented files
5335 Details : If there was a heavy load and highly-fragmented OST filesystems
5336 it was possible to have all the OST threads deadlock waiting on
5337 allocation of biovecs, because the biovecs were not released
5338 until the entire RPC IO was completed. Instead, release biovecs
5339 as soon as they are complete to ensure forward IO progress.
5341 Severity : enhancement
5343 Description: Support for specifying external journal device at mount
5344 Details : If an OST or MDS device is formatted with an external journal
5345 device, this device major/minor is stored in the ext3 superblock
5346 and may not be valid for failover. Allow detecting and
5347 specifying the external journal at mount time.
5352 Description: Mounting an MDS with pending unlinked files may cause oops
5353 Details : target_finish_recovery() calls mds_postrecov() which returned
5354 the number of orphans unlinked. mds_lov_connect->mds_postsetup()
5355 considers this an error and immediately begins cleaning up the
5356 lov, just after starting the mds_lov process
5358 Severity : enhancement
5360 Description: Implement 'lfs df' to report actual free space on per-OST basis
5361 Details : Add sub-command 'df' on 'lfs' to report the disk space usage of
5362 MDS/OSDs. Usage: lfs df [-i][-h]. Command Options: '-i' to report
5363 usage of objects; '-h' to report in human readable format.
5365 ------------------------------------------------------------------------------
5367 08-26-2005 Cluster File Systems, Inc. <info@clusterfs.com>
5374 Description: Mounting an ldiskfs file system with mballoc may crash OST node.
5375 Details : ldiskfs mballoc code may reference an uninitialized buddy struct
5376 at startup during orphan unlinking. Instead, skip buddy update
5377 before setup, as it will be regenerated after recovery is complete.
5382 Description: If an OST is inactive, its locks might reference stale inodes.
5383 Details : lov_change_cbdata() must iterate over all namespaces, even if
5384 they are inactive to clear inode references from the lock.
5386 Severity : enhancement
5387 Frequency : occasional, if non-standard max_dirty_mb used
5389 Description: Client will block write RPCs if not enough grant
5390 Details : If a client has max_dirty_mb smaller than max_rpcs_in_flight,
5391 then the client will block writes while waiting for another RPC
5392 to complete instead of consuming its dirty limit. With change
5393 we get improved performance when max_dirty_mb is small.
5395 Severity : enhancement
5396 Bugzilla : 3389, 6253
5397 Description: Add support for supplementary groups on the MDS.
5398 Details : The MDS has an upcall /proc/fs/lustre/mds/{mds}/group_upcall
5399 (set to /usr/sbin/l_getgroups if enabled) which will do MDS-side
5400 lookups for user supplementary groups into a cache.
5404 Description: O_CREAT|O_EXCL open flags in liblustre always return -EEXIST
5405 Details : Make libsysio to not enforce O_EXCL by clearing the flag,
5406 for liblustre O_EXCL is enforced by MDS.
5410 Description: readdir never returns NULL in liblustre.
5411 Details : Corrected llu_iop_getdirentries logic, to return offset of next
5412 dentry in struct dirent.
5416 Frequency : liblustre only, depends on application IO pattern
5417 Description: liblustre clients evicted if not contacting servers
5418 Details : Don't put liblustre clients into the ping_evictor list, so
5419 they will not be evicted by the pinger ever.
5421 Severity : enhancement
5423 Description: Add ability to evict clients by NID from MDS.
5424 Details : By echoing "nid:$NID" string into
5425 /proc/fs/lustre/mds/.../evict_client client with nid that equals to
5426 $NID would be instantly evicted from this MDS and from all active
5427 OSTs connected to it.
5431 Description: Do not query file size twice, somewhat slowing stat(2) calls.
5432 Details : lookup_it_finish() used to query file size from OSTs that was not
5437 Description: service threads change working directory to that of init
5438 Details : Starting lustre service threads may pin the working directory
5439 of the parent thread, making that filesystem busy. Threads
5440 now change to the working directory of init to avoid this.
5444 Frequency : during shutdown only
5445 Description: shutdown with a failed MDS or OST can cause unmount to hang
5446 Details : Don't resend DISCONNECT messages in ptlrpc_disconnect_import()
5451 Frequency : 2.6 only
5452 Description: chmod/chown may include an extra supplementary group
5453 Details : ll{,u}_mdc_pack_op_data() does not properly initialize the
5454 supplementary group and if none is specified this is used.
5457 Bugzilla : 5479 (6816)
5459 Description: Racing open + rm can assert client in mdc_set_open_replay_data()
5460 Details : If lookup is in progress on a file that is unlinked we might try
5461 to revalidate the inode and fail in revalidate after lookup is
5462 complete and ll_file_open() enqueues the open again but
5463 it_open_error() was not checking DISP_OPEN_OPEN errors correctly.
5466 Frequency : always, if lconf --abort_recovery used
5468 Description: lconf --abort_recovery fails with 'Operation not supported'
5469 Details : lconf was attempting to abort recovery on the MDT device and not
5472 ------------------------------------------------------------------------------
5474 2005-08-08 Cluster File Systems, Inc. <info@clusterfs.com>
5479 Frequency : rare (only unsupported configurations with a node running as an
5481 Bugzilla : 6514, 5137
5482 Description: Mounting a Lustre file system on a node running as an OST could
5484 Details : OSTs now preallocates memory needed to write out data at
5485 startup, instead of when needed, to avoid having to
5486 allocate memory in possibly low memory situations.
5487 Specifically, if the file system is mounted on on OST,
5488 memory pressure could force it to try to write out data,
5489 which it needed to allocate memory to do. Due to the low
5490 memory, it would be unable to do so and the node would
5491 become unresponsive.
5493 Severity : enhancement
5495 Description: Addition of lconf --service command line option
5496 Details : lconf now accepts a '--service <arg>' option, which is
5497 shorthand for 'lconf --group <arg> --select <arg>=<hostname>'
5499 Severity : enhancement
5501 Description: Failover mode is now the default for OSTs.
5502 Details : By default, OSTs will now run in failover mode. To return to
5503 the old behaviour, add '--failout' to the lmc line for OSTs.
5505 Severity : enhancement
5507 Description: Health checks are now provided for MDS and OSTs
5508 Details : Additional detailed health check information on MSD and OSTs
5509 is now provided through the procfs health_check value.
5512 Frequency : occasional, depends on IO load
5514 Description: Disk fragmentation on the OSTs could eventually cause slowdowns
5515 after numerous create/delete cycles
5516 Details : The ext3 inode allocation policy would not allocate new inodes
5517 very well on the OSTs because there are no new directories
5518 being created. Instead we look for groups with free space if
5519 the parent directories are nearly full.
5524 Description: Network or server problems during mount may cause partially
5525 mounted clients instead of returning an error.
5526 Details : The config llog parsing code may overwrite the error return
5527 code during mount error handling, returning success instead
5533 Description: MDS can fail to allocate large reply buffers
5534 Details : After long uptimes the MDS can fail to allocate large reply
5535 buffers (e.g. zconf client mount config records) due to memory
5536 fragmentation or consumption by the buffer cache. Preallocate
5537 some large reply buffers so that these replies can be sent even
5538 under memory pressure.
5542 Frequency : rare (liblustre)
5543 Description: fsx running with liblustre complained that using truncate() to
5544 extend the file doesn't work. This patch corrects that issue.
5545 Details : This is the liblustre equivalent of the fix for bug 6196. Fixes
5546 ATTR_SIZE and lsm use in llu_setattr_raw.
5550 Frequency : rare, only 2.6 kernels
5551 Description: Unusual file access patterns on the MDS may result in inode
5552 data being lost in very rare circumstances.
5553 Details : Bad interaction between the ea-in-inode patch and the "no-read"
5554 code in the 2.6 kernel caused the inode and/or EA data not to
5555 be read from disk, causing single-file corruption.
5559 Frequency : rare, only 2.6 filesystems using extents
5560 Description: Heavy concurrent write and delete load may cause data corruption.
5561 Details : It was possible under high-load situations to have an extent
5562 metadata block in the block device cache from a just-unlinked
5563 file overwrite a newly-allocated data block. We now unmap any
5564 metadata buffers that alias just-allocated data blocks.
5568 Frequency : filesystems with default stripe_count larger than 77
5569 Description: lconf+mke2fs fail when formatting filesystem with > 77 stripes
5570 Details : lconf specifies an inode size of 4096 bytes when the default
5571 stripe_count is larger than 77. This conflicts with the default
5572 inode density of 1 per 4096 bytes. Allocate smaller inodes in
5573 this case to avoid pinning too much memory for large EAs.
5575 ------------------------------------------------------------------------------
5577 2005-07-07 Cluster File Systems, Inc. <info@clusterfs.com>
5582 Frequency : rare (extremely heavy IO load with hundreds of clients)
5584 Description: Client is evicted, gets IO error writing to file
5585 Details : lock ordering changes for bug 5492 reintroduced bug 3267 and
5586 caused clients to be evicted for AST timeouts. The fixes in
5587 bug 5192 mean we no longer need to have such short AST timeouts
5588 so ldlm_timeout has been increased.
