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 4.18.0-425.10.1.el8 (RHEL8.7)
7 * Other server kernels known to build and work at some point (others may also work):
8 3.10.0-862.14.4.el7 (RHEL7.5)
9 3.10.0-957.27.2.el7 (RHEL7.6)
10 3.10.0-1062.18.1.el7 (RHEL7.7)
11 3.10.0-1127.19.1.el7 (RHEL7.8)
12 3.10.0-1160.83.1.el7 (RHEL7.9)
13 4.18.0-80.11.2.el8 (RHEL8.0)
14 4.18.0-147.8.1.el8 (RHEL8.1)
15 4.18.0-193.28.1.el8 (RHEL8.2)
16 4.18.0-240.22.1.el8 (RHEL8.3)
17 4.18.0-305.25.1.el8 (RHEL8.4)
18 4.18.0-348.23.1.el8 (RHEL8.5)
19 4.18.0-372.32.1.el8 (RHEL8.6)
20 4.4.120-92.70 (SLES12 SP2)
21 4.4.180-94.100 (SLES12 SP3)
22 4.4.0-131 (Ubuntu 16.04)
23 4.15.0-32 (Ubuntu 18.04)
24 5.4.0-48 (Ubuntu 20.04)
25 vanilla linux 5.4.0 (ZFS + ldiskfs)
26 vanilla linux 5.4.21 (ZFS + ldiskfs)
27 vanilla linux 5.4.136 (ZFS + ldiskfs)
28 5.10.0-60.79.0.103.oe2203 (openEuler 22.03 LTS)
29 * ldiskfs needs an ldiskfs patch series for that kernel, ZFS does not
30 * Client primary kernels built and tested during release cycle:
31 5.14.0-162.18.1.el9 (RHEL9.1)
32 4.18.0-425.10.1.el8 (RHEL8.7)
33 5.4.0-37 (Ubuntu 20.04)
34 5.14.21-150400.24.46 (SLES15 SP4)
35 * Other clients known to build on these kernels at some point (others may also work):
36 3.10.0-862.14.4.el7 (RHEL7.5)
37 3.10.0-957.27.2.el7 (RHEL7.6)
38 4.14.0-49.13.1.el7a (RHEL7.5)
39 4.14.0-115.2.2.el7a (RHEL7.6)
40 3.10.0-1062.18.1.el7 (RHEL7.7)
41 3.10.0-1127.19.1.el7 (RHEL7.8)
42 3.10.0-1160.83.1.el7 (RHEL7.9)
43 4.18.0-80.11.2.el8 (RHEL8.0)
44 4.18.0-147.8.1.el8 (RHEL8.1)
45 4.18.0-193.28.1.el8 (RHEL8.2)
46 4.18.0-240.22.1.el8 (RHEL8.3)
47 4.18.0-305.25.1.el8 (RHEL8.4)
48 4.18.0-348.23.1.el8 (RHEL8.5)
49 4.18.0-372.32.1.el8 (RHEL8.6)
50 5.14.0-70.30.1.el9 (RHEL9.0)
51 4.4.120-92.70 (SLES12 SP2)
52 4.4.180-94.100 (SLES12 SP3)
53 4.12.14-95.48 (SLES12 SP4)
54 4.12.14-122.133 (SLES12 SP5)
55 4.12.14-197.75 (SLES15 SP1)
56 5.3.18-24.96 (SLES15 SP2)
57 5.3.18-150300.59.93 (SLES15 SP3)
58 4.4.0-131 (Ubuntu 16.04)
59 4.15.0-48 (Ubuntu 18.04)
60 5.8.0-53 (Ubuntu 20.04.2 HWE)
61 5.11.0-31 (Ubuntu 20.04.3 HWE)
62 5.11.0 (vanilla kernel.org)
63 5.10.0 (openEuler 22.03 LTS)
64 * Recommended e2fsprogs version: 1.46.6-wc1 or newer
65 * Recommended ZFS version: 2.1.5
66 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
67 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
68 stack overflow. For more information, please refer to bugzilla 17630.
69 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
70 the re-exporting nodes to be mounted with "32bitapi" mount option
72 --------------------------------------------------------------------------------
77 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
78 for currently supported client and server kernel versions.
79 * Server primary kernels built and tested during release cycle:
80 4.18.0-348.2.1.el8 (RHEL8.5)
81 * Other server kernels known to build and work at some point (others may also work):
82 3.10.0-862.14.4.el7 (RHEL7.5)
83 3.10.0-957.27.2.el7 (RHEL7.6)
84 3.10.0-1062.18.1.el7 (RHEL7.7)
85 3.10.0-1127.19.1.el7 (RHEL7.8)
86 3.10.0-1160.49.1.el7 (RHEL7.9)
87 4.18.0-80.11.2.el8 (RHEL8.0)
88 4.18.0-147.8.1.el8 (RHEL8.1)
89 4.18.0-193.28.1.el8 (RHEL8.2)
90 4.18.0-240.22.1.el8 (RHEL8.3)
91 4.18.0-305.25.1.el8 (RHEL8.4)
92 4.4.120-92.70 (SLES12 SP2)
93 4.4.180-94.100 (SLES12 SP3)
94 4.4.0-131 (Ubuntu 16.04)
95 4.15.0-32 (Ubuntu 18.04)
96 5.4.0-48 (Ubuntu 20.04)
97 vanilla linux 5.4.0 (ZFS + ldiskfs)
98 vanilla linux 5.4.21 (ZFS + ldiskfs)
99 vanilla linux 5.4.136 (ZFS + ldiskfs)
100 * ldiskfs needs an ldiskfs patch series for that kernel, ZFS does not
101 * Client primary kernels built and tested during release cycle:
102 4.18.0-348.2.1.el8 (RHEL8.5)
103 5.4.0-37 (Ubuntu 20.04)
104 5.3.18-59.27 (SLES15 SP3)
105 * Other clients known to build on these kernels at some point (others may also work):
106 3.10.0-862.14.4.el7 (RHEL7.5)
107 3.10.0-957.27.2.el7 (RHEL7.6)
108 4.14.0-49.13.1.el7a (RHEL7.5)
109 4.14.0-115.2.2.el7a (RHEL7.6)
110 3.10.0-1062.18.1.el7 (RHEL7.7)
111 3.10.0-1127.19.1.el7 (RHEL7.8)
112 3.10.0-1160.49.1.el7 (RHEL7.9)
113 4.18.0-80.11.2.el8 (RHEL8.0)
114 4.18.0-147.8.1.el8 (RHEL8.1)
115 4.18.0-193.28.1.el8 (RHEL8.2)
116 4.18.0-240.22.1.el8 (RHEL8.3)
117 4.18.0-305.25.1.el8 (RHEL8.4)
118 4.4.120-92.70 (SLES12 SP2)
119 4.4.180-94.100 (SLES12 SP3)
120 4.12.14-95.48 (SLES12 SP4)
121 4.12.14-122.91 (SLES12 SP5)
122 4.12.14-197.75 (SLES15 SP1)
123 5.3.18-24.96 (SLES15 SP2)
124 4.4.0-131 (Ubuntu 16.04)
125 4.15.0-48 (Ubuntu 18.04)
126 5.8.0-53 (Ubuntu 20.04.2 HWE)
127 5.11.0-31 (Ubuntu 20.04.3 HWE)
128 5.11.0 (vanilla kernel.org)
129 * Recommended e2fsprogs version: 1.46.2.wc5 or newer
130 * Recommended ZFS version: 2.0.7
131 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
132 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
133 stack overflow. For more information, please refer to bugzilla 17630.
134 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
135 the re-exporting nodes to be mounted with "32bitapi" mount option
137 --------------------------------------------------------------------------------
141 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
142 for currently supported client and server kernel versions.
143 * Server primary kernels built and tested during release cycle:
144 3.10.0-1062.18.1.el7 (RHEL7.7)
145 3.10.0-1127.19.1.el7 (RHEL7.8)
146 3.10.0-1160.6.1.el7 (RHEL7.9)
147 * Other server kernels known to build and work at some point (others may also work):
148 3.10.0-862.14.4.el7 (RHEL7.5)
149 3.10.0-957.27.2.el7 (RHEL7.6)
150 4.18.0-80.11.2.el8 (RHEL8.0)
151 4.18.0-147.8.1.el8 (RHEL8.1)
152 4.18.0-193.28.1.el8 (RHEL8.2)
153 4.18.0-240.1.1.el8 (RHEL8.3)
154 4.4.120-92.70 (SLES12 SP2)
155 4.4.180-94.100 (SLES12 SP3)
156 4.4.0-131 (Ubuntu 16.04)
157 4.15.0-32 (Ubuntu 18.04)
158 5.4.0-48 (Ubuntu 20.04)
159 vanilla linux 5.4.0 (ZFS + ldiskfs)
160 * ldiskfs needs an ldiskfs patch series for that kernel, ZFS does not
161 * Client primary kernels built and tested during release cycle:
162 3.10.0-1062.18.1.el7 (RHEL7.7)
163 3.10.0-1127.19.1.el7 (RHEL7.8)
164 3.10.0-1160.6.1.el7 (RHEL7.9)
165 4.12.14-95.48 (SLES12 SP4)
166 4.12.14-122.20 (SLES12 SP5)
167 4.12.14-197.75 (SLES15 SP1)
168 5.3.18-24.24 (SLES15 SP2)
169 4.15.0-48 (Ubuntu 18.04)
170 * Other clients known to build on these kernels at some point (others may also work):
171 3.10.0-862.14.4.el7 (RHEL7.5)
172 3.10.0-957.27.2.el7 (RHEL7.6)
173 4.14.0-49.13.1.el7a (RHEL7.5)
174 4.14.0-115.2.2.el7a (RHEL7.6)
175 4.18.0-80.11.2.el8 (RHEL8.0)
176 4.18.0-147.8.1.el8 (RHEL8.1)
177 4.18.0-193.28.1.el8 (RHEL8.2)
178 4.18.0-240.1.1.el8 (RHEL8.3)
179 4.4.120-92.70 (SLES12 SP2)
180 4.4.180-94.100 (SLES12 SP3)
181 4.4.0-131 (Ubuntu 16.04)
182 5.4.0-37 (Ubuntu 20.04)
183 5.4.0 (vanilla kernel.org)
184 * Recommended e2fsprogs version: 1.45.6.wc5 or newer
185 * Recommended ZFS version: 2.0.0
186 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
187 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
188 stack overflow. For more information, please refer to bugzilla 17630.
189 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
190 the re-exporting nodes to be mounted with "32bitapi" mount option
192 --------------------------------------------------------------------------------
196 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
197 for currently supported client and server kernel versions.
198 * Server primary kernels built and tested during release cycle:
199 3.10.0-1062.1.1.el7 (RHEL7.7)
200 * Other server kernels known to build and work at some point (others may also work):
201 3.10.0-862.14.4.el7 (RHEL7.5)
202 3.10.0-957.27.2.el7 (RHEL7.6)
203 4.18.0-80.7.1.el8 (RHEL8.0)
204 4.4.120-92.70 (SLES12 SP2)
205 4.4.180-94.100 (SLES12 SP3)
206 4.4.0-131 (Ubuntu 16.04)
207 4.15.0-32 (Ubuntu 18.04)
208 vanilla linux 4.6.7 (ZFS only)
209 * ldiskfs needs an ldiskfs patch series for that kernel, ZFS does not
210 * Client primary unpatched kernels built and tested during release cycle:
211 3.10.0-1062.1.1.el7 (RHEL7.7)
212 4.12.14-95.29 (SLES12 SP4)
213 4.15.0-48 (Ubuntu 18.04)
214 * Other clients known to build on unpatched kernels at some point (others may also work):
215 3.10.0-862.14.4.el7 (RHEL7.5)
216 3.10.0-957.27.2.el7 (RHEL7.6)
217 4.14.0-49.13.1.el7a (RHEL7.5)
218 4.14.0-115.2.2.el7a (RHEL7.6)
219 4.18.0-80.7.1.el8 (RHEL8.0)
220 4.4.120-92.70 (SLES12 SP2)
221 4.4.180-94.100 (SLES12 SP3)
222 4.4.0-131 (Ubuntu 16.04)
223 vanilla linux 4.15.0 (ZFS only)
224 * Recommended e2fsprogs version: 1.45.2.wc1 or newer
225 * Recommended ZFS / SPL version: 0.7.13
226 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
227 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
228 stack overflow. For more information, please refer to bugzilla 17630.
229 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
230 the re-exporting nodes to be mounted with "32bitapi" mount option
231 * Note for Ubuntu 14.04.5:
232 The 4.4.0 series is the HWE Kernel derived from 16.04 (Xenial).
233 See: https://wiki.ubuntu.com/Kernel/LTSEnablementStack
235 --------------------------------------------------------------------------------
239 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
240 for currently supported client and server kernel versions.
241 * Server known to build on the following kernels (others may also work):
242 2.6.32-696.18.7.el6 (RHEL6.9)
243 3.10.0-862.14.4.el7 (RHEL7.5)
244 3.10.0-957.el7 (RHEL7.6)
245 4.4.120-92.70 (SLES12 SP2)
246 4.4.162-94.69 (SLES12 SP3)
247 4.4.0-131 (Ubuntu 16.04)
248 4.15.0-32 (Ubuntu 18.04)
249 vanilla linux 4.6.7 (ZFS only)
250 * ldiskfs needs an ldiskfs patch series for that kernel, ZFS does not
251 * Client known to build on unpatched kernels (others may also work):
252 2.6.32-696.18.7.el6 (RHEL6.9)
253 2.6.32-754.6.3.el6 (RHEL6.10)
254 3.10.0-862.14.4.el7 (RHEL7.5)
255 3.10.0-957.el7 (RHEL7.6)
256 4.14.0-49.13.1.el7a (RHEL7.5)
257 4.14.0-115.2.2.el7a (RHEL7.6)
258 4.4.120-92.70 (SLES12 SP2)
259 4.4.162-94.69 (SLES12 SP3)
260 4.4.0-131 (Ubuntu 16.04)
261 4.15.0-32 (Ubuntu 18.04)
262 vanilla linux 4.15.0 (ZFS only)
263 * Recommended e2fsprogs version: 1.44.3.wc1 or newer
264 * Recommended ZFS / SPL version: 0.7.9
265 * Tested with ZFS / SPL version: 0.7.9
266 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
267 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
268 stack overflow. For more information, please refer to bugzilla 17630.
269 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
270 the re-exporting nodes to be mounted with "32bitapi" mount option
271 * Note for Ubuntu 14.04.5:
272 The 4.4.0 series is the HWE Kernel derived from 16.04 (Xenial).
273 See: https://wiki.ubuntu.com/Kernel/LTSEnablementStack
275 --------------------------------------------------------------------------------
277 2018-04-03 Intel Corporation
279 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
280 for currently supported client and server kernel versions.
281 * Server known to build on patched kernels:
282 2.6.32-431.29.2.el6 (RHEL6.5)
283 2.6.32-504.30.3.el6 (RHEL6.6)
284 2.6.32-573.26.1.el6 (RHEL6.7)
285 2.6.32-642.15.1.el6 (RHEL6.8)
286 2.6.32-696.18.7.el6 (RHEL6.9)
287 3.10.0-693.21.1.el7 (RHEL7.4)
288 3.0.101-0.47.71 (SLES11 SP3)
289 3.0.101-107 (SLES11 SP4)
290 3.12.74-60.64.40 (SLES12 SP1)
291 4.4.114-92.64 (SLES12 SP2)
292 4.4.114-94.11 (SLES12 SP3)
293 3.13.0-101 (Ubuntu 14.04, ZFS only)
294 4.4.0-85.108 (Ubuntu 14.04.5 LTS)
295 4.4.0-116 (Ubuntu 16.04)
296 vanilla linux 4.6.7 (ZFS only)
297 * Client known to build on unpatched kernels:
298 2.6.32-431.29.2.el6 (RHEL6.5)
299 2.6.32-504.30.3.el6 (RHEL6.6)
300 2.6.32-573.26.1.el6 (RHEL6.7)
301 2.6.32-642.15.1.el6 (RHEL6.8)
302 2.6.32-696.18.7.el6 (RHEL6.9)
303 3.10.0-693.21.1.el7 (RHEL7.4)
304 3.0.101-0.47.71 (SLES11 SP3)
305 3.0.101-107 (SLES11 SP4)
306 3.12.74-60.64.40 (SLES12 SP1)
307 4.4.114-92.64 (SLES12 SP2)
308 4.4.114-94.11 (SLES12 SP3)
309 3.13.0-101 (Ubuntu 14.04)
310 4.4.0-85.108 (Ubuntu 14.04.5 LTS)
311 4.4.0-116 (Ubuntu 16.04)
313 * Recommended e2fsprogs version: 1.42.13.wc6 or newer
314 * Recommended ZFS / SPL version: 0.7.6
315 * Tested with ZFS / SPL version: 0.7.6
316 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
317 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
318 stack overflow. For more information, please refer to bugzilla 17630.
319 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
320 the re-exporting nodes to be mounted with "32bitapi" mount option
321 * Note for Ubuntu 14.04.5:
322 The 4.4.0 series is the HWE Kernel derived from 16.04 (Xenial).
323 See: https://wiki.ubuntu.com/Kernel/LTSEnablementStack
325 --------------------------------------------------------------------------------
327 2017-07-05 Intel Corporation
329 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
330 for currently supported client and server kernel versions.
331 * Server known to build on patched kernels:
332 2.6.32-431.29.2.el6 (RHEL6.5)
333 2.6.32-504.30.3.el6 (RHEL6.6)
334 2.6.32-573.26.1.el6 (RHEL6.7)
335 2.6.32-642.15.1.el6 (RHEL6.8)
336 2.6.32-696.3.1.el6 (RHEL6.9)
337 3.10.0-514.21.1.el7 (RHEL7.3)
338 3.0.101-0.47.71 (SLES11 SP3)
339 3.0.101-100 (SLES11 SP4)
340 3.12.74-60.64.40 (SLES12 SP1)
341 4.4.59-92.17 (SLES12 SP2)
342 vanilla linux 4.6.7 (ZFS only)
343 * Client known to build on unpatched kernels:
344 2.6.32-431.29.2.el6 (RHEL6.5)
345 2.6.32-504.30.3.el6 (RHEL6.6)
346 2.6.32-573.26.1.el6 (RHEL6.7)
347 2.6.32-642.15.1.el6 (RHEL6.8)
348 2.6.32-696.3.1.el6 (RHEL6.9)
349 3.10.0-514.21.1.el7 (RHEL7.3)
350 3.0.101-0.47.71 (SLES11 SP3)
351 3.0.101-100 (SLES11 SP4)
352 3.12.74-60.64.40 (SLES12 SP1)
353 4.4.59-92.17 (SLES12 SP2)
355 * Recommended e2fsprogs version: 1.42.13.wc4 or newer
356 * Recommended ZFS / SPL version: 0.6.5.9
357 * Tested with ZFS / SPL version: 0.6.5.9
358 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
359 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
360 stack overflow. For more information, please refer to bugzilla 17630.
361 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
362 the re-exporting nodes to be mounted with "32bitapi" mount option
364 --------------------------------------------------------------------------------
366 12-07-2016 Intel Corporation
368 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
369 for currently supported client and server kernel versions.
370 * Server known to build on patched kernels:
371 2.6.32-431.29.2.el6 (RHEL6.5)
372 2.6.32-504.30.3.el6 (RHEL6.6)
373 2.6.32-573.26.1.el6 (RHEL6.7)
374 2.6.32-642.11.1.el6 (RHEL6.8)
375 3.10.0-514.el7 (RHEL7.3)
376 3.0.101-0.47.71 (SLES11 SP3)
377 3.0.101-80 (SLES11 SP4)
378 3.12.62-60.64.8 (SLES12 SP1)
379 vanilla linux 4.5.7 (ZFS only)
380 * Client known to build on unpatched kernels:
381 2.6.32-431.29.2.el6 (RHEL6.5)
382 2.6.32-504.30.3.el6 (RHEL6.6)
383 2.6.32-573.26.1.el6 (RHEL6.7)
384 2.6.32-642.11.1.el6 (RHEL6.8)
385 3.10.0-514.el7 (RHEL7.3)
386 3.0.101-0.47.71 (SLES11 SP3)
387 3.0.101-80 (SLES11 SP4)
388 3.12.62-60.64.8 (SLES12 SP1)
390 * Recommended e2fsprogs version: 1.42.13.wc5 or newer
391 * Recommended ZFS / SPL version: 0.6.5.8
392 * Tested with ZFS / SPL version: 0.6.5.8
393 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
394 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
395 stack overflow. For more information, please refer to bugzilla 17630.
