3 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
4 for currently supported client and server kernel versions.
5 * Server primary kernels built and tested during release cycle:
6 3.10.0-1062.18.1.el7 (RHEL7.7)
7 3.10.0-1127.8.2.el7 (RHEL7.8)
8 * Other server kernels known to build and work at some point (others may also work):
9 3.10.0-862.14.4.el7 (RHEL7.5)
10 3.10.0-957.27.2.el7 (RHEL7.6)
11 4.18.0-80.11.2.el8 (RHEL8.0)
12 4.18.0-147.8.1.el8 (RHEL8.1)
13 4.18.0-193.6.3.el8 (RHEL8.2)
14 4.4.120-92.70 (SLES12 SP2)
15 4.4.180-94.100 (SLES12 SP3)
16 4.4.0-131 (Ubuntu 16.04)
17 4.15.0-32 (Ubuntu 18.04)
18 vanilla linux 4.6.7 (ZFS only)
19 * ldiskfs needs an ldiskfs patch series for that kernel, ZFS does not
20 * Client primary unpatched kernels built and tested during release cycle:
21 3.10.0-1062.18.1.el7 (RHEL7.7)
22 3.10.0-1127.8.2.el7 (RHEL7.8)
23 4.12.14-95.48 (SLES12 SP4)
24 4.12.14-122.20 (SLES12 SP5)
25 4.12.14-197.37 (SLES15 SP1)
26 4.15.0-48 (Ubuntu 18.04)
27 * Other clients known to build on unpatched kernels at some point (others may also work):
28 3.10.0-862.14.4.el7 (RHEL7.5)
29 3.10.0-957.27.2.el7 (RHEL7.6)
30 4.14.0-49.13.1.el7a (RHEL7.5)
31 4.14.0-115.2.2.el7a (RHEL7.6)
32 4.18.0-80.11.2.el8 (RHEL8.0)
33 4.18.0-147.8.1.el8 (RHEL8.1)
34 4.18.0-193.6.3.el8 (RHEL8.2)
35 4.4.120-92.70 (SLES12 SP2)
36 4.4.180-94.100 (SLES12 SP3)
37 4.4.0-131 (Ubuntu 16.04)
38 vanilla linux 4.15.0 (ZFS only)
39 * Recommended e2fsprogs version: 1.45.6.wc1 or newer
40 * Recommended ZFS version: 0.8.3
41 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
42 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
43 stack overflow. For more information, please refer to bugzilla 17630.
44 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
45 the re-exporting nodes to be mounted with "32bitapi" mount option
46 * Note for Ubuntu 14.04.5:
47 The 4.4.0 series is the HWE Kernel derived from 16.04 (Xenial).
48 See: https://wiki.ubuntu.com/Kernel/LTSEnablementStack
50 --------------------------------------------------------------------------------
54 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
55 for currently supported client and server kernel versions.
56 * Server primary kernels built and tested during release cycle:
57 3.10.0-1062.1.1.el7 (RHEL7.7)
58 * Other server kernels known to build and work at some point (others may also work):
59 3.10.0-862.14.4.el7 (RHEL7.5)
60 3.10.0-957.27.2.el7 (RHEL7.6)
61 4.18.0-80.7.1.el8 (RHEL8.0)
62 4.4.120-92.70 (SLES12 SP2)
63 4.4.180-94.100 (SLES12 SP3)
64 4.4.0-131 (Ubuntu 16.04)
65 4.15.0-32 (Ubuntu 18.04)
66 vanilla linux 4.6.7 (ZFS only)
67 * ldiskfs needs an ldiskfs patch series for that kernel, ZFS does not
68 * Client primary unpatched kernels built and tested during release cycle:
69 3.10.0-1062.1.1.el7 (RHEL7.7)
70 4.12.14-95.29 (SLES12 SP4)
71 4.15.0-48 (Ubuntu 18.04)
72 * Other clients known to build on unpatched kernels at some point (others may also work):
73 3.10.0-862.14.4.el7 (RHEL7.5)
74 3.10.0-957.27.2.el7 (RHEL7.6)
75 4.14.0-49.13.1.el7a (RHEL7.5)
76 4.14.0-115.2.2.el7a (RHEL7.6)
77 4.18.0-80.7.1.el8 (RHEL8.0)
78 4.4.120-92.70 (SLES12 SP2)
79 4.4.180-94.100 (SLES12 SP3)
80 4.4.0-131 (Ubuntu 16.04)
81 vanilla linux 4.15.0 (ZFS only)
82 * Recommended e2fsprogs version: 1.45.2.wc1 or newer
83 * Recommended ZFS / SPL version: 0.7.13
84 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
85 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
86 stack overflow. For more information, please refer to bugzilla 17630.
87 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
88 the re-exporting nodes to be mounted with "32bitapi" mount option
89 * Note for Ubuntu 14.04.5:
90 The 4.4.0 series is the HWE Kernel derived from 16.04 (Xenial).
91 See: https://wiki.ubuntu.com/Kernel/LTSEnablementStack
93 --------------------------------------------------------------------------------
97 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
98 for currently supported client and server kernel versions.
99 * Server known to build on the following kernels (others may also work):
100 2.6.32-696.18.7.el6 (RHEL6.9)
101 3.10.0-862.14.4.el7 (RHEL7.5)
102 3.10.0-957.el7 (RHEL7.6)
103 4.4.120-92.70 (SLES12 SP2)
104 4.4.162-94.69 (SLES12 SP3)
105 4.4.0-131 (Ubuntu 16.04)
106 4.15.0-32 (Ubuntu 18.04)
107 vanilla linux 4.6.7 (ZFS only)
108 * ldiskfs needs an ldiskfs patch series for that kernel, ZFS does not
109 * Client known to build on unpatched kernels (others may also work):
110 2.6.32-696.18.7.el6 (RHEL6.9)
111 2.6.32-754.6.3.el6 (RHEL6.10)
112 3.10.0-862.14.4.el7 (RHEL7.5)
113 3.10.0-957.el7 (RHEL7.6)
114 4.14.0-49.13.1.el7a (RHEL7.5)
115 4.14.0-115.2.2.el7a (RHEL7.6)
116 4.4.120-92.70 (SLES12 SP2)
117 4.4.162-94.69 (SLES12 SP3)
118 4.4.0-131 (Ubuntu 16.04)
119 4.15.0-32 (Ubuntu 18.04)
120 vanilla linux 4.15.0 (ZFS only)
121 * Recommended e2fsprogs version: 1.44.3.wc1 or newer
122 * Recommended ZFS / SPL version: 0.7.9
123 * Tested with ZFS / SPL version: 0.7.9
124 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
125 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
126 stack overflow. For more information, please refer to bugzilla 17630.
127 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
128 the re-exporting nodes to be mounted with "32bitapi" mount option
129 * Note for Ubuntu 14.04.5:
130 The 4.4.0 series is the HWE Kernel derived from 16.04 (Xenial).
131 See: https://wiki.ubuntu.com/Kernel/LTSEnablementStack
133 --------------------------------------------------------------------------------
135 2018-04-03 Intel Corporation
137 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
138 for currently supported client and server kernel versions.
139 * Server known to build on patched kernels:
140 2.6.32-431.29.2.el6 (RHEL6.5)
141 2.6.32-504.30.3.el6 (RHEL6.6)
142 2.6.32-573.26.1.el6 (RHEL6.7)
143 2.6.32-642.15.1.el6 (RHEL6.8)
144 2.6.32-696.18.7.el6 (RHEL6.9)
145 3.10.0-693.21.1.el7 (RHEL7.4)
146 3.0.101-0.47.71 (SLES11 SP3)
147 3.0.101-107 (SLES11 SP4)
148 3.12.74-60.64.40 (SLES12 SP1)
149 4.4.114-92.64 (SLES12 SP2)
150 4.4.114-94.11 (SLES12 SP3)
151 3.13.0-101 (Ubuntu 14.04, ZFS only)
152 4.4.0-85.108 (Ubuntu 14.04.5 LTS)
153 4.4.0-116 (Ubuntu 16.04)
154 vanilla linux 4.6.7 (ZFS only)
155 * Client known to build on unpatched kernels:
156 2.6.32-431.29.2.el6 (RHEL6.5)
157 2.6.32-504.30.3.el6 (RHEL6.6)
158 2.6.32-573.26.1.el6 (RHEL6.7)
159 2.6.32-642.15.1.el6 (RHEL6.8)
160 2.6.32-696.18.7.el6 (RHEL6.9)
161 3.10.0-693.21.1.el7 (RHEL7.4)
162 3.0.101-0.47.71 (SLES11 SP3)
163 3.0.101-107 (SLES11 SP4)
164 3.12.74-60.64.40 (SLES12 SP1)
165 4.4.114-92.64 (SLES12 SP2)
166 4.4.114-94.11 (SLES12 SP3)
167 3.13.0-101 (Ubuntu 14.04)
168 4.4.0-85.108 (Ubuntu 14.04.5 LTS)
169 4.4.0-116 (Ubuntu 16.04)
171 * Recommended e2fsprogs version: 1.42.13.wc6 or newer
172 * Recommended ZFS / SPL version: 0.7.6
173 * Tested with ZFS / SPL version: 0.7.6
174 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
175 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
176 stack overflow. For more information, please refer to bugzilla 17630.
177 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
178 the re-exporting nodes to be mounted with "32bitapi" mount option
179 * Note for Ubuntu 14.04.5:
180 The 4.4.0 series is the HWE Kernel derived from 16.04 (Xenial).
181 See: https://wiki.ubuntu.com/Kernel/LTSEnablementStack
183 --------------------------------------------------------------------------------
185 2017-07-05 Intel Corporation
187 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
188 for currently supported client and server kernel versions.
189 * Server known to build on patched kernels:
190 2.6.32-431.29.2.el6 (RHEL6.5)
191 2.6.32-504.30.3.el6 (RHEL6.6)
192 2.6.32-573.26.1.el6 (RHEL6.7)
193 2.6.32-642.15.1.el6 (RHEL6.8)
194 2.6.32-696.3.1.el6 (RHEL6.9)
195 3.10.0-514.21.1.el7 (RHEL7.3)
196 3.0.101-0.47.71 (SLES11 SP3)
197 3.0.101-100 (SLES11 SP4)
198 3.12.74-60.64.40 (SLES12 SP1)
199 4.4.59-92.17 (SLES12 SP2)
200 vanilla linux 4.6.7 (ZFS only)
201 * Client known to build on unpatched kernels:
202 2.6.32-431.29.2.el6 (RHEL6.5)
203 2.6.32-504.30.3.el6 (RHEL6.6)
204 2.6.32-573.26.1.el6 (RHEL6.7)
205 2.6.32-642.15.1.el6 (RHEL6.8)
206 2.6.32-696.3.1.el6 (RHEL6.9)
207 3.10.0-514.21.1.el7 (RHEL7.3)
208 3.0.101-0.47.71 (SLES11 SP3)
209 3.0.101-100 (SLES11 SP4)
210 3.12.74-60.64.40 (SLES12 SP1)
211 4.4.59-92.17 (SLES12 SP2)
213 * Recommended e2fsprogs version: 1.42.13.wc4 or newer
214 * Recommended ZFS / SPL version: 0.6.5.9
215 * Tested with ZFS / SPL version: 0.6.5.9
216 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
217 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
218 stack overflow. For more information, please refer to bugzilla 17630.
219 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
220 the re-exporting nodes to be mounted with "32bitapi" mount option
222 --------------------------------------------------------------------------------
224 12-07-2016 Intel Corporation
226 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
227 for currently supported client and server kernel versions.
228 * Server known to build on patched kernels:
229 2.6.32-431.29.2.el6 (RHEL6.5)
230 2.6.32-504.30.3.el6 (RHEL6.6)
231 2.6.32-573.26.1.el6 (RHEL6.7)
232 2.6.32-642.11.1.el6 (RHEL6.8)
233 3.10.0-514.el7 (RHEL7.3)
234 3.0.101-0.47.71 (SLES11 SP3)
235 3.0.101-80 (SLES11 SP4)
236 3.12.62-60.64.8 (SLES12 SP1)
237 vanilla linux 4.5.7 (ZFS only)
238 * Client known to build on unpatched kernels:
239 2.6.32-431.29.2.el6 (RHEL6.5)
240 2.6.32-504.30.3.el6 (RHEL6.6)
241 2.6.32-573.26.1.el6 (RHEL6.7)
242 2.6.32-642.11.1.el6 (RHEL6.8)
243 3.10.0-514.el7 (RHEL7.3)
244 3.0.101-0.47.71 (SLES11 SP3)
245 3.0.101-80 (SLES11 SP4)
246 3.12.62-60.64.8 (SLES12 SP1)
248 * Recommended e2fsprogs version: 1.42.13.wc5 or newer
249 * Recommended ZFS / SPL version: 0.6.5.8
250 * Tested with ZFS / SPL version: 0.6.5.8
251 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
252 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
253 stack overflow. For more information, please refer to bugzilla 17630.
254 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
255 the re-exporting nodes to be mounted with "32bitapi" mount option
257 --------------------------------------------------------------------------------
259 02-29-2016 Intel Corporation
261 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
262 for currently supported client and server kernel versions.
263 * Server known to build on patched kernels:
264 2.6.32-431.29.2.el6 (RHEL6.5)
265 2.6.32-504.30.3.el6 (RHEL6.6)
266 2.6.32-573.12.1.el6 (RHEL6.7)
267 3.10.0-327.3.1.el7 (RHEL7.2)
268 3.0.101-0.47.71 (SLES11 SP3)
269 3.0.101-68 (SLES11 SP4)
270 vanilla linux 4.2.1 (ZFS only)
271 * Client known to build on unpatched kernels:
272 2.6.32-431.29.2.el6 (RHEL6.5)
273 2.6.32-504.30.3.el6 (RHEL6.6)
274 2.6.32-573.12.1.el6 (RHEL6.7)
275 3.10.0-327.3.1.el7 (RHEL7.2)
276 3.0.101-0.47.71 (SLES11 SP3)
277 3.0.101-68 (SLES11 SP4)
280 * Recommended e2fsprogs version: 1.42.13.wc4 or newer
281 * Recommended ZFS / SPL version: 0.6.4.2
282 * Tested with ZFS / SPL version: 0.6.4.2
283 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
284 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
285 stack overflow. For more information, please refer to bugzilla 17630.
286 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
287 the re-exporting nodes to be mounted with "32bitapi" mount option
289 --------------------------------------------------------------------------------
291 03-10-2015 Intel Corporation
293 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
294 for currently supported client and server kernel versions.
295 * Server known to build on patched kernels:
296 2.6.32-431.29.2.el6 (RHEL6.5)
297 2.6.32-504.8.1.el6 (RHEL6.6)
298 3.0.101-0.46 (SLES11 SP3)
299 * Client known to build on unpatched kernels:
300 2.6.32-431.29.2.el6 (RHEL6.5)
301 2.6.32-504.8.1.el6 (RHEL6.6)
302 3.10.0-123.20.1.el7 (RHEL7)
303 3.0.101-0.46 (SLES11 SP3)
304 * Recommended e2fsprogs version: 1.42.9.wc1 or newer
305 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
306 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
307 stack overflow. For more information, please refer to bugzilla 17630.
308 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
309 the re-exporting nodes to be mounted with "32bitapi" mount option
311 Severity : enhancement
313 Description: control OST-index in IDIF via ROCOMPAT flag.
314 Details : Introduce new flag OBD_ROCOMPAT_IDX_IN_IDIF that is stored in the
315 last_rcvd file. For new formatted OST device, it will be auto set;
316 for the case of upgrading from old OST device, you can enable it
317 via the lproc interface osd-ldiskfs.index_in_idif. With such flag
318 enabled, for new created OST-object, its IDIF-in-LMA will contain
319 the OST-index; for the existing OST-object, the OSD will convert
320 old format IDIF as new format IDIF with OST-index stored in the
321 LMA EA when accessing such OST-object or via OI scrub. Once such
322 flag is enabled, it cannot be reverted back, so the system cannot
323 be downgraded to the orignal incompatible version.
325 --------------------------------------------------------------------------------
327 07-30-2014 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 2.6.32-431.20.3.el6 (RHEL6.5)
333 3.0.101-0.31 (SLES11 SP3)
334 * Client known to build on unpatched kernels:
335 2.6.32-431.20.3.el6 (RHEL6.5)
336 3.0.101-0.31 (SLES11 SP3)
339 * Recommended e2fsprogs version: 1.42.9.wc1 or newer
340 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
341 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
342 stack overflow. For more information, please refer to bugzilla 17630.
343 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
344 the re-exporting nodes to be mounted with "32bitapi" mount option
346 --------------------------------------------------------------------------------
348 10-31-2013 Intel Corporation
350 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
351 * Server builds for kernels:
352 2.6.32-358.18.1.el6 (RHEL6.4)
353 * Client builds for unpatched kernels:
354 2.6.32-358.18.1.el6 (RHEL6.4)
355 3.0.92-0.8 (SLES11 SP3)
357 * Recommended e2fsprogs version: 1.42.7.wc2 or newer
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 05-30-2013 Intel Corporation
368 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
369 * Server builds for kernels:
370 2.6.32-358.6.2.el6 (RHEL6.4)
371 * Client builds for unpatched kernels:
372 2.6.32-358.6.2.el6 (RHEL6.4)
373 3.0.74-0.6.6 (SLES11 SP2)
375 * Recommended e2fsprogs version: 1.42.7.wc1 or newer
376 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
377 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
378 stack overflow. For more information, please refer to bugzilla 17630.
379 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
380 the re-exporting nodes to be mounted with "32bitapi" mount option
382 --------------------------------------------------------------------------------
384 9-30-2012 Whamcloud, Inc.
386 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
387 * Server support for kernels:
388 2.6.32-279.5.1.el6 (RHEL6)
389 * Client support for unpatched kernels:
390 2.6.32-279.5.1.el6 (RHEL6)
392 2.6.18-238.19.1.el5 (RHEL5)
393 2.6.16.60-0.69.1 (SLES10)
394 * Recommended e2fsprogs version: 1.41.90.wc4
395 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
396 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
397 stack overflow. For more information, please refer to bugzilla 17630.
398 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
399 the re-exporting nodes to be mounted with "32bitapi" mount option
401 Severity : enhancement
403 Description: Update RHEL6.3 kernel to 2.6.32-279.5.1.el6.
405 Severity : enhancement
407 Description: Update RHEL6.3 kernel to 2.6.32-279.2.1.el6.
409 Severity : enhancement
411 Description: Update RHEL6.2 kernel to 2.6.32-220.17.1.el6
413 Severity : enhancement
415 Description: Update RHEL6.2 kernel to 2.6.32-220.13.1.el6
417 Severity : enhancement
419 Description: Update RHEL6.2 kernel to 2.6.32-220.7.1.el6
421 --------------------------------------------------------------------------------
423 04-30-2012 Whamcloud, Inc.