5591 Frequency : occasional during --force or --failover shutdown under load
5592 Bugzilla : 5949, 4834
5593 Description: Server oops/LBUG if stopped with --force or --failover under load
5594 Details : a collection of import/export refcount and cleanup ordering
5595 issues fixed for safer force cleanup
5598 Frequency : only filesystems larger than 120 OSTs
5599 Bugzilla : 5990, 6223
5600 Description: lfs getstripe would oops on a very large filesystem
5601 Details : lov_getconfig used kfree on vmalloc'd memory
5604 Frequency : only filesystems exporting via NFS to Solaris 10 clients
5605 Bugzilla : 6242, 6243
5606 Description: reading from files that had been truncated to a non-zero size
5607 but never opened returned no data
5608 Details : ll_file_read() reads zeros from no-object files to EOF
5613 Description: A bug in MDS/OSS recovery could cause the OSS to fail an assertion
5614 Details : There's little harm in aborting MDS/OSS recovery and letting it
5615 try again, so I removed the LASSERT and return an error instead.
5617 Severity : enhancement
5619 Description: New debugging infrastructure for tracking down data corruption
5620 Details : The I/O checksum code was replaced to: (a) control it at runtime,
5621 (b) cover more of the client-side code path, and (c) try to narrow
5622 down where problems occurred
5626 Bugzilla : 3819, 4364, 4397, 6313
5627 Description: Racing close and eviction MDS could cause assertion in mds_close
5628 Details : It was possible to get multiple mfd references during close and
5629 client eviction, leading to one thread referencing a freed mfd.
5631 Severity: : enhancement
5632 Bugzilla : 3262, 6359
5633 Description: Attempts to reconnect to servers are now more aggressive.
5634 Details : This builds on the enhanced upcall-less recovery that was added
5635 in 1.4.2. When trying to reconnect to servers, clients will
5636 now try each server in the failover group every 10 seconds. By
5637 default, clients would previously try one server every 25 seconds.
5642 Description: After recovery, certain operations trigger a failed
5643 assertion on a client.
5644 Details : Failing over an mds, using lconf -d --failover, while a
5645 client was doing a readdir() call would cause the client to
5646 LBUG after recovery completed and the readdir() was resent.
5648 Severity : enhancement
5650 Description: Default groups are now added by lconf
5651 Details : You can now run lconf --group <servicename> without having to
5652 manually add groups with lmc.
5655 Frequency : occasional
5657 Description: Nodes with an elan id of 0 trigger a failed assertion
5660 Frequency : always when accessing e.g. tty/console device nodes
5662 Description: tty and some other devices nodes cannot be used on lustre
5663 Details : file's private_data field is used by device data and lustre
5664 values in there got lost. New field was added to struct file to
5665 store fs-specific private data.
5668 Frequency : when exporting Lustre via NFS
5670 Description: NFSD failed occasionally when looking up a path component
5671 Details : NFSD is looking up ".." which was broken in ext3 directories
5672 that had grown large enough to become hashed.
5675 Frequency : Clusters with multiple interfaces not on the same subnet
5677 Description: Nodes will repeatedly try to reconnect to an interface which it
5678 cannot reach and report an error to the log.
5679 Details : Extra peer list entries will be created by lconf with some peers
5680 unreachable. lconf now validates the peer before adding it.
5683 Frequency : Only if a default stripe is set on the filesystem root.
5685 Description: Setting a default stripe on the filesystem root prevented the
5686 filesystem from being remounted.
5687 Details : The client was sending extra request flags in the root getattr
5688 request and did not allocate a reply buffer for the dir EA.
5691 Frequency : occasional, higher if lots of files are accessed by one client
5692 Bugzilla : 6159, 6097
5693 Description: Client trips assertion regarding lsm mismatch/magic
5694 Details : While revalidating inodes the VFS looks up inodes with ifind()
5695 and in rare cases can find an inode that is being freed.
5696 The ll_test_inode() code will free the lsm during ifind()
5697 when it finds an existing inode and then the VFS later attaches
5698 this free lsm to a new inode.
5702 Bugzilla : 6422, 7030
5703 Description: MDS deadlock between mkdir and client eviction
5704 Details : Creating a new file via mkdir or mknod (starting a transaction
5705 and getting the ns lock) can deadlock with client eviction
5706 (gets ns lock and trying to finish a synchronous transaction).
5709 Frequency : occasional
5710 Description: While starting a server, the fsfilt_ext3 module could not be
5712 Details : CFS's improved ext3 filesystem is named ldiskfs for 2.6
5713 kernels. Previously, lconf would still use the ext3 name
5714 when trying to load modules. Now, it will correctly use
5715 ext3 on 2.4 and ldiskfs on 2.6.
5717 Severity : enhancement
5718 Description: The default stripe count has been changed to 1
5719 Details : The interpretation of the default stripe count (0, to lfs
5720 or lmc) has been changed to mean striping across a single
5721 OST, rather than all available. For general usage we have
5722 found a stripe count of 1 or 2 works best.
5724 Severity : enhancement
5725 Description: Add support for compiling against Cray portals.
5726 Details : Conditional compiling for some areas that are different
5730 Frequency : occasional
5731 Bugzilla : 6409, 6834
5732 Description: Creating files with an explicit stripe count may lead to
5733 a failed assertion on the MDS
5734 Details : If some OSTs are full or unavailable, creating files may
5735 trigger a failed assertion on the MDS. Now, Lustre will
5736 try to use other servers or return an error to the
5740 Frequency : occasional
5742 Description: Multiple concurrent overlapping read+write on multiple SMP nodes
5743 caused lock timeout during readahead (since 1.4.2).
5744 Details : Processes doing readahead might match a lock that hasn't been
5745 granted yet if there are overlapping and conflicting lock
5746 requests. The readahead process waits on ungranted lock
5747 (original lock is CBPENDING), while OST waits for that process
5748 to cancel CBPENDING read lock and eventually evicts client.
5750 Severity : enhancement
5752 Description: Initial enabling of flock support for clients
5753 Details : Implements fcntl advisory locking and file status functions.
5754 This feature is provided as an optional mount flag (default
5755 off), and is NOT CURRENTLY SUPPORTED. Not all types of record
5756 locking are implemented yet, and those that are are not guaranteed
5757 to be completely correct in production environments.
5758 mount -t lustre -o [flock|noflock] ...
5761 Frequency : occasional
5763 Description: OSTs running 2.4 kernels but with extents enabled might trip an
5764 assertion in the ext3 JBD (journaling) layer.
5765 Details : The b_committed_data struct is protected by the big kernel lock
5766 in 2.4 kernels, serializing journal_commit_transaction() and
5767 ext3_get_block_handle->ext3_new_block->find_next_usable_block()
5768 access to this struct. In 2.6 kernels there is finer grained
5769 locking to improve SMP performance of the JBD layer.
5773 Description: Changes the "SCSI I/O Stats" kernel patch to default to "enabled"
5778 Description: merge and cleanup kernel patches.
5781 -----------------------------------------------------------------------------
5783 2005-05-05 Cluster File Systems, Inc. <info@clusterfs.com>
5785 NOTE: Lustre 1.4.2 uses an incompatible network protocol than previous
5786 versions of Lustre. Please update all servers and clients to
5787 version 1.4.2 or later at the same time. You must also run
5788 "lconf --write-conf {config}.xml" on the MDS while it is stopped
5789 to update the configuration logs.
5791 - fix for HPUX NFS client breakage when NFS exporting Lustre (5781)
5792 - mdc_enqueue does not need max_mds_easize request buffer on send (5707)
5793 - swab llog records of type '0' so we get proper header size/idx (5861)
5794 - send llog cancel req to DLM cancel portal instead of cb portal (5515)
5795 - fix rename of one directory over another leaking an inode (5953)
5796 - avoid SetPageDirty on 2.6 (5981)
5797 - don't re-add just-being-destroyed locks to the waiting list (5653)
5798 - when creating new directories, inherit the parent's custom
5799 striping settings if present parent (3048)
5800 - flush buffers from cache before direct IO in 2.6 obdfilter (4982)
5801 - don't hold i_size_sem in ll_nopage() and ll_ap_refresh_count (6077)
5802 - don't hold client locks on temporary worklist from l_lru (5666)
5803 - handle IO errors in 2.6 obdfilter bio completion routine (6046)
5804 - automatically evict dead clients (5921)
5805 - Update file size properly in create+truncate+fstat case (6196)
5806 - Do not unhash mountpoint dentries, do not allow removal of
5808 - Avoid lock ordering deadlock issue with write/truncate (6203,5654)
5809 - reserve enough journal credits in fsfilt_start_log for setattr (4554)
5810 - ldlm_enqueue freed-export error path would always LBUG (6149,6184)
5811 - don't reference lr_lvb_data until after we hold lr_lvb_sem (6170)
5812 - don't overwrite last_rcvd if there is a *_client_add() error (6086)
5813 - Correctly handle reads of files with no objects (6243)
5814 - lctl recover will also mark a device active if deactivate used (5933)
5816 - by default create 1 inode per 4kB space on MDS, per 16kB on OSTs
5817 - allow --write-conf on an MDS with different nettype than client (5619)
5818 - don't write config llogs to MDS for mounts not from that MDS (5617)
5819 - lconf should create multiple TCP connections from a client (5201)
5820 - init scripts are now turned off by default; run chkconfig --on
5821 lustre and chkconfig --on lustrefs to use them
5822 - upcalls are no longer needed for clients to recover to failover
5824 - add --abort-recovery option to lconf to abort recovery on device
5826 - add support for an arbitrary number of OSTs (3026)
5827 - Quota support protocol changes.