396 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
397 the re-exporting nodes to be mounted with "32bitapi" mount option
399 --------------------------------------------------------------------------------
401 02-29-2016 Intel Corporation
403 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
404 for currently supported client and server kernel versions.
405 * Server known to build on patched kernels:
406 2.6.32-431.29.2.el6 (RHEL6.5)
407 2.6.32-504.30.3.el6 (RHEL6.6)
408 2.6.32-573.12.1.el6 (RHEL6.7)
409 3.10.0-327.3.1.el7 (RHEL7.2)
410 3.0.101-0.47.71 (SLES11 SP3)
411 3.0.101-68 (SLES11 SP4)
412 vanilla linux 4.2.1 (ZFS only)
413 * Client known to build on unpatched kernels:
414 2.6.32-431.29.2.el6 (RHEL6.5)
415 2.6.32-504.30.3.el6 (RHEL6.6)
416 2.6.32-573.12.1.el6 (RHEL6.7)
417 3.10.0-327.3.1.el7 (RHEL7.2)
418 3.0.101-0.47.71 (SLES11 SP3)
419 3.0.101-68 (SLES11 SP4)
422 * Recommended e2fsprogs version: 1.42.13.wc4 or newer
423 * Recommended ZFS / SPL version: 0.6.4.2
424 * Tested with ZFS / SPL version: 0.6.4.2
425 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
426 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
427 stack overflow. For more information, please refer to bugzilla 17630.
428 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
429 the re-exporting nodes to be mounted with "32bitapi" mount option
431 --------------------------------------------------------------------------------
433 03-10-2015 Intel Corporation
435 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
436 for currently supported client and server kernel versions.
437 * Server known to build on patched kernels:
438 2.6.32-431.29.2.el6 (RHEL6.5)
439 2.6.32-504.8.1.el6 (RHEL6.6)
440 3.0.101-0.46 (SLES11 SP3)
441 * Client known to build on unpatched kernels:
442 2.6.32-431.29.2.el6 (RHEL6.5)
443 2.6.32-504.8.1.el6 (RHEL6.6)
444 3.10.0-123.20.1.el7 (RHEL7)
445 3.0.101-0.46 (SLES11 SP3)
446 * Recommended e2fsprogs version: 1.42.9.wc1 or newer
447 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
448 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
449 stack overflow. For more information, please refer to bugzilla 17630.
450 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
451 the re-exporting nodes to be mounted with "32bitapi" mount option
453 Severity : enhancement
455 Description: control OST-index in IDIF via ROCOMPAT flag.
456 Details : Introduce new flag OBD_ROCOMPAT_IDX_IN_IDIF that is stored in the
457 last_rcvd file. For new formatted OST device, it will be auto set;
458 for the case of upgrading from old OST device, you can enable it
459 via the lproc interface osd-ldiskfs.index_in_idif. With such flag
460 enabled, for new created OST-object, its IDIF-in-LMA will contain
461 the OST-index; for the existing OST-object, the OSD will convert
462 old format IDIF as new format IDIF with OST-index stored in the
463 LMA EA when accessing such OST-object or via OI scrub. Once such
464 flag is enabled, it cannot be reverted back, so the system cannot
465 be downgraded to the orignal incompatible version.
467 --------------------------------------------------------------------------------
469 07-30-2014 Intel Corporation
471 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
472 for currently supported client and server kernel versions.
473 * Server known to build on patched kernels:
474 2.6.32-431.20.3.el6 (RHEL6.5)
475 3.0.101-0.31 (SLES11 SP3)
476 * Client known to build on unpatched kernels:
477 2.6.32-431.20.3.el6 (RHEL6.5)
478 3.0.101-0.31 (SLES11 SP3)
481 * Recommended e2fsprogs version: 1.42.9.wc1 or newer
482 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
483 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
484 stack overflow. For more information, please refer to bugzilla 17630.
485 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
486 the re-exporting nodes to be mounted with "32bitapi" mount option
488 --------------------------------------------------------------------------------
490 10-31-2013 Intel Corporation
492 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
493 * Server builds for kernels:
494 2.6.32-358.18.1.el6 (RHEL6.4)
495 * Client builds for unpatched kernels:
496 2.6.32-358.18.1.el6 (RHEL6.4)
497 3.0.92-0.8 (SLES11 SP3)
499 * Recommended e2fsprogs version: 1.42.7.wc2 or newer
500 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
501 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
502 stack overflow. For more information, please refer to bugzilla 17630.
503 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
504 the re-exporting nodes to be mounted with "32bitapi" mount option
506 --------------------------------------------------------------------------------
508 05-30-2013 Intel Corporation
510 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
511 * Server builds for kernels:
512 2.6.32-358.6.2.el6 (RHEL6.4)
513 * Client builds for unpatched kernels:
514 2.6.32-358.6.2.el6 (RHEL6.4)
515 3.0.74-0.6.6 (SLES11 SP2)
517 * Recommended e2fsprogs version: 1.42.7.wc1 or newer
518 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
519 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
520 stack overflow. For more information, please refer to bugzilla 17630.
521 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
522 the re-exporting nodes to be mounted with "32bitapi" mount option
524 --------------------------------------------------------------------------------
526 9-30-2012 Whamcloud, Inc.
528 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
529 * Server support for kernels:
530 2.6.32-279.5.1.el6 (RHEL6)
531 * Client support for unpatched kernels:
532 2.6.32-279.5.1.el6 (RHEL6)
534 2.6.18-238.19.1.el5 (RHEL5)
535 2.6.16.60-0.69.1 (SLES10)
536 * Recommended e2fsprogs version: 1.41.90.wc4
537 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
538 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
539 stack overflow. For more information, please refer to bugzilla 17630.
540 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
541 the re-exporting nodes to be mounted with "32bitapi" mount option
543 Severity : enhancement
545 Description: Update RHEL6.3 kernel to 2.6.32-279.5.1.el6.
547 Severity : enhancement
549 Description: Update RHEL6.3 kernel to 2.6.32-279.2.1.el6.
551 Severity : enhancement
553 Description: Update RHEL6.2 kernel to 2.6.32-220.17.1.el6
555 Severity : enhancement
557 Description: Update RHEL6.2 kernel to 2.6.32-220.13.1.el6
559 Severity : enhancement
561 Description: Update RHEL6.2 kernel to 2.6.32-220.7.1.el6
563 --------------------------------------------------------------------------------
565 04-30-2012 Whamcloud, Inc.
567 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
568 * Server support for kernels:
569 2.6.32-220.4.2.el6 (RHEL6)
570 * Client support for unpatched kernels:
571 2.6.32-220.4.2.el6 (RHEL6)
573 2.6.18-238.19.1.el5 (RHEL5)
574 2.6.16.60-0.69.1 (SLES10)
575 * Recommended e2fsprogs version: 1.41.90.wc4
576 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
577 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
578 stack overflow. For more information, please refer to bugzilla 17630.
579 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
580 the re-exporting nodes to be mounted with "32bitapi" mount option
582 Severity : enhancement
584 Description: Update RHEL6.2 kernel to 2.6.32-220.4.2.el6
586 Severity : enhancement
588 Description: Update RHEL6.2 kernel to 2.6.32-220.4.1.el6
590 Severity : enhancement
592 Description: Update RHEL6.2 kernel to 2.6.32-220.el6
594 Severity : enhancement
596 Description: Update RHEL6.1 kernel to 2.6.32-131.17.1.el6
598 --------------------------------------------------------------------------------
600 09-30-2011 Whamcloud, Inc.
602 * Server support for kernels:
603 2.6.18-238.19.1.el5 (RHEL5)
604 2.6.32-131.6.1.el6 (RHEL6)
605 * Client support for unpatched kernels:
606 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
607 2.6.18-238.19.1.el5 (RHEL5)
608 2.6.32-131.6.1.el6 (RHEL6)
609 2.6.29.4-167.fc11 (FC11)
610 2.6.16.60-0.69.1 (SLES10)
612 * Recommended e2fsprogs version: 1.41.90.wc3
613 * NFS export disabled when stack size < 8192. Since the NFSv4 export of
614 Lustre filesystem with 4K stack may cause a stack overflow. For more
615 information, please refer to bugzilla 17630.
616 * NFSv4 reexport requires Lustre client on the re-exporting nodes to be
617 mounted with "32bitapi" mount option
619 Severity : enhancement
620 Description: Update RHEL6.1 kernel to 2.6.32-131.6.1.el6
622 Severity : enhancement
623 Description: Update RHEL5.6 kernel to 2.6.18-238.19.1.el5
625 Severity : enhancement
626 Description: Update RHEL5.6 kernel to 2.6.18-238.9.1.el5
629 Frequency : only with SLES10
631 Description: Use OFED "KMP" provided by Novell
632 Details : SLES10 SP3 ships with OFED in a separate "KMP" package.
633 Lustre is now built against this package. That means you need to
634 install the ofed-kmp package from Novell for the patchless client.
635 Note that the ofed-kmp that Novell ships may not exactly match the
636 kernel version but should still be compatible.
638 Severity : enhancement
640 Description: Update RHEL5.5 kernel to 2.6.18-194.17.1.el5
641 Update OEL5.5 kernel to 2.6.18-194.17.1.0.1.el5.
643 --------------------------------------------------------------------------------
645 2010-07-15 Oracle, Inc.
647 * Server support for kernels:
648 2.6.18-164.11.1.el5 (RHEL 5)
649 2.6.18-164.11.1.0.1.el5 (OEL 5)
650 * Client support for unpatched kernels:
651 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
652 2.6.18-164.11.1.el5 (RHEL 5),
653 2.6.18-164.11.1.0.1.el5 (OEL 5)
654 2.6.16.60-0.42.8 (SLES 10),
656 2.6.29.4-167.fc11 (FC11)
657 * Recommended e2fsprogs version: 1.41.10-sun2
659 Severity : enhancement
661 Description: Distribute lfs_migrate script with lustre
662 Details : The lfs_migrate mentioned in the manual is now included in the
663 lustre tree in lustre/scripts and will distributed in lustre RPMs.
665 Severity : enhancement
667 Description: Failover nid list added to import proc file.
669 Severity : enhancement
671 Description: "som_preview" mount option required to use SOM
672 Details : The clients will need to use som_preview mount option
676 Description: Lustre does not do 1MB IOs to HW RAID
677 Details : Bump MAX_PHYS/HW_SEGMENTS and SG_ALL to 256 in the RHEL5 kernel.
679 Severity : enhancement
681 Description: lfs quota output is not very convenient for awk/sed-parsing
682 Details : Some positions in lfs quota output table could be empty or
683 non-empty which made it hard to parse it with scripts, now
684 a dash is put instead of space where there is not supposed
689 Description: Update to OFED 1.5.1
690 Details : OFED 1.5.1 will be used in environments that don't use
691 vendor supplied OFED.
695 Description: Remove lnet. prefix from lctl params display
696 Details : Also renamed lnet.memused to lnet_memused to
701 Description: Use conf_param -d to remove permanant settings.
705 Description: Add list_param -R
706 Details : lctl list_param now supports recursive mode.
709 Bugzilla : 21829, 22312
710 Description: Fix llobdstat.
711 Details : 1. Now updating the write_bytes counter
712 2. Port script fixes from 1.8.
716 Description: decrease the usage of memory on clients.
717 Details : 1. On clients, recycle dentries and inodes unused.
718 2. Delete the code related to ll_deathrow(att 6215 in bug 1443). It
723 Description: llobdstat fix and enhancement
724 Details : add a counter to set a limit to how many samples will be returned
725 fix a wildcard in the path to limit to obdfilter stats only
729 Description: lfs check only allowed for root
730 Details : Code cleanups and added a regresion test of non-root use of
735 Description: MDS Oopses in ptlrpc_dispatch_difficult_reply()/__list_add()
736 Details : Fix hr->hr_index overload.
740 Description: Fix FIEMAP functionality.
744 Description: FC11 patchless client support added
748 Description: Update RHEL5.4 kernel to 2.6.18-164.11.1.el5 and
749 OEL5.4 kernel to 2.6.18-164.11.1.0.1.el5.
753 Description: New acc-sm script and various test-framework enhancments.
757 Description: Previously landed patch was reverted.
761 Description: Update SLES10 kernel to 2.6.16.60-0.42.8 (patchless)
765 Description: Add ldiskfs maxdirsize mount option.
769 Description: Fix mkfs.lustre support for 16TB LUNs.
773 Description: Print mmp_check_interval
774 Details : print mmp_check_interval and make it possible to abort
775 mount operation in case it takes too long.
779 Description: Add cascading_rw to lustre mpi tests.
783 Description: Store SOM data in EA.
784 Details : Also several SOM related cleanups.
788 Description: Add refcount to md_open_data
792 Description: Use refcount on lov_request_set to prevent memory corruption.
796 Description: Statahead: use parent dir's i_mutex for synchronization.
800 Description: loadgen improvements.
804 Description: Enable peer health detection in ksocklnd and ko2iblnd.
808 Description: Control DCACHE_LUSTRE_INVALID flag with MDS_INODE_LOOKCUP lock
812 Description: Fix race on obd_devcie:md_stats
816 Description: Fix SLES 11 client builds.
817 Details : Resolved uninitalized pointer gcc warnings.
821 Description: Allow non-root access for "lfs check".
822 Details : Added a check in obd_class_ioctl() for OBD_IOC_PING_TARGET.
826 Description: Clear imp_force_reconnect correctly in ptlrpc_connect_interpret()
828 Severity : enhancement
830 Description: Update kernel to RHEL5.4 2.6.18-164.6.1.el5 and
831 OEL5 2.6.18-164.6.1.0.1.el5(Both in-kernel OFED enabled).
835 Description: Sanity test_27p fails.
836 Details : OBD_FAIL_OST_ENOSPC has been moved to filter_statfs().
837 Sanity tests 27 were cleaned up and improved.
841 Description: Unify req format on client/servers
842 Details : Use new req_capsule API [almost] everywhere instead of old PTLRPC
843 buffers and swabbers approach..
848 Description: drop join file stuff.
850 Severity : enhancement
852 Description: Add 2.6.27 and SLES11 (patchless client) support
856 Description: Fix in ptlrpc_expire_one_request() to print the signed time
859 Severity : enhancement
861 Description: Build kernels (RHEL5, OEL5 and SLES10/11) using the vendor's own
864 Severity : enhancement
866 Description: Remove set_info(KEY_UNLINKED) from MDS/OSC
868 Severity : enhancement
870 Description: correctly handle big reply message.
871 Details : send LNet event if reply is bigger than buffer and adjust this buffer
876 Description: Drop unnecessary __GFP_NOMEMALLOC flag from filter_get_page()
878 Severity : enhancement
880 Description: Add support for OEL5.
884 Description: Conf-sanity.sh 50g test - deactivated OST should not cause a panic.
886 Severity : enhancement
888 Description: "weak-modules" support
889 Details : Implement "weak-modules" support which enables kernel modules
890 to be used with any kernel that implements the same kABI. In
891 order to achieve this modules are now installed in
892 /lib/modules/$(uname -r)/updates/kernel on all distributions.
896 Description: client could not reconnect to OST because of an active request.
897 Details : The client now retries to reconnect to the same server, if a
898 connect request failed with EBUSY or -EAGAIN. Test case 35b
899 conf-sanity.sh has been created.
903 Description: Don't allow make backward step on assignin osc next id.
904 Details : race between allocation next id and ll_sync thread can be cause
905 of set wrong osc next id and can be kill valid ost objects.
909 Description: Changes in raid5-large-io-rhel5.patch to calculate sectors properly
913 Description: Increase the default BLK_DEF_MAX_SECTORS value for RHEL5 and SLES11
917 Description: Error handling in osc_statfs_interpret() has been improved.
918 Details : Check in osc_statfs_interpret() for EBADR.
922 Description: Do not send statfs() requests to OSTs disabled by administrator.
923 Details : Check in lov_prep_statfs_set() for non-NULL ltd_exp.
925 Severity : enhancement
927 Description: Added sanity-quota test to for lfs(1) to work with numeric uid/gid.
929 Severity : enhancement
931 Description: Update kernel to RHEL5 2.6.18-128.7.1.el5.
933 Severity : enhancement
935 Description: Enhancement for lfs(1) command to use numeric uid/gid.
939 Description: Improved cleanup procedure for conf-sanity test_46a.
943 Description: Do not start bulk transfer, if client tried to reconnect.
944 Details : Do not start bulk transfer, if client tried to reconnect,
945 but failed due to active rpcs. Check for exp_abort_active_req
946 has been implemented.
950 Description: Prevent inconsistences between linux and lustre mount structures.
951 Details : Wait indefinitely in server_wait_finished() until mnt_count drops.
952 Make the sleep interruptible.
954 Severity : enhancement
956 Description: provide server to client comms path
957 Details : server to client communications path via new LDLM_SET_INFO rpc
961 Description: truncate starts GFP_FS allocation under transaction causing deadlock
962 Details : ldiskfs_truncate calls grab_cache_page which may start page
963 allocation under an open transaction. This may lead to
964 calling prune_icache with consequent lustre reentrance.
968 Description: send correctly lsm on open replay
969 Details : MDS is trust to LSM size on replay open, but client can set wrong size
974 Description: enable client interface failover
975 Details : When a child reconnects from another NID, properly update export
976 nid hash position and ldlm reverse import.
978 Severity : enhancement
980 Description: Communicate OST degraded/readonly state via statfs to MDS
981 Details : Flags in the statfs returned from OSTs indicate whether the
982 OST is in a degraded RAID state, or if the filesystem has
983 turned read-only after a filesystem error is detected.
988 Description: don't panic if EPROTO was hit when reading symlink
989 Details : correctly handling request reference in error cases.
991 Severity : enhancement
993 Description: Add LustreNetLink, a kernel-userspace communcation path.
995 Severity : enhancement
997 Description: Update kernel to SLES10 SP2 2.6.16.60-0.39.3.
1002 Description: access to llog context before init.
1003 Details : move handling CATALOGS file at osc layer and forbid access to llog
1004 context before init.
1007 Frequency : rare, only when client is writing to OST on local node
1009 Description: Avoid deadlock for local client writes
1010 Details : Use new OBD_BRW_MEMALLOC flag to notify OST about writes in the
1011 memory freeing context. This allows OST threads to set the
1012 PF_MEMALLOC flag on task structures in order to allocate memory
1013 from reserved pools and complete IO.
1014 Use GFP_HIGHUSER for OST allocations for non-local client writes,
1015 so that the OST threads generate memory pressure and allow
1016 inactive pages to be reclaimed.
1018 Severity : enhancement
1020 Description: Update kernel to RHEL5.3 2.6.18-128.1.14.el5.
1025 Description: lock ordering violation between &cli->cl_sem and _lprocfs_lock
1026 Details : move ldlm namespace creation in setup phase to avoid grab
1027 _lprocfs_lock with cli_sem held.
1031 Description: Temporarily disable grant shrink.
1032 Details : Disable the feature for debugging.
1035 Frequency : only when formatting test filesystems
1037 Description: Unable to run several mkfs.lustre on loop devices at the same time
1038 Details : mkfs.lustre returns error 256 on the concurrent loop devices
1039 formatting. The solution is to proper handle the error.
1041 Severity : enhancement
1043 Description: Update kernel to RHEL5.3 2.6.18-128.1.6.el5.
1045 Severity : enhancement
1047 Description: Update kernel to SLES10 SP2 2.6.16.60-0.37.
1052 Description: resolve race between obd_disconnect and class_disconnect_exports
1053 Details : if obd_disconnect will be called to already disconnected export he
1054 forget release one reference and osc module can't unloaded.
1056 Severity : enhancement
1058 Description: Allow tuning service thread via /proc
1059 Details : For each service a new
1060 /proc/fs/lustre/{service}/*/thread_{min,max,started} entry is
1061 created that can be used to set min/max thread counts, and get the
1062 current number of running threads.
1067 Descriptoin: don't return error if have partially created objects for file.
1068 Details : lov_update_create_set uses set->set_success as index for created
1069 objects, so if some requests will be failed, they will have hole
1070 at end of array and qos_shrink_lsm for allocate correct lsm.
