3 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
4 for currently supported client and server kernel versions.
5 * Server primary kernels built and tested during release cycle:
6 3.10.0-957.27.2.el7 (RHEL7.6)
7 3.10.0-1062.18.1.el7 (RHEL7.7)
8 3.10.0-1127.8.2.el7 (RHEL7.8)
9 * Server known to build on the following kernels (others may also work):
10 2.6.32-696.18.7.el6 (RHEL6.9)
11 3.10.0-862.14.4.el7 (RHEL7.5)
12 4.4.120-92.70 (SLES12 SP2)
13 4.4.180-94.97 (SLES12 SP3)
14 4.4.0-142 (Ubuntu 16.04)
15 4.15.0-45 (Ubuntu 18.04)
16 vanilla linux 4.6.7 (ZFS only)
17 * ldiskfs needs an ldiskfs patch series for that kernel, ZFS does not
18 * Client primary unpatched kernels built and tested during release cycle:
19 3.10.0-957.27.2.el7 (RHEL7.6)
20 3.10.0-1062.18.1.el7 (RHEL7.7)
21 3.10.0-1127.8.2.el7 (RHEL7.8)
22 4.18.0-80.11.2.el8 (RHEL8.0)
23 4.4.180-94.97 (SLES12 SP3)
24 4.12.14-95.45 (SLES12 SP4)
25 4.12.14-122.20 (SLES12 SP5)
26 4.15.0-45 (Ubuntu 18.04)
27 * Client known to build on unpatched kernels (others may also work):
28 2.6.32-696.18.7.el6 (RHEL6.9)
29 2.6.32-754.6.3.el6 (RHEL6.10)
30 3.10.0-862.14.4.el7 (RHEL7.5)
31 4.14.0-49.13.1.el7a (RHEL7.5)
32 4.14.0-115.2.2.el7a (RHEL7.6)
33 4.18.0-147.3.1.el8 (RHEL8.1)
34 4.18.0-193.1.2.el8 (RHEL8.2)
35 4.4.120-92.70 (SLES12 SP2)
36 4.4.0-142 (Ubuntu 16.04)
37 vanilla linux 4.15.0 (ZFS only)
38 * Recommended e2fsprogs version: 1.45.2.wc1 or newer
39 * Recommended ZFS / SPL version: 0.7.13
40 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
41 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
42 stack overflow. For more information, please refer to bugzilla 17630.
43 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
44 the re-exporting nodes to be mounted with "32bitapi" mount option
45 * Note for Ubuntu 14.04.5:
46 The 4.4.0 series is the HWE Kernel derived from 16.04 (Xenial).
47 See: https://wiki.ubuntu.com/Kernel/LTSEnablementStack
51 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
52 for currently supported client and server kernel versions.
53 * Server primary kernels built and tested during release cycle:
54 3.10.0-957.27.2.el7 (RHEL7.6)
55 3.10.0-1062.9.1.el7 (RHEL7.7)
56 * Server known to build on the following kernels (others may also work):
57 2.6.32-696.18.7.el6 (RHEL6.9)
58 3.10.0-862.14.4.el7 (RHEL7.5)
59 4.4.120-92.70 (SLES12 SP2)
60 4.4.180-94.97 (SLES12 SP3)
61 4.4.0-142 (Ubuntu 16.04)
62 4.15.0-45 (Ubuntu 18.04)
63 vanilla linux 4.6.7 (ZFS only)
64 * ldiskfs needs an ldiskfs patch series for that kernel, ZFS does not
65 * Client primary unpatched kernels built and tested during release cycle:
66 3.10.0-957.27.2.el7 (RHEL7.6)
67 3.10.0-1062.9.1.el7 (RHEL7.7)
68 4.18.0-80.11.2.el8 (RHEL8.0)
69 4.4.180-94.97 (SLES12 SP3)
70 4.12.14-95.29 (SLES12 SP4)
71 4.15.0-45 (Ubuntu 18.04)
72 * Client known to build on unpatched kernels (others may also work):
73 2.6.32-696.18.7.el6 (RHEL6.9)
74 2.6.32-754.6.3.el6 (RHEL6.10)
75 3.10.0-862.14.4.el7 (RHEL7.5)
76 4.14.0-49.13.1.el7a (RHEL7.5)
77 4.14.0-115.2.2.el7a (RHEL7.6)
78 4.18.0-147.3.1.el8 (RHEL8.1)
79 4.4.120-92.70 (SLES12 SP2)
80 4.4.0-142 (Ubuntu 16.04)
81 vanilla linux 4.15.0 (ZFS only)
82 * Recommended e2fsprogs version: 1.45.2.wc1 or newer
83 * Recommended ZFS / SPL version: 0.7.13
84 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
85 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
86 stack overflow. For more information, please refer to bugzilla 17630.
87 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
88 the re-exporting nodes to be mounted with "32bitapi" mount option
89 * Note for Ubuntu 14.04.5:
90 The 4.4.0 series is the HWE Kernel derived from 16.04 (Xenial).
91 See: https://wiki.ubuntu.com/Kernel/LTSEnablementStack
95 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
96 for currently supported client and server kernel versions.
97 * Server primary kernels built and tested during release cycle:
98 3.10.0-957.27.2.el7 (RHEL7.6)
99 3.10.0-1062.1.1.el7 (RHEL7.7)
100 * Server known to build on the following kernels (others may also work):
101 2.6.32-696.18.7.el6 (RHEL6.9)
102 3.10.0-862.14.4.el7 (RHEL7.5)
103 4.4.120-92.70 (SLES12 SP2)
104 4.4.180-94.97 (SLES12 SP3)
105 4.4.0-142 (Ubuntu 16.04)
106 4.15.0-45 (Ubuntu 18.04)
107 vanilla linux 4.6.7 (ZFS only)
108 * ldiskfs needs an ldiskfs patch series for that kernel, ZFS does not
109 * Client primary unpatched kernels built and tested during release cycle:
110 3.10.0-957.27.2.el7 (RHEL7.6)
111 3.10.0-1062.1.1.el7 (RHEL7.7)
112 4.18.0-80.7.1.el8 (RHEL8.0)
113 4.4.180-94.97 (SLES12 SP3)
114 4.12.14-95.29 (SLES12 SP4)
115 4.15.0-45 (Ubuntu 18.04)
116 * Client known to build on unpatched kernels (others may also work):
117 2.6.32-696.18.7.el6 (RHEL6.9)
118 2.6.32-754.6.3.el6 (RHEL6.10)
119 3.10.0-862.14.4.el7 (RHEL7.5)
120 4.14.0-49.13.1.el7a (RHEL7.5)
121 4.14.0-115.2.2.el7a (RHEL7.6)
122 4.4.120-92.70 (SLES12 SP2)
123 4.4.0-142 (Ubuntu 16.04)
124 vanilla linux 4.15.0 (ZFS only)
125 * Recommended e2fsprogs version: 1.45.2.wc1 or newer
126 * Recommended ZFS / SPL version: 0.7.13
127 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
128 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
129 stack overflow. For more information, please refer to bugzilla 17630.
130 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
131 the re-exporting nodes to be mounted with "32bitapi" mount option
132 * Note for Ubuntu 14.04.5:
133 The 4.4.0 series is the HWE Kernel derived from 16.04 (Xenial).
134 See: https://wiki.ubuntu.com/Kernel/LTSEnablementStack
138 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
139 for currently supported client and server kernel versions.
140 * Server known to build on the following kernels (others may also work):
141 2.6.32-696.18.7.el6 (RHEL6.9)
142 3.10.0-862.14.4.el7 (RHEL7.5)
143 3.10.0-957.10.1.el7 (RHEL7.6)
144 4.4.120-92.70 (SLES12 SP2)
145 4.4.162-94.69 (SLES12 SP3)
146 4.4.0-142 (Ubuntu 16.04)
147 4.15.0-45 (Ubuntu 18.04)
148 vanilla linux 4.6.7 (ZFS only)
149 * ldiskfs needs an ldiskfs patch series for that kernel, ZFS does not
150 * Client known to build on unpatched kernels (others may also work):
151 2.6.32-696.18.7.el6 (RHEL6.9)
152 2.6.32-754.6.3.el6 (RHEL6.10)
153 3.10.0-862.14.4.el7 (RHEL7.5)
154 3.10.0-957.10.1.el7 (RHEL7.6)
155 4.14.0-49.13.1.el7a (RHEL7.5)
156 4.14.0-115.2.2.el7a (RHEL7.6)
157 4.4.120-92.70 (SLES12 SP2)
158 4.4.162-94.69 (SLES12 SP3)
159 4.12.14-95.13 (SLES12 SP4)
160 4.4.0-142 (Ubuntu 16.04)
161 4.15.0-45 (Ubuntu 18.04)
162 vanilla linux 4.15.0 (ZFS only)
163 * Recommended e2fsprogs version: 1.44.5.wc1 or newer
164 * Recommended ZFS / SPL version: 0.7.13
165 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
166 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
167 stack overflow. For more information, please refer to bugzilla 17630.
168 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
169 the re-exporting nodes to be mounted with "32bitapi" mount option
170 * Note for Ubuntu 14.04.5:
171 The 4.4.0 series is the HWE Kernel derived from 16.04 (Xenial).
172 See: https://wiki.ubuntu.com/Kernel/LTSEnablementStack
177 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
178 for currently supported client and server kernel versions.
179 * Server known to build on the following kernels (others may also work):
180 2.6.32-696.18.7.el6 (RHEL6.9)
181 3.10.0-862.14.4.el7 (RHEL7.5)
182 3.10.0-957.10.1.el7 (RHEL7.6)
183 4.4.120-92.70 (SLES12 SP2)
184 4.4.162-94.69 (SLES12 SP3)
185 4.4.0-142 (Ubuntu 16.04)
186 4.15.0-45 (Ubuntu 18.04)
187 vanilla linux 4.6.7 (ZFS only)
188 * ldiskfs needs an ldiskfs patch series for that kernel, ZFS does not
189 * Client known to build on unpatched kernels (others may also work):
190 2.6.32-696.18.7.el6 (RHEL6.9)
191 2.6.32-754.6.3.el6 (RHEL6.10)
192 3.10.0-862.14.4.el7 (RHEL7.5)
193 3.10.0-957.10.1.el7 (RHEL7.6)
194 4.14.0-49.13.1.el7a (RHEL7.5)
195 4.14.0-115.2.2.el7a (RHEL7.6)
196 4.4.120-92.70 (SLES12 SP2)
197 4.4.162-94.69 (SLES12 SP3)
198 4.12.14-95.13 (SLES12 SP4)
199 4.4.0-142 (Ubuntu 16.04)
200 4.15.0-45 (Ubuntu 18.04)
201 vanilla linux 4.15.0 (ZFS only)
202 * Recommended e2fsprogs version: 1.44.5.wc1 or newer
203 * Recommended ZFS / SPL version: 0.7.13
204 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
205 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
206 stack overflow. For more information, please refer to bugzilla 17630.
207 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
208 the re-exporting nodes to be mounted with "32bitapi" mount option
209 * Note for Ubuntu 14.04.5:
210 The 4.4.0 series is the HWE Kernel derived from 16.04 (Xenial).
211 See: https://wiki.ubuntu.com/Kernel/LTSEnablementStack
215 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
216 for currently supported client and server kernel versions.
217 * Server known to build on the following kernels (others may also work):
218 2.6.32-696.18.7.el6 (RHEL6.9)
219 3.10.0-862.14.4.el7 (RHEL7.5)
220 3.10.0-957.el7 (RHEL7.6)
221 4.4.120-92.70 (SLES12 SP2)
222 4.4.162-94.69 (SLES12 SP3)
223 4.4.0-131 (Ubuntu 16.04)
224 4.15.0-32 (Ubuntu 18.04)
225 vanilla linux 4.6.7 (ZFS only)
226 * ldiskfs needs an ldiskfs patch series for that kernel, ZFS does not
227 * Client known to build on unpatched kernels (others may also work):
228 2.6.32-696.18.7.el6 (RHEL6.9)
229 2.6.32-754.6.3.el6 (RHEL6.10)
230 3.10.0-862.14.4.el7 (RHEL7.5)
231 3.10.0-957.el7 (RHEL7.6)
232 4.14.0-49.13.1.el7a (RHEL7.5)
233 4.14.0-115.2.2.el7a (RHEL7.6)
234 4.4.120-92.70 (SLES12 SP2)
235 4.4.162-94.69 (SLES12 SP3)
236 4.4.0-131 (Ubuntu 16.04)
237 4.15.0-32 (Ubuntu 18.04)
238 vanilla linux 4.15.0 (ZFS only)
239 * Recommended e2fsprogs version: 1.44.3.wc1 or newer
240 * Recommended ZFS / SPL version: 0.7.9
241 * Tested with ZFS / SPL version: 0.7.9
242 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
243 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
244 stack overflow. For more information, please refer to bugzilla 17630.
245 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
246 the re-exporting nodes to be mounted with "32bitapi" mount option
247 * Note for Ubuntu 14.04.5:
248 The 4.4.0 series is the HWE Kernel derived from 16.04 (Xenial).
249 See: https://wiki.ubuntu.com/Kernel/LTSEnablementStack
251 --------------------------------------------------------------------------------
254 2018-04-03 Intel Corporation
256 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
257 for currently supported client and server kernel versions.
258 * Server known to build on patched kernels:
259 2.6.32-431.29.2.el6 (RHEL6.5)
260 2.6.32-504.30.3.el6 (RHEL6.6)
261 2.6.32-573.26.1.el6 (RHEL6.7)
262 2.6.32-642.15.1.el6 (RHEL6.8)
263 2.6.32-696.18.7.el6 (RHEL6.9)
264 3.10.0-693.21.1.el7 (RHEL7.4)
265 3.0.101-0.47.71 (SLES11 SP3)
266 3.0.101-107 (SLES11 SP4)
267 3.12.74-60.64.40 (SLES12 SP1)
268 4.4.114-92.64 (SLES12 SP2)
269 4.4.114-94.11 (SLES12 SP3)
270 3.13.0-101 (Ubuntu 14.04, ZFS only)
271 4.4.0-85.108 (Ubuntu 14.04.5 LTS)
272 4.4.0-116 (Ubuntu 16.04)
273 vanilla linux 4.6.7 (ZFS only)
274 * Client known to build on unpatched kernels:
275 2.6.32-431.29.2.el6 (RHEL6.5)
276 2.6.32-504.30.3.el6 (RHEL6.6)
277 2.6.32-573.26.1.el6 (RHEL6.7)
278 2.6.32-642.15.1.el6 (RHEL6.8)
279 2.6.32-696.18.7.el6 (RHEL6.9)
280 3.10.0-693.21.1.el7 (RHEL7.4)
281 3.0.101-0.47.71 (SLES11 SP3)
282 3.0.101-107 (SLES11 SP4)
283 3.12.74-60.64.40 (SLES12 SP1)
284 4.4.114-92.64 (SLES12 SP2)
285 4.4.114-94.11 (SLES12 SP3)
286 3.13.0-101 (Ubuntu 14.04)
287 4.4.0-85.108 (Ubuntu 14.04.5 LTS)
288 4.4.0-116 (Ubuntu 16.04)
290 * Recommended e2fsprogs version: 1.42.13.wc6 or newer
291 * Recommended ZFS / SPL version: 0.7.6
292 * Tested with ZFS / SPL version: 0.7.6
293 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
294 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
295 stack overflow. For more information, please refer to bugzilla 17630.
296 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
297 the re-exporting nodes to be mounted with "32bitapi" mount option
298 * Note for Ubuntu 14.04.5:
299 The 4.4.0 series is the HWE Kernel derived from 16.04 (Xenial).
300 See: https://wiki.ubuntu.com/Kernel/LTSEnablementStack
302 --------------------------------------------------------------------------------
304 2017-07-05 Intel Corporation
306 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
307 for currently supported client and server kernel versions.
308 * Server known to build on patched kernels:
309 2.6.32-431.29.2.el6 (RHEL6.5)
310 2.6.32-504.30.3.el6 (RHEL6.6)
311 2.6.32-573.26.1.el6 (RHEL6.7)
312 2.6.32-642.15.1.el6 (RHEL6.8)
313 2.6.32-696.3.1.el6 (RHEL6.9)
314 3.10.0-514.21.1.el7 (RHEL7.3)
315 3.0.101-0.47.71 (SLES11 SP3)
316 3.0.101-100 (SLES11 SP4)
317 3.12.74-60.64.40 (SLES12 SP1)
318 4.4.59-92.17 (SLES12 SP2)
319 vanilla linux 4.6.7 (ZFS only)
320 * Client known to build on unpatched kernels:
321 2.6.32-431.29.2.el6 (RHEL6.5)
322 2.6.32-504.30.3.el6 (RHEL6.6)
323 2.6.32-573.26.1.el6 (RHEL6.7)
324 2.6.32-642.15.1.el6 (RHEL6.8)
325 2.6.32-696.3.1.el6 (RHEL6.9)
326 3.10.0-514.21.1.el7 (RHEL7.3)
327 3.0.101-0.47.71 (SLES11 SP3)
328 3.0.101-100 (SLES11 SP4)
329 3.12.74-60.64.40 (SLES12 SP1)
330 4.4.59-92.17 (SLES12 SP2)
332 * Recommended e2fsprogs version: 1.42.13.wc4 or newer
333 * Recommended ZFS / SPL version: 0.6.5.9
334 * Tested with ZFS / SPL version: 0.6.5.9
335 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
336 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
337 stack overflow. For more information, please refer to bugzilla 17630.
338 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
339 the re-exporting nodes to be mounted with "32bitapi" mount option
341 --------------------------------------------------------------------------------
343 12-07-2016 Intel Corporation
345 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
346 for currently supported client and server kernel versions.
347 * Server known to build on patched kernels:
348 2.6.32-431.29.2.el6 (RHEL6.5)
349 2.6.32-504.30.3.el6 (RHEL6.6)
350 2.6.32-573.26.1.el6 (RHEL6.7)
351 2.6.32-642.11.1.el6 (RHEL6.8)
352 3.10.0-514.el7 (RHEL7.3)
353 3.0.101-0.47.71 (SLES11 SP3)
354 3.0.101-80 (SLES11 SP4)
355 3.12.62-60.64.8 (SLES12 SP1)
356 vanilla linux 4.5.7 (ZFS only)
357 * Client known to build on unpatched kernels:
358 2.6.32-431.29.2.el6 (RHEL6.5)
359 2.6.32-504.30.3.el6 (RHEL6.6)
360 2.6.32-573.26.1.el6 (RHEL6.7)
361 2.6.32-642.11.1.el6 (RHEL6.8)
362 3.10.0-514.el7 (RHEL7.3)
363 3.0.101-0.47.71 (SLES11 SP3)
364 3.0.101-80 (SLES11 SP4)
365 3.12.62-60.64.8 (SLES12 SP1)
367 * Recommended e2fsprogs version: 1.42.13.wc5 or newer
368 * Recommended ZFS / SPL version: 0.6.5.8
369 * Tested with ZFS / SPL version: 0.6.5.8
370 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
371 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
372 stack overflow. For more information, please refer to bugzilla 17630.
373 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
374 the re-exporting nodes to be mounted with "32bitapi" mount option
376 --------------------------------------------------------------------------------
378 02-29-2016 Intel Corporation
380 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
381 for currently supported client and server kernel versions.
382 * Server known to build on patched kernels:
383 2.6.32-431.29.2.el6 (RHEL6.5)
384 2.6.32-504.30.3.el6 (RHEL6.6)
385 2.6.32-573.12.1.el6 (RHEL6.7)
386 3.10.0-327.3.1.el7 (RHEL7.2)
387 3.0.101-0.47.71 (SLES11 SP3)
388 3.0.101-68 (SLES11 SP4)
389 vanilla linux 4.2.1 (ZFS only)
390 * Client known to build on unpatched kernels:
391 2.6.32-431.29.2.el6 (RHEL6.5)
392 2.6.32-504.30.3.el6 (RHEL6.6)
393 2.6.32-573.12.1.el6 (RHEL6.7)
394 3.10.0-327.3.1.el7 (RHEL7.2)
395 3.0.101-0.47.71 (SLES11 SP3)
396 3.0.101-68 (SLES11 SP4)
399 * Recommended e2fsprogs version: 1.42.13.wc4 or newer
400 * Recommended ZFS / SPL version: 0.6.4.2
401 * Tested with ZFS / SPL version: 0.6.4.2
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 03-10-2015 Intel Corporation
412 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
413 for currently supported client and server kernel versions.
414 * Server known to build on patched kernels:
415 2.6.32-431.29.2.el6 (RHEL6.5)
416 2.6.32-504.8.1.el6 (RHEL6.6)
417 3.0.101-0.46 (SLES11 SP3)
418 * Client known to build on unpatched kernels:
419 2.6.32-431.29.2.el6 (RHEL6.5)
420 2.6.32-504.8.1.el6 (RHEL6.6)
421 3.10.0-123.20.1.el7 (RHEL7)
422 3.0.101-0.46 (SLES11 SP3)
423 * Recommended e2fsprogs version: 1.42.9.wc1 or newer
424 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
425 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
426 stack overflow. For more information, please refer to bugzilla 17630.
427 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
428 the re-exporting nodes to be mounted with "32bitapi" mount option
430 Severity : enhancement
432 Description: control OST-index in IDIF via ROCOMPAT flag.
433 Details : Introduce new flag OBD_ROCOMPAT_IDX_IN_IDIF that is stored in the
434 last_rcvd file. For new formatted OST device, it will be auto set;
435 for the case of upgrading from old OST device, you can enable it
436 via the lproc interface osd-ldiskfs.index_in_idif. With such flag
437 enabled, for new created OST-object, its IDIF-in-LMA will contain
438 the OST-index; for the existing OST-object, the OSD will convert
439 old format IDIF as new format IDIF with OST-index stored in the
440 LMA EA when accessing such OST-object or via OI scrub. Once such
441 flag is enabled, it cannot be reverted back, so the system cannot
442 be downgraded to the orignal incompatible version.
444 --------------------------------------------------------------------------------
446 07-30-2014 Intel Corporation
448 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
449 for currently supported client and server kernel versions.
450 * Server known to build on patched kernels:
451 2.6.32-431.20.3.el6 (RHEL6.5)
452 3.0.101-0.31 (SLES11 SP3)
453 * Client known to build on unpatched kernels:
454 2.6.32-431.20.3.el6 (RHEL6.5)
455 3.0.101-0.31 (SLES11 SP3)
458 * Recommended e2fsprogs version: 1.42.9.wc1 or newer
459 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
460 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
461 stack overflow. For more information, please refer to bugzilla 17630.
462 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
463 the re-exporting nodes to be mounted with "32bitapi" mount option
465 --------------------------------------------------------------------------------
467 10-31-2013 Intel Corporation
469 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
470 * Server builds for kernels:
471 2.6.32-358.18.1.el6 (RHEL6.4)
472 * Client builds for unpatched kernels:
473 2.6.32-358.18.1.el6 (RHEL6.4)
474 3.0.92-0.8 (SLES11 SP3)
476 * Recommended e2fsprogs version: 1.42.7.wc2 or newer
477 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
478 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
479 stack overflow. For more information, please refer to bugzilla 17630.
480 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
481 the re-exporting nodes to be mounted with "32bitapi" mount option
483 --------------------------------------------------------------------------------
485 05-30-2013 Intel Corporation
487 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
488 * Server builds for kernels:
489 2.6.32-358.6.2.el6 (RHEL6.4)
490 * Client builds for unpatched kernels:
491 2.6.32-358.6.2.el6 (RHEL6.4)
492 3.0.74-0.6.6 (SLES11 SP2)
494 * Recommended e2fsprogs version: 1.42.7.wc1 or newer
495 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
496 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
497 stack overflow. For more information, please refer to bugzilla 17630.
498 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
499 the re-exporting nodes to be mounted with "32bitapi" mount option
501 --------------------------------------------------------------------------------
503 9-30-2012 Whamcloud, Inc.
505 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
506 * Server support for kernels:
507 2.6.32-279.5.1.el6 (RHEL6)
508 * Client support for unpatched kernels:
509 2.6.32-279.5.1.el6 (RHEL6)
511 2.6.18-238.19.1.el5 (RHEL5)
512 2.6.16.60-0.69.1 (SLES10)
513 * Recommended e2fsprogs version: 1.41.90.wc4
514 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
515 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
516 stack overflow. For more information, please refer to bugzilla 17630.
517 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
518 the re-exporting nodes to be mounted with "32bitapi" mount option
520 Severity : enhancement
522 Description: Update RHEL6.3 kernel to 2.6.32-279.5.1.el6.
524 Severity : enhancement
526 Description: Update RHEL6.3 kernel to 2.6.32-279.2.1.el6.
528 Severity : enhancement
530 Description: Update RHEL6.2 kernel to 2.6.32-220.17.1.el6
532 Severity : enhancement
534 Description: Update RHEL6.2 kernel to 2.6.32-220.13.1.el6
536 Severity : enhancement
538 Description: Update RHEL6.2 kernel to 2.6.32-220.7.1.el6
540 --------------------------------------------------------------------------------
542 04-30-2012 Whamcloud, Inc.
544 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
545 * Server support for kernels:
546 2.6.32-220.4.2.el6 (RHEL6)
547 * Client support for unpatched kernels:
548 2.6.32-220.4.2.el6 (RHEL6)
550 2.6.18-238.19.1.el5 (RHEL5)
551 2.6.16.60-0.69.1 (SLES10)
552 * Recommended e2fsprogs version: 1.41.90.wc4
553 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
554 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
555 stack overflow. For more information, please refer to bugzilla 17630.
556 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
557 the re-exporting nodes to be mounted with "32bitapi" mount option
559 Severity : enhancement
561 Description: Update RHEL6.2 kernel to 2.6.32-220.4.2.el6
563 Severity : enhancement
565 Description: Update RHEL6.2 kernel to 2.6.32-220.4.1.el6
567 Severity : enhancement
569 Description: Update RHEL6.2 kernel to 2.6.32-220.el6
571 Severity : enhancement
573 Description: Update RHEL6.1 kernel to 2.6.32-131.17.1.el6
575 --------------------------------------------------------------------------------
577 09-30-2011 Whamcloud, Inc.
579 * Server support for kernels:
580 2.6.18-238.19.1.el5 (RHEL5)
581 2.6.32-131.6.1.el6 (RHEL6)
582 * Client support for unpatched kernels:
583 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
584 2.6.18-238.19.1.el5 (RHEL5)
585 2.6.32-131.6.1.el6 (RHEL6)
586 2.6.29.4-167.fc11 (FC11)
587 2.6.16.60-0.69.1 (SLES10)
589 * Recommended e2fsprogs version: 1.41.90.wc3
590 * NFS export disabled when stack size < 8192. Since the NFSv4 export of
591 Lustre filesystem with 4K stack may cause a stack overflow. For more
592 information, please refer to bugzilla 17630.
593 * NFSv4 reexport requires Lustre client on the re-exporting nodes to be
594 mounted with "32bitapi" mount option
596 Severity : enhancement
597 Description: Update RHEL6.1 kernel to 2.6.32-131.6.1.el6
599 Severity : enhancement
600 Description: Update RHEL5.6 kernel to 2.6.18-238.19.1.el5
602 Severity : enhancement
603 Description: Update RHEL5.6 kernel to 2.6.18-238.9.1.el5
606 Frequency : only with SLES10
608 Description: Use OFED "KMP" provided by Novell
609 Details : SLES10 SP3 ships with OFED in a separate "KMP" package.
610 Lustre is now built against this package. That means you need to
611 install the ofed-kmp package from Novell for the patchless client.
612 Note that the ofed-kmp that Novell ships may not exactly match the
613 kernel version but should still be compatible.
615 Severity : enhancement
617 Description: Update RHEL5.5 kernel to 2.6.18-194.17.1.el5
618 Update OEL5.5 kernel to 2.6.18-194.17.1.0.1.el5.
620 --------------------------------------------------------------------------------
622 2010-07-15 Oracle, Inc.
624 * Server support for kernels:
625 2.6.18-164.11.1.el5 (RHEL 5)
626 2.6.18-164.11.1.0.1.el5 (OEL 5)
627 * Client support for unpatched kernels:
628 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
629 2.6.18-164.11.1.el5 (RHEL 5),
630 2.6.18-164.11.1.0.1.el5 (OEL 5)
631 2.6.16.60-0.42.8 (SLES 10),
633 2.6.29.4-167.fc11 (FC11)
634 * Recommended e2fsprogs version: 1.41.10-sun2
636 Severity : enhancement
638 Description: Distribute lfs_migrate script with lustre
639 Details : The lfs_migrate mentioned in the manual is now included in the
640 lustre tree in lustre/scripts and will distributed in lustre RPMs.
642 Severity : enhancement
644 Description: Failover nid list added to import proc file.
646 Severity : enhancement
648 Description: "som_preview" mount option required to use SOM
649 Details : The clients will need to use som_preview mount option
653 Description: Lustre does not do 1MB IOs to HW RAID
654 Details : Bump MAX_PHYS/HW_SEGMENTS and SG_ALL to 256 in the RHEL5 kernel.
656 Severity : enhancement
658 Description: lfs quota output is not very convenient for awk/sed-parsing
659 Details : Some positions in lfs quota output table could be empty or
660 non-empty which made it hard to parse it with scripts, now
661 a dash is put instead of space where there is not supposed
666 Description: Update to OFED 1.5.1
667 Details : OFED 1.5.1 will be used in environments that don't use
668 vendor supplied OFED.
672 Description: Remove lnet. prefix from lctl params display
673 Details : Also renamed lnet.memused to lnet_memused to
678 Description: Use conf_param -d to remove permanant settings.
682 Description: Add list_param -R
683 Details : lctl list_param now supports recursive mode.
686 Bugzilla : 21829, 22312
687 Description: Fix llobdstat.
688 Details : 1. Now updating the write_bytes counter
689 2. Port script fixes from 1.8.
693 Description: decrease the usage of memory on clients.
694 Details : 1. On clients, recycle dentries and inodes unused.
695 2. Delete the code related to ll_deathrow(att 6215 in bug 1443). It
700 Description: llobdstat fix and enhancement
701 Details : add a counter to set a limit to how many samples will be returned
702 fix a wildcard in the path to limit to obdfilter stats only
706 Description: lfs check only allowed for root
707 Details : Code cleanups and added a regresion test of non-root use of
712 Description: MDS Oopses in ptlrpc_dispatch_difficult_reply()/__list_add()
713 Details : Fix hr->hr_index overload.
717 Description: Fix FIEMAP functionality.
721 Description: FC11 patchless client support added
725 Description: Update RHEL5.4 kernel to 2.6.18-164.11.1.el5 and
726 OEL5.4 kernel to 2.6.18-164.11.1.0.1.el5.
730 Description: New acc-sm script and various test-framework enhancments.
734 Description: Previously landed patch was reverted.
738 Description: Update SLES10 kernel to 2.6.16.60-0.42.8 (patchless)
742 Description: Add ldiskfs maxdirsize mount option.
746 Description: Fix mkfs.lustre support for 16TB LUNs.
750 Description: Print mmp_check_interval
751 Details : print mmp_check_interval and make it possible to abort
752 mount operation in case it takes too long.