5828 - forward compatibility changes to wire structs (6007)
5829 - rmmod NALs that might be loaded because of /etc/modules.conf (6133)
5830 - support for mountfsoptions and clientoptions to the Lustre LDAP (5873)
5831 - improved "lustre status" script
5832 - initialize blocksize for non-regular files (6062)
5833 - added --disable-server and --disable-client configure options (5782)
5834 - introduce a lookup cache for lconf to avoid repeated DB scans (6204)
5835 - Vanilla 2.4.29 support
5836 - increase maximum number of obd devices to 520 (6242)
5837 - remove the tcp-zero-copy patch from the suse-2.4 series (5902)
5838 - Quadrics Elan drivers are now included for the RHEL 3 2.4.21 and
5839 SLES 9 2.6.5 kernels
5840 - limit stripes per file to 160 (the maximum EA size) (6093)
5842 2005-03-22 Cluster File Systems, Inc. <info@clusterfs.com>
5845 - don't LASSERT in ll_release on NULL lld with NFS export (4655, 5760)
5846 - hold NS lock when calling handle_ast_error->del_waiting_lock (5746)
5847 - fix setattr mtime regression from lovcleanup merge (4829, 5669)
5848 - workaround for 2.6 crash in ll_unhash_aliases (5687, 5210)
5849 - small ext3 extents cleanups and fixes (5733)
5850 - improved mballoc code, several small races and bugs fixed (5733, 5638)
5851 - kernel version 43 - fix remove_suid bugs in both 2.4 and 2.6 (5695)
5852 - avoid needless client->OST connect, fix handle mismatch (5317)
5853 - fix DLM error path that led to out-of-sync client, long delays (5779)
5854 - support common vfs-enforced mount options (nodev,nosuid,noexec) (5637)
5855 - fix several locking issues related to i_size (5492,5624,5654,5672)
5856 - don't move pending lock onto export if it is already evicted (5683)
5857 - fix kernel oops when creating .foo in unlinked directory (5548)
5858 - fix deadlock in obdfilter statistics vs. object create (5811)
5859 - use time_{before,after} to avoid timer jiffies wrap (5882)
5860 - shutdown --force/--failover stability (3607,3651,4797,5203,4834)
5861 - Do not leak request if server was not able to process it (5154)
5862 - If mds_open unable to find parent dir, make that negative lookup(5154)
5863 - don't create new directories with extent-mapping (5909, 5936)
5865 - fix lustre/lustrefs init scripts for SuSE (patch from Scali, 5702)
5866 - don't hold the pinger_sem in ptlrpc_pinger_sending_on_import
5867 - change obd_increase_kms to obd_adjust_kms (up or down) (5654)
5868 - lconf, lmc search both /usr/lib and /usr/lib64 for Python libs (5800)
5869 - support for RHEL4 kernel on i686 (5773)
5870 - provide error messages when incompatible logs are encountered (5898)
5872 2005-02-18 Cluster File Systems, Inc. <info@clusterfs.com>
5873 * version 1.4.0.10 (1.4.1 release candidate 1)
5875 - don't keep a lock reference when lock is not granted (4238)
5876 - unsafe list practices (rarely) led to infinite eviction loop (4908)
5877 - add per-fs limit of Lustre pages in page cache, avoid OOM (4699)
5878 - drop import inflight refcount on signal_completed_replay error (5255)
5879 - unlock page after async write error during send (3677)
5880 - handle missing objects in filter_preprw_read properly (5265)
5881 - no transno return for symlink open, don't save no-trasno open (3440)
5882 - don't try to complete elan receive that already failed (4012)
5883 - free RPC server reply state on error (5406)
5884 - clean up thread from ptlrpc_start_thread() on error (5160)
5885 - readahead could read extra page into cache that wasn't ejected (5388)
5886 - prevent races in class_attach/setup/cleanup/detach (5260)
5887 - don't dereference de->d_inode after l_dput of de (5458)
5888 - use "int" for stripe value returned from lock_to_stripe (5544)
5889 - mballoc allocation and error-checking fixes in 2.6 (5504)
5890 - block device patches to fix I/O request sizes in 2.6 (5482)
5891 - look up hostnames for IB nals (5602)
5892 - 2.6 changed lock ordering of 2 semaphores, caused deadlock (5654)
5893 - don't start multiple acceptors for the same port (5277)
5894 - fix incorrect LASSERT in mds_getattr_name (5635)
5895 - export a proc file for general "ping" checking (5628)
5896 - fix "lfs check" to not block when the MDS is down (5628)
5898 - service request history (4965)
5899 - put {ll,lov,osc}_async_page structs in a single slab (4699)
5900 - create an "evict_client" /proc entry on OSTs, like the MDS has
5901 - fix mount usage message, return errors per mount(8) (5168)
5902 - change grep [] to grep "[]" in tests so they work in more UMLs
5903 - fix ppc64/x86_64 spec to use %{_libdir} instead of /usr/lib (5389)
5904 - remove ancient LOV_MAGIC_V0 EA support (5047)
5905 - add "disk I/Os in flight" and "I/O req time" stats in obdfilter
5906 - align r/w RPCs to PTLRPC_MAX_BRW_SIZE boundary for performance (3451)
5907 - allow readahead allocations to fail when low on memory (5383)
5908 - mmap locking landed again, after considerable improvement (2828)
5909 - add get_hostaddr() to lustreDB.py for LDAP support (5459)
5911 2004-11-23 Cluster File Systems, Inc. <info@clusterfs.com>
5914 - send OST transaction number in read/write reply to free req (4966)
5915 - don't ASSERT in ptl_send_rpc() if we run out of memory (5119)
5916 - lock /proc/sys/portals/routes internal state, avoiding oops (4827)
5917 - the watchdog thread now runs as interruptible (5246)
5918 - flock/lockf fixes (but it's still disabled, pending 5135)
5919 - don't use EXT3 constants in llite code (5094)
5920 - memory shortage at startup could cause assertion (5176)
5922 - reorganization of lov code
5923 - single portals codebase
5925 - add extents/mballoc support (5025)
5926 - direct I/O reads in the obdfilter (4048)
5927 - kernel patches from LNXI for 2.6 (bluesmoke, perfctr, mtd, kexec)
5929 tbd Cluster File Systems, Inc. <info@clusterfs.com>
5932 - send OST transaction number in read/write reply to free req (4966)
5933 - don't ASSERT in ptl_send_rpc() if we run out of memory (5119)
5934 - lock /proc/sys/portals/routes internal state, avoiding oops (4827)
5935 - the watchdog thread now runs as interruptible (5246)
5936 - handle missing objects in filter_preprw_read properly (5265)
5937 - unsafe list practices (rarely) led to infinite eviction loop (4908)
5938 - drop import inflight refcount on signal_completed_replay error (5255)
5939 - unlock page after async write error during send (3677)
5940 - return original error code on reconstructed replies (3761)
5941 - no transno return for symlink open, don't save no-trasno open (3440)
5943 - add pid to ldlm debugging output (4922)
5944 - bump the watchdog timeouts -- we can't handle 30sec yet
5945 - extra debugging for orphan dentry/inode bug (5259)
5947 2004-11-16 Cluster File Systems, Inc. <info@clusterfs.com>
5950 - fix TCP_NODELAY bug, which caused extreme perf regression (5134)
5951 - allocate qswnal tx descriptors singly to avoid fragmentation (4504)
5952 - don't LBUG on obdo_alloc() failure, use OBD_SLAB_ALLOC() (4800)
5953 - fix NULL dereference in /proc/sys/portals/routes (4827)
5954 - allow failed mdc_close() operations to be interrupted (4561)
5955 - stop precreate on OST before MDS would time out on it (4778)
5956 - don't send partial-page writes before EOF from client (4410)
5957 - discard client grant for sub-page writes on large-page clients (4520)
5958 - don't free dentries not owned by NFS code, check generation (4806)
5959 - fix lsm leak if mds_create_objects() fails (4801)
5960 - limit debug_daemon file size, always print CERROR messages (4789)
5961 - use transno after validating reply (3892)
5962 - process timed out requests if import state changes (3754)
5963 - update mtime on OST during writes, return in glimpse (4829)
5964 - add mkfsoptions to LDAP (4679)
5965 - use ->max_readahead method instead of zapping global ra (5039)
5966 - don't interrupt __l_wait_event() during strace
5968 - add software watchdogs to catch hung threads quickly (4941)
5969 - make lustrefs init script start after nfs is mounted
5970 - fix CWARN/ERROR duplication (4930)
5971 - return async write errors to application if possible (2248)
5972 - add /proc/sys/portal/memused (bytes allocated by PORTALS_ALLOC)
5973 - print NAL number in %x format (4645)
5974 - update barely-supported suse-2.4.21-171 series (4842)
5975 - support for sles 9 %post scripts
5976 - support for building 2.6 kernel-source packages
5977 - support for sles km_* packages
5979 2004-10-07 Cluster File Systems, Inc. <info@clusterfs.com>
5982 - ignore -ENOENT errors in osc_destroy (3639)
5983 - notify osc create thread that OSC is being cleaned up (4600)