1072 Severity : enhancement
1074 Description: Update OFED support to 1.4.1
1079 Description: Reduce small size read RPC
1080 Details : Set read-ahead limit for every file and only do read-ahead when
1081 available read-ahead pages are bigger than 1M to avoid small size
1084 Severity : enhancement
1086 Description: move AT tunable parameters for more consistent usage
1087 Details : add AT tunables under /proc/sys/lustre, add to conf_param parsing
1089 Severity : enhancement
1091 Description: add lazystatfs mount option to allow statfs(2) to skip down OSTs
1092 Details : allow skip disconnected ost for send statfs request and hide error
1098 Description: fix racy locking of mballoc block bitmaps causing BUG
1099 Details : The locking of the mballoc buddy bitmap and the in-memory
1100 block bitmap was using two different spin locks in some
1101 cases. This made it possible to incorrectly access the
1102 mballoc bitmap while another process was modifying it,
1103 causing a sanity assertion to fail. While no on-disk corruption
1104 was reported, there was some risk of this happening.
1107 Frequency : rare, on llog test 6
1109 Descriptoin: don't allow connect to already connected import
1110 Details : allowing connect to already connected import is hide connecting problem.
1115 Descriptoin: allow kill process which wait statahead result
1116 Details : for some reasons 'ls' can stick in waiting result from statahead,
1117 in this case need way for kill this process.
1119 Severity : enhancement
1121 Description: Add state history info file, enhance import info file
1122 Details : Track import connection state changes in a new osc/mdc proc file;
1123 add overview-type data to the osc/mdc import proc file.
1125 Severity : enhancement
1127 Description: MDS create should not wait for statfs RPC while holding DLM lock.
1129 Severity : enhancement
1131 Description: Update to RHEL5U3 kernel-2.6.18-128.1.1.el5.
1136 Descriptoin: OST grant too much space to client even there are not enough space.
1137 Details : Client will shrink its grant cache to OST if there are no write
1138 activity over 6 mins (GRANT_SHRINK_INTERVAL), and OST will retrieve
1139 this grant cache if there are already not enough avaible space
1140 (left_space < total_clients * 32M).
1143 Frequency : start MDS on uncleanly shutdowned MDS device
1145 Descriptoin: ll_sync thread stay in waiting mds<>ost recovery finished
1146 Details : stay in waiting mds<>ost recovery finished produce random bugs
1147 due race between two ll_sync thread for one lov target. send
1148 ACTIVATE event only if connect realy finished and import have
1152 Frequency : rare, connect and disconnect target at same time
1154 Descriptoin: ASSERTION(atomic_read(&imp->imp_inflight) == 0
1155 Details : don't call obd_disconnect under lov_lock. this long time
1156 operation and can block ptlrpcd which answer to connect request.
1161 Descriptoin: don't lose wakeup for imp_recovery_waitq
1162 Details : recover_import_no_retry or invalidate_import and import_close can
1163 both sleep on imp_recovery_waitq, but we was send only one wakeup
1167 Frequency : always with long access acl
1169 Descriptoin: mds can't pack reply with long acl.
1170 Details : mds don't control size of acl but they limited by reint/getattr
1173 Severity : enhancement
1175 Description: Update to SLES10 kernel-2.6.16.60-0.33.
1177 Severity : enhancement
1179 Description: Update to RHEL5 kernel-2.6.18-92.1.22.el5.
1182 Frequency : start MDS on uncleanly shutdowned MDS device
1184 Descriptoin: aborting recovery hang on MDS
1185 Details : don't throttle destroy RPCs for the MDT.
1188 Frequency : on remount
1190 Description: external journal device not working after the remount
1191 Details : clear dev_rdonly flag for external journal devices in
1197 Description: shutdown vs evict race
1198 Details : client_disconnect_export vs connect request race.
1199 if client will evicted at this time - we start invalidate
1200 thread without referece to import and import can be freed
1204 Frequency : rare, need acl's on inode.
1206 Description: client can't handle ost additional correctly
1207 Details : if ost was added after client connected to mds client can have
1208 hit lnet_try_match_md ... to big messages to wide striped files.
1209 in this case need teach client to handle config events about add
1210 lov target and update client max ea size at that event.
1212 Severity : enhancement
1214 Description: Changelogs
1215 Details : Changelogs are a lightweight mechanism to track filesystem
1216 metadata and namespace changes. The changelog is recorded
1217 permanently on the MDTs, and is periodically "consumed" / purged
1218 when records are no longer needed.
1220 Severity : enhancement
1222 Description: compact fld format with extents
1223 Details : Store range of seq rather than every seq in FLD. Seq
1224 controller update FLD rather than clients. In Case of CMD, mdt0
1225 has FLD, all other metadata server act as non persistent proxy
1226 for FLD queries and cache fld entries in fld cache.
1231 Description: don't skip ost target if they assigned to file
1232 Details : Drop slow OSCs if we can, but not for requested start idx.
1233 This means "if OSC is slow and it is not the requested
1234 start OST, then it can be skipped, otherwise skip it only
1235 if it is inactive/recovering/out-of-space.
1239 Description: more cleanup in mds_lov
1240 Details : not send LOV EA under replay, we can't know about they size at this
1241 time. Don't allow client connect to mds before any ost connected,
1242 for avoid problems with LOV EA size and returning EIO to client.
1244 Severity : enhancement
1246 Description: Interoperability at server side (Disk interoperability)
1248 Severity : enhancement
1250 Description: Update to RHEL5 kernel-2.6.18-92.1.17.el5.
1252 Severity : enhancement
1254 Description: Update to SLES10 SP2 kernel-2.6.16.60-0.31.
1256 Severity : enhancement
1258 Description: New client IO stack (CLIO).
1260 Severity : enhancement
1262 Description: Commit on sharing. Eliminate inter-client dependencies between
1263 uncommitted transactions by doing transaction commits.
1264 Thereby clients may recovery independently.
1267 Frequency : Create a symlink file with a very long name
1269 Description: ldlm_cancel_pack()) ASSERTION(max >= dlm->lock_count + count)
1270 Details : If there is no extra space in the request for early cancels,
1271 ldlm_req_handles_avail() returns 0 instead of a negative value.
1273 Severity : enhancement
1275 Description: Add /proc entry for import status
1276 Details : The mdc, osc, and mgc import directories now have
1277 an import directory that contains useful import data for debugging
1278 connection problems.
1280 Severity : enhancement
1282 Description: Re-disable certain /proc logging
1283 Details : Enable and disable client's offset_stats, extents_stats and
1284 extents_stats_per_process stats logging on the fly.
1287 Frequency : Only on FC kernels 2.6.22+
1289 Description: oops in statahead
1290 Details : Do not drop reference count for the dentry from VFS when lookup,
1291 VFS will do that by itself.
1293 Severity : enhancement
1295 Description: Generic /proc file permissions
1296 Details : Set /Proc file permissions in a more generic way to enable non-
1297 root users operate on some /proc files.
1301 Description: Hitting mdc_commit_close() ASSERTION
1302 Details : Properly handle request reference release in
1303 ll_release_openhandle().
1307 Description: quota recovery deadlock during mds failover
1308 Details : This patch includes att18982, att18236, att18237 in bz14840.
1310 1. fix osts hang when mds does failover with quotaon
1311 2. prevent watchdog storm when osts threads wait for the
1316 Frequency : only patchless client
1317 Description: add workaround for race between add/remove dentry from hash
1319 Severity : enhancement
1321 Description: Allow OST glimpses to return PW locks
1325 Description: LBUG when llog conf file is full
1326 Details : When llog bitmap is full, ENOSPC should be returned for plain log.
1330 Description: Prevent import from entering FULL state when server in recovery
1334 Description: service mount cannot take device name with ":"
1335 Details : Only when device name contains ":/" will mount treat it as
1341 Description: replace ptlrpcd with the statahead thread to interpret the async
1342 statahead RPC callback
1346 Frequency : on recovery
1347 Description: I/O failures after umount during fail back
1348 Details : if client reconnected to restarted server we need join to recovery
1349 instead of find server handler is changed and process self
1350 eviction with cancel all locks.
1352 Severity : enhancement
1354 Description: Update to RHEL5 kernel-2.6.18-92.1.10.el5.
1356 Severity : enhancement
1358 Description: Update to SLES10 SP2 kernel-2.6.16.60-0.27.
1360 Severity : enhancement
1362 Description: Upcall on Lustre log has been dumped
1363 Details : Allow for a user mode script to be called once a Lustre log has
1364 been dumped. It passes the filename of the dumped log to the
1365 script, the location of the script can be specified via
1366 /proc/sys/lnet/debug_log_upcall.
1371 Description: avoid idr_remove called for id which is not allocated.
1372 Details : Move assigment s_dev for clustered nfs to end of initialization,
1373 to avoid problem with error handling.
1378 Description: avoid Already found the key in hash [CONN_UNUSED_HASH] messages
1379 Details : When connection is reused this not moved from CONN_UNUSED_HASH
1380 into CONN_USED_HASH and this prodice warning when put connection
1381 again in unused hash.
1384 Severity : enhancement
1386 Description: File striping can now be set to use an arbitrary pool of OSTs.
1388 Severity : enhancement
1390 Description: Export bytes_read/bytes_write count on OSC/OST.
1394 Description: Early reply size mismatch, MGC loses connection
1395 Details : Apply the MGS_CONNECT_SUPPORTED mask at reconnect time so
1396 the connect flags are properly negotiated.
1401 Description: quotas are not honored with O_DIRECT
1402 Details : all writes with the flag O_DIRECT will use grants which leads to
1403 this problem. Now using OBD_BRW_SYNC to guard this.
1407 Description: add quota statistics
1408 Details : 1. sort out quota proc entries and proc code.
1409 2. add quota statistics
1411 Severity : enhancement
1413 Description: enable quota support for HEAD.
1417 Description: Properly propagate oinfo flags from lov to osc for statfs
1418 Details : restore missing copy oi_flags to lov requests.
1420 Severity : enhancement
1422 Description: Add man pages for llobdstat(8), llstat(8), plot-llstat(8),
1423 : l_getgroups(8), lst(8), routerstat(8)
1424 Details : included man pages for llobdstat(8), llstat(8),
1425 : plot-llstat(8), l_getgroups(8), lst(8), routerstat(8)
1427 Severity : enhancement
1429 Description: configure's --enable-quota should check the
1430 : kernel .config for CONFIG_QUOTA
1431 Details : configure is terminated if --enable-quota is passed but
1432 : no quota support is in kernel
1436 Description: Remove portals compatibility
1437 Details : Remove portals compatibility, not interoperable with releases
1442 Description: Resolve device initialization race
1443 Details : Prevent proc handler from accessing devices added to the
1444 obd_devs array but yet be intialized.
1446 Severity : enhancement
1448 Description: Update to SLES10 SP2 kernel-2.6.16.60-0.23.
1450 Severity : enhancement
1452 Description: Update to RHEL5 kernel-2.6.18-92.1.6.el5.
1457 Description: Using wrong pointer in osc_brw_prep_request
1458 Details : Access to array[-1] can produce panic if kernel compiled with
1459 CONFIG_PAGE_ALLOC enabled
1463 Description: Client runs out of low memory
1464 Details : Consider only lowmem when counting initial number of llap pages
1468 Description: *optional* service tags registration
1469 Details : if the "service tags" package is installed on a Lustre node
1470 When the filesystem is mounted, a local-node service tag will
1471 be created. See http://inventory.sun.com/ for more information
1472 about the Service Tags asset management system.
1476 Description: Kernel BUG tries to release flock
1477 Details : Lustre does not destroy flock lock before last reference goes
1478 away. So always drop flock locks when client is evicted and
1479 perform unlock regardless of successfulness of speaking to MDS.
1483 Description: add refcount protection for osc callbacks, avoid panic on shutdown
1487 Description: sanity test 65a fails if stripecount of -1 is set
1488 Details : handle -1 striping on filesystem in ll_dirstripe_verify
1493 Description: ASSERTION(CheckWriteback(page,cmd)) failed
1494 Details : badly clear PG_Writeback bit in ll_ap_completion can produce false
1497 Severity : enhancement
1499 Description: Update to RHEL5 kernel-2.6.18-53.1.21.el5.
1503 Description: do not process already freed flock
1504 Details : flock can possibly be freed by another thread before it reaches
1505 to ldlm_flock_completion_ast.
1509 Description: LBUG during stress test
1510 Details : Need properly lock accesses the flock deadlock detection list.
1514 Description: oops in page fault handler
1515 Details : kernel page fault handler can return two special 'pages' in error
1516 case, don't try dereference NOPAGE_SIGBUS and NOPAGE_OMM.
1520 Description: timeout with invalidate import.
1521 Details : ptlrpcd_check call obd_zombie_impexp_cull and wait request which
1522 should be handled by ptlrpcd. This produce long age waiting and
1523 -ETIMEOUT ptlrpc_invalidate_import and as result LASSERT.
1525 Severity : enhancement
1527 Description: Update to RHEL5 kernel-2.6.18-53.1.19.el5.
1531 Description: enable MGS and MDT services start separately
1532 Details : add a 'nomgs' option in mount.lustre to enable start a MDT with
1533 a co-located MGS without starting the MGS, which is a complement
1534 to 'nosvc' mount option.
1538 Frequency : after recovery
1539 Description: precreate to many object's after del orphan.
1540 Details : del orphan st in oscc last_id == next_id and this triger growing
1541 count of precreated objects. Set flag LOW to skip increase count
1542 of precreated objects.
1546 Frequency : rare, on clear nid stats
1547 Description: ASSERTION(client_stat->nid_exp_ref_count == 0)
1548 Details : when clean nid stats sometimes try destroy live entry,
1549 and this produce panic in free.
1553 Description: Stack overflow during MDS log replay
1554 ease stack pressure by using a thread dealing llog_process.
1558 Description: wait until IO finished before start new when do lock cancel.
1559 Details : VM protocol want old IO finished before start new, in this case
1560 need wait until PG_writeback is cleared until check dirty flag and
1561 call writepages in lock cancel callback.
1563 Severity : enhancement
1565 Description: using special macro for print time and cleanup in includes.
1569 Description: mds_mfd_close() ASSERTION(rc == 0)
1570 Details : In mds_mfd_close(), we need protect inode's writecount change
1571 within its orphan write semaphore to prevent possible races.
1575 Description: Obsolete CURRENT_SECONDS and use cfs_time_current_sec() instead.
1579 Frequency : rare, on shutdown ost
1580 Description: don't hit live lock with umount ost.
1581 Details : shrink_dcache_parent can be in long loop with destroy dentries,
1582 use shrink_dcache_sb instead.
1586 Description: don't panic with use echo client
1587 Details : echo client pass NULL as client nid pointer and this produce null
1588 pointer dereference.
1592 Description: fix build on ppc32
1593 Details : compile code with -m64 flag produce wrong object file for ppc32.
1597 Description: add message levels for liblustreapi
1601 Description: fix for occasional failure case of -ENOSPC in recovery-small tests
1602 Details : Move the 'good_osts' check before the 'total_bavail' check. This
1603 will result in an -EAGAIN and in the exit call path we call
1604 alloc_rr() which will with increasing aggressiveness attempt to
1605 aquire precreated objects on the minimum number of required OSCs.
1609 Description: Use old size assignment to avoid deadlock
1610 Details : Reverts the changes in bugs 2369 and bug 14138 that introduced
1611 the scheduling while holding a spinlock. We do not need locking
1612 for size in ll_update_inode() because size is only updated from
1613 the MDS for directories or files without objects, so there is no
1614 other place to do the update, and concurrent access to such inodes
1615 are protected by the inode lock.
1619 Description: resolve "_IOWR redefined" build error on SLES10
1623 Description: dump the memory debugging after all modules are unloaded to
1624 suppress false negative in conf_sanity test 39
1626 Severity : enhancement
1628 Description: build kernel-ib packages for OFED 1.3 in our release cycle
1633 Description: fix SLES kernel versioning
1634 Details : the kernel version for our SLES 10 kernel did not include a "-"
1635 before the "smp" at the end. while this was not a problem in
1636 general, it did mean that software trying to use the kernel
1637 version to try to detect a vendor specific kernel would fail.
1638 this was most evident by the OFED build scripts.
1642 Description: Don't update lov_desc members until making sure they are valid
1643 Details : When updating lov_desc members via proc fs, need fix their
1644 validities before doing the real update.
1648 Description: don't put request into delay list while invalidate in flight.
1649 Details : ptlrpc_delay_request sometimes put in delay list while invalidate
1650 import in flight. this produce timeout for invalidate and sometimes
1651 can cause stale data.
1655 Frequency : on ppc only
1656 Description: not convert ost objects for directory because it's not exist.
1657 Details : ll_dir_getstripe assume dirrectory has ost objects but this wrong.
1661 Description: Add FMODE_EXEC file flag for SLES10 SP1 kernel.
1663 Severity : enhancement
1665 Description: Update to support 2.6.22.14 vanilla kernel.
1669 Frequency : rare, on recovery
1670 Description: read procfs can produce deadlock in some situation
1671 Details : Holding lprocfs lock which send rpc can produce block for destroy
1672 obd objects and this also block reconnect with -EALREADY. This
1673 isn't fix all lprocfs bugs - but make it rare.
1675 Severity : enhancement
1677 Description: Update kernel to RHEL5 2.6.18-53.1.14.el5.
1680 Frequency : frequent on X2 node
1682 Description: mdc_set_open_replay_data LBUG
1683 Details : Set replay data for requests that are eligible for replay.
1687 Description: lustre_mgs: operation 101 on unconnected MGS
1688 Details : When MGC is disconnected from MGS long enough, MGS will evict the
1689 MGC, and late on MGC cannot successfully connect to MGS and a lot
1690 of the error messages complaining that MGS is not connected.
1694 Frequency : on network error
1695 Description: panic with double free request if network error
1696 Details : mdc_finish_enqueue is finish request if any network error ocuring,
1697 but it's true only for synchronus enqueue, for async enqueue
1698 (via ptlrpcd) this incorrect and ptlrpcd want finish request
1701 Severity : enhancement
1703 Description: client-side metadata stat-ahead during readdir(directory readahead)
1704 Details : perform client-side metadata stat-ahead when the client detects
1705 readdir and sequential stat of dir entries therein
1708 Frequency : on start mds
1710 Description: Implement get_info(last_id) in obdfilter.
1713 Frequency : occasional
1715 Description: Correctly check stale fid, not start epoch if ost not support SOM
1716 Details : open with flag O_CREATE need set old fid in op_fid3 because
1717 op_fid2 was overwritten with new generated fid, but mds can answer
1718 with one of these two fids and both is not stale. Setattr
1719 incorrectly started an epoch and assume will be called
1720 done_writing, but without SOM done_writing ever being called.
1723 Frequency : rare, depends on device drivers and load
1725 Description: MDS or OSS nodes crash due to stack overflow
1726 Details : Code changes in 1.8.0 increased the stack usage of some functions.
1727 In some cases, in conjunction with device drivers that use a lot
1728 of stack the MDS (or possibly OSS) service threads could overflow
1729 the stack. One change which was identified to consume additional
1730 stack has been reworked to avoid the extra stack usage.
1733 Frequency : occasional
1735 Description: Do not fail import if osc_interpret_create gets -EAGAIN
1736 Details : If osc_interpret_create got -EAGAIN it immediately exits and
1737 wakeup oscc_waitq. After wakeup oscc_wait_for_objects call
1738 oscc_has_objects and see OSC has no objests and call
1739 oscc_internal_create to resend create request.
1741 Severity : enhancement
1743 Description: Update to SLES10 SP1 latest kernel-2.6.16.54-0.2.5.
1745 Severity : enhancement
1747 Description: Update to RHEL5 latest kernel-2.6.18-53.1.13.el5.
1750 Frequency : very rare
1752 Description: Fix replay if there is an un-replied request and open
1753 Details : In some cases, older replay request will revert the
1754 mcd->mcd_last_xid on MDS which is used to record the client's
1755 latest sent request.
1757 Severity : enhancement
1759 Description: Update to RHEL5 latest kernel-2.6.18-53.1.6.el5.
1761 Severity : enhancement
1763 Description: Add rhel5 support to HEAD.
1765 Serverity : enhancement
1767 Description: Update RHEL4 kernel to 2.6.9-67.0.4.
1772 Description: Don't allow skipping OSTs if index has been specified.
1773 Details : Don't allow skipping OSTs if index has been specified, make
1774 locking in internal create lots better.
1778 Description: LBUG in ptlrpc_check_set() bad phase ebc0de00
1779 Details : access to bitfield in structure is always rounded to long
1780 and this produce problem with not atomic change any bit.
1784 Description: Lustre make rpms failed.
1785 Details : Remove ldiskfs spec file to avoids rpmbuild be confused when
1786 builds Lustre rpms from tarball.
1789 Frequency : rare on shutdown ost
1791 Description: If llog cancel was not send before clean_exports phase, this can
1792 produce deadlock in llog code.