756 Description: Add cascading_rw to lustre mpi tests.
760 Description: Store SOM data in EA.
761 Details : Also several SOM related cleanups.
765 Description: Add refcount to md_open_data
769 Description: Use refcount on lov_request_set to prevent memory corruption.
773 Description: Statahead: use parent dir's i_mutex for synchronization.
777 Description: loadgen improvements.
781 Description: Enable peer health detection in ksocklnd and ko2iblnd.
785 Description: Control DCACHE_LUSTRE_INVALID flag with MDS_INODE_LOOKCUP lock
789 Description: Fix race on obd_devcie:md_stats
793 Description: Fix SLES 11 client builds.
794 Details : Resolved uninitalized pointer gcc warnings.
798 Description: Allow non-root access for "lfs check".
799 Details : Added a check in obd_class_ioctl() for OBD_IOC_PING_TARGET.
803 Description: Clear imp_force_reconnect correctly in ptlrpc_connect_interpret()
805 Severity : enhancement
807 Description: Update kernel to RHEL5.4 2.6.18-164.6.1.el5 and
808 OEL5 2.6.18-164.6.1.0.1.el5(Both in-kernel OFED enabled).
812 Description: Sanity test_27p fails.
813 Details : OBD_FAIL_OST_ENOSPC has been moved to filter_statfs().
814 Sanity tests 27 were cleaned up and improved.
818 Description: Unify req format on client/servers
819 Details : Use new req_capsule API [almost] everywhere instead of old PTLRPC
820 buffers and swabbers approach..
825 Description: drop join file stuff.
827 Severity : enhancement
829 Description: Add 2.6.27 and SLES11 (patchless client) support
833 Description: Fix in ptlrpc_expire_one_request() to print the signed time
836 Severity : enhancement
838 Description: Build kernels (RHEL5, OEL5 and SLES10/11) using the vendor's own
841 Severity : enhancement
843 Description: Remove set_info(KEY_UNLINKED) from MDS/OSC
845 Severity : enhancement
847 Description: correctly handle big reply message.
848 Details : send LNet event if reply is bigger than buffer and adjust this buffer
853 Description: Drop unnecessary __GFP_NOMEMALLOC flag from filter_get_page()
855 Severity : enhancement
857 Description: Add support for OEL5.
861 Description: Conf-sanity.sh 50g test - deactivated OST should not cause a panic.
863 Severity : enhancement
865 Description: "weak-modules" support
866 Details : Implement "weak-modules" support which enables kernel modules
867 to be used with any kernel that implements the same kABI. In
868 order to achieve this modules are now installed in
869 /lib/modules/$(uname -r)/updates/kernel on all distributions.
873 Description: client could not reconnect to OST because of an active request.
874 Details : The client now retries to reconnect to the same server, if a
875 connect request failed with EBUSY or -EAGAIN. Test case 35b
876 conf-sanity.sh has been created.
880 Description: Don't allow make backward step on assignin osc next id.
881 Details : race between allocation next id and ll_sync thread can be cause
882 of set wrong osc next id and can be kill valid ost objects.
886 Description: Changes in raid5-large-io-rhel5.patch to calculate sectors properly
890 Description: Increase the default BLK_DEF_MAX_SECTORS value for RHEL5 and SLES11
894 Description: Error handling in osc_statfs_interpret() has been improved.
895 Details : Check in osc_statfs_interpret() for EBADR.
899 Description: Do not send statfs() requests to OSTs disabled by administrator.
900 Details : Check in lov_prep_statfs_set() for non-NULL ltd_exp.
902 Severity : enhancement
904 Description: Added sanity-quota test to for lfs(1) to work with numeric uid/gid.
906 Severity : enhancement
908 Description: Update kernel to RHEL5 2.6.18-128.7.1.el5.
910 Severity : enhancement
912 Description: Enhancement for lfs(1) command to use numeric uid/gid.
916 Description: Improved cleanup procedure for conf-sanity test_46a.
920 Description: Do not start bulk transfer, if client tried to reconnect.
921 Details : Do not start bulk transfer, if client tried to reconnect,
922 but failed due to active rpcs. Check for exp_abort_active_req
923 has been implemented.
927 Description: Prevent inconsistences between linux and lustre mount structures.
928 Details : Wait indefinitely in server_wait_finished() until mnt_count drops.
929 Make the sleep interruptible.
931 Severity : enhancement
933 Description: provide server to client comms path
934 Details : server to client communications path via new LDLM_SET_INFO rpc
938 Description: truncate starts GFP_FS allocation under transaction causing deadlock
939 Details : ldiskfs_truncate calls grab_cache_page which may start page
940 allocation under an open transaction. This may lead to
941 calling prune_icache with consequent lustre reentrance.
945 Description: send correctly lsm on open replay
946 Details : MDS is trust to LSM size on replay open, but client can set wrong size
951 Description: enable client interface failover
952 Details : When a child reconnects from another NID, properly update export
953 nid hash position and ldlm reverse import.
955 Severity : enhancement
957 Description: Communicate OST degraded/readonly state via statfs to MDS
958 Details : Flags in the statfs returned from OSTs indicate whether the
959 OST is in a degraded RAID state, or if the filesystem has
960 turned read-only after a filesystem error is detected.
965 Description: don't panic if EPROTO was hit when reading symlink
966 Details : correctly handling request reference in error cases.
968 Severity : enhancement
970 Description: Add LustreNetLink, a kernel-userspace communcation path.
972 Severity : enhancement
974 Description: Update kernel to SLES10 SP2 2.6.16.60-0.39.3.
979 Description: access to llog context before init.
980 Details : move handling CATALOGS file at osc layer and forbid access to llog
984 Frequency : rare, only when client is writing to OST on local node
986 Description: Avoid deadlock for local client writes
987 Details : Use new OBD_BRW_MEMALLOC flag to notify OST about writes in the
988 memory freeing context. This allows OST threads to set the
989 PF_MEMALLOC flag on task structures in order to allocate memory
990 from reserved pools and complete IO.
991 Use GFP_HIGHUSER for OST allocations for non-local client writes,
992 so that the OST threads generate memory pressure and allow
993 inactive pages to be reclaimed.
995 Severity : enhancement
997 Description: Update kernel to RHEL5.3 2.6.18-128.1.14.el5.
1002 Description: lock ordering violation between &cli->cl_sem and _lprocfs_lock
1003 Details : move ldlm namespace creation in setup phase to avoid grab
1004 _lprocfs_lock with cli_sem held.
1008 Description: Temporarily disable grant shrink.
1009 Details : Disable the feature for debugging.
1012 Frequency : only when formatting test filesystems
1014 Description: Unable to run several mkfs.lustre on loop devices at the same time
1015 Details : mkfs.lustre returns error 256 on the concurrent loop devices
1016 formatting. The solution is to proper handle the error.
1018 Severity : enhancement
1020 Description: Update kernel to RHEL5.3 2.6.18-128.1.6.el5.
1022 Severity : enhancement
1024 Description: Update kernel to SLES10 SP2 2.6.16.60-0.37.
1029 Description: resolve race between obd_disconnect and class_disconnect_exports
1030 Details : if obd_disconnect will be called to already disconnected export he
1031 forget release one reference and osc module can't unloaded.
1033 Severity : enhancement
1035 Description: Allow tuning service thread via /proc
1036 Details : For each service a new
1037 /proc/fs/lustre/{service}/*/thread_{min,max,started} entry is
1038 created that can be used to set min/max thread counts, and get the
1039 current number of running threads.
1044 Descriptoin: don't return error if have partially created objects for file.
1045 Details : lov_update_create_set uses set->set_success as index for created
1046 objects, so if some requests will be failed, they will have hole
1047 at end of array and qos_shrink_lsm for allocate correct lsm.
1049 Severity : enhancement
1051 Description: Update OFED support to 1.4.1
1056 Description: Reduce small size read RPC
1057 Details : Set read-ahead limit for every file and only do read-ahead when
1058 available read-ahead pages are bigger than 1M to avoid small size
1061 Severity : enhancement
1063 Description: move AT tunable parameters for more consistent usage
1064 Details : add AT tunables under /proc/sys/lustre, add to conf_param parsing
1066 Severity : enhancement
1068 Description: add lazystatfs mount option to allow statfs(2) to skip down OSTs
1069 Details : allow skip disconnected ost for send statfs request and hide error
1075 Description: fix racy locking of mballoc block bitmaps causing BUG
1076 Details : The locking of the mballoc buddy bitmap and the in-memory
1077 block bitmap was using two different spin locks in some
1078 cases. This made it possible to incorrectly access the
1079 mballoc bitmap while another process was modifying it,
1080 causing a sanity assertion to fail. While no on-disk corruption
1081 was reported, there was some risk of this happening.
1084 Frequency : rare, on llog test 6
1086 Descriptoin: don't allow connect to already connected import
1087 Details : allowing connect to already connected import is hide connecting problem.
1092 Descriptoin: allow kill process which wait statahead result
1093 Details : for some reasons 'ls' can stick in waiting result from statahead,
1094 in this case need way for kill this process.
1096 Severity : enhancement
1098 Description: Add state history info file, enhance import info file
1099 Details : Track import connection state changes in a new osc/mdc proc file;
1100 add overview-type data to the osc/mdc import proc file.
1102 Severity : enhancement
1104 Description: MDS create should not wait for statfs RPC while holding DLM lock.
1106 Severity : enhancement
1108 Description: Update to RHEL5U3 kernel-2.6.18-128.1.1.el5.
1113 Descriptoin: OST grant too much space to client even there are not enough space.
1114 Details : Client will shrink its grant cache to OST if there are no write
1115 activity over 6 mins (GRANT_SHRINK_INTERVAL), and OST will retrieve
1116 this grant cache if there are already not enough avaible space
1117 (left_space < total_clients * 32M).
1120 Frequency : start MDS on uncleanly shutdowned MDS device
1122 Descriptoin: ll_sync thread stay in waiting mds<>ost recovery finished
1123 Details : stay in waiting mds<>ost recovery finished produce random bugs
1124 due race between two ll_sync thread for one lov target. send
1125 ACTIVATE event only if connect realy finished and import have
1129 Frequency : rare, connect and disconnect target at same time
1131 Descriptoin: ASSERTION(atomic_read(&imp->imp_inflight) == 0
1132 Details : don't call obd_disconnect under lov_lock. this long time
1133 operation and can block ptlrpcd which answer to connect request.
1138 Descriptoin: don't lose wakeup for imp_recovery_waitq
1139 Details : recover_import_no_retry or invalidate_import and import_close can
1140 both sleep on imp_recovery_waitq, but we was send only one wakeup
1144 Frequency : always with long access acl
1146 Descriptoin: mds can't pack reply with long acl.
1147 Details : mds don't control size of acl but they limited by reint/getattr
1150 Severity : enhancement
1152 Description: Update to SLES10 kernel-2.6.16.60-0.33.
1154 Severity : enhancement
1156 Description: Update to RHEL5 kernel-2.6.18-92.1.22.el5.
1159 Frequency : start MDS on uncleanly shutdowned MDS device
1161 Descriptoin: aborting recovery hang on MDS
1162 Details : don't throttle destroy RPCs for the MDT.
1165 Frequency : on remount
1167 Description: external journal device not working after the remount
1168 Details : clear dev_rdonly flag for external journal devices in
1174 Description: shutdown vs evict race
1175 Details : client_disconnect_export vs connect request race.
1176 if client will evicted at this time - we start invalidate
1177 thread without referece to import and import can be freed
1181 Frequency : rare, need acl's on inode.
1183 Description: client can't handle ost additional correctly
1184 Details : if ost was added after client connected to mds client can have
1185 hit lnet_try_match_md ... to big messages to wide striped files.
1186 in this case need teach client to handle config events about add
1187 lov target and update client max ea size at that event.
1189 Severity : enhancement
1191 Description: Changelogs
1192 Details : Changelogs are a lightweight mechanism to track filesystem
1193 metadata and namespace changes. The changelog is recorded
1194 permanently on the MDTs, and is periodically "consumed" / purged
1195 when records are no longer needed.
1197 Severity : enhancement
1199 Description: compact fld format with extents
1200 Details : Store range of seq rather than every seq in FLD. Seq
1201 controller update FLD rather than clients. In Case of CMD, mdt0
1202 has FLD, all other metadata server act as non persistent proxy
1203 for FLD queries and cache fld entries in fld cache.
1208 Description: don't skip ost target if they assigned to file
1209 Details : Drop slow OSCs if we can, but not for requested start idx.
1210 This means "if OSC is slow and it is not the requested
1211 start OST, then it can be skipped, otherwise skip it only
1212 if it is inactive/recovering/out-of-space.
1216 Description: more cleanup in mds_lov
1217 Details : not send LOV EA under replay, we can't know about they size at this
1218 time. Don't allow client connect to mds before any ost connected,
1219 for avoid problems with LOV EA size and returning EIO to client.
1221 Severity : enhancement
1223 Description: Interoperability at server side (Disk interoperability)
1225 Severity : enhancement
1227 Description: Update to RHEL5 kernel-2.6.18-92.1.17.el5.
1229 Severity : enhancement
1231 Description: Update to SLES10 SP2 kernel-2.6.16.60-0.31.
1233 Severity : enhancement
1235 Description: New client IO stack (CLIO).
1237 Severity : enhancement
1239 Description: Commit on sharing. Eliminate inter-client dependencies between
1240 uncommitted transactions by doing transaction commits.
1241 Thereby clients may recovery independently.
1244 Frequency : Create a symlink file with a very long name
1246 Description: ldlm_cancel_pack()) ASSERTION(max >= dlm->lock_count + count)
1247 Details : If there is no extra space in the request for early cancels,
1248 ldlm_req_handles_avail() returns 0 instead of a negative value.
1250 Severity : enhancement
1252 Description: Add /proc entry for import status
1253 Details : The mdc, osc, and mgc import directories now have
1254 an import directory that contains useful import data for debugging
1255 connection problems.
1257 Severity : enhancement
1259 Description: Re-disable certain /proc logging
1260 Details : Enable and disable client's offset_stats, extents_stats and
1261 extents_stats_per_process stats logging on the fly.
1264 Frequency : Only on FC kernels 2.6.22+
1266 Description: oops in statahead
1267 Details : Do not drop reference count for the dentry from VFS when lookup,
1268 VFS will do that by itself.
1270 Severity : enhancement
1272 Description: Generic /proc file permissions
1273 Details : Set /Proc file permissions in a more generic way to enable non-
1274 root users operate on some /proc files.
1278 Description: Hitting mdc_commit_close() ASSERTION
1279 Details : Properly handle request reference release in
1280 ll_release_openhandle().
1284 Description: quota recovery deadlock during mds failover
1285 Details : This patch includes att18982, att18236, att18237 in bz14840.
1287 1. fix osts hang when mds does failover with quotaon
1288 2. prevent watchdog storm when osts threads wait for the
1293 Frequency : only patchless client
1294 Description: add workaround for race between add/remove dentry from hash
1296 Severity : enhancement
1298 Description: Allow OST glimpses to return PW locks
1302 Description: LBUG when llog conf file is full
1303 Details : When llog bitmap is full, ENOSPC should be returned for plain log.
1307 Description: Prevent import from entering FULL state when server in recovery
1311 Description: service mount cannot take device name with ":"
1312 Details : Only when device name contains ":/" will mount treat it as
1318 Description: replace ptlrpcd with the statahead thread to interpret the async
1319 statahead RPC callback
1323 Frequency : on recovery
1324 Description: I/O failures after umount during fail back
1325 Details : if client reconnected to restarted server we need join to recovery
1326 instead of find server handler is changed and process self
1327 eviction with cancel all locks.
1329 Severity : enhancement
1331 Description: Update to RHEL5 kernel-2.6.18-92.1.10.el5.
1333 Severity : enhancement
1335 Description: Update to SLES10 SP2 kernel-2.6.16.60-0.27.
1337 Severity : enhancement
1339 Description: Upcall on Lustre log has been dumped
1340 Details : Allow for a user mode script to be called once a Lustre log has
1341 been dumped. It passes the filename of the dumped log to the
1342 script, the location of the script can be specified via
1343 /proc/sys/lnet/debug_log_upcall.
1348 Description: avoid idr_remove called for id which is not allocated.
1349 Details : Move assigment s_dev for clustered nfs to end of initialization,
1350 to avoid problem with error handling.
1355 Description: avoid Already found the key in hash [CONN_UNUSED_HASH] messages
1356 Details : When connection is reused this not moved from CONN_UNUSED_HASH
1357 into CONN_USED_HASH and this prodice warning when put connection
1358 again in unused hash.
1361 Severity : enhancement
1363 Description: File striping can now be set to use an arbitrary pool of OSTs.
1365 Severity : enhancement
1367 Description: Export bytes_read/bytes_write count on OSC/OST.
1371 Description: Early reply size mismatch, MGC loses connection
1372 Details : Apply the MGS_CONNECT_SUPPORTED mask at reconnect time so
1373 the connect flags are properly negotiated.
1378 Description: quotas are not honored with O_DIRECT
1379 Details : all writes with the flag O_DIRECT will use grants which leads to
1380 this problem. Now using OBD_BRW_SYNC to guard this.
1384 Description: add quota statistics
1385 Details : 1. sort out quota proc entries and proc code.
1386 2. add quota statistics
1388 Severity : enhancement
1390 Description: enable quota support for HEAD.
1394 Description: Properly propagate oinfo flags from lov to osc for statfs
1395 Details : restore missing copy oi_flags to lov requests.
1397 Severity : enhancement
1399 Description: Add man pages for llobdstat(8), llstat(8), plot-llstat(8),
1400 : l_getgroups(8), lst(8), routerstat(8)
1401 Details : included man pages for llobdstat(8), llstat(8),
1402 : plot-llstat(8), l_getgroups(8), lst(8), routerstat(8)
1404 Severity : enhancement
1406 Description: configure's --enable-quota should check the
1407 : kernel .config for CONFIG_QUOTA
1408 Details : configure is terminated if --enable-quota is passed but
1409 : no quota support is in kernel
1413 Description: Remove portals compatibility
1414 Details : Remove portals compatibility, not interoperable with releases
1419 Description: Resolve device initialization race
1420 Details : Prevent proc handler from accessing devices added to the
1421 obd_devs array but yet be intialized.
1423 Severity : enhancement
1425 Description: Update to SLES10 SP2 kernel-2.6.16.60-0.23.
1427 Severity : enhancement
1429 Description: Update to RHEL5 kernel-2.6.18-92.1.6.el5.
1434 Description: Using wrong pointer in osc_brw_prep_request
1435 Details : Access to array[-1] can produce panic if kernel compiled with
1436 CONFIG_PAGE_ALLOC enabled
1440 Description: Client runs out of low memory
1441 Details : Consider only lowmem when counting initial number of llap pages
1445 Description: *optional* service tags registration
1446 Details : if the "service tags" package is installed on a Lustre node
1447 When the filesystem is mounted, a local-node service tag will
1448 be created. See http://inventory.sun.com/ for more information
1449 about the Service Tags asset management system.
1453 Description: Kernel BUG tries to release flock
1454 Details : Lustre does not destroy flock lock before last reference goes
1455 away. So always drop flock locks when client is evicted and
1456 perform unlock regardless of successfulness of speaking to MDS.
1460 Description: add refcount protection for osc callbacks, avoid panic on shutdown
1464 Description: sanity test 65a fails if stripecount of -1 is set
1465 Details : handle -1 striping on filesystem in ll_dirstripe_verify
1470 Description: ASSERTION(CheckWriteback(page,cmd)) failed
1471 Details : badly clear PG_Writeback bit in ll_ap_completion can produce false
1474 Severity : enhancement
1476 Description: Update to RHEL5 kernel-2.6.18-53.1.21.el5.
1480 Description: do not process already freed flock
1481 Details : flock can possibly be freed by another thread before it reaches
1482 to ldlm_flock_completion_ast.
1486 Description: LBUG during stress test
1487 Details : Need properly lock accesses the flock deadlock detection list.
1491 Description: oops in page fault handler
1492 Details : kernel page fault handler can return two special 'pages' in error
1493 case, don't try dereference NOPAGE_SIGBUS and NOPAGE_OMM.
1497 Description: timeout with invalidate import.
1498 Details : ptlrpcd_check call obd_zombie_impexp_cull and wait request which
1499 should be handled by ptlrpcd. This produce long age waiting and
1500 -ETIMEOUT ptlrpc_invalidate_import and as result LASSERT.
1502 Severity : enhancement
1504 Description: Update to RHEL5 kernel-2.6.18-53.1.19.el5.
1508 Description: enable MGS and MDT services start separately
1509 Details : add a 'nomgs' option in mount.lustre to enable start a MDT with
1510 a co-located MGS without starting the MGS, which is a complement
1511 to 'nosvc' mount option.
1515 Frequency : after recovery
1516 Description: precreate to many object's after del orphan.
1517 Details : del orphan st in oscc last_id == next_id and this triger growing
1518 count of precreated objects. Set flag LOW to skip increase count
1519 of precreated objects.
1523 Frequency : rare, on clear nid stats
1524 Description: ASSERTION(client_stat->nid_exp_ref_count == 0)
1525 Details : when clean nid stats sometimes try destroy live entry,
1526 and this produce panic in free.
1530 Description: Stack overflow during MDS log replay
1531 ease stack pressure by using a thread dealing llog_process.
1535 Description: wait until IO finished before start new when do lock cancel.
1536 Details : VM protocol want old IO finished before start new, in this case
1537 need wait until PG_writeback is cleared until check dirty flag and
1538 call writepages in lock cancel callback.
1540 Severity : enhancement
1542 Description: using special macro for print time and cleanup in includes.
1546 Description: mds_mfd_close() ASSERTION(rc == 0)
1547 Details : In mds_mfd_close(), we need protect inode's writecount change
1548 within its orphan write semaphore to prevent possible races.
1552 Description: Obsolete CURRENT_SECONDS and use cfs_time_current_sec() instead.
1556 Frequency : rare, on shutdown ost
1557 Description: don't hit live lock with umount ost.
1558 Details : shrink_dcache_parent can be in long loop with destroy dentries,
1559 use shrink_dcache_sb instead.
1563 Description: don't panic with use echo client
1564 Details : echo client pass NULL as client nid pointer and this produce null
1565 pointer dereference.
1569 Description: fix build on ppc32
1570 Details : compile code with -m64 flag produce wrong object file for ppc32.
1574 Description: add message levels for liblustreapi
1578 Description: fix for occasional failure case of -ENOSPC in recovery-small tests
1579 Details : Move the 'good_osts' check before the 'total_bavail' check. This
1580 will result in an -EAGAIN and in the exit call path we call
1581 alloc_rr() which will with increasing aggressiveness attempt to
1582 aquire precreated objects on the minimum number of required OSCs.
1586 Description: Use old size assignment to avoid deadlock
1587 Details : Reverts the changes in bugs 2369 and bug 14138 that introduced
1588 the scheduling while holding a spinlock. We do not need locking
1589 for size in ll_update_inode() because size is only updated from
1590 the MDS for directories or files without objects, so there is no
1591 other place to do the update, and concurrent access to such inodes
1592 are protected by the inode lock.
1596 Description: resolve "_IOWR redefined" build error on SLES10
1600 Description: dump the memory debugging after all modules are unloaded to
1601 suppress false negative in conf_sanity test 39
1603 Severity : enhancement
1605 Description: build kernel-ib packages for OFED 1.3 in our release cycle
1610 Description: fix SLES kernel versioning
1611 Details : the kernel version for our SLES 10 kernel did not include a "-"
1612 before the "smp" at the end. while this was not a problem in
1613 general, it did mean that software trying to use the kernel
1614 version to try to detect a vendor specific kernel would fail.
1615 this was most evident by the OFED build scripts.
1619 Description: Don't update lov_desc members until making sure they are valid
1620 Details : When updating lov_desc members via proc fs, need fix their
1621 validities before doing the real update.
1625 Description: don't put request into delay list while invalidate in flight.
1626 Details : ptlrpc_delay_request sometimes put in delay list while invalidate
1627 import in flight. this produce timeout for invalidate and sometimes
1628 can cause stale data.
1632 Frequency : on ppc only
1633 Description: not convert ost objects for directory because it's not exist.
1634 Details : ll_dir_getstripe assume dirrectory has ost objects but this wrong.
1638 Description: Add FMODE_EXEC file flag for SLES10 SP1 kernel.
1640 Severity : enhancement
1642 Description: Update to support 2.6.22.14 vanilla kernel.
1646 Frequency : rare, on recovery
1647 Description: read procfs can produce deadlock in some situation
1648 Details : Holding lprocfs lock which send rpc can produce block for destroy
1649 obd objects and this also block reconnect with -EALREADY. This
1650 isn't fix all lprocfs bugs - but make it rare.
1652 Severity : enhancement
1654 Description: Update kernel to RHEL5 2.6.18-53.1.14.el5.
1657 Frequency : frequent on X2 node
1659 Description: mdc_set_open_replay_data LBUG
1660 Details : Set replay data for requests that are eligible for replay.
1664 Description: lustre_mgs: operation 101 on unconnected MGS
1665 Details : When MGC is disconnected from MGS long enough, MGS will evict the
1666 MGC, and late on MGC cannot successfully connect to MGS and a lot
1667 of the error messages complaining that MGS is not connected.
1671 Frequency : on network error
1672 Description: panic with double free request if network error
1673 Details : mdc_finish_enqueue is finish request if any network error ocuring,
1674 but it's true only for synchronus enqueue, for async enqueue
1675 (via ptlrpcd) this incorrect and ptlrpcd want finish request
1678 Severity : enhancement
1680 Description: client-side metadata stat-ahead during readdir(directory readahead)
1681 Details : perform client-side metadata stat-ahead when the client detects
1682 readdir and sequential stat of dir entries therein
1685 Frequency : on start mds
1687 Description: Implement get_info(last_id) in obdfilter.
1690 Frequency : occasional
1692 Description: Correctly check stale fid, not start epoch if ost not support SOM
1693 Details : open with flag O_CREATE need set old fid in op_fid3 because
1694 op_fid2 was overwritten with new generated fid, but mds can answer
1695 with one of these two fids and both is not stale. Setattr
1696 incorrectly started an epoch and assume will be called
1697 done_writing, but without SOM done_writing ever being called.
1700 Frequency : rare, depends on device drivers and load
1702 Description: MDS or OSS nodes crash due to stack overflow
1703 Details : Code changes in 1.8.0 increased the stack usage of some functions.
1704 In some cases, in conjunction with device drivers that use a lot
1705 of stack the MDS (or possibly OSS) service threads could overflow
1706 the stack. One change which was identified to consume additional
1707 stack has been reworked to avoid the extra stack usage.
1710 Frequency : occasional
1712 Description: Do not fail import if osc_interpret_create gets -EAGAIN
1713 Details : If osc_interpret_create got -EAGAIN it immediately exits and
1714 wakeup oscc_waitq. After wakeup oscc_wait_for_objects call
1715 oscc_has_objects and see OSC has no objests and call
1716 oscc_internal_create to resend create request.
1718 Severity : enhancement
1720 Description: Update to SLES10 SP1 latest kernel-2.6.16.54-0.2.5.
1722 Severity : enhancement
1724 Description: Update to RHEL5 latest kernel-2.6.18-53.1.13.el5.
1727 Frequency : very rare
1729 Description: Fix replay if there is an un-replied request and open
1730 Details : In some cases, older replay request will revert the
1731 mcd->mcd_last_xid on MDS which is used to record the client's
1732 latest sent request.
1734 Severity : enhancement
1736 Description: Update to RHEL5 latest kernel-2.6.18-53.1.6.el5.
1738 Severity : enhancement
1740 Description: Add rhel5 support to HEAD.
1742 Serverity : enhancement
1744 Description: Update RHEL4 kernel to 2.6.9-67.0.4.
1749 Description: Don't allow skipping OSTs if index has been specified.
1750 Details : Don't allow skipping OSTs if index has been specified, make
1751 locking in internal create lots better.
1755 Description: LBUG in ptlrpc_check_set() bad phase ebc0de00
1756 Details : access to bitfield in structure is always rounded to long
1757 and this produce problem with not atomic change any bit.
1761 Description: Lustre make rpms failed.
1762 Details : Remove ldiskfs spec file to avoids rpmbuild be confused when
1763 builds Lustre rpms from tarball.
1766 Frequency : rare on shutdown ost
1768 Description: If llog cancel was not send before clean_exports phase, this can
1769 produce deadlock in llog code.
1770 Details : If llog thread has last reference to obd and call class_import_put
1771 this produce deadlock because llog_cleanup_commit_master wait when
1772 last llog_commit_thread exited, but this never success because was
1773 called from llog_commit_thread.
1777 Description: allow userland application know is lost one of stripes.
1778 Details : fill lvb_blocks with error code on ost and return it to
1779 application if error flag found.
1783 Description: NULL lov_tgts causing MDS oops
1784 Details : more safe checks for NULL lov_tgts for avoid oops.
1786 Severity : enhancement
1788 Description: Update to RHEL4 latest kernel-2.6.9-67.0.1.EL.
1792 Descriptoin: make lov_create() will not stuck in obd_statfs_rqset()
1793 Details : If an OST is down the MDS will hang indefinitely in
1794 obd_statfs_rqset() waiting for the statfs data. While for
1795 MDS QOS usage of statfs, it should not stuck in waiting.
1797 Severity : enhancement
1799 Description: remote_acl support
1800 Details : Support ACL-based permission check for remote user.
1801 Support setfacl/getfacl for remote user with the utils
1802 "lfs {l,r}{s,g}etfacl" which follow the same parameter format as
1803 the system "{s,g}etfacl" utils.
1805 Severity : enhancement
1807 Description: Update to RHEL4 U6 kernel-2.6.9-67.EL.
1809 Severity : enhancement
1811 Description: Update to RHEL5 latest kernel-2.6.18-53.1.4.el5.
1815 Description: make mgs_setparam() handle fsname containing dash
1816 Details : fsname containing a dash does not work with lctl conf_param
1818 Severity : enhancement
1820 Description: Update to SLES10 SP1 latest kernel-2.6.16.54-0.2.3.
1822 Severity : enhancement
1824 Description: Update to RHEL5 Update-1 kernel 2.6.18-53.el5.
1828 Frequency : rare, at shutdown
1829 Description: access already free / zero obd_namespace.
1830 Details : if client_disconnect_export was called without force flag set,
1831 and exist connect request in flight, this can produce access to
1832 NULL pointer (or already free pointer) when connect_interpret
1833 store ocd flags in obd_namespace.
1837 Frequency : only at startup
1838 Description: not alloc memory with spinlock held.
1839 Details : allocation memory with GFP_KERNEL can produce sleep deadlock,
1840 if any spinlock held.
1842 Severity : enhancement
1844 Description: make lustre randomly fail allocating memory
1845 Details : Make lustre randomly failed allocating memory for testing purpose.
1847 Severity : enhancement
1849 Description: lost problems with lov objid file
1850 Details : Fixes some scability and access to not inited memory problems
1851 in work with lov objdid file.
1856 Description: lfs find does not continue on file error
1857 Details : Continue other files processing when a file/dir is absent.
1861 Description: Inconsistent usage of lustre_pack_reply()
1862 Details : Standardize the usage of lustre_pack_reply() such that it
1863 always generate a CERROR on failure.
1866 Frequency : occasional
1868 Description: MDS hang or stay in waiting lock
1869 Details : If client receive lock with CBPENDING flag ldlm need send lock
1870 cancel as separate rpc, to avoid situation when cancel request
1871 can't processed due all i/o threads stay in wait lock.