425 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
426 * Server support for kernels:
427 2.6.32-220.4.2.el6 (RHEL6)
428 * Client support for unpatched kernels:
429 2.6.32-220.4.2.el6 (RHEL6)
431 2.6.18-238.19.1.el5 (RHEL5)
432 2.6.16.60-0.69.1 (SLES10)
433 * Recommended e2fsprogs version: 1.41.90.wc4
434 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
435 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
436 stack overflow. For more information, please refer to bugzilla 17630.
437 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
438 the re-exporting nodes to be mounted with "32bitapi" mount option
440 Severity : enhancement
442 Description: Update RHEL6.2 kernel to 2.6.32-220.4.2.el6
444 Severity : enhancement
446 Description: Update RHEL6.2 kernel to 2.6.32-220.4.1.el6
448 Severity : enhancement
450 Description: Update RHEL6.2 kernel to 2.6.32-220.el6
452 Severity : enhancement
454 Description: Update RHEL6.1 kernel to 2.6.32-131.17.1.el6
456 --------------------------------------------------------------------------------
458 09-30-2011 Whamcloud, Inc.
460 * Server support for kernels:
461 2.6.18-238.19.1.el5 (RHEL5)
462 2.6.32-131.6.1.el6 (RHEL6)
463 * Client support for unpatched kernels:
464 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
465 2.6.18-238.19.1.el5 (RHEL5)
466 2.6.32-131.6.1.el6 (RHEL6)
467 2.6.29.4-167.fc11 (FC11)
468 2.6.16.60-0.69.1 (SLES10)
470 * Recommended e2fsprogs version: 1.41.90.wc3
471 * NFS export disabled when stack size < 8192. Since the NFSv4 export of
472 Lustre filesystem with 4K stack may cause a stack overflow. For more
473 information, please refer to bugzilla 17630.
474 * NFSv4 reexport requires Lustre client on the re-exporting nodes to be
475 mounted with "32bitapi" mount option
477 Severity : enhancement
478 Description: Update RHEL6.1 kernel to 2.6.32-131.6.1.el6
480 Severity : enhancement
481 Description: Update RHEL5.6 kernel to 2.6.18-238.19.1.el5
483 Severity : enhancement
484 Description: Update RHEL5.6 kernel to 2.6.18-238.9.1.el5
487 Frequency : only with SLES10
489 Description: Use OFED "KMP" provided by Novell
490 Details : SLES10 SP3 ships with OFED in a separate "KMP" package.
491 Lustre is now built against this package. That means you need to
492 install the ofed-kmp package from Novell for the patchless client.
493 Note that the ofed-kmp that Novell ships may not exactly match the
494 kernel version but should still be compatible.
496 Severity : enhancement
498 Description: Update RHEL5.5 kernel to 2.6.18-194.17.1.el5
499 Update OEL5.5 kernel to 2.6.18-194.17.1.0.1.el5.
501 --------------------------------------------------------------------------------
503 2010-07-15 Oracle, Inc.
505 * Server support for kernels:
506 2.6.18-164.11.1.el5 (RHEL 5)
507 2.6.18-164.11.1.0.1.el5 (OEL 5)
508 * Client support for unpatched kernels:
509 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
510 2.6.18-164.11.1.el5 (RHEL 5),
511 2.6.18-164.11.1.0.1.el5 (OEL 5)
512 2.6.16.60-0.42.8 (SLES 10),
514 2.6.29.4-167.fc11 (FC11)
515 * Recommended e2fsprogs version: 1.41.10-sun2
517 Severity : enhancement
519 Description: Distribute lfs_migrate script with lustre
520 Details : The lfs_migrate mentioned in the manual is now included in the
521 lustre tree in lustre/scripts and will distributed in lustre RPMs.
523 Severity : enhancement
525 Description: Failover nid list added to import proc file.
527 Severity : enhancement
529 Description: "som_preview" mount option required to use SOM
530 Details : The clients will need to use som_preview mount option
534 Description: Lustre does not do 1MB IOs to HW RAID
535 Details : Bump MAX_PHYS/HW_SEGMENTS and SG_ALL to 256 in the RHEL5 kernel.
537 Severity : enhancement
539 Description: lfs quota output is not very convenient for awk/sed-parsing
540 Details : Some positions in lfs quota output table could be empty or
541 non-empty which made it hard to parse it with scripts, now
542 a dash is put instead of space where there is not supposed
547 Description: Update to OFED 1.5.1
548 Details : OFED 1.5.1 will be used in environments that don't use
549 vendor supplied OFED.
553 Description: Remove lnet. prefix from lctl params display
554 Details : Also renamed lnet.memused to lnet_memused to
559 Description: Use conf_param -d to remove permanant settings.
563 Description: Add list_param -R
564 Details : lctl list_param now supports recursive mode.
567 Bugzilla : 21829, 22312
568 Description: Fix llobdstat.
569 Details : 1. Now updating the write_bytes counter
570 2. Port script fixes from 1.8.
574 Description: decrease the usage of memory on clients.
575 Details : 1. On clients, recycle dentries and inodes unused.
576 2. Delete the code related to ll_deathrow(att 6215 in bug 1443). It
581 Description: llobdstat fix and enhancement
582 Details : add a counter to set a limit to how many samples will be returned
583 fix a wildcard in the path to limit to obdfilter stats only
587 Description: lfs check only allowed for root
588 Details : Code cleanups and added a regresion test of non-root use of
593 Description: MDS Oopses in ptlrpc_dispatch_difficult_reply()/__list_add()
594 Details : Fix hr->hr_index overload.
598 Description: Fix FIEMAP functionality.
602 Description: FC11 patchless client support added
606 Description: Update RHEL5.4 kernel to 2.6.18-164.11.1.el5 and
607 OEL5.4 kernel to 2.6.18-164.11.1.0.1.el5.
611 Description: New acc-sm script and various test-framework enhancments.
615 Description: Previously landed patch was reverted.
619 Description: Update SLES10 kernel to 2.6.16.60-0.42.8 (patchless)
623 Description: Add ldiskfs maxdirsize mount option.
627 Description: Fix mkfs.lustre support for 16TB LUNs.
631 Description: Print mmp_check_interval
632 Details : print mmp_check_interval and make it possible to abort
633 mount operation in case it takes too long.
637 Description: Add cascading_rw to lustre mpi tests.
641 Description: Store SOM data in EA.
642 Details : Also several SOM related cleanups.
646 Description: Add refcount to md_open_data
650 Description: Use refcount on lov_request_set to prevent memory corruption.
654 Description: Statahead: use parent dir's i_mutex for synchronization.
658 Description: loadgen improvements.
662 Description: Enable peer health detection in ksocklnd and ko2iblnd.
666 Description: Control DCACHE_LUSTRE_INVALID flag with MDS_INODE_LOOKCUP lock
670 Description: Fix race on obd_devcie:md_stats
674 Description: Fix SLES 11 client builds.
675 Details : Resolved uninitalized pointer gcc warnings.
679 Description: Allow non-root access for "lfs check".
680 Details : Added a check in obd_class_ioctl() for OBD_IOC_PING_TARGET.
684 Description: Clear imp_force_reconnect correctly in ptlrpc_connect_interpret()
686 Severity : enhancement
688 Description: Update kernel to RHEL5.4 2.6.18-164.6.1.el5 and
689 OEL5 2.6.18-164.6.1.0.1.el5(Both in-kernel OFED enabled).
693 Description: Sanity test_27p fails.
694 Details : OBD_FAIL_OST_ENOSPC has been moved to filter_statfs().
695 Sanity tests 27 were cleaned up and improved.
699 Description: Unify req format on client/servers
700 Details : Use new req_capsule API [almost] everywhere instead of old PTLRPC
701 buffers and swabbers approach..
706 Description: drop join file stuff.
708 Severity : enhancement
710 Description: Add 2.6.27 and SLES11 (patchless client) support
714 Description: Fix in ptlrpc_expire_one_request() to print the signed time
717 Severity : enhancement
719 Description: Build kernels (RHEL5, OEL5 and SLES10/11) using the vendor's own
722 Severity : enhancement
724 Description: Remove set_info(KEY_UNLINKED) from MDS/OSC
726 Severity : enhancement
728 Description: correctly handle big reply message.
729 Details : send LNet event if reply is bigger than buffer and adjust this buffer
734 Description: Drop unnecessary __GFP_NOMEMALLOC flag from filter_get_page()
736 Severity : enhancement
738 Description: Add support for OEL5.
742 Description: Conf-sanity.sh 50g test - deactivated OST should not cause a panic.
744 Severity : enhancement
746 Description: "weak-modules" support
747 Details : Implement "weak-modules" support which enables kernel modules
748 to be used with any kernel that implements the same kABI. In
749 order to achieve this modules are now installed in
750 /lib/modules/$(uname -r)/updates/kernel on all distributions.
754 Description: client could not reconnect to OST because of an active request.
755 Details : The client now retries to reconnect to the same server, if a
756 connect request failed with EBUSY or -EAGAIN. Test case 35b
757 conf-sanity.sh has been created.
761 Description: Don't allow make backward step on assignin osc next id.
762 Details : race between allocation next id and ll_sync thread can be cause
763 of set wrong osc next id and can be kill valid ost objects.
767 Description: Changes in raid5-large-io-rhel5.patch to calculate sectors properly
771 Description: Increase the default BLK_DEF_MAX_SECTORS value for RHEL5 and SLES11
775 Description: Error handling in osc_statfs_interpret() has been improved.
776 Details : Check in osc_statfs_interpret() for EBADR.
780 Description: Do not send statfs() requests to OSTs disabled by administrator.
781 Details : Check in lov_prep_statfs_set() for non-NULL ltd_exp.
783 Severity : enhancement
785 Description: Added sanity-quota test to for lfs(1) to work with numeric uid/gid.
787 Severity : enhancement
789 Description: Update kernel to RHEL5 2.6.18-128.7.1.el5.
791 Severity : enhancement
793 Description: Enhancement for lfs(1) command to use numeric uid/gid.
797 Description: Improved cleanup procedure for conf-sanity test_46a.
801 Description: Do not start bulk transfer, if client tried to reconnect.
802 Details : Do not start bulk transfer, if client tried to reconnect,
803 but failed due to active rpcs. Check for exp_abort_active_req
804 has been implemented.
808 Description: Prevent inconsistences between linux and lustre mount structures.
809 Details : Wait indefinitely in server_wait_finished() until mnt_count drops.
810 Make the sleep interruptible.
812 Severity : enhancement
814 Description: provide server to client comms path
815 Details : server to client communications path via new LDLM_SET_INFO rpc
819 Description: truncate starts GFP_FS allocation under transaction causing deadlock
820 Details : ldiskfs_truncate calls grab_cache_page which may start page
821 allocation under an open transaction. This may lead to
822 calling prune_icache with consequent lustre reentrance.
826 Description: send correctly lsm on open replay
827 Details : MDS is trust to LSM size on replay open, but client can set wrong size
832 Description: enable client interface failover
833 Details : When a child reconnects from another NID, properly update export
834 nid hash position and ldlm reverse import.
836 Severity : enhancement
838 Description: Communicate OST degraded/readonly state via statfs to MDS
839 Details : Flags in the statfs returned from OSTs indicate whether the
840 OST is in a degraded RAID state, or if the filesystem has
841 turned read-only after a filesystem error is detected.
846 Description: don't panic if EPROTO was hit when reading symlink
847 Details : correctly handling request reference in error cases.
849 Severity : enhancement
851 Description: Add LustreNetLink, a kernel-userspace communcation path.
853 Severity : enhancement
855 Description: Update kernel to SLES10 SP2 2.6.16.60-0.39.3.
860 Description: access to llog context before init.
861 Details : move handling CATALOGS file at osc layer and forbid access to llog
865 Frequency : rare, only when client is writing to OST on local node
867 Description: Avoid deadlock for local client writes
868 Details : Use new OBD_BRW_MEMALLOC flag to notify OST about writes in the
869 memory freeing context. This allows OST threads to set the
870 PF_MEMALLOC flag on task structures in order to allocate memory
871 from reserved pools and complete IO.
872 Use GFP_HIGHUSER for OST allocations for non-local client writes,
873 so that the OST threads generate memory pressure and allow
874 inactive pages to be reclaimed.
876 Severity : enhancement
878 Description: Update kernel to RHEL5.3 2.6.18-128.1.14.el5.
883 Description: lock ordering violation between &cli->cl_sem and _lprocfs_lock
884 Details : move ldlm namespace creation in setup phase to avoid grab
885 _lprocfs_lock with cli_sem held.
889 Description: Temporarily disable grant shrink.
890 Details : Disable the feature for debugging.
893 Frequency : only when formatting test filesystems
895 Description: Unable to run several mkfs.lustre on loop devices at the same time
896 Details : mkfs.lustre returns error 256 on the concurrent loop devices
897 formatting. The solution is to proper handle the error.
899 Severity : enhancement
901 Description: Update kernel to RHEL5.3 2.6.18-128.1.6.el5.
903 Severity : enhancement
905 Description: Update kernel to SLES10 SP2 2.6.16.60-0.37.
910 Description: resolve race between obd_disconnect and class_disconnect_exports
911 Details : if obd_disconnect will be called to already disconnected export he
912 forget release one reference and osc module can't unloaded.
914 Severity : enhancement
916 Description: Allow tuning service thread via /proc
917 Details : For each service a new
918 /proc/fs/lustre/{service}/*/thread_{min,max,started} entry is
919 created that can be used to set min/max thread counts, and get the
920 current number of running threads.
925 Descriptoin: don't return error if have partially created objects for file.
926 Details : lov_update_create_set uses set->set_success as index for created
927 objects, so if some requests will be failed, they will have hole
928 at end of array and qos_shrink_lsm for allocate correct lsm.
930 Severity : enhancement
932 Description: Update OFED support to 1.4.1
937 Description: Reduce small size read RPC
938 Details : Set read-ahead limit for every file and only do read-ahead when
939 available read-ahead pages are bigger than 1M to avoid small size
942 Severity : enhancement
944 Description: move AT tunable parameters for more consistent usage
945 Details : add AT tunables under /proc/sys/lustre, add to conf_param parsing
947 Severity : enhancement
949 Description: add lazystatfs mount option to allow statfs(2) to skip down OSTs
950 Details : allow skip disconnected ost for send statfs request and hide error
956 Description: fix racy locking of mballoc block bitmaps causing BUG
957 Details : The locking of the mballoc buddy bitmap and the in-memory
958 block bitmap was using two different spin locks in some
959 cases. This made it possible to incorrectly access the
960 mballoc bitmap while another process was modifying it,
961 causing a sanity assertion to fail. While no on-disk corruption
962 was reported, there was some risk of this happening.
965 Frequency : rare, on llog test 6
967 Descriptoin: don't allow connect to already connected import
968 Details : allowing connect to already connected import is hide connecting problem.
973 Descriptoin: allow kill process which wait statahead result
974 Details : for some reasons 'ls' can stick in waiting result from statahead,
975 in this case need way for kill this process.
977 Severity : enhancement
979 Description: Add state history info file, enhance import info file
980 Details : Track import connection state changes in a new osc/mdc proc file;
981 add overview-type data to the osc/mdc import proc file.
983 Severity : enhancement
985 Description: MDS create should not wait for statfs RPC while holding DLM lock.
987 Severity : enhancement
989 Description: Update to RHEL5U3 kernel-2.6.18-128.1.1.el5.
994 Descriptoin: OST grant too much space to client even there are not enough space.
995 Details : Client will shrink its grant cache to OST if there are no write
996 activity over 6 mins (GRANT_SHRINK_INTERVAL), and OST will retrieve
997 this grant cache if there are already not enough avaible space
998 (left_space < total_clients * 32M).
1001 Frequency : start MDS on uncleanly shutdowned MDS device
1003 Descriptoin: ll_sync thread stay in waiting mds<>ost recovery finished
1004 Details : stay in waiting mds<>ost recovery finished produce random bugs
1005 due race between two ll_sync thread for one lov target. send
1006 ACTIVATE event only if connect realy finished and import have
1010 Frequency : rare, connect and disconnect target at same time
1012 Descriptoin: ASSERTION(atomic_read(&imp->imp_inflight) == 0
1013 Details : don't call obd_disconnect under lov_lock. this long time
1014 operation and can block ptlrpcd which answer to connect request.
1019 Descriptoin: don't lose wakeup for imp_recovery_waitq
1020 Details : recover_import_no_retry or invalidate_import and import_close can
1021 both sleep on imp_recovery_waitq, but we was send only one wakeup
1025 Frequency : always with long access acl
1027 Descriptoin: mds can't pack reply with long acl.
1028 Details : mds don't control size of acl but they limited by reint/getattr
1031 Severity : enhancement
1033 Description: Update to SLES10 kernel-2.6.16.60-0.33.
1035 Severity : enhancement
1037 Description: Update to RHEL5 kernel-2.6.18-92.1.22.el5.
1040 Frequency : start MDS on uncleanly shutdowned MDS device
1042 Descriptoin: aborting recovery hang on MDS
1043 Details : don't throttle destroy RPCs for the MDT.
1046 Frequency : on remount
1048 Description: external journal device not working after the remount
1049 Details : clear dev_rdonly flag for external journal devices in
1055 Description: shutdown vs evict race
1056 Details : client_disconnect_export vs connect request race.
1057 if client will evicted at this time - we start invalidate
1058 thread without referece to import and import can be freed
1062 Frequency : rare, need acl's on inode.
1064 Description: client can't handle ost additional correctly
1065 Details : if ost was added after client connected to mds client can have
1066 hit lnet_try_match_md ... to big messages to wide striped files.
1067 in this case need teach client to handle config events about add
1068 lov target and update client max ea size at that event.
1070 Severity : enhancement
1072 Description: Changelogs
1073 Details : Changelogs are a lightweight mechanism to track filesystem
1074 metadata and namespace changes. The changelog is recorded
1075 permanently on the MDTs, and is periodically "consumed" / purged
1076 when records are no longer needed.
1078 Severity : enhancement
1080 Description: compact fld format with extents
1081 Details : Store range of seq rather than every seq in FLD. Seq
1082 controller update FLD rather than clients. In Case of CMD, mdt0
1083 has FLD, all other metadata server act as non persistent proxy
1084 for FLD queries and cache fld entries in fld cache.
1089 Description: don't skip ost target if they assigned to file
1090 Details : Drop slow OSCs if we can, but not for requested start idx.
1091 This means "if OSC is slow and it is not the requested
1092 start OST, then it can be skipped, otherwise skip it only
1093 if it is inactive/recovering/out-of-space.
1097 Description: more cleanup in mds_lov
1098 Details : not send LOV EA under replay, we can't know about they size at this
1099 time. Don't allow client connect to mds before any ost connected,
1100 for avoid problems with LOV EA size and returning EIO to client.