5984 - add nettype argument for llmount in #5d in conf-sanity.sh (3936)
5985 - reconstruct ost_handle() like mds_handle() (4657)
5986 - create a new thread to do import eviction to avoid deadlock (3969)
5987 - let lconf resolve symlinked-to devices (4629)
5988 - don't unlink "objects" from directory with default EA (4554)
5989 - hold socknal file ref over connect in case target is down (4394)
5990 - allow more than 32000 subdirectories in a single directory (3244)
5991 - fix blocks count for O_DIRECT writes (3751)
5992 - OST returns ENOSPC from object create when no space left (4539)
5993 - don't send truncate RPC if file size isn't changing (4410)
5994 - limit OSC precreate to 1/2 of value OST considers bogus (4778)
5995 - bind to privileged port in socknal and tcpnal (3689)
5997 - rate limit CERROR/CWARN console message to avoid overload (4519)
5998 - GETFILEINFO dir ioctl returns LOV EA + MDS stat in 1 call (3327)
5999 - basic mmap support (3918)
6000 - kernel patch series update from b1_4 (4711)
6002 2004-09-16 Cluster File Systems, Inc. <info@clusterfs.com>
6005 - avoid crash during MDS cleanup with OST shut down (2775)
6006 - fix loi_list_lock/oig_lock inversion on interrupted IO (4136)
6007 - don't use bad inodes on the MDS (3744)
6008 - dynamic object preallocation to improve recovery speed (4236)
6009 - don't hold spinlock over lock dumping or change debug flags (4401)
6010 - don't zero obd_dev when it is force cleaned (3651)
6011 - print grants to console if they go negative (4431)
6012 - "lctl deactivate" will stop automatic recovery attempts (3406)
6013 - look for existing locks in ldlm_handle_enqueue() (3764)
6014 - don't resolve lock handle twice in recovery avoiding race (4401)
6015 - revalidate should check working dir is a directory (4134)
6017 - don't always mark "slow" obdfilter messages as errors (4418)
6019 2004-08-24 Cluster File Systems, Inc. <info@clusterfs.com>
6022 - don't close LustreDB during write_conf until it is done (3860)
6023 - fix typo in lconf for_each_profile (3821)
6024 - allow dumping logs from multiple threads at one time (3820)
6025 - don't allow multiple threads in OSC recovery (3812)
6026 - fix debug_size parameters (3864)
6027 - fix mds_postrecov to initialize import for llog ctxt (3121)
6028 - replace config semaphore with spinlock (3306)
6029 - be sure to send a reply for a CANCEL rpc with bad export (3863)
6030 - don't allow enqueue to complete on a destroyed export (3822)
6031 - down write_lock before checking llog header bitmap (3825)
6032 - recover from lock replay timeout (3764)
6033 - up llog sem before sending rpc (3652)
6034 - reduce ns lock hold times when setting kms (3267)
6035 - change a dlm LBUG to LASSERTF, to maybe learn something (4228)
6036 - fix NULL deref and obd_dev leak on setup error (3312)
6037 - replace some LBUG about llog ops with error handling (3841)
6038 - don't match INVALID dentries from d_lookup and spin (3784)
6039 - hold dcache_lock while marking dentries INVALID and hashing (4255)
6040 - fix invalid assertion in ptlrpc_set_wait (3880)
6042 - add libwrap support for the TCP acceptor (3996)
6043 - add /proc/sys/portals/routes for non-root route listing (3994)
6044 - allow setting MDS UUID in .xml (2580)
6045 - print the stack of a process that LBUGs (4228)
6047 2004-07-14 Cluster File Systems, Inc. <info@clusterfs.com>
6050 - don't cleanup request in ll_file_open() on failed MDS open (3430)
6051 - make sure to unset replay flag from failed open requests (3440)
6052 - if default stripe count is 0, use OST count for inode size (3636)
6053 - update parent mtime/ctime on client for create/unlink (2611)
6054 - drop dentry ref in ext3_add_link from open_connect_dentry (3266)
6055 - free recovery state on server during a forced cleanup (3571)
6056 - unregister_reply for resent reqs (3063)
6057 - loop back devices mounting and status check on 2.6 (3563)
6058 - fix resource-creation race that can provoke i_size == 0 (3513)
6059 - don't try to use bad inodes returned from MDS/OST fs lookup (3688)
6060 - more debugging for page-accounting assertion (3746)
6061 - return -ENOENT instead of asserting if ost getattr+unlink race (3558)
6062 - avoid deadlock after precreation failure (3758)
6063 - fix race and lock order deadlock in orphan handling (3450, 3750)
6064 - add validity checks when grabbing inodes from l_ast_data (3599)
6066 - add /proc/.../recovery_status to obdfilter (3428)
6067 - lightweight CDEBUG infrastructure, debug daemon (3668)
6068 - change default OSC RPC parameters to be better on small clusters
6069 - turn off OST read cache for files smaller than 32MB
6070 - install man pages and include them in rpms (3100)
6071 - add new init script for (un)mounting lustre filesystems (2593)
6072 - run chkconfig in %post for init scripts (3701)
6073 - drop scimac NAL (unmaintained)
6075 2004-06-17 Cluster File Systems, Inc. <info@clusterfs.com>
6078 - clean kiobufs before and after use (3485)
6079 - strip trailing '/'s before comparing paths with /proc/mounts (3486)
6080 - remove assertions to work around "in-flight rpcs" recovery bug (3063)
6081 - change init script to fail more clearly if not run as root (1528)
6082 - allow clients to reconnect during replay (1742)
6083 - fix ns_lock/i_sem lock ordering deadlock for kms update (3477)
6084 - don't do DNS lookups on NIDs too small for IP addresses (3442)
6085 - re-awaken ptlrpcd if new requests arrive during check_set (3554)
6086 - fix cond_resched (3554)
6087 - only evict unfinished clients after recovery (3515)
6088 - allow bulk resend, prevent data loss (3570)
6089 - dynamic ptlrpc request buffer allocation (2102)
6090 - don't allow unlinking open directory if it isn't empty (2904)
6091 - set MDS/OST threads to umask 0 to not clobber client modes (3359)
6092 - remove extraneous obd dereference causing LASSERT failure (3334)
6093 - don't use get_cycles() when creating temp. files on the mds (3156)
6094 - hold i_sem when setting i_size in ll_extent_lock() (3564)
6095 - handle EEXIST for set-stripe, set proper directory name (3336)
6097 - servers can dump a log evicting a client - lustre.dump_on_timeout=1
6098 - fix ksocknal_fmb_callback() error messages (2918)
6100 2004-05-27 Cluster File Systems, Inc. <info@clusterfs.com>
6103 - don't copy lvb into (possibly NULL) reply on error (2983)
6104 - don't deref dentry after dput, don't free lvb on error (2922)
6105 - use the kms to determine writeback rpc length (2947)
6106 - increment oti_logcookies when osc is inactive (2948)
6107 - update client's i_blocks count via lvb messages (2543)
6108 - handle intent open/close of special files properly (1557)
6109 - mount MDS with errors=remount-ro, like obdfilter (2009)
6110 - initialize lock handle to avoid ASSERT on error cleanup (3057)
6111 - don't use cancelling-locks' kms values (2947)
6112 - use highest lock extent for kms, not last one (2925)
6113 - don't dereference ERR_PTR() dentry in error handling path (3107)
6114 - fix thread race in portals_debug_dumplog() (3122)
6115 - create lprocfs device entries at setup instead of at attach (1519)
6116 - common AST error handler, don't evict client on completion race (3145)
6117 - zero nameidata in detach_mnt in 2.6 (3118)
6118 - verify d_inode after revalidate_special is valid in 2.6 (3116)
6119 - use lustre_put_super() to handle zconf unmounts in 2.6 (3064)
6120 - initialize RPC timeout timer earlier for 2.6 (3219)
6121 - don't dereference NULL reply buffer if mdc_close was never sent (2410)
6122 - print nal/nid for unknown nid (3258)
6123 - additional checks for oscc recovery before doing precreate (3284)
6124 - fix ll_extent_lock() error return code for 64-bit systems (3043)
6125 - don't crash in mdc_close for bad permissions on open (3285)
6126 - zero i_rdev for non-device files (3147)
6127 - clear page->private before handing to FS, better assertion (3119)
6128 - tune the read pipeline (3236)
6129 - fix incorrect decref of invalidated dentry (2350)
6130 - provide read-ahead stats and refine rpc in flight stats (3328)
6131 - don't hold journal transaction open across create RPC (3313)
6132 - update atime on MDS at close time (3265)
6133 - close LDAP connection when recovering to avoid server load (3315)
6134 - update iopen-2.6 patch with fixes from 2399,2517,2904 (3301)
6135 - don't leak open file on MDS after open resend (3325)
6136 - serialize filter_precreate and filter_destroy_precreated (3329)
6137 - loop device shouldn't call sync_dev() for nul device (3092)
6138 - clear page cache after eviction (2766)