1875 Description: Update to RHEL5 kernel 2.6.18-8.1.15.el5.
1879 Description: Update to SLES10 SP1 kernel 2.6.16.53-0.16
1883 Description: Update to SLES9 kernel-2.6.5-7.287.3.
1887 Description: Update to RHEL4 latest kernel.
1889 Severity : enhancement
1891 Description: Build SLES10 patchless client fails
1892 Details : The configure was broken by run ./configure with
1893 --with-linux-obj=.... argument for patchless client. When the
1894 configure use --with-linux-obj, the LINUXINCLUDE= -Iinclude
1895 can't search header adequately. Use absolute path such as
1896 -I($LINUX)/include instead.
1900 Description: interrupt oig_wait produce painc on resend.
1901 Details : brw_redo_request can be used for resend requests from ptlrpcd and
1902 private set, and this produce situation when rq_ptlrpcd_data not
1903 copyed to new allocated request and triggered LBUG on assert
1904 req->rq_ptlrpcd_data != NULL. But this member used only for wakeup
1905 ptlrpcd set if request is changed and can be safety changed to use
1910 Description: LASSERT_{REQ,REP}SWAB macros are buggy
1911 Details : If SWAB_PARANOIA is disabled, the LASSERT_REQSWAB and
1912 LASSERT_REPSWAB macros become no-ops, which is incorrect. Drop
1913 these macros and replace them with their difinitions instead.
1917 Description: Update kernel patches for SLES10 2.6.16.53-0.8.
1918 Details : Update which_patch & target file for SLES10 latest kernel.
1921 Description: Remove client patches from SLES 10 kernel.
1922 Details : This causes SLES 10 clients to behave as patchless clients
1923 even on a Lustre-patched (server) kernel.
1925 Severity : enhancement
1927 Description: self-adjustable client's lru lists
1928 Details : use adaptive algorithm for managing client cached locks lru
1929 lists according to current server load, other client's work
1930 pattern, memory activities, etc. Both, server and client
1931 side namespaces provide number of proc tunables for controlling
1934 Severity : enhancement
1936 Description: light-weight GSS support
1937 Details : Support krb5n and krb5a mode, which keep Kerberos 5 authentication
1938 and reduce performance overhead.
1940 Severity : enhancement
1942 Description: Linux keyring support
1943 Details : Support using service of Linux keyring for Lustre GSS internal
1944 context refresh/cache mechanism.
1948 Description: Fix errors in lfs documentation
1949 Details : Fixes man pages
1953 Description: don't use GFP_* in generic Lustre code.
1954 Details : Use cfs_alloc_* functions and CFS_* flags for code portability.
1958 Description: obdclass is limited by single OBD_ALLOC(idarray)
1959 Details : replace OBD_ALLOC/OBD_FREE with OBD_VMALLOC/OBD_VFREE
1963 Description: warnings with build patchless client with vanila 2.6.19 and up
1964 Details : change old ctl_table style and replace ctl_table/ctl_table_header
1965 with cfs_sysctl_table_t/cfs_sysctl_table_header_t
1969 Frequency : Only for SLES
1970 Description: sanity_quota fail test_1
1971 Details : There are multiple occurences of $TSTUSR in SLES's /etc/group
1972 file, which makes TSTID[2] inunique.
1976 Frequency : Only for SLES9
1977 Description: Kernel patches for SLES9 2.6.5-7.286 kernel
1978 Details : Update target/ChangeLog/which_patch .
1982 Frequency : Only for test_10 in sanity-quota.sh
1983 Description: a bug in quota test code
1984 Details : checking a test flag in a wrong place
1988 Frequency : Only for test_13 in sanity-quota.sh
1989 Description: a bug in quota test script
1990 Details : shouldn't check overfull allocated quota
1994 Description: jbd statistics
1995 Details : Port older jbd statistics patch for sles10
1997 Severity : enhancement
1999 Description: Add printing inode info into message about error in writepage.
2004 Description: reply_lock_interpret crash due to race with it and lock cancel.
2005 Details : Do not replay locks that are being cancelled. Do not reference
2006 locks by their address during replay, just by their handle.
2010 Frequency : When flocks are used.
2011 Description: assertion failure in ldlm_cli_enquque_fini for non NULL lock.
2012 Details : Flock locks might destroy just granted lock if it could be merged
2013 with another existing flock, this is done in completion handler,
2014 so teach ldlm_cli_enquque_fini that this is a valid case for
2020 Description: Oops in read and write path when failing to allocate lock.
2021 Details : Check if lock allocation failed and return error back.
2025 Description: lstripe command fails for valid OST index
2026 Details : The stripe offset is compared to lov->desc.ld_tgt_count
2027 instead of lov->desc.ld_active_tgt_count.
2031 Description: sanity.sh failed test 103
2032 Details : RHEL mis-interpret setfacl "-X" param, so we won't test setfacl
2037 Description: df doesn't work properly if diskfs blocksize != 4K
2038 Details : Choose biggest blocksize of OST's as the LOV's blocksize.
2042 Description: lfs find on -1 stripe looping in lsm_lmm_verify_common()
2043 Details : Avoid lov_verify_lmm_common() on directory with -1 stripe count.
2045 Severity : enhancement
2047 Description: Adaptive timeouts
2048 Details : RPC timeouts adapt to changing server load and network
2049 conditions to reduce resend attempts and improve recovery time.
2053 Description: llapi_file_create() does not allow some changes
2054 Details : add llapi_file_open() that allows specifying the mode and
2055 open flags, and also returns an open file handle.
2059 Description: merge and cleanup kernel patches.
2060 Details : Remove mnt_lustre_list in vfs_intent-2.6-rhel4.patch.
2064 Description: Add journal checksum support.(Kernel part)
2065 Details : The journal checksum feature adds two new flags i.e
2066 JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT and
2067 JBD2_FEATURE_COMPAT_CHECKSUM. JBD2_FEATURE_CHECKSUM flag
2068 indicates that the commit block contains the checksum for
2069 the blocks described by the descriptor blocks. Now commit
2070 record can be sent to disk without waiting for descriptor
2071 blocks to be written to disk. This behavior is controlled
2072 using JBD2_FEATURE_ASYNC_COMMIT flag.
2076 Description: OSS needs mutliple precreate threads
2077 Details : Add ability to start more than one create thread per OSS.
2081 Description: Sanity.sh test_65k failed: setstripe should have succeeded
2082 Details : execute lfs setstripe on client
2086 Description: mds_obd_create error creating tmp object
2087 Details : When the user sets quota on root, llog will be affected and can't
2088 create files and write files.
2091 Frequency : Always on ia64 patchless client, and possibly others.
2093 Description: Add EXPORT_SYMBOL check for node_to_cpumask symbol.
2094 Details : This allows the patchless client to be loaded on architectures
2095 without this export.
2099 Description: RedHat Update kernel for RHEL5
2100 Details : Add the kernel config file for RHEL5.
2104 Description: RedHat Update kernel for RHEL5
2105 Details : Modify the kernel config file more closer RHEL5.
2109 Description: Build failure against Centos5 (RHEL5)
2110 Details : Define PAGE_SIZE when it isn't present.
2114 Description: test_7 on sanity-quota.sh can't be used on separate machines
2115 Details : add facet to handle it
2119 Description: "ll_intent_file_open()) lock enqueue: err: -13" with nfs
2120 Details : with NFS, the anon dentry's parent was set to itself in
2121 d_alloc_anon(), so in MDS, we use rec->ur_fid1 to find the
2122 corresponding dentry other than use rec->ur_name.
2124 Severity : enhancement
2126 Description: lfs setstripe enhancement
2127 Details : Make lfs setstripe understand 'k', 'm' and 'g' for stripe size.
2131 Description: enable data checksumming by default
2132 Details : enable checksum by default, allow --disable-checksum
2133 configure option and "-o nochecksum" mount option. Checksums
2134 can also be disabled at runtime via $LPROC/osc/*/checksum_pages.
2138 Description: lustre support for RHEL5
2139 Details : Add support for RHEL5.
2143 Description: Client eviction due to failover config
2144 Details : after a connection loss, the lustre client should attempt to
2145 reconnect to the last active server first before trying the
2146 other potential connections.
2150 Description: when mds and osts use different quota unit(32bit and 64bit),
2151 quota will be released repeatly.
2152 Details : void sending multiple quota reqs to mds, which will keep the
2153 status between the reqs.
2157 Description: osts not allocated evenly to files
2158 Details : change the condition to increase offset_idx
2161 Frequency : Always for filesystems larger than 2TB on 32-bit systems.
2163 Description: Data corruption for OSTs that are formatted larger than 2TB
2165 Details : When generating the bio request for lustre file writes the
2166 sector number would overflow a temporary variable before being
2167 used for the IO. The data reads correctly from Lustre (which
2168 will overflow in a similar manner) but other file data or
2169 filesystem metadata may be corrupted in some cases.
2173 Description: Tune the kernel for good SCSI performance.
2174 Details : Set the value of /sys/block/{dev}/queue/max_sectors_kb
2175 to the value of /sys/block/{dev}/queue/max_hw_sectors_kb
2180 Description: rewrite ext2-derived code in obdclass/uuid.c
2181 Details : rewrite inherited code (uuid parsing code from ext2 utils)
2182 from scratch preserving functionality.
2186 Description: "lfs find -obd UUID" prints directories
2187 Details : "lfs find -obd UUID" will return all directory names instead
2188 of just file names. It is incorrect because the directories
2189 do not reside on the OSTs.
2193 Description: use i_size_read and i_size_write in 2.6 port
2194 Details : replace inode->i_size access with i_size_read/write()
2197 Frequency : when removing large files
2199 Description: scheduling issue during removal of large Lustre files
2200 Details : Don't take the BKL in fsfilt_ext3_setattr() for 2.6 kernels.
2201 It causes scheduling issues when removing large files (17TB in the
2205 Frequency : only with liblustre clients on XT3
2207 Description: evictions taking too long
2208 Details : allow llrd to evict clients directly on OSTs
2211 Frequency : only on ppc
2213 Description: /proc/fs/lustre/devices broken on ppc
2214 Details : The patch as applied to 1.6.2 doesn't look correct for all arches.
2215 We should make sure the type of 'index' is loff_t and then cast
2216 explicitly as needed below. Do not assign an explicitly cast
2220 Frequency : only for rhel5
2222 Description: Kernel patches update for RHEL5 2.6.18-8.1.10.el5.
2223 Details : Modify the target file & which_kernel.
2225 Severity : enhancement
2227 Description: omit set fsid for export NFS
2228 Details : fix set/restore device id for avoid EMFILE error and mark lustre
2229 fs as FS_REQUIRES_DEV for avoid problems with generate fsid.
2233 Frequency : Always, for kernels after 2.6.16
2234 Description: Fix warning idr_remove called for id=.. which is not allocated.
2235 Details : Last kernels save old s_dev before kill super and not allow
2236 to restore from callback - restore it before call kill_anon_super.
2240 Description: buffer overruns could theoretically occur
2241 Details : llapi_semantic_traverse() modifies the "path" argument by
2242 appending values to the end of the origin string, and a
2243 overrun may occur. Adding buffer overrun check in liblustreapi.
2247 Description: Fix error on 'ls .' at the top of the Lustre mount.
2248 Details : Don't revalidate dentry if it is a root dentry.
2252 Description: Kernel patches update for RHEL4 2.6.9-55.0.6.
2253 Details : Modify vm-tunables-rhel4.patch.
2257 Description: Kernel config for 2.6.18-vanilla.
2258 Details : Modify targets/2.6-vanilla.target.in.
2259 Add config file kernel-2.6.18-2.6-vanilla-i686.config.
2260 Add config file kernel-2.6.18-2.6-vanilla-i686-smp.config.
2261 Add config file kernel-2.6.18-2.6-vanilla-x86_64.config.
2262 Add config file kernel-2.6.18-2.6-vanilla-x86_64-smp.config.
2266 Description: improve handling recoverable errors
2267 Details : if request processig with error which can be recoverable on server
2268 request should be resend, otherwise page released from cache and
2273 Description: Kernel patches update for RHEL5 2.6.18-8.1.14.el5.
2274 Details : Modify target file & which_patch.
2275 A flaw was found in the IA32 system call emulation provided
2276 on AMD64 and Intel 64 platforms. An improperly validated 64-bit
2277 value could be stored in the %RAX register, which could trigger an
2278 out-of-bounds system call table access. An untrusted local user
2279 could exploit this flaw to run code in the kernel
2280 (ie a root privilege escalation). (CVE-2007-4573).
2284 Description: O_DIRECT bypasses client statistics.
2285 Details : When running with O_DIRECT I/O, neither the client rpc_stats nor
2286 read_ahead_stats were updated. copied the stats section from
2287 osc_send_oap_rpc() into async_internal().
2291 Description: Add jbd statistics patch for RHEL5 and 2.6.18-vanilla
2295 Description: change order of libsysio includes
2296 Details : '#include sysio.h' should always come before '#include xtio.h'
2300 Description: handle "serious error: objid * already exists" more gracefully
2301 Details : If LAST_ID value on disk is smaller than the objects existing in
2302 the O/0/d* directories, it indicates disk corruption and causes an
2303 LBUG(). If the object is 0-length, then we should use the existing
2304 object. This will help to avoid a full fsck in most cases.
2306 Severity : enhancement
2308 Description: adapt the lustre_config script to support the upgrade case
2309 Details : Add "-u" option for lustre_config script to support upgrading 1.4
2310 server targets to 1.6 in parallel.
2314 Description: To avoid grant space > avaible space when the disk is almost
2315 full. Without this patch you might see the error "grant XXXX >
2316 available" or some LBUG about grant, when the disk is almost
2318 Details : In filter_check_grant, for non_grant cache write, we should
2319 check the left space by if (*left > ungranted + bytes), instead
2320 of (*left > ungranted), because only we are sure the left space
2321 is enough for another "bytes", then the ungrant space should be
2322 increase. In client, we should update cl_avail_grant only there
2323 is OBD_MD_FLGRANT in the reply.
2327 Description: Update RHEL 4 kernel to fix local root privilege escalation.
2328 Details : Update to the latest RHEL 4 kernel to fix the vulnerability
2329 described in CVE-2007-4573. This problem could allow untrusted
2330 local users to gain root access.
2333 Frequency : when using O_DIRECT and quotas
2335 Description: Incorrect file ownership on O_DIRECT output files
2336 Details : block usage reported by 'lfs quota' does not take into account
2337 files that have been written with O_DIRECT.
2342 Description: touch file failed when fs is not full
2343 Details : OST in recovery should not be discarded by MDS in alloc_qos(),
2344 otherwise we can get ENOSP while fs is not full.
2348 Description: parallel lock callbacks
2349 Details : Instead of sending blocking and completion callbacks as separated
2350 requests, adding them to a set and sending in parallel.
2353 Frequency : only for Cray XT3
2354 Bugzilla : 12829/13455
2355 Description: Changing primary group doesn't change the group lustre assigns to
2357 Details : When CRAY_XT3 is defined, the fsgid supplied by the client is
2358 overridden with the primary group provided by the group upcall,
2359 whereas the supplied fsgid can be trusted if it is in the list of
2360 supplementary groups returned by the group upcall.
2362 Severity : enhancement
2364 Description: Allow masking D_WARNING, D_ERROR messages from console
2365 Details : Console messages can now be disabled via lnet.printk.
2369 Description: User code with malformed file open parameter crashes client node
2370 Details : Before packing join_file req, all the related reference should be
2371 checked carefully in case some malformed flags cause fake
2372 join_file req on client.
2376 Description: LDLM_ENQUEUE races with LDLM_CP_CALLBACK
2377 Details : ldlm_completion_ast() assumes that a lock is granted when the req
2378 mode is equal to the granted mode. However, it should also check
2379 that LDLM_FL_CP_REQD is not set.
2383 Description: Heavy nfs access might result in deadlocks
2384 Details : After ELC code landed, it is now improper to enqueue any mds
2385 locks under och_sem, because enqueue might want to decide to
2386 cancel open locks for same inode we are holding och_sem for.
2390 Description: Client eviction while running blogbench
2391 Details : A lot of unlink operations with concurrent I/O can lead to a
2392 deadlock causing evictions. To address the problem, the number of
2393 oustanding OST_DESTROY requests is now throttled to
2394 max_rpcs_in_flight per OSC and LDLM_FL_DISCARD_DATA blocking
2395 callbacks are processed in priority.
2399 Description: enable ACLs on MDS by default
2400 Details : ACLs must be enabled on MDS by default.
2403 Frequency : PPC/PPC64 only
2405 Description: conflicts between asm-ppc64/types.h and lustre_types.h
2406 Details : fix duplicated definitions between asm-ppc64/types.h and
2407 lustre_types.h on PPC.
2410 Frequency : PPC/PPC64 only
2412 Description: asm-ppc/segment.h does not exist
2413 Details : fix compile issue on PPC.
2417 Description: better handle error messages in extents code
2420 Frequency : RHEL4 only
2422 Description: mkfs is very slow on IA64/RHEL4
2423 Details : A performance regression has been discovered in the MPT Fusion
2424 driver between versions 3.02.73rh and 3.02.99.00rh. As a
2425 consequence, we have downgraded the MPT Fusion driver in the RHEL4
2426 kernel from 3.02.99.00 to 3.02.73 until this problem is fixed.
2428 Severity : enhancement
2430 Description: SNMP support enhancement
2431 Details : Adding total number of sampled request for an MDS node in snmp
2434 Severity : enhancement
2436 Description: Optimize ldlm waiting list processing for PR extent locks
2437 Details : When processing waiting list for read extent lock and meeting read
2438 lock that is same or wider to it that is not contended, skip
2439 processing rest of the list and immediatelly return current
2440 status of conflictness, since we are guaranteed there are no
2441 conflicting locks in the rest of the list.
2445 Description: Time out and refuse to reconnect
2446 Details : When the failover node is the primary node, it is possible
2447 to have two identical connections in imp_conn_list. We must
2448 compare not conn's pointers but NIDs, otherwise we can defeat
2449 connection throttling.
2453 Description: port llog fixes from b1_6 into HEAD
2454 Details : Port llog reference couting and some llog cleanups from b1_6
2455 (bug 10800) into HEAD, for protect from panic and access to already
2456 free llog structures.
2460 Description: Detect stride IO mode in read-ahead
2461 Details : When a client does stride read, read-ahead should detect that and
2462 read-ahead pages according to the detected stride pattern.
2466 Description: data checksumming impacts single node performance
2467 Details : add support for several checksum algorithm. Currently, only CRC32
2468 and Adler-32 are supported. The checksum type can be changed on
2469 the fly via /proc/fs/lustre/osc/*/checksum_type.
2473 Description: use adler32 for page checksums
2474 Details : when available, use the Adler-32 algorithm instead of CRC32 for
2479 Description: build for x2 fails
2480 Details : fix compile issue on Cray systems.
2484 Description: Properly match for duplicate locks
2485 Details : Due to different lock order from skiplists code, we need to
2486 traverse entire list for now
2489 Frequency : only on PPC/SLES10
2491 Description: "BITS_PER_LONG is not 32 or 64" in linux/idr.h
2492 Details : On SLES10/PPC, fs.h includes idr.h which requires BITS_PER_LONG to
2493 be defined. Add a hack in mkfs_lustre.c to work around this compile
2498 Description: LASSERT on MDS when client holding flock lock dies
2499 Details : ldlm pool logic depends on number of granted locks equal to
2500 number of released locks which is not true for flock locks, so
2501 just exclude such locks from consideration.
2505 Description: MDS deadlock with many ll_sync_lov threads and I/O stalled
2506 Details : Use fsfilt_sync() for both the whole filesystem sync and
2507 individual file sync to eliminate dangerous inode locking
2508 with I_LOCK that can lead to a deadlock.
2512 Description: performance in 1.6.3
2513 Details : Force q->max_phys_segments to MAX_PHYS_SEGMENTS on SLES10 to be
2514 sure that 1MB requests are not fragmented by the block layer.
2516 Severity : enhancement
2518 Description: organize the server-side client stats on per-nid basis
2519 Details : Change the structure of stats under obdfilter and mds to
2527 The "uuid"s file would list the uuids of _active_ exports.
2528 And the clear entry is to clear all stats and stale nids.
2530 Severity : enhancement
2532 Description: eliminate client locks in face of contention
2533 Details : file contention detection and lockless i/o implementation
2534 for contended files.
2538 Description: Reinitialize optind to 0 so that interactive lfs works in all cases
2541 Frequency : very rare, if additional xattrs are used on kernels >= 2.6.12
2543 Description: MDS may lose file striping (and hence file data) in some cases
2544 Details : If there are additional extended attributes stored on the MDS,
2545 in particular ACLs, SELinux, or user attributes (if user_xattr
2546 is specified for the client mount options) then there is a risk
2547 of attribute loss. Additionally, the Lustre file striping
2548 needs to be larger than default (e.g. striped over all OSTs),
2549 and an additional attribute must be stored initially in the
2550 inode and then increase in size enough to be moved to the
2551 external attribute block (e.g. ACL growing in size) for file
2556 Description: skiplist implementation simplification
2557 Details : skiplists are used to group compatible locks on granted list
2558 that was implemented as tracking first and last lock of each
2559 lock group the patch changes that to using doubly linked lists
2563 Description: MDS LBUG: ASSERTION(!IS_ERR(dchild))
2564 Details : Change LASSERTs to client eviction (i.e. abort client's recovery)
2565 because LASSERT on both the data supplied by a client, and the
2566 data on disk is dangerous and incorrect.
2568 Severity : enhancement
2570 Description: Slow truncate/writes to huge files at high offsets.
2571 Details : Directly associate cached pages to lock that protect those pages,
2572 this allows us to quickly find what pages to write and remove
2573 once lock callback is received.
2577 Description: more ldlm soft lockups
2578 Details : In ldlm_resource_add_lock(), call to ldlm_resource_dump()
2579 starve other threads from the resource lock for a long time in
2580 case of long waiting queue, so change the debug level from
2581 D_OTHER to the less frequently used D_INFO.
2583 Severity : enhancement
2585 Description: add -gid, -group, -uid, -user options to lfs find
2589 Description: Hung threads in invalidate_inode_pages2_range
2590 Details : The direct IO path doesn't call check_rpcs to submit a new RPC
2591 once one is completed. As a result, some RPCs are stuck in the
2592 queue and are never sent.
2596 Description: filter threads hungs on waiting journal commit
2597 Details : Cleanup filter group llog code, then only filter group llog will
2598 be only created in the MDS/OST syncing process.
2602 Description: Procfs and llog threads access destoryed import sometimes.
2603 Details : Sync the import destoryed process with procfs and llog threads by
2604 the import refcount and semaphore.
2606 Severity : enhancement
2608 Description: openlock cache of b1_6 port to HEAD
2613 Description: kernel BUG at ldiskfs2_ext_new_extent_cb
2614 Details : If insertion of an extent fails, then discard the inode
2615 preallocation and free data blocks else it can lead to duplicate
2620 Description: don't always update ctime in ext3_xattr_set_handle()
2621 Details : Current xattr code updates inode ctime in ext3_xattr_set_handle.
2622 In some cases the ctime should not be updated, for example for
2623 2.0->1.8 compatibility it is necessary to delete an xattr and it
2624 should not update the ctime.
2628 Bugzilla : 15713/16362
2629 Description: Assertion in iopen_connect_dentry in 1.6.3
2630 Details : looking up an inode via iopen with the wrong generation number can
2631 populate the dcache with a disconneced dentry while the inode
2632 number is in the process of being reallocated. This causes an
2633 assertion failure in iopen since the inode's dentry list contains
2634 both a connected and disconnected dentry.
2638 Description: assertion failure in ldlm_handle2lock()
2639 Details : fix a race between class_handle_unhash() and class_handle2object()
2640 introduced in lustre 1.6.5 by bug 13622.
2645 Description: Kernel BUG: sd_iostats_bump: unexpected disk index
2646 Details : remove the limit of 256 scsi disks in the sd_iostat patch
2651 Description: oops in sd_iostats_seq_show()
2652 Details : unloading/reloading the scsi low level driver triggers a kernel
2653 bug when trying to access the sd iostat file.
2658 Description: Kernel panics during QLogic driver reload
2659 Details : REQ_BLOCK_PC requests are not handled properly in the sd iostat
2660 patch, causing memory corruption.
2665 Description: journal_dev option does not work in b1_6
2666 Details : pass mount option during pre-mount.
2668 Severity : enhancement
2670 Description: Add a FIEMAP(FIle Extent MAP) ioctl
2671 Details : FIEMAP ioctl will allow an application to efficiently fetch the
2672 extent information of a file. It can be used to map logical blocks
2673 in a file to physical blocks in the block device.
2677 Description: LDLM soft lockups - improvement
2678 Details : It is be possible to send the lock handle along with each read
2679 or write request because the client is already doing a lock match
2680 itself so there isn't any reason the OST should have to re-do that
2684 Frequency : only on Cray X2
2686 Description: X2 build failures
2687 Details : fix build failures on Cray X2.
2691 Description: xid & resent requests
2692 Details : Initialize RPC XID from clock at startup (randomly if clock is
2695 Severity : enhancement
2697 Description: Add lustre_start utility to start or stop multiple Lustre servers
2702 Description: Lustre GPF in {:ptlrpc:ptlrpc_server_free_request+373}
2703 Details : In case of memory pressure, list_del() can be called twice on
2704 req->rq_history_list, causing a kernel oops.
2708 Description: kptllnd_peer_check_sends()) ASSERTION(!in_interrupt()) failed
2709 Details : fix stack overflow in the distributed lock manager by defering
2710 export eviction after a failed AST to the elt thread instead of
2711 handling it in the dlm interpret routine.
2715 Description: Convert some comments to new format.
2716 Details : Update documenting comments to match doxygen conventions.
2720 Description: Grammar fixes.
2721 Details : A couple of trivial spelling fixes.
2725 Description: OSD_COUNTERS-mandatory
2726 Details : Make previously optional ->oti_{w,r}_locks sanity checks mandatory
2727 to simplify the code.
2731 Description: simplify cmm_device freeing logic.
2732 Details : Call cmm_device_free() in the failure path of cmm_device_alloc().
2736 Description: Add lockdep support to dt_object_operations locking interface.
2737 Details : Augment ->do_{read,write}_lock() prototypes with a `role'
2738 parameter indicating lock ordering. Update mdd code to use new
2743 Description: Introduce failloc constants for lockless IO tests.
2744 Details : Add two new failloc constants to test lockless IO. Only one of
2745 them in implemented---another is checked in yet to be landed
2750 Description: Add lockdep support for inode mutex.
2751 Details : Introduce and use new LOCK_INODE_MUTEX_PARENT() macro to be used
2752 in the situations where Lustre has to lock more than one inode
2757 Description: Add optional invariants checking support.
2758 Details : Add new LINVRNT() macro, optional on new --enable-invariants
2759 configure switch. This macro is to be used for consistency and
2760 sanity checks that are too expensive to be left in `production'
2765 Description: Zap lock->l_granted_mode with explicit LCK_MINMODE.
2766 Details : Use LCK_MINMODE rather than 0 to reset lock->l_granted_mode to
2771 Description: Add lockdep support for ldlm_lock and ldlm_resource.
2772 Details : Use spin_lock_nested() in (the only) situation where more than
2773 one ldlm_lock is locked simultaneously. Also, fix possible
2774 dead-lock in ldlm_lock_change_resource() by enforcing particular
2779 Description: Use struct ldlm_callback_suite in ldlm_lock_create().
2780 Details : Instead of specifying each ldlm_lock call-back through separate
2781 parameter, wrap them into struct ldlm_callback_suite.
2785 Description: Kill join_lru obd method and its callers.
2786 Details : CLIO uses lock weighting policy to keep locks over mmapped regions
2787 in memory---a requirement implemented through ->o_join_lru() obd
2788 method in HEAD. Remove this method and its users.
2792 Description: Add asynchronous ldlm ENQUEUE completion handler.
2793 Details : CLIO posts ENQUEUE requests asynchronously through ptlrpcd---a
2794 case that stock ldlm_completion_ast() cannot handle as it waits
2795 until lock is granted. Introduce new ldlm_completion_ast_async()
2796 for this. Also comment ldlm_completion_ast().
2800 Description: ldlm_error <-> errno conversion.
2801 Details : Add functions to map (rather arbitrary) between LDLM error codes
2802 and standard errno values. CLIO needs this to prevent LDLM specific
2803 constants from escaping ldlm and osc.
2807 Description: Kill unused ldlm_handle2lock_ns() function.
2808 Details : Kill unused ldlm_handle2lock_ns() function.
2812 Description: Add lu_ref support to ldlm_lock
2813 Details : lu_ref support for ldlm_lock and ldlm_resource. See lu_ref patch.
2814 lu_ref fields ->l_reference and ->lr_reference are added to ldlm_lock
2815 and ldlm_resource. LDLM interface has to be changed, because code that
2816 releases a reference on a lock, has to "know" what reference this is.
2817 In the most frequent case
2819 lock = ldlm_handle2lock(handle);
2821 LDLM_LOCK_PUT(lock);
2823 no changes are required. When any other reference (received _not_ from
2824 ldlm_handle2lock()) is released, LDLM_LOCK_RELEASE() has to be called
2825 instead of LDLM_LOCK_PUT().
2827 Arguably, changes are pervasive, and interface requires some discipline
2828 for proper use. On the other hand, it was very instrumental in finding
2829 a few leaked lock references.
2833 Description: Add ldlm_lock_addref_try().
2834 Details : Introduce ldlm_lock_addref_try() function (used by CLIO) that
2835 attempts to addref a lock that might be being canceled
2840 Description: Add ldlm_weigh_callback().
2841 Details : Add new ->l_weigh_ast() call-back to ldlm_lock. It is called
2842 by ldlm_cancel_shrink_policy() to estimate lock "value", instead of
2843 hard-coded `number of pages' logic.
2847 Description: Add lockdep annotations to llog code.
2848 Details : Use appropriately tagged _nested() locking calls in the places
2849 where llog takes more than one ->lgh_lock lock.
2853 Description: Add loi_kms_set().
2854 Details : Wrap kms updates into a helper function.
2858 Description: Constify instances of struct lsm_operations.
2859 Details : Constify instances of struct lsm_operations.
2863 Description: lu_conf support.
2864 Details : On a server, a file system object is uniquely identified
2865 by a fid, which is sufficient to locate and load all object
2866 state (inode). On a client, on the other hand, more data are
2867 necessary instantiate an object. Change lu_object_find() and
2868 friends to take additional `lu_conf' argument describing object.
2869 Typically this includes layout information.
2873 Description: lu_context fixes.
2874 Details : Introduce new lu_context functions that are needed on the client
2875 side, where some system threads (ptlrpcd) are shared by multiple
2876 modules, and so cannot be stopped during module shutdown.
2880 Description: Add start and stop methods to lu_device_type_operations.
2881 Details : Introduce two new methods in lu_device_type_operations, that are
2882 invoked when first instance of a given type is created and last one
2883 is destroyed respectively. This is need by CLIO.
2887 Description: Add lu_ref support to struct lu_device.
2888 Details : Add lu_ref support to lu_object and lu_device. lu_ref is used to
2889 track leaked references.
2893 Description: Introduce lu_kmem_descr.
2894 Details : lu_kmem_descr and its companion interface allow to create
2895 and destroy a number of kmem caches at once.