1102 Severity : enhancement
1104 Description: Interoperability at server side (Disk interoperability)
1106 Severity : enhancement
1108 Description: Update to RHEL5 kernel-2.6.18-92.1.17.el5.
1110 Severity : enhancement
1112 Description: Update to SLES10 SP2 kernel-2.6.16.60-0.31.
1114 Severity : enhancement
1116 Description: New client IO stack (CLIO).
1118 Severity : enhancement
1120 Description: Commit on sharing. Eliminate inter-client dependencies between
1121 uncommitted transactions by doing transaction commits.
1122 Thereby clients may recovery independently.
1125 Frequency : Create a symlink file with a very long name
1127 Description: ldlm_cancel_pack()) ASSERTION(max >= dlm->lock_count + count)
1128 Details : If there is no extra space in the request for early cancels,
1129 ldlm_req_handles_avail() returns 0 instead of a negative value.
1131 Severity : enhancement
1133 Description: Add /proc entry for import status
1134 Details : The mdc, osc, and mgc import directories now have
1135 an import directory that contains useful import data for debugging
1136 connection problems.
1138 Severity : enhancement
1140 Description: Re-disable certain /proc logging
1141 Details : Enable and disable client's offset_stats, extents_stats and
1142 extents_stats_per_process stats logging on the fly.
1145 Frequency : Only on FC kernels 2.6.22+
1147 Description: oops in statahead
1148 Details : Do not drop reference count for the dentry from VFS when lookup,
1149 VFS will do that by itself.
1151 Severity : enhancement
1153 Description: Generic /proc file permissions
1154 Details : Set /Proc file permissions in a more generic way to enable non-
1155 root users operate on some /proc files.
1159 Description: Hitting mdc_commit_close() ASSERTION
1160 Details : Properly handle request reference release in
1161 ll_release_openhandle().
1165 Description: quota recovery deadlock during mds failover
1166 Details : This patch includes att18982, att18236, att18237 in bz14840.
1168 1. fix osts hang when mds does failover with quotaon
1169 2. prevent watchdog storm when osts threads wait for the
1174 Frequency : only patchless client
1175 Description: add workaround for race between add/remove dentry from hash
1177 Severity : enhancement
1179 Description: Allow OST glimpses to return PW locks
1183 Description: LBUG when llog conf file is full
1184 Details : When llog bitmap is full, ENOSPC should be returned for plain log.
1188 Description: Prevent import from entering FULL state when server in recovery
1192 Description: service mount cannot take device name with ":"
1193 Details : Only when device name contains ":/" will mount treat it as
1199 Description: replace ptlrpcd with the statahead thread to interpret the async
1200 statahead RPC callback
1204 Frequency : on recovery
1205 Description: I/O failures after umount during fail back
1206 Details : if client reconnected to restarted server we need join to recovery
1207 instead of find server handler is changed and process self
1208 eviction with cancel all locks.
1210 Severity : enhancement
1212 Description: Update to RHEL5 kernel-2.6.18-92.1.10.el5.
1214 Severity : enhancement
1216 Description: Update to SLES10 SP2 kernel-2.6.16.60-0.27.
1218 Severity : enhancement
1220 Description: Upcall on Lustre log has been dumped
1221 Details : Allow for a user mode script to be called once a Lustre log has
1222 been dumped. It passes the filename of the dumped log to the
1223 script, the location of the script can be specified via
1224 /proc/sys/lnet/debug_log_upcall.
1229 Description: avoid idr_remove called for id which is not allocated.
1230 Details : Move assigment s_dev for clustered nfs to end of initialization,
1231 to avoid problem with error handling.
1236 Description: avoid Already found the key in hash [CONN_UNUSED_HASH] messages
1237 Details : When connection is reused this not moved from CONN_UNUSED_HASH
1238 into CONN_USED_HASH and this prodice warning when put connection
1239 again in unused hash.
1242 Severity : enhancement
1244 Description: File striping can now be set to use an arbitrary pool of OSTs.
1246 Severity : enhancement
1248 Description: Export bytes_read/bytes_write count on OSC/OST.
1252 Description: Early reply size mismatch, MGC loses connection
1253 Details : Apply the MGS_CONNECT_SUPPORTED mask at reconnect time so
1254 the connect flags are properly negotiated.
1259 Description: quotas are not honored with O_DIRECT
1260 Details : all writes with the flag O_DIRECT will use grants which leads to
1261 this problem. Now using OBD_BRW_SYNC to guard this.
1265 Description: add quota statistics
1266 Details : 1. sort out quota proc entries and proc code.
1267 2. add quota statistics
1269 Severity : enhancement
1271 Description: enable quota support for HEAD.
1275 Description: Properly propagate oinfo flags from lov to osc for statfs
1276 Details : restore missing copy oi_flags to lov requests.
1278 Severity : enhancement
1280 Description: Add man pages for llobdstat(8), llstat(8), plot-llstat(8),
1281 : l_getgroups(8), lst(8), routerstat(8)
1282 Details : included man pages for llobdstat(8), llstat(8),
1283 : plot-llstat(8), l_getgroups(8), lst(8), routerstat(8)
1285 Severity : enhancement
1287 Description: configure's --enable-quota should check the
1288 : kernel .config for CONFIG_QUOTA
1289 Details : configure is terminated if --enable-quota is passed but
1290 : no quota support is in kernel
1294 Description: Remove portals compatibility
1295 Details : Remove portals compatibility, not interoperable with releases
1300 Description: Resolve device initialization race
1301 Details : Prevent proc handler from accessing devices added to the
1302 obd_devs array but yet be intialized.
1304 Severity : enhancement
1306 Description: Update to SLES10 SP2 kernel-2.6.16.60-0.23.
1308 Severity : enhancement
1310 Description: Update to RHEL5 kernel-2.6.18-92.1.6.el5.
1315 Description: Using wrong pointer in osc_brw_prep_request
1316 Details : Access to array[-1] can produce panic if kernel compiled with
1317 CONFIG_PAGE_ALLOC enabled
1321 Description: Client runs out of low memory
1322 Details : Consider only lowmem when counting initial number of llap pages
1326 Description: *optional* service tags registration
1327 Details : if the "service tags" package is installed on a Lustre node
1328 When the filesystem is mounted, a local-node service tag will
1329 be created. See http://inventory.sun.com/ for more information
1330 about the Service Tags asset management system.
1334 Description: Kernel BUG tries to release flock
1335 Details : Lustre does not destroy flock lock before last reference goes
1336 away. So always drop flock locks when client is evicted and
1337 perform unlock regardless of successfulness of speaking to MDS.
1341 Description: add refcount protection for osc callbacks, avoid panic on shutdown
1345 Description: sanity test 65a fails if stripecount of -1 is set
1346 Details : handle -1 striping on filesystem in ll_dirstripe_verify
1351 Description: ASSERTION(CheckWriteback(page,cmd)) failed
1352 Details : badly clear PG_Writeback bit in ll_ap_completion can produce false
1355 Severity : enhancement
1357 Description: Update to RHEL5 kernel-2.6.18-53.1.21.el5.
1361 Description: do not process already freed flock
1362 Details : flock can possibly be freed by another thread before it reaches
1363 to ldlm_flock_completion_ast.
1367 Description: LBUG during stress test
1368 Details : Need properly lock accesses the flock deadlock detection list.
1372 Description: oops in page fault handler
1373 Details : kernel page fault handler can return two special 'pages' in error
1374 case, don't try dereference NOPAGE_SIGBUS and NOPAGE_OMM.
1378 Description: timeout with invalidate import.
1379 Details : ptlrpcd_check call obd_zombie_impexp_cull and wait request which
1380 should be handled by ptlrpcd. This produce long age waiting and
1381 -ETIMEOUT ptlrpc_invalidate_import and as result LASSERT.
1383 Severity : enhancement
1385 Description: Update to RHEL5 kernel-2.6.18-53.1.19.el5.
1389 Description: enable MGS and MDT services start separately
1390 Details : add a 'nomgs' option in mount.lustre to enable start a MDT with
1391 a co-located MGS without starting the MGS, which is a complement
1392 to 'nosvc' mount option.
1396 Frequency : after recovery
1397 Description: precreate to many object's after del orphan.
1398 Details : del orphan st in oscc last_id == next_id and this triger growing
1399 count of precreated objects. Set flag LOW to skip increase count
1400 of precreated objects.
1404 Frequency : rare, on clear nid stats
1405 Description: ASSERTION(client_stat->nid_exp_ref_count == 0)
1406 Details : when clean nid stats sometimes try destroy live entry,
1407 and this produce panic in free.
1411 Description: Stack overflow during MDS log replay
1412 ease stack pressure by using a thread dealing llog_process.
1416 Description: wait until IO finished before start new when do lock cancel.
1417 Details : VM protocol want old IO finished before start new, in this case
1418 need wait until PG_writeback is cleared until check dirty flag and
1419 call writepages in lock cancel callback.
1421 Severity : enhancement
1423 Description: using special macro for print time and cleanup in includes.
1427 Description: mds_mfd_close() ASSERTION(rc == 0)
1428 Details : In mds_mfd_close(), we need protect inode's writecount change
1429 within its orphan write semaphore to prevent possible races.
1433 Description: Obsolete CURRENT_SECONDS and use cfs_time_current_sec() instead.
1437 Frequency : rare, on shutdown ost
1438 Description: don't hit live lock with umount ost.
1439 Details : shrink_dcache_parent can be in long loop with destroy dentries,
1440 use shrink_dcache_sb instead.
1444 Description: don't panic with use echo client
1445 Details : echo client pass NULL as client nid pointer and this produce null
1446 pointer dereference.
1450 Description: fix build on ppc32
1451 Details : compile code with -m64 flag produce wrong object file for ppc32.
1455 Description: add message levels for liblustreapi
1459 Description: fix for occasional failure case of -ENOSPC in recovery-small tests
1460 Details : Move the 'good_osts' check before the 'total_bavail' check. This
1461 will result in an -EAGAIN and in the exit call path we call
1462 alloc_rr() which will with increasing aggressiveness attempt to
1463 aquire precreated objects on the minimum number of required OSCs.
1467 Description: Use old size assignment to avoid deadlock
1468 Details : Reverts the changes in bugs 2369 and bug 14138 that introduced
1469 the scheduling while holding a spinlock. We do not need locking
1470 for size in ll_update_inode() because size is only updated from
1471 the MDS for directories or files without objects, so there is no
1472 other place to do the update, and concurrent access to such inodes
1473 are protected by the inode lock.
1477 Description: resolve "_IOWR redefined" build error on SLES10
1481 Description: dump the memory debugging after all modules are unloaded to
1482 suppress false negative in conf_sanity test 39
1484 Severity : enhancement
1486 Description: build kernel-ib packages for OFED 1.3 in our release cycle
1491 Description: fix SLES kernel versioning
1492 Details : the kernel version for our SLES 10 kernel did not include a "-"
1493 before the "smp" at the end. while this was not a problem in
1494 general, it did mean that software trying to use the kernel
1495 version to try to detect a vendor specific kernel would fail.
1496 this was most evident by the OFED build scripts.
1500 Description: Don't update lov_desc members until making sure they are valid
1501 Details : When updating lov_desc members via proc fs, need fix their
1502 validities before doing the real update.
1506 Description: don't put request into delay list while invalidate in flight.
1507 Details : ptlrpc_delay_request sometimes put in delay list while invalidate
1508 import in flight. this produce timeout for invalidate and sometimes
1509 can cause stale data.
1513 Frequency : on ppc only
1514 Description: not convert ost objects for directory because it's not exist.
1515 Details : ll_dir_getstripe assume dirrectory has ost objects but this wrong.
1519 Description: Add FMODE_EXEC file flag for SLES10 SP1 kernel.
1521 Severity : enhancement
1523 Description: Update to support 2.6.22.14 vanilla kernel.
1527 Frequency : rare, on recovery
1528 Description: read procfs can produce deadlock in some situation
1529 Details : Holding lprocfs lock which send rpc can produce block for destroy
1530 obd objects and this also block reconnect with -EALREADY. This
1531 isn't fix all lprocfs bugs - but make it rare.
1533 Severity : enhancement
1535 Description: Update kernel to RHEL5 2.6.18-53.1.14.el5.
1538 Frequency : frequent on X2 node
1540 Description: mdc_set_open_replay_data LBUG
1541 Details : Set replay data for requests that are eligible for replay.
1545 Description: lustre_mgs: operation 101 on unconnected MGS
1546 Details : When MGC is disconnected from MGS long enough, MGS will evict the
1547 MGC, and late on MGC cannot successfully connect to MGS and a lot
1548 of the error messages complaining that MGS is not connected.
1552 Frequency : on network error
1553 Description: panic with double free request if network error
1554 Details : mdc_finish_enqueue is finish request if any network error ocuring,
1555 but it's true only for synchronus enqueue, for async enqueue
1556 (via ptlrpcd) this incorrect and ptlrpcd want finish request
1559 Severity : enhancement
1561 Description: client-side metadata stat-ahead during readdir(directory readahead)
1562 Details : perform client-side metadata stat-ahead when the client detects
1563 readdir and sequential stat of dir entries therein
1566 Frequency : on start mds
1568 Description: Implement get_info(last_id) in obdfilter.
1571 Frequency : occasional
1573 Description: Correctly check stale fid, not start epoch if ost not support SOM
1574 Details : open with flag O_CREATE need set old fid in op_fid3 because
1575 op_fid2 was overwritten with new generated fid, but mds can answer
1576 with one of these two fids and both is not stale. Setattr
1577 incorrectly started an epoch and assume will be called
1578 done_writing, but without SOM done_writing ever being called.
1581 Frequency : rare, depends on device drivers and load
1583 Description: MDS or OSS nodes crash due to stack overflow
1584 Details : Code changes in 1.8.0 increased the stack usage of some functions.
1585 In some cases, in conjunction with device drivers that use a lot
1586 of stack the MDS (or possibly OSS) service threads could overflow
1587 the stack. One change which was identified to consume additional
1588 stack has been reworked to avoid the extra stack usage.
1591 Frequency : occasional
1593 Description: Do not fail import if osc_interpret_create gets -EAGAIN
1594 Details : If osc_interpret_create got -EAGAIN it immediately exits and
1595 wakeup oscc_waitq. After wakeup oscc_wait_for_objects call
1596 oscc_has_objects and see OSC has no objests and call
1597 oscc_internal_create to resend create request.
1599 Severity : enhancement
1601 Description: Update to SLES10 SP1 latest kernel-2.6.16.54-0.2.5.
1603 Severity : enhancement
1605 Description: Update to RHEL5 latest kernel-2.6.18-53.1.13.el5.
1608 Frequency : very rare
1610 Description: Fix replay if there is an un-replied request and open
1611 Details : In some cases, older replay request will revert the
1612 mcd->mcd_last_xid on MDS which is used to record the client's
1613 latest sent request.
1615 Severity : enhancement
1617 Description: Update to RHEL5 latest kernel-2.6.18-53.1.6.el5.
1619 Severity : enhancement
1621 Description: Add rhel5 support to HEAD.
1623 Serverity : enhancement
1625 Description: Update RHEL4 kernel to 2.6.9-67.0.4.
1630 Description: Don't allow skipping OSTs if index has been specified.
1631 Details : Don't allow skipping OSTs if index has been specified, make
1632 locking in internal create lots better.
1636 Description: LBUG in ptlrpc_check_set() bad phase ebc0de00
1637 Details : access to bitfield in structure is always rounded to long
1638 and this produce problem with not atomic change any bit.
1642 Description: Lustre make rpms failed.
1643 Details : Remove ldiskfs spec file to avoids rpmbuild be confused when
1644 builds Lustre rpms from tarball.
1647 Frequency : rare on shutdown ost
1649 Description: If llog cancel was not send before clean_exports phase, this can
1650 produce deadlock in llog code.
1651 Details : If llog thread has last reference to obd and call class_import_put
1652 this produce deadlock because llog_cleanup_commit_master wait when
1653 last llog_commit_thread exited, but this never success because was
1654 called from llog_commit_thread.
1658 Description: allow userland application know is lost one of stripes.
1659 Details : fill lvb_blocks with error code on ost and return it to
1660 application if error flag found.
1664 Description: NULL lov_tgts causing MDS oops
1665 Details : more safe checks for NULL lov_tgts for avoid oops.
1667 Severity : enhancement
1669 Description: Update to RHEL4 latest kernel-2.6.9-67.0.1.EL.
1673 Descriptoin: make lov_create() will not stuck in obd_statfs_rqset()
1674 Details : If an OST is down the MDS will hang indefinitely in
1675 obd_statfs_rqset() waiting for the statfs data. While for
1676 MDS QOS usage of statfs, it should not stuck in waiting.
1678 Severity : enhancement
1680 Description: remote_acl support
1681 Details : Support ACL-based permission check for remote user.
1682 Support setfacl/getfacl for remote user with the utils
1683 "lfs {l,r}{s,g}etfacl" which follow the same parameter format as
1684 the system "{s,g}etfacl" utils.
1686 Severity : enhancement
1688 Description: Update to RHEL4 U6 kernel-2.6.9-67.EL.
1690 Severity : enhancement
1692 Description: Update to RHEL5 latest kernel-2.6.18-53.1.4.el5.
1696 Description: make mgs_setparam() handle fsname containing dash
1697 Details : fsname containing a dash does not work with lctl conf_param
1699 Severity : enhancement
1701 Description: Update to SLES10 SP1 latest kernel-2.6.16.54-0.2.3.
1703 Severity : enhancement
1705 Description: Update to RHEL5 Update-1 kernel 2.6.18-53.el5.
1709 Frequency : rare, at shutdown
1710 Description: access already free / zero obd_namespace.
1711 Details : if client_disconnect_export was called without force flag set,
1712 and exist connect request in flight, this can produce access to
1713 NULL pointer (or already free pointer) when connect_interpret
1714 store ocd flags in obd_namespace.
1718 Frequency : only at startup
1719 Description: not alloc memory with spinlock held.
1720 Details : allocation memory with GFP_KERNEL can produce sleep deadlock,
1721 if any spinlock held.
1723 Severity : enhancement
1725 Description: make lustre randomly fail allocating memory
1726 Details : Make lustre randomly failed allocating memory for testing purpose.
1728 Severity : enhancement
1730 Description: lost problems with lov objid file
1731 Details : Fixes some scability and access to not inited memory problems
1732 in work with lov objdid file.
1737 Description: lfs find does not continue on file error
1738 Details : Continue other files processing when a file/dir is absent.
1742 Description: Inconsistent usage of lustre_pack_reply()
1743 Details : Standardize the usage of lustre_pack_reply() such that it
1744 always generate a CERROR on failure.
1747 Frequency : occasional
1749 Description: MDS hang or stay in waiting lock
1750 Details : If client receive lock with CBPENDING flag ldlm need send lock
1751 cancel as separate rpc, to avoid situation when cancel request
1752 can't processed due all i/o threads stay in wait lock.