6139 - resynchronize MDS->OST in background (2824)
6140 - refuse to mount the same filesystem twice on same mountpoint (3394)
6141 - allow llmount to create routes for mounting behind routers (3320)
6142 - push lock cancellation to blocking thread for glimpse ASTs (3409)
6143 - don't call osc_set_data_with_check() for TEST_LOCK matches (3159)
6144 - fix rare problem with rename on htree directories (3417)
6146 - allow default OST striping configuration per directory (1414)
6147 - fix compilation for qswnal for 2.6 kernels (3125)
6148 - increase maximum number of MDS request buffers for large systems
6149 - change liblustreapi to be useful for external progs like lfsck (3098)
6150 - increase local configuration timeout for slow disks (3353)
6151 - allow configuring ldlm AST timeout - lustre.ldlm_timeout=<seconds>
6153 2004-03-22 Cluster File Systems, Inc. <info@clusterfs.com>
6156 - fixes for glimpse AST timeouts / incorrectly 0-sized files (2818)
6157 - don't overwrite extent policy data in reply if lock was blocked (2901)
6158 - drop filter export grants atomically with removal from device (2663)
6159 - del obd_self_export from work_list in class_disconnect_exports (2908)
6160 - don't LBUG if MDS recovery times out during orphan cleanup (2530)
6161 - swab reply message in mdc_close, other PPC fixes (2464)
6162 - fix destroying of named logs (2325)
6163 - overwrite old logs when running lconf --write_conf (2264)
6164 - bump LLOG_CHUNKSIZE to 8k to allow for larger clusters (2306)
6165 - fix race in target_handle_connect (2898)
6166 - mds_reint_create() should take same inode create lock (2926)
6167 - correct journal credits calculated for CANCEL_UNLINK_LOG (2931)
6168 - don't close files for self_export to avoid uninitialized obd (2936)
6169 - allow MDS with the same name as client node (2939)
6170 - hold dentry reference for closed log files for unlink (2325)
6171 - reserve space for all logs during transactions (2059)
6172 - don't evict page beyond end of stripe extent (2925)
6173 - don't oops on a deleted current working directory (2399)
6174 - handle hard links to targets without a parent properly (2517)
6175 - don't dereference NULL lock when racing during eviction (2867)
6176 - don't grow lock extents when lots of conflicting locks (2919)
6178 2004-03-04 Cluster File Systems, Inc. <info@clusterfs.com>
6181 - account for cache space usage on clients to avoid data loss (974)
6182 - lfsck support in lustre kernel code (2349)
6183 - reduce journal credits needed for BRW writes (2370)
6184 - orphan handling to avoid losing space on client/server crashes
6185 - ptlrpcd can be blocked, stopping ALL progress (2477)
6186 - use lock value blocks to assist in proper KMS, faster stat (1021)
6187 - takes i_sem instead of DLM locks internally on obdfilter (2720)
6188 - recovery for initial connections (2355)
6189 - fixes for mds_cleanup_orphans (1934)
6190 - abort_recovery crashes MDS in b_eq (mds_unlink_orphan) (2584)
6191 - block all file creations until orphan recovery completes (1901)
6192 - client remove rq_connection from request struct (2423)
6193 - conf-sanity test_5, proper cleanup in umount log not availale (2640)
6194 - recovery timer race (2670)
6195 - mdc_close recovey bug (2532)
6196 - ptlrpc cleanup bug (2710)
6197 - mds timeout on local locks (2588)
6198 - namespace lock held during RPCs (2431)
6199 - handle interrupted sync write properly (2503)
6200 - don't try to handle a message that hasn't been replied to (2699)
6201 - client assert failure during cleanup after abort recovery (2701)
6202 - leak mdc device after failed mount (2712)
6203 - ptlrpc_check_set allows timedout requests to complete (2714)
6204 - wait for inflight reqs when ptlrpcd finishes (2710)
6205 - make sure unregistered services are removed from the srv_list
6206 - reset bulk XID's when resending them (caught by 1138 test)
6207 - unregister_bulk after timeout
6208 - fix lconf error (2694)
6209 - handle write after unfinished setstripe, stripe-only getstripe (2388)
6210 - readahead locks pages, leaves pending causing memory pressure (2673)
6211 - increase OST request buffers to 4096 on large machines (2729)
6212 - fix up permission of existing directories in simple_mkdir (2661)
6213 - init deleted item, add assertions ptlrpc_abort_inflight() (2725)
6214 - don't assign transno to errored transactions (2742)
6215 - don't delete objects on OST if given a bogus objid from MDS (2751)
6216 - handle large client PAGE_SIZE readdir on small PAGE_SIZE MDS (2777)
6217 - if rq_no_resend, then timeout request after recovery (2432)
6218 - fix MDS llog_logid record size, 64-bit array alignment (2733)
6219 - don't call usermode_helper from ptlrpcd, DEFAULT upcall (2773)
6220 - put magic in mount.lustre data, check for bad/NULL mount data (2529)
6221 - MDS recovery shouldn't delete objects that it has given out (2730)
6222 - if enqueue arrives after completion, don't clobber LVB (2819)
6223 - don't unlock pages twice when trigger_group_io returns error (2814)
6224 - don't deref NULL rq_repmsg if ldlm_handle_enqueue failed (2822)
6225 - don't write pages to disk if there was an error (1450)
6226 - don't ping imports that have recovery disabled (2676)
6227 - take buffered bytes into account when balancing socknal conn (2817)
6228 - hold a DLM lock over readdir always, use truncate_inode_pages (2706)
6229 - reconnect unlink llog connection after MDS reconnects to OST (2816)
6230 - remove little-endian swabbing of llog records (1987)
6231 - set/limit i_blksize to LL_MAX_BLKSIZE on client (2884)
6232 - retry reposting request buffers if they fail (1191)
6233 - grow extent at grant time to avoid granting a revoked lock (2809)
6234 - lock revoke doesn't evict page if covered by a second lock (2765)
6235 - disable VM readahead to avoid reading outside lock extents (2805)
6237 - return LL_SUPER_MAGIC from statfs for the filesystem type (1972)
6238 - updated kernel patches for hp-2.4.20 kernel (2681)
6240 2004-02-07 Cluster File Systems, Inc. <info@clusterfs.com>
6243 - fix truncated write corruption (2366)
6244 - fix for failed assertion in iopen_connect_dentry (1792,2517)
6246 - don't flag the ptlrpcd thread with PF_MEMALLOC (2636)
6247 - ensure len(uuid) < 37 in lmc (1171)
6248 - fix ia64 OOPS in llog_test (2255)
6249 - zero end of page at obdfilter for partial page writes (2648)
6250 - don't leave stale dentries around after renames (bug 2428)
6251 - fix timeouts when evicting a client with a single lock held (2642)
6252 - set deadline for the initial HELLO message to drain (2634)
6253 - print out dotted-quad IP addresses in the socknal (2302)
6255 - additional debugging for MDS client eviction problem (2443)
6256 - fix mkfsoptions support for osts (2603, 2604)
6258 2004-01-27 Cluster File Systems, Inc. <info@clusterfs.com>
6261 - add series for the vanilla 2.6.0 kernel
6262 - add series for the vanilla 2.4.24 kernel
6263 - add series for a cray x86/64 UL kernel drop
6264 - fix xattr patches for the vanilla 2.4.19 series
6266 - generate true UUIDs in lmc (1171)
6267 - have portals stack dumping break in UML (2466)
6268 - avoid bad dchild deref; avoid inum lock w/o creation (2362)
6269 - allocate with _NOFS in ldlm to avoid deadlock (1933)
6270 - wake callback waiting threads on client eviction (2460)
6271 - Add --ptldebug and --subsystem to lmc (1719)
6272 - update assertion to allow safe interrupt allocation
6273 - set rq_no_resend for cancel requests (2432)
6274 - recalculate ptlrpcd timeout after resend (2494)
6275 - call vfs_rmdir when removing pending directories (2368)
6276 - fix renaming a file to itself (2429)
6277 - lmc creates a default one-stripe lov (2454)
6278 - expand procfs space to handle large clusters (2326)
6279 - increase UML stack to avoid overflow
6280 - update lconf's list of debug and subsystem masks
6281 - fix lfs find --obd (2510)
6282 - /proc tunable for disabling filter read caching (2591)
6283 - stop rpm packages from altering slapd.conf (2301)
6284 - disable nagle in the socknal under 0conf (2578)
6285 - choose mds inode size based on stripe count (2572)
6286 - fix kernel-source rpm problems (2516)
6288 - add --disable-doc to avoid pdf generation (2421)
6289 - update documentation, tests, type-os, comments
6290 - avoid format warnings on ia64
6291 - remove the TOE NAL
6292 - tiny code cleanups by removing unused fields
6294 2004-01-07 Cluster File Systems, Inc. <info@clusterfs.com>
6297 - fix obvious semaphore misuse in as-yet-unused setattr path (2348)
6298 - remove the most blatant lies from BUILDING file (2371)