1793 Details : If llog thread has last reference to obd and call class_import_put
1794 this produce deadlock because llog_cleanup_commit_master wait when
1795 last llog_commit_thread exited, but this never success because was
1796 called from llog_commit_thread.
1800 Description: allow userland application know is lost one of stripes.
1801 Details : fill lvb_blocks with error code on ost and return it to
1802 application if error flag found.
1806 Description: NULL lov_tgts causing MDS oops
1807 Details : more safe checks for NULL lov_tgts for avoid oops.
1809 Severity : enhancement
1811 Description: Update to RHEL4 latest kernel-2.6.9-67.0.1.EL.
1815 Descriptoin: make lov_create() will not stuck in obd_statfs_rqset()
1816 Details : If an OST is down the MDS will hang indefinitely in
1817 obd_statfs_rqset() waiting for the statfs data. While for
1818 MDS QOS usage of statfs, it should not stuck in waiting.
1820 Severity : enhancement
1822 Description: remote_acl support
1823 Details : Support ACL-based permission check for remote user.
1824 Support setfacl/getfacl for remote user with the utils
1825 "lfs {l,r}{s,g}etfacl" which follow the same parameter format as
1826 the system "{s,g}etfacl" utils.
1828 Severity : enhancement
1830 Description: Update to RHEL4 U6 kernel-2.6.9-67.EL.
1832 Severity : enhancement
1834 Description: Update to RHEL5 latest kernel-2.6.18-53.1.4.el5.
1838 Description: make mgs_setparam() handle fsname containing dash
1839 Details : fsname containing a dash does not work with lctl conf_param
1841 Severity : enhancement
1843 Description: Update to SLES10 SP1 latest kernel-2.6.16.54-0.2.3.
1845 Severity : enhancement
1847 Description: Update to RHEL5 Update-1 kernel 2.6.18-53.el5.
1851 Frequency : rare, at shutdown
1852 Description: access already free / zero obd_namespace.
1853 Details : if client_disconnect_export was called without force flag set,
1854 and exist connect request in flight, this can produce access to
1855 NULL pointer (or already free pointer) when connect_interpret
1856 store ocd flags in obd_namespace.
1860 Frequency : only at startup
1861 Description: not alloc memory with spinlock held.
1862 Details : allocation memory with GFP_KERNEL can produce sleep deadlock,
1863 if any spinlock held.
1865 Severity : enhancement
1867 Description: make lustre randomly fail allocating memory
1868 Details : Make lustre randomly failed allocating memory for testing purpose.
1870 Severity : enhancement
1872 Description: lost problems with lov objid file
1873 Details : Fixes some scability and access to not inited memory problems
1874 in work with lov objdid file.
1879 Description: lfs find does not continue on file error
1880 Details : Continue other files processing when a file/dir is absent.
1884 Description: Inconsistent usage of lustre_pack_reply()
1885 Details : Standardize the usage of lustre_pack_reply() such that it
1886 always generate a CERROR on failure.
1889 Frequency : occasional
1891 Description: MDS hang or stay in waiting lock
1892 Details : If client receive lock with CBPENDING flag ldlm need send lock
1893 cancel as separate rpc, to avoid situation when cancel request
1894 can't processed due all i/o threads stay in wait lock.
1898 Description: Update to RHEL5 kernel 2.6.18-8.1.15.el5.
1902 Description: Update to SLES10 SP1 kernel 2.6.16.53-0.16
1906 Description: Update to SLES9 kernel-2.6.5-7.287.3.
1910 Description: Update to RHEL4 latest kernel.
1912 Severity : enhancement
1914 Description: Build SLES10 patchless client fails
1915 Details : The configure was broken by run ./configure with
1916 --with-linux-obj=.... argument for patchless client. When the
1917 configure use --with-linux-obj, the LINUXINCLUDE= -Iinclude
1918 can't search header adequately. Use absolute path such as
1919 -I($LINUX)/include instead.
1923 Description: interrupt oig_wait produce painc on resend.
1924 Details : brw_redo_request can be used for resend requests from ptlrpcd and
1925 private set, and this produce situation when rq_ptlrpcd_data not
1926 copyed to new allocated request and triggered LBUG on assert
1927 req->rq_ptlrpcd_data != NULL. But this member used only for wakeup
1928 ptlrpcd set if request is changed and can be safety changed to use
1933 Description: LASSERT_{REQ,REP}SWAB macros are buggy
1934 Details : If SWAB_PARANOIA is disabled, the LASSERT_REQSWAB and
1935 LASSERT_REPSWAB macros become no-ops, which is incorrect. Drop
1936 these macros and replace them with their difinitions instead.
1940 Description: Update kernel patches for SLES10 2.6.16.53-0.8.
1941 Details : Update which_patch & target file for SLES10 latest kernel.
1944 Description: Remove client patches from SLES 10 kernel.
1945 Details : This causes SLES 10 clients to behave as patchless clients
1946 even on a Lustre-patched (server) kernel.
1948 Severity : enhancement
1950 Description: self-adjustable client's lru lists
1951 Details : use adaptive algorithm for managing client cached locks lru
1952 lists according to current server load, other client's work
1953 pattern, memory activities, etc. Both, server and client
1954 side namespaces provide number of proc tunables for controlling
1957 Severity : enhancement
1959 Description: light-weight GSS support
1960 Details : Support krb5n and krb5a mode, which keep Kerberos 5 authentication
1961 and reduce performance overhead.
1963 Severity : enhancement
1965 Description: Linux keyring support
1966 Details : Support using service of Linux keyring for Lustre GSS internal
1967 context refresh/cache mechanism.
1971 Description: Fix errors in lfs documentation
1972 Details : Fixes man pages
1976 Description: don't use GFP_* in generic Lustre code.
1977 Details : Use cfs_alloc_* functions and CFS_* flags for code portability.
1981 Description: obdclass is limited by single OBD_ALLOC(idarray)
1982 Details : replace OBD_ALLOC/OBD_FREE with OBD_VMALLOC/OBD_VFREE
1986 Description: warnings with build patchless client with vanila 2.6.19 and up
1987 Details : change old ctl_table style and replace ctl_table/ctl_table_header
1988 with cfs_sysctl_table_t/cfs_sysctl_table_header_t
1992 Frequency : Only for SLES
1993 Description: sanity_quota fail test_1
1994 Details : There are multiple occurences of $TSTUSR in SLES's /etc/group
1995 file, which makes TSTID[2] inunique.
1999 Frequency : Only for SLES9
2000 Description: Kernel patches for SLES9 2.6.5-7.286 kernel
2001 Details : Update target/ChangeLog/which_patch .
2005 Frequency : Only for test_10 in sanity-quota.sh
2006 Description: a bug in quota test code
2007 Details : checking a test flag in a wrong place
2011 Frequency : Only for test_13 in sanity-quota.sh
2012 Description: a bug in quota test script
2013 Details : shouldn't check overfull allocated quota
2017 Description: jbd statistics
2018 Details : Port older jbd statistics patch for sles10
2020 Severity : enhancement
2022 Description: Add printing inode info into message about error in writepage.
2027 Description: reply_lock_interpret crash due to race with it and lock cancel.
2028 Details : Do not replay locks that are being cancelled. Do not reference
2029 locks by their address during replay, just by their handle.
2033 Frequency : When flocks are used.
2034 Description: assertion failure in ldlm_cli_enquque_fini for non NULL lock.
2035 Details : Flock locks might destroy just granted lock if it could be merged
2036 with another existing flock, this is done in completion handler,
2037 so teach ldlm_cli_enquque_fini that this is a valid case for
2043 Description: Oops in read and write path when failing to allocate lock.
2044 Details : Check if lock allocation failed and return error back.
2048 Description: lstripe command fails for valid OST index
2049 Details : The stripe offset is compared to lov->desc.ld_tgt_count
2050 instead of lov->desc.ld_active_tgt_count.
2054 Description: sanity.sh failed test 103
2055 Details : RHEL mis-interpret setfacl "-X" param, so we won't test setfacl
2060 Description: df doesn't work properly if diskfs blocksize != 4K
2061 Details : Choose biggest blocksize of OST's as the LOV's blocksize.
2065 Description: lfs find on -1 stripe looping in lsm_lmm_verify_common()
2066 Details : Avoid lov_verify_lmm_common() on directory with -1 stripe count.
2068 Severity : enhancement
2070 Description: Adaptive timeouts
2071 Details : RPC timeouts adapt to changing server load and network
2072 conditions to reduce resend attempts and improve recovery time.
2076 Description: llapi_file_create() does not allow some changes
2077 Details : add llapi_file_open() that allows specifying the mode and
2078 open flags, and also returns an open file handle.
2082 Description: merge and cleanup kernel patches.
2083 Details : Remove mnt_lustre_list in vfs_intent-2.6-rhel4.patch.
2087 Description: Add journal checksum support.(Kernel part)
2088 Details : The journal checksum feature adds two new flags i.e
2089 JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT and
2090 JBD2_FEATURE_COMPAT_CHECKSUM. JBD2_FEATURE_CHECKSUM flag
2091 indicates that the commit block contains the checksum for
2092 the blocks described by the descriptor blocks. Now commit
2093 record can be sent to disk without waiting for descriptor
2094 blocks to be written to disk. This behavior is controlled
2095 using JBD2_FEATURE_ASYNC_COMMIT flag.
2099 Description: OSS needs mutliple precreate threads
2100 Details : Add ability to start more than one create thread per OSS.
2104 Description: Sanity.sh test_65k failed: setstripe should have succeeded
2105 Details : execute lfs setstripe on client
2109 Description: mds_obd_create error creating tmp object
2110 Details : When the user sets quota on root, llog will be affected and can't
2111 create files and write files.
2114 Frequency : Always on ia64 patchless client, and possibly others.
2116 Description: Add EXPORT_SYMBOL check for node_to_cpumask symbol.
2117 Details : This allows the patchless client to be loaded on architectures
2118 without this export.
2122 Description: RedHat Update kernel for RHEL5
2123 Details : Add the kernel config file for RHEL5.
2127 Description: RedHat Update kernel for RHEL5
2128 Details : Modify the kernel config file more closer RHEL5.
2132 Description: Build failure against Centos5 (RHEL5)
2133 Details : Define PAGE_SIZE when it isn't present.
2137 Description: test_7 on sanity-quota.sh can't be used on separate machines
2138 Details : add facet to handle it
2142 Description: "ll_intent_file_open()) lock enqueue: err: -13" with nfs
2143 Details : with NFS, the anon dentry's parent was set to itself in
2144 d_alloc_anon(), so in MDS, we use rec->ur_fid1 to find the
2145 corresponding dentry other than use rec->ur_name.
2147 Severity : enhancement
2149 Description: lfs setstripe enhancement
2150 Details : Make lfs setstripe understand 'k', 'm' and 'g' for stripe size.
2154 Description: enable data checksumming by default
2155 Details : enable checksum by default, allow --disable-checksum
2156 configure option and "-o nochecksum" mount option. Checksums
2157 can also be disabled at runtime via $LPROC/osc/*/checksum_pages.
2161 Description: lustre support for RHEL5
2162 Details : Add support for RHEL5.
2166 Description: Client eviction due to failover config
2167 Details : after a connection loss, the lustre client should attempt to
2168 reconnect to the last active server first before trying the
2169 other potential connections.
2173 Description: when mds and osts use different quota unit(32bit and 64bit),
2174 quota will be released repeatly.
2175 Details : void sending multiple quota reqs to mds, which will keep the
2176 status between the reqs.
2180 Description: osts not allocated evenly to files
2181 Details : change the condition to increase offset_idx
2184 Frequency : Always for filesystems larger than 2TB on 32-bit systems.
2186 Description: Data corruption for OSTs that are formatted larger than 2TB
2188 Details : When generating the bio request for lustre file writes the
2189 sector number would overflow a temporary variable before being
2190 used for the IO. The data reads correctly from Lustre (which
2191 will overflow in a similar manner) but other file data or
2192 filesystem metadata may be corrupted in some cases.
2196 Description: Tune the kernel for good SCSI performance.
2197 Details : Set the value of /sys/block/{dev}/queue/max_sectors_kb
2198 to the value of /sys/block/{dev}/queue/max_hw_sectors_kb
2203 Description: rewrite ext2-derived code in obdclass/uuid.c
2204 Details : rewrite inherited code (uuid parsing code from ext2 utils)
2205 from scratch preserving functionality.
2209 Description: "lfs find -obd UUID" prints directories
2210 Details : "lfs find -obd UUID" will return all directory names instead
2211 of just file names. It is incorrect because the directories
2212 do not reside on the OSTs.
2216 Description: use i_size_read and i_size_write in 2.6 port
2217 Details : replace inode->i_size access with i_size_read/write()
2220 Frequency : when removing large files
2222 Description: scheduling issue during removal of large Lustre files
2223 Details : Don't take the BKL in fsfilt_ext3_setattr() for 2.6 kernels.
2224 It causes scheduling issues when removing large files (17TB in the
2228 Frequency : only with liblustre clients on XT3
2230 Description: evictions taking too long
2231 Details : allow llrd to evict clients directly on OSTs
2234 Frequency : only on ppc
2236 Description: /proc/fs/lustre/devices broken on ppc
2237 Details : The patch as applied to 1.6.2 doesn't look correct for all arches.
2238 We should make sure the type of 'index' is loff_t and then cast
2239 explicitly as needed below. Do not assign an explicitly cast
2243 Frequency : only for rhel5
2245 Description: Kernel patches update for RHEL5 2.6.18-8.1.10.el5.
2246 Details : Modify the target file & which_kernel.
2248 Severity : enhancement
2250 Description: omit set fsid for export NFS
2251 Details : fix set/restore device id for avoid EMFILE error and mark lustre
2252 fs as FS_REQUIRES_DEV for avoid problems with generate fsid.
2256 Frequency : Always, for kernels after 2.6.16
2257 Description: Fix warning idr_remove called for id=.. which is not allocated.
2258 Details : Last kernels save old s_dev before kill super and not allow
2259 to restore from callback - restore it before call kill_anon_super.
2263 Description: buffer overruns could theoretically occur
2264 Details : llapi_semantic_traverse() modifies the "path" argument by
2265 appending values to the end of the origin string, and a
2266 overrun may occur. Adding buffer overrun check in liblustreapi.
2270 Description: Fix error on 'ls .' at the top of the Lustre mount.
2271 Details : Don't revalidate dentry if it is a root dentry.
2275 Description: Kernel patches update for RHEL4 2.6.9-55.0.6.
2276 Details : Modify vm-tunables-rhel4.patch.
2280 Description: Kernel config for 2.6.18-vanilla.
2281 Details : Modify targets/2.6-vanilla.target.in.
2282 Add config file kernel-2.6.18-2.6-vanilla-i686.config.
2283 Add config file kernel-2.6.18-2.6-vanilla-i686-smp.config.
2284 Add config file kernel-2.6.18-2.6-vanilla-x86_64.config.
2285 Add config file kernel-2.6.18-2.6-vanilla-x86_64-smp.config.
2289 Description: improve handling recoverable errors
2290 Details : if request processig with error which can be recoverable on server
2291 request should be resend, otherwise page released from cache and
2296 Description: Kernel patches update for RHEL5 2.6.18-8.1.14.el5.
2297 Details : Modify target file & which_patch.
2298 A flaw was found in the IA32 system call emulation provided
2299 on AMD64 and Intel 64 platforms. An improperly validated 64-bit
2300 value could be stored in the %RAX register, which could trigger an
2301 out-of-bounds system call table access. An untrusted local user
2302 could exploit this flaw to run code in the kernel
2303 (ie a root privilege escalation). (CVE-2007-4573).
2307 Description: O_DIRECT bypasses client statistics.
2308 Details : When running with O_DIRECT I/O, neither the client rpc_stats nor
2309 read_ahead_stats were updated. copied the stats section from
2310 osc_send_oap_rpc() into async_internal().
2314 Description: Add jbd statistics patch for RHEL5 and 2.6.18-vanilla
2318 Description: change order of libsysio includes
2319 Details : '#include sysio.h' should always come before '#include xtio.h'
2323 Description: handle "serious error: objid * already exists" more gracefully
2324 Details : If LAST_ID value on disk is smaller than the objects existing in
2325 the O/0/d* directories, it indicates disk corruption and causes an
2326 LBUG(). If the object is 0-length, then we should use the existing
2327 object. This will help to avoid a full fsck in most cases.
2329 Severity : enhancement
2331 Description: adapt the lustre_config script to support the upgrade case
2332 Details : Add "-u" option for lustre_config script to support upgrading 1.4
2333 server targets to 1.6 in parallel.
2337 Description: To avoid grant space > avaible space when the disk is almost
2338 full. Without this patch you might see the error "grant XXXX >
2339 available" or some LBUG about grant, when the disk is almost
2341 Details : In filter_check_grant, for non_grant cache write, we should
2342 check the left space by if (*left > ungranted + bytes), instead
2343 of (*left > ungranted), because only we are sure the left space
2344 is enough for another "bytes", then the ungrant space should be
2345 increase. In client, we should update cl_avail_grant only there
2346 is OBD_MD_FLGRANT in the reply.
2350 Description: Update RHEL 4 kernel to fix local root privilege escalation.
2351 Details : Update to the latest RHEL 4 kernel to fix the vulnerability
2352 described in CVE-2007-4573. This problem could allow untrusted
2353 local users to gain root access.
2356 Frequency : when using O_DIRECT and quotas
2358 Description: Incorrect file ownership on O_DIRECT output files
2359 Details : block usage reported by 'lfs quota' does not take into account
2360 files that have been written with O_DIRECT.
2365 Description: touch file failed when fs is not full
2366 Details : OST in recovery should not be discarded by MDS in alloc_qos(),
2367 otherwise we can get ENOSP while fs is not full.
2371 Description: parallel lock callbacks
2372 Details : Instead of sending blocking and completion callbacks as separated
2373 requests, adding them to a set and sending in parallel.
2376 Frequency : only for Cray XT3
2377 Bugzilla : 12829/13455
2378 Description: Changing primary group doesn't change the group lustre assigns to
2380 Details : When CRAY_XT3 is defined, the fsgid supplied by the client is
2381 overridden with the primary group provided by the group upcall,
2382 whereas the supplied fsgid can be trusted if it is in the list of
2383 supplementary groups returned by the group upcall.
2385 Severity : enhancement
2387 Description: Allow masking D_WARNING, D_ERROR messages from console
2388 Details : Console messages can now be disabled via lnet.printk.
2392 Description: User code with malformed file open parameter crashes client node
2393 Details : Before packing join_file req, all the related reference should be
2394 checked carefully in case some malformed flags cause fake
2395 join_file req on client.
2399 Description: LDLM_ENQUEUE races with LDLM_CP_CALLBACK
2400 Details : ldlm_completion_ast() assumes that a lock is granted when the req
2401 mode is equal to the granted mode. However, it should also check
2402 that LDLM_FL_CP_REQD is not set.
2406 Description: Heavy nfs access might result in deadlocks
2407 Details : After ELC code landed, it is now improper to enqueue any mds
2408 locks under och_sem, because enqueue might want to decide to
2409 cancel open locks for same inode we are holding och_sem for.
2413 Description: Client eviction while running blogbench
2414 Details : A lot of unlink operations with concurrent I/O can lead to a
2415 deadlock causing evictions. To address the problem, the number of
2416 oustanding OST_DESTROY requests is now throttled to
2417 max_rpcs_in_flight per OSC and LDLM_FL_DISCARD_DATA blocking
2418 callbacks are processed in priority.
2422 Description: enable ACLs on MDS by default
2423 Details : ACLs must be enabled on MDS by default.
2426 Frequency : PPC/PPC64 only
2428 Description: conflicts between asm-ppc64/types.h and lustre_types.h
2429 Details : fix duplicated definitions between asm-ppc64/types.h and
2430 lustre_types.h on PPC.
2433 Frequency : PPC/PPC64 only
2435 Description: asm-ppc/segment.h does not exist
2436 Details : fix compile issue on PPC.
2440 Description: better handle error messages in extents code
2443 Frequency : RHEL4 only
2445 Description: mkfs is very slow on IA64/RHEL4
2446 Details : A performance regression has been discovered in the MPT Fusion
2447 driver between versions 3.02.73rh and 3.02.99.00rh. As a
2448 consequence, we have downgraded the MPT Fusion driver in the RHEL4
2449 kernel from 3.02.99.00 to 3.02.73 until this problem is fixed.