2899 Description: Fix lu_object finalization race.
2900 Details : Fix a race between lu_object_find() finding an object and its
2901 concurrent finalization. This race is (most likely) not possible
2902 on the server, but might happen on the client.
2906 Description: Introduce lu_ref interface.
2907 Details : lu_ref is a debugging module allowing to track references to
2908 a given object. It is quite cpu expensive, and has to be
2909 explicitly enabled with --enable-lu_ref. See usage description
2914 Description: Factor lu_site procfs stats into a separate function.
2915 Details : Separate lu_site stats printing code into a separate function
2916 that can be reused on a client.
2920 Description: Constify instances of struct {lu,dt,md}_device_operations.
2921 Details : Constify instances of struct {lu,dt,md}_device_operations.
2925 Description: Introduce struct md_site and move meta-data specific parts of
2926 struct lu_site here.
2927 Details : Move md-specific fields out of struct lu_site into special struct
2928 md_site, so that lu_site can be used on a client.
2932 Description: Kill mdd_lov_destroy().
2933 Details : Remove unused mdd code.
2937 Description: Add st_block checking to multistat.c.
2938 Details : Add st_block checking to multistat.c.
2942 Description: Add lu_ref support to struct obd_device.
2943 Details : Add lu_ref tracking to obd_device.
2947 Description: Kill obd_set_fail_loc().
2948 Details : Remove unused code.
2952 Description: Add special type for ptlrpc_request interpret functions.
2953 Details : Add lu_env parameter to ->rq_interpreter call-back. NULL is passed
2954 there. Actual usage will be in CLIO.
2958 Description: Replace RW_LOCK_UNLOCKED() macro with rwlock_init().
2959 Details : Replace RW_LOCK_UNLOCKED() with rwlock_init() as the former
2960 doesn't work with lockdep.
2964 Description: Add rwv.c test program.
2965 Details : New testing program exercising readv(2) and writev(2) (Qian).
2969 Description: Add sendfile.c test program.
2970 Details : New testing program exercising sendfile(2) (Jay).
2974 Description: Ratelimit a message that can be very frequent.
2975 Details : Ratelimit a memory allocation failure message that can
2980 Description: Use cdebug_show() in CDEBUG-style macros defined outside of libcfs.
2981 Details : Use cdebug_show() in CDEBUG-style macros defined outside of libcfs.
2985 Description: Liblustre build fixes.
2986 Details : Liblustre build fixes.
2990 Description: libcfs: add cfs_{need,cond}_resched() interface.
2991 Details : libcfs: add cfs_{need,cond}_resched() definition and
2992 implementations for Linux, NT, and liblustre.
2994 Severity : enhancement
2996 Description: More exported tunables for mballoc
2997 Details : Add support for tunable preallocation window and new tunables for
2998 large/small requests
3002 Description: Detect corruption of block bitmap and checking for preallocations
3003 Details : Checks validity of on-disk block bitmap. Also it does better
3004 checking of number of applied preallocations. When corruption is
3005 found, it turns filesystem readonly to prevent further corruptions.
3009 Description: (rw.c:1323:ll_read_ahead_pages()) ASSERTION(page_idx > ria->ria_stoff) failed
3010 Details : Once the unmatched stride IO mode is detected, shrink the stride-ahead
3011 window to 0. If it does hit cache miss, and read-pattern is still
3012 stride-io mode, does not reset the stride window, but also does not
3013 increase the stride window length in this case.
3017 Frequency : only for big-endian servers
3018 Description: Check if system is big-endian while mounting fs with extents feature
3019 Details : Mounting a filesystem with extents feature will fail on big-endian
3020 systems since ext3-based ldiskfs is not supported on big-endian
3021 systems. This can be over-riden with "bigendian_extents" mount option.
3023 Severity : enhancement
3025 Description: The root squash functionality
3026 Details : A security feature, which is to prevent users from being able
3027 to mount lustre on their desktop, run as root, and delete
3028 all of the files in the filesystem. The goal is accomplished by
3029 remapping user id (UID) and group id (GID) of the root user to
3030 a UID and GID specified by the system administartor via Lustre
3031 configuration management server (MGS). The functionality also
3032 allows to specify sets of clients for which the remapping does
3037 Description: Excessive recovery window
3038 Details : With AT enabled, the recovery window can be excessively long (6000+
3039 seconds). To address this problem, we no longer use
3040 OBD_RECOVERY_FACTOR when extending the recovery window (the connect
3041 timeout no longer depends on the service time, it is set to
3042 INITIAL_CONNECT_TIMEOUT now) and clients report the old service
3043 time via pb_service_time.
3047 Description: Watchdog triggered on MDS failover
3048 Details : enable OBD_CONNECT_MDT flag when connecting from the MDS so that
3049 the OSTs know that the MDS "UUID" can be reused for the same export
3050 from a different NID, so we do not need to wait for the export to be
3054 Frequency : rare, only if using MMP with Linux RAID
3056 Description: MMP doesn't work with Linux RAID
3057 Details : While using HA for Lustre servers with Linux RAID, it is possible
3058 that MMP will not detect multiple mounts. To make this work we
3059 need to unplug the device queue in RAID when the MMP block is being
3060 written. Also while reading the MMP block, we should read it from
3061 disk and not the cached one.
3063 Severity : enhancement
3065 Description: open file using fid
3066 Details : A file can be opened using just its fid, like
3067 <mntpt>/.lustre/fid/SEQ:OID:VER - this is needed for HSM and replication
3070 Frequency : Only in RHEL5 when mounting multiple ext3 filesystems
3073 Description: "kmem_cache_create: duplicate cache jbd_4k" error message
3074 Details : add proper locking for creation of jbd_4k slab cache
3078 Description: MMP check in ext3_remount() fails without displaying any error
3079 Details : When multiple mount protection fails during remount, proper error
3082 Severity : enhancement
3084 Description: Allow stripe size to be up to 4G-64k
3085 Details : Fix math logic to allow large stripe sizes.
3089 Description: add check for >8TB ldiskfs filesystems
3090 Details : ext3-based ldiskfs does not support greater than 8TB LUNs.
3091 Don't allow >8TB ldiskfs filesystems to be mounted without
3092 force_over_8tb mount option
3097 Descriptoin: OSS complains "too many blocks" and crashes
3098 Details : In filter_commitrw_write():
3099 1. lquota_chkquota();
3100 2. obdo_from_inode();
3101 3. lquota_pending_commit();
3102 "2." has changed uid and gid in this case, then caused
3103 lquota_chkquota() and lquota_pending_commit() to use
3104 different uid and gid. That is the root of the bug.
3108 Descriptoin: getstripe doesn't report stripe count = -1 sanely
3109 Details : Detect and print out the 'stripe count = -1' case properly.
3113 Description: Out or order replies might be lost on replay
3114 Details : In ptlrpc_retain_replayable_request if we cannot find retained
3115 request with tid smaller than one currently being added, add it
3116 to the start, not end of the list.
3118 --------------------------------------------------------------------------------
3120 2007-08-10 Cluster File Systems, Inc. <info@clusterfs.com>
3122 * Support for kernels:
3123 2.6.5-7.283 (SLES 9),
3124 2.6.9-55.EL (RHEL 4),
3125 2.6.16.46-0.14 (SLES 10),
3126 2.6.18.8 vanilla (kernel.org)
3127 * Client support for unpatched kernels:
3128 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
3129 2.6.16 - 2.6.22 vanilla (kernel.org)
3130 * Due to recently discovered recovery problems, we do not recommend
3131 using patchless RHEL 4 clients with this or any earlier release.
3132 * Recommended e2fsprogs version: 1.39.cfs8
3133 * Note that reiserfs quotas are disabled on SLES 10 in this kernel.
3134 * Starting with this release, the ldiskfs backing filesystem required
3135 by Lustre is now in its own package, lustre-ldiskfs. This package
3136 should be installed. It is versioned separately from Lustre and
3137 may be released separately in future.
3141 Description: block reactivating mgc import until all deactivates complete
3142 Details : Fix race when failing back MDT/MGS to itself (testing)
3145 Frequency : at statup only
3147 Description: mds_lov_synchronize race leads to various problems
3148 Details : simultaneous MDT->OST connections at startup can cause the
3149 sync to abort, leaving the OSC in a bad state.
3151 Severity : enhancement
3153 Description: add optional extra BUILD_VERSION info
3154 Details : add a new environment variable (namely LUSTRE_VERS) which allows
3155 to override the lustre version.
3158 Frequency : 2.6.18 servers only
3160 Description: ll_kern_mount() doesn't release the module reference
3161 Details : The ldiskfs module reference count never drops down to 0
3162 because ll_kern_mount() doesn't release the module reference.
3167 Description: server LBUG when using old ost_num_threads parameter
3168 Details : Accept the old ost_num_threads parameter but warn that it
3169 is deprecated, and fix an off-by-one error that caused an LBUG.
3174 Description: Transient SCSI error results in persistent IO issue
3175 Details : iobuf->dr_error is not reinitialized to 0 between two
3179 Frequency : sometimes when underlying device returns I/O errors
3181 Description: OSTs not going read-only during write failures
3182 Details : OSTs are not remounted read-only when the journal commit threads
3183 get I/O errors because fsfilt_ext3 calls journal_start/stop()
3184 instead of the ext3 wrappers.
3188 Description: poor connect scaling with increasing client count
3189 Details : Don't run filter_grant_sanity_check for more than 100 exports
3190 to improve scaling for large numbers of clients.
3193 Frequency : SLES10 only
3195 Description: sanity-quota.sh quotacheck failed: rc = -22
3196 Details : Quotas cannot be enabled on SLES10.
3199 Frequency : liblustre clients only
3201 Description: getdirentries does not give error when run on compute nodes
3202 Details : getdirentries does not fail when the size specified as an argument
3203 is too small to contain at least one entry
3205 Severity : enhancement
3207 Description: Add LNET router traceability for debug purposes
3208 Details : If a checksum failure occurs with a router as part of the
3209 IO path, the NID of the last router that forwarded the bulk data
3210 is printed so it can be identified.
3215 Description: OST "spontaneously" evicts client; client has imp_pingable == 0
3216 Details : Due to a race condition, liblustre clients were occasionally
3217 evicted incorrectly.
3219 Severity : enhancement
3221 Description: lfs setstripe use optional parameters instead of postional
3224 Severity : enhancement
3226 Description: Nanosecond timestamp support for ldiskfs
3227 Details : The on-disk ldiskfs filesystem has added support for nanosecond
3228 resolution timestamps. There is not yet support for this at
3229 the Lustre filesystem level.
3232 Frequency : during server recovery
3234 Description: MDS failing to send precreate requests due to OSCC_FLAG_RECOVERING
3235 Details : request with rq_no_resend flag not awake l_wait_event if they get
3239 Frequency : nfs export on patchless client
3241 Description: connectathon hang when test nfs export over patchless client
3242 Details : Disconnected dentry cannot be found with lookup, so we do not need
3243 to unhash it or make it invalid
3246 Description: fix llapi_lov_get_uuids() to allow many OSTs to be returned
3247 Details: : Change llapi_lov_get_uuids() to read the UUIDs from /proc instead
3248 of using an ioctl. This allows lfsck for > 160 OSTs to succeed.
3253 Description: open req refcounting wrong on reconnect
3254 Details : If reconnect happened between getting open reply from server and
3255 call to mdc_set_replay_data in ll_file_open, we will schedule
3256 replay for unreferenced request that we are about to free.
3257 Subsequent close will crash in variety of ways.
3258 Check that request is still eligible for replay in
3259 mdc_set_replay_data().
3264 Description: disable writes to filesystem when reading health_check file
3265 Details : the default for reading the health_check proc file has changed
3266 to NOT do a journal transaction and write to disk, because this
3267 can cause reads of the /proc file to hang and block HA state
3268 checking on a healthy but otherwise heavily loaded system. It
3269 is possible to return to the previous behaviour during configure
3270 with --enable-health-write.
3272 Severity : enhancement
3274 Description: 64-bit inode version
3275 Details: : Add a on-disk 64-bit inode version for ext3 to track changes made
3276 to the inode. This will be required for version-based recovery.
3281 Description: MDS fails to start if a duplicate client export is detected
3282 Details : in some rare cases it was possible for a client to connect to
3283 an MDS multiple times. Upon recovery the MDS would detect this
3284 and fail during startup. Handle this more gracefully.
3286 Severity : enhancement
3288 Description: Add -o localflock option to simulate old noflock
3290 Details : This will achieve local-only flock/fcntl locks
3296 Description: log_commit_thread vs filter_destroy race leads to crash
3297 Details : Take import reference before releasing llog record semaphore
3302 Description: Wrong request locking in request set processing
3303 Details : ptlrpc_check_set wrongly uses req->rq_lock for proctect add to
3304 imp_delayed_list, in this place should be used imp_lock.
3307 Frequency : when reconnection
3309 Description: Grant Leak when osc reconnect to OST
3310 Details : When osc reconnect ost, OST(filter) should check whether it
3311 should grant more space to client by comparing fed_grant and
3312 cl_avail_grant, and return the granted space to client instead
3313 of "new granted" space, because client will call osc_init_grant
3314 to update the client grant space info.
3317 Frequency : when client reconnect to OST
3319 Description: Grant Leak when osc do resend and replay bulk write
3320 Details : When osc reconnect to OST, OST(filter)should clear grant info of
3321 bulk write request, because the grant info while be sync between
3322 OSC and OST when reconnect, and we should ignore the grant info
3323 these of resend/replay write req.
3328 Description: Grant space more than available left space sometimes.
3329 Details : When then OST is about to be full, if two bulk writing from
3330 different clients came to OST. Accord the avaliable space of the
3331 OST, the first req should be permitted, and the second one
3332 should be denied by ENOSPC. But if the seconde arrived before
3333 the first one is committed. The OST might wrongly permit second
3334 writing, which will cause grant space > available space.
3337 Frequency : when client is evicted
3339 Description: Grant might be wrongly erased when osc is evicted by OST
3340 Details : when the import is evicted by server, it will fork another
3341 thread ptlrpc_invalidate_import_thread to invalidate the
3342 import, where the grant will be set to 0. While the original
3343 thread will update the grant it got when connecting. So if
3344 the former happened latter, the grant will be wrongly errased
3345 because of this race.
3350 Description: Checking Stale with correct fid
3351 Details : ll_revalidate_it should uses de_inode instead of op_data.fid2
3352 to check whether it is stale, because sometimes, we want the
3353 enqueue happened anyway, and op_data.fid2 will not be initialized.
3355 Severity : enhancement
3357 Description: update patchless client
3358 Details : Add support for patchless client with 2.6.20, 2.6.21 and RHEL 5
3361 Frequency : only with 2.4 kernel
3363 Description: random memory corruption
3364 Details : size of struct ll_inode_info is to big for union inode.u and this
3365 can be cause of random memory corruption.
3370 Description: Memory leak in recovery
3371 Details : Lov_mds_md was not free in an error handler in mds_create_object.
3372 It should also check obd_fail before fsfilt_start, otherwise if
3373 fsfilt_start return -EROFS,(failover mds during mds recovery).
3374 then the req will return with repmsg->transno = 0 and rc = EROFS.
3375 and we met hit the assert LASSERT(req->rq_reqmsg->transno ==
3376 req->rq_repmsg->transno) in ptlrpc_replay_interpret. Fcc should
3377 be freed no matter whether fsfilt_commit success or not.
3380 Frequency : only with huge count clients
3382 Description: Prevents from taking the superblock lock in llap_from_page for
3384 Details : using LL_ORIGIN_REMOVEPAGE origin flag instead of LL_ORIGIN_UNKNOW
3385 for llap_from_page call in ll_removepage prevents from taking the
3386 superblock lock for a soon died page.
3391 Description: Not check open intent error before release open handle
3392 Details : in some rare cases, the open intent error is not checked before
3393 release open handle, which may cause
3394 ASSERTION(open_req->rq_transno != 0), because it tries to release
3395 the failed open handle.
3400 Description: Set cat log bitmap only after create log success.
3401 Details : in some rare cases, the cat log bitmap is set too early. and it
3402 should be set only after create log success.
3406 Description: Accessing a block bevice can re-enable I/O when Lustre is
3407 tearing down a device.
3408 Details : dev_clear_rdonly(bdev) must be called in kill_bdev() instead of
3413 Description: service threads may hog cpus when there are a lot of requests
3414 Details : Insert cond_resched to give other threads a chance to use some CPU
3419 Description: the cat log was not initialized in recovery
3420 Details : When mds(mgs) do recovery, the tgt_count might be zero, so the
3421 unlink log on mds will not be initialized until mds post
3422 recovery. And also in mds post recovery, the unlink log will
3423 initialization will be done asynchronausly, so there will be race
3424 between add unlink log and unlink log initialization.
3428 Description: brw_stats were being printed incorrectly
3429 Details : brw_stats were being printed as log2 but all of them were not
3430 recorded as log2. Also remove some code duplication arising from
3431 filter_tally_{read,write}.
3435 Frequency : rare, only in recovery.
3436 Description: ASSERTION(req->rq_type != LI_POISON) failed
3437 Details : imp_lock should be held while iterating over imp_sending_list for
3438 prevent destroy request after get timeout in ptlrpc_queue_wait.
3442 Description: replay-single.sh test 52 fails
3443 Details : A lock's skiplist need to be cleanup when it being unlinked
3444 from its resource list.
3448 Description: Short directio read returns full requested size rather than
3450 Details : Direct I/O operations should return actual amount of bytes
3451 transferred rather than requested size.
3453 Severity : enhancement
3455 Description: metadata RPC reduction (e.g. for rm performance)
3456 Details : decrease the amount of synchronous RPC between clients and servers
3457 by canceling conflicing lock before the operation on the client
3458 and packing thier handles into the main operation RPC to server.
3460 Severity : enhancement
3462 Description: Async OSC create to avoid the blocking unnecessarily.
3463 Details : If an OST has no remaining object, system will block on the
3464 creation when it needs to create a new object on this OST. Now,
3465 ways use pre-created objects when available, instead of blocking on
3466 an empty osc while others are not empty. If we must block, we
3467 block for the shortest possible period of time.
3471 Description: improve handling recoverable errors
3472 Details : if request processig with error which can be recoverable on server
3473 request should be resend, otherwise page released from cache and
3476 Severity : enhancement
3478 Description: refine locking for avoid write wrong info into lov_objid file
3479 Details : fix possible races with add new target and write/update data in
3482 --------------------------------------------------------------------------------
3484 2007-05-03 Cluster File Systems, Inc. <info@clusterfs.com>
3489 Frequency : on some architectures
3491 Description: 1.6 client sometimes fails to mount from a 1.4 MDT
3492 Details : Uninitialized flags sometimes cause configuration commands to
3496 Frequency : patchless clients only
3498 Description: missing __iget() symbol export
3499 Details : The __iget() symbol export is missing. To avoid the need for
3500 this on patchless clients the deathrow inode reaper is turned
3501 off, and we depend on the VM to clean up old inodes. This
3502 dependency was during via the fix for bug 12181.
3504 --------------------------------------------------------------------------------
3506 2007-04-19 Cluster File Systems, Inc. <info@clusterfs.com>
3508 * CONFIGURATION CHANGE. This version of Lustre WILL NOT
3509 INTEROPERATE with older versions automatically. In many cases a
3510 special upgrade step is needed. Please read the
3511 user documentation before upgrading any part of a live system.
3512 * WIRE PROTOCOL CHANGE from previous 1.6 beta versions. This
3513 version will not interoperate with 1.6 betas before beta5 (1.5.95).
3514 * WARNING: Lustre configuration and startup changes are required with
3515 this release. See https://mail.clusterfs.com/wikis/lustre/MountConf
3519 Severity : enhancement
3521 Description: MountConf
3522 Details : Lustre configuration is now managed via mkfs and mount
3523 commands instead of lmc and lconf. New obd types (MGS, MGC)
3524 are added for dynamic configuration management. See
3525 https://mail.clusterfs.com/wikis/lustre/MountConf for
3528 Severity : enhancement
3530 Description: dynamic OST addition
3531 Details : OSTs can now be added to a live filesystem
3533 Severity : enhancement
3535 Description: startup order invariance
3536 Details : MDTs and OSTs can be started in any order. Clients only
3537 require the MDT to complete startup.
3539 Severity : enhancement
3541 Description: parallel, asynchronous orphan cleanup
3542 Details : orphan cleanup is now performed in separate threads for each
3543 OST, allowing parallel non-blocking operation.
3545 Severity : enhancement
3547 Description: optimized stripe assignment
3548 Details : stripe assignments are now made based on ost space available,
3549 ost previous usage, and OSS previous usage, in order to try
3550 to optimize storage space and networking resources.
3552 Severity : enhancement
3554 Description: Permanently set tunables
3555 Details : All writable /proc/fs/lustre tunables can now be permanently
3556 set on a per-server basis, at mkfs time or on a live system.
3558 Severity : enhancement
3560 Description: Lustre message v2
3561 Details : Add lustre message format v2.
3563 Severity : enhancement
3565 Description: client OST exclusion list
3566 Details : Clients can be started with a list of OSTs that should be
3567 declared "inactive" for known non-responsive OSTs.
3571 Description: SPEC SFS validation failure on NFS v2 over lustre.
3572 Details : Changes the blocksize for regular files to be 2x RPC size,
3573 and not depend on stripe size.
3575 Severity : enhancement
3577 Description: Multiple MD RPCs in flight.
3578 Details : Further unserialise some read-only MDS RPCs - learn about intents.
3579 To avoid overly-overloading MDS, introduce a limit on number of
3580 MDS RPCs in flight for a single client and add /proc controls
3581 to adjust this limit.
3583 Severity : enhancement
3585 Description: client read/write statistics
3586 Details : Add client read/write call usage stats for performance
3587 analysis of user processes.
3588 /proc/fs/lustre/llite/*/offset_stats shows non-sequential
3589 file access. extents_stats shows chunk size distribution.
3590 extents_stats_per_process show chunk size distribution per
3593 Severity : enhancement
3595 Description: mds statistics
3596 Details : Add detailed mds operations statistics in
3597 /proc/fs/lustre/mds/*/stats.
3601 Description: Failure of copying files with lustre special EAs.
3602 Details : Client side always return success for setxattr call for lustre
3603 special xattr (currently only "trusted.lov").
3608 Description: Refcount LNET uuids
3609 Details : The global LNET uuid list grew linearly with every startup;
3610 refcount repeated list entries instead of always adding to
3613 Severity : enhancement
3615 Description: Dynamic service threads
3616 Details : Within a small range, start extra service threads
3617 automatically when the request queue builds up.
3620 Frequency : mixed-endian client/server environments
3622 Description: mixed-endian crashes
3623 Details : The new msg_v2 system had some failures in mixed-endian
3626 Severity : enhancement
3628 Description: Easy OST removal
3629 Details : OSTs can be permanently deactivated with e.g. 'lctl
3630 conf_param lustre-OST0001.osc.active=0'
3632 Severity : enhancement
3634 Description: MGS proc entries
3635 Details : Added basic proc entries for the MGS showing what filesystems
3638 Severity : enhancement
3640 Description: provide MGS failover
3641 Details : Added config lock reacquisition after MGS server failover.
3643 Severity : enhancement
3645 Description: add Linux 2.4 support
3646 Details : Added support for RHEL 2.4.21 kernel for 1.6 servers and clients
3650 Description: a large application tries to do I/O to the same resource and dies
3651 in the middle of it.
3652 Details : Check the req->rq_arrival time after the call to
3653 ost_brw_lock_get(), but before we do anything about
3654 processing it & sending the BULK transfer request. This
3655 should help move old stale pending locks off the queue as
3656 quickly as obd_timeout.
3659 Frequency : when an incorrect nid is specified during startup
3661 Description: ptlrpc connect to non-existant node causes kernel crash
3662 Details : LNET can't be re-entered from an event callback, which
3663 happened when we expire a message after the export has been
3664 cleaned up. Instead, hand the zombie cleanup off to another
3667 Severity : enhancement
3669 Description: plain/inodebits lock performance improvement
3670 Details : Grouping plain/inodebits in granted list by their request modes
3671 and bits policy, thus improving the performance of search through
3675 Frequency : only if OST filesystem is corrupted
3677 Description: client incorrectly hits assertion in ptlrpc_replay_req()
3678 Details : for a short time RPCs with bulk IO are in the replay list,
3679 but replay of bulk IOs is unimplemented. If the OST filesystem
3680 is corrupted due to disk cache incoherency and then replay is
3681 started it is possible to trip an assertion. Avoid putting
3682 committed RPCs into the replay list at all to avoid this issue.
3685 Frequency : liblustre (e.g. catamount) on a large cluster with >= 8 OSTs/OSS
3687 Description: System hang on startup
3688 Details : This bug allowed the liblustre (e.g. catamount) client to
3689 return to the app before handling all startup RPCs. This
3690 could leave the node unresponsive to lustre network traffic
3691 and manifested as a server ptllnd timeout.
3693 Severity : enhancement
3695 Description: Add "/proc/sys/lustre/debug_peer_on_timeout"
3696 Details : liblustre envirable: LIBLUSTRE_DEBUG_PEER_ON_TIMEOUT
3697 boolean to control whether to print peer debug info when a
3698 client's RPC times out.
3701 Frequency : only for kernels with patches from Lustre below 1.4.3
3703 Description: Remove old rdonly API
3704 Details : Remove old rdonly API which unused from at least lustre 1.4.3
3707 Frequency : only for devices with external journals
3709 Description: Set external device read-only also
3710 Details : During a commanded failover stop, we set the disk device
3711 read-only while the server shuts down. We now also set any
3712 external journal device read-only at the same time.
3715 Frequency : when upgrading from 1.4 while trying to change parameters
3717 Description: The wrong (new) MDC name was used when setting parameters for
3718 upgraded MDT's. Also allows changing of OSC (and MDC)
3719 parameters if --writeconf is specified at tunefs upgrade time.
3722 Frequency : when setting specific ost indicies
3724 Description: QOS code breaks on skipped indicies
3725 Details : Add checks for missing OST indicies in the QOS code, so OSTs
3726 created with --index need not be sequential.
3728 Severity : enhancement
3730 Description: Add uninit_groups feature to ldiskfs2 to speed up e2fsck
3731 Details : The uninit_groups feature works in conjunction with the kernel
3732 filesystem code (ldiskfs2 only) and e2fsprogs-1.39-cfs6 to speed
3733 up the pass1 processing of e2fsck. This is a read-only feature
3734 in ldiskfs2 only, so older kernels and current ldiskfs cannot
3735 mount filesystems that have had this feature enabled.
3737 Severity : enhancement
3739 Description: Improve multi-block allocation algorithm to avoid fragmentation
3740 Details : The mballoc3 code (ldiskfs2 only) adds new mechanisms to improve
3741 allocation locality and avoid filesystem fragmentation.
3743 ------------------------------------------------------------------------------
3745 2007-04-01 Cluster File Systems, Inc. <info@clusterfs.com>
3747 * Support for kernels:
3748 2.4.21-47.0.1.EL (RHEL 3)
3749 2.6.5-7.283 (SLES 9)
3750 2.6.9-42.0.10.EL (RHEL 4)
3751 2.6.12.6 vanilla (kernel.org)
3752 2.6.16.27-0.9 (SLES 10)
3753 * Recommended e2fsprogs version: 1.39.cfs5
3755 * Note that reiserfs quotas are disabled on SLES 10 in this kernel
3759 Frequency : occasional, depends on client load and configuration
3760 Bugzilla : 12181, 12203
3761 Description: data loss for recently-modified files
3763 Details : In some cases it is possible that recently written or created
3764 files may not be written to disk in a timely manner (this should
3765 normally be within 30s unless client IO load is very high).
3766 The problem appears as zero-length files or files that are a
3767 multiple of 1MB in size after a client crash or client eviction
3768 that are missing data at the end of the file.
3770 This problem is more likely to be hit on clients where files are
3771 repeatedly created and unlinked in the same directory, clients
3772 have a large amount of RAM, have many CPUs, the filesystem has
3773 many OSTs, the clients are rebooted frequently, and/or the files
3774 are not accessed by other nodes after being written.
3776 The presence of the problem can be detected by looking at
3777 /proc/sys/fs/inode-state. If the first number (nr_inodes) is
3778 smaller than the second (nr_unused) then dirty files will not
3779 be flushed automatically to disk. "sync; sleep 10" should be
3780 run several times on the node before unmounting it to update
3781 Lustre (this is also safe to run on nodes without this problem).
3783 There is also a related kernel bug in the RHEL4 4 2.6.9 kernel
3784 that can cause this same problem, so customers using that kernel
3785 also need to update the kernel in addition to Lustre. In order
3786 to properly fix this bug, the RHEL3 2.4.21 kernel is also updated.
3788 It is normal that files written just before a client crash (less
3789 than 30s) may not yet have been flushed to disk, even for local
3793 Frequency : frequent on thin XT3 nodes
3795 Description: UUID collision on thin XT3 Linux nodes
3796 Details : UUIDs on Compute Node Linux XT3 nodes were not generated
3797 randomly, since we relied on an insufficiently-seeded PRNG.
3802 Description: OSS hangs after "All ost request buffers busy"
3803 Details : A deadlock between quota and journal operations caused OSS
3804 hangs after printing "All ost request buffers busy."
3807 Frequency : always on liblustre builds
3809 Description: Cleanup compiler warnings on liblustre
3812 Frequency : always on liblustre builds on XT3
3814 Description: LC_CONFIG_CDEBUG don't run while build liblustre on XT3.
3818 Description: Addition of EXT3_FEATURE_RO_COMPAT_DIR_NLINKS flag for
3819 > 32000 subdirectories
3820 Details : Add EXT3_FEATURE_RO_COMPAT_DIR_NLINK flag to
3821 EXT3_FEATURE_RO_COMPAT_SUPP. This flag will be set whenever
3822 subdirectory count crosses 32000. This will aid e2fsck to
3823 correctly handle more than 32000 subdirectories.
3826 Frequency : liblustre (e.g. catamount) on a large cluster with >= 8 OSTs/OSS
3828 Description: System hang on startup
3829 Details : This bug allowed the liblustre (e.g. catamount) client to
3830 return to the app before handling all startup RPCs. This
3831 could leave the node unresponsive to lustre network traffic
3832 and manifested as a server ptllnd timeout.
3834 Severity : enhancement
3836 Description: Add "/proc/sys/lustre/debug_peer_on_timeout"
3837 (liblustre envirable: LIBLUSTRE_DEBUG_PEER_ON_TIMEOUT)
3838 boolean to control whether to print peer debug info when a
3839 client's RPC times out.
3844 Description: make O_SYNC working on 2.6 kernels
3845 Details : 2.6 kernels use different method for mark pages for write,
3846 so need add a code to lustre for O_SYNC work.
3851 Description: Failure to close file and release space on NFS
3852 Details : Put inode details into lock acquired in ll_intent_file_open.
3853 Use mdc_intent_lock in ll_intent_open to properly
3854 detect all kind of errors unhandled by mdc_enqueue.
3859 Description: proc file read during shutdown sometimes raced obd removal,
3861 Details : Add lock to prevent obd access after proc file removal.
3864 Frequency : Only for files larger than 4GB on 32-bit clients.
3866 Description: improperly doing page alignment of locks
3867 Details : Modify lustre core code to use CFS_PAGE_* defines instead of
3868 PAGE_*. Make CFS_PAGE_MASK a 64-bit mask.
3873 Description: RPCs being resent when they shouldn't be
3874 Details : Some RPCs that should not be resent are being resent. This
3875 can cause inconsistencies in the RPC state machine. Do not
3876 resend such requests.