6299 - change default debug level to reasonable production setting
6300 - reduce client side cache size to reduce cache flush time
6301 - reduce max RPCs in flight to avoid unnecessary file fragmentation
6302 - make TCP zerocopy and pinger support enabled by default (2476)
6303 - sync writes completed after process exits caused crashes (2319)
6304 - maintain correct mount count on the MDS (2356)
6305 - backout 1557, because 2316 wasn't really fixed
6306 - better file I/O statistics gathering in /proc
6307 - don't take unnecessary, deadlock-inducing bug in readpage (2383)
6308 - another kernel patch to fix zero-copy TCP function export
6309 - don't take duplicate lock when processing re-sent getattr (2420)
6310 - lctl uses obd_self_export instead of creating new conn (2353)
6311 - MDS/OST recovery case which requires object creation asserted (2425)
6312 - move lfs from /usr/sbin to /usr/bin in packages
6313 - fix race between mds_client_add and mds_client_free (2417)
6314 - use kmalloc instead of slabs in portals (2430)
6315 - don't create duplicate records when a failover MDS is present (2442)
6316 - remove unnecessary mount age check (2332)
6317 - don't remove directory inodes from locks prematurely (2451)
6318 - don't break if MDS service name is the same as hostname (2103)
6319 - fix races in client write RPC generation when cache full (2482)
6321 2003-12-13 Cluster File Systems, Inc. <info@clusterfs.com>
6324 - remove now-unused request->rq_obd (278)
6325 - if an allocation fails, print out how much memory we've used (1933)
6326 - use PORTAL_SLAB_ALLOC for structures, to get GFP_MEMALLOC (1933)
6327 - add the "configurable stack size" patch to most series files (1256)
6328 - ability to write large log records, for 100+ OST configs (2306)
6329 - fix NULL deref when filter_prep fails (2314)
6330 - fix operator precedence error in filter_sync
6331 - dynamic allocation of socknal TX descriptors (2315)
6332 - fix a missed case in the GFP_MEMALLOC patch, can cause deadlock (2310)
6333 - fix gcc 2.96 compilation problem in xattr kernel patch (2294)
6334 - ensure that CWARN messages in Portals always get to the syslog
6335 - __init/__exit are not for prototype decls (ldlm_init/exit)
6336 - x86-64 compile warning fixes
6337 - fix gateway LMC keyword conflict (2318)
6338 - fix MDS lock inversions in getattr/reint paths (1844)
6339 - fix a rare lock re-ordering bug, which caused deadlock (2322)
6340 - fix i_sem/journal inversion in fsfilt_ext3_write_record (2306)
6341 - DLM race condition prevented some lock evictions (2328)
6342 - ENOMEM detection and retry on socknal sends (2230)
6343 - use GFP_NOFS throughout Lustre, to combat ENOMEM (2230)
6344 - move osc_rpcd into ptlrpc, for use in MDC and others (2329)
6345 - protect MDS inode fsdata with stronger locking; fixes assertion (2313)
6346 - better error messages when a client is rejected during recovery (1505)
6347 - avoid cancelling locks which were never granted, after failure (2330)
6348 - fix i_sem/journal inversion in mds_client_add (2333)
6349 - fix truncate/getattr lock cycle deadlock (2334)
6350 - use rpcd to send close; allows resend after timeout, avoid leak (1897)
6351 - fix two rare exit paths which could leak an l_lock() ref (2321)
6352 - fencepost error in MDS/OST orphan recovery (2226)
6353 - make log record alignment 8 bytes (1988)
6354 - lstripe now fails when requested offset > ost_count (2237)
6355 - ensure that all kernel series have a complete list.h (1607)
6356 - fix crashes in special-file operations (2316)
6357 - lctl create/brw OID mismatch, caused by obsolete filter loop (2339)
6359 - allow configurable automake binary, for testing new versions
6360 - small update to the lfs documentation
6362 2003-12-03 Cluster File Systems, Inc. <info@clusterfs.com>
6364 * fix negative export reference count in fsfilt_sync (2312)
6366 2003-12-01 Cluster File Systems, Inc. <info@clusterfs.com>
6367 * release candidate 0.9.1
6369 - orphans are moved into the PENDING directory for possible recovery
6370 - replayed opens now open by fid for orphan/rename safety (1042)
6371 - last close of an orphan inode generates a transno (683)
6372 - chdir() and mount() now pin the directory entry (1020)
6373 - avoid CERROR in normal ll_setattr_raw() error case (1500)
6374 - discard very old requests without processing them (1502)
6375 - remove some common, well-understood CERRORs (1505)
6376 - require O_DIRECT I/O to be page-sized to workaround IA64 crash (1609)
6377 - clear "grant" flags in OST replies until OST grant code lands (1644)
6378 - fix read performance by not clobbering i_blksize on client (1598)
6379 - fix __ldlm_handle2lock oops by not dereferencing lock after PUT (1625)
6380 - make LRU size a /proc tunable, clears locks when reduced (707)
6381 - fix some lprocfs rot that prevented ptlbd from loading (1732)
6382 - server locks take references on exports now (1558)
6383 - build fixes for 2.4.20-rh trees (1663)
6384 - return an error from lov_create if all OSCs are inactive (1751)
6385 - fix import levels when a reconnect happens without a timeout (1597)
6386 - exit early from mds_open if we get a lookup error (1749)
6387 - partial page read at EOF wouldn't wait for disk before sending (1642)
6388 - avoid NULL deref in obdfilter when reading page past EOF (1592)
6389 - avoid LASSERT in ll_intent_lock if server failed very early (1090)
6390 - fix LBUG in ll_it_open_error with rc = -2 (1861)
6391 - write/truncate lock inversion (1639)
6392 - Don't auto-load obdclass, portals modules during cleanup (1495)
6393 - fix timestamps from jumping to "now" (1763)
6394 - extra journal assertions (1648)
6395 - add an extra multiunlink test (1771)
6396 - fix read_record/write_record API (1776)
6397 - fix leak of offset_extent, possible incorrect i_size later (1772)
6398 - fix lasserts in mis-matched transnos during open-unlink testing (1541)
6399 - Debugging for the kqswnal_get_idle_tx problems (1820)
6400 - Allow recovery to be attempted multiple times (1536)
6401 - Write out MDS last_rcvd file after it is first created (1600)
6402 - Fix tx_descriptor leak in failed transmit situations (1827)
6403 - ext3 journaling fixes for assertion failure after IO error (1871)
6404 - class_export_put() on freed export after completion AST error (1896)
6405 - Fix revalidate looping in VFS (1322)
6406 - Don't access a freed export during MDS_REINT timeout (1521)
6407 - Add open-unlink recovery support on the MDS (1673,1764)
6408 - Return an error if no MDS data was read from last_rcvd (1946)
6409 - Fix for lookup "." or ".." crash on error (1932,1931,1935)
6410 - Don't setup a disk device that doesn't match exported UUID (317)
6411 - Reduce bulk RPC timeout to avoid cascading client/OST failures (1845)
6412 - avoid committing NULL handle in force close
6413 - local.sh is now a one-stripe LOV configuration
6414 - POSIX utime.4 -EPERM on FIFO not owned by user (56)
6415 - fix ext3 htree duplicate directory entry corruption (1516)
6416 - POSIX creat.13, fstat.1, open.18, stat.3 new file atime/mtime (2020)
6417 - update to new LOV EA format (2097)
6418 - interoperability for different PAGE_SIZE/wordsize (686,1821,1343,2042)
6420 2003-06-15 Phil Schwan <phil@clusterfs.com>
6423 - imports and exports cleanup too early, need refcounts (349, 879, 1045)
6424 - per-import/export recovery handling (958, 931, 959)
6425 - multiple last-rcvd slots, for serving multiple FSes (949)
6426 - connections are again shared between multiple imp/exports (963, 964)
6427 - "umount -f" would hang if any requests needed to be sent (393, 978)
6428 - avoid pinning large req buffer by copying for queued messages (989)
6429 - add "uuid" to "lctl device" command to help upcalls (991)
6430 - "open" RPCs with transnos would confuse recovery counters (1037)
6431 - do proper endian conversion of all wire messages (288, 340, 891)
6432 - remove OST bulk get LBUGs, fix ost_brw_write cleanup (1126)
6433 - call waiting locks callback from LDLM recovery thread (1127, 1151)
6434 - fix ptlrpc_connection leak in target_handle_connect (1174)
6435 - fix import refcounting bug in OST and MDS cleanup (1134)
6436 - if an invalid-at-open-time OSC returned before close(), LBUG (1150)
6437 - fix very unlikely obd_types race condition (501)
6438 - remove osc_open hack for echo_client (1187)
6439 - we leaked exports/dlmimps for forcibly disconnected clients (1143)
6440 - a failure in read_inode2 leads to deadlock (1139)
6441 - cancel ack-locks as soon as transaction is committed (1072)
6442 - fix major leaks and crashes in the bulk I/O path (937, 1057)