2451 Severity : enhancement
2453 Description: SNMP support enhancement
2454 Details : Adding total number of sampled request for an MDS node in snmp
2457 Severity : enhancement
2459 Description: Optimize ldlm waiting list processing for PR extent locks
2460 Details : When processing waiting list for read extent lock and meeting read
2461 lock that is same or wider to it that is not contended, skip
2462 processing rest of the list and immediatelly return current
2463 status of conflictness, since we are guaranteed there are no
2464 conflicting locks in the rest of the list.
2468 Description: Time out and refuse to reconnect
2469 Details : When the failover node is the primary node, it is possible
2470 to have two identical connections in imp_conn_list. We must
2471 compare not conn's pointers but NIDs, otherwise we can defeat
2472 connection throttling.
2476 Description: port llog fixes from b1_6 into HEAD
2477 Details : Port llog reference couting and some llog cleanups from b1_6
2478 (bug 10800) into HEAD, for protect from panic and access to already
2479 free llog structures.
2483 Description: Detect stride IO mode in read-ahead
2484 Details : When a client does stride read, read-ahead should detect that and
2485 read-ahead pages according to the detected stride pattern.
2489 Description: data checksumming impacts single node performance
2490 Details : add support for several checksum algorithm. Currently, only CRC32
2491 and Adler-32 are supported. The checksum type can be changed on
2492 the fly via /proc/fs/lustre/osc/*/checksum_type.
2496 Description: use adler32 for page checksums
2497 Details : when available, use the Adler-32 algorithm instead of CRC32 for
2502 Description: build for x2 fails
2503 Details : fix compile issue on Cray systems.
2507 Description: Properly match for duplicate locks
2508 Details : Due to different lock order from skiplists code, we need to
2509 traverse entire list for now
2512 Frequency : only on PPC/SLES10
2514 Description: "BITS_PER_LONG is not 32 or 64" in linux/idr.h
2515 Details : On SLES10/PPC, fs.h includes idr.h which requires BITS_PER_LONG to
2516 be defined. Add a hack in mkfs_lustre.c to work around this compile
2521 Description: LASSERT on MDS when client holding flock lock dies
2522 Details : ldlm pool logic depends on number of granted locks equal to
2523 number of released locks which is not true for flock locks, so
2524 just exclude such locks from consideration.
2528 Description: MDS deadlock with many ll_sync_lov threads and I/O stalled
2529 Details : Use fsfilt_sync() for both the whole filesystem sync and
2530 individual file sync to eliminate dangerous inode locking
2531 with I_LOCK that can lead to a deadlock.
2535 Description: performance in 1.6.3
2536 Details : Force q->max_phys_segments to MAX_PHYS_SEGMENTS on SLES10 to be
2537 sure that 1MB requests are not fragmented by the block layer.
2539 Severity : enhancement
2541 Description: organize the server-side client stats on per-nid basis
2542 Details : Change the structure of stats under obdfilter and mds to
2550 The "uuid"s file would list the uuids of _active_ exports.
2551 And the clear entry is to clear all stats and stale nids.
2553 Severity : enhancement
2555 Description: eliminate client locks in face of contention
2556 Details : file contention detection and lockless i/o implementation
2557 for contended files.
2561 Description: Reinitialize optind to 0 so that interactive lfs works in all cases
2564 Frequency : very rare, if additional xattrs are used on kernels >= 2.6.12
2566 Description: MDS may lose file striping (and hence file data) in some cases
2567 Details : If there are additional extended attributes stored on the MDS,
2568 in particular ACLs, SELinux, or user attributes (if user_xattr
2569 is specified for the client mount options) then there is a risk
2570 of attribute loss. Additionally, the Lustre file striping
2571 needs to be larger than default (e.g. striped over all OSTs),
2572 and an additional attribute must be stored initially in the
2573 inode and then increase in size enough to be moved to the
2574 external attribute block (e.g. ACL growing in size) for file
2579 Description: skiplist implementation simplification
2580 Details : skiplists are used to group compatible locks on granted list
2581 that was implemented as tracking first and last lock of each
2582 lock group the patch changes that to using doubly linked lists
2586 Description: MDS LBUG: ASSERTION(!IS_ERR(dchild))
2587 Details : Change LASSERTs to client eviction (i.e. abort client's recovery)
2588 because LASSERT on both the data supplied by a client, and the
2589 data on disk is dangerous and incorrect.
2591 Severity : enhancement
2593 Description: Slow truncate/writes to huge files at high offsets.
2594 Details : Directly associate cached pages to lock that protect those pages,
2595 this allows us to quickly find what pages to write and remove
2596 once lock callback is received.
2600 Description: more ldlm soft lockups
2601 Details : In ldlm_resource_add_lock(), call to ldlm_resource_dump()
2602 starve other threads from the resource lock for a long time in
2603 case of long waiting queue, so change the debug level from
2604 D_OTHER to the less frequently used D_INFO.
2606 Severity : enhancement
2608 Description: add -gid, -group, -uid, -user options to lfs find
2612 Description: Hung threads in invalidate_inode_pages2_range
2613 Details : The direct IO path doesn't call check_rpcs to submit a new RPC
2614 once one is completed. As a result, some RPCs are stuck in the
2615 queue and are never sent.
2619 Description: filter threads hungs on waiting journal commit
2620 Details : Cleanup filter group llog code, then only filter group llog will
2621 be only created in the MDS/OST syncing process.
2625 Description: Procfs and llog threads access destoryed import sometimes.
2626 Details : Sync the import destoryed process with procfs and llog threads by
2627 the import refcount and semaphore.
2629 Severity : enhancement
2631 Description: openlock cache of b1_6 port to HEAD
2636 Description: kernel BUG at ldiskfs2_ext_new_extent_cb
2637 Details : If insertion of an extent fails, then discard the inode
2638 preallocation and free data blocks else it can lead to duplicate
2643 Description: don't always update ctime in ext3_xattr_set_handle()
2644 Details : Current xattr code updates inode ctime in ext3_xattr_set_handle.
2645 In some cases the ctime should not be updated, for example for
2646 2.0->1.8 compatibility it is necessary to delete an xattr and it
2647 should not update the ctime.
2651 Bugzilla : 15713/16362
2652 Description: Assertion in iopen_connect_dentry in 1.6.3
2653 Details : looking up an inode via iopen with the wrong generation number can
2654 populate the dcache with a disconneced dentry while the inode
2655 number is in the process of being reallocated. This causes an
2656 assertion failure in iopen since the inode's dentry list contains
2657 both a connected and disconnected dentry.
2661 Description: assertion failure in ldlm_handle2lock()
2662 Details : fix a race between class_handle_unhash() and class_handle2object()
2663 introduced in lustre 1.6.5 by bug 13622.
2668 Description: Kernel BUG: sd_iostats_bump: unexpected disk index
2669 Details : remove the limit of 256 scsi disks in the sd_iostat patch
2674 Description: oops in sd_iostats_seq_show()
2675 Details : unloading/reloading the scsi low level driver triggers a kernel
2676 bug when trying to access the sd iostat file.
2681 Description: Kernel panics during QLogic driver reload
2682 Details : REQ_BLOCK_PC requests are not handled properly in the sd iostat
2683 patch, causing memory corruption.
2688 Description: journal_dev option does not work in b1_6
2689 Details : pass mount option during pre-mount.
2691 Severity : enhancement
2693 Description: Add a FIEMAP(FIle Extent MAP) ioctl
2694 Details : FIEMAP ioctl will allow an application to efficiently fetch the
2695 extent information of a file. It can be used to map logical blocks
2696 in a file to physical blocks in the block device.
2700 Description: LDLM soft lockups - improvement
2701 Details : It is be possible to send the lock handle along with each read
2702 or write request because the client is already doing a lock match
2703 itself so there isn't any reason the OST should have to re-do that
2707 Frequency : only on Cray X2
2709 Description: X2 build failures
2710 Details : fix build failures on Cray X2.
2714 Description: xid & resent requests
2715 Details : Initialize RPC XID from clock at startup (randomly if clock is
2718 Severity : enhancement
2720 Description: Add lustre_start utility to start or stop multiple Lustre servers
2725 Description: Lustre GPF in {:ptlrpc:ptlrpc_server_free_request+373}
2726 Details : In case of memory pressure, list_del() can be called twice on
2727 req->rq_history_list, causing a kernel oops.
2731 Description: kptllnd_peer_check_sends()) ASSERTION(!in_interrupt()) failed
2732 Details : fix stack overflow in the distributed lock manager by defering
2733 export eviction after a failed AST to the elt thread instead of
2734 handling it in the dlm interpret routine.
2738 Description: Convert some comments to new format.
2739 Details : Update documenting comments to match doxygen conventions.
2743 Description: Grammar fixes.
2744 Details : A couple of trivial spelling fixes.
2748 Description: OSD_COUNTERS-mandatory
2749 Details : Make previously optional ->oti_{w,r}_locks sanity checks mandatory
2750 to simplify the code.
2754 Description: simplify cmm_device freeing logic.
2755 Details : Call cmm_device_free() in the failure path of cmm_device_alloc().
2759 Description: Add lockdep support to dt_object_operations locking interface.
2760 Details : Augment ->do_{read,write}_lock() prototypes with a `role'
2761 parameter indicating lock ordering. Update mdd code to use new
2766 Description: Introduce failloc constants for lockless IO tests.
2767 Details : Add two new failloc constants to test lockless IO. Only one of
2768 them in implemented---another is checked in yet to be landed
2773 Description: Add lockdep support for inode mutex.
2774 Details : Introduce and use new LOCK_INODE_MUTEX_PARENT() macro to be used
2775 in the situations where Lustre has to lock more than one inode
2780 Description: Add optional invariants checking support.
2781 Details : Add new LINVRNT() macro, optional on new --enable-invariants
2782 configure switch. This macro is to be used for consistency and
2783 sanity checks that are too expensive to be left in `production'
2788 Description: Zap lock->l_granted_mode with explicit LCK_MINMODE.
2789 Details : Use LCK_MINMODE rather than 0 to reset lock->l_granted_mode to
2794 Description: Add lockdep support for ldlm_lock and ldlm_resource.
2795 Details : Use spin_lock_nested() in (the only) situation where more than
2796 one ldlm_lock is locked simultaneously. Also, fix possible
2797 dead-lock in ldlm_lock_change_resource() by enforcing particular
2802 Description: Use struct ldlm_callback_suite in ldlm_lock_create().
2803 Details : Instead of specifying each ldlm_lock call-back through separate
2804 parameter, wrap them into struct ldlm_callback_suite.
2808 Description: Kill join_lru obd method and its callers.
2809 Details : CLIO uses lock weighting policy to keep locks over mmapped regions
2810 in memory---a requirement implemented through ->o_join_lru() obd
2811 method in HEAD. Remove this method and its users.
2815 Description: Add asynchronous ldlm ENQUEUE completion handler.
2816 Details : CLIO posts ENQUEUE requests asynchronously through ptlrpcd---a
2817 case that stock ldlm_completion_ast() cannot handle as it waits
2818 until lock is granted. Introduce new ldlm_completion_ast_async()
2819 for this. Also comment ldlm_completion_ast().
2823 Description: ldlm_error <-> errno conversion.
2824 Details : Add functions to map (rather arbitrary) between LDLM error codes
2825 and standard errno values. CLIO needs this to prevent LDLM specific
2826 constants from escaping ldlm and osc.
2830 Description: Kill unused ldlm_handle2lock_ns() function.
2831 Details : Kill unused ldlm_handle2lock_ns() function.
2835 Description: Add lu_ref support to ldlm_lock
2836 Details : lu_ref support for ldlm_lock and ldlm_resource. See lu_ref patch.
2837 lu_ref fields ->l_reference and ->lr_reference are added to ldlm_lock
2838 and ldlm_resource. LDLM interface has to be changed, because code that
2839 releases a reference on a lock, has to "know" what reference this is.
2840 In the most frequent case
2842 lock = ldlm_handle2lock(handle);
2844 LDLM_LOCK_PUT(lock);
2846 no changes are required. When any other reference (received _not_ from
2847 ldlm_handle2lock()) is released, LDLM_LOCK_RELEASE() has to be called
2848 instead of LDLM_LOCK_PUT().
2850 Arguably, changes are pervasive, and interface requires some discipline
2851 for proper use. On the other hand, it was very instrumental in finding
2852 a few leaked lock references.
2856 Description: Add ldlm_lock_addref_try().
2857 Details : Introduce ldlm_lock_addref_try() function (used by CLIO) that
2858 attempts to addref a lock that might be being canceled
2863 Description: Add ldlm_weigh_callback().
2864 Details : Add new ->l_weigh_ast() call-back to ldlm_lock. It is called
2865 by ldlm_cancel_shrink_policy() to estimate lock "value", instead of
2866 hard-coded `number of pages' logic.
2870 Description: Add lockdep annotations to llog code.
2871 Details : Use appropriately tagged _nested() locking calls in the places
2872 where llog takes more than one ->lgh_lock lock.
2876 Description: Add loi_kms_set().
2877 Details : Wrap kms updates into a helper function.
2881 Description: Constify instances of struct lsm_operations.
2882 Details : Constify instances of struct lsm_operations.
2886 Description: lu_conf support.
2887 Details : On a server, a file system object is uniquely identified
2888 by a fid, which is sufficient to locate and load all object
2889 state (inode). On a client, on the other hand, more data are
2890 necessary instantiate an object. Change lu_object_find() and
2891 friends to take additional `lu_conf' argument describing object.
2892 Typically this includes layout information.
2896 Description: lu_context fixes.
2897 Details : Introduce new lu_context functions that are needed on the client
2898 side, where some system threads (ptlrpcd) are shared by multiple
2899 modules, and so cannot be stopped during module shutdown.
2903 Description: Add start and stop methods to lu_device_type_operations.
2904 Details : Introduce two new methods in lu_device_type_operations, that are
2905 invoked when first instance of a given type is created and last one
2906 is destroyed respectively. This is need by CLIO.
2910 Description: Add lu_ref support to struct lu_device.
2911 Details : Add lu_ref support to lu_object and lu_device. lu_ref is used to
2912 track leaked references.
2916 Description: Introduce lu_kmem_descr.
2917 Details : lu_kmem_descr and its companion interface allow to create
2918 and destroy a number of kmem caches at once.
2922 Description: Fix lu_object finalization race.
2923 Details : Fix a race between lu_object_find() finding an object and its
2924 concurrent finalization. This race is (most likely) not possible
2925 on the server, but might happen on the client.
2929 Description: Introduce lu_ref interface.
2930 Details : lu_ref is a debugging module allowing to track references to
2931 a given object. It is quite cpu expensive, and has to be
2932 explicitly enabled with --enable-lu_ref. See usage description
2937 Description: Factor lu_site procfs stats into a separate function.
2938 Details : Separate lu_site stats printing code into a separate function
2939 that can be reused on a client.
2943 Description: Constify instances of struct {lu,dt,md}_device_operations.
2944 Details : Constify instances of struct {lu,dt,md}_device_operations.
2948 Description: Introduce struct md_site and move meta-data specific parts of
2949 struct lu_site here.
2950 Details : Move md-specific fields out of struct lu_site into special struct
2951 md_site, so that lu_site can be used on a client.
2955 Description: Kill mdd_lov_destroy().
2956 Details : Remove unused mdd code.
2960 Description: Add st_block checking to multistat.c.
2961 Details : Add st_block checking to multistat.c.
2965 Description: Add lu_ref support to struct obd_device.
2966 Details : Add lu_ref tracking to obd_device.
2970 Description: Kill obd_set_fail_loc().
2971 Details : Remove unused code.
2975 Description: Add special type for ptlrpc_request interpret functions.
2976 Details : Add lu_env parameter to ->rq_interpreter call-back. NULL is passed
2977 there. Actual usage will be in CLIO.
2981 Description: Replace RW_LOCK_UNLOCKED() macro with rwlock_init().
2982 Details : Replace RW_LOCK_UNLOCKED() with rwlock_init() as the former
2983 doesn't work with lockdep.
2987 Description: Add rwv.c test program.
2988 Details : New testing program exercising readv(2) and writev(2) (Qian).
2992 Description: Add sendfile.c test program.
2993 Details : New testing program exercising sendfile(2) (Jay).
2997 Description: Ratelimit a message that can be very frequent.
2998 Details : Ratelimit a memory allocation failure message that can
3003 Description: Use cdebug_show() in CDEBUG-style macros defined outside of libcfs.
3004 Details : Use cdebug_show() in CDEBUG-style macros defined outside of libcfs.
3008 Description: Liblustre build fixes.
3009 Details : Liblustre build fixes.
3013 Description: libcfs: add cfs_{need,cond}_resched() interface.
3014 Details : libcfs: add cfs_{need,cond}_resched() definition and
3015 implementations for Linux, NT, and liblustre.
3017 Severity : enhancement
3019 Description: More exported tunables for mballoc
3020 Details : Add support for tunable preallocation window and new tunables for
3021 large/small requests
3025 Description: Detect corruption of block bitmap and checking for preallocations
3026 Details : Checks validity of on-disk block bitmap. Also it does better
3027 checking of number of applied preallocations. When corruption is
3028 found, it turns filesystem readonly to prevent further corruptions.
3032 Description: (rw.c:1323:ll_read_ahead_pages()) ASSERTION(page_idx > ria->ria_stoff) failed
3033 Details : Once the unmatched stride IO mode is detected, shrink the stride-ahead
3034 window to 0. If it does hit cache miss, and read-pattern is still
3035 stride-io mode, does not reset the stride window, but also does not
3036 increase the stride window length in this case.
3040 Frequency : only for big-endian servers
3041 Description: Check if system is big-endian while mounting fs with extents feature
3042 Details : Mounting a filesystem with extents feature will fail on big-endian
3043 systems since ext3-based ldiskfs is not supported on big-endian
3044 systems. This can be over-riden with "bigendian_extents" mount option.
3046 Severity : enhancement
3048 Description: The root squash functionality
3049 Details : A security feature, which is to prevent users from being able
3050 to mount lustre on their desktop, run as root, and delete
3051 all of the files in the filesystem. The goal is accomplished by
3052 remapping user id (UID) and group id (GID) of the root user to
3053 a UID and GID specified by the system administartor via Lustre
3054 configuration management server (MGS). The functionality also
3055 allows to specify sets of clients for which the remapping does
3060 Description: Excessive recovery window
3061 Details : With AT enabled, the recovery window can be excessively long (6000+
3062 seconds). To address this problem, we no longer use
3063 OBD_RECOVERY_FACTOR when extending the recovery window (the connect
3064 timeout no longer depends on the service time, it is set to
3065 INITIAL_CONNECT_TIMEOUT now) and clients report the old service
3066 time via pb_service_time.
3070 Description: Watchdog triggered on MDS failover
3071 Details : enable OBD_CONNECT_MDT flag when connecting from the MDS so that
3072 the OSTs know that the MDS "UUID" can be reused for the same export
3073 from a different NID, so we do not need to wait for the export to be
3077 Frequency : rare, only if using MMP with Linux RAID
3079 Description: MMP doesn't work with Linux RAID
3080 Details : While using HA for Lustre servers with Linux RAID, it is possible
3081 that MMP will not detect multiple mounts. To make this work we
3082 need to unplug the device queue in RAID when the MMP block is being
3083 written. Also while reading the MMP block, we should read it from
3084 disk and not the cached one.
3086 Severity : enhancement
3088 Description: open file using fid
3089 Details : A file can be opened using just its fid, like
3090 <mntpt>/.lustre/fid/SEQ:OID:VER - this is needed for HSM and replication
3093 Frequency : Only in RHEL5 when mounting multiple ext3 filesystems
3096 Description: "kmem_cache_create: duplicate cache jbd_4k" error message
3097 Details : add proper locking for creation of jbd_4k slab cache
3101 Description: MMP check in ext3_remount() fails without displaying any error
3102 Details : When multiple mount protection fails during remount, proper error
3105 Severity : enhancement
3107 Description: Allow stripe size to be up to 4G-64k
3108 Details : Fix math logic to allow large stripe sizes.
3112 Description: add check for >8TB ldiskfs filesystems
3113 Details : ext3-based ldiskfs does not support greater than 8TB LUNs.
3114 Don't allow >8TB ldiskfs filesystems to be mounted without
3115 force_over_8tb mount option
3120 Descriptoin: OSS complains "too many blocks" and crashes
3121 Details : In filter_commitrw_write():
3122 1. lquota_chkquota();
3123 2. obdo_from_inode();
3124 3. lquota_pending_commit();
3125 "2." has changed uid and gid in this case, then caused
3126 lquota_chkquota() and lquota_pending_commit() to use
3127 different uid and gid. That is the root of the bug.