1756 Description: Update to RHEL5 kernel 2.6.18-8.1.15.el5.
1760 Description: Update to SLES10 SP1 kernel 2.6.16.53-0.16
1764 Description: Update to SLES9 kernel-2.6.5-7.287.3.
1768 Description: Update to RHEL4 latest kernel.
1770 Severity : enhancement
1772 Description: Build SLES10 patchless client fails
1773 Details : The configure was broken by run ./configure with
1774 --with-linux-obj=.... argument for patchless client. When the
1775 configure use --with-linux-obj, the LINUXINCLUDE= -Iinclude
1776 can't search header adequately. Use absolute path such as
1777 -I($LINUX)/include instead.
1781 Description: interrupt oig_wait produce painc on resend.
1782 Details : brw_redo_request can be used for resend requests from ptlrpcd and
1783 private set, and this produce situation when rq_ptlrpcd_data not
1784 copyed to new allocated request and triggered LBUG on assert
1785 req->rq_ptlrpcd_data != NULL. But this member used only for wakeup
1786 ptlrpcd set if request is changed and can be safety changed to use
1791 Description: LASSERT_{REQ,REP}SWAB macros are buggy
1792 Details : If SWAB_PARANOIA is disabled, the LASSERT_REQSWAB and
1793 LASSERT_REPSWAB macros become no-ops, which is incorrect. Drop
1794 these macros and replace them with their difinitions instead.
1798 Description: Update kernel patches for SLES10 2.6.16.53-0.8.
1799 Details : Update which_patch & target file for SLES10 latest kernel.
1802 Description: Remove client patches from SLES 10 kernel.
1803 Details : This causes SLES 10 clients to behave as patchless clients
1804 even on a Lustre-patched (server) kernel.
1806 Severity : enhancement
1808 Description: self-adjustable client's lru lists
1809 Details : use adaptive algorithm for managing client cached locks lru
1810 lists according to current server load, other client's work
1811 pattern, memory activities, etc. Both, server and client
1812 side namespaces provide number of proc tunables for controlling
1815 Severity : enhancement
1817 Description: light-weight GSS support
1818 Details : Support krb5n and krb5a mode, which keep Kerberos 5 authentication
1819 and reduce performance overhead.
1821 Severity : enhancement
1823 Description: Linux keyring support
1824 Details : Support using service of Linux keyring for Lustre GSS internal
1825 context refresh/cache mechanism.
1829 Description: Fix errors in lfs documentation
1830 Details : Fixes man pages
1834 Description: don't use GFP_* in generic Lustre code.
1835 Details : Use cfs_alloc_* functions and CFS_* flags for code portability.
1839 Description: obdclass is limited by single OBD_ALLOC(idarray)
1840 Details : replace OBD_ALLOC/OBD_FREE with OBD_VMALLOC/OBD_VFREE
1844 Description: warnings with build patchless client with vanila 2.6.19 and up
1845 Details : change old ctl_table style and replace ctl_table/ctl_table_header
1846 with cfs_sysctl_table_t/cfs_sysctl_table_header_t
1850 Frequency : Only for SLES
1851 Description: sanity_quota fail test_1
1852 Details : There are multiple occurences of $TSTUSR in SLES's /etc/group
1853 file, which makes TSTID[2] inunique.
1857 Frequency : Only for SLES9
1858 Description: Kernel patches for SLES9 2.6.5-7.286 kernel
1859 Details : Update target/ChangeLog/which_patch .
1863 Frequency : Only for test_10 in sanity-quota.sh
1864 Description: a bug in quota test code
1865 Details : checking a test flag in a wrong place
1869 Frequency : Only for test_13 in sanity-quota.sh
1870 Description: a bug in quota test script
1871 Details : shouldn't check overfull allocated quota
1875 Description: jbd statistics
1876 Details : Port older jbd statistics patch for sles10
1878 Severity : enhancement
1880 Description: Add printing inode info into message about error in writepage.
1885 Description: reply_lock_interpret crash due to race with it and lock cancel.
1886 Details : Do not replay locks that are being cancelled. Do not reference
1887 locks by their address during replay, just by their handle.
1891 Frequency : When flocks are used.
1892 Description: assertion failure in ldlm_cli_enquque_fini for non NULL lock.
1893 Details : Flock locks might destroy just granted lock if it could be merged
1894 with another existing flock, this is done in completion handler,
1895 so teach ldlm_cli_enquque_fini that this is a valid case for
1901 Description: Oops in read and write path when failing to allocate lock.
1902 Details : Check if lock allocation failed and return error back.
1906 Description: lstripe command fails for valid OST index
1907 Details : The stripe offset is compared to lov->desc.ld_tgt_count
1908 instead of lov->desc.ld_active_tgt_count.
1912 Description: sanity.sh failed test 103
1913 Details : RHEL mis-interpret setfacl "-X" param, so we won't test setfacl
1918 Description: df doesn't work properly if diskfs blocksize != 4K
1919 Details : Choose biggest blocksize of OST's as the LOV's blocksize.
1923 Description: lfs find on -1 stripe looping in lsm_lmm_verify_common()
1924 Details : Avoid lov_verify_lmm_common() on directory with -1 stripe count.
1926 Severity : enhancement
1928 Description: Adaptive timeouts
1929 Details : RPC timeouts adapt to changing server load and network
1930 conditions to reduce resend attempts and improve recovery time.
1934 Description: llapi_file_create() does not allow some changes
1935 Details : add llapi_file_open() that allows specifying the mode and
1936 open flags, and also returns an open file handle.
1940 Description: merge and cleanup kernel patches.
1941 Details : Remove mnt_lustre_list in vfs_intent-2.6-rhel4.patch.
1945 Description: Add journal checksum support.(Kernel part)
1946 Details : The journal checksum feature adds two new flags i.e
1947 JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT and
1948 JBD2_FEATURE_COMPAT_CHECKSUM. JBD2_FEATURE_CHECKSUM flag
1949 indicates that the commit block contains the checksum for
1950 the blocks described by the descriptor blocks. Now commit
1951 record can be sent to disk without waiting for descriptor
1952 blocks to be written to disk. This behavior is controlled
1953 using JBD2_FEATURE_ASYNC_COMMIT flag.
1957 Description: OSS needs mutliple precreate threads
1958 Details : Add ability to start more than one create thread per OSS.
1962 Description: Sanity.sh test_65k failed: setstripe should have succeeded
1963 Details : execute lfs setstripe on client
1967 Description: mds_obd_create error creating tmp object
1968 Details : When the user sets quota on root, llog will be affected and can't
1969 create files and write files.
1972 Frequency : Always on ia64 patchless client, and possibly others.
1974 Description: Add EXPORT_SYMBOL check for node_to_cpumask symbol.
1975 Details : This allows the patchless client to be loaded on architectures
1976 without this export.
1980 Description: RedHat Update kernel for RHEL5
1981 Details : Add the kernel config file for RHEL5.
1985 Description: RedHat Update kernel for RHEL5
1986 Details : Modify the kernel config file more closer RHEL5.
1990 Description: Build failure against Centos5 (RHEL5)
1991 Details : Define PAGE_SIZE when it isn't present.
1995 Description: test_7 on sanity-quota.sh can't be used on separate machines
1996 Details : add facet to handle it
2000 Description: "ll_intent_file_open()) lock enqueue: err: -13" with nfs
2001 Details : with NFS, the anon dentry's parent was set to itself in
2002 d_alloc_anon(), so in MDS, we use rec->ur_fid1 to find the
2003 corresponding dentry other than use rec->ur_name.
2005 Severity : enhancement
2007 Description: lfs setstripe enhancement
2008 Details : Make lfs setstripe understand 'k', 'm' and 'g' for stripe size.
2012 Description: enable data checksumming by default
2013 Details : enable checksum by default, allow --disable-checksum
2014 configure option and "-o nochecksum" mount option. Checksums
2015 can also be disabled at runtime via $LPROC/osc/*/checksum_pages.
2019 Description: lustre support for RHEL5
2020 Details : Add support for RHEL5.
2024 Description: Client eviction due to failover config
2025 Details : after a connection loss, the lustre client should attempt to
2026 reconnect to the last active server first before trying the
2027 other potential connections.
2031 Description: when mds and osts use different quota unit(32bit and 64bit),
2032 quota will be released repeatly.
2033 Details : void sending multiple quota reqs to mds, which will keep the
2034 status between the reqs.
2038 Description: osts not allocated evenly to files
2039 Details : change the condition to increase offset_idx
2042 Frequency : Always for filesystems larger than 2TB on 32-bit systems.
2044 Description: Data corruption for OSTs that are formatted larger than 2TB
2046 Details : When generating the bio request for lustre file writes the
2047 sector number would overflow a temporary variable before being
2048 used for the IO. The data reads correctly from Lustre (which
2049 will overflow in a similar manner) but other file data or
2050 filesystem metadata may be corrupted in some cases.
2054 Description: Tune the kernel for good SCSI performance.
2055 Details : Set the value of /sys/block/{dev}/queue/max_sectors_kb
2056 to the value of /sys/block/{dev}/queue/max_hw_sectors_kb
2061 Description: rewrite ext2-derived code in obdclass/uuid.c
2062 Details : rewrite inherited code (uuid parsing code from ext2 utils)
2063 from scratch preserving functionality.
2067 Description: "lfs find -obd UUID" prints directories
2068 Details : "lfs find -obd UUID" will return all directory names instead
2069 of just file names. It is incorrect because the directories
2070 do not reside on the OSTs.
2074 Description: use i_size_read and i_size_write in 2.6 port
2075 Details : replace inode->i_size access with i_size_read/write()
2078 Frequency : when removing large files
2080 Description: scheduling issue during removal of large Lustre files
2081 Details : Don't take the BKL in fsfilt_ext3_setattr() for 2.6 kernels.
2082 It causes scheduling issues when removing large files (17TB in the
2086 Frequency : only with liblustre clients on XT3
2088 Description: evictions taking too long
2089 Details : allow llrd to evict clients directly on OSTs
2092 Frequency : only on ppc
2094 Description: /proc/fs/lustre/devices broken on ppc
2095 Details : The patch as applied to 1.6.2 doesn't look correct for all arches.
2096 We should make sure the type of 'index' is loff_t and then cast
2097 explicitly as needed below. Do not assign an explicitly cast
2101 Frequency : only for rhel5
2103 Description: Kernel patches update for RHEL5 2.6.18-8.1.10.el5.
2104 Details : Modify the target file & which_kernel.
2106 Severity : enhancement
2108 Description: omit set fsid for export NFS
2109 Details : fix set/restore device id for avoid EMFILE error and mark lustre
2110 fs as FS_REQUIRES_DEV for avoid problems with generate fsid.
2114 Frequency : Always, for kernels after 2.6.16
2115 Description: Fix warning idr_remove called for id=.. which is not allocated.
2116 Details : Last kernels save old s_dev before kill super and not allow
2117 to restore from callback - restore it before call kill_anon_super.
2121 Description: buffer overruns could theoretically occur
2122 Details : llapi_semantic_traverse() modifies the "path" argument by
2123 appending values to the end of the origin string, and a
2124 overrun may occur. Adding buffer overrun check in liblustreapi.
2128 Description: Fix error on 'ls .' at the top of the Lustre mount.
2129 Details : Don't revalidate dentry if it is a root dentry.
2133 Description: Kernel patches update for RHEL4 2.6.9-55.0.6.
2134 Details : Modify vm-tunables-rhel4.patch.
2138 Description: Kernel config for 2.6.18-vanilla.
2139 Details : Modify targets/2.6-vanilla.target.in.
2140 Add config file kernel-2.6.18-2.6-vanilla-i686.config.
2141 Add config file kernel-2.6.18-2.6-vanilla-i686-smp.config.
2142 Add config file kernel-2.6.18-2.6-vanilla-x86_64.config.
2143 Add config file kernel-2.6.18-2.6-vanilla-x86_64-smp.config.
2147 Description: improve handling recoverable errors
2148 Details : if request processig with error which can be recoverable on server
2149 request should be resend, otherwise page released from cache and
2154 Description: Kernel patches update for RHEL5 2.6.18-8.1.14.el5.
2155 Details : Modify target file & which_patch.
2156 A flaw was found in the IA32 system call emulation provided
2157 on AMD64 and Intel 64 platforms. An improperly validated 64-bit
2158 value could be stored in the %RAX register, which could trigger an
2159 out-of-bounds system call table access. An untrusted local user
2160 could exploit this flaw to run code in the kernel
2161 (ie a root privilege escalation). (CVE-2007-4573).
2165 Description: O_DIRECT bypasses client statistics.
2166 Details : When running with O_DIRECT I/O, neither the client rpc_stats nor
2167 read_ahead_stats were updated. copied the stats section from
2168 osc_send_oap_rpc() into async_internal().
2172 Description: Add jbd statistics patch for RHEL5 and 2.6.18-vanilla
2176 Description: change order of libsysio includes
2177 Details : '#include sysio.h' should always come before '#include xtio.h'
2181 Description: handle "serious error: objid * already exists" more gracefully
2182 Details : If LAST_ID value on disk is smaller than the objects existing in
2183 the O/0/d* directories, it indicates disk corruption and causes an
2184 LBUG(). If the object is 0-length, then we should use the existing
2185 object. This will help to avoid a full fsck in most cases.
2187 Severity : enhancement
2189 Description: adapt the lustre_config script to support the upgrade case
2190 Details : Add "-u" option for lustre_config script to support upgrading 1.4
2191 server targets to 1.6 in parallel.
2195 Description: To avoid grant space > avaible space when the disk is almost
2196 full. Without this patch you might see the error "grant XXXX >
2197 available" or some LBUG about grant, when the disk is almost
2199 Details : In filter_check_grant, for non_grant cache write, we should
2200 check the left space by if (*left > ungranted + bytes), instead
2201 of (*left > ungranted), because only we are sure the left space
2202 is enough for another "bytes", then the ungrant space should be
2203 increase. In client, we should update cl_avail_grant only there
2204 is OBD_MD_FLGRANT in the reply.
2208 Description: Update RHEL 4 kernel to fix local root privilege escalation.
2209 Details : Update to the latest RHEL 4 kernel to fix the vulnerability
2210 described in CVE-2007-4573. This problem could allow untrusted
2211 local users to gain root access.
2214 Frequency : when using O_DIRECT and quotas
2216 Description: Incorrect file ownership on O_DIRECT output files
2217 Details : block usage reported by 'lfs quota' does not take into account
2218 files that have been written with O_DIRECT.
2223 Description: touch file failed when fs is not full
2224 Details : OST in recovery should not be discarded by MDS in alloc_qos(),
2225 otherwise we can get ENOSP while fs is not full.
2229 Description: parallel lock callbacks
2230 Details : Instead of sending blocking and completion callbacks as separated
2231 requests, adding them to a set and sending in parallel.
2234 Frequency : only for Cray XT3
2235 Bugzilla : 12829/13455
2236 Description: Changing primary group doesn't change the group lustre assigns to
2238 Details : When CRAY_XT3 is defined, the fsgid supplied by the client is
2239 overridden with the primary group provided by the group upcall,
2240 whereas the supplied fsgid can be trusted if it is in the list of
2241 supplementary groups returned by the group upcall.
2243 Severity : enhancement
2245 Description: Allow masking D_WARNING, D_ERROR messages from console
2246 Details : Console messages can now be disabled via lnet.printk.
2250 Description: User code with malformed file open parameter crashes client node
2251 Details : Before packing join_file req, all the related reference should be
2252 checked carefully in case some malformed flags cause fake
2253 join_file req on client.
2257 Description: LDLM_ENQUEUE races with LDLM_CP_CALLBACK
2258 Details : ldlm_completion_ast() assumes that a lock is granted when the req
2259 mode is equal to the granted mode. However, it should also check
2260 that LDLM_FL_CP_REQD is not set.
2264 Description: Heavy nfs access might result in deadlocks
2265 Details : After ELC code landed, it is now improper to enqueue any mds
2266 locks under och_sem, because enqueue might want to decide to
2267 cancel open locks for same inode we are holding och_sem for.
2271 Description: Client eviction while running blogbench
2272 Details : A lot of unlink operations with concurrent I/O can lead to a
2273 deadlock causing evictions. To address the problem, the number of
2274 oustanding OST_DESTROY requests is now throttled to
2275 max_rpcs_in_flight per OSC and LDLM_FL_DISCARD_DATA blocking
2276 callbacks are processed in priority.
2280 Description: enable ACLs on MDS by default
2281 Details : ACLs must be enabled on MDS by default.
2284 Frequency : PPC/PPC64 only
2286 Description: conflicts between asm-ppc64/types.h and lustre_types.h
2287 Details : fix duplicated definitions between asm-ppc64/types.h and
2288 lustre_types.h on PPC.
2291 Frequency : PPC/PPC64 only
2293 Description: asm-ppc/segment.h does not exist
2294 Details : fix compile issue on PPC.
2298 Description: better handle error messages in extents code
2301 Frequency : RHEL4 only
2303 Description: mkfs is very slow on IA64/RHEL4
2304 Details : A performance regression has been discovered in the MPT Fusion
2305 driver between versions 3.02.73rh and 3.02.99.00rh. As a
2306 consequence, we have downgraded the MPT Fusion driver in the RHEL4
2307 kernel from 3.02.99.00 to 3.02.73 until this problem is fixed.
2309 Severity : enhancement
2311 Description: SNMP support enhancement
2312 Details : Adding total number of sampled request for an MDS node in snmp
2315 Severity : enhancement
2317 Description: Optimize ldlm waiting list processing for PR extent locks
2318 Details : When processing waiting list for read extent lock and meeting read
2319 lock that is same or wider to it that is not contended, skip
2320 processing rest of the list and immediatelly return current
2321 status of conflictness, since we are guaranteed there are no
2322 conflicting locks in the rest of the list.
2326 Description: Time out and refuse to reconnect
2327 Details : When the failover node is the primary node, it is possible
2328 to have two identical connections in imp_conn_list. We must
2329 compare not conn's pointers but NIDs, otherwise we can defeat
2330 connection throttling.
2334 Description: port llog fixes from b1_6 into HEAD
2335 Details : Port llog reference couting and some llog cleanups from b1_6
2336 (bug 10800) into HEAD, for protect from panic and access to already
2337 free llog structures.
2341 Description: Detect stride IO mode in read-ahead
2342 Details : When a client does stride read, read-ahead should detect that and
2343 read-ahead pages according to the detected stride pattern.
2347 Description: data checksumming impacts single node performance
2348 Details : add support for several checksum algorithm. Currently, only CRC32
2349 and Adler-32 are supported. The checksum type can be changed on
2350 the fly via /proc/fs/lustre/osc/*/checksum_type.