6443 - make sure to commitrw after any preprw to avoid deadlock (1162)
6444 - failing to execute a file in a lustre FS would lock inode (1203)
6445 - small DEBUG_REQ fix to avoid dereferencing a NULL (1227)
6446 - don't ASSERT while cleaning up an incompletely-setup obd (1248)
6447 - obd_uuid2tgt would walk off the end of the list (1255)
6448 - on IA64 the osc would give portals incorrect bulk size (1258)
6449 - fix debug daemon ioctl interface; allows daemon on ia64 (1274)
6450 - fix lock inversion caused by new llite matching code (1282)
6451 - limit the number of dirty pages on a client to 10MB (1286)
6452 - timed out locks were not being corrected cancelled (1289)
6453 - fix O_DIRECT above 4GB on IA-32 (1292)
6454 * major user-visible changes
6455 - fail out/fail over policy now controlled by the upcall (993)
6457 - add OBD_PING to check server availability and failure (954)
6458 - lustre messages are now sent in sending host order (288, 340, 891)
6459 - add eadatalen to MDS getattr reply (340)
6460 - OST read replies may contain second buffer, with per-page status (593)
6462 2003-03-11 Phil Schwan <phil@clusterfs.com>
6465 - LDLM_DEBUG macro fix, for gcc 3.2 (850)
6466 - failed open()s could cause deadlock; fixed (867, 869)
6467 - stop cancelling OST locks when files are closed (481)
6468 - overlapping XID spaces caused network corruption (851, 853)
6469 - fix unsafe fsfilt counter arithmetic; change to atomic_t
6470 - setattr_raw added, to do single-RPC, server-side setattrs
6471 - lmc/lconf syntax change for OST UUIDs
6472 - fix crashy race condition between ptlrpc_free_req and osc_close
6473 - don't use request in mdc_enqueue if we hit a timeout (889)
6474 - don't set the inode i_size for regular files from the MDS (896)
6475 - handle out of order completion AST (842)
6476 - don't LBUG if a lock request times out after receiving AST (913)
6477 - avoid d_rehash race in ll_find_alias by rehashing inside dcache_lock
6478 - if a bad lock AST arrives, send an error instead of dropping entirely
6479 - return 0 from revalidate2 if ll_intent_lock returns -EINTR (912)
6480 - fix leak in bulk IO when only partially completed (899, 900, 926)
6481 - fix O_DIRECT for ia64 (55)
6482 - (almost) eliminate Lustre-kernel-thread effects on load average (722)
6483 - C-z after timeout could hang a process forever; fixed (977)
6485 - client-side I/O cache (678, 924, 929, 941, 970)
6487 - READPAGE and SETATTRs which don't take server-side locks get
6490 2003-02-11 Phil Schwan <phil@clusterfs.com>
6493 - Fix ldlm_lock_match on the MDS to avoid matching remote locks (592)
6494 - Fix fsfilt_extN_readpage() to read a full page of directory
6495 entries, or fake the remainder if PAGE_SIZE != blocksize (500)
6496 - Avoid extra mdc_getattr() in ll_intent_lock when possible (534, 604)
6497 - Fix imbalanced LOV object allocation and out-of-bound access (469)
6498 - Most intent operations were removed, in favour of a new RPC mode
6499 that does a single RPC to the server and bypasses most of the VFS
6500 - All LDLM resource ID arrays were removed in favour of ldlm_res_id
6501 - Aggressively cancel local locks on DLM servers
6502 - mds_reint_unlink sends EA to the client if it's the last nlink.
6503 client uses that EA to unlink OST objects.
6504 - mds_reint_{rename,unlink,link} were rewritten to take ordered locks
6505 - recursive symlinks were fixed (439)
6506 - fixed NULL deref in DEBUG_REQ
6507 - filter_update_lastobjid no longer calls sync, which annoyed extN
6508 - fixed multi-client small-writes to a single file problem (445)
6509 - fixed mtime updates during file writes (607)
6510 - fixed vector writes on obdfilter causing problems when ENOSPC (670)
6511 - fixed bug in obd_brw_read/write() (under guise of testing 367)
6512 - fixed Linux OST size reporting problem (444, 656)
6513 - OST now updates object mtime with writes or setattr (607, 619)
6514 - client verifies file size before zeroing page past EOF (445)
6515 - OST now writes last allocated objid to disk with allocation (108)
6516 - LOV on echo now works (409)
6518 - mds_reint_unlink sends a new buffer, with the EA included. this
6519 buffer is only valid if body->valid & OBD_MD_FLEASIZE, which is only
6520 set if a regular file was being unlinked, and it was the last link
6521 - use PtlGet from the target for bulk writes (315)
6522 - OST now updates object mtime with writes or setattr (607, 619)
6523 - LDLM now has a grant-time callback to revalidate locked items, if
6525 - Many MDS operations were reorganized to combat race conditions
6527 - Merge b_intel branch (updated lprocfs code) - now at /proc/fs/lustre
6528 - configure check to avoid gcc version 2.96 20000731-2.96-98) (606)
6530 2003-01-06 Andreas Dilger <adilger@clusterfs.com>
6533 - Fully reactivate OST imports after reconnection (512, others)
6534 - Make sure client sees our -ENOTCONN from mds_handle (513 - partial)
6535 - More graceful error handling for truncating on dead OST (515)
6536 - Don't error out unless we're actually accessing dead stripes (474)
6537 - Fix garbage sizes when stripes are missing (410)
6538 - LRU counters were broken, causing constant lock purge (433, 432)
6539 - garbage on read from stripes with failed OSTs (441)
6540 - mark OSCs as active before reconnecting during recovery (438)
6541 - lov_enqueue and lov_cancel need to handle inactive OSTs (403)
6542 - lfind did not preserve OST order in output (443)
6543 - symlinks cause hung clients, incorrect data (439)
6544 - stop dereferencing request after dropping refcount (457)
6545 - don't LASSERT(spin_is_locked) on non-SMP (455)
6546 - fixes for many rename() bugs
6547 - fstat didn't correctly synchronize attributes (399)
6548 - server must handle lock cancellation during blocking AST prep (487)
6549 - bulk descriptors were free()d too soon (511)
6550 - fix paths in lconf, which would load incorrect modules (451, 507)
6551 - fix confusing lconf 'host not found' error message (386)
6552 - fix lock order deadlock on OST (O/R i_sem before journal ops, 478)
6553 - fix race condition in mdc_blocking_ast() for inode access (526)
6554 - fix lov_unpackmd() unpacking wrong number of stripes (537)
6555 - fix lov_set_osc_active() marking wrong OSC inactive (440)
6556 - fix bad lstripe lov_unpackmd() assertion (fix layering too) (527)
6557 - fix multiple writes of stripe MD to MDS (358, maybe 519)
6558 - fix lstripe in several ways (kernel side) (527)
6559 - fix request leak in ldlm_cli_enqueue (262)
6560 - incorrect OSC was marked inactive after OST failure
6561 - call mds_fs_cleanup before unmounting filesystem (524)
6562 - fix races between taking ns_lock and ldlm_lock_change_resource
6563 - fix races updating LOV export open file list
6564 - fix lov_enqueue error path, avoid decref-ing bad lock handle
6565 - fix recovery NULL deref in ldlm_cli_cancel_unused
6566 - fix some DLM races by using new hash table for lock handles (419)
6567 - permit the client to specify desired inodes, at replay
6568 - duplicate requests when we queue them for replay reintegration
6569 - fix last_rcvd offset calculation
6570 - sync after each recovered transaction, so we always make progress
6571 - never, not always, ERESTART requests without transnos
6572 - store the lov_desc in the MDS, so we don't depend on getlovinfo to
6574 - skip replay if the MDS says that the client is already connected
6575 - don't check for a recovery-enabled export to match lctl's UUID
6576 - don't INC_USE_COUNT for phantom exports
6577 - don't crash when cleaning up phantom exports (567)
6578 - don't double-finish or set replay data for errored mdc_open requests
6579 - abort requests when they time out, so we don't get old replies
6580 - send/receive replies for AST messages again
6581 - if the client says that it doesn't have the lock, cancel it on the
6583 - if we timeout during I/O, don't try to cancel an in-use lock; instead
6584 mark it as destroyed, it will all work out when decref is called
6585 - fix module use counts (22, 581)
6587 - ASTs now expect a reply (server cancels lock on error reply)
6589 2002-12-02 Andreas Dilger <adilger@clusterfs.com>
6592 - fix many simultaneous client startup (392)
6593 - fix dentry->d_it clobbering
6594 - credentials weren't being shipped for readdir/getattr operations
6595 - remove invalid assertions triggered during some concurrent MD
6597 - proper Lustre versions added (336, 389)
6598 - fix memory leak for create error case (398)
6599 - fix LOV locking bug that would get cli/srv out of sync
6600 - fix echo client over LOV (409)