3879 Frequency : rare, only with NFS export
3881 Description: Crash on NFS re-export node
3882 Details : under very unusual load conditions an assertion is hit in
3883 ll_intent_file_open()
3886 Frequency : only if OST filesystem is corrupted
3888 Description: client incorrectly hits assertion in ptlrpc_replay_req()
3889 Details : for a short time RPCs with bulk IO are in the replay list,
3890 but replay of bulk IOs is unimplemented. If the OST filesystem
3891 is corrupted due to disk cache incoherency and then replay is
3892 started it is possible to trip an assertion. Avoid putting
3893 committed RPCs into the replay list at all to avoid this issue.
3898 Description: large O_DIRECT requests fail under memory pressure/fragmentation
3899 Details : Large single O_DIRECT read and write calls can fail to allocate
3900 a sufficiently large buffer to process the request. In case of
3901 allocation failure the allocation is retried with a smaller
3902 buffer and broken into smaller requests.
3904 Severity : enhancement
3906 Description: Add -o localflock option to simulate old noflock behaviour.
3907 Details : This will achieve local-only flock/fcntl locks coherentness.
3912 Description: versioning check is incomplete
3913 Details : Checking the version difference of client vs. server, report
3914 error if the gap is too big.
3919 Description: add support PG_writeback bit
3920 Details : add support for PG_writeback bit for Lustre, for more carefull
3921 work with page cache in 2.6 kernel. This also fix some deadlocks
3922 and remove hack for work O_SYNC with 2.6 kernel.
3924 Severity : enhancement
3926 Description: Add uninit_groups feature to ldiskfs2 to speed up e2fsck
3927 Details : The uninit_groups feature works in conjunction with the kernel
3928 filesystem code (ldiskfs2 only) and e2fsprogs-1.39-cfs6 to speed
3929 up the pass1 processing of e2fsck. This is a read-only feature
3930 in ldiskfs2 only, so older kernels and current ldiskfs cannot
3931 mount filesystems that have had this feature enabled.
3933 Severity : enhancement
3935 Description: Improve multi-block allocation algorithm to avoid fragmentation
3936 Details : The mballoc3 code (ldiskfs2 only) adds new mechanisms to improve
3937 allocation locality and avoid filesystem fragmentation.
3939 ------------------------------------------------------------------------------
3941 2007-02-09 Cluster File Systems, Inc. <info@clusterfs.com>
3943 * Support for kernels:
3944 2.6.9-42.0.3.EL (RHEL 4)
3945 2.6.5-7.276 (SLES 9)
3946 2.4.21-47.0.1.EL (RHEL 3)
3947 2.6.12.6 vanilla (kernel.org)
3948 2.6.16.21-0.8 (SLES10)
3949 * Recommended e2fsprogs version: 1.39.cfs2-0
3951 * The backwards-compatible /proc/sys/portals symlink has been removed
3952 in this release. Before upgrading, please ensure that you change
3953 any configuration scripts or /etc/sysctl.conf files that access
3954 /proc/sys/portals/* or sysctl portals.* to use the corresponding
3955 entry in /proc/sys/lnet or sysctl lnet.*. This change can be made
3956 in advance of the upgrade on any system running Lustre 1.4.6 or
3957 newer, since /proc/sys/lnet was added in that version.
3958 * Note that reiserfs quotas are disabled on SLES 10 in this kernel
3962 Frequency : only when quota is used
3964 Description: avoid scanning export list for quota master
3965 Details : Change the algorithms to avoid scanning export list in order
3966 to improve the efficiency.
3969 Frequency : MDS failover only, very rarely
3971 Description: "went back in time" messages on mds failover
3972 Details : The greatest transno may be lost when the current operation
3973 finishes with an error (transno==0) and the client's last_rcvd
3974 record is over-written. Save the greatest transno in the
3975 mds_last_transno for this case.
3978 Frequency : always for specific kernels and striping counts
3980 Description: client may get "Matching packet too big" without ACL support
3981 Details : Clients compiled without CONFIG_FS_POSIX_ACL get an error message
3982 when trying to access files in certain configurations. The
3983 clients should in fact be denied when mounting because they do
3984 not understand ACLs.
3987 Frequency : Cray XT3 with more than 4000 clients and multiple jobs
3989 Description: many clients connecting with IO in progress causes connect timeouts
3990 Details : Avoid synchronous journal commits to avoid delays caused by many
3991 clients connecting/disconnecting when bulk IO is in progress.
3992 Queue liblustre connect requests on OST_REQUEST_PORTAL instead of
3993 OST_IO_PORTAL to avoid delays behind potentially many pending
3997 Frequency : occasionally with multiple writers to a single file
3999 Description: shared writes to file may result in wrong size reported by stat()
4000 Details : Allow growing of kms when extent lock is cancelled
4003 Frequency : always with random mmap IO to multi-striped file
4005 Description: mmap write might be lost if we are writing to a 'hole' in stripe
4006 Details : Only if the hole is at the end of OST object so that kms is too
4007 small. Fix is to increase kms accordingly in ll_nopage.
4010 Frequency : rare, only if OST filesystem is inconsistent with MDS filesystem
4012 Description: writes to a missing object would leak memory on the OST
4013 Details : If there is an inconsistency between the MDS and OST filesystems,
4014 such that the MDS references an object that doesn't exist, writes
4015 to that object will leak memory due to incorrect cleanup in the
4016 error handling path, eventually running out of memory on the OST.
4021 Description: Creating too long symlink causes lustre errors
4022 Details : Check symlink and name lengths before sending requests to MDS.
4025 Frequency : only if flock is enabled (not on by default)
4027 Description: posix locks not released on fd closure on 2.6.9+
4028 Details : We failed to add posix locks to list of inode locks on 2.6.9+
4029 kernels, this caused such locks not to be released on fd close and
4030 then assertions on fs unmount about still used locks.
4033 Frequency : MDS failover only, very rarely
4035 Description: clients may get ASSERTION(granted_lock != NULL)
4036 Details : When request was taking a long time, and a client was resending
4037 a getattr by name lock request. The were multiple lock requests
4038 with the same client lock handle and
4039 mds_getattr_name->fixup_handle_for_resent_request found one of the
4040 lock handles but later failed with ASSERTION(granted_lock != NULL).
4045 Description: handle->h_buffer_credits > 0, assertion failure
4046 Details : h_buffer_credits is zero after truncate, causing assertion
4047 failure. This patch extends the transaction or creates a new
4051 Frequency : NFS re-export or patchless client
4052 Bugzilla : 11179, 10796
4053 Description: Crash on NFS re-export node (__d_move)
4054 Details : We do not want to hash the dentry if we don't have a lock.
4055 But if this dentry is later used in d_move, we'd hit uninitialised
4056 list head d_hash, so we just do this to init d_hash field but
4057 leave dentry unhashed.
4060 Frequency : NFS re-export or patchless client
4062 Description: NFS exports has problem with symbolic link
4063 Details : lustre client didn't properly install dentry when re-exported
4064 to NFS or running patchless client.
4067 Frequency : NFS re-export or patchless client
4069 Description: Various nfs/patchless fixes.
4070 Details : fixes reuse disconected alias for lookup process - this fixes
4071 warning "find_exported_dentry: npd != pd",
4072 fix permission error with open files at nfs.
4073 fix apply umask when do revalidate.
4076 Frequency : occasional
4078 Description: Crash on NFS re-export node
4079 Details : calling clear_page() on the wrong pointer triggered oops in
4080 generic_mapping_read().
4083 Frequency : rarely, using O_DIRECT IO
4085 Description: unaligned directio crashes client with LASSERT
4086 Details : check for unaligned buffers before trying any requests.
4089 Frequency : rarely, using CFS RAID5 patches in non-standard kernel series
4091 Description: stale data returned from RAID cache
4092 Details : If only a small amount of IO is done to the RAID device before
4093 reading it again it is possible to get stale data from the RAID
4094 cache instead of reading it from disk.
4097 Frequency : always for sles10 kernel
4099 Description: sles10 support
4100 Details : ll_follow_link: compile fixes and using of nd_set_link
4101 under newer kernels.
4104 Frequency : depends on arch, kernel and compiler version, always on sles10
4107 Description: recursive or deep enough symlinks cause stack overflow
4108 Details : getting rid of large stack-allocated variable in
4112 Frequency : depends on hardware
4114 Description: lustre write performance loss in the SLES10 kernel
4115 Details : the performance loss is caused by using of write barriers in the
4116 ext3 code. The SLES10 kernel turns barrier support on by
4117 default. The fix is to undo that change for ldiskfs.
4119 ------------------------------------------------------------------------------
4121 2006-12-09 Cluster File Systems, Inc. <info@clusterfs.com>
4123 * Support for kernels:
4124 2.6.9-42.0.3EL (RHEL 4)
4125 2.6.5-7.276 (SLES 9)
4126 2.4.21-47.0.1.EL (RHEL 3)
4127 2.6.12.6 vanilla (kernel.org)
4131 Frequency : quota enabled and large files being deleted
4133 Description: releasing more than 4GB of quota at once hangs OST
4134 Details : If a user deletes more than 4GB of files on a single OST it
4135 will cause the OST to spin in an infinite loop. Release
4136 quota in < 4GB chunks, or use a 64-bit value for 1.4.7.1+.
4141 Description: statfs data retrieved from /proc may be stale or zero
4142 Details : When reading per-device statfs data from /proc, in the
4143 {kbytes,files}_{total,free,avail} files, it may appear
4144 as zero or be out of date.
4147 Frequency : systems with MD RAID1 external journal devices
4149 Description: lconf's call to blkid is confused by RAID1 journal devices
4150 Details : Use the "blkid -l" flag to locate the MD RAID device instead
4151 of returning all block devices that match the journal UUID.
4154 Frequency : always, for aggregate stripe size over 4GB
4156 Description: "lfs setstripe" fails assertion when setting 4GB+ stripe width
4157 Details : Using "lfs setstripe" to set stripe size * stripe count over 4GB
4158 will fail the kernel with "ASSERTION(lsm->lsm_xfersize != 0)"
4161 Frequency : always if "lfs find" used on a local file/directory
4163 Description: "lfs find" segfaults if used on a local file/directory
4164 Details : The case where a directory component was not specified wasn't
4165 handled correctly. Handle this properly.
4168 Frequency : always on ppc64
4170 Description: the write to an ext3 filesystem mounted with mballoc got stuck
4171 Details : ext3_mb_generate_buddy() uses find_next_bit() which does not
4172 perform endianness conversion.
4175 Frequency : rarely (truncate to non-zero file size after write under load)
4176 Bugzilla : 10730, 10687
4177 Description: Files padded with zeros to next 4K multiple
4178 Details : With filesystems mounted using the "extents" option (2.6 kernels)
4179 it is possible that files that are truncated to a non-zero size
4180 immediately after being written are filled with zero bytes beyond
4181 the truncated size. No file data is lost.
4183 Severity : enhancement
4185 Description: Allow recovery/failover for liblustre clients.
4186 Details : liblustre clients were unaware of failover configurations until
4189 Severity : enhancement
4191 Description: user file locks should fail when not mounting with flock option
4192 Details : Set up an error-returning stub in ll_file_operations.lock field
4193 to prevent incorrect behaviour when client is mounted without
4194 flock option. Also, set up properly f_op->flock field for
4198 Frequency : always on ia64
4200 Description: "lfs df" loops on printing out MDS statfs information
4201 Details : The obd_ioctl_data was not initialized and in some systems
4202 this caused a failure during the ioctl that did not return
4203 an error. Initialize the struct and return an error on failure.
4206 Frequency : SLES 9 only
4208 Description: Error of copying files with lustre special EAs as root
4209 Details : Client side always return success for setxattr call for lustre
4210 special xattr (currently only "trusted.lov").
4213 Frequency : rarely on clusters with both ia64+i386 clients
4215 Description: ia64+i686 clients doing shared IO on the same file may LBUG
4216 Details : In rare cases when both ia64+i686 (or other mixed-PAGE_SIZE)
4217 clients are doing concurrent writes to the same file it is
4218 possible that the ia64 clients may LASSERT because the OST
4219 extent locks are not PAGE_SIZE aligned. Ensure that grown
4220 locks are always aligned on the request boundary.
4223 Frequency : specific use, occasional
4225 Description: Overwriting in use executable truncates on-disk binary image
4226 Details : If one node attempts to overwrite an executable in use by
4227 another node, we now correctly return ETXTBSY instead of
4228 truncating the file.
4230 Severity : enhancement
4232 Description: Async OSC create to avoid the blocking unnecessarily.
4233 Details : If an OST has no remaining object, system will block on the creating
4234 when need to create a new object on this OST. Now, ways use
4235 pre-created objects when available, instead of blocking on an
4236 empty osc while others are not empty. If we must block, we block
4237 for the shortest possible period of time.
4242 Description: chmod on Lustre root is propagated to other clients
4243 Details : Re-validate root's dentry in ll_lookup_it to avoid having it
4244 invalid by the follow_mount time.
4247 Frequency : liblustre clients only
4249 Description: Race in 'instant cancel' lock handling could lead to such locks
4250 never to be granted in case of SMP MDS
4251 Details : Do not destroy not yet granted but cbpending locks in
4255 Frequency : replay/resend of open
4257 Description: non null lock assetion failure in mds_intent_policy
4258 Details : Trying to replay/resend lockless open requests resulted in
4259 mds_open() returning 0 with no lock. Now it sets a flag if
4260 a lock is going to be returned.
4262 Severity : enhancement
4264 Description: Checksum enhancements
4265 Details : New checksum enhancements allow for resending RPCs that failed
4268 Severity : enhancement
4270 Description: Tunables on number of dirty pages in cacche
4271 Details : Allow to set limit on number of dirty pages cached.
4276 Description: client crash on unmount - lock still has references
4277 Details : In some error handling cases it was possible to leak a lock
4278 reference on a client while accessing a file. This was not
4279 harmful to the client during operation, but would cause the
4280 client to crash when the filesystem is unmounted.
4283 Frequency : specific case, rare
4285 Description: ETXTBSY on mds though file not in use by client
4286 Details : ETXTBSY is no longer incorrectly returned when attempting to
4287 chmod or chown a directory that the user previously tried to
4288 execute or a currently-executing binary.
4291 Frequency : extremely rare except on liblustre-based clients
4293 Description: Lustre space not freed when files are deleted
4294 Details : Clean up open-unlinked files after client eviction. Previously
4295 the unlink was skipped and the files remained as orphans.
4300 Description: OST failure "would be an LBUG" in waiting_locks_callback()
4301 Details : In some cases it was possible to send a blocking callback to
4302 a client doing a glimpse, even though that client didn't get
4303 a lock granted. When the glimpse lock is cancelled on the OST
4304 the freed lock is left on the waiting list and corrupted the list.
4307 Frequency : all core dumps
4309 Description: Broke core dumps to lustre
4310 Details : Negative dentry may be unhashed if parent does not have UPDATE
4311 lock, but some callers, e.g. do_coredump, expect dentry to be
4312 hashed after successful create, hash it in ll_create_it.
4314 ------------------------------------------------------------------------------
4316 2006-09-13 Cluster File Systems, Inc. <info@clusterfs.com>
4318 * Support for kernels:
4319 2.6.9-42.0.2.EL (RHEL 4)
4320 2.6.5-7.276 (SLES 9)
4321 2.4.21-40.EL (RHEL 3)
4322 2.6.12.6 vanilla (kernel.org)
4326 Frequency : always on RHEL 3
4328 Description: Number of open files grows over time
4329 Details : The number of open files grows over time, whether or not
4330 Lustre is started. This was due to a filp leak introduced
4331 by one of our kernel patches.
4333 ------------------------------------------------------------------------------
4335 08-20-2006 Cluster File Systems, Inc. <info@clusterfs.com>
4337 * Support for kernels:
4338 2.6.9-42.EL (RHEL 4)
4339 2.6.5-7.276 (SLES 9)
4340 2.4.21-40.EL (RHEL 3)
4341 2.6.12.6 vanilla (kernel.org)
4346 Bugzilla : 5719, 9635, 9792, 9684
4347 Description: OST (or MDS) trips assertions in (re)connection under heavy load
4348 Details : If a server is under heavy load and cannot reply to new
4349 connection requests before the client resends the (re)connect,
4350 the connection handling code can behave badly if two service
4351 threads are concurrently handing separate (re)connections from
4352 the same client. Add better locking to the connection handling
4353 code, and ensure that only a single connection will be processed
4354 for a given client UUID, even if the lock is dropped.
4356 Severity : enhancement
4358 Description: add TCP zero-copy support to kernel
4359 Details : Add support to the kernel TCP stack to allow zero-copy bulk
4360 sends if the hardware supports scatter-gather and checksumming.
4361 This allows socklnd to do client-write and server-read more
4362 efficiently and reduce CPU utilization from skbuf copying.
4365 Frequency : only if NFS exporting from client
4367 Description: NULL pointer deref in ll_iocontrol() if chattr mknod file
4368 Details : If setting attributes on a file created under NFS that had
4369 never been opened it would be possible to oops the client
4370 if the file had no objects.
4373 Frequency : always for liblustre
4375 Description: liblustre client does MDS+OSTs setattr RPC for each write
4376 Details : When doing a write from a liblustre client, the client
4377 incorrectly issued an RPC to the MDS and each OST the file was
4378 striped over in order to update the timestamps. When writing
4379 with small chunks and many clients this could overwhelm the MDS
4380 with RPCs. In all cases it would slow down the write because
4381 these RPCs are unnecessary.
4383 Severity : enhancement
4385 Description: allow number of MDS service threads to be changed at module load
4386 Details : It is now possible to change the number of MDS service threads
4387 running. Adding "options mds mds_num_threads={N}" to the MDS's
4388 /etc/modprobe.conf will set the number of threads for the next
4389 time Lustre is restarted (assuming the "mds" module is also
4390 reloaded at that time). The default number of threads will
4391 stay the same, 32 for most systems.
4396 Description: OST crash if filesystem is unformatted or corrupt
4397 Details : If an OST is started on a device that has never been formatted
4398 or if the filesystem is corrupt and cannot even mount then the
4399 error handling cleanup routines would dereference a NULL pointer.
4404 Description: NULL pointer deref in llap_from_page.
4405 Details : get_cache_page_nowait can return a page with NULL (or otherwise
4406 incorrect) mapping if the page was truncated/reclaimed while it was
4407 searched for. Check for this condition and skip such pages when
4408 doing readahead. Introduce extra check to llap_from_page() to
4409 verify page->mapping->host is non-NULL (so page is not anonymous).
4412 Frequency : Sometimes when using sys_sendfile
4414 Description: "page not covered by a lock" warnings from ll_readpage
4415 Details : sendfile called ll_readpage without right page locks present.
4416 Now we introduced ll_file_sendfile that does necessary locking
4417 around call to generic_file_sendfile() much like we do in
4421 Frequency : with certain MDS communication failures at client mount time
4423 Description: NULL pointer deref after failed client mount
4424 Details : a client connection request may delayed by the network layer
4425 and not be sent until after the PTLRPC layer has timed out the
4426 request. If the client fails the mount immediately it will try
4427 to clean up before the network times out the request. Add a
4428 reference from the request import to the obd device and delay
4429 the cleanup until the network drops the request.
4432 Frequency : occasionally during client (re)connect
4434 Description: assertion failure during client (re)connect
4435 Details : processing a client connection request may be delayed by the
4436 client or server longer than the client connect timeout. This
4437 causes the client to resend the connection request. If the
4438 original connection request is replied in this interval, the
4439 client may trip an assertion failure in ptlrpc_connect_interpret()
4440 which thought it would be the only running connect process.
4443 Frequency : only with obd_echo servers and clients that are rebooted
4445 Description: kernel BUG accessing uninitialized data structure
4446 Details : When running an obd_echo server it did not start the ping_evictor
4447 thread, and when a client was evicted an uninitialized data
4448 structure was accessed. Start the ping_evictor in the RPC
4449 service startup instead of the OBD startup.
4451 Severity : enhancement
4452 Bugzilla : 10193 (patchless)
4453 Description: Remove dependency on various unexported kernel interfaces.
4454 Details : No longer need reparent_to_init, exit_mm, exit_files,
4455 sock_getsockopt, filemap_populate, FMODE_EXEC, put_filp.
4458 Frequency : rare (only users of deprecated and unsupported LDAP config)
4460 Description: write_conf for zeroconf mount queried LDAP incorrectly for client
4461 Details : LDAP apparently contains 'lustreName' attributes instead of
4462 'name'. A simple remapping of the name is sufficient.
4465 Frequency : rare (only with non-default dump_on_timeout debug enabled)
4467 Description: waiting_locks_callback trips kernel BUG if client is evicted
4468 Details : Running with the dump_on_timeout debug flag turned on makes
4469 it possible that the waiting_locks_callback() can try to dump
4470 the Lustre kernel debug logs from an interrupt handler. Defer
4471 this log dumping to the expired_lock_main() thread.
4473 Severity : enhancement
4475 Description: Support NFS exporting on 2.6 kernels.
4476 Details : Implement non-rawops metadata methods for NFS server to use without
4477 changing NFS server code.
4480 Frequency : very rare (synthetic metadata workload only)
4482 Description: two racing renames might cause an MDS thread to deadlock
4483 Details : Running the "racer" program may cause one MDS thread to rename
4484 a file from being the source of a rename to being the target of
4485 a rename at exactly the same time that another thread is doing
4486 so, and the second thread has already enqueued these locks after
4487 doing a lookup of the target and is trying to relock them in
4488 order. Ensure that we don't try to re-lock the same resource.
4491 Frequency : only very large systems with liblustre clients
4493 Description: slow eviction of liblustre clients with the "evict_by_nid" RPC
4494 Details : Use asynchronous set_info RPCs to send the "evict_by_nid" to
4495 all OSTs in parallel. This allows the eviction of stale liblustre
4496 clients to proceed much faster than if they were done in series,
4497 and also offers similar improvements for other set_info RPCs.
4502 Description: excessive CPU usage during initial read phase on client
4503 Details : During the initial read phase on a client, it would agressively
4504 retry readahead on the file, consuming too much CPU and impacting
4505 performance (since 1.4.5.8). Improve the readahead algorithm
4506 to avoid this, and also improve some other common cases (read
4507 of small files in particular, where "small" is files smaller than
4508 /proc/fs/lustre/llite/*/max_read_ahead_whole_mb, 2MB by default).
4513 Description: MDS crash when receiving packet with unknown intent.
4514 Details : Do not LBUG in unknown intent case, just return -EFAULT
4516 Severity : enhancement
4517 Bugzilla : 9293, 9385
4518 Description: MDS RPCs are serialised on client. This is unnecessary for some.
4519 Details : Do not serialize getattr (non-intent version) and statfs.
4522 Frequency : occasional, when OST network is overloaded/intermittent
4524 Description: client evicted by OST after bulk IO timeout
4525 Details : If a client sends a bulk IO request (read or write) the OST
4526 may evict the client if it is unresposive to its data GET/PUT
4527 request. This is incorrect if the network is overloaded (takes
4528 too long to transfer the RPC data) or dropped the OST GET/PUT
4529 request. There is no need to evict the client at all, since
4530 the pinger and/or lock callbacks will handle this, and the
4531 client can restart the bulk request.
4534 Frequency : Always when mmapping file with no objects
4536 Description: client crashes when mmapping file with no objects
4537 Details : Check that we actually have objects in a file before doing any
4538 operations on objects in ll_vm_open, ll_vm_close and
4544 Description: Request leak when working with deleted CWD
4545 Details : Introduce advanced request refcount tracking for requests
4546 referenced from lustre intent.
4548 Severity : Enhancement
4550 Description: Cache open file handles on client.
4551 Details : MDS now will return special lock along with openhandle, if
4552 requested and client is allowed to hold openhandle, even if unused,
4553 until such a lock is revoked. Helps NFS a lot, since NFS is opening
4554 closing files for every read/write openration.
4556 Severity : Enhancement
4558 Description: Cache open negative dentries on client when possible.
4559 Details : Guard negative dentries with UPDATE lock on parent dir, drop
4560 negative dentries on lock revocation.
4565 Description: Remounting a client read-only wasn't possible with a zconf mount
4566 Details : It wasn't possible to remount a client read-only with llmount.
4568 Severity : enhancement
4569 Description: Include MPICH 1.2.6 Lustre ADIO interface patch
4570 Details : In lustre/contrib/ or /usr/share/lustre in RPM a patch for
4571 MPICH is included to add Lustre-specific ADIO interfaces.
4572 This is based closely on the UFS ADIO layer and only differs
4573 in file creation, in order to allow the OST striping to be set.
4574 This is user-contributed code and not supported by CFS.
4579 Description: extended inode attributes (immutable, append-only) work improperly
4580 when 2.4 and 2.6 kernels are used on client/server or vice versa
4581 Details : Introduce kernel-independent values for these flags.
4583 Severity : enhancement
4586 Description: Allow fractional MB tunings for lustre in /proc/ filesystem.
4587 Details : Many of the /proc/ tunables can only be tuned at a megabyte
4588 granularity. Now, Fractional MB granularity is be supported,
4589 this is very useful for low memory system.
4591 Severity : enhancement
4593 Description: Getattr by fid
4594 Details : Getting a file attributes by its fid, obtaining UPDATE|LOOKUP
4595 locks, avoids extra getattr rpc requests to MDS, allows '/' to
4596 have locks and avoids getattr rpc requests for it on every stat.
4599 Frequency : Always, for filesystems larger than 2TB
4601 Description: ldiskfs crash at mount for filesystem larger than 2TB with mballoc
4602 Details : Kenrel kmalloc limits allocations to 128kB and this prevents
4603 filesystems larger than 2TB to be mounted with mballoc enabled.
4606 Frequency : Always, for 32-bit kernel without CONFIG_LBD and filesystem > 2TB
4608 Description: ldiskfs crash at mount for filesystem larger than 2TB with mballoc
4609 Details : If a 32-bit kernel is compiled without CONFIG_LBD enabled and a
4610 filesystems larger than 2TB is mounted then the kernel will
4611 silently corrupt the start of the filesystem. CONFIG_LBD is
4612 enabled for all CFS-supported kernels, but the possibility of
4613 this happening with a modified kernel config exists.
4615 Severity : enhancement
4617 Description: add client O_DIRECT support for 2.6 kernels
4618 Details : It is now possible to do O_DIRECT reads and writes to files
4619 in the Lustre client mountpoint on 2.6 kernel clients.
4621 Severity : enhancement
4623 Description: parallel glimpse, setattr, statfs, punch, destroy requests
4624 Details : Sends glimpse, setattr, statfs, punch, destroy requests to OSTs in
4625 parallel, not waiting for response from every OST before sending
4626 a rpc to the next OST.
4631 Description: setattr vs write race when updating file timestamps
4632 Details : Client processes that update a file timestamp into the past
4633 right after writing to the file (e.g. tar) it is possible that
4634 the updated file modification time can be reset to the current
4635 time due to a race between processing the setattr and write RPC.
4637 Severity : enhancement
4639 Description: Bring 'lfs find' closer in line with regular Linux find.
4640 Details : lfs find util supports -atime, -mtime, -ctime, -maxdepth, -print,
4641 -print0 options and obtains all the needed info through the lustre
4644 Severity : enhancement
4646 Description: support up to 1024 configured devices on one node
4647 Details : change obd_dev array from statically allocated to dynamically
4648 allocated structs as they are first used to reduce memory usage
4653 Description: Flush dirty partially truncated pages during truncate
4654 Details : Immediatelly flush partially truncated pages in filter_setattr,
4655 this way we completely avoid having any pages in page cache on OST
4656 and can retire ugly workarounds during writes to flush such pages.
4661 Description: i_sem vs transaction deadlock in mds_obd_destroy during unlink.
4662 Details : protect inode from truncation within vfs_unlink() context
4663 just take a reference before calling vfs_unlink() and release it
4664 when parent's i_sem is free.
4669 Description: last_id value checked outside lock on OST caused LASSERT failure
4670 Details : If there were multiple MDS->OST object precreate requests in
4671 flight, it was possible that the OST's last object id was checked
4672 outside a lock and incorrectly tripped an assertion. Move checks
4673 inside locks, and discard old precreate requests.
4676 Frequency : always, if extents are used on OSTs
4678 Description: index ei_leaf_hi (48-bit extension) is not zeroed in extent index
4679 Details : OSTs using the extents format would not zero the high 16 bits of
4680 the index physical block number. This is not a problem for any
4681 OST filesystems smaller than 16TB, and no kernels support ext3
4682 filesystems larger than 16TB yet. This is fixed in 1.4.7 (all
4683 new/modified files) and can be fixed for existing filesystems
4684 with e2fsprogs-1.39-cfs1.
4689 Description: import connection selection may be incorrect if timer wraps
4690 Details : Using a 32-bit jiffies timer with HZ=1000 may cause backup
4691 import connections to be ignored if the 32-bit jiffies counter
4692 wraps. Use a 64-bit jiffies counter.
4695 Frequency : very large clusters immediately after boot
4697 Description: LNET request buffers exhausted under heavy short-term load
4698 Details : If a large number of client requests are generated on a service
4699 that has previously never seen so many requests it is possible
4700 that the request buffer growth cannot keep up with the spike in
4701 demand. Instead of dropping incoming requests, they are held in
4702 the LND until the RPC service can accept more requests.
4705 Frequency : Sometimes during replay
4707 Description: Assertion failure in ll_local_open after replay.
4708 Details : If replay happened on an open request reply before we were able
4709 to set replay handler, reply will become not swabbed tripping the
4710 assertion in ll_local_open. Now we set the handler right after
4711 recognising of open request
4714 Frequency : very rare
4716 Description: kernel reports "badness in vsnprintf"
4717 Details : Reading from the "recovery_status" /proc file in small chunks
4718 may cause a negative length in lprocfs_obd_rd_recovery_status()
4719 call to vsnprintf() (which is otherwise harmless). Exit early
4720 if there is no more space in the output buffer.
4722 Severity : enhancement
4724 Description: clear OBD RPC statistics by writing to them
4725 Details : It is now possible to clear the OBD RPC statistics by writing
4726 to the "stats" file.
4731 Description: Client mtime is not the same on different clients after utimes
4732 Details : In some cases, the client was using the utimes() syscall on
4733 a file cached on another node. The clients now validate the
4734 ctime from the MDS + OSTs to determine which one is right.
4739 Description: Inability to activate failout mode
4740 Details : lconf script incorrectly assumed that in pythong string's numeric
4741 value is used in comparisons.
4744 Frequency : always with multiple stripes per file
4746 Description: Inefficient object allocation for mutli-stripe files
4747 Details : When selecting which OSTs to stripe files over, for files with
4748 a stripe count that divides evenly into the number of OSTs,
4749 the MDS is always picking the same starting OST for each file.
4750 Return the OST selection heuristic to the original design.
4755 Description: mount failures may take full timeout to return an error
4756 Details : Under some heavy load conditions it is possible that a
4757 failed mount can wait for the full obd_timeout interval,
4758 possibly several minutes, before reporting an error.
4759 Instead return an error as soon as the status is known.
4761 Frequency : quota enabled and large files being deleted
4763 Description: releasing more than 4GB of quota at once hangs OST
4764 Details : If a user deletes more than 4GB of files on a single OST it
4765 will cause the OST to spin in an infinite loop. Release
4766 quota in < 4GB chunks, or use a 64-bit value for 1.4.7.1+.