2354 Description: use adler32 for page checksums
2355 Details : when available, use the Adler-32 algorithm instead of CRC32 for
2360 Description: build for x2 fails
2361 Details : fix compile issue on Cray systems.
2365 Description: Properly match for duplicate locks
2366 Details : Due to different lock order from skiplists code, we need to
2367 traverse entire list for now
2370 Frequency : only on PPC/SLES10
2372 Description: "BITS_PER_LONG is not 32 or 64" in linux/idr.h
2373 Details : On SLES10/PPC, fs.h includes idr.h which requires BITS_PER_LONG to
2374 be defined. Add a hack in mkfs_lustre.c to work around this compile
2379 Description: LASSERT on MDS when client holding flock lock dies
2380 Details : ldlm pool logic depends on number of granted locks equal to
2381 number of released locks which is not true for flock locks, so
2382 just exclude such locks from consideration.
2386 Description: MDS deadlock with many ll_sync_lov threads and I/O stalled
2387 Details : Use fsfilt_sync() for both the whole filesystem sync and
2388 individual file sync to eliminate dangerous inode locking
2389 with I_LOCK that can lead to a deadlock.
2393 Description: performance in 1.6.3
2394 Details : Force q->max_phys_segments to MAX_PHYS_SEGMENTS on SLES10 to be
2395 sure that 1MB requests are not fragmented by the block layer.
2397 Severity : enhancement
2399 Description: organize the server-side client stats on per-nid basis
2400 Details : Change the structure of stats under obdfilter and mds to
2408 The "uuid"s file would list the uuids of _active_ exports.
2409 And the clear entry is to clear all stats and stale nids.
2411 Severity : enhancement
2413 Description: eliminate client locks in face of contention
2414 Details : file contention detection and lockless i/o implementation
2415 for contended files.
2419 Description: Reinitialize optind to 0 so that interactive lfs works in all cases
2422 Frequency : very rare, if additional xattrs are used on kernels >= 2.6.12
2424 Description: MDS may lose file striping (and hence file data) in some cases
2425 Details : If there are additional extended attributes stored on the MDS,
2426 in particular ACLs, SELinux, or user attributes (if user_xattr
2427 is specified for the client mount options) then there is a risk
2428 of attribute loss. Additionally, the Lustre file striping
2429 needs to be larger than default (e.g. striped over all OSTs),
2430 and an additional attribute must be stored initially in the
2431 inode and then increase in size enough to be moved to the
2432 external attribute block (e.g. ACL growing in size) for file
2437 Description: skiplist implementation simplification
2438 Details : skiplists are used to group compatible locks on granted list
2439 that was implemented as tracking first and last lock of each
2440 lock group the patch changes that to using doubly linked lists
2444 Description: MDS LBUG: ASSERTION(!IS_ERR(dchild))
2445 Details : Change LASSERTs to client eviction (i.e. abort client's recovery)
2446 because LASSERT on both the data supplied by a client, and the
2447 data on disk is dangerous and incorrect.
2449 Severity : enhancement
2451 Description: Slow truncate/writes to huge files at high offsets.
2452 Details : Directly associate cached pages to lock that protect those pages,
2453 this allows us to quickly find what pages to write and remove
2454 once lock callback is received.
2458 Description: more ldlm soft lockups
2459 Details : In ldlm_resource_add_lock(), call to ldlm_resource_dump()
2460 starve other threads from the resource lock for a long time in
2461 case of long waiting queue, so change the debug level from
2462 D_OTHER to the less frequently used D_INFO.
2464 Severity : enhancement
2466 Description: add -gid, -group, -uid, -user options to lfs find
2470 Description: Hung threads in invalidate_inode_pages2_range
2471 Details : The direct IO path doesn't call check_rpcs to submit a new RPC
2472 once one is completed. As a result, some RPCs are stuck in the
2473 queue and are never sent.
2477 Description: filter threads hungs on waiting journal commit
2478 Details : Cleanup filter group llog code, then only filter group llog will
2479 be only created in the MDS/OST syncing process.
2483 Description: Procfs and llog threads access destoryed import sometimes.
2484 Details : Sync the import destoryed process with procfs and llog threads by
2485 the import refcount and semaphore.
2487 Severity : enhancement
2489 Description: openlock cache of b1_6 port to HEAD
2494 Description: kernel BUG at ldiskfs2_ext_new_extent_cb
2495 Details : If insertion of an extent fails, then discard the inode
2496 preallocation and free data blocks else it can lead to duplicate
2501 Description: don't always update ctime in ext3_xattr_set_handle()
2502 Details : Current xattr code updates inode ctime in ext3_xattr_set_handle.
2503 In some cases the ctime should not be updated, for example for
2504 2.0->1.8 compatibility it is necessary to delete an xattr and it
2505 should not update the ctime.
2509 Bugzilla : 15713/16362
2510 Description: Assertion in iopen_connect_dentry in 1.6.3
2511 Details : looking up an inode via iopen with the wrong generation number can
2512 populate the dcache with a disconneced dentry while the inode
2513 number is in the process of being reallocated. This causes an
2514 assertion failure in iopen since the inode's dentry list contains
2515 both a connected and disconnected dentry.
2519 Description: assertion failure in ldlm_handle2lock()
2520 Details : fix a race between class_handle_unhash() and class_handle2object()
2521 introduced in lustre 1.6.5 by bug 13622.
2526 Description: Kernel BUG: sd_iostats_bump: unexpected disk index
2527 Details : remove the limit of 256 scsi disks in the sd_iostat patch
2532 Description: oops in sd_iostats_seq_show()
2533 Details : unloading/reloading the scsi low level driver triggers a kernel
2534 bug when trying to access the sd iostat file.
2539 Description: Kernel panics during QLogic driver reload
2540 Details : REQ_BLOCK_PC requests are not handled properly in the sd iostat
2541 patch, causing memory corruption.
2546 Description: journal_dev option does not work in b1_6
2547 Details : pass mount option during pre-mount.
2549 Severity : enhancement
2551 Description: Add a FIEMAP(FIle Extent MAP) ioctl
2552 Details : FIEMAP ioctl will allow an application to efficiently fetch the
2553 extent information of a file. It can be used to map logical blocks
2554 in a file to physical blocks in the block device.
2558 Description: LDLM soft lockups - improvement
2559 Details : It is be possible to send the lock handle along with each read
2560 or write request because the client is already doing a lock match
2561 itself so there isn't any reason the OST should have to re-do that
2565 Frequency : only on Cray X2
2567 Description: X2 build failures
2568 Details : fix build failures on Cray X2.
2572 Description: xid & resent requests
2573 Details : Initialize RPC XID from clock at startup (randomly if clock is
2576 Severity : enhancement
2578 Description: Add lustre_start utility to start or stop multiple Lustre servers
2583 Description: Lustre GPF in {:ptlrpc:ptlrpc_server_free_request+373}
2584 Details : In case of memory pressure, list_del() can be called twice on
2585 req->rq_history_list, causing a kernel oops.
2589 Description: kptllnd_peer_check_sends()) ASSERTION(!in_interrupt()) failed
2590 Details : fix stack overflow in the distributed lock manager by defering
2591 export eviction after a failed AST to the elt thread instead of
2592 handling it in the dlm interpret routine.
2596 Description: Convert some comments to new format.
2597 Details : Update documenting comments to match doxygen conventions.
2601 Description: Grammar fixes.
2602 Details : A couple of trivial spelling fixes.
2606 Description: OSD_COUNTERS-mandatory
2607 Details : Make previously optional ->oti_{w,r}_locks sanity checks mandatory
2608 to simplify the code.
2612 Description: simplify cmm_device freeing logic.
2613 Details : Call cmm_device_free() in the failure path of cmm_device_alloc().
2617 Description: Add lockdep support to dt_object_operations locking interface.
2618 Details : Augment ->do_{read,write}_lock() prototypes with a `role'
2619 parameter indicating lock ordering. Update mdd code to use new
2624 Description: Introduce failloc constants for lockless IO tests.
2625 Details : Add two new failloc constants to test lockless IO. Only one of
2626 them in implemented---another is checked in yet to be landed
2631 Description: Add lockdep support for inode mutex.
2632 Details : Introduce and use new LOCK_INODE_MUTEX_PARENT() macro to be used
2633 in the situations where Lustre has to lock more than one inode
2638 Description: Add optional invariants checking support.
2639 Details : Add new LINVRNT() macro, optional on new --enable-invariants
2640 configure switch. This macro is to be used for consistency and
2641 sanity checks that are too expensive to be left in `production'
2646 Description: Zap lock->l_granted_mode with explicit LCK_MINMODE.
2647 Details : Use LCK_MINMODE rather than 0 to reset lock->l_granted_mode to
2652 Description: Add lockdep support for ldlm_lock and ldlm_resource.
2653 Details : Use spin_lock_nested() in (the only) situation where more than
2654 one ldlm_lock is locked simultaneously. Also, fix possible
2655 dead-lock in ldlm_lock_change_resource() by enforcing particular
2660 Description: Use struct ldlm_callback_suite in ldlm_lock_create().
2661 Details : Instead of specifying each ldlm_lock call-back through separate
2662 parameter, wrap them into struct ldlm_callback_suite.
2666 Description: Kill join_lru obd method and its callers.
2667 Details : CLIO uses lock weighting policy to keep locks over mmapped regions
2668 in memory---a requirement implemented through ->o_join_lru() obd
2669 method in HEAD. Remove this method and its users.
2673 Description: Add asynchronous ldlm ENQUEUE completion handler.
2674 Details : CLIO posts ENQUEUE requests asynchronously through ptlrpcd---a
2675 case that stock ldlm_completion_ast() cannot handle as it waits
2676 until lock is granted. Introduce new ldlm_completion_ast_async()
2677 for this. Also comment ldlm_completion_ast().
2681 Description: ldlm_error <-> errno conversion.
2682 Details : Add functions to map (rather arbitrary) between LDLM error codes
2683 and standard errno values. CLIO needs this to prevent LDLM specific
2684 constants from escaping ldlm and osc.
2688 Description: Kill unused ldlm_handle2lock_ns() function.
2689 Details : Kill unused ldlm_handle2lock_ns() function.
2693 Description: Add lu_ref support to ldlm_lock
2694 Details : lu_ref support for ldlm_lock and ldlm_resource. See lu_ref patch.
2695 lu_ref fields ->l_reference and ->lr_reference are added to ldlm_lock
2696 and ldlm_resource. LDLM interface has to be changed, because code that
2697 releases a reference on a lock, has to "know" what reference this is.
2698 In the most frequent case
2700 lock = ldlm_handle2lock(handle);
2702 LDLM_LOCK_PUT(lock);
2704 no changes are required. When any other reference (received _not_ from
2705 ldlm_handle2lock()) is released, LDLM_LOCK_RELEASE() has to be called
2706 instead of LDLM_LOCK_PUT().
2708 Arguably, changes are pervasive, and interface requires some discipline
2709 for proper use. On the other hand, it was very instrumental in finding
2710 a few leaked lock references.
2714 Description: Add ldlm_lock_addref_try().
2715 Details : Introduce ldlm_lock_addref_try() function (used by CLIO) that
2716 attempts to addref a lock that might be being canceled
2721 Description: Add ldlm_weigh_callback().
2722 Details : Add new ->l_weigh_ast() call-back to ldlm_lock. It is called
2723 by ldlm_cancel_shrink_policy() to estimate lock "value", instead of
2724 hard-coded `number of pages' logic.
2728 Description: Add lockdep annotations to llog code.
2729 Details : Use appropriately tagged _nested() locking calls in the places
2730 where llog takes more than one ->lgh_lock lock.
2734 Description: Add loi_kms_set().
2735 Details : Wrap kms updates into a helper function.
2739 Description: Constify instances of struct lsm_operations.
2740 Details : Constify instances of struct lsm_operations.
2744 Description: lu_conf support.
2745 Details : On a server, a file system object is uniquely identified
2746 by a fid, which is sufficient to locate and load all object
2747 state (inode). On a client, on the other hand, more data are
2748 necessary instantiate an object. Change lu_object_find() and
2749 friends to take additional `lu_conf' argument describing object.
2750 Typically this includes layout information.
2754 Description: lu_context fixes.
2755 Details : Introduce new lu_context functions that are needed on the client
2756 side, where some system threads (ptlrpcd) are shared by multiple
2757 modules, and so cannot be stopped during module shutdown.
2761 Description: Add start and stop methods to lu_device_type_operations.
2762 Details : Introduce two new methods in lu_device_type_operations, that are
2763 invoked when first instance of a given type is created and last one
2764 is destroyed respectively. This is need by CLIO.
2768 Description: Add lu_ref support to struct lu_device.
2769 Details : Add lu_ref support to lu_object and lu_device. lu_ref is used to
2770 track leaked references.
2774 Description: Introduce lu_kmem_descr.
2775 Details : lu_kmem_descr and its companion interface allow to create
2776 and destroy a number of kmem caches at once.
2780 Description: Fix lu_object finalization race.
2781 Details : Fix a race between lu_object_find() finding an object and its
2782 concurrent finalization. This race is (most likely) not possible
2783 on the server, but might happen on the client.
2787 Description: Introduce lu_ref interface.
2788 Details : lu_ref is a debugging module allowing to track references to
2789 a given object. It is quite cpu expensive, and has to be
2790 explicitly enabled with --enable-lu_ref. See usage description
2795 Description: Factor lu_site procfs stats into a separate function.
2796 Details : Separate lu_site stats printing code into a separate function
2797 that can be reused on a client.
2801 Description: Constify instances of struct {lu,dt,md}_device_operations.
2802 Details : Constify instances of struct {lu,dt,md}_device_operations.
2806 Description: Introduce struct md_site and move meta-data specific parts of
2807 struct lu_site here.
2808 Details : Move md-specific fields out of struct lu_site into special struct
2809 md_site, so that lu_site can be used on a client.
2813 Description: Kill mdd_lov_destroy().
2814 Details : Remove unused mdd code.
2818 Description: Add st_block checking to multistat.c.
2819 Details : Add st_block checking to multistat.c.
2823 Description: Add lu_ref support to struct obd_device.
2824 Details : Add lu_ref tracking to obd_device.
2828 Description: Kill obd_set_fail_loc().
2829 Details : Remove unused code.
2833 Description: Add special type for ptlrpc_request interpret functions.
2834 Details : Add lu_env parameter to ->rq_interpreter call-back. NULL is passed
2835 there. Actual usage will be in CLIO.
2839 Description: Replace RW_LOCK_UNLOCKED() macro with rwlock_init().
2840 Details : Replace RW_LOCK_UNLOCKED() with rwlock_init() as the former
2841 doesn't work with lockdep.
2845 Description: Add rwv.c test program.
2846 Details : New testing program exercising readv(2) and writev(2) (Qian).
2850 Description: Add sendfile.c test program.
2851 Details : New testing program exercising sendfile(2) (Jay).
2855 Description: Ratelimit a message that can be very frequent.
2856 Details : Ratelimit a memory allocation failure message that can
2861 Description: Use cdebug_show() in CDEBUG-style macros defined outside of libcfs.
2862 Details : Use cdebug_show() in CDEBUG-style macros defined outside of libcfs.
2866 Description: Liblustre build fixes.
2867 Details : Liblustre build fixes.
2871 Description: libcfs: add cfs_{need,cond}_resched() interface.
2872 Details : libcfs: add cfs_{need,cond}_resched() definition and
2873 implementations for Linux, NT, and liblustre.
2875 Severity : enhancement
2877 Description: More exported tunables for mballoc
2878 Details : Add support for tunable preallocation window and new tunables for
2879 large/small requests
2883 Description: Detect corruption of block bitmap and checking for preallocations
2884 Details : Checks validity of on-disk block bitmap. Also it does better
2885 checking of number of applied preallocations. When corruption is
2886 found, it turns filesystem readonly to prevent further corruptions.
2890 Description: (rw.c:1323:ll_read_ahead_pages()) ASSERTION(page_idx > ria->ria_stoff) failed
2891 Details : Once the unmatched stride IO mode is detected, shrink the stride-ahead
2892 window to 0. If it does hit cache miss, and read-pattern is still
2893 stride-io mode, does not reset the stride window, but also does not
2894 increase the stride window length in this case.
2898 Frequency : only for big-endian servers
2899 Description: Check if system is big-endian while mounting fs with extents feature
2900 Details : Mounting a filesystem with extents feature will fail on big-endian
2901 systems since ext3-based ldiskfs is not supported on big-endian
2902 systems. This can be over-riden with "bigendian_extents" mount option.
2904 Severity : enhancement
2906 Description: The root squash functionality
2907 Details : A security feature, which is to prevent users from being able
2908 to mount lustre on their desktop, run as root, and delete
2909 all of the files in the filesystem. The goal is accomplished by
2910 remapping user id (UID) and group id (GID) of the root user to
2911 a UID and GID specified by the system administartor via Lustre
2912 configuration management server (MGS). The functionality also
2913 allows to specify sets of clients for which the remapping does
2918 Description: Excessive recovery window
2919 Details : With AT enabled, the recovery window can be excessively long (6000+
2920 seconds). To address this problem, we no longer use
2921 OBD_RECOVERY_FACTOR when extending the recovery window (the connect
2922 timeout no longer depends on the service time, it is set to
2923 INITIAL_CONNECT_TIMEOUT now) and clients report the old service
2924 time via pb_service_time.
2928 Description: Watchdog triggered on MDS failover
2929 Details : enable OBD_CONNECT_MDT flag when connecting from the MDS so that
2930 the OSTs know that the MDS "UUID" can be reused for the same export
2931 from a different NID, so we do not need to wait for the export to be
2935 Frequency : rare, only if using MMP with Linux RAID
2937 Description: MMP doesn't work with Linux RAID
2938 Details : While using HA for Lustre servers with Linux RAID, it is possible
2939 that MMP will not detect multiple mounts. To make this work we
2940 need to unplug the device queue in RAID when the MMP block is being
2941 written. Also while reading the MMP block, we should read it from
2942 disk and not the cached one.
2944 Severity : enhancement
2946 Description: open file using fid
2947 Details : A file can be opened using just its fid, like
2948 <mntpt>/.lustre/fid/SEQ:OID:VER - this is needed for HSM and replication
2951 Frequency : Only in RHEL5 when mounting multiple ext3 filesystems
2954 Description: "kmem_cache_create: duplicate cache jbd_4k" error message
2955 Details : add proper locking for creation of jbd_4k slab cache
2959 Description: MMP check in ext3_remount() fails without displaying any error
2960 Details : When multiple mount protection fails during remount, proper error
2963 Severity : enhancement
2965 Description: Allow stripe size to be up to 4G-64k
2966 Details : Fix math logic to allow large stripe sizes.
2970 Description: add check for >8TB ldiskfs filesystems
2971 Details : ext3-based ldiskfs does not support greater than 8TB LUNs.
2972 Don't allow >8TB ldiskfs filesystems to be mounted without
2973 force_over_8tb mount option
2978 Descriptoin: OSS complains "too many blocks" and crashes
2979 Details : In filter_commitrw_write():
2980 1. lquota_chkquota();
2981 2. obdo_from_inode();
2982 3. lquota_pending_commit();
2983 "2." has changed uid and gid in this case, then caused
2984 lquota_chkquota() and lquota_pending_commit() to use
2985 different uid and gid. That is the root of the bug.