3131 Descriptoin: getstripe doesn't report stripe count = -1 sanely
3132 Details : Detect and print out the 'stripe count = -1' case properly.
3136 Description: Out or order replies might be lost on replay
3137 Details : In ptlrpc_retain_replayable_request if we cannot find retained
3138 request with tid smaller than one currently being added, add it
3139 to the start, not end of the list.
3141 --------------------------------------------------------------------------------
3143 2007-08-10 Cluster File Systems, Inc. <info@clusterfs.com>
3145 * Support for kernels:
3146 2.6.5-7.283 (SLES 9),
3147 2.6.9-55.EL (RHEL 4),
3148 2.6.16.46-0.14 (SLES 10),
3149 2.6.18.8 vanilla (kernel.org)
3150 * Client support for unpatched kernels:
3151 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
3152 2.6.16 - 2.6.22 vanilla (kernel.org)
3153 * Due to recently discovered recovery problems, we do not recommend
3154 using patchless RHEL 4 clients with this or any earlier release.
3155 * Recommended e2fsprogs version: 1.39.cfs8
3156 * Note that reiserfs quotas are disabled on SLES 10 in this kernel.
3157 * Starting with this release, the ldiskfs backing filesystem required
3158 by Lustre is now in its own package, lustre-ldiskfs. This package
3159 should be installed. It is versioned separately from Lustre and
3160 may be released separately in future.
3164 Description: block reactivating mgc import until all deactivates complete
3165 Details : Fix race when failing back MDT/MGS to itself (testing)
3168 Frequency : at statup only
3170 Description: mds_lov_synchronize race leads to various problems
3171 Details : simultaneous MDT->OST connections at startup can cause the
3172 sync to abort, leaving the OSC in a bad state.
3174 Severity : enhancement
3176 Description: add optional extra BUILD_VERSION info
3177 Details : add a new environment variable (namely LUSTRE_VERS) which allows
3178 to override the lustre version.
3181 Frequency : 2.6.18 servers only
3183 Description: ll_kern_mount() doesn't release the module reference
3184 Details : The ldiskfs module reference count never drops down to 0
3185 because ll_kern_mount() doesn't release the module reference.
3190 Description: server LBUG when using old ost_num_threads parameter
3191 Details : Accept the old ost_num_threads parameter but warn that it
3192 is deprecated, and fix an off-by-one error that caused an LBUG.
3197 Description: Transient SCSI error results in persistent IO issue
3198 Details : iobuf->dr_error is not reinitialized to 0 between two
3202 Frequency : sometimes when underlying device returns I/O errors
3204 Description: OSTs not going read-only during write failures
3205 Details : OSTs are not remounted read-only when the journal commit threads
3206 get I/O errors because fsfilt_ext3 calls journal_start/stop()
3207 instead of the ext3 wrappers.
3211 Description: poor connect scaling with increasing client count
3212 Details : Don't run filter_grant_sanity_check for more than 100 exports
3213 to improve scaling for large numbers of clients.
3216 Frequency : SLES10 only
3218 Description: sanity-quota.sh quotacheck failed: rc = -22
3219 Details : Quotas cannot be enabled on SLES10.
3222 Frequency : liblustre clients only
3224 Description: getdirentries does not give error when run on compute nodes
3225 Details : getdirentries does not fail when the size specified as an argument
3226 is too small to contain at least one entry
3228 Severity : enhancement
3230 Description: Add LNET router traceability for debug purposes
3231 Details : If a checksum failure occurs with a router as part of the
3232 IO path, the NID of the last router that forwarded the bulk data
3233 is printed so it can be identified.
3238 Description: OST "spontaneously" evicts client; client has imp_pingable == 0
3239 Details : Due to a race condition, liblustre clients were occasionally
3240 evicted incorrectly.
3242 Severity : enhancement
3244 Description: lfs setstripe use optional parameters instead of postional
3247 Severity : enhancement
3249 Description: Nanosecond timestamp support for ldiskfs
3250 Details : The on-disk ldiskfs filesystem has added support for nanosecond
3251 resolution timestamps. There is not yet support for this at
3252 the Lustre filesystem level.
3255 Frequency : during server recovery
3257 Description: MDS failing to send precreate requests due to OSCC_FLAG_RECOVERING
3258 Details : request with rq_no_resend flag not awake l_wait_event if they get
3262 Frequency : nfs export on patchless client
3264 Description: connectathon hang when test nfs export over patchless client
3265 Details : Disconnected dentry cannot be found with lookup, so we do not need
3266 to unhash it or make it invalid
3269 Description: fix llapi_lov_get_uuids() to allow many OSTs to be returned
3270 Details: : Change llapi_lov_get_uuids() to read the UUIDs from /proc instead
3271 of using an ioctl. This allows lfsck for > 160 OSTs to succeed.
3276 Description: open req refcounting wrong on reconnect
3277 Details : If reconnect happened between getting open reply from server and
3278 call to mdc_set_replay_data in ll_file_open, we will schedule
3279 replay for unreferenced request that we are about to free.
3280 Subsequent close will crash in variety of ways.
3281 Check that request is still eligible for replay in
3282 mdc_set_replay_data().
3287 Description: disable writes to filesystem when reading health_check file
3288 Details : the default for reading the health_check proc file has changed
3289 to NOT do a journal transaction and write to disk, because this
3290 can cause reads of the /proc file to hang and block HA state
3291 checking on a healthy but otherwise heavily loaded system. It
3292 is possible to return to the previous behaviour during configure
3293 with --enable-health-write.
3295 Severity : enhancement
3297 Description: 64-bit inode version
3298 Details: : Add a on-disk 64-bit inode version for ext3 to track changes made
3299 to the inode. This will be required for version-based recovery.
3304 Description: MDS fails to start if a duplicate client export is detected
3305 Details : in some rare cases it was possible for a client to connect to
3306 an MDS multiple times. Upon recovery the MDS would detect this
3307 and fail during startup. Handle this more gracefully.
3309 Severity : enhancement
3311 Description: Add -o localflock option to simulate old noflock
3313 Details : This will achieve local-only flock/fcntl locks
3319 Description: log_commit_thread vs filter_destroy race leads to crash
3320 Details : Take import reference before releasing llog record semaphore
3325 Description: Wrong request locking in request set processing
3326 Details : ptlrpc_check_set wrongly uses req->rq_lock for proctect add to
3327 imp_delayed_list, in this place should be used imp_lock.
3330 Frequency : when reconnection
3332 Description: Grant Leak when osc reconnect to OST
3333 Details : When osc reconnect ost, OST(filter) should check whether it
3334 should grant more space to client by comparing fed_grant and
3335 cl_avail_grant, and return the granted space to client instead
3336 of "new granted" space, because client will call osc_init_grant
3337 to update the client grant space info.
3340 Frequency : when client reconnect to OST
3342 Description: Grant Leak when osc do resend and replay bulk write
3343 Details : When osc reconnect to OST, OST(filter)should clear grant info of
3344 bulk write request, because the grant info while be sync between
3345 OSC and OST when reconnect, and we should ignore the grant info
3346 these of resend/replay write req.
3351 Description: Grant space more than available left space sometimes.
3352 Details : When then OST is about to be full, if two bulk writing from
3353 different clients came to OST. Accord the avaliable space of the
3354 OST, the first req should be permitted, and the second one
3355 should be denied by ENOSPC. But if the seconde arrived before
3356 the first one is committed. The OST might wrongly permit second
3357 writing, which will cause grant space > available space.
3360 Frequency : when client is evicted
3362 Description: Grant might be wrongly erased when osc is evicted by OST
3363 Details : when the import is evicted by server, it will fork another
3364 thread ptlrpc_invalidate_import_thread to invalidate the
3365 import, where the grant will be set to 0. While the original
3366 thread will update the grant it got when connecting. So if
3367 the former happened latter, the grant will be wrongly errased
3368 because of this race.
3373 Description: Checking Stale with correct fid
3374 Details : ll_revalidate_it should uses de_inode instead of op_data.fid2
3375 to check whether it is stale, because sometimes, we want the
3376 enqueue happened anyway, and op_data.fid2 will not be initialized.
3378 Severity : enhancement
3380 Description: update patchless client
3381 Details : Add support for patchless client with 2.6.20, 2.6.21 and RHEL 5
3384 Frequency : only with 2.4 kernel
3386 Description: random memory corruption
3387 Details : size of struct ll_inode_info is to big for union inode.u and this
3388 can be cause of random memory corruption.
3393 Description: Memory leak in recovery
3394 Details : Lov_mds_md was not free in an error handler in mds_create_object.
3395 It should also check obd_fail before fsfilt_start, otherwise if
3396 fsfilt_start return -EROFS,(failover mds during mds recovery).
3397 then the req will return with repmsg->transno = 0 and rc = EROFS.
3398 and we met hit the assert LASSERT(req->rq_reqmsg->transno ==
3399 req->rq_repmsg->transno) in ptlrpc_replay_interpret. Fcc should
3400 be freed no matter whether fsfilt_commit success or not.
3403 Frequency : only with huge count clients
3405 Description: Prevents from taking the superblock lock in llap_from_page for
3407 Details : using LL_ORIGIN_REMOVEPAGE origin flag instead of LL_ORIGIN_UNKNOW
3408 for llap_from_page call in ll_removepage prevents from taking the
3409 superblock lock for a soon died page.
3414 Description: Not check open intent error before release open handle
3415 Details : in some rare cases, the open intent error is not checked before
3416 release open handle, which may cause
3417 ASSERTION(open_req->rq_transno != 0), because it tries to release
3418 the failed open handle.
3423 Description: Set cat log bitmap only after create log success.
3424 Details : in some rare cases, the cat log bitmap is set too early. and it
3425 should be set only after create log success.
3429 Description: Accessing a block bevice can re-enable I/O when Lustre is
3430 tearing down a device.
3431 Details : dev_clear_rdonly(bdev) must be called in kill_bdev() instead of
3436 Description: service threads may hog cpus when there are a lot of requests
3437 Details : Insert cond_resched to give other threads a chance to use some CPU
3442 Description: the cat log was not initialized in recovery
3443 Details : When mds(mgs) do recovery, the tgt_count might be zero, so the
3444 unlink log on mds will not be initialized until mds post
3445 recovery. And also in mds post recovery, the unlink log will
3446 initialization will be done asynchronausly, so there will be race
3447 between add unlink log and unlink log initialization.
3451 Description: brw_stats were being printed incorrectly
3452 Details : brw_stats were being printed as log2 but all of them were not
3453 recorded as log2. Also remove some code duplication arising from
3454 filter_tally_{read,write}.
3458 Frequency : rare, only in recovery.
3459 Description: ASSERTION(req->rq_type != LI_POISON) failed
3460 Details : imp_lock should be held while iterating over imp_sending_list for
3461 prevent destroy request after get timeout in ptlrpc_queue_wait.
3465 Description: replay-single.sh test 52 fails
3466 Details : A lock's skiplist need to be cleanup when it being unlinked
3467 from its resource list.
3471 Description: Short directio read returns full requested size rather than
3473 Details : Direct I/O operations should return actual amount of bytes
3474 transferred rather than requested size.
3476 Severity : enhancement
3478 Description: metadata RPC reduction (e.g. for rm performance)
3479 Details : decrease the amount of synchronous RPC between clients and servers
3480 by canceling conflicing lock before the operation on the client
3481 and packing thier handles into the main operation RPC to server.
3483 Severity : enhancement
3485 Description: Async OSC create to avoid the blocking unnecessarily.
3486 Details : If an OST has no remaining object, system will block on the
3487 creation when it needs to create a new object on this OST. Now,
3488 ways use pre-created objects when available, instead of blocking on
3489 an empty osc while others are not empty. If we must block, we
3490 block for the shortest possible period of time.
3494 Description: improve handling recoverable errors
3495 Details : if request processig with error which can be recoverable on server
3496 request should be resend, otherwise page released from cache and
3499 Severity : enhancement
3501 Description: refine locking for avoid write wrong info into lov_objid file
3502 Details : fix possible races with add new target and write/update data in
3505 --------------------------------------------------------------------------------
3507 2007-05-03 Cluster File Systems, Inc. <info@clusterfs.com>
3512 Frequency : on some architectures
3514 Description: 1.6 client sometimes fails to mount from a 1.4 MDT
3515 Details : Uninitialized flags sometimes cause configuration commands to
3519 Frequency : patchless clients only
3521 Description: missing __iget() symbol export
3522 Details : The __iget() symbol export is missing. To avoid the need for
3523 this on patchless clients the deathrow inode reaper is turned
3524 off, and we depend on the VM to clean up old inodes. This
3525 dependency was during via the fix for bug 12181.
3527 --------------------------------------------------------------------------------
3529 2007-04-19 Cluster File Systems, Inc. <info@clusterfs.com>
3531 * CONFIGURATION CHANGE. This version of Lustre WILL NOT
3532 INTEROPERATE with older versions automatically. In many cases a
3533 special upgrade step is needed. Please read the
3534 user documentation before upgrading any part of a live system.
3535 * WIRE PROTOCOL CHANGE from previous 1.6 beta versions. This
3536 version will not interoperate with 1.6 betas before beta5 (1.5.95).
3537 * WARNING: Lustre configuration and startup changes are required with
3538 this release. See https://mail.clusterfs.com/wikis/lustre/MountConf
3542 Severity : enhancement
3544 Description: MountConf
3545 Details : Lustre configuration is now managed via mkfs and mount
3546 commands instead of lmc and lconf. New obd types (MGS, MGC)
3547 are added for dynamic configuration management. See
3548 https://mail.clusterfs.com/wikis/lustre/MountConf for
3551 Severity : enhancement
3553 Description: dynamic OST addition
3554 Details : OSTs can now be added to a live filesystem
3556 Severity : enhancement
3558 Description: startup order invariance
3559 Details : MDTs and OSTs can be started in any order. Clients only
3560 require the MDT to complete startup.
3562 Severity : enhancement
3564 Description: parallel, asynchronous orphan cleanup
3565 Details : orphan cleanup is now performed in separate threads for each
3566 OST, allowing parallel non-blocking operation.
3568 Severity : enhancement
3570 Description: optimized stripe assignment
3571 Details : stripe assignments are now made based on ost space available,
3572 ost previous usage, and OSS previous usage, in order to try
3573 to optimize storage space and networking resources.
3575 Severity : enhancement
3577 Description: Permanently set tunables
3578 Details : All writable /proc/fs/lustre tunables can now be permanently
3579 set on a per-server basis, at mkfs time or on a live system.
3581 Severity : enhancement
3583 Description: Lustre message v2
3584 Details : Add lustre message format v2.
3586 Severity : enhancement
3588 Description: client OST exclusion list
3589 Details : Clients can be started with a list of OSTs that should be
3590 declared "inactive" for known non-responsive OSTs.
3594 Description: SPEC SFS validation failure on NFS v2 over lustre.
3595 Details : Changes the blocksize for regular files to be 2x RPC size,
3596 and not depend on stripe size.
3598 Severity : enhancement
3600 Description: Multiple MD RPCs in flight.
3601 Details : Further unserialise some read-only MDS RPCs - learn about intents.
3602 To avoid overly-overloading MDS, introduce a limit on number of
3603 MDS RPCs in flight for a single client and add /proc controls
3604 to adjust this limit.
3606 Severity : enhancement
3608 Description: client read/write statistics
3609 Details : Add client read/write call usage stats for performance
3610 analysis of user processes.
3611 /proc/fs/lustre/llite/*/offset_stats shows non-sequential
3612 file access. extents_stats shows chunk size distribution.
3613 extents_stats_per_process show chunk size distribution per
3616 Severity : enhancement
3618 Description: mds statistics
3619 Details : Add detailed mds operations statistics in
3620 /proc/fs/lustre/mds/*/stats.
3624 Description: Failure of copying files with lustre special EAs.
3625 Details : Client side always return success for setxattr call for lustre
3626 special xattr (currently only "trusted.lov").
3631 Description: Refcount LNET uuids
3632 Details : The global LNET uuid list grew linearly with every startup;
3633 refcount repeated list entries instead of always adding to
3636 Severity : enhancement
3638 Description: Dynamic service threads
3639 Details : Within a small range, start extra service threads
3640 automatically when the request queue builds up.
3643 Frequency : mixed-endian client/server environments
3645 Description: mixed-endian crashes
3646 Details : The new msg_v2 system had some failures in mixed-endian
3649 Severity : enhancement
3651 Description: Easy OST removal
3652 Details : OSTs can be permanently deactivated with e.g. 'lctl
3653 conf_param lustre-OST0001.osc.active=0'
3655 Severity : enhancement
3657 Description: MGS proc entries
3658 Details : Added basic proc entries for the MGS showing what filesystems
3661 Severity : enhancement
3663 Description: provide MGS failover
3664 Details : Added config lock reacquisition after MGS server failover.
3666 Severity : enhancement
3668 Description: add Linux 2.4 support
3669 Details : Added support for RHEL 2.4.21 kernel for 1.6 servers and clients
3673 Description: a large application tries to do I/O to the same resource and dies
3674 in the middle of it.
3675 Details : Check the req->rq_arrival time after the call to
3676 ost_brw_lock_get(), but before we do anything about
3677 processing it & sending the BULK transfer request. This
3678 should help move old stale pending locks off the queue as
3679 quickly as obd_timeout.
3682 Frequency : when an incorrect nid is specified during startup
3684 Description: ptlrpc connect to non-existant node causes kernel crash
3685 Details : LNET can't be re-entered from an event callback, which
3686 happened when we expire a message after the export has been
3687 cleaned up. Instead, hand the zombie cleanup off to another
3690 Severity : enhancement
3692 Description: plain/inodebits lock performance improvement
3693 Details : Grouping plain/inodebits in granted list by their request modes
3694 and bits policy, thus improving the performance of search through
3698 Frequency : only if OST filesystem is corrupted
3700 Description: client incorrectly hits assertion in ptlrpc_replay_req()
3701 Details : for a short time RPCs with bulk IO are in the replay list,
3702 but replay of bulk IOs is unimplemented. If the OST filesystem
3703 is corrupted due to disk cache incoherency and then replay is
3704 started it is possible to trip an assertion. Avoid putting
3705 committed RPCs into the replay list at all to avoid this issue.
3708 Frequency : liblustre (e.g. catamount) on a large cluster with >= 8 OSTs/OSS
3710 Description: System hang on startup
3711 Details : This bug allowed the liblustre (e.g. catamount) client to
3712 return to the app before handling all startup RPCs. This
3713 could leave the node unresponsive to lustre network traffic
3714 and manifested as a server ptllnd timeout.
3716 Severity : enhancement
3718 Description: Add "/proc/sys/lustre/debug_peer_on_timeout"
3719 Details : liblustre envirable: LIBLUSTRE_DEBUG_PEER_ON_TIMEOUT
3720 boolean to control whether to print peer debug info when a
3721 client's RPC times out.
3724 Frequency : only for kernels with patches from Lustre below 1.4.3
3726 Description: Remove old rdonly API
3727 Details : Remove old rdonly API which unused from at least lustre 1.4.3
3730 Frequency : only for devices with external journals
3732 Description: Set external device read-only also
3733 Details : During a commanded failover stop, we set the disk device
3734 read-only while the server shuts down. We now also set any
3735 external journal device read-only at the same time.
3738 Frequency : when upgrading from 1.4 while trying to change parameters
3740 Description: The wrong (new) MDC name was used when setting parameters for
3741 upgraded MDT's. Also allows changing of OSC (and MDC)
3742 parameters if --writeconf is specified at tunefs upgrade time.
3745 Frequency : when setting specific ost indicies
3747 Description: QOS code breaks on skipped indicies
3748 Details : Add checks for missing OST indicies in the QOS code, so OSTs
3749 created with --index need not be sequential.
3751 Severity : enhancement
3753 Description: Add uninit_groups feature to ldiskfs2 to speed up e2fsck
3754 Details : The uninit_groups feature works in conjunction with the kernel
3755 filesystem code (ldiskfs2 only) and e2fsprogs-1.39-cfs6 to speed
3756 up the pass1 processing of e2fsck. This is a read-only feature
3757 in ldiskfs2 only, so older kernels and current ldiskfs cannot
3758 mount filesystems that have had this feature enabled.
3760 Severity : enhancement
3762 Description: Improve multi-block allocation algorithm to avoid fragmentation
3763 Details : The mballoc3 code (ldiskfs2 only) adds new mechanisms to improve
3764 allocation locality and avoid filesystem fragmentation.