4771 Description: statfs data retrieved from /proc may be stale or zero
4772 Details : When reading per-device statfs data from /proc, in the
4773 {kbytes,files}_{total,free,avail} files, it may appear
4774 as zero or be out of date.
4777 Frequency : systems with MD RAID1 external journal devices
4779 Description: lconf's call to blkid is confused by RAID1 journal devices
4780 Details : Use the "blkid -l" flag to locate the MD RAID device instead
4781 of returning all block devices that match the journal UUID.
4784 Frequency : always, for aggregate stripe size over 4GB
4786 Description: assertion fails when trying to use 4GB stripe size
4787 Details : Use "setstripe" to set stripe size over 4GB will fail the kernel,
4788 complaining "ASSERTION(lsm->lsm_xfersize != 0)"
4791 Frequency : always on ppc64
4793 Description: the first write on an ext3 filesystem with mballoc got stuck
4794 Details : ext3_mb_generate_buddy() uses find_next_bit() which does not
4795 perform endianness conversion.
4797 ------------------------------------------------------------------------------
4799 02-14-2006 Cluster File Systems, Inc. <info@clusterfs.com>
4801 * WIRE PROTOCOL CHANGE. This version of Lustre networking WILL NOT
4802 INTEROPERATE with older versions automatically. Please read the
4803 user documentation before upgrading any part of a live system.
4804 * WARNING: Lustre networking configuration changes are required with
4805 this release. See https://bugzilla.clusterfs.com/show_bug.cgi?id=10052
4808 * Support for kernels:
4809 2.6.9-22.0.2.EL (RHEL 4)
4810 2.6.5-7.244 (SLES 9)
4811 2.6.12.6 vanilla (kernel.org)
4814 Severity : enhancement
4815 Bugzilla : 7981/8208
4816 Description: Introduced Lustre Networking (LNET)
4817 Details : LNET is new networking infrastructure for Lustre, it includes
4818 a reorganized network configuration mode (see the user
4819 documentation for full details) as well as support for routing
4820 between different network fabrics. Lustre Networking Devices
4821 (LNDs) for the supported network fabrics have also been
4822 created for this new infrastructure.
4824 Severity : enhancement
4825 Description: Introduced Access control lists
4826 Details : clients can set ACLs on files and directories in order to have
4827 more fine-grained permissions than the standard Unix UGO+RWX.
4828 The MDS must be started with the "-o acl" mount option.
4830 Severity : enhancement
4831 Description: Introduced filesystem quotas
4832 Details : Administrators may now establish per-user quotas on the
4835 Severity : enhancement
4837 Description: Configuration change for the XT3
4838 The PTLLND is now used to run Lustre over Portals on the XT3
4839 The configure option(s) --with-cray-portals are no longer used.
4840 Rather --with-portals=<path-to-portals-includes> is used to
4841 enable building on the XT3. In addition to enable XT3 specific
4842 features the option --enable-cray-xt3 must be used.
4847 Description: Running on many-way SMP OSTs can trigger oops in llcd_send()
4848 Details : A race between allocating a new llcd and re-getting the llcd_lock
4849 allowed another thread to grab newly-allocated llcd.
4851 Severity : enhancement
4853 Description: 2.6 OST async journal commit and locking fix to improve performance
4854 Details : The filter_direct_io()+filter_commitrw_write() journal commits for
4855 2.6 kernels are now async as they already were in 2.4 kernels so
4856 that they can commit concurrently with the network bulk transfer.
4857 For block-allocated files the filter allocation semaphore is held
4858 to avoid filesystem fragmentation during allocation. BKL lock
4859 removed for 2.6 xattr operations where it is no longer needed.
4864 Description: lconf incorrectly determined whether two IP networks could talk
4865 Details : In some more complicated routing and multiple-network
4866 configurations, lconf will avoid trying to make a network
4867 connection to a disjoint part of the IP space. It was doing the
4868 math incorrectly for one set of cases.
4873 Description: Fix for potential infinite loop processing records in an llog.
4874 Details : If an llog record is corrupted/zeroed, it is possible to loop
4875 forever in llog_process(). Validate the llog record length
4876 and skip the remainder of the block on error.
4879 Frequency : occasional (liblustre only)
4881 Description: liblustre could not open files whose last component is a symlink
4882 Details : sysio_path_walk() would incorrectly pass the open intent to
4883 intermediate path components.
4886 Frequency : rare (liblustre only with non-standard tuning)
4887 Bugzilla : 7201 (7350)
4888 Description: Tuning the MDC DLM LRU size to zero triggers client LASSERT
4889 Details : llu_lookup_finish_locks() tries to set lock data on a lock
4890 after it has been released, only do this for referenced locks
4892 Severity : enhancement
4894 Description: specifying an (invalid) directory default stripe_size of -1
4895 would reset the directory default striping
4896 Details : stripe_size -1 was used internally to signal directory stripe
4897 removal, now use "all default" to signal dir stripe removal
4898 as a directory striping of "all default" is not useful
4901 Frequency : common for large clusters running liblustre clients
4903 Description: doing an ls when liblustre clients are running is slow
4904 Details : sending a glimpse AST to a liblustre client waits for every AST
4905 to time out, as liblustre clients will not respond. Since they
4906 cannot cache data we refresh the OST lock LVB from disk instead.
4908 Severity : enhancement
4910 Description: doing an ls at the same time as file IO can be slow
4911 Details : enqueue and other "small" requests can be blocked behind many
4912 large IO requests. Create a new OST IO portal for non-IO
4913 requests so they can be processed faster.
4916 Frequency : rare (only HPUX clients mounting unsupported re-exported NFS vol)
4918 Description: an HPUX NFS client would get -EACCESS when ftruncate()ing a newly
4919 created file with mode 000
4920 Details : the Linux NFS server relies on an MDS_OPEN_OWNEROVERRIDE hack to
4921 allow an ftruncate() as a non-root user to a file with mode 000.
4922 Lustre now respects this flag to disable mode checks when
4923 truncating a file owned by the user
4926 Frequency : liblustre-only, when liblustre client dies unexpectedly or becomes
4929 Description: Revoking locks from clients that went dead or catatonic might take
4931 Details : New lock flags FL_CANCEL_ON_BLOCK used by liblustre makes
4932 cancellation of such locks instant on servers without waiting for
4933 any reply from clients. Clients drops these locks when cancel
4934 notification from server is received without replying.
4937 Frequency : liblustre-only, when liblustre client dies or becomes busy
4939 Description: Doing ls on Linux clients can take a long time with active
4941 Details : Liblustre client cannot handle ASTs in timely manner, so avoid
4942 granting such locks to it in the first place if possible. Locks
4943 are taken by proxy on the OST during the read or write and
4944 dropped immediately afterward. Add connect flags handling, do
4945 not grant locks to liblustre clients for glimpse ASTs.
4947 Severity : enhancement
4949 Description: Improve read-ahead algorithm to avoid excessive IO for random reads
4950 Details : Existing read-ahead algorithm is tuned for the case of streamlined
4951 sequential reads and behaves badly with applications doing random
4952 reads. Improve it by reading ahead at least read region, and
4953 avoiding excessive large RPC for small reads.
4955 Severity : enhancement
4957 Description: Creating more than 1000 files for a single job may cause a load
4958 imbalance on the OSTs if there are also a large number of OSTs.
4959 Details : qos_prep_create() uses an OST index reseed value that is an
4960 even multiple of the number of available OSTs so that if the
4961 reseed happens in the middle of the object allocation it will
4962 still utilize the OSTs as uniformly as possible.
4967 Description: OST or MDS may oops in ping_evictor_main()
4968 Details : ping_evictor_main() drops obd_dev_lock if deleting a stale export
4969 but doesn't restart at beginning of obd_exports_timed list
4972 Severity : enhancement
4974 Description: improve by-nid export eviction on the MDS and OST
4975 Details : allow multiple exports with the same NID to be evicted at one
4976 time without re-searching the exports list.
4979 Frequency : rare, only with supplementary groups enabled on SMP 2.6 kernels
4981 Description: MDS may oops in groups_free()
4982 Details : in rare race conditions a newly allocated group_info struct is
4983 freed again, and this can be NULL. The 2.4 compatibility code
4984 for groups_free() checked for a NULL pointer, but 2.6 did not.
4987 Frequency : common for liblustre clients doing little filesystem IO
4988 Bugzilla : 9352, 7313
4989 Description: server may evict liblustre clients accessing contended locks
4990 Details : if a client is granted a lock or receives a completion AST
4991 with a blocking AST already set it would not reply to the AST
4992 for LDLM_FL_CANCEL_ON_BLOCK locks. It now replies to such ASTs.
4995 Frequency : lfs setstripe, only systems with more than 160 OSTs
4997 Description: unable to set striping with a starting offset beyond OST 160
4998 Details : llapi_create_file() incorrectly limited the starting stripe
4999 index to the maximum single-file stripe count.
5002 Frequency : LDAP users only
5004 Description: lconf did not handle in-kernel recovery with LDAP properly
5005 Details : lconf/LustreDB get_refs() is searching the wrong namespace
5007 Severity : enhancement
5009 Description: bind OST threads to NUMA nodes to improve performance
5010 Details : all OST threads are uniformly bound to CPUs on a single NUMA
5011 node and do their allocations there to localize memory access
5013 Severity : enhancement
5015 Description: llmount can determine client NID directly from Myrinet (GM)
5016 Details : the client NID code from gmnalnid was moved directly into
5017 llmount, removing the need to use this or specifying the
5018 client NID explicitly when mounting GM clients with zeroconf
5021 Frequency : if client is started with down MDS
5023 Description: if client is started with down MDS mount hangs in ptlrpc_queue_wait
5024 Details : Having an LWI_INTR() wait event (interruptible, but no timeout)
5025 will wait indefinitely in ptlrpc_queue_wait->l_wait_event() after
5026 ptlrpc_import_delayed_req() because we didn't check if the
5027 request was interrupted, and we also didn't break out of the
5028 event loop if there was no timeout
5033 Description: data loss during non-page-aligned writes to a single file from
5034 both multiple nodes and multiple threads on one node at same time
5035 Details : updates to KMS and lsm weren't protected by common lock. Resulting
5036 inconsistency led to false short-reads, that were cached and later
5037 used by ->prepare_write() to fill in partially written page,
5038 leading to data loss.
5041 Frequency : always, if lconf --abort_recovery used
5043 Description: lconf --abort_recovery fails with 'Operation not supported'
5044 Details : lconf was attempting to abort recovery on the MDT device and not
5047 Severity : enhancement
5049 Description: remove cleanup logs
5050 Details : replace lconf-generated cleanup logs with lustre internal
5051 cleanup routines. Eliminates the need for client-cleanup and
5054 Severity : enhancement
5056 Description: add support for EAs (user and system) on lustre filesystems
5057 Details : it is now possible to store extended attributes in the Lustre
5058 client filesystem, and with the user_xattr mount option it
5059 is possible to allow users to store EAs on their files also
5061 Severity : enhancement
5063 Description: Add possibility (config option) to show minimal available OST free
5065 Details : When compiled with --enable-mindf configure option, statfs(2)
5066 (and so, df) will return least minimal free space available from
5067 all OSTs as amount of free space on FS, instead of summary of
5068 free spaces of all OSTs.
5070 Severity : enhancement
5072 Description: do not expand extent locks acquired on OST-side
5073 Details : Modify ldlm_extent_policy() to not expand local locks, acquired
5074 by server: they are not cached anyway.
5077 Frequency : when mmap is used/binaries executed from Lustre
5079 Description: Unmmap pages before throwing them away from read cache.
5080 Details : llap_shrink cache now attempts to unmap pages before discarding
5081 them (if unmapping failed - do not discard). SLES9 kernel has
5082 extra checks that trigger if this unmapping is not done first.
5087 Description: lconf didn't resolve symlinks before checking to see whether a
5088 given mountpoint was already in use
5091 Frequency : when migrating failover services
5092 Bugzilla : 6395, 9514
5093 Description: When migrating a subset of services from a node (e.g. failback
5094 from a failover service node) the remaining services would
5095 time out and evict clients.
5096 Details : lconf --force (implied by --failover) sets the global obd_timeout
5097 to 5 seconds in order to quickly disconnect, but this caused
5098 other RPCs to time out too quickly. Do not change the global
5099 obd_timeout for force cleanup, only set it for DISCONNECT RPCs.
5101 Severity : enhancement
5102 Frequency : if MDS is started with down OST
5103 Bugzilla : 9439,5706
5104 Description: Allow startup/shutdown of an MDS without depending on the
5105 availability of the OSTs.
5106 Details : Asynchronously call mds_lov_synchronize during MDS startup.
5107 Add appropriate locking and lov-osc refcounts for safe
5108 cleaning. Add osc abort_inflight calls in case the
5109 synchronize never started.
5112 Frequency : occasional (Cray XT3 only)
5114 Description: root not authorized to access files in CRAY_PORTALS environment
5115 Details : The client process capabilities were not honoured on the MDS in
5116 a CRAY_PORTALS/CRAY_XT3 environment. If the file had previously
5117 been accessed by an authorized user then root was able to access
5118 the file on the local client also. The root user capabilities
5119 are now allowed on the MDS, as this environment has secure UID.
5122 Frequency : occasional
5124 Description: ldiskfs "too long searching" message happens too often
5125 Details : A debugging message (otherwise harmless) prints too often on
5126 the OST console. This has been reduced to only happen when
5127 there are fragmentation problems on the filesystem.
5132 Description: Division by zero in statfs when all OSCs are inactive
5133 Details : lov_get_stripecnt() returns zero due to incorrect order of checks,
5134 lov_statfs divides by value returned by lov_get_stripecnt().
5138 Bugzilla : 9489, 3273
5139 Description: First write from each client to each OST was only 4kB in size,
5140 to initialize client writeback cache, which caused sub-optimal
5141 RPCs and poor layout on disk for the first written file.
5142 Details : Clients now request an initial cache grant at (re)connect time
5143 and so that they can start streaming writes to the cache right
5144 away and always do full-sized RPCs if there is enough data.
5145 If the OST is rebooted the client also re-establishes its grant
5146 so that client cached writes will be honoured under the grant.
5151 Description: Slow ls (and stat(2) syscall) on files residing on IO-loaded OSTs
5152 Details : Now I/O RPCs go to different portal number and (presumably) fast
5153 lock requests (and glimses) and other RPCs get their own service
5154 threads pool that should be able to service those RPCs
5157 Severity : enhancement
5159 Description: Ability to exchange lustre version between client and servers and
5160 issue warnings at client side if client is too old. Also for
5161 liblustre clients there is ability to refuse connection of too old
5163 Details : New 'version' field is added to connect data structure that is
5164 filled with version info. That info is later checked by server and
5168 Frequency : rare, liblustre only.
5169 Bugzilla : 9296, 9581
5170 Description: Two simultaneous writes from liblustre at offset within same page
5171 might proceed at the same time overwriting eachother with stale
5173 Details : I/O lock withing llu_file_prwv was released too early, before data
5174 actually was hitting the wire. Extended lock-holding time until
5175 server acknowledges receiving data.
5178 Frequency : extremely rare. Never observed in practice.
5180 Description: avoid generating lustre_handle cookie of 0.
5181 Details : class_handle_hash() generates handle cookies by incrementing
5182 global counter, and can hit 0 occasionaly (this is unlikely, but
5183 not impossible, because initial value of cookie counter is
5184 selected randonly). Value of 0 is used as a sentinel meaning
5185 "unassigned handle" --- avoid it. Also coalesce two critical
5186 sections in this function into one.
5188 Severity : enhancement
5190 Description: allow liblustre clients to delegate truncate locking to OST
5191 Details : To avoid overhead of locking, liblustre client instructs OST to
5192 take extent lock in ost_punch() on client's behalf. New connection
5193 flag is added to handle backward compatibility.
5195 Severity : enhancement
5196 Bugzilla : 4928, 7341, 9758
5197 Description: allow number of OST service threads to be specified
5198 Details : a module parameter allows the number of OST service threads
5199 to be specified via "options ost ost_num_threads={N}" in the
5200 OSS's /etc/modules.conf or /etc/modprobe.conf.
5204 Bugzilla : 6146, 9635, 9895
5205 Description: servers crash with bad pointer in target_handle_connect()
5206 Details : In rare cases when a client is reconnecting it was possible that
5207 the connection request was the last reference for that export.
5208 We would temporarily drop the export reference and get a new
5209 one, but this may have been the last reference and the export
5210 was just destroyed. Get new reference before dropping old one.
5212 Severity : enhancement
5213 Frequency : if client is started with failover MDS
5215 Description: Allow multiple MDS hostnames in the mount command
5216 Details : Try to read the configuration from all specified MDS
5217 hostnames during a client mount in case the "primary"
5220 Severity : enhancement
5222 Description: Stop sending data to evicted clients as soon as possible.
5223 Details : Check if the client we are about to send or are sending data to
5224 was evicted already. (Check is done every second of waiting,
5225 for which l_wait_event interface was extended to allow checking
5226 of exit condition at specified intervals).
5229 Frequency : rare, normally only when NFS exporting is done from client
5231 Description: 'bad disk LOV MAGIC: 0x00000000' error when chown'ing files
5233 Details : Make mds_get_md() recognise empty md case and set lmm size to 0.
5236 Frequency : always, if srand() is called before liblustre initialization
5238 Description: Liblustre uses system PRNG disturbing its usage by user application
5239 Details : Introduce internal to lustre fast and high-quality PRNG for
5240 lustre usage and make liblustre and some other places in generic
5241 lustre code to use it.
5243 Severity : enhancement
5244 Bugzilla : 9477, 9557, 9870
5245 Description: Verify that the MDS configuration logs are updated when xml is
5246 Details : Check if the .xml configuration logs are newer than the config
5247 logs stored on the MDS and report an error if this is the case.
5248 Request --write-conf, or allow starting with --old_conf.
5250 Severity : enhancement
5252 Description: Handle symlinks in the path when checking if Lustre is mounted.
5253 Details : Resolve intermediate symlinks when checking if a client has
5254 mounted a filesystem to avoid duplicate client mounts.
5259 Description: lconf can hit an error exception but still return success.
5260 Details : The lconf command catches the Command error exception at the top
5261 level script context and will exit with the associated exit
5262 status, but doesn't ensure that this exit status is non-zero.
5267 Description: failure of ptlrpc thread startup can cause oops
5268 Details : Starting a ptlrpc service thread can fail if there are a large
5269 number of threads or the server memory is very fragmented.
5270 Handle this without oopsing.
5273 Frequency : always, only if liblustre and non-default acceptor port was used
5275 Description: liblustre cannot connect to servers with non-default acceptor port
5276 Details : tcpnal_set_default_params() was not called and was therefore
5277 ignoring the environment varaible TCPNAL_PORT, as well as other
5278 TCPNAL_ environment variables
5283 Description: two objects could be created on the same OST for a single file
5284 Details : If an OST is down, in some cases it was possible to create two
5285 objects on a single OST for a single file. No problems other
5286 than potential performance impact and spurious error messages.
5290 Bugzilla : 5681, 9562
5291 Description: Client may oops in ll_unhash_aliases
5292 Details : Client dcache may become inconsistent in race condition.
5293 In some cases "getcwd" can fail if the current directory is
5299 Description: Inode refcounting problems in NFS export code
5300 Details : link_raw functions used to call d_instantiate without obtaining
5301 extra inode reference first.
5305 Bugzilla : 9942, 9903
5306 Description: Referencing freed requests leading to crash, memleaks with NFS.
5307 Details : We used to require that call to ll_revalidate_it was always
5308 followed by ll_lookup_it. Also with revalidate_special() it is
5309 possible to call ll_revalidate_it() twice for the same dentry
5310 even if first occurence returned success. This fix changes semantic
5311 between DISP_ENQ_COMPLETE disposition flag to mean there is extra
5312 reference on a request referred from the intent.
5313 ll_intent_release() then releases such a request.
5316 Frequency : rare, normally benchmark loads only
5318 Description: unlinked inodes were kept in memory on the client
5319 Details : If a client is repeatedly creating and unlinking files it
5320 can accumulate a lot of stale inodes in the inode slab cache.
5321 If there is no other client load running this can cause the
5322 client node to run out of memory. Instead flush old inodes
5323 from client cache that have the same inode number as a new inode.
5326 Frequency : SLES9 2.6.5 kernel and long filenames only
5327 Bugzilla : 9969, 10379
5328 Description: utime reports stale NFS file handle
5329 Details : SLES9 uses out-of-dentry names in some cases, which confused
5330 the lustre dentry revalidation. Change it to always use the
5334 Frequency : rare, unless heavy write-truncate concurrency is continuous
5335 Bugzilla : 4180, 6984, 7171, 9963, 9331
5336 Description: OST becomes very slow and/or deadlocked during object unlink
5337 Details : filter_destroy() was holding onto the parent directory lock
5338 while truncating+unlinking objects. For very large objects this
5339 may block other threads for a long time and slow overall OST
5340 responsiveness. It may also be possible to get a lock ordering
5341 deadlock in this case, or run out of journal credits because of
5342 the combined truncate+unlink. Solution is to do object truncate
5343 first in one transaction without parent lock, and then do the
5344 final unlink in a new transaction with the parent lock. This
5345 reduces the lock hold time dramatically.
5348 Frequency : rare, 2.4 kernels only
5350 Description: MDS or OST cleanup may trip kernel BUG when dropping kernel lock
5351 Details : mds_cleanup() and filter_cleanup() need to drop the kernel lock
5352 before unmounting their filesystem in order to avoid deadlock.
5353 The kernel_locked() function in 2.4 kernels only checks whether
5354 the kernel lock is held, not whether it is this process that is
5355 holding it as 2.6 kernels do.
5360 Description: MDS or OST may oops/LBUG if a client is connecting multiple times
5361 Details : The client ptlrpc code may be trying to reconnect to a down
5362 server before a previous connection attempt has timed out.
5363 Increase the reconnect interval to be longer than the connection
5364 timeout interval to avoid sending duplicate connections to
5368 Frequency : echo_client brw_test command
5370 Description: fix echo_client to work with OST preallocated code
5371 Details : OST preallocation code (5137) didn't take echo_client IO path
5372 into account: echo_client calls filter methods outside of any
5373 OST thread and, hence, there is no per-thread preallocated
5374 pages and buffers to use. Solution: hijack pga pages for IO. As
5375 a byproduct, this avoids unnecessary data copying.
5379 Bugzilla : 3555, 5962, 6025, 6155, 6296, 9574
5380 Description: Client can oops in mdc_commit_close() after open replay
5381 Details : It was possible for the MDS to return an open request with no
5382 transaction number in mds_finish_transno() if the client was
5383 evicted, but without actually returning an error. Clients
5384 would later try to replay that open and may trip an assertion
5385 Simplify the client close codepath, and always return an error
5386 from the MDS in case the open is not successful.
5389 Frequency : rare, 2.6 OSTs only
5391 Description: OST may deadlock under high load on fragmented files
5392 Details : If there was a heavy load and highly-fragmented OST filesystems
5393 it was possible to have all the OST threads deadlock waiting on
5394 allocation of biovecs, because the biovecs were not released
5395 until the entire RPC IO was completed. Instead, release biovecs
5396 as soon as they are complete to ensure forward IO progress.
5398 Severity : enhancement
5400 Description: Support for specifying external journal device at mount
5401 Details : If an OST or MDS device is formatted with an external journal
5402 device, this device major/minor is stored in the ext3 superblock
5403 and may not be valid for failover. Allow detecting and
5404 specifying the external journal at mount time.
5409 Description: Mounting an MDS with pending unlinked files may cause oops
5410 Details : target_finish_recovery() calls mds_postrecov() which returned
5411 the number of orphans unlinked. mds_lov_connect->mds_postsetup()
5412 considers this an error and immediately begins cleaning up the
5413 lov, just after starting the mds_lov process
5415 Severity : enhancement
5417 Description: Implement 'lfs df' to report actual free space on per-OST basis
5418 Details : Add sub-command 'df' on 'lfs' to report the disk space usage of
5419 MDS/OSDs. Usage: lfs df [-i][-h]. Command Options: '-i' to report
5420 usage of objects; '-h' to report in human readable format.
5422 ------------------------------------------------------------------------------
5424 08-26-2005 Cluster File Systems, Inc. <info@clusterfs.com>
5431 Description: Mounting an ldiskfs file system with mballoc may crash OST node.
5432 Details : ldiskfs mballoc code may reference an uninitialized buddy struct
5433 at startup during orphan unlinking. Instead, skip buddy update
5434 before setup, as it will be regenerated after recovery is complete.
5439 Description: If an OST is inactive, its locks might reference stale inodes.
5440 Details : lov_change_cbdata() must iterate over all namespaces, even if
5441 they are inactive to clear inode references from the lock.
5443 Severity : enhancement
5444 Frequency : occasional, if non-standard max_dirty_mb used
5446 Description: Client will block write RPCs if not enough grant
5447 Details : If a client has max_dirty_mb smaller than max_rpcs_in_flight,
5448 then the client will block writes while waiting for another RPC
5449 to complete instead of consuming its dirty limit. With change
5450 we get improved performance when max_dirty_mb is small.
5452 Severity : enhancement
5453 Bugzilla : 3389, 6253
5454 Description: Add support for supplementary groups on the MDS.
5455 Details : The MDS has an upcall /proc/fs/lustre/mds/{mds}/group_upcall
5456 (set to /usr/sbin/l_getgroups if enabled) which will do MDS-side
5457 lookups for user supplementary groups into a cache.
5461 Description: O_CREAT|O_EXCL open flags in liblustre always return -EEXIST
5462 Details : Make libsysio to not enforce O_EXCL by clearing the flag,
5463 for liblustre O_EXCL is enforced by MDS.
5467 Description: readdir never returns NULL in liblustre.
5468 Details : Corrected llu_iop_getdirentries logic, to return offset of next
5469 dentry in struct dirent.
5473 Frequency : liblustre only, depends on application IO pattern
5474 Description: liblustre clients evicted if not contacting servers
5475 Details : Don't put liblustre clients into the ping_evictor list, so
5476 they will not be evicted by the pinger ever.
5478 Severity : enhancement
5480 Description: Add ability to evict clients by NID from MDS.
5481 Details : By echoing "nid:$NID" string into
5482 /proc/fs/lustre/mds/.../evict_client client with nid that equals to
5483 $NID would be instantly evicted from this MDS and from all active
5484 OSTs connected to it.
5488 Description: Do not query file size twice, somewhat slowing stat(2) calls.
5489 Details : lookup_it_finish() used to query file size from OSTs that was not
5494 Description: service threads change working directory to that of init
5495 Details : Starting lustre service threads may pin the working directory
5496 of the parent thread, making that filesystem busy. Threads
5497 now change to the working directory of init to avoid this.
5501 Frequency : during shutdown only
5502 Description: shutdown with a failed MDS or OST can cause unmount to hang
5503 Details : Don't resend DISCONNECT messages in ptlrpc_disconnect_import()
5508 Frequency : 2.6 only
5509 Description: chmod/chown may include an extra supplementary group
5510 Details : ll{,u}_mdc_pack_op_data() does not properly initialize the
5511 supplementary group and if none is specified this is used.
5514 Bugzilla : 5479 (6816)
5516 Description: Racing open + rm can assert client in mdc_set_open_replay_data()
5517 Details : If lookup is in progress on a file that is unlinked we might try
5518 to revalidate the inode and fail in revalidate after lookup is
5519 complete and ll_file_open() enqueues the open again but
5520 it_open_error() was not checking DISP_OPEN_OPEN errors correctly.
5523 Frequency : always, if lconf --abort_recovery used
5525 Description: lconf --abort_recovery fails with 'Operation not supported'
5526 Details : lconf was attempting to abort recovery on the MDT device and not
5529 ------------------------------------------------------------------------------
5531 2005-08-08 Cluster File Systems, Inc. <info@clusterfs.com>
5536 Frequency : rare (only unsupported configurations with a node running as an
5538 Bugzilla : 6514, 5137
5539 Description: Mounting a Lustre file system on a node running as an OST could
5541 Details : OSTs now preallocates memory needed to write out data at
5542 startup, instead of when needed, to avoid having to
5543 allocate memory in possibly low memory situations.
5544 Specifically, if the file system is mounted on on OST,
5545 memory pressure could force it to try to write out data,
5546 which it needed to allocate memory to do. Due to the low
5547 memory, it would be unable to do so and the node would
5548 become unresponsive.
5550 Severity : enhancement
5552 Description: Addition of lconf --service command line option
5553 Details : lconf now accepts a '--service <arg>' option, which is
5554 shorthand for 'lconf --group <arg> --select <arg>=<hostname>'
5556 Severity : enhancement
5558 Description: Failover mode is now the default for OSTs.
5559 Details : By default, OSTs will now run in failover mode. To return to
5560 the old behaviour, add '--failout' to the lmc line for OSTs.
5562 Severity : enhancement
5564 Description: Health checks are now provided for MDS and OSTs
5565 Details : Additional detailed health check information on MSD and OSTs
5566 is now provided through the procfs health_check value.
5569 Frequency : occasional, depends on IO load
5571 Description: Disk fragmentation on the OSTs could eventually cause slowdowns
5572 after numerous create/delete cycles
5573 Details : The ext3 inode allocation policy would not allocate new inodes
5574 very well on the OSTs because there are no new directories
5575 being created. Instead we look for groups with free space if
5576 the parent directories are nearly full.
5581 Description: Network or server problems during mount may cause partially
5582 mounted clients instead of returning an error.
5583 Details : The config llog parsing code may overwrite the error return
5584 code during mount error handling, returning success instead
5590 Description: MDS can fail to allocate large reply buffers
5591 Details : After long uptimes the MDS can fail to allocate large reply
5592 buffers (e.g. zconf client mount config records) due to memory
5593 fragmentation or consumption by the buffer cache. Preallocate
5594 some large reply buffers so that these replies can be sent even
5595 under memory pressure.
5599 Frequency : rare (liblustre)
5600 Description: fsx running with liblustre complained that using truncate() to
5601 extend the file doesn't work. This patch corrects that issue.
5602 Details : This is the liblustre equivalent of the fix for bug 6196. Fixes
5603 ATTR_SIZE and lsm use in llu_setattr_raw.
5607 Frequency : rare, only 2.6 kernels
5608 Description: Unusual file access patterns on the MDS may result in inode
5609 data being lost in very rare circumstances.
5610 Details : Bad interaction between the ea-in-inode patch and the "no-read"
5611 code in the 2.6 kernel caused the inode and/or EA data not to
5612 be read from disk, causing single-file corruption.
5616 Frequency : rare, only 2.6 filesystems using extents
5617 Description: Heavy concurrent write and delete load may cause data corruption.
5618 Details : It was possible under high-load situations to have an extent
5619 metadata block in the block device cache from a just-unlinked
5620 file overwrite a newly-allocated data block. We now unmap any
5621 metadata buffers that alias just-allocated data blocks.
5625 Frequency : filesystems with default stripe_count larger than 77
5626 Description: lconf+mke2fs fail when formatting filesystem with > 77 stripes
5627 Details : lconf specifies an inode size of 4096 bytes when the default
5628 stripe_count is larger than 77. This conflicts with the default
5629 inode density of 1 per 4096 bytes. Allocate smaller inodes in
5630 this case to avoid pinning too much memory for large EAs.