2989 Descriptoin: getstripe doesn't report stripe count = -1 sanely
2990 Details : Detect and print out the 'stripe count = -1' case properly.
2994 Description: Out or order replies might be lost on replay
2995 Details : In ptlrpc_retain_replayable_request if we cannot find retained
2996 request with tid smaller than one currently being added, add it
2997 to the start, not end of the list.
2999 --------------------------------------------------------------------------------
3001 2007-08-10 Cluster File Systems, Inc. <info@clusterfs.com>
3003 * Support for kernels:
3004 2.6.5-7.283 (SLES 9),
3005 2.6.9-55.EL (RHEL 4),
3006 2.6.16.46-0.14 (SLES 10),
3007 2.6.18.8 vanilla (kernel.org)
3008 * Client support for unpatched kernels:
3009 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
3010 2.6.16 - 2.6.22 vanilla (kernel.org)
3011 * Due to recently discovered recovery problems, we do not recommend
3012 using patchless RHEL 4 clients with this or any earlier release.
3013 * Recommended e2fsprogs version: 1.39.cfs8
3014 * Note that reiserfs quotas are disabled on SLES 10 in this kernel.
3015 * Starting with this release, the ldiskfs backing filesystem required
3016 by Lustre is now in its own package, lustre-ldiskfs. This package
3017 should be installed. It is versioned separately from Lustre and
3018 may be released separately in future.
3022 Description: block reactivating mgc import until all deactivates complete
3023 Details : Fix race when failing back MDT/MGS to itself (testing)
3026 Frequency : at statup only
3028 Description: mds_lov_synchronize race leads to various problems
3029 Details : simultaneous MDT->OST connections at startup can cause the
3030 sync to abort, leaving the OSC in a bad state.
3032 Severity : enhancement
3034 Description: add optional extra BUILD_VERSION info
3035 Details : add a new environment variable (namely LUSTRE_VERS) which allows
3036 to override the lustre version.
3039 Frequency : 2.6.18 servers only
3041 Description: ll_kern_mount() doesn't release the module reference
3042 Details : The ldiskfs module reference count never drops down to 0
3043 because ll_kern_mount() doesn't release the module reference.
3048 Description: server LBUG when using old ost_num_threads parameter
3049 Details : Accept the old ost_num_threads parameter but warn that it
3050 is deprecated, and fix an off-by-one error that caused an LBUG.
3055 Description: Transient SCSI error results in persistent IO issue
3056 Details : iobuf->dr_error is not reinitialized to 0 between two
3060 Frequency : sometimes when underlying device returns I/O errors
3062 Description: OSTs not going read-only during write failures
3063 Details : OSTs are not remounted read-only when the journal commit threads
3064 get I/O errors because fsfilt_ext3 calls journal_start/stop()
3065 instead of the ext3 wrappers.
3069 Description: poor connect scaling with increasing client count
3070 Details : Don't run filter_grant_sanity_check for more than 100 exports
3071 to improve scaling for large numbers of clients.
3074 Frequency : SLES10 only
3076 Description: sanity-quota.sh quotacheck failed: rc = -22
3077 Details : Quotas cannot be enabled on SLES10.
3080 Frequency : liblustre clients only
3082 Description: getdirentries does not give error when run on compute nodes
3083 Details : getdirentries does not fail when the size specified as an argument
3084 is too small to contain at least one entry
3086 Severity : enhancement
3088 Description: Add LNET router traceability for debug purposes
3089 Details : If a checksum failure occurs with a router as part of the
3090 IO path, the NID of the last router that forwarded the bulk data
3091 is printed so it can be identified.
3096 Description: OST "spontaneously" evicts client; client has imp_pingable == 0
3097 Details : Due to a race condition, liblustre clients were occasionally
3098 evicted incorrectly.
3100 Severity : enhancement
3102 Description: lfs setstripe use optional parameters instead of postional
3105 Severity : enhancement
3107 Description: Nanosecond timestamp support for ldiskfs
3108 Details : The on-disk ldiskfs filesystem has added support for nanosecond
3109 resolution timestamps. There is not yet support for this at
3110 the Lustre filesystem level.
3113 Frequency : during server recovery
3115 Description: MDS failing to send precreate requests due to OSCC_FLAG_RECOVERING
3116 Details : request with rq_no_resend flag not awake l_wait_event if they get
3120 Frequency : nfs export on patchless client
3122 Description: connectathon hang when test nfs export over patchless client
3123 Details : Disconnected dentry cannot be found with lookup, so we do not need
3124 to unhash it or make it invalid
3127 Description: fix llapi_lov_get_uuids() to allow many OSTs to be returned
3128 Details: : Change llapi_lov_get_uuids() to read the UUIDs from /proc instead
3129 of using an ioctl. This allows lfsck for > 160 OSTs to succeed.
3134 Description: open req refcounting wrong on reconnect
3135 Details : If reconnect happened between getting open reply from server and
3136 call to mdc_set_replay_data in ll_file_open, we will schedule
3137 replay for unreferenced request that we are about to free.
3138 Subsequent close will crash in variety of ways.
3139 Check that request is still eligible for replay in
3140 mdc_set_replay_data().
3145 Description: disable writes to filesystem when reading health_check file
3146 Details : the default for reading the health_check proc file has changed
3147 to NOT do a journal transaction and write to disk, because this
3148 can cause reads of the /proc file to hang and block HA state
3149 checking on a healthy but otherwise heavily loaded system. It
3150 is possible to return to the previous behaviour during configure
3151 with --enable-health-write.
3153 Severity : enhancement
3155 Description: 64-bit inode version
3156 Details: : Add a on-disk 64-bit inode version for ext3 to track changes made
3157 to the inode. This will be required for version-based recovery.
3162 Description: MDS fails to start if a duplicate client export is detected
3163 Details : in some rare cases it was possible for a client to connect to
3164 an MDS multiple times. Upon recovery the MDS would detect this
3165 and fail during startup. Handle this more gracefully.
3167 Severity : enhancement
3169 Description: Add -o localflock option to simulate old noflock
3171 Details : This will achieve local-only flock/fcntl locks
3177 Description: log_commit_thread vs filter_destroy race leads to crash
3178 Details : Take import reference before releasing llog record semaphore
3183 Description: Wrong request locking in request set processing
3184 Details : ptlrpc_check_set wrongly uses req->rq_lock for proctect add to
3185 imp_delayed_list, in this place should be used imp_lock.
3188 Frequency : when reconnection
3190 Description: Grant Leak when osc reconnect to OST
3191 Details : When osc reconnect ost, OST(filter) should check whether it
3192 should grant more space to client by comparing fed_grant and
3193 cl_avail_grant, and return the granted space to client instead
3194 of "new granted" space, because client will call osc_init_grant
3195 to update the client grant space info.
3198 Frequency : when client reconnect to OST
3200 Description: Grant Leak when osc do resend and replay bulk write
3201 Details : When osc reconnect to OST, OST(filter)should clear grant info of
3202 bulk write request, because the grant info while be sync between
3203 OSC and OST when reconnect, and we should ignore the grant info
3204 these of resend/replay write req.
3209 Description: Grant space more than available left space sometimes.
3210 Details : When then OST is about to be full, if two bulk writing from
3211 different clients came to OST. Accord the avaliable space of the
3212 OST, the first req should be permitted, and the second one
3213 should be denied by ENOSPC. But if the seconde arrived before
3214 the first one is committed. The OST might wrongly permit second
3215 writing, which will cause grant space > available space.
3218 Frequency : when client is evicted
3220 Description: Grant might be wrongly erased when osc is evicted by OST
3221 Details : when the import is evicted by server, it will fork another
3222 thread ptlrpc_invalidate_import_thread to invalidate the
3223 import, where the grant will be set to 0. While the original
3224 thread will update the grant it got when connecting. So if
3225 the former happened latter, the grant will be wrongly errased
3226 because of this race.
3231 Description: Checking Stale with correct fid
3232 Details : ll_revalidate_it should uses de_inode instead of op_data.fid2
3233 to check whether it is stale, because sometimes, we want the
3234 enqueue happened anyway, and op_data.fid2 will not be initialized.
3236 Severity : enhancement
3238 Description: update patchless client
3239 Details : Add support for patchless client with 2.6.20, 2.6.21 and RHEL 5
3242 Frequency : only with 2.4 kernel
3244 Description: random memory corruption
3245 Details : size of struct ll_inode_info is to big for union inode.u and this
3246 can be cause of random memory corruption.
3251 Description: Memory leak in recovery
3252 Details : Lov_mds_md was not free in an error handler in mds_create_object.
3253 It should also check obd_fail before fsfilt_start, otherwise if
3254 fsfilt_start return -EROFS,(failover mds during mds recovery).
3255 then the req will return with repmsg->transno = 0 and rc = EROFS.
3256 and we met hit the assert LASSERT(req->rq_reqmsg->transno ==
3257 req->rq_repmsg->transno) in ptlrpc_replay_interpret. Fcc should
3258 be freed no matter whether fsfilt_commit success or not.
3261 Frequency : only with huge count clients
3263 Description: Prevents from taking the superblock lock in llap_from_page for
3265 Details : using LL_ORIGIN_REMOVEPAGE origin flag instead of LL_ORIGIN_UNKNOW
3266 for llap_from_page call in ll_removepage prevents from taking the
3267 superblock lock for a soon died page.
3272 Description: Not check open intent error before release open handle
3273 Details : in some rare cases, the open intent error is not checked before
3274 release open handle, which may cause
3275 ASSERTION(open_req->rq_transno != 0), because it tries to release
3276 the failed open handle.
3281 Description: Set cat log bitmap only after create log success.
3282 Details : in some rare cases, the cat log bitmap is set too early. and it
3283 should be set only after create log success.
3287 Description: Accessing a block bevice can re-enable I/O when Lustre is
3288 tearing down a device.
3289 Details : dev_clear_rdonly(bdev) must be called in kill_bdev() instead of
3294 Description: service threads may hog cpus when there are a lot of requests
3295 Details : Insert cond_resched to give other threads a chance to use some CPU
3300 Description: the cat log was not initialized in recovery
3301 Details : When mds(mgs) do recovery, the tgt_count might be zero, so the
3302 unlink log on mds will not be initialized until mds post
3303 recovery. And also in mds post recovery, the unlink log will
3304 initialization will be done asynchronausly, so there will be race
3305 between add unlink log and unlink log initialization.
3309 Description: brw_stats were being printed incorrectly
3310 Details : brw_stats were being printed as log2 but all of them were not
3311 recorded as log2. Also remove some code duplication arising from
3312 filter_tally_{read,write}.
3316 Frequency : rare, only in recovery.
3317 Description: ASSERTION(req->rq_type != LI_POISON) failed
3318 Details : imp_lock should be held while iterating over imp_sending_list for
3319 prevent destroy request after get timeout in ptlrpc_queue_wait.
3323 Description: replay-single.sh test 52 fails
3324 Details : A lock's skiplist need to be cleanup when it being unlinked
3325 from its resource list.
3329 Description: Short directio read returns full requested size rather than
3331 Details : Direct I/O operations should return actual amount of bytes
3332 transferred rather than requested size.
3334 Severity : enhancement
3336 Description: metadata RPC reduction (e.g. for rm performance)
3337 Details : decrease the amount of synchronous RPC between clients and servers
3338 by canceling conflicing lock before the operation on the client
3339 and packing thier handles into the main operation RPC to server.
3341 Severity : enhancement
3343 Description: Async OSC create to avoid the blocking unnecessarily.
3344 Details : If an OST has no remaining object, system will block on the
3345 creation when it needs to create a new object on this OST. Now,
3346 ways use pre-created objects when available, instead of blocking on
3347 an empty osc while others are not empty. If we must block, we
3348 block for the shortest possible period of time.
3352 Description: improve handling recoverable errors
3353 Details : if request processig with error which can be recoverable on server
3354 request should be resend, otherwise page released from cache and
3357 Severity : enhancement
3359 Description: refine locking for avoid write wrong info into lov_objid file
3360 Details : fix possible races with add new target and write/update data in
3363 --------------------------------------------------------------------------------
3365 2007-05-03 Cluster File Systems, Inc. <info@clusterfs.com>
3370 Frequency : on some architectures
3372 Description: 1.6 client sometimes fails to mount from a 1.4 MDT
3373 Details : Uninitialized flags sometimes cause configuration commands to
3377 Frequency : patchless clients only
3379 Description: missing __iget() symbol export
3380 Details : The __iget() symbol export is missing. To avoid the need for
3381 this on patchless clients the deathrow inode reaper is turned
3382 off, and we depend on the VM to clean up old inodes. This
3383 dependency was during via the fix for bug 12181.
3385 --------------------------------------------------------------------------------
3387 2007-04-19 Cluster File Systems, Inc. <info@clusterfs.com>
3389 * CONFIGURATION CHANGE. This version of Lustre WILL NOT
3390 INTEROPERATE with older versions automatically. In many cases a
3391 special upgrade step is needed. Please read the
3392 user documentation before upgrading any part of a live system.
3393 * WIRE PROTOCOL CHANGE from previous 1.6 beta versions. This
3394 version will not interoperate with 1.6 betas before beta5 (1.5.95).
3395 * WARNING: Lustre configuration and startup changes are required with
3396 this release. See https://mail.clusterfs.com/wikis/lustre/MountConf
3400 Severity : enhancement
3402 Description: MountConf
3403 Details : Lustre configuration is now managed via mkfs and mount
3404 commands instead of lmc and lconf. New obd types (MGS, MGC)
3405 are added for dynamic configuration management. See
3406 https://mail.clusterfs.com/wikis/lustre/MountConf for
3409 Severity : enhancement
3411 Description: dynamic OST addition
3412 Details : OSTs can now be added to a live filesystem
3414 Severity : enhancement
3416 Description: startup order invariance
3417 Details : MDTs and OSTs can be started in any order. Clients only
3418 require the MDT to complete startup.
3420 Severity : enhancement
3422 Description: parallel, asynchronous orphan cleanup
3423 Details : orphan cleanup is now performed in separate threads for each
3424 OST, allowing parallel non-blocking operation.
3426 Severity : enhancement
3428 Description: optimized stripe assignment
3429 Details : stripe assignments are now made based on ost space available,
3430 ost previous usage, and OSS previous usage, in order to try
3431 to optimize storage space and networking resources.
3433 Severity : enhancement
3435 Description: Permanently set tunables
3436 Details : All writable /proc/fs/lustre tunables can now be permanently
3437 set on a per-server basis, at mkfs time or on a live system.
3439 Severity : enhancement
3441 Description: Lustre message v2
3442 Details : Add lustre message format v2.
3444 Severity : enhancement
3446 Description: client OST exclusion list
3447 Details : Clients can be started with a list of OSTs that should be
3448 declared "inactive" for known non-responsive OSTs.
3452 Description: SPEC SFS validation failure on NFS v2 over lustre.
3453 Details : Changes the blocksize for regular files to be 2x RPC size,
3454 and not depend on stripe size.
3456 Severity : enhancement
3458 Description: Multiple MD RPCs in flight.
3459 Details : Further unserialise some read-only MDS RPCs - learn about intents.
3460 To avoid overly-overloading MDS, introduce a limit on number of
3461 MDS RPCs in flight for a single client and add /proc controls
3462 to adjust this limit.
3464 Severity : enhancement
3466 Description: client read/write statistics
3467 Details : Add client read/write call usage stats for performance
3468 analysis of user processes.
3469 /proc/fs/lustre/llite/*/offset_stats shows non-sequential
3470 file access. extents_stats shows chunk size distribution.
3471 extents_stats_per_process show chunk size distribution per
3474 Severity : enhancement
3476 Description: mds statistics
3477 Details : Add detailed mds operations statistics in
3478 /proc/fs/lustre/mds/*/stats.
3482 Description: Failure of copying files with lustre special EAs.
3483 Details : Client side always return success for setxattr call for lustre
3484 special xattr (currently only "trusted.lov").
3489 Description: Refcount LNET uuids
3490 Details : The global LNET uuid list grew linearly with every startup;
3491 refcount repeated list entries instead of always adding to
3494 Severity : enhancement
3496 Description: Dynamic service threads
3497 Details : Within a small range, start extra service threads
3498 automatically when the request queue builds up.
3501 Frequency : mixed-endian client/server environments
3503 Description: mixed-endian crashes
3504 Details : The new msg_v2 system had some failures in mixed-endian
3507 Severity : enhancement
3509 Description: Easy OST removal
3510 Details : OSTs can be permanently deactivated with e.g. 'lctl
3511 conf_param lustre-OST0001.osc.active=0'
3513 Severity : enhancement
3515 Description: MGS proc entries
3516 Details : Added basic proc entries for the MGS showing what filesystems
3519 Severity : enhancement
3521 Description: provide MGS failover
3522 Details : Added config lock reacquisition after MGS server failover.
3524 Severity : enhancement
3526 Description: add Linux 2.4 support
3527 Details : Added support for RHEL 2.4.21 kernel for 1.6 servers and clients
3531 Description: a large application tries to do I/O to the same resource and dies
3532 in the middle of it.
3533 Details : Check the req->rq_arrival time after the call to
3534 ost_brw_lock_get(), but before we do anything about
3535 processing it & sending the BULK transfer request. This
3536 should help move old stale pending locks off the queue as
3537 quickly as obd_timeout.
3540 Frequency : when an incorrect nid is specified during startup
3542 Description: ptlrpc connect to non-existant node causes kernel crash
3543 Details : LNET can't be re-entered from an event callback, which
3544 happened when we expire a message after the export has been
3545 cleaned up. Instead, hand the zombie cleanup off to another
3548 Severity : enhancement
3550 Description: plain/inodebits lock performance improvement
3551 Details : Grouping plain/inodebits in granted list by their request modes
3552 and bits policy, thus improving the performance of search through
3556 Frequency : only if OST filesystem is corrupted
3558 Description: client incorrectly hits assertion in ptlrpc_replay_req()
3559 Details : for a short time RPCs with bulk IO are in the replay list,
3560 but replay of bulk IOs is unimplemented. If the OST filesystem
3561 is corrupted due to disk cache incoherency and then replay is
3562 started it is possible to trip an assertion. Avoid putting
3563 committed RPCs into the replay list at all to avoid this issue.
3566 Frequency : liblustre (e.g. catamount) on a large cluster with >= 8 OSTs/OSS
3568 Description: System hang on startup
3569 Details : This bug allowed the liblustre (e.g. catamount) client to
3570 return to the app before handling all startup RPCs. This
3571 could leave the node unresponsive to lustre network traffic
3572 and manifested as a server ptllnd timeout.
3574 Severity : enhancement
3576 Description: Add "/proc/sys/lustre/debug_peer_on_timeout"
3577 Details : liblustre envirable: LIBLUSTRE_DEBUG_PEER_ON_TIMEOUT
3578 boolean to control whether to print peer debug info when a
3579 client's RPC times out.