3766 ------------------------------------------------------------------------------
3768 2007-04-01 Cluster File Systems, Inc. <info@clusterfs.com>
3770 * Support for kernels:
3771 2.4.21-47.0.1.EL (RHEL 3)
3772 2.6.5-7.283 (SLES 9)
3773 2.6.9-42.0.10.EL (RHEL 4)
3774 2.6.12.6 vanilla (kernel.org)
3775 2.6.16.27-0.9 (SLES 10)
3776 * Recommended e2fsprogs version: 1.39.cfs5
3778 * Note that reiserfs quotas are disabled on SLES 10 in this kernel
3782 Frequency : occasional, depends on client load and configuration
3783 Bugzilla : 12181, 12203
3784 Description: data loss for recently-modified files
3786 Details : In some cases it is possible that recently written or created
3787 files may not be written to disk in a timely manner (this should
3788 normally be within 30s unless client IO load is very high).
3789 The problem appears as zero-length files or files that are a
3790 multiple of 1MB in size after a client crash or client eviction
3791 that are missing data at the end of the file.
3793 This problem is more likely to be hit on clients where files are
3794 repeatedly created and unlinked in the same directory, clients
3795 have a large amount of RAM, have many CPUs, the filesystem has
3796 many OSTs, the clients are rebooted frequently, and/or the files
3797 are not accessed by other nodes after being written.
3799 The presence of the problem can be detected by looking at
3800 /proc/sys/fs/inode-state. If the first number (nr_inodes) is
3801 smaller than the second (nr_unused) then dirty files will not
3802 be flushed automatically to disk. "sync; sleep 10" should be
3803 run several times on the node before unmounting it to update
3804 Lustre (this is also safe to run on nodes without this problem).
3806 There is also a related kernel bug in the RHEL4 4 2.6.9 kernel
3807 that can cause this same problem, so customers using that kernel
3808 also need to update the kernel in addition to Lustre. In order
3809 to properly fix this bug, the RHEL3 2.4.21 kernel is also updated.
3811 It is normal that files written just before a client crash (less
3812 than 30s) may not yet have been flushed to disk, even for local
3816 Frequency : frequent on thin XT3 nodes
3818 Description: UUID collision on thin XT3 Linux nodes
3819 Details : UUIDs on Compute Node Linux XT3 nodes were not generated
3820 randomly, since we relied on an insufficiently-seeded PRNG.
3825 Description: OSS hangs after "All ost request buffers busy"
3826 Details : A deadlock between quota and journal operations caused OSS
3827 hangs after printing "All ost request buffers busy."
3830 Frequency : always on liblustre builds
3832 Description: Cleanup compiler warnings on liblustre
3835 Frequency : always on liblustre builds on XT3
3837 Description: LC_CONFIG_CDEBUG don't run while build liblustre on XT3.
3841 Description: Addition of EXT3_FEATURE_RO_COMPAT_DIR_NLINKS flag for
3842 > 32000 subdirectories
3843 Details : Add EXT3_FEATURE_RO_COMPAT_DIR_NLINK flag to
3844 EXT3_FEATURE_RO_COMPAT_SUPP. This flag will be set whenever
3845 subdirectory count crosses 32000. This will aid e2fsck to
3846 correctly handle more than 32000 subdirectories.
3849 Frequency : liblustre (e.g. catamount) on a large cluster with >= 8 OSTs/OSS
3851 Description: System hang on startup
3852 Details : This bug allowed the liblustre (e.g. catamount) client to
3853 return to the app before handling all startup RPCs. This
3854 could leave the node unresponsive to lustre network traffic
3855 and manifested as a server ptllnd timeout.
3857 Severity : enhancement
3859 Description: Add "/proc/sys/lustre/debug_peer_on_timeout"
3860 (liblustre envirable: LIBLUSTRE_DEBUG_PEER_ON_TIMEOUT)
3861 boolean to control whether to print peer debug info when a
3862 client's RPC times out.
3867 Description: make O_SYNC working on 2.6 kernels
3868 Details : 2.6 kernels use different method for mark pages for write,
3869 so need add a code to lustre for O_SYNC work.
3874 Description: Failure to close file and release space on NFS
3875 Details : Put inode details into lock acquired in ll_intent_file_open.
3876 Use mdc_intent_lock in ll_intent_open to properly
3877 detect all kind of errors unhandled by mdc_enqueue.
3882 Description: proc file read during shutdown sometimes raced obd removal,
3884 Details : Add lock to prevent obd access after proc file removal.
3887 Frequency : Only for files larger than 4GB on 32-bit clients.
3889 Description: improperly doing page alignment of locks
3890 Details : Modify lustre core code to use CFS_PAGE_* defines instead of
3891 PAGE_*. Make CFS_PAGE_MASK a 64-bit mask.
3896 Description: RPCs being resent when they shouldn't be
3897 Details : Some RPCs that should not be resent are being resent. This
3898 can cause inconsistencies in the RPC state machine. Do not
3899 resend such requests.
3902 Frequency : rare, only with NFS export
3904 Description: Crash on NFS re-export node
3905 Details : under very unusual load conditions an assertion is hit in
3906 ll_intent_file_open()
3909 Frequency : only if OST filesystem is corrupted
3911 Description: client incorrectly hits assertion in ptlrpc_replay_req()
3912 Details : for a short time RPCs with bulk IO are in the replay list,
3913 but replay of bulk IOs is unimplemented. If the OST filesystem
3914 is corrupted due to disk cache incoherency and then replay is
3915 started it is possible to trip an assertion. Avoid putting
3916 committed RPCs into the replay list at all to avoid this issue.
3921 Description: large O_DIRECT requests fail under memory pressure/fragmentation
3922 Details : Large single O_DIRECT read and write calls can fail to allocate
3923 a sufficiently large buffer to process the request. In case of
3924 allocation failure the allocation is retried with a smaller
3925 buffer and broken into smaller requests.
3927 Severity : enhancement
3929 Description: Add -o localflock option to simulate old noflock behaviour.
3930 Details : This will achieve local-only flock/fcntl locks coherentness.
3935 Description: versioning check is incomplete
3936 Details : Checking the version difference of client vs. server, report
3937 error if the gap is too big.
3942 Description: add support PG_writeback bit
3943 Details : add support for PG_writeback bit for Lustre, for more carefull
3944 work with page cache in 2.6 kernel. This also fix some deadlocks
3945 and remove hack for work O_SYNC with 2.6 kernel.
3947 Severity : enhancement
3949 Description: Add uninit_groups feature to ldiskfs2 to speed up e2fsck
3950 Details : The uninit_groups feature works in conjunction with the kernel
3951 filesystem code (ldiskfs2 only) and e2fsprogs-1.39-cfs6 to speed
3952 up the pass1 processing of e2fsck. This is a read-only feature
3953 in ldiskfs2 only, so older kernels and current ldiskfs cannot
3954 mount filesystems that have had this feature enabled.
3956 Severity : enhancement
3958 Description: Improve multi-block allocation algorithm to avoid fragmentation
3959 Details : The mballoc3 code (ldiskfs2 only) adds new mechanisms to improve
3960 allocation locality and avoid filesystem fragmentation.
3962 ------------------------------------------------------------------------------
3964 2007-02-09 Cluster File Systems, Inc. <info@clusterfs.com>
3966 * Support for kernels:
3967 2.6.9-42.0.3.EL (RHEL 4)
3968 2.6.5-7.276 (SLES 9)
3969 2.4.21-47.0.1.EL (RHEL 3)
3970 2.6.12.6 vanilla (kernel.org)
3971 2.6.16.21-0.8 (SLES10)
3972 * Recommended e2fsprogs version: 1.39.cfs2-0
3974 * The backwards-compatible /proc/sys/portals symlink has been removed
3975 in this release. Before upgrading, please ensure that you change
3976 any configuration scripts or /etc/sysctl.conf files that access
3977 /proc/sys/portals/* or sysctl portals.* to use the corresponding
3978 entry in /proc/sys/lnet or sysctl lnet.*. This change can be made
3979 in advance of the upgrade on any system running Lustre 1.4.6 or
3980 newer, since /proc/sys/lnet was added in that version.
3981 * Note that reiserfs quotas are disabled on SLES 10 in this kernel
3985 Frequency : only when quota is used
3987 Description: avoid scanning export list for quota master
3988 Details : Change the algorithms to avoid scanning export list in order
3989 to improve the efficiency.
3992 Frequency : MDS failover only, very rarely
3994 Description: "went back in time" messages on mds failover
3995 Details : The greatest transno may be lost when the current operation
3996 finishes with an error (transno==0) and the client's last_rcvd
3997 record is over-written. Save the greatest transno in the
3998 mds_last_transno for this case.
4001 Frequency : always for specific kernels and striping counts
4003 Description: client may get "Matching packet too big" without ACL support
4004 Details : Clients compiled without CONFIG_FS_POSIX_ACL get an error message
4005 when trying to access files in certain configurations. The
4006 clients should in fact be denied when mounting because they do
4007 not understand ACLs.
4010 Frequency : Cray XT3 with more than 4000 clients and multiple jobs
4012 Description: many clients connecting with IO in progress causes connect timeouts
4013 Details : Avoid synchronous journal commits to avoid delays caused by many
4014 clients connecting/disconnecting when bulk IO is in progress.
4015 Queue liblustre connect requests on OST_REQUEST_PORTAL instead of
4016 OST_IO_PORTAL to avoid delays behind potentially many pending
4020 Frequency : occasionally with multiple writers to a single file
4022 Description: shared writes to file may result in wrong size reported by stat()
4023 Details : Allow growing of kms when extent lock is cancelled
4026 Frequency : always with random mmap IO to multi-striped file
4028 Description: mmap write might be lost if we are writing to a 'hole' in stripe
4029 Details : Only if the hole is at the end of OST object so that kms is too
4030 small. Fix is to increase kms accordingly in ll_nopage.
4033 Frequency : rare, only if OST filesystem is inconsistent with MDS filesystem
4035 Description: writes to a missing object would leak memory on the OST
4036 Details : If there is an inconsistency between the MDS and OST filesystems,
4037 such that the MDS references an object that doesn't exist, writes
4038 to that object will leak memory due to incorrect cleanup in the
4039 error handling path, eventually running out of memory on the OST.
4044 Description: Creating too long symlink causes lustre errors
4045 Details : Check symlink and name lengths before sending requests to MDS.
4048 Frequency : only if flock is enabled (not on by default)
4050 Description: posix locks not released on fd closure on 2.6.9+
4051 Details : We failed to add posix locks to list of inode locks on 2.6.9+
4052 kernels, this caused such locks not to be released on fd close and
4053 then assertions on fs unmount about still used locks.
4056 Frequency : MDS failover only, very rarely
4058 Description: clients may get ASSERTION(granted_lock != NULL)
4059 Details : When request was taking a long time, and a client was resending
4060 a getattr by name lock request. The were multiple lock requests
4061 with the same client lock handle and
4062 mds_getattr_name->fixup_handle_for_resent_request found one of the
4063 lock handles but later failed with ASSERTION(granted_lock != NULL).
4068 Description: handle->h_buffer_credits > 0, assertion failure
4069 Details : h_buffer_credits is zero after truncate, causing assertion
4070 failure. This patch extends the transaction or creates a new
4074 Frequency : NFS re-export or patchless client
4075 Bugzilla : 11179, 10796
4076 Description: Crash on NFS re-export node (__d_move)
4077 Details : We do not want to hash the dentry if we don't have a lock.
4078 But if this dentry is later used in d_move, we'd hit uninitialised
4079 list head d_hash, so we just do this to init d_hash field but
4080 leave dentry unhashed.
4083 Frequency : NFS re-export or patchless client
4085 Description: NFS exports has problem with symbolic link
4086 Details : lustre client didn't properly install dentry when re-exported
4087 to NFS or running patchless client.
4090 Frequency : NFS re-export or patchless client
4092 Description: Various nfs/patchless fixes.
4093 Details : fixes reuse disconected alias for lookup process - this fixes
4094 warning "find_exported_dentry: npd != pd",
4095 fix permission error with open files at nfs.
4096 fix apply umask when do revalidate.
4099 Frequency : occasional
4101 Description: Crash on NFS re-export node
4102 Details : calling clear_page() on the wrong pointer triggered oops in
4103 generic_mapping_read().
4106 Frequency : rarely, using O_DIRECT IO
4108 Description: unaligned directio crashes client with LASSERT
4109 Details : check for unaligned buffers before trying any requests.
4112 Frequency : rarely, using CFS RAID5 patches in non-standard kernel series
4114 Description: stale data returned from RAID cache
4115 Details : If only a small amount of IO is done to the RAID device before
4116 reading it again it is possible to get stale data from the RAID
4117 cache instead of reading it from disk.
4120 Frequency : always for sles10 kernel
4122 Description: sles10 support
4123 Details : ll_follow_link: compile fixes and using of nd_set_link
4124 under newer kernels.
4127 Frequency : depends on arch, kernel and compiler version, always on sles10
4130 Description: recursive or deep enough symlinks cause stack overflow
4131 Details : getting rid of large stack-allocated variable in
4135 Frequency : depends on hardware
4137 Description: lustre write performance loss in the SLES10 kernel
4138 Details : the performance loss is caused by using of write barriers in the
4139 ext3 code. The SLES10 kernel turns barrier support on by
4140 default. The fix is to undo that change for ldiskfs.
4142 ------------------------------------------------------------------------------
4144 2006-12-09 Cluster File Systems, Inc. <info@clusterfs.com>
4146 * Support for kernels:
4147 2.6.9-42.0.3EL (RHEL 4)
4148 2.6.5-7.276 (SLES 9)
4149 2.4.21-47.0.1.EL (RHEL 3)
4150 2.6.12.6 vanilla (kernel.org)
4154 Frequency : quota enabled and large files being deleted
4156 Description: releasing more than 4GB of quota at once hangs OST
4157 Details : If a user deletes more than 4GB of files on a single OST it
4158 will cause the OST to spin in an infinite loop. Release
4159 quota in < 4GB chunks, or use a 64-bit value for 1.4.7.1+.
4164 Description: statfs data retrieved from /proc may be stale or zero
4165 Details : When reading per-device statfs data from /proc, in the
4166 {kbytes,files}_{total,free,avail} files, it may appear
4167 as zero or be out of date.
4170 Frequency : systems with MD RAID1 external journal devices
4172 Description: lconf's call to blkid is confused by RAID1 journal devices
4173 Details : Use the "blkid -l" flag to locate the MD RAID device instead
4174 of returning all block devices that match the journal UUID.
4177 Frequency : always, for aggregate stripe size over 4GB
4179 Description: "lfs setstripe" fails assertion when setting 4GB+ stripe width
4180 Details : Using "lfs setstripe" to set stripe size * stripe count over 4GB
4181 will fail the kernel with "ASSERTION(lsm->lsm_xfersize != 0)"
4184 Frequency : always if "lfs find" used on a local file/directory
4186 Description: "lfs find" segfaults if used on a local file/directory
4187 Details : The case where a directory component was not specified wasn't
4188 handled correctly. Handle this properly.
4191 Frequency : always on ppc64
4193 Description: the write to an ext3 filesystem mounted with mballoc got stuck
4194 Details : ext3_mb_generate_buddy() uses find_next_bit() which does not
4195 perform endianness conversion.
4198 Frequency : rarely (truncate to non-zero file size after write under load)
4199 Bugzilla : 10730, 10687
4200 Description: Files padded with zeros to next 4K multiple
4201 Details : With filesystems mounted using the "extents" option (2.6 kernels)
4202 it is possible that files that are truncated to a non-zero size
4203 immediately after being written are filled with zero bytes beyond
4204 the truncated size. No file data is lost.
4206 Severity : enhancement
4208 Description: Allow recovery/failover for liblustre clients.
4209 Details : liblustre clients were unaware of failover configurations until
4212 Severity : enhancement
4214 Description: user file locks should fail when not mounting with flock option
4215 Details : Set up an error-returning stub in ll_file_operations.lock field
4216 to prevent incorrect behaviour when client is mounted without
4217 flock option. Also, set up properly f_op->flock field for
4221 Frequency : always on ia64
4223 Description: "lfs df" loops on printing out MDS statfs information
4224 Details : The obd_ioctl_data was not initialized and in some systems
4225 this caused a failure during the ioctl that did not return
4226 an error. Initialize the struct and return an error on failure.
4229 Frequency : SLES 9 only
4231 Description: Error of copying files with lustre special EAs as root
4232 Details : Client side always return success for setxattr call for lustre
4233 special xattr (currently only "trusted.lov").
4236 Frequency : rarely on clusters with both ia64+i386 clients
4238 Description: ia64+i686 clients doing shared IO on the same file may LBUG
4239 Details : In rare cases when both ia64+i686 (or other mixed-PAGE_SIZE)
4240 clients are doing concurrent writes to the same file it is
4241 possible that the ia64 clients may LASSERT because the OST
4242 extent locks are not PAGE_SIZE aligned. Ensure that grown
4243 locks are always aligned on the request boundary.
4246 Frequency : specific use, occasional
4248 Description: Overwriting in use executable truncates on-disk binary image
4249 Details : If one node attempts to overwrite an executable in use by
4250 another node, we now correctly return ETXTBSY instead of
4251 truncating the file.
4253 Severity : enhancement
4255 Description: Async OSC create to avoid the blocking unnecessarily.
4256 Details : If an OST has no remaining object, system will block on the creating
4257 when need to create a new object on this OST. Now, ways use
4258 pre-created objects when available, instead of blocking on an
4259 empty osc while others are not empty. If we must block, we block
4260 for the shortest possible period of time.
4265 Description: chmod on Lustre root is propagated to other clients
4266 Details : Re-validate root's dentry in ll_lookup_it to avoid having it
4267 invalid by the follow_mount time.
4270 Frequency : liblustre clients only
4272 Description: Race in 'instant cancel' lock handling could lead to such locks
4273 never to be granted in case of SMP MDS
4274 Details : Do not destroy not yet granted but cbpending locks in
4278 Frequency : replay/resend of open
4280 Description: non null lock assetion failure in mds_intent_policy
4281 Details : Trying to replay/resend lockless open requests resulted in
4282 mds_open() returning 0 with no lock. Now it sets a flag if
4283 a lock is going to be returned.
4285 Severity : enhancement
4287 Description: Checksum enhancements
4288 Details : New checksum enhancements allow for resending RPCs that failed
4291 Severity : enhancement
4293 Description: Tunables on number of dirty pages in cacche
4294 Details : Allow to set limit on number of dirty pages cached.
4299 Description: client crash on unmount - lock still has references
4300 Details : In some error handling cases it was possible to leak a lock
4301 reference on a client while accessing a file. This was not
4302 harmful to the client during operation, but would cause the
4303 client to crash when the filesystem is unmounted.
4306 Frequency : specific case, rare
4308 Description: ETXTBSY on mds though file not in use by client
4309 Details : ETXTBSY is no longer incorrectly returned when attempting to
4310 chmod or chown a directory that the user previously tried to
4311 execute or a currently-executing binary.
4314 Frequency : extremely rare except on liblustre-based clients
4316 Description: Lustre space not freed when files are deleted
4317 Details : Clean up open-unlinked files after client eviction. Previously
4318 the unlink was skipped and the files remained as orphans.
4323 Description: OST failure "would be an LBUG" in waiting_locks_callback()
4324 Details : In some cases it was possible to send a blocking callback to
4325 a client doing a glimpse, even though that client didn't get
4326 a lock granted. When the glimpse lock is cancelled on the OST
4327 the freed lock is left on the waiting list and corrupted the list.
4330 Frequency : all core dumps
4332 Description: Broke core dumps to lustre
4333 Details : Negative dentry may be unhashed if parent does not have UPDATE
4334 lock, but some callers, e.g. do_coredump, expect dentry to be
4335 hashed after successful create, hash it in ll_create_it.
4337 ------------------------------------------------------------------------------
4339 2006-09-13 Cluster File Systems, Inc. <info@clusterfs.com>
4341 * Support for kernels:
4342 2.6.9-42.0.2.EL (RHEL 4)
4343 2.6.5-7.276 (SLES 9)
4344 2.4.21-40.EL (RHEL 3)
4345 2.6.12.6 vanilla (kernel.org)
4349 Frequency : always on RHEL 3
4351 Description: Number of open files grows over time
4352 Details : The number of open files grows over time, whether or not
4353 Lustre is started. This was due to a filp leak introduced
4354 by one of our kernel patches.