6601 - fix dbench 2, extN refcount problem (170, 258, 356, 418)
6602 - fix double-O_EXCL intent crash (424)
6603 - avoid sending multiple lock CANCELs (352)
6605 - MDS can do multi-client recovery (modulo bugs in new code)
6607 - many updates, edits, cleanups
6609 2002-11-18 Phil Schwan <phil@clusterfs.com>
6612 - fix null d_it dereference (346)
6613 - fix full OST/dbench hang (333)
6614 - fix permission problem with file removal (286)
6615 - fix removal of OSCs from LOV when they fail
6616 - fix NULL deref during bulk timeout (214)
6617 - fix problems related to multiple filesystems on one MDS (241)
6618 - fixed serious subtle metadata locking bugs
6619 - free locks on clients when inodes are removed due to memory
6621 - fix inode pointer in lock data (285)
6622 - partial support for multiple MDS on a single host (241)
6623 - data locks weren't cancelled at clear_inode time (290, 311)
6624 - intent locks could lead to unbounded lock growth (205)
6625 - added a maximum lock count, an LRU list, and a flusher
6626 - fix multiple rename (365)
6627 - properly abstracted the echo client
6628 - OSC locked 1 byte too many; fixed
6629 - rewrote brw callback code:
6630 - fixed recovery bugs related to LOVs (306)
6631 - fixed too-many-pages-in-one-write crash (191)
6632 - fixed (again) crash in sync_io_timeout (214)
6633 - probably fixed callback-related race (385)
6635 - Add capability to MDS protocol
6636 - LDLM cancellations and callbacks on different portals
6638 2002-10-28 Andreas Dilger <adilger@clusterfs.com>
6641 - limit client IOV size to PTL_MD_MAX_IOV (611336, 191)
6642 - defer open object destruction to close time (601981, 138)
6643 - open/close OST file handle in obdo (OBD_MD_FLHANDLE) (601981, 138)
6644 - move LDLM_ENQUEUE/CONVERT back to MDS portal (625069)
6645 - abstract ll_lookup2, fix ll_revalidate2 to use abstraction (256)
6646 - don't call obd_setattr in ll_file_release for destroyed objects
6647 * protocol change to lustre_msg: move |version| and add |flags|
6648 * protocol change to osc_punch: "start" in "o_size", "end" in "o_blocks"
6649 * lock replay: for LDLM_FL_REPLAY trust client to do right thing
6650 * added replay of create, unlink, link and rename operations during
6651 MDS failover; recovery should be much more robust now
6652 * remove failed OSCs from LOVs (only lov_create uses this so far)
6653 * the lustre-HOWTO was brought (more) up to date (582544)
6655 2002-10-23 Phil Schwan <phil@clusterfs.com>
6658 - in-use dentries weren't being reused properly (617851)
6659 - prevent multiple LDLM setup (599178)
6660 - fix LOV size calculations for truncate (617853)
6661 - fix client handling of MDS intent errors (POSIX)
6662 - fix permission bug in lovstripe.c test (624321)
6663 - fix MDS thread deadlock - move LDLM handler to DLM portal (625069)
6664 - truncate past end of file could corrupt data
6665 - proper cleanup after timeouts, crashes, etc (592524, 550815)
6666 - a race in recovery could return ETIMEDOUT to apps (623947)
6667 - building outside the source directory was fixed
6668 * the lustre-HOWTO was brought (more) up to date (582544)
6669 * major progress was made on recovery functionality
6671 2002-10-10 Phil Schwan <phil@clusterfs.com>
6674 - recovery deadlock fix
6675 - rm -rf causes LBUG fix (617817)
6676 - file open by multiple tasks fix (618962)
6677 - directory permissions bugs (602707 and 620007)
6678 - journal_stop fixed with locking (611313)
6679 - O_APPEND failures resolved (618273, perhaps 614459)
6680 - lconf PATH fix (619770)
6681 - IA64 build fix (621450)
6682 - RPC buffer sizes scale with amount of memory
6684 2002-10-01 Phil Schwan <phil@clusterfs.com>
6687 - locks would be cancelled without throwing away data pages,
6688 resulting in inconsistent data (605627)
6689 - inode attributes were not always being refreshed (605627, 612449)
6690 - lconf now continues to cleanup after lctl reports an error
6691 - MDS now enforces user permissions (602707)
6692 - lprocfs cleanup fixed, but not yet enabled (614157)
6693 - fixed infinite server hang, should a client not respond to an AST
6694 - avoid going into recovery if user calls readlink() with a buffer
6695 that's too small (613941)
6696 - AST RPCs no longer require replies (614867) -- this may be changed
6697 - don't crash server if client sends an IOV that's too big (611336)
6698 - fixed lock conversion deadlock (611892)
6699 - fixed the following of symlinks (614622)
6700 * recovery: the server can remove locks from a client that dies, other
6701 clients can make progress
6702 * more extN patch fixes
6703 * compile-time configurable ptlrpc buffer allocations
6705 - collaborative read cache document
6706 - Lustre Lite Performance CDR document-in-progress
6708 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
6711 - fix typo in patch-2.4.18
6713 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
6716 - clear ptlrpc request each time in handle_incoming_request()
6717 - unlink of files now destroys the object on the OST
6719 2002-09-19 Peter Braam <braam@clusterfs.com>
6721 * add hard link support
6722 * change obdfile creation method
6723 * kernel patch changed
6725 2002-09-19 Peter Braam <braam@clusterfs.com>
6728 - stack overflow bug in extN fixed
6730 2002-09-18 Andreas Dilger <adilger@clusterfs.com>
6732 * documentation updates
6733 - add man pages for config tools
6734 - update tests/README to describe testing with new config tools
6735 - finish metadata API descriptions
6736 * bug fixes and cleanups
6737 - statfs workaround for 16TB limit
6738 - LOV stripe allocation improved, can stripe on subset of OSTs
6739 - LOV file size/IO offset was wrong for files > 4GB in size
6740 - object EA data was being dropped, caused files to be unreadable
6741 - memory overflow with non-LOV OST caused memory corruption
6742 - fixed regression tests to work with new config tools, obdfilter
6743 - fixed bug when directory size became larger than 1 block
6744 - fixed bug (for single client case) when PWD was deleted
6745 - invalidate local directory pages when doing intent-based ops
6746 - avoid LDLM oops when lock callback contained bad data
6748 2002-09-09 Andreas Dilger <adilger@clusterfs.com>
6750 * documentation updates
6751 * bug fixes and cleanups
6752 - configuration tools
6755 - 64-bit compile warnings
6756 - 64-bit internal statfs data
6758 * test_brw on persistent OST devices
6760 * lprocfs (disabled)
6762 2002-09-04 Andreas Dilger <adilger@clusterfs.com>
6764 * documentation updates
6765 * bug fixes and cleanups
6766 * configuration tools
6768 2002-08-30 Peter J. Braam <braam@clusterfs.com>
6771 * many small fixes to 0_5_4
6772 * io/network handling
6773 * thinkos in MDS operations
6775 2002-08-24 Peter J. Braam <braam@clusterfs.com>
6778 * crucial basic fixes to 0.5.3
6779 * IOR, Iozone work over Elan
6782 2002-08-07 Phil Schwan <phil@clusterfs.com>
6783 * version 0_5_3, our first alpha
6784 * we use the new Portals iovs
6785 * documentation updates
6786 * bug fixes and cleanups
6787 * small changes in the DLM wire protocol
6789 2002-07-25 Peter J. Braam <braam@clusterfs.com>
6790 * version 0_5_1 with some initial stability,
6791 * locking on MD and file I/O.
6792 * documentation updates
6793 * several bug fixes since 0.5.0
6794 * small changes in wire protocol
6796 2002-07-18 Phil Schwan <phil@clusterfs.com>
6798 * delivered as Lustre Light Alpha
6799 * fixed a crash after handling invalid MDS requests
6800 * fixed directory pages for architectures with non-4k pages sizes
6802 2002-07-11 Andreas Dilger <adilger@clusterfs.com>
6803 * release version v0_4_4
6804 * Moves TCP acceptor to be on port 2432 (unused Coda port) instead
6806 * Fixes a number of interruption problems with OST operations.
6807 * Update documentation for portals header changes
6808 * Move all wire protocol structs/defines to lustre_idl.h
6809 * Fixes symlink length bug.
6810 * Add tcpdump to repository.
6812 2002-07-05 Andreas Dilger <adilger@clusterfs.com>
6813 * release version v0_4_3
6814 * Fixes statfs for inodes on extN.
6815 * Fixes bug in runtests which would delete /etc/hosts.
6816 * Use 64-bit object IDs wherever possible (not into VFS though)
6817 Remove ost_get_info, which is unused by lustre, and out of date.
6819 2002-07-03 Peter Braam <braam@clusterfs.com>
6820 * release version v0_4_2 Fixes a lookup error (type not passed)
6821 * move forward to head of Portals
6822 * move forward to latest Lustre kernel
6824 2002-06-25 Peter Braam <braam@clusterfs.com>
6825 * release version v0_4_1. Hopefully stable on single node use.