3582 Frequency : only for kernels with patches from Lustre below 1.4.3
3584 Description: Remove old rdonly API
3585 Details : Remove old rdonly API which unused from at least lustre 1.4.3
3588 Frequency : only for devices with external journals
3590 Description: Set external device read-only also
3591 Details : During a commanded failover stop, we set the disk device
3592 read-only while the server shuts down. We now also set any
3593 external journal device read-only at the same time.
3596 Frequency : when upgrading from 1.4 while trying to change parameters
3598 Description: The wrong (new) MDC name was used when setting parameters for
3599 upgraded MDT's. Also allows changing of OSC (and MDC)
3600 parameters if --writeconf is specified at tunefs upgrade time.
3603 Frequency : when setting specific ost indicies
3605 Description: QOS code breaks on skipped indicies
3606 Details : Add checks for missing OST indicies in the QOS code, so OSTs
3607 created with --index need not be sequential.
3609 Severity : enhancement
3611 Description: Add uninit_groups feature to ldiskfs2 to speed up e2fsck
3612 Details : The uninit_groups feature works in conjunction with the kernel
3613 filesystem code (ldiskfs2 only) and e2fsprogs-1.39-cfs6 to speed
3614 up the pass1 processing of e2fsck. This is a read-only feature
3615 in ldiskfs2 only, so older kernels and current ldiskfs cannot
3616 mount filesystems that have had this feature enabled.
3618 Severity : enhancement
3620 Description: Improve multi-block allocation algorithm to avoid fragmentation
3621 Details : The mballoc3 code (ldiskfs2 only) adds new mechanisms to improve
3622 allocation locality and avoid filesystem fragmentation.
3624 ------------------------------------------------------------------------------
3626 2007-04-01 Cluster File Systems, Inc. <info@clusterfs.com>
3628 * Support for kernels:
3629 2.4.21-47.0.1.EL (RHEL 3)
3630 2.6.5-7.283 (SLES 9)
3631 2.6.9-42.0.10.EL (RHEL 4)
3632 2.6.12.6 vanilla (kernel.org)
3633 2.6.16.27-0.9 (SLES 10)
3634 * Recommended e2fsprogs version: 1.39.cfs5
3636 * Note that reiserfs quotas are disabled on SLES 10 in this kernel
3640 Frequency : occasional, depends on client load and configuration
3641 Bugzilla : 12181, 12203
3642 Description: data loss for recently-modified files
3644 Details : In some cases it is possible that recently written or created
3645 files may not be written to disk in a timely manner (this should
3646 normally be within 30s unless client IO load is very high).
3647 The problem appears as zero-length files or files that are a
3648 multiple of 1MB in size after a client crash or client eviction
3649 that are missing data at the end of the file.
3651 This problem is more likely to be hit on clients where files are
3652 repeatedly created and unlinked in the same directory, clients
3653 have a large amount of RAM, have many CPUs, the filesystem has
3654 many OSTs, the clients are rebooted frequently, and/or the files
3655 are not accessed by other nodes after being written.
3657 The presence of the problem can be detected by looking at
3658 /proc/sys/fs/inode-state. If the first number (nr_inodes) is
3659 smaller than the second (nr_unused) then dirty files will not
3660 be flushed automatically to disk. "sync; sleep 10" should be
3661 run several times on the node before unmounting it to update
3662 Lustre (this is also safe to run on nodes without this problem).
3664 There is also a related kernel bug in the RHEL4 4 2.6.9 kernel
3665 that can cause this same problem, so customers using that kernel
3666 also need to update the kernel in addition to Lustre. In order
3667 to properly fix this bug, the RHEL3 2.4.21 kernel is also updated.
3669 It is normal that files written just before a client crash (less
3670 than 30s) may not yet have been flushed to disk, even for local
3674 Frequency : frequent on thin XT3 nodes
3676 Description: UUID collision on thin XT3 Linux nodes
3677 Details : UUIDs on Compute Node Linux XT3 nodes were not generated
3678 randomly, since we relied on an insufficiently-seeded PRNG.
3683 Description: OSS hangs after "All ost request buffers busy"
3684 Details : A deadlock between quota and journal operations caused OSS
3685 hangs after printing "All ost request buffers busy."
3688 Frequency : always on liblustre builds
3690 Description: Cleanup compiler warnings on liblustre
3693 Frequency : always on liblustre builds on XT3
3695 Description: LC_CONFIG_CDEBUG don't run while build liblustre on XT3.
3699 Description: Addition of EXT3_FEATURE_RO_COMPAT_DIR_NLINKS flag for
3700 > 32000 subdirectories
3701 Details : Add EXT3_FEATURE_RO_COMPAT_DIR_NLINK flag to
3702 EXT3_FEATURE_RO_COMPAT_SUPP. This flag will be set whenever
3703 subdirectory count crosses 32000. This will aid e2fsck to
3704 correctly handle more than 32000 subdirectories.
3707 Frequency : liblustre (e.g. catamount) on a large cluster with >= 8 OSTs/OSS
3709 Description: System hang on startup
3710 Details : This bug allowed the liblustre (e.g. catamount) client to
3711 return to the app before handling all startup RPCs. This
3712 could leave the node unresponsive to lustre network traffic
3713 and manifested as a server ptllnd timeout.
3715 Severity : enhancement
3717 Description: Add "/proc/sys/lustre/debug_peer_on_timeout"
3718 (liblustre envirable: LIBLUSTRE_DEBUG_PEER_ON_TIMEOUT)
3719 boolean to control whether to print peer debug info when a
3720 client's RPC times out.
3725 Description: make O_SYNC working on 2.6 kernels
3726 Details : 2.6 kernels use different method for mark pages for write,
3727 so need add a code to lustre for O_SYNC work.
3732 Description: Failure to close file and release space on NFS
3733 Details : Put inode details into lock acquired in ll_intent_file_open.
3734 Use mdc_intent_lock in ll_intent_open to properly
3735 detect all kind of errors unhandled by mdc_enqueue.
3740 Description: proc file read during shutdown sometimes raced obd removal,
3742 Details : Add lock to prevent obd access after proc file removal.
3745 Frequency : Only for files larger than 4GB on 32-bit clients.
3747 Description: improperly doing page alignment of locks
3748 Details : Modify lustre core code to use CFS_PAGE_* defines instead of
3749 PAGE_*. Make CFS_PAGE_MASK a 64-bit mask.
3754 Description: RPCs being resent when they shouldn't be
3755 Details : Some RPCs that should not be resent are being resent. This
3756 can cause inconsistencies in the RPC state machine. Do not
3757 resend such requests.
3760 Frequency : rare, only with NFS export
3762 Description: Crash on NFS re-export node
3763 Details : under very unusual load conditions an assertion is hit in
3764 ll_intent_file_open()
3767 Frequency : only if OST filesystem is corrupted
3769 Description: client incorrectly hits assertion in ptlrpc_replay_req()
3770 Details : for a short time RPCs with bulk IO are in the replay list,
3771 but replay of bulk IOs is unimplemented. If the OST filesystem
3772 is corrupted due to disk cache incoherency and then replay is
3773 started it is possible to trip an assertion. Avoid putting
3774 committed RPCs into the replay list at all to avoid this issue.
3779 Description: large O_DIRECT requests fail under memory pressure/fragmentation
3780 Details : Large single O_DIRECT read and write calls can fail to allocate
3781 a sufficiently large buffer to process the request. In case of
3782 allocation failure the allocation is retried with a smaller
3783 buffer and broken into smaller requests.
3785 Severity : enhancement
3787 Description: Add -o localflock option to simulate old noflock behaviour.
3788 Details : This will achieve local-only flock/fcntl locks coherentness.
3793 Description: versioning check is incomplete
3794 Details : Checking the version difference of client vs. server, report
3795 error if the gap is too big.
3800 Description: add support PG_writeback bit
3801 Details : add support for PG_writeback bit for Lustre, for more carefull
3802 work with page cache in 2.6 kernel. This also fix some deadlocks
3803 and remove hack for work O_SYNC with 2.6 kernel.
3805 Severity : enhancement
3807 Description: Add uninit_groups feature to ldiskfs2 to speed up e2fsck
3808 Details : The uninit_groups feature works in conjunction with the kernel
3809 filesystem code (ldiskfs2 only) and e2fsprogs-1.39-cfs6 to speed
3810 up the pass1 processing of e2fsck. This is a read-only feature
3811 in ldiskfs2 only, so older kernels and current ldiskfs cannot
3812 mount filesystems that have had this feature enabled.
3814 Severity : enhancement
3816 Description: Improve multi-block allocation algorithm to avoid fragmentation
3817 Details : The mballoc3 code (ldiskfs2 only) adds new mechanisms to improve
3818 allocation locality and avoid filesystem fragmentation.
3820 ------------------------------------------------------------------------------
3822 2007-02-09 Cluster File Systems, Inc. <info@clusterfs.com>
3824 * Support for kernels:
3825 2.6.9-42.0.3.EL (RHEL 4)
3826 2.6.5-7.276 (SLES 9)
3827 2.4.21-47.0.1.EL (RHEL 3)
3828 2.6.12.6 vanilla (kernel.org)
3829 2.6.16.21-0.8 (SLES10)
3830 * Recommended e2fsprogs version: 1.39.cfs2-0
3832 * The backwards-compatible /proc/sys/portals symlink has been removed
3833 in this release. Before upgrading, please ensure that you change
3834 any configuration scripts or /etc/sysctl.conf files that access
3835 /proc/sys/portals/* or sysctl portals.* to use the corresponding
3836 entry in /proc/sys/lnet or sysctl lnet.*. This change can be made
3837 in advance of the upgrade on any system running Lustre 1.4.6 or
3838 newer, since /proc/sys/lnet was added in that version.
3839 * Note that reiserfs quotas are disabled on SLES 10 in this kernel
3843 Frequency : only when quota is used
3845 Description: avoid scanning export list for quota master
3846 Details : Change the algorithms to avoid scanning export list in order
3847 to improve the efficiency.
3850 Frequency : MDS failover only, very rarely
3852 Description: "went back in time" messages on mds failover
3853 Details : The greatest transno may be lost when the current operation
3854 finishes with an error (transno==0) and the client's last_rcvd
3855 record is over-written. Save the greatest transno in the
3856 mds_last_transno for this case.
3859 Frequency : always for specific kernels and striping counts
3861 Description: client may get "Matching packet too big" without ACL support
3862 Details : Clients compiled without CONFIG_FS_POSIX_ACL get an error message
3863 when trying to access files in certain configurations. The
3864 clients should in fact be denied when mounting because they do
3865 not understand ACLs.
3868 Frequency : Cray XT3 with more than 4000 clients and multiple jobs
3870 Description: many clients connecting with IO in progress causes connect timeouts
3871 Details : Avoid synchronous journal commits to avoid delays caused by many
3872 clients connecting/disconnecting when bulk IO is in progress.
3873 Queue liblustre connect requests on OST_REQUEST_PORTAL instead of
3874 OST_IO_PORTAL to avoid delays behind potentially many pending
3878 Frequency : occasionally with multiple writers to a single file
3880 Description: shared writes to file may result in wrong size reported by stat()
3881 Details : Allow growing of kms when extent lock is cancelled
3884 Frequency : always with random mmap IO to multi-striped file
3886 Description: mmap write might be lost if we are writing to a 'hole' in stripe
3887 Details : Only if the hole is at the end of OST object so that kms is too
3888 small. Fix is to increase kms accordingly in ll_nopage.
3891 Frequency : rare, only if OST filesystem is inconsistent with MDS filesystem
3893 Description: writes to a missing object would leak memory on the OST
3894 Details : If there is an inconsistency between the MDS and OST filesystems,
3895 such that the MDS references an object that doesn't exist, writes
3896 to that object will leak memory due to incorrect cleanup in the
3897 error handling path, eventually running out of memory on the OST.
3902 Description: Creating too long symlink causes lustre errors
3903 Details : Check symlink and name lengths before sending requests to MDS.
3906 Frequency : only if flock is enabled (not on by default)
3908 Description: posix locks not released on fd closure on 2.6.9+
3909 Details : We failed to add posix locks to list of inode locks on 2.6.9+
3910 kernels, this caused such locks not to be released on fd close and
3911 then assertions on fs unmount about still used locks.
3914 Frequency : MDS failover only, very rarely
3916 Description: clients may get ASSERTION(granted_lock != NULL)
3917 Details : When request was taking a long time, and a client was resending
3918 a getattr by name lock request. The were multiple lock requests
3919 with the same client lock handle and
3920 mds_getattr_name->fixup_handle_for_resent_request found one of the
3921 lock handles but later failed with ASSERTION(granted_lock != NULL).
3926 Description: handle->h_buffer_credits > 0, assertion failure
3927 Details : h_buffer_credits is zero after truncate, causing assertion
3928 failure. This patch extends the transaction or creates a new
3932 Frequency : NFS re-export or patchless client
3933 Bugzilla : 11179, 10796
3934 Description: Crash on NFS re-export node (__d_move)
3935 Details : We do not want to hash the dentry if we don't have a lock.
3936 But if this dentry is later used in d_move, we'd hit uninitialised
3937 list head d_hash, so we just do this to init d_hash field but
3938 leave dentry unhashed.
3941 Frequency : NFS re-export or patchless client
3943 Description: NFS exports has problem with symbolic link
3944 Details : lustre client didn't properly install dentry when re-exported
3945 to NFS or running patchless client.
3948 Frequency : NFS re-export or patchless client
3950 Description: Various nfs/patchless fixes.
3951 Details : fixes reuse disconected alias for lookup process - this fixes
3952 warning "find_exported_dentry: npd != pd",
3953 fix permission error with open files at nfs.
3954 fix apply umask when do revalidate.
3957 Frequency : occasional
3959 Description: Crash on NFS re-export node
3960 Details : calling clear_page() on the wrong pointer triggered oops in
3961 generic_mapping_read().
3964 Frequency : rarely, using O_DIRECT IO
3966 Description: unaligned directio crashes client with LASSERT
3967 Details : check for unaligned buffers before trying any requests.
3970 Frequency : rarely, using CFS RAID5 patches in non-standard kernel series
3972 Description: stale data returned from RAID cache
3973 Details : If only a small amount of IO is done to the RAID device before
3974 reading it again it is possible to get stale data from the RAID
3975 cache instead of reading it from disk.
3978 Frequency : always for sles10 kernel
3980 Description: sles10 support
3981 Details : ll_follow_link: compile fixes and using of nd_set_link
3982 under newer kernels.
3985 Frequency : depends on arch, kernel and compiler version, always on sles10
3988 Description: recursive or deep enough symlinks cause stack overflow
3989 Details : getting rid of large stack-allocated variable in
3993 Frequency : depends on hardware
3995 Description: lustre write performance loss in the SLES10 kernel
3996 Details : the performance loss is caused by using of write barriers in the
3997 ext3 code. The SLES10 kernel turns barrier support on by
3998 default. The fix is to undo that change for ldiskfs.
4000 ------------------------------------------------------------------------------
4002 2006-12-09 Cluster File Systems, Inc. <info@clusterfs.com>
4004 * Support for kernels:
4005 2.6.9-42.0.3EL (RHEL 4)
4006 2.6.5-7.276 (SLES 9)
4007 2.4.21-47.0.1.EL (RHEL 3)
4008 2.6.12.6 vanilla (kernel.org)
4012 Frequency : quota enabled and large files being deleted
4014 Description: releasing more than 4GB of quota at once hangs OST
4015 Details : If a user deletes more than 4GB of files on a single OST it
4016 will cause the OST to spin in an infinite loop. Release
4017 quota in < 4GB chunks, or use a 64-bit value for 1.4.7.1+.
4022 Description: statfs data retrieved from /proc may be stale or zero
4023 Details : When reading per-device statfs data from /proc, in the
4024 {kbytes,files}_{total,free,avail} files, it may appear
4025 as zero or be out of date.
4028 Frequency : systems with MD RAID1 external journal devices
4030 Description: lconf's call to blkid is confused by RAID1 journal devices
4031 Details : Use the "blkid -l" flag to locate the MD RAID device instead
4032 of returning all block devices that match the journal UUID.
4035 Frequency : always, for aggregate stripe size over 4GB
4037 Description: "lfs setstripe" fails assertion when setting 4GB+ stripe width
4038 Details : Using "lfs setstripe" to set stripe size * stripe count over 4GB
4039 will fail the kernel with "ASSERTION(lsm->lsm_xfersize != 0)"
4042 Frequency : always if "lfs find" used on a local file/directory
4044 Description: "lfs find" segfaults if used on a local file/directory
4045 Details : The case where a directory component was not specified wasn't
4046 handled correctly. Handle this properly.
4049 Frequency : always on ppc64
4051 Description: the write to an ext3 filesystem mounted with mballoc got stuck
4052 Details : ext3_mb_generate_buddy() uses find_next_bit() which does not
4053 perform endianness conversion.
4056 Frequency : rarely (truncate to non-zero file size after write under load)
4057 Bugzilla : 10730, 10687
4058 Description: Files padded with zeros to next 4K multiple
4059 Details : With filesystems mounted using the "extents" option (2.6 kernels)
4060 it is possible that files that are truncated to a non-zero size
4061 immediately after being written are filled with zero bytes beyond
4062 the truncated size. No file data is lost.
4064 Severity : enhancement
4066 Description: Allow recovery/failover for liblustre clients.
4067 Details : liblustre clients were unaware of failover configurations until
4070 Severity : enhancement
4072 Description: user file locks should fail when not mounting with flock option
4073 Details : Set up an error-returning stub in ll_file_operations.lock field
4074 to prevent incorrect behaviour when client is mounted without
4075 flock option. Also, set up properly f_op->flock field for
4079 Frequency : always on ia64
4081 Description: "lfs df" loops on printing out MDS statfs information
4082 Details : The obd_ioctl_data was not initialized and in some systems
4083 this caused a failure during the ioctl that did not return
4084 an error. Initialize the struct and return an error on failure.
4087 Frequency : SLES 9 only
4089 Description: Error of copying files with lustre special EAs as root
4090 Details : Client side always return success for setxattr call for lustre
4091 special xattr (currently only "trusted.lov").
4094 Frequency : rarely on clusters with both ia64+i386 clients
4096 Description: ia64+i686 clients doing shared IO on the same file may LBUG
4097 Details : In rare cases when both ia64+i686 (or other mixed-PAGE_SIZE)
4098 clients are doing concurrent writes to the same file it is
4099 possible that the ia64 clients may LASSERT because the OST
4100 extent locks are not PAGE_SIZE aligned. Ensure that grown
4101 locks are always aligned on the request boundary.
4104 Frequency : specific use, occasional
4106 Description: Overwriting in use executable truncates on-disk binary image
4107 Details : If one node attempts to overwrite an executable in use by
4108 another node, we now correctly return ETXTBSY instead of
4109 truncating the file.