4356 ------------------------------------------------------------------------------
4358 08-20-2006 Cluster File Systems, Inc. <info@clusterfs.com>
4360 * Support for kernels:
4361 2.6.9-42.EL (RHEL 4)
4362 2.6.5-7.276 (SLES 9)
4363 2.4.21-40.EL (RHEL 3)
4364 2.6.12.6 vanilla (kernel.org)
4369 Bugzilla : 5719, 9635, 9792, 9684
4370 Description: OST (or MDS) trips assertions in (re)connection under heavy load
4371 Details : If a server is under heavy load and cannot reply to new
4372 connection requests before the client resends the (re)connect,
4373 the connection handling code can behave badly if two service
4374 threads are concurrently handing separate (re)connections from
4375 the same client. Add better locking to the connection handling
4376 code, and ensure that only a single connection will be processed
4377 for a given client UUID, even if the lock is dropped.
4379 Severity : enhancement
4381 Description: add TCP zero-copy support to kernel
4382 Details : Add support to the kernel TCP stack to allow zero-copy bulk
4383 sends if the hardware supports scatter-gather and checksumming.
4384 This allows socklnd to do client-write and server-read more
4385 efficiently and reduce CPU utilization from skbuf copying.
4388 Frequency : only if NFS exporting from client
4390 Description: NULL pointer deref in ll_iocontrol() if chattr mknod file
4391 Details : If setting attributes on a file created under NFS that had
4392 never been opened it would be possible to oops the client
4393 if the file had no objects.
4396 Frequency : always for liblustre
4398 Description: liblustre client does MDS+OSTs setattr RPC for each write
4399 Details : When doing a write from a liblustre client, the client
4400 incorrectly issued an RPC to the MDS and each OST the file was
4401 striped over in order to update the timestamps. When writing
4402 with small chunks and many clients this could overwhelm the MDS
4403 with RPCs. In all cases it would slow down the write because
4404 these RPCs are unnecessary.
4406 Severity : enhancement
4408 Description: allow number of MDS service threads to be changed at module load
4409 Details : It is now possible to change the number of MDS service threads
4410 running. Adding "options mds mds_num_threads={N}" to the MDS's
4411 /etc/modprobe.conf will set the number of threads for the next
4412 time Lustre is restarted (assuming the "mds" module is also
4413 reloaded at that time). The default number of threads will
4414 stay the same, 32 for most systems.
4419 Description: OST crash if filesystem is unformatted or corrupt
4420 Details : If an OST is started on a device that has never been formatted
4421 or if the filesystem is corrupt and cannot even mount then the
4422 error handling cleanup routines would dereference a NULL pointer.
4427 Description: NULL pointer deref in llap_from_page.
4428 Details : get_cache_page_nowait can return a page with NULL (or otherwise
4429 incorrect) mapping if the page was truncated/reclaimed while it was
4430 searched for. Check for this condition and skip such pages when
4431 doing readahead. Introduce extra check to llap_from_page() to
4432 verify page->mapping->host is non-NULL (so page is not anonymous).
4435 Frequency : Sometimes when using sys_sendfile
4437 Description: "page not covered by a lock" warnings from ll_readpage
4438 Details : sendfile called ll_readpage without right page locks present.
4439 Now we introduced ll_file_sendfile that does necessary locking
4440 around call to generic_file_sendfile() much like we do in
4444 Frequency : with certain MDS communication failures at client mount time
4446 Description: NULL pointer deref after failed client mount
4447 Details : a client connection request may delayed by the network layer
4448 and not be sent until after the PTLRPC layer has timed out the
4449 request. If the client fails the mount immediately it will try
4450 to clean up before the network times out the request. Add a
4451 reference from the request import to the obd device and delay
4452 the cleanup until the network drops the request.
4455 Frequency : occasionally during client (re)connect
4457 Description: assertion failure during client (re)connect
4458 Details : processing a client connection request may be delayed by the
4459 client or server longer than the client connect timeout. This
4460 causes the client to resend the connection request. If the
4461 original connection request is replied in this interval, the
4462 client may trip an assertion failure in ptlrpc_connect_interpret()
4463 which thought it would be the only running connect process.
4466 Frequency : only with obd_echo servers and clients that are rebooted
4468 Description: kernel BUG accessing uninitialized data structure
4469 Details : When running an obd_echo server it did not start the ping_evictor
4470 thread, and when a client was evicted an uninitialized data
4471 structure was accessed. Start the ping_evictor in the RPC
4472 service startup instead of the OBD startup.
4474 Severity : enhancement
4475 Bugzilla : 10193 (patchless)
4476 Description: Remove dependency on various unexported kernel interfaces.
4477 Details : No longer need reparent_to_init, exit_mm, exit_files,
4478 sock_getsockopt, filemap_populate, FMODE_EXEC, put_filp.
4481 Frequency : rare (only users of deprecated and unsupported LDAP config)
4483 Description: write_conf for zeroconf mount queried LDAP incorrectly for client
4484 Details : LDAP apparently contains 'lustreName' attributes instead of
4485 'name'. A simple remapping of the name is sufficient.
4488 Frequency : rare (only with non-default dump_on_timeout debug enabled)
4490 Description: waiting_locks_callback trips kernel BUG if client is evicted
4491 Details : Running with the dump_on_timeout debug flag turned on makes
4492 it possible that the waiting_locks_callback() can try to dump
4493 the Lustre kernel debug logs from an interrupt handler. Defer
4494 this log dumping to the expired_lock_main() thread.
4496 Severity : enhancement
4498 Description: Support NFS exporting on 2.6 kernels.
4499 Details : Implement non-rawops metadata methods for NFS server to use without
4500 changing NFS server code.
4503 Frequency : very rare (synthetic metadata workload only)
4505 Description: two racing renames might cause an MDS thread to deadlock
4506 Details : Running the "racer" program may cause one MDS thread to rename
4507 a file from being the source of a rename to being the target of
4508 a rename at exactly the same time that another thread is doing
4509 so, and the second thread has already enqueued these locks after
4510 doing a lookup of the target and is trying to relock them in
4511 order. Ensure that we don't try to re-lock the same resource.
4514 Frequency : only very large systems with liblustre clients
4516 Description: slow eviction of liblustre clients with the "evict_by_nid" RPC
4517 Details : Use asynchronous set_info RPCs to send the "evict_by_nid" to
4518 all OSTs in parallel. This allows the eviction of stale liblustre
4519 clients to proceed much faster than if they were done in series,
4520 and also offers similar improvements for other set_info RPCs.
4525 Description: excessive CPU usage during initial read phase on client
4526 Details : During the initial read phase on a client, it would agressively
4527 retry readahead on the file, consuming too much CPU and impacting
4528 performance (since 1.4.5.8). Improve the readahead algorithm
4529 to avoid this, and also improve some other common cases (read
4530 of small files in particular, where "small" is files smaller than
4531 /proc/fs/lustre/llite/*/max_read_ahead_whole_mb, 2MB by default).
4536 Description: MDS crash when receiving packet with unknown intent.
4537 Details : Do not LBUG in unknown intent case, just return -EFAULT
4539 Severity : enhancement
4540 Bugzilla : 9293, 9385
4541 Description: MDS RPCs are serialised on client. This is unnecessary for some.
4542 Details : Do not serialize getattr (non-intent version) and statfs.
4545 Frequency : occasional, when OST network is overloaded/intermittent
4547 Description: client evicted by OST after bulk IO timeout
4548 Details : If a client sends a bulk IO request (read or write) the OST
4549 may evict the client if it is unresposive to its data GET/PUT
4550 request. This is incorrect if the network is overloaded (takes
4551 too long to transfer the RPC data) or dropped the OST GET/PUT
4552 request. There is no need to evict the client at all, since
4553 the pinger and/or lock callbacks will handle this, and the
4554 client can restart the bulk request.
4557 Frequency : Always when mmapping file with no objects
4559 Description: client crashes when mmapping file with no objects
4560 Details : Check that we actually have objects in a file before doing any
4561 operations on objects in ll_vm_open, ll_vm_close and
4567 Description: Request leak when working with deleted CWD
4568 Details : Introduce advanced request refcount tracking for requests
4569 referenced from lustre intent.
4571 Severity : Enhancement
4573 Description: Cache open file handles on client.
4574 Details : MDS now will return special lock along with openhandle, if
4575 requested and client is allowed to hold openhandle, even if unused,
4576 until such a lock is revoked. Helps NFS a lot, since NFS is opening
4577 closing files for every read/write openration.
4579 Severity : Enhancement
4581 Description: Cache open negative dentries on client when possible.
4582 Details : Guard negative dentries with UPDATE lock on parent dir, drop
4583 negative dentries on lock revocation.
4588 Description: Remounting a client read-only wasn't possible with a zconf mount
4589 Details : It wasn't possible to remount a client read-only with llmount.
4591 Severity : enhancement
4592 Description: Include MPICH 1.2.6 Lustre ADIO interface patch
4593 Details : In lustre/contrib/ or /usr/share/lustre in RPM a patch for
4594 MPICH is included to add Lustre-specific ADIO interfaces.
4595 This is based closely on the UFS ADIO layer and only differs
4596 in file creation, in order to allow the OST striping to be set.
4597 This is user-contributed code and not supported by CFS.
4602 Description: extended inode attributes (immutable, append-only) work improperly
4603 when 2.4 and 2.6 kernels are used on client/server or vice versa
4604 Details : Introduce kernel-independent values for these flags.
4606 Severity : enhancement
4609 Description: Allow fractional MB tunings for lustre in /proc/ filesystem.
4610 Details : Many of the /proc/ tunables can only be tuned at a megabyte
4611 granularity. Now, Fractional MB granularity is be supported,
4612 this is very useful for low memory system.
4614 Severity : enhancement
4616 Description: Getattr by fid
4617 Details : Getting a file attributes by its fid, obtaining UPDATE|LOOKUP
4618 locks, avoids extra getattr rpc requests to MDS, allows '/' to
4619 have locks and avoids getattr rpc requests for it on every stat.
4622 Frequency : Always, for filesystems larger than 2TB
4624 Description: ldiskfs crash at mount for filesystem larger than 2TB with mballoc
4625 Details : Kenrel kmalloc limits allocations to 128kB and this prevents
4626 filesystems larger than 2TB to be mounted with mballoc enabled.
4629 Frequency : Always, for 32-bit kernel without CONFIG_LBD and filesystem > 2TB
4631 Description: ldiskfs crash at mount for filesystem larger than 2TB with mballoc
4632 Details : If a 32-bit kernel is compiled without CONFIG_LBD enabled and a
4633 filesystems larger than 2TB is mounted then the kernel will
4634 silently corrupt the start of the filesystem. CONFIG_LBD is
4635 enabled for all CFS-supported kernels, but the possibility of
4636 this happening with a modified kernel config exists.
4638 Severity : enhancement
4640 Description: add client O_DIRECT support for 2.6 kernels
4641 Details : It is now possible to do O_DIRECT reads and writes to files
4642 in the Lustre client mountpoint on 2.6 kernel clients.
4644 Severity : enhancement
4646 Description: parallel glimpse, setattr, statfs, punch, destroy requests
4647 Details : Sends glimpse, setattr, statfs, punch, destroy requests to OSTs in
4648 parallel, not waiting for response from every OST before sending
4649 a rpc to the next OST.
4654 Description: setattr vs write race when updating file timestamps
4655 Details : Client processes that update a file timestamp into the past
4656 right after writing to the file (e.g. tar) it is possible that
4657 the updated file modification time can be reset to the current
4658 time due to a race between processing the setattr and write RPC.
4660 Severity : enhancement
4662 Description: Bring 'lfs find' closer in line with regular Linux find.
4663 Details : lfs find util supports -atime, -mtime, -ctime, -maxdepth, -print,
4664 -print0 options and obtains all the needed info through the lustre
4667 Severity : enhancement
4669 Description: support up to 1024 configured devices on one node
4670 Details : change obd_dev array from statically allocated to dynamically
4671 allocated structs as they are first used to reduce memory usage
4676 Description: Flush dirty partially truncated pages during truncate
4677 Details : Immediatelly flush partially truncated pages in filter_setattr,
4678 this way we completely avoid having any pages in page cache on OST
4679 and can retire ugly workarounds during writes to flush such pages.
4684 Description: i_sem vs transaction deadlock in mds_obd_destroy during unlink.
4685 Details : protect inode from truncation within vfs_unlink() context
4686 just take a reference before calling vfs_unlink() and release it
4687 when parent's i_sem is free.
4692 Description: last_id value checked outside lock on OST caused LASSERT failure
4693 Details : If there were multiple MDS->OST object precreate requests in
4694 flight, it was possible that the OST's last object id was checked
4695 outside a lock and incorrectly tripped an assertion. Move checks
4696 inside locks, and discard old precreate requests.
4699 Frequency : always, if extents are used on OSTs
4701 Description: index ei_leaf_hi (48-bit extension) is not zeroed in extent index
4702 Details : OSTs using the extents format would not zero the high 16 bits of
4703 the index physical block number. This is not a problem for any
4704 OST filesystems smaller than 16TB, and no kernels support ext3
4705 filesystems larger than 16TB yet. This is fixed in 1.4.7 (all
4706 new/modified files) and can be fixed for existing filesystems
4707 with e2fsprogs-1.39-cfs1.
4712 Description: import connection selection may be incorrect if timer wraps
4713 Details : Using a 32-bit jiffies timer with HZ=1000 may cause backup
4714 import connections to be ignored if the 32-bit jiffies counter
4715 wraps. Use a 64-bit jiffies counter.
4718 Frequency : very large clusters immediately after boot
4720 Description: LNET request buffers exhausted under heavy short-term load
4721 Details : If a large number of client requests are generated on a service
4722 that has previously never seen so many requests it is possible
4723 that the request buffer growth cannot keep up with the spike in
4724 demand. Instead of dropping incoming requests, they are held in
4725 the LND until the RPC service can accept more requests.
4728 Frequency : Sometimes during replay
4730 Description: Assertion failure in ll_local_open after replay.
4731 Details : If replay happened on an open request reply before we were able
4732 to set replay handler, reply will become not swabbed tripping the
4733 assertion in ll_local_open. Now we set the handler right after
4734 recognising of open request
4737 Frequency : very rare
4739 Description: kernel reports "badness in vsnprintf"
4740 Details : Reading from the "recovery_status" /proc file in small chunks
4741 may cause a negative length in lprocfs_obd_rd_recovery_status()
4742 call to vsnprintf() (which is otherwise harmless). Exit early
4743 if there is no more space in the output buffer.
4745 Severity : enhancement
4747 Description: clear OBD RPC statistics by writing to them
4748 Details : It is now possible to clear the OBD RPC statistics by writing
4749 to the "stats" file.
4754 Description: Client mtime is not the same on different clients after utimes
4755 Details : In some cases, the client was using the utimes() syscall on
4756 a file cached on another node. The clients now validate the
4757 ctime from the MDS + OSTs to determine which one is right.
4762 Description: Inability to activate failout mode
4763 Details : lconf script incorrectly assumed that in pythong string's numeric
4764 value is used in comparisons.
4767 Frequency : always with multiple stripes per file
4769 Description: Inefficient object allocation for mutli-stripe files
4770 Details : When selecting which OSTs to stripe files over, for files with
4771 a stripe count that divides evenly into the number of OSTs,
4772 the MDS is always picking the same starting OST for each file.
4773 Return the OST selection heuristic to the original design.
4778 Description: mount failures may take full timeout to return an error
4779 Details : Under some heavy load conditions it is possible that a
4780 failed mount can wait for the full obd_timeout interval,
4781 possibly several minutes, before reporting an error.
4782 Instead return an error as soon as the status is known.
4784 Frequency : quota enabled and large files being deleted
4786 Description: releasing more than 4GB of quota at once hangs OST
4787 Details : If a user deletes more than 4GB of files on a single OST it
4788 will cause the OST to spin in an infinite loop. Release
4789 quota in < 4GB chunks, or use a 64-bit value for 1.4.7.1+.
4794 Description: statfs data retrieved from /proc may be stale or zero
4795 Details : When reading per-device statfs data from /proc, in the
4796 {kbytes,files}_{total,free,avail} files, it may appear
4797 as zero or be out of date.
4800 Frequency : systems with MD RAID1 external journal devices
4802 Description: lconf's call to blkid is confused by RAID1 journal devices
4803 Details : Use the "blkid -l" flag to locate the MD RAID device instead
4804 of returning all block devices that match the journal UUID.
4807 Frequency : always, for aggregate stripe size over 4GB
4809 Description: assertion fails when trying to use 4GB stripe size
4810 Details : Use "setstripe" to set stripe size over 4GB will fail the kernel,
4811 complaining "ASSERTION(lsm->lsm_xfersize != 0)"
4814 Frequency : always on ppc64
4816 Description: the first write on an ext3 filesystem with mballoc got stuck
4817 Details : ext3_mb_generate_buddy() uses find_next_bit() which does not
4818 perform endianness conversion.
4820 ------------------------------------------------------------------------------
4822 02-14-2006 Cluster File Systems, Inc. <info@clusterfs.com>
4824 * WIRE PROTOCOL CHANGE. This version of Lustre networking WILL NOT
4825 INTEROPERATE with older versions automatically. Please read the
4826 user documentation before upgrading any part of a live system.
4827 * WARNING: Lustre networking configuration changes are required with
4828 this release. See https://bugzilla.clusterfs.com/show_bug.cgi?id=10052
4831 * Support for kernels:
4832 2.6.9-22.0.2.EL (RHEL 4)
4833 2.6.5-7.244 (SLES 9)
4834 2.6.12.6 vanilla (kernel.org)
4837 Severity : enhancement
4838 Bugzilla : 7981/8208
4839 Description: Introduced Lustre Networking (LNET)
4840 Details : LNET is new networking infrastructure for Lustre, it includes
4841 a reorganized network configuration mode (see the user
4842 documentation for full details) as well as support for routing
4843 between different network fabrics. Lustre Networking Devices
4844 (LNDs) for the supported network fabrics have also been
4845 created for this new infrastructure.
4847 Severity : enhancement
4848 Description: Introduced Access control lists
4849 Details : clients can set ACLs on files and directories in order to have
4850 more fine-grained permissions than the standard Unix UGO+RWX.
4851 The MDS must be started with the "-o acl" mount option.
4853 Severity : enhancement
4854 Description: Introduced filesystem quotas
4855 Details : Administrators may now establish per-user quotas on the
4858 Severity : enhancement
4860 Description: Configuration change for the XT3
4861 The PTLLND is now used to run Lustre over Portals on the XT3
4862 The configure option(s) --with-cray-portals are no longer used.
4863 Rather --with-portals=<path-to-portals-includes> is used to
4864 enable building on the XT3. In addition to enable XT3 specific
4865 features the option --enable-cray-xt3 must be used.
4870 Description: Running on many-way SMP OSTs can trigger oops in llcd_send()
4871 Details : A race between allocating a new llcd and re-getting the llcd_lock
4872 allowed another thread to grab newly-allocated llcd.
4874 Severity : enhancement
4876 Description: 2.6 OST async journal commit and locking fix to improve performance
4877 Details : The filter_direct_io()+filter_commitrw_write() journal commits for
4878 2.6 kernels are now async as they already were in 2.4 kernels so
4879 that they can commit concurrently with the network bulk transfer.
4880 For block-allocated files the filter allocation semaphore is held
4881 to avoid filesystem fragmentation during allocation. BKL lock
4882 removed for 2.6 xattr operations where it is no longer needed.
4887 Description: lconf incorrectly determined whether two IP networks could talk
4888 Details : In some more complicated routing and multiple-network
4889 configurations, lconf will avoid trying to make a network
4890 connection to a disjoint part of the IP space. It was doing the
4891 math incorrectly for one set of cases.
4896 Description: Fix for potential infinite loop processing records in an llog.
4897 Details : If an llog record is corrupted/zeroed, it is possible to loop
4898 forever in llog_process(). Validate the llog record length
4899 and skip the remainder of the block on error.
4902 Frequency : occasional (liblustre only)
4904 Description: liblustre could not open files whose last component is a symlink
4905 Details : sysio_path_walk() would incorrectly pass the open intent to
4906 intermediate path components.
4909 Frequency : rare (liblustre only with non-standard tuning)
4910 Bugzilla : 7201 (7350)
4911 Description: Tuning the MDC DLM LRU size to zero triggers client LASSERT
4912 Details : llu_lookup_finish_locks() tries to set lock data on a lock
4913 after it has been released, only do this for referenced locks
4915 Severity : enhancement
4917 Description: specifying an (invalid) directory default stripe_size of -1
4918 would reset the directory default striping
4919 Details : stripe_size -1 was used internally to signal directory stripe
git://git.whamcloud.com / fs / lustre-release.git / blob