5632 ------------------------------------------------------------------------------
5634 2005-07-07 Cluster File Systems, Inc. <info@clusterfs.com>
5639 Frequency : rare (extremely heavy IO load with hundreds of clients)
5641 Description: Client is evicted, gets IO error writing to file
5642 Details : lock ordering changes for bug 5492 reintroduced bug 3267 and
5643 caused clients to be evicted for AST timeouts. The fixes in
5644 bug 5192 mean we no longer need to have such short AST timeouts
5645 so ldlm_timeout has been increased.
5648 Frequency : occasional during --force or --failover shutdown under load
5649 Bugzilla : 5949, 4834
5650 Description: Server oops/LBUG if stopped with --force or --failover under load
5651 Details : a collection of import/export refcount and cleanup ordering
5652 issues fixed for safer force cleanup
5655 Frequency : only filesystems larger than 120 OSTs
5656 Bugzilla : 5990, 6223
5657 Description: lfs getstripe would oops on a very large filesystem
5658 Details : lov_getconfig used kfree on vmalloc'd memory
5661 Frequency : only filesystems exporting via NFS to Solaris 10 clients
5662 Bugzilla : 6242, 6243
5663 Description: reading from files that had been truncated to a non-zero size
5664 but never opened returned no data
5665 Details : ll_file_read() reads zeros from no-object files to EOF
5670 Description: A bug in MDS/OSS recovery could cause the OSS to fail an assertion
5671 Details : There's little harm in aborting MDS/OSS recovery and letting it
5672 try again, so I removed the LASSERT and return an error instead.
5674 Severity : enhancement
5676 Description: New debugging infrastructure for tracking down data corruption
5677 Details : The I/O checksum code was replaced to: (a) control it at runtime,
5678 (b) cover more of the client-side code path, and (c) try to narrow
5679 down where problems occurred
5683 Bugzilla : 3819, 4364, 4397, 6313
5684 Description: Racing close and eviction MDS could cause assertion in mds_close
5685 Details : It was possible to get multiple mfd references during close and
5686 client eviction, leading to one thread referencing a freed mfd.
5688 Severity: : enhancement
5689 Bugzilla : 3262, 6359
5690 Description: Attempts to reconnect to servers are now more aggressive.
5691 Details : This builds on the enhanced upcall-less recovery that was added
5692 in 1.4.2. When trying to reconnect to servers, clients will
5693 now try each server in the failover group every 10 seconds. By
5694 default, clients would previously try one server every 25 seconds.
5699 Description: After recovery, certain operations trigger a failed
5700 assertion on a client.
5701 Details : Failing over an mds, using lconf -d --failover, while a
5702 client was doing a readdir() call would cause the client to
5703 LBUG after recovery completed and the readdir() was resent.
5705 Severity : enhancement
5707 Description: Default groups are now added by lconf
5708 Details : You can now run lconf --group <servicename> without having to
5709 manually add groups with lmc.
5712 Frequency : occasional
5714 Description: Nodes with an elan id of 0 trigger a failed assertion
5717 Frequency : always when accessing e.g. tty/console device nodes
5719 Description: tty and some other devices nodes cannot be used on lustre
5720 Details : file's private_data field is used by device data and lustre
5721 values in there got lost. New field was added to struct file to
5722 store fs-specific private data.
5725 Frequency : when exporting Lustre via NFS
5727 Description: NFSD failed occasionally when looking up a path component
5728 Details : NFSD is looking up ".." which was broken in ext3 directories
5729 that had grown large enough to become hashed.
5732 Frequency : Clusters with multiple interfaces not on the same subnet
5734 Description: Nodes will repeatedly try to reconnect to an interface which it
5735 cannot reach and report an error to the log.
5736 Details : Extra peer list entries will be created by lconf with some peers
5737 unreachable. lconf now validates the peer before adding it.
5740 Frequency : Only if a default stripe is set on the filesystem root.
5742 Description: Setting a default stripe on the filesystem root prevented the
5743 filesystem from being remounted.
5744 Details : The client was sending extra request flags in the root getattr
5745 request and did not allocate a reply buffer for the dir EA.
5748 Frequency : occasional, higher if lots of files are accessed by one client
5749 Bugzilla : 6159, 6097
5750 Description: Client trips assertion regarding lsm mismatch/magic
5751 Details : While revalidating inodes the VFS looks up inodes with ifind()
5752 and in rare cases can find an inode that is being freed.
5753 The ll_test_inode() code will free the lsm during ifind()
5754 when it finds an existing inode and then the VFS later attaches
5755 this free lsm to a new inode.
5759 Bugzilla : 6422, 7030
5760 Description: MDS deadlock between mkdir and client eviction
5761 Details : Creating a new file via mkdir or mknod (starting a transaction
5762 and getting the ns lock) can deadlock with client eviction
5763 (gets ns lock and trying to finish a synchronous transaction).
5766 Frequency : occasional
5767 Description: While starting a server, the fsfilt_ext3 module could not be
5769 Details : CFS's improved ext3 filesystem is named ldiskfs for 2.6
5770 kernels. Previously, lconf would still use the ext3 name
5771 when trying to load modules. Now, it will correctly use
5772 ext3 on 2.4 and ldiskfs on 2.6.
5774 Severity : enhancement
5775 Description: The default stripe count has been changed to 1
5776 Details : The interpretation of the default stripe count (0, to lfs
5777 or lmc) has been changed to mean striping across a single
5778 OST, rather than all available. For general usage we have
5779 found a stripe count of 1 or 2 works best.
5781 Severity : enhancement
5782 Description: Add support for compiling against Cray portals.
5783 Details : Conditional compiling for some areas that are different
5787 Frequency : occasional
5788 Bugzilla : 6409, 6834
5789 Description: Creating files with an explicit stripe count may lead to
5790 a failed assertion on the MDS
5791 Details : If some OSTs are full or unavailable, creating files may
5792 trigger a failed assertion on the MDS. Now, Lustre will
5793 try to use other servers or return an error to the
5797 Frequency : occasional
5799 Description: Multiple concurrent overlapping read+write on multiple SMP nodes
5800 caused lock timeout during readahead (since 1.4.2).
5801 Details : Processes doing readahead might match a lock that hasn't been
5802 granted yet if there are overlapping and conflicting lock
5803 requests. The readahead process waits on ungranted lock
5804 (original lock is CBPENDING), while OST waits for that process
5805 to cancel CBPENDING read lock and eventually evicts client.
5807 Severity : enhancement
5809 Description: Initial enabling of flock support for clients
5810 Details : Implements fcntl advisory locking and file status functions.
5811 This feature is provided as an optional mount flag (default
5812 off), and is NOT CURRENTLY SUPPORTED. Not all types of record
5813 locking are implemented yet, and those that are are not guaranteed
5814 to be completely correct in production environments.
5815 mount -t lustre -o [flock|noflock] ...
5818 Frequency : occasional
5820 Description: OSTs running 2.4 kernels but with extents enabled might trip an
5821 assertion in the ext3 JBD (journaling) layer.
5822 Details : The b_committed_data struct is protected by the big kernel lock
5823 in 2.4 kernels, serializing journal_commit_transaction() and
5824 ext3_get_block_handle->ext3_new_block->find_next_usable_block()
5825 access to this struct. In 2.6 kernels there is finer grained
5826 locking to improve SMP performance of the JBD layer.
5830 Description: Changes the "SCSI I/O Stats" kernel patch to default to "enabled"
5835 Description: merge and cleanup kernel patches.
5838 -----------------------------------------------------------------------------
5840 2005-05-05 Cluster File Systems, Inc. <info@clusterfs.com>
5842 NOTE: Lustre 1.4.2 uses an incompatible network protocol than previous
5843 versions of Lustre. Please update all servers and clients to
5844 version 1.4.2 or later at the same time. You must also run
5845 "lconf --write-conf {config}.xml" on the MDS while it is stopped
5846 to update the configuration logs.
5848 - fix for HPUX NFS client breakage when NFS exporting Lustre (5781)
5849 - mdc_enqueue does not need max_mds_easize request buffer on send (5707)
5850 - swab llog records of type '0' so we get proper header size/idx (5861)
5851 - send llog cancel req to DLM cancel portal instead of cb portal (5515)
5852 - fix rename of one directory over another leaking an inode (5953)
5853 - avoid SetPageDirty on 2.6 (5981)
5854 - don't re-add just-being-destroyed locks to the waiting list (5653)
5855 - when creating new directories, inherit the parent's custom
5856 striping settings if present parent (3048)
5857 - flush buffers from cache before direct IO in 2.6 obdfilter (4982)
5858 - don't hold i_size_sem in ll_nopage() and ll_ap_refresh_count (6077)
5859 - don't hold client locks on temporary worklist from l_lru (5666)
5860 - handle IO errors in 2.6 obdfilter bio completion routine (6046)
5861 - automatically evict dead clients (5921)
5862 - Update file size properly in create+truncate+fstat case (6196)
5863 - Do not unhash mountpoint dentries, do not allow removal of
5865 - Avoid lock ordering deadlock issue with write/truncate (6203,5654)
5866 - reserve enough journal credits in fsfilt_start_log for setattr (4554)
5867 - ldlm_enqueue freed-export error path would always LBUG (6149,6184)
5868 - don't reference lr_lvb_data until after we hold lr_lvb_sem (6170)
5869 - don't overwrite last_rcvd if there is a *_client_add() error (6086)
5870 - Correctly handle reads of files with no objects (6243)
5871 - lctl recover will also mark a device active if deactivate used (5933)
5873 - by default create 1 inode per 4kB space on MDS, per 16kB on OSTs
5874 - allow --write-conf on an MDS with different nettype than client (5619)
5875 - don't write config llogs to MDS for mounts not from that MDS (5617)
5876 - lconf should create multiple TCP connections from a client (5201)
5877 - init scripts are now turned off by default; run chkconfig --on
5878 lustre and chkconfig --on lustrefs to use them
5879 - upcalls are no longer needed for clients to recover to failover
5881 - add --abort-recovery option to lconf to abort recovery on device
5883 - add support for an arbitrary number of OSTs (3026)
5884 - Quota support protocol changes.
5885 - forward compatibility changes to wire structs (6007)
5886 - rmmod NALs that might be loaded because of /etc/modules.conf (6133)
5887 - support for mountfsoptions and clientoptions to the Lustre LDAP (5873)
5888 - improved "lustre status" script
5889 - initialize blocksize for non-regular files (6062)
5890 - added --disable-server and --disable-client configure options (5782)
5891 - introduce a lookup cache for lconf to avoid repeated DB scans (6204)
5892 - Vanilla 2.4.29 support
5893 - increase maximum number of obd devices to 520 (6242)
5894 - remove the tcp-zero-copy patch from the suse-2.4 series (5902)
5895 - Quadrics Elan drivers are now included for the RHEL 3 2.4.21 and
5896 SLES 9 2.6.5 kernels
5897 - limit stripes per file to 160 (the maximum EA size) (6093)
5899 2005-03-22 Cluster File Systems, Inc. <info@clusterfs.com>
5902 - don't LASSERT in ll_release on NULL lld with NFS export (4655, 5760)
5903 - hold NS lock when calling handle_ast_error->del_waiting_lock (5746)
5904 - fix setattr mtime regression from lovcleanup merge (4829, 5669)
5905 - workaround for 2.6 crash in ll_unhash_aliases (5687, 5210)
5906 - small ext3 extents cleanups and fixes (5733)
5907 - improved mballoc code, several small races and bugs fixed (5733, 5638)
5908 - kernel version 43 - fix remove_suid bugs in both 2.4 and 2.6 (5695)
5909 - avoid needless client->OST connect, fix handle mismatch (5317)
5910 - fix DLM error path that led to out-of-sync client, long delays (5779)
5911 - support common vfs-enforced mount options (nodev,nosuid,noexec) (5637)
5912 - fix several locking issues related to i_size (5492,5624,5654,5672)
5913 - don't move pending lock onto export if it is already evicted (5683)
5914 - fix kernel oops when creating .foo in unlinked directory (5548)
5915 - fix deadlock in obdfilter statistics vs. object create (5811)
5916 - use time_{before,after} to avoid timer jiffies wrap (5882)
5917 - shutdown --force/--failover stability (3607,3651,4797,5203,4834)
5918 - Do not leak request if server was not able to process it (5154)
5919 - If mds_open unable to find parent dir, make that negative lookup(5154)
5920 - don't create new directories with extent-mapping (5909, 5936)
5922 - fix lustre/lustrefs init scripts for SuSE (patch from Scali, 5702)
5923 - don't hold the pinger_sem in ptlrpc_pinger_sending_on_import
5924 - change obd_increase_kms to obd_adjust_kms (up or down) (5654)
5925 - lconf, lmc search both /usr/lib and /usr/lib64 for Python libs (5800)
5926 - support for RHEL4 kernel on i686 (5773)
5927 - provide error messages when incompatible logs are encountered (5898)
5929 2005-02-18 Cluster File Systems, Inc. <info@clusterfs.com>
5930 * version 1.4.0.10 (1.4.1 release candidate 1)
5932 - don't keep a lock reference when lock is not granted (4238)
5933 - unsafe list practices (rarely) led to infinite eviction loop (4908)
5934 - add per-fs limit of Lustre pages in page cache, avoid OOM (4699)
5935 - drop import inflight refcount on signal_completed_replay error (5255)
5936 - unlock page after async write error during send (3677)
5937 - handle missing objects in filter_preprw_read properly (5265)
5938 - no transno return for symlink open, don't save no-trasno open (3440)
5939 - don't try to complete elan receive that already failed (4012)
5940 - free RPC server reply state on error (5406)
5941 - clean up thread from ptlrpc_start_thread() on error (5160)
5942 - readahead could read extra page into cache that wasn't ejected (5388)
5943 - prevent races in class_attach/setup/cleanup/detach (5260)
5944 - don't dereference de->d_inode after l_dput of de (5458)
5945 - use "int" for stripe value returned from lock_to_stripe (5544)
5946 - mballoc allocation and error-checking fixes in 2.6 (5504)
5947 - block device patches to fix I/O request sizes in 2.6 (5482)
5948 - look up hostnames for IB nals (5602)
5949 - 2.6 changed lock ordering of 2 semaphores, caused deadlock (5654)
5950 - don't start multiple acceptors for the same port (5277)
5951 - fix incorrect LASSERT in mds_getattr_name (5635)
5952 - export a proc file for general "ping" checking (5628)
5953 - fix "lfs check" to not block when the MDS is down (5628)
5955 - service request history (4965)
5956 - put {ll,lov,osc}_async_page structs in a single slab (4699)
5957 - create an "evict_client" /proc entry on OSTs, like the MDS has
5958 - fix mount usage message, return errors per mount(8) (5168)
5959 - change grep [] to grep "[]" in tests so they work in more UMLs
5960 - fix ppc64/x86_64 spec to use %{_libdir} instead of /usr/lib (5389)
5961 - remove ancient LOV_MAGIC_V0 EA support (5047)
5962 - add "disk I/Os in flight" and "I/O req time" stats in obdfilter
5963 - align r/w RPCs to PTLRPC_MAX_BRW_SIZE boundary for performance (3451)
5964 - allow readahead allocations to fail when low on memory (5383)
5965 - mmap locking landed again, after considerable improvement (2828)
5966 - add get_hostaddr() to lustreDB.py for LDAP support (5459)
5968 2004-11-23 Cluster File Systems, Inc. <info@clusterfs.com>
5971 - send OST transaction number in read/write reply to free req (4966)
5972 - don't ASSERT in ptl_send_rpc() if we run out of memory (5119)
5973 - lock /proc/sys/portals/routes internal state, avoiding oops (4827)
5974 - the watchdog thread now runs as interruptible (5246)
5975 - flock/lockf fixes (but it's still disabled, pending 5135)
5976 - don't use EXT3 constants in llite code (5094)
5977 - memory shortage at startup could cause assertion (5176)
5979 - reorganization of lov code
5980 - single portals codebase
5982 - add extents/mballoc support (5025)
5983 - direct I/O reads in the obdfilter (4048)
5984 - kernel patches from LNXI for 2.6 (bluesmoke, perfctr, mtd, kexec)
5986 tbd Cluster File Systems, Inc. <info@clusterfs.com>
5989 - send OST transaction number in read/write reply to free req (4966)
5990 - don't ASSERT in ptl_send_rpc() if we run out of memory (5119)
5991 - lock /proc/sys/portals/routes internal state, avoiding oops (4827)
5992 - the watchdog thread now runs as interruptible (5246)
5993 - handle missing objects in filter_preprw_read properly (5265)
5994 - unsafe list practices (rarely) led to infinite eviction loop (4908)
5995 - drop import inflight refcount on signal_completed_replay error (5255)
5996 - unlock page after async write error during send (3677)
5997 - return original error code on reconstructed replies (3761)
5998 - no transno return for symlink open, don't save no-trasno open (3440)
6000 - add pid to ldlm debugging output (4922)
6001 - bump the watchdog timeouts -- we can't handle 30sec yet
6002 - extra debugging for orphan dentry/inode bug (5259)
6004 2004-11-16 Cluster File Systems, Inc. <info@clusterfs.com>
6007 - fix TCP_NODELAY bug, which caused extreme perf regression (5134)
6008 - allocate qswnal tx descriptors singly to avoid fragmentation (4504)
6009 - don't LBUG on obdo_alloc() failure, use OBD_SLAB_ALLOC() (4800)
6010 - fix NULL dereference in /proc/sys/portals/routes (4827)
6011 - allow failed mdc_close() operations to be interrupted (4561)
6012 - stop precreate on OST before MDS would time out on it (4778)
6013 - don't send partial-page writes before EOF from client (4410)
6014 - discard client grant for sub-page writes on large-page clients (4520)
6015 - don't free dentries not owned by NFS code, check generation (4806)
6016 - fix lsm leak if mds_create_objects() fails (4801)
6017 - limit debug_daemon file size, always print CERROR messages (4789)
6018 - use transno after validating reply (3892)
6019 - process timed out requests if import state changes (3754)
6020 - update mtime on OST during writes, return in glimpse (4829)
6021 - add mkfsoptions to LDAP (4679)
6022 - use ->max_readahead method instead of zapping global ra (5039)
6023 - don't interrupt __l_wait_event() during strace
6025 - add software watchdogs to catch hung threads quickly (4941)
6026 - make lustrefs init script start after nfs is mounted
6027 - fix CWARN/ERROR duplication (4930)
6028 - return async write errors to application if possible (2248)
6029 - add /proc/sys/portal/memused (bytes allocated by PORTALS_ALLOC)
6030 - print NAL number in %x format (4645)
6031 - update barely-supported suse-2.4.21-171 series (4842)
6032 - support for sles 9 %post scripts
6033 - support for building 2.6 kernel-source packages
6034 - support for sles km_* packages
6036 2004-10-07 Cluster File Systems, Inc. <info@clusterfs.com>
6039 - ignore -ENOENT errors in osc_destroy (3639)
6040 - notify osc create thread that OSC is being cleaned up (4600)
6041 - add nettype argument for llmount in #5d in conf-sanity.sh (3936)
6042 - reconstruct ost_handle() like mds_handle() (4657)
6043 - create a new thread to do import eviction to avoid deadlock (3969)
6044 - let lconf resolve symlinked-to devices (4629)
6045 - don't unlink "objects" from directory with default EA (4554)
6046 - hold socknal file ref over connect in case target is down (4394)
6047 - allow more than 32000 subdirectories in a single directory (3244)
6048 - fix blocks count for O_DIRECT writes (3751)
6049 - OST returns ENOSPC from object create when no space left (4539)
6050 - don't send truncate RPC if file size isn't changing (4410)
6051 - limit OSC precreate to 1/2 of value OST considers bogus (4778)
6052 - bind to privileged port in socknal and tcpnal (3689)
6054 - rate limit CERROR/CWARN console message to avoid overload (4519)
6055 - GETFILEINFO dir ioctl returns LOV EA + MDS stat in 1 call (3327)
6056 - basic mmap support (3918)
6057 - kernel patch series update from b1_4 (4711)
6059 2004-09-16 Cluster File Systems, Inc. <info@clusterfs.com>
6062 - avoid crash during MDS cleanup with OST shut down (2775)
6063 - fix loi_list_lock/oig_lock inversion on interrupted IO (4136)
6064 - don't use bad inodes on the MDS (3744)
6065 - dynamic object preallocation to improve recovery speed (4236)
6066 - don't hold spinlock over lock dumping or change debug flags (4401)
6067 - don't zero obd_dev when it is force cleaned (3651)
6068 - print grants to console if they go negative (4431)
6069 - "lctl deactivate" will stop automatic recovery attempts (3406)
6070 - look for existing locks in ldlm_handle_enqueue() (3764)
6071 - don't resolve lock handle twice in recovery avoiding race (4401)
6072 - revalidate should check working dir is a directory (4134)
6074 - don't always mark "slow" obdfilter messages as errors (4418)
6076 2004-08-24 Cluster File Systems, Inc. <info@clusterfs.com>
6079 - don't close LustreDB during write_conf until it is done (3860)
6080 - fix typo in lconf for_each_profile (3821)
6081 - allow dumping logs from multiple threads at one time (3820)
6082 - don't allow multiple threads in OSC recovery (3812)
6083 - fix debug_size parameters (3864)
6084 - fix mds_postrecov to initialize import for llog ctxt (3121)
6085 - replace config semaphore with spinlock (3306)
6086 - be sure to send a reply for a CANCEL rpc with bad export (3863)
6087 - don't allow enqueue to complete on a destroyed export (3822)
6088 - down write_lock before checking llog header bitmap (3825)
6089 - recover from lock replay timeout (3764)
6090 - up llog sem before sending rpc (3652)
6091 - reduce ns lock hold times when setting kms (3267)
6092 - change a dlm LBUG to LASSERTF, to maybe learn something (4228)
6093 - fix NULL deref and obd_dev leak on setup error (3312)
6094 - replace some LBUG about llog ops with error handling (3841)
6095 - don't match INVALID dentries from d_lookup and spin (3784)
6096 - hold dcache_lock while marking dentries INVALID and hashing (4255)
6097 - fix invalid assertion in ptlrpc_set_wait (3880)
6099 - add libwrap support for the TCP acceptor (3996)
6100 - add /proc/sys/portals/routes for non-root route listing (3994)
6101 - allow setting MDS UUID in .xml (2580)
6102 - print the stack of a process that LBUGs (4228)
6104 2004-07-14 Cluster File Systems, Inc. <info@clusterfs.com>
6107 - don't cleanup request in ll_file_open() on failed MDS open (3430)
6108 - make sure to unset replay flag from failed open requests (3440)
6109 - if default stripe count is 0, use OST count for inode size (3636)
6110 - update parent mtime/ctime on client for create/unlink (2611)
6111 - drop dentry ref in ext3_add_link from open_connect_dentry (3266)
6112 - free recovery state on server during a forced cleanup (3571)
6113 - unregister_reply for resent reqs (3063)
6114 - loop back devices mounting and status check on 2.6 (3563)
6115 - fix resource-creation race that can provoke i_size == 0 (3513)
6116 - don't try to use bad inodes returned from MDS/OST fs lookup (3688)
6117 - more debugging for page-accounting assertion (3746)
6118 - return -ENOENT instead of asserting if ost getattr+unlink race (3558)
6119 - avoid deadlock after precreation failure (3758)
6120 - fix race and lock order deadlock in orphan handling (3450, 3750)
6121 - add validity checks when grabbing inodes from l_ast_data (3599)
6123 - add /proc/.../recovery_status to obdfilter (3428)
6124 - lightweight CDEBUG infrastructure, debug daemon (3668)
6125 - change default OSC RPC parameters to be better on small clusters
6126 - turn off OST read cache for files smaller than 32MB
6127 - install man pages and include them in rpms (3100)
6128 - add new init script for (un)mounting lustre filesystems (2593)
6129 - run chkconfig in %post for init scripts (3701)
6130 - drop scimac NAL (unmaintained)
6132 2004-06-17 Cluster File Systems, Inc. <info@clusterfs.com>
6135 - clean kiobufs before and after use (3485)
6136 - strip trailing '/'s before comparing paths with /proc/mounts (3486)
6137 - remove assertions to work around "in-flight rpcs" recovery bug (3063)
6138 - change init script to fail more clearly if not run as root (1528)
6139 - allow clients to reconnect during replay (1742)
6140 - fix ns_lock/i_sem lock ordering deadlock for kms update (3477)
6141 - don't do DNS lookups on NIDs too small for IP addresses (3442)
6142 - re-awaken ptlrpcd if new requests arrive during check_set (3554)
6143 - fix cond_resched (3554)
6144 - only evict unfinished clients after recovery (3515)
6145 - allow bulk resend, prevent data loss (3570)
6146 - dynamic ptlrpc request buffer allocation (2102)
6147 - don't allow unlinking open directory if it isn't empty (2904)
6148 - set MDS/OST threads to umask 0 to not clobber client modes (3359)
6149 - remove extraneous obd dereference causing LASSERT failure (3334)
6150 - don't use get_cycles() when creating temp. files on the mds (3156)
6151 - hold i_sem when setting i_size in ll_extent_lock() (3564)
6152 - handle EEXIST for set-stripe, set proper directory name (3336)
6154 - servers can dump a log evicting a client - lustre.dump_on_timeout=1
6155 - fix ksocknal_fmb_callback() error messages (2918)
6157 2004-05-27 Cluster File Systems, Inc. <info@clusterfs.com>
6160 - don't copy lvb into (possibly NULL) reply on error (2983)
6161 - don't deref dentry after dput, don't free lvb on error (2922)
6162 - use the kms to determine writeback rpc length (2947)
6163 - increment oti_logcookies when osc is inactive (2948)
6164 - update client's i_blocks count via lvb messages (2543)
6165 - handle intent open/close of special files properly (1557)
6166 - mount MDS with errors=remount-ro, like obdfilter (2009)
6167 - initialize lock handle to avoid ASSERT on error cleanup (3057)
6168 - don't use cancelling-locks' kms values (2947)
6169 - use highest lock extent for kms, not last one (2925)
6170 - don't dereference ERR_PTR() dentry in error handling path (3107)
6171 - fix thread race in portals_debug_dumplog() (3122)
6172 - create lprocfs device entries at setup instead of at attach (1519)
6173 - common AST error handler, don't evict client on completion race (3145)
6174 - zero nameidata in detach_mnt in 2.6 (3118)
6175 - verify d_inode after revalidate_special is valid in 2.6 (3116)
6176 - use lustre_put_super() to handle zconf unmounts in 2.6 (3064)
6177 - initialize RPC timeout timer earlier for 2.6 (3219)
6178 - don't dereference NULL reply buffer if mdc_close was never sent (2410)
6179 - print nal/nid for unknown nid (3258)
6180 - additional checks for oscc recovery before doing precreate (3284)
6181 - fix ll_extent_lock() error return code for 64-bit systems (3043)
6182 - don't crash in mdc_close for bad permissions on open (3285)
6183 - zero i_rdev for non-device files (3147)
6184 - clear page->private before handing to FS, better assertion (3119)
6185 - tune the read pipeline (3236)
6186 - fix incorrect decref of invalidated dentry (2350)
6187 - provide read-ahead stats and refine rpc in flight stats (3328)
6188 - don't hold journal transaction open across create RPC (3313)
6189 - update atime on MDS at close time (3265)
6190 - close LDAP connection when recovering to avoid server load (3315)
6191 - update iopen-2.6 patch with fixes from 2399,2517,2904 (3301)
6192 - don't leak open file on MDS after open resend (3325)
6193 - serialize filter_precreate and filter_destroy_precreated (3329)
6194 - loop device shouldn't call sync_dev() for nul device (3092)
6195 - clear page cache after eviction (2766)
6196 - resynchronize MDS->OST in background (2824)
6197 - refuse to mount the same filesystem twice on same mountpoint (3394)
6198 - allow llmount to create routes for mounting behind routers (3320)
6199 - push lock cancellation to blocking thread for glimpse ASTs (3409)
6200 - don't call osc_set_data_with_check() for TEST_LOCK matches (3159)
6201 - fix rare problem with rename on htree directories (3417)
6203 - allow default OST striping configuration per directory (1414)
6204 - fix compilation for qswnal for 2.6 kernels (3125)
6205 - increase maximum number of MDS request buffers for large systems
6206 - change liblustreapi to be useful for external progs like lfsck (3098)
6207 - increase local configuration timeout for slow disks (3353)
6208 - allow configuring ldlm AST timeout - lustre.ldlm_timeout=<seconds>
6210 2004-03-22 Cluster File Systems, Inc. <info@clusterfs.com>
6213 - fixes for glimpse AST timeouts / incorrectly 0-sized files (2818)
6214 - don't overwrite extent policy data in reply if lock was blocked (2901)
6215 - drop filter export grants atomically with removal from device (2663)
6216 - del obd_self_export from work_list in class_disconnect_exports (2908)
6217 - don't LBUG if MDS recovery times out during orphan cleanup (2530)
6218 - swab reply message in mdc_close, other PPC fixes (2464)
6219 - fix destroying of named logs (2325)
6220 - overwrite old logs when running lconf --write_conf (2264)
6221 - bump LLOG_CHUNKSIZE to 8k to allow for larger clusters (2306)
6222 - fix race in target_handle_connect (2898)
6223 - mds_reint_create() should take same inode create lock (2926)
6224 - correct journal credits calculated for CANCEL_UNLINK_LOG (2931)
6225 - don't close files for self_export to avoid uninitialized obd (2936)
6226 - allow MDS with the same name as client node (2939)
6227 - hold dentry reference for closed log files for unlink (2325)
6228 - reserve space for all logs during transactions (2059)
6229 - don't evict page beyond end of stripe extent (2925)
6230 - don't oops on a deleted current working directory (2399)
6231 - handle hard links to targets without a parent properly (2517)
6232 - don't dereference NULL lock when racing during eviction (2867)
6233 - don't grow lock extents when lots of conflicting locks (2919)
6235 2004-03-04 Cluster File Systems, Inc. <info@clusterfs.com>
6238 - account for cache space usage on clients to avoid data loss (974)
6239 - lfsck support in lustre kernel code (2349)
6240 - reduce journal credits needed for BRW writes (2370)
6241 - orphan handling to avoid losing space on client/server crashes
6242 - ptlrpcd can be blocked, stopping ALL progress (2477)
6243 - use lock value blocks to assist in proper KMS, faster stat (1021)
6244 - takes i_sem instead of DLM locks internally on obdfilter (2720)
6245 - recovery for initial connections (2355)
6246 - fixes for mds_cleanup_orphans (1934)
6247 - abort_recovery crashes MDS in b_eq (mds_unlink_orphan) (2584)
6248 - block all file creations until orphan recovery completes (1901)
6249 - client remove rq_connection from request struct (2423)
6250 - conf-sanity test_5, proper cleanup in umount log not availale (2640)
6251 - recovery timer race (2670)
6252 - mdc_close recovey bug (2532)
6253 - ptlrpc cleanup bug (2710)
6254 - mds timeout on local locks (2588)
6255 - namespace lock held during RPCs (2431)
6256 - handle interrupted sync write properly (2503)
6257 - don't try to handle a message that hasn't been replied to (2699)
6258 - client assert failure during cleanup after abort recovery (2701)
6259 - leak mdc device after failed mount (2712)
6260 - ptlrpc_check_set allows timedout requests to complete (2714)
6261 - wait for inflight reqs when ptlrpcd finishes (2710)
6262 - make sure unregistered services are removed from the srv_list
6263 - reset bulk XID's when resending them (caught by 1138 test)
6264 - unregister_bulk after timeout
6265 - fix lconf error (2694)
6266 - handle write after unfinished setstripe, stripe-only getstripe (2388)
6267 - readahead locks pages, leaves pending causing memory pressure (2673)
6268 - increase OST request buffers to 4096 on large machines (2729)
6269 - fix up permission of existing directories in simple_mkdir (2661)
6270 - init deleted item, add assertions ptlrpc_abort_inflight() (2725)
6271 - don't assign transno to errored transactions (2742)
6272 - don't delete objects on OST if given a bogus objid from MDS (2751)
6273 - handle large client PAGE_SIZE readdir on small PAGE_SIZE MDS (2777)
6274 - if rq_no_resend, then timeout request after recovery (2432)
6275 - fix MDS llog_logid record size, 64-bit array alignment (2733)
6276 - don't call usermode_helper from ptlrpcd, DEFAULT upcall (2773)
6277 - put magic in mount.lustre data, check for bad/NULL mount data (2529)
6278 - MDS recovery shouldn't delete objects that it has given out (2730)
6279 - if enqueue arrives after completion, don't clobber LVB (2819)
6280 - don't unlock pages twice when trigger_group_io returns error (2814)
6281 - don't deref NULL rq_repmsg if ldlm_handle_enqueue failed (2822)
6282 - don't write pages to disk if there was an error (1450)
6283 - don't ping imports that have recovery disabled (2676)
6284 - take buffered bytes into account when balancing socknal conn (2817)
6285 - hold a DLM lock over readdir always, use truncate_inode_pages (2706)
6286 - reconnect unlink llog connection after MDS reconnects to OST (2816)
6287 - remove little-endian swabbing of llog records (1987)
6288 - set/limit i_blksize to LL_MAX_BLKSIZE on client (2884)
6289 - retry reposting request buffers if they fail (1191)
6290 - grow extent at grant time to avoid granting a revoked lock (2809)
6291 - lock revoke doesn't evict page if covered by a second lock (2765)
6292 - disable VM readahead to avoid reading outside lock extents (2805)
6294 - return LL_SUPER_MAGIC from statfs for the filesystem type (1972)
6295 - updated kernel patches for hp-2.4.20 kernel (2681)
6297 2004-02-07 Cluster File Systems, Inc. <info@clusterfs.com>
6300 - fix truncated write corruption (2366)
6301 - fix for failed assertion in iopen_connect_dentry (1792,2517)
6303 - don't flag the ptlrpcd thread with PF_MEMALLOC (2636)
6304 - ensure len(uuid) < 37 in lmc (1171)
6305 - fix ia64 OOPS in llog_test (2255)
6306 - zero end of page at obdfilter for partial page writes (2648)
6307 - don't leave stale dentries around after renames (bug 2428)
6308 - fix timeouts when evicting a client with a single lock held (2642)
6309 - set deadline for the initial HELLO message to drain (2634)
6310 - print out dotted-quad IP addresses in the socknal (2302)
6312 - additional debugging for MDS client eviction problem (2443)
6313 - fix mkfsoptions support for osts (2603, 2604)
6315 2004-01-27 Cluster File Systems, Inc. <info@clusterfs.com>
6318 - add series for the vanilla 2.6.0 kernel
6319 - add series for the vanilla 2.4.24 kernel
6320 - add series for a cray x86/64 UL kernel drop
6321 - fix xattr patches for the vanilla 2.4.19 series
6323 - generate true UUIDs in lmc (1171)
6324 - have portals stack dumping break in UML (2466)
6325 - avoid bad dchild deref; avoid inum lock w/o creation (2362)
6326 - allocate with _NOFS in ldlm to avoid deadlock (1933)
6327 - wake callback waiting threads on client eviction (2460)
6328 - Add --ptldebug and --subsystem to lmc (1719)
6329 - update assertion to allow safe interrupt allocation
6330 - set rq_no_resend for cancel requests (2432)
6331 - recalculate ptlrpcd timeout after resend (2494)
6332 - call vfs_rmdir when removing pending directories (2368)
6333 - fix renaming a file to itself (2429)
6334 - lmc creates a default one-stripe lov (2454)
6335 - expand procfs space to handle large clusters (2326)
6336 - increase UML stack to avoid overflow
6337 - update lconf's list of debug and subsystem masks
6338 - fix lfs find --obd (2510)
6339 - /proc tunable for disabling filter read caching (2591)
6340 - stop rpm packages from altering slapd.conf (2301)
6341 - disable nagle in the socknal under 0conf (2578)
6342 - choose mds inode size based on stripe count (2572)
6343 - fix kernel-source rpm problems (2516)
6345 - add --disable-doc to avoid pdf generation (2421)
6346 - update documentation, tests, type-os, comments
6347 - avoid format warnings on ia64
6348 - remove the TOE NAL
6349 - tiny code cleanups by removing unused fields
6351 2004-01-07 Cluster File Systems, Inc. <info@clusterfs.com>
6354 - fix obvious semaphore misuse in as-yet-unused setattr path (2348)
6355 - remove the most blatant lies from BUILDING file (2371)
6356 - change default debug level to reasonable production setting
6357 - reduce client side cache size to reduce cache flush time
6358 - reduce max RPCs in flight to avoid unnecessary file fragmentation
6359 - make TCP zerocopy and pinger support enabled by default (2476)
6360 - sync writes completed after process exits caused crashes (2319)
6361 - maintain correct mount count on the MDS (2356)
6362 - backout 1557, because 2316 wasn't really fixed
6363 - better file I/O statistics gathering in /proc
6364 - don't take unnecessary, deadlock-inducing bug in readpage (2383)
6365 - another kernel patch to fix zero-copy TCP function export
6366 - don't take duplicate lock when processing re-sent getattr (2420)
6367 - lctl uses obd_self_export instead of creating new conn (2353)
6368 - MDS/OST recovery case which requires object creation asserted (2425)
6369 - move lfs from /usr/sbin to /usr/bin in packages
6370 - fix race between mds_client_add and mds_client_free (2417)
6371 - use kmalloc instead of slabs in portals (2430)
6372 - don't create duplicate records when a failover MDS is present (2442)
6373 - remove unnecessary mount age check (2332)
6374 - don't remove directory inodes from locks prematurely (2451)
6375 - don't break if MDS service name is the same as hostname (2103)
6376 - fix races in client write RPC generation when cache full (2482)
6378 2003-12-13 Cluster File Systems, Inc. <info@clusterfs.com>
6381 - remove now-unused request->rq_obd (278)
6382 - if an allocation fails, print out how much memory we've used (1933)
6383 - use PORTAL_SLAB_ALLOC for structures, to get GFP_MEMALLOC (1933)
6384 - add the "configurable stack size" patch to most series files (1256)
6385 - ability to write large log records, for 100+ OST configs (2306)
6386 - fix NULL deref when filter_prep fails (2314)
6387 - fix operator precedence error in filter_sync
6388 - dynamic allocation of socknal TX descriptors (2315)
6389 - fix a missed case in the GFP_MEMALLOC patch, can cause deadlock (2310)
6390 - fix gcc 2.96 compilation problem in xattr kernel patch (2294)
6391 - ensure that CWARN messages in Portals always get to the syslog
6392 - __init/__exit are not for prototype decls (ldlm_init/exit)
6393 - x86-64 compile warning fixes
6394 - fix gateway LMC keyword conflict (2318)
6395 - fix MDS lock inversions in getattr/reint paths (1844)
6396 - fix a rare lock re-ordering bug, which caused deadlock (2322)
6397 - fix i_sem/journal inversion in fsfilt_ext3_write_record (2306)
6398 - DLM race condition prevented some lock evictions (2328)
6399 - ENOMEM detection and retry on socknal sends (2230)
6400 - use GFP_NOFS throughout Lustre, to combat ENOMEM (2230)
6401 - move osc_rpcd into ptlrpc, for use in MDC and others (2329)
6402 - protect MDS inode fsdata with stronger locking; fixes assertion (2313)
6403 - better error messages when a client is rejected during recovery (1505)
6404 - avoid cancelling locks which were never granted, after failure (2330)
6405 - fix i_sem/journal inversion in mds_client_add (2333)
6406 - fix truncate/getattr lock cycle deadlock (2334)
6407 - use rpcd to send close; allows resend after timeout, avoid leak (1897)
6408 - fix two rare exit paths which could leak an l_lock() ref (2321)
6409 - fencepost error in MDS/OST orphan recovery (2226)
6410 - make log record alignment 8 bytes (1988)
6411 - lstripe now fails when requested offset > ost_count (2237)
6412 - ensure that all kernel series have a complete list.h (1607)
6413 - fix crashes in special-file operations (2316)
6414 - lctl create/brw OID mismatch, caused by obsolete filter loop (2339)
6416 - allow configurable automake binary, for testing new versions
6417 - small update to the lfs documentation
6419 2003-12-03 Cluster File Systems, Inc. <info@clusterfs.com>
6421 * fix negative export reference count in fsfilt_sync (2312)
6423 2003-12-01 Cluster File Systems, Inc. <info@clusterfs.com>
6424 * release candidate 0.9.1
6426 - orphans are moved into the PENDING directory for possible recovery
6427 - replayed opens now open by fid for orphan/rename safety (1042)
6428 - last close of an orphan inode generates a transno (683)
6429 - chdir() and mount() now pin the directory entry (1020)
6430 - avoid CERROR in normal ll_setattr_raw() error case (1500)
6431 - discard very old requests without processing them (1502)
6432 - remove some common, well-understood CERRORs (1505)
6433 - require O_DIRECT I/O to be page-sized to workaround IA64 crash (1609)
6434 - clear "grant" flags in OST replies until OST grant code lands (1644)
6435 - fix read performance by not clobbering i_blksize on client (1598)
6436 - fix __ldlm_handle2lock oops by not dereferencing lock after PUT (1625)
6437 - make LRU size a /proc tunable, clears locks when reduced (707)
6438 - fix some lprocfs rot that prevented ptlbd from loading (1732)
6439 - server locks take references on exports now (1558)
6440 - build fixes for 2.4.20-rh trees (1663)
6441 - return an error from lov_create if all OSCs are inactive (1751)
6442 - fix import levels when a reconnect happens without a timeout (1597)
6443 - exit early from mds_open if we get a lookup error (1749)
6444 - partial page read at EOF wouldn't wait for disk before sending (1642)
6445 - avoid NULL deref in obdfilter when reading page past EOF (1592)
6446 - avoid LASSERT in ll_intent_lock if server failed very early (1090)
6447 - fix LBUG in ll_it_open_error with rc = -2 (1861)
6448 - write/truncate lock inversion (1639)
6449 - Don't auto-load obdclass, portals modules during cleanup (1495)
6450 - fix timestamps from jumping to "now" (1763)
6451 - extra journal assertions (1648)
6452 - add an extra multiunlink test (1771)
6453 - fix read_record/write_record API (1776)
6454 - fix leak of offset_extent, possible incorrect i_size later (1772)
6455 - fix lasserts in mis-matched transnos during open-unlink testing (1541)
6456 - Debugging for the kqswnal_get_idle_tx problems (1820)
6457 - Allow recovery to be attempted multiple times (1536)
6458 - Write out MDS last_rcvd file after it is first created (1600)
6459 - Fix tx_descriptor leak in failed transmit situations (1827)
6460 - ext3 journaling fixes for assertion failure after IO error (1871)
6461 - class_export_put() on freed export after completion AST error (1896)
6462 - Fix revalidate looping in VFS (1322)
6463 - Don't access a freed export during MDS_REINT timeout (1521)
6464 - Add open-unlink recovery support on the MDS (1673,1764)
6465 - Return an error if no MDS data was read from last_rcvd (1946)
6466 - Fix for lookup "." or ".." crash on error (1932,1931,1935)
6467 - Don't setup a disk device that doesn't match exported UUID (317)
6468 - Reduce bulk RPC timeout to avoid cascading client/OST failures (1845)
6469 - avoid committing NULL handle in force close
6470 - local.sh is now a one-stripe LOV configuration
6471 - POSIX utime.4 -EPERM on FIFO not owned by user (56)
6472 - fix ext3 htree duplicate directory entry corruption (1516)
6473 - POSIX creat.13, fstat.1, open.18, stat.3 new file atime/mtime (2020)
6474 - update to new LOV EA format (2097)
6475 - interoperability for different PAGE_SIZE/wordsize (686,1821,1343,2042)
6477 2003-06-15 Phil Schwan <phil@clusterfs.com>
6480 - imports and exports cleanup too early, need refcounts (349, 879, 1045)
6481 - per-import/export recovery handling (958, 931, 959)
6482 - multiple last-rcvd slots, for serving multiple FSes (949)
6483 - connections are again shared between multiple imp/exports (963, 964)
6484 - "umount -f" would hang if any requests needed to be sent (393, 978)
6485 - avoid pinning large req buffer by copying for queued messages (989)
6486 - add "uuid" to "lctl device" command to help upcalls (991)
6487 - "open" RPCs with transnos would confuse recovery counters (1037)
6488 - do proper endian conversion of all wire messages (288, 340, 891)
6489 - remove OST bulk get LBUGs, fix ost_brw_write cleanup (1126)
6490 - call waiting locks callback from LDLM recovery thread (1127, 1151)
6491 - fix ptlrpc_connection leak in target_handle_connect (1174)
6492 - fix import refcounting bug in OST and MDS cleanup (1134)
6493 - if an invalid-at-open-time OSC returned before close(), LBUG (1150)
6494 - fix very unlikely obd_types race condition (501)
6495 - remove osc_open hack for echo_client (1187)
6496 - we leaked exports/dlmimps for forcibly disconnected clients (1143)
6497 - a failure in read_inode2 leads to deadlock (1139)
6498 - cancel ack-locks as soon as transaction is committed (1072)
6499 - fix major leaks and crashes in the bulk I/O path (937, 1057)
6500 - make sure to commitrw after any preprw to avoid deadlock (1162)
6501 - failing to execute a file in a lustre FS would lock inode (1203)
6502 - small DEBUG_REQ fix to avoid dereferencing a NULL (1227)
6503 - don't ASSERT while cleaning up an incompletely-setup obd (1248)
6504 - obd_uuid2tgt would walk off the end of the list (1255)
6505 - on IA64 the osc would give portals incorrect bulk size (1258)
6506 - fix debug daemon ioctl interface; allows daemon on ia64 (1274)
6507 - fix lock inversion caused by new llite matching code (1282)
6508 - limit the number of dirty pages on a client to 10MB (1286)
6509 - timed out locks were not being corrected cancelled (1289)
6510 - fix O_DIRECT above 4GB on IA-32 (1292)
6511 * major user-visible changes
6512 - fail out/fail over policy now controlled by the upcall (993)
6514 - add OBD_PING to check server availability and failure (954)
6515 - lustre messages are now sent in sending host order (288, 340, 891)
6516 - add eadatalen to MDS getattr reply (340)
6517 - OST read replies may contain second buffer, with per-page status (593)
6519 2003-03-11 Phil Schwan <phil@clusterfs.com>
6522 - LDLM_DEBUG macro fix, for gcc 3.2 (850)
6523 - failed open()s could cause deadlock; fixed (867, 869)
6524 - stop cancelling OST locks when files are closed (481)
6525 - overlapping XID spaces caused network corruption (851, 853)
6526 - fix unsafe fsfilt counter arithmetic; change to atomic_t
6527 - setattr_raw added, to do single-RPC, server-side setattrs
6528 - lmc/lconf syntax change for OST UUIDs
6529 - fix crashy race condition between ptlrpc_free_req and osc_close
6530 - don't use request in mdc_enqueue if we hit a timeout (889)
6531 - don't set the inode i_size for regular files from the MDS (896)
6532 - handle out of order completion AST (842)
6533 - don't LBUG if a lock request times out after receiving AST (913)
6534 - avoid d_rehash race in ll_find_alias by rehashing inside dcache_lock
6535 - if a bad lock AST arrives, send an error instead of dropping entirely
6536 - return 0 from revalidate2 if ll_intent_lock returns -EINTR (912)
6537 - fix leak in bulk IO when only partially completed (899, 900, 926)
6538 - fix O_DIRECT for ia64 (55)
6539 - (almost) eliminate Lustre-kernel-thread effects on load average (722)
6540 - C-z after timeout could hang a process forever; fixed (977)
6542 - client-side I/O cache (678, 924, 929, 941, 970)
6544 - READPAGE and SETATTRs which don't take server-side locks get
6547 2003-02-11 Phil Schwan <phil@clusterfs.com>
6550 - Fix ldlm_lock_match on the MDS to avoid matching remote locks (592)
6551 - Fix fsfilt_extN_readpage() to read a full page of directory
6552 entries, or fake the remainder if PAGE_SIZE != blocksize (500)
6553 - Avoid extra mdc_getattr() in ll_intent_lock when possible (534, 604)
6554 - Fix imbalanced LOV object allocation and out-of-bound access (469)
6555 - Most intent operations were removed, in favour of a new RPC mode
6556 that does a single RPC to the server and bypasses most of the VFS
6557 - All LDLM resource ID arrays were removed in favour of ldlm_res_id
6558 - Aggressively cancel local locks on DLM servers
6559 - mds_reint_unlink sends EA to the client if it's the last nlink.
6560 client uses that EA to unlink OST objects.
6561 - mds_reint_{rename,unlink,link} were rewritten to take ordered locks
6562 - recursive symlinks were fixed (439)
6563 - fixed NULL deref in DEBUG_REQ
6564 - filter_update_lastobjid no longer calls sync, which annoyed extN
6565 - fixed multi-client small-writes to a single file problem (445)
6566 - fixed mtime updates during file writes (607)
6567 - fixed vector writes on obdfilter causing problems when ENOSPC (670)
6568 - fixed bug in obd_brw_read/write() (under guise of testing 367)
6569 - fixed Linux OST size reporting problem (444, 656)
6570 - OST now updates object mtime with writes or setattr (607, 619)
6571 - client verifies file size before zeroing page past EOF (445)
6572 - OST now writes last allocated objid to disk with allocation (108)
6573 - LOV on echo now works (409)
6575 - mds_reint_unlink sends a new buffer, with the EA included. this
6576 buffer is only valid if body->valid & OBD_MD_FLEASIZE, which is only
6577 set if a regular file was being unlinked, and it was the last link
6578 - use PtlGet from the target for bulk writes (315)
6579 - OST now updates object mtime with writes or setattr (607, 619)
6580 - LDLM now has a grant-time callback to revalidate locked items, if
6582 - Many MDS operations were reorganized to combat race conditions
6584 - Merge b_intel branch (updated lprocfs code) - now at /proc/fs/lustre
6585 - configure check to avoid gcc version 2.96 20000731-2.96-98) (606)
6587 2003-01-06 Andreas Dilger <adilger@clusterfs.com>
6590 - Fully reactivate OST imports after reconnection (512, others)
6591 - Make sure client sees our -ENOTCONN from mds_handle (513 - partial)
6592 - More graceful error handling for truncating on dead OST (515)
6593 - Don't error out unless we're actually accessing dead stripes (474)
6594 - Fix garbage sizes when stripes are missing (410)
6595 - LRU counters were broken, causing constant lock purge (433, 432)
6596 - garbage on read from stripes with failed OSTs (441)
6597 - mark OSCs as active before reconnecting during recovery (438)
6598 - lov_enqueue and lov_cancel need to handle inactive OSTs (403)
6599 - lfind did not preserve OST order in output (443)
6600 - symlinks cause hung clients, incorrect data (439)
6601 - stop dereferencing request after dropping refcount (457)
6602 - don't LASSERT(spin_is_locked) on non-SMP (455)
6603 - fixes for many rename() bugs
6604 - fstat didn't correctly synchronize attributes (399)
6605 - server must handle lock cancellation during blocking AST prep (487)
6606 - bulk descriptors were free()d too soon (511)
6607 - fix paths in lconf, which would load incorrect modules (451, 507)
6608 - fix confusing lconf 'host not found' error message (386)
6609 - fix lock order deadlock on OST (O/R i_sem before journal ops, 478)
6610 - fix race condition in mdc_blocking_ast() for inode access (526)
6611 - fix lov_unpackmd() unpacking wrong number of stripes (537)
6612 - fix lov_set_osc_active() marking wrong OSC inactive (440)
6613 - fix bad lstripe lov_unpackmd() assertion (fix layering too) (527)
6614 - fix multiple writes of stripe MD to MDS (358, maybe 519)
6615 - fix lstripe in several ways (kernel side) (527)
6616 - fix request leak in ldlm_cli_enqueue (262)
6617 - incorrect OSC was marked inactive after OST failure
6618 - call mds_fs_cleanup before unmounting filesystem (524)
6619 - fix races between taking ns_lock and ldlm_lock_change_resource
6620 - fix races updating LOV export open file list
6621 - fix lov_enqueue error path, avoid decref-ing bad lock handle
6622 - fix recovery NULL deref in ldlm_cli_cancel_unused
6623 - fix some DLM races by using new hash table for lock handles (419)
6624 - permit the client to specify desired inodes, at replay
6625 - duplicate requests when we queue them for replay reintegration
6626 - fix last_rcvd offset calculation
6627 - sync after each recovered transaction, so we always make progress
6628 - never, not always, ERESTART requests without transnos
6629 - store the lov_desc in the MDS, so we don't depend on getlovinfo to
6631 - skip replay if the MDS says that the client is already connected
6632 - don't check for a recovery-enabled export to match lctl's UUID
6633 - don't INC_USE_COUNT for phantom exports
6634 - don't crash when cleaning up phantom exports (567)
6635 - don't double-finish or set replay data for errored mdc_open requests
6636 - abort requests when they time out, so we don't get old replies
6637 - send/receive replies for AST messages again
6638 - if the client says that it doesn't have the lock, cancel it on the
6640 - if we timeout during I/O, don't try to cancel an in-use lock; instead
6641 mark it as destroyed, it will all work out when decref is called
6642 - fix module use counts (22, 581)
6644 - ASTs now expect a reply (server cancels lock on error reply)
6646 2002-12-02 Andreas Dilger <adilger@clusterfs.com>
6649 - fix many simultaneous client startup (392)
6650 - fix dentry->d_it clobbering
6651 - credentials weren't being shipped for readdir/getattr operations
6652 - remove invalid assertions triggered during some concurrent MD
6654 - proper Lustre versions added (336, 389)
6655 - fix memory leak for create error case (398)
6656 - fix LOV locking bug that would get cli/srv out of sync
6657 - fix echo client over LOV (409)
6658 - fix dbench 2, extN refcount problem (170, 258, 356, 418)
6659 - fix double-O_EXCL intent crash (424)
6660 - avoid sending multiple lock CANCELs (352)
6662 - MDS can do multi-client recovery (modulo bugs in new code)
6664 - many updates, edits, cleanups
6666 2002-11-18 Phil Schwan <phil@clusterfs.com>
6669 - fix null d_it dereference (346)
6670 - fix full OST/dbench hang (333)
6671 - fix permission problem with file removal (286)
6672 - fix removal of OSCs from LOV when they fail
6673 - fix NULL deref during bulk timeout (214)
6674 - fix problems related to multiple filesystems on one MDS (241)
6675 - fixed serious subtle metadata locking bugs
6676 - free locks on clients when inodes are removed due to memory
6678 - fix inode pointer in lock data (285)
6679 - partial support for multiple MDS on a single host (241)
6680 - data locks weren't cancelled at clear_inode time (290, 311)
6681 - intent locks could lead to unbounded lock growth (205)
6682 - added a maximum lock count, an LRU list, and a flusher
6683 - fix multiple rename (365)
6684 - properly abstracted the echo client
6685 - OSC locked 1 byte too many; fixed
6686 - rewrote brw callback code:
6687 - fixed recovery bugs related to LOVs (306)
6688 - fixed too-many-pages-in-one-write crash (191)
6689 - fixed (again) crash in sync_io_timeout (214)
6690 - probably fixed callback-related race (385)
6692 - Add capability to MDS protocol
6693 - LDLM cancellations and callbacks on different portals
6695 2002-10-28 Andreas Dilger <adilger@clusterfs.com>
6698 - limit client IOV size to PTL_MD_MAX_IOV (611336, 191)
6699 - defer open object destruction to close time (601981, 138)
6700 - open/close OST file handle in obdo (OBD_MD_FLHANDLE) (601981, 138)
6701 - move LDLM_ENQUEUE/CONVERT back to MDS portal (625069)
6702 - abstract ll_lookup2, fix ll_revalidate2 to use abstraction (256)
6703 - don't call obd_setattr in ll_file_release for destroyed objects
6704 * protocol change to lustre_msg: move |version| and add |flags|
6705 * protocol change to osc_punch: "start" in "o_size", "end" in "o_blocks"
6706 * lock replay: for LDLM_FL_REPLAY trust client to do right thing
6707 * added replay of create, unlink, link and rename operations during
6708 MDS failover; recovery should be much more robust now
6709 * remove failed OSCs from LOVs (only lov_create uses this so far)
6710 * the lustre-HOWTO was brought (more) up to date (582544)
6712 2002-10-23 Phil Schwan <phil@clusterfs.com>
6715 - in-use dentries weren't being reused properly (617851)
6716 - prevent multiple LDLM setup (599178)
6717 - fix LOV size calculations for truncate (617853)
6718 - fix client handling of MDS intent errors (POSIX)
6719 - fix permission bug in lovstripe.c test (624321)
6720 - fix MDS thread deadlock - move LDLM handler to DLM portal (625069)
6721 - truncate past end of file could corrupt data
6722 - proper cleanup after timeouts, crashes, etc (592524, 550815)
6723 - a race in recovery could return ETIMEDOUT to apps (623947)
6724 - building outside the source directory was fixed
6725 * the lustre-HOWTO was brought (more) up to date (582544)
6726 * major progress was made on recovery functionality
6728 2002-10-10 Phil Schwan <phil@clusterfs.com>
6731 - recovery deadlock fix
6732 - rm -rf causes LBUG fix (617817)
6733 - file open by multiple tasks fix (618962)
6734 - directory permissions bugs (602707 and 620007)
6735 - journal_stop fixed with locking (611313)
6736 - O_APPEND failures resolved (618273, perhaps 614459)
6737 - lconf PATH fix (619770)
6738 - IA64 build fix (621450)
6739 - RPC buffer sizes scale with amount of memory
6741 2002-10-01 Phil Schwan <phil@clusterfs.com>
6744 - locks would be cancelled without throwing away data pages,
6745 resulting in inconsistent data (605627)
6746 - inode attributes were not always being refreshed (605627, 612449)
6747 - lconf now continues to cleanup after lctl reports an error
6748 - MDS now enforces user permissions (602707)
6749 - lprocfs cleanup fixed, but not yet enabled (614157)
6750 - fixed infinite server hang, should a client not respond to an AST
6751 - avoid going into recovery if user calls readlink() with a buffer
6752 that's too small (613941)
6753 - AST RPCs no longer require replies (614867) -- this may be changed
6754 - don't crash server if client sends an IOV that's too big (611336)
6755 - fixed lock conversion deadlock (611892)
6756 - fixed the following of symlinks (614622)
6757 * recovery: the server can remove locks from a client that dies, other
6758 clients can make progress
6759 * more extN patch fixes
6760 * compile-time configurable ptlrpc buffer allocations
6762 - collaborative read cache document
6763 - Lustre Lite Performance CDR document-in-progress
6765 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
6768 - fix typo in patch-2.4.18
6770 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
6773 - clear ptlrpc request each time in handle_incoming_request()
6774 - unlink of files now destroys the object on the OST
6776 2002-09-19 Peter Braam <braam@clusterfs.com>
6778 * add hard link support
6779 * change obdfile creation method
6780 * kernel patch changed
6782 2002-09-19 Peter Braam <braam@clusterfs.com>
6785 - stack overflow bug in extN fixed
6787 2002-09-18 Andreas Dilger <adilger@clusterfs.com>
6789 * documentation updates
6790 - add man pages for config tools
6791 - update tests/README to describe testing with new config tools
6792 - finish metadata API descriptions
6793 * bug fixes and cleanups
6794 - statfs workaround for 16TB limit
6795 - LOV stripe allocation improved, can stripe on subset of OSTs
6796 - LOV file size/IO offset was wrong for files > 4GB in size
6797 - object EA data was being dropped, caused files to be unreadable
6798 - memory overflow with non-LOV OST caused memory corruption
6799 - fixed regression tests to work with new config tools, obdfilter
6800 - fixed bug when directory size became larger than 1 block
6801 - fixed bug (for single client case) when PWD was deleted
6802 - invalidate local directory pages when doing intent-based ops
6803 - avoid LDLM oops when lock callback contained bad data
6805 2002-09-09 Andreas Dilger <adilger@clusterfs.com>
6807 * documentation updates
6808 * bug fixes and cleanups
6809 - configuration tools
6812 - 64-bit compile warnings
6813 - 64-bit internal statfs data
6815 * test_brw on persistent OST devices
6817 * lprocfs (disabled)
6819 2002-09-04 Andreas Dilger <adilger@clusterfs.com>
6821 * documentation updates
6822 * bug fixes and cleanups
6823 * configuration tools
6825 2002-08-30 Peter J. Braam <braam@clusterfs.com>
6828 * many small fixes to 0_5_4
6829 * io/network handling
6830 * thinkos in MDS operations
6832 2002-08-24 Peter J. Braam <braam@clusterfs.com>
6835 * crucial basic fixes to 0.5.3
6836 * IOR, Iozone work over Elan
6839 2002-08-07 Phil Schwan <phil@clusterfs.com>
6840 * version 0_5_3, our first alpha
6841 * we use the new Portals iovs
6842 * documentation updates
6843 * bug fixes and cleanups
6844 * small changes in the DLM wire protocol
6846 2002-07-25 Peter J. Braam <braam@clusterfs.com>
6847 * version 0_5_1 with some initial stability,
6848 * locking on MD and file I/O.
6849 * documentation updates
6850 * several bug fixes since 0.5.0
6851 * small changes in wire protocol
6853 2002-07-18 Phil Schwan <phil@clusterfs.com>
6855 * delivered as Lustre Light Alpha
6856 * fixed a crash after handling invalid MDS requests
6857 * fixed directory pages for architectures with non-4k pages sizes
6859 2002-07-11 Andreas Dilger <adilger@clusterfs.com>
6860 * release version v0_4_4
6861 * Moves TCP acceptor to be on port 2432 (unused Coda port) instead
6863 * Fixes a number of interruption problems with OST operations.
6864 * Update documentation for portals header changes
6865 * Move all wire protocol structs/defines to lustre_idl.h
6866 * Fixes symlink length bug.
6867 * Add tcpdump to repository.
6869 2002-07-05 Andreas Dilger <adilger@clusterfs.com>
6870 * release version v0_4_3
6871 * Fixes statfs for inodes on extN.
6872 * Fixes bug in runtests which would delete /etc/hosts.
6873 * Use 64-bit object IDs wherever possible (not into VFS though)
6874 Remove ost_get_info, which is unused by lustre, and out of date.
6876 2002-07-03 Peter Braam <braam@clusterfs.com>
6877 * release version v0_4_2 Fixes a lookup error (type not passed)
6878 * move forward to head of Portals
6879 * move forward to latest Lustre kernel
6881 2002-06-25 Peter Braam <braam@clusterfs.com>
6882 * release version v0_4_1. Hopefully stable on single node use.