4111 Severity : enhancement
4113 Description: Async OSC create to avoid the blocking unnecessarily.
4114 Details : If an OST has no remaining object, system will block on the creating
4115 when need to create a new object on this OST. Now, ways use
4116 pre-created objects when available, instead of blocking on an
4117 empty osc while others are not empty. If we must block, we block
4118 for the shortest possible period of time.
4123 Description: chmod on Lustre root is propagated to other clients
4124 Details : Re-validate root's dentry in ll_lookup_it to avoid having it
4125 invalid by the follow_mount time.
4128 Frequency : liblustre clients only
4130 Description: Race in 'instant cancel' lock handling could lead to such locks
4131 never to be granted in case of SMP MDS
4132 Details : Do not destroy not yet granted but cbpending locks in
4136 Frequency : replay/resend of open
4138 Description: non null lock assetion failure in mds_intent_policy
4139 Details : Trying to replay/resend lockless open requests resulted in
4140 mds_open() returning 0 with no lock. Now it sets a flag if
4141 a lock is going to be returned.
4143 Severity : enhancement
4145 Description: Checksum enhancements
4146 Details : New checksum enhancements allow for resending RPCs that failed
4149 Severity : enhancement
4151 Description: Tunables on number of dirty pages in cacche
4152 Details : Allow to set limit on number of dirty pages cached.
4157 Description: client crash on unmount - lock still has references
4158 Details : In some error handling cases it was possible to leak a lock
4159 reference on a client while accessing a file. This was not
4160 harmful to the client during operation, but would cause the
4161 client to crash when the filesystem is unmounted.
4164 Frequency : specific case, rare
4166 Description: ETXTBSY on mds though file not in use by client
4167 Details : ETXTBSY is no longer incorrectly returned when attempting to
4168 chmod or chown a directory that the user previously tried to
4169 execute or a currently-executing binary.
4172 Frequency : extremely rare except on liblustre-based clients
4174 Description: Lustre space not freed when files are deleted
4175 Details : Clean up open-unlinked files after client eviction. Previously
4176 the unlink was skipped and the files remained as orphans.
4181 Description: OST failure "would be an LBUG" in waiting_locks_callback()
4182 Details : In some cases it was possible to send a blocking callback to
4183 a client doing a glimpse, even though that client didn't get
4184 a lock granted. When the glimpse lock is cancelled on the OST
4185 the freed lock is left on the waiting list and corrupted the list.
4188 Frequency : all core dumps
4190 Description: Broke core dumps to lustre
4191 Details : Negative dentry may be unhashed if parent does not have UPDATE
4192 lock, but some callers, e.g. do_coredump, expect dentry to be
4193 hashed after successful create, hash it in ll_create_it.
4195 ------------------------------------------------------------------------------
4197 2006-09-13 Cluster File Systems, Inc. <info@clusterfs.com>
4199 * Support for kernels:
4200 2.6.9-42.0.2.EL (RHEL 4)
4201 2.6.5-7.276 (SLES 9)
4202 2.4.21-40.EL (RHEL 3)
4203 2.6.12.6 vanilla (kernel.org)
4207 Frequency : always on RHEL 3
4209 Description: Number of open files grows over time
4210 Details : The number of open files grows over time, whether or not
4211 Lustre is started. This was due to a filp leak introduced
4212 by one of our kernel patches.
4214 ------------------------------------------------------------------------------
4216 08-20-2006 Cluster File Systems, Inc. <info@clusterfs.com>
4218 * Support for kernels:
4219 2.6.9-42.EL (RHEL 4)
4220 2.6.5-7.276 (SLES 9)
4221 2.4.21-40.EL (RHEL 3)
4222 2.6.12.6 vanilla (kernel.org)
4227 Bugzilla : 5719, 9635, 9792, 9684
4228 Description: OST (or MDS) trips assertions in (re)connection under heavy load
4229 Details : If a server is under heavy load and cannot reply to new
4230 connection requests before the client resends the (re)connect,
4231 the connection handling code can behave badly if two service
4232 threads are concurrently handing separate (re)connections from
4233 the same client. Add better locking to the connection handling
4234 code, and ensure that only a single connection will be processed
4235 for a given client UUID, even if the lock is dropped.
4237 Severity : enhancement
4239 Description: add TCP zero-copy support to kernel
4240 Details : Add support to the kernel TCP stack to allow zero-copy bulk
4241 sends if the hardware supports scatter-gather and checksumming.
4242 This allows socklnd to do client-write and server-read more
4243 efficiently and reduce CPU utilization from skbuf copying.
4246 Frequency : only if NFS exporting from client
4248 Description: NULL pointer deref in ll_iocontrol() if chattr mknod file
4249 Details : If setting attributes on a file created under NFS that had
4250 never been opened it would be possible to oops the client
4251 if the file had no objects.
4254 Frequency : always for liblustre
4256 Description: liblustre client does MDS+OSTs setattr RPC for each write
4257 Details : When doing a write from a liblustre client, the client
4258 incorrectly issued an RPC to the MDS and each OST the file was
4259 striped over in order to update the timestamps. When writing
4260 with small chunks and many clients this could overwhelm the MDS
4261 with RPCs. In all cases it would slow down the write because
4262 these RPCs are unnecessary.
4264 Severity : enhancement
4266 Description: allow number of MDS service threads to be changed at module load
4267 Details : It is now possible to change the number of MDS service threads
4268 running. Adding "options mds mds_num_threads={N}" to the MDS's
4269 /etc/modprobe.conf will set the number of threads for the next
4270 time Lustre is restarted (assuming the "mds" module is also
4271 reloaded at that time). The default number of threads will
4272 stay the same, 32 for most systems.
4277 Description: OST crash if filesystem is unformatted or corrupt
4278 Details : If an OST is started on a device that has never been formatted
4279 or if the filesystem is corrupt and cannot even mount then the
4280 error handling cleanup routines would dereference a NULL pointer.
4285 Description: NULL pointer deref in llap_from_page.
4286 Details : get_cache_page_nowait can return a page with NULL (or otherwise
4287 incorrect) mapping if the page was truncated/reclaimed while it was
4288 searched for. Check for this condition and skip such pages when
4289 doing readahead. Introduce extra check to llap_from_page() to
4290 verify page->mapping->host is non-NULL (so page is not anonymous).
4293 Frequency : Sometimes when using sys_sendfile
4295 Description: "page not covered by a lock" warnings from ll_readpage
4296 Details : sendfile called ll_readpage without right page locks present.
4297 Now we introduced ll_file_sendfile that does necessary locking
4298 around call to generic_file_sendfile() much like we do in
4302 Frequency : with certain MDS communication failures at client mount time
4304 Description: NULL pointer deref after failed client mount
4305 Details : a client connection request may delayed by the network layer
4306 and not be sent until after the PTLRPC layer has timed out the
4307 request. If the client fails the mount immediately it will try
4308 to clean up before the network times out the request. Add a
4309 reference from the request import to the obd device and delay
4310 the cleanup until the network drops the request.
4313 Frequency : occasionally during client (re)connect
4315 Description: assertion failure during client (re)connect
4316 Details : processing a client connection request may be delayed by the
4317 client or server longer than the client connect timeout. This
4318 causes the client to resend the connection request. If the
4319 original connection request is replied in this interval, the
4320 client may trip an assertion failure in ptlrpc_connect_interpret()
4321 which thought it would be the only running connect process.
4324 Frequency : only with obd_echo servers and clients that are rebooted
4326 Description: kernel BUG accessing uninitialized data structure
4327 Details : When running an obd_echo server it did not start the ping_evictor
4328 thread, and when a client was evicted an uninitialized data
4329 structure was accessed. Start the ping_evictor in the RPC
4330 service startup instead of the OBD startup.
4332 Severity : enhancement
4333 Bugzilla : 10193 (patchless)
4334 Description: Remove dependency on various unexported kernel interfaces.
4335 Details : No longer need reparent_to_init, exit_mm, exit_files,
4336 sock_getsockopt, filemap_populate, FMODE_EXEC, put_filp.
4339 Frequency : rare (only users of deprecated and unsupported LDAP config)
4341 Description: write_conf for zeroconf mount queried LDAP incorrectly for client
4342 Details : LDAP apparently contains 'lustreName' attributes instead of
4343 'name'. A simple remapping of the name is sufficient.
4346 Frequency : rare (only with non-default dump_on_timeout debug enabled)
4348 Description: waiting_locks_callback trips kernel BUG if client is evicted
4349 Details : Running with the dump_on_timeout debug flag turned on makes
4350 it possible that the waiting_locks_callback() can try to dump
4351 the Lustre kernel debug logs from an interrupt handler. Defer
4352 this log dumping to the expired_lock_main() thread.
4354 Severity : enhancement
4356 Description: Support NFS exporting on 2.6 kernels.
4357 Details : Implement non-rawops metadata methods for NFS server to use without
4358 changing NFS server code.
4361 Frequency : very rare (synthetic metadata workload only)
4363 Description: two racing renames might cause an MDS thread to deadlock
4364 Details : Running the "racer" program may cause one MDS thread to rename
4365 a file from being the source of a rename to being the target of
4366 a rename at exactly the same time that another thread is doing
4367 so, and the second thread has already enqueued these locks after
4368 doing a lookup of the target and is trying to relock them in
4369 order. Ensure that we don't try to re-lock the same resource.
4372 Frequency : only very large systems with liblustre clients
4374 Description: slow eviction of liblustre clients with the "evict_by_nid" RPC
4375 Details : Use asynchronous set_info RPCs to send the "evict_by_nid" to
4376 all OSTs in parallel. This allows the eviction of stale liblustre
4377 clients to proceed much faster than if they were done in series,
4378 and also offers similar improvements for other set_info RPCs.
4383 Description: excessive CPU usage during initial read phase on client
4384 Details : During the initial read phase on a client, it would agressively
4385 retry readahead on the file, consuming too much CPU and impacting
4386 performance (since 1.4.5.8). Improve the readahead algorithm
4387 to avoid this, and also improve some other common cases (read
4388 of small files in particular, where "small" is files smaller than
4389 /proc/fs/lustre/llite/*/max_read_ahead_whole_mb, 2MB by default).
4394 Description: MDS crash when receiving packet with unknown intent.
4395 Details : Do not LBUG in unknown intent case, just return -EFAULT
4397 Severity : enhancement
4398 Bugzilla : 9293, 9385
4399 Description: MDS RPCs are serialised on client. This is unnecessary for some.
4400 Details : Do not serialize getattr (non-intent version) and statfs.
4403 Frequency : occasional, when OST network is overloaded/intermittent
4405 Description: client evicted by OST after bulk IO timeout
4406 Details : If a client sends a bulk IO request (read or write) the OST
4407 may evict the client if it is unresposive to its data GET/PUT
4408 request. This is incorrect if the network is overloaded (takes
4409 too long to transfer the RPC data) or dropped the OST GET/PUT
4410 request. There is no need to evict the client at all, since
4411 the pinger and/or lock callbacks will handle this, and the
4412 client can restart the bulk request.
4415 Frequency : Always when mmapping file with no objects
4417 Description: client crashes when mmapping file with no objects
4418 Details : Check that we actually have objects in a file before doing any
4419 operations on objects in ll_vm_open, ll_vm_close and
4425 Description: Request leak when working with deleted CWD
4426 Details : Introduce advanced request refcount tracking for requests
4427 referenced from lustre intent.
4429 Severity : Enhancement
4431 Description: Cache open file handles on client.
4432 Details : MDS now will return special lock along with openhandle, if
4433 requested and client is allowed to hold openhandle, even if unused,
4434 until such a lock is revoked. Helps NFS a lot, since NFS is opening
4435 closing files for every read/write openration.
4437 Severity : Enhancement
4439 Description: Cache open negative dentries on client when possible.
4440 Details : Guard negative dentries with UPDATE lock on parent dir, drop
4441 negative dentries on lock revocation.
4446 Description: Remounting a client read-only wasn't possible with a zconf mount
4447 Details : It wasn't possible to remount a client read-only with llmount.
4449 Severity : enhancement
4450 Description: Include MPICH 1.2.6 Lustre ADIO interface patch
4451 Details : In lustre/contrib/ or /usr/share/lustre in RPM a patch for
4452 MPICH is included to add Lustre-specific ADIO interfaces.
4453 This is based closely on the UFS ADIO layer and only differs
4454 in file creation, in order to allow the OST striping to be set.
4455 This is user-contributed code and not supported by CFS.
4460 Description: extended inode attributes (immutable, append-only) work improperly
4461 when 2.4 and 2.6 kernels are used on client/server or vice versa
4462 Details : Introduce kernel-independent values for these flags.
4464 Severity : enhancement
4467 Description: Allow fractional MB tunings for lustre in /proc/ filesystem.
4468 Details : Many of the /proc/ tunables can only be tuned at a megabyte
4469 granularity. Now, Fractional MB granularity is be supported,
4470 this is very useful for low memory system.
4472 Severity : enhancement
4474 Description: Getattr by fid
4475 Details : Getting a file attributes by its fid, obtaining UPDATE|LOOKUP
4476 locks, avoids extra getattr rpc requests to MDS, allows '/' to
4477 have locks and avoids getattr rpc requests for it on every stat.
4480 Frequency : Always, for filesystems larger than 2TB
4482 Description: ldiskfs crash at mount for filesystem larger than 2TB with mballoc
4483 Details : Kenrel kmalloc limits allocations to 128kB and this prevents
4484 filesystems larger than 2TB to be mounted with mballoc enabled.
4487 Frequency : Always, for 32-bit kernel without CONFIG_LBD and filesystem > 2TB
4489 Description: ldiskfs crash at mount for filesystem larger than 2TB with mballoc
4490 Details : If a 32-bit kernel is compiled without CONFIG_LBD enabled and a
4491 filesystems larger than 2TB is mounted then the kernel will
4492 silently corrupt the start of the filesystem. CONFIG_LBD is
4493 enabled for all CFS-supported kernels, but the possibility of
4494 this happening with a modified kernel config exists.
4496 Severity : enhancement
4498 Description: add client O_DIRECT support for 2.6 kernels
4499 Details : It is now possible to do O_DIRECT reads and writes to files
4500 in the Lustre client mountpoint on 2.6 kernel clients.
4502 Severity : enhancement
4504 Description: parallel glimpse, setattr, statfs, punch, destroy requests
4505 Details : Sends glimpse, setattr, statfs, punch, destroy requests to OSTs in
4506 parallel, not waiting for response from every OST before sending
4507 a rpc to the next OST.
4512 Description: setattr vs write race when updating file timestamps
4513 Details : Client processes that update a file timestamp into the past
4514 right after writing to the file (e.g. tar) it is possible that
4515 the updated file modification time can be reset to the current
4516 time due to a race between processing the setattr and write RPC.
4518 Severity : enhancement
4520 Description: Bring 'lfs find' closer in line with regular Linux find.
4521 Details : lfs find util supports -atime, -mtime, -ctime, -maxdepth, -print,
4522 -print0 options and obtains all the needed info through the lustre
4525 Severity : enhancement
4527 Description: support up to 1024 configured devices on one node
4528 Details : change obd_dev array from statically allocated to dynamically
4529 allocated structs as they are first used to reduce memory usage
4534 Description: Flush dirty partially truncated pages during truncate
4535 Details : Immediatelly flush partially truncated pages in filter_setattr,
4536 this way we completely avoid having any pages in page cache on OST
4537 and can retire ugly workarounds during writes to flush such pages.
4542 Description: i_sem vs transaction deadlock in mds_obd_destroy during unlink.
4543 Details : protect inode from truncation within vfs_unlink() context
4544 just take a reference before calling vfs_unlink() and release it
4545 when parent's i_sem is free.
4550 Description: last_id value checked outside lock on OST caused LASSERT failure
4551 Details : If there were multiple MDS->OST object precreate requests in
4552 flight, it was possible that the OST's last object id was checked
4553 outside a lock and incorrectly tripped an assertion. Move checks
4554 inside locks, and discard old precreate requests.
4557 Frequency : always, if extents are used on OSTs
4559 Description: index ei_leaf_hi (48-bit extension) is not zeroed in extent index
4560 Details : OSTs using the extents format would not zero the high 16 bits of
4561 the index physical block number. This is not a problem for any
4562 OST filesystems smaller than 16TB, and no kernels support ext3
4563 filesystems larger than 16TB yet. This is fixed in 1.4.7 (all
4564 new/modified files) and can be fixed for existing filesystems
4565 with e2fsprogs-1.39-cfs1.
4570 Description: import connection selection may be incorrect if timer wraps
4571 Details : Using a 32-bit jiffies timer with HZ=1000 may cause backup
4572 import connections to be ignored if the 32-bit jiffies counter
4573 wraps. Use a 64-bit jiffies counter.
4576 Frequency : very large clusters immediately after boot
4578 Description: LNET request buffers exhausted under heavy short-term load
4579 Details : If a large number of client requests are generated on a service
4580 that has previously never seen so many requests it is possible
4581 that the request buffer growth cannot keep up with the spike in
4582 demand. Instead of dropping incoming requests, they are held in
4583 the LND until the RPC service can accept more requests.
4586 Frequency : Sometimes during replay
4588 Description: Assertion failure in ll_local_open after replay.
4589 Details : If replay happened on an open request reply before we were able
4590 to set replay handler, reply will become not swabbed tripping the
4591 assertion in ll_local_open. Now we set the handler right after
4592 recognising of open request
4595 Frequency : very rare
4597 Description: kernel reports "badness in vsnprintf"
4598 Details : Reading from the "recovery_status" /proc file in small chunks
4599 may cause a negative length in lprocfs_obd_rd_recovery_status()
4600 call to vsnprintf() (which is otherwise harmless). Exit early
4601 if there is no more space in the output buffer.
4603 Severity : enhancement
4605 Description: clear OBD RPC statistics by writing to them
4606 Details : It is now possible to clear the OBD RPC statistics by writing
4607 to the "stats" file.
4612 Description: Client mtime is not the same on different clients after utimes
4613 Details : In some cases, the client was using the utimes() syscall on
4614 a file cached on another node. The clients now validate the
4615 ctime from the MDS + OSTs to determine which one is right.
4620 Description: Inability to activate failout mode
4621 Details : lconf script incorrectly assumed that in pythong string's numeric
4622 value is used in comparisons.
4625 Frequency : always with multiple stripes per file
4627 Description: Inefficient object allocation for mutli-stripe files
4628 Details : When selecting which OSTs to stripe files over, for files with
4629 a stripe count that divides evenly into the number of OSTs,
4630 the MDS is always picking the same starting OST for each file.
4631 Return the OST selection heuristic to the original design.
4636 Description: mount failures may take full timeout to return an error
4637 Details : Under some heavy load conditions it is possible that a
4638 failed mount can wait for the full obd_timeout interval,
4639 possibly several minutes, before reporting an error.
4640 Instead return an error as soon as the status is known.
4642 Frequency : quota enabled and large files being deleted
4644 Description: releasing more than 4GB of quota at once hangs OST
4645 Details : If a user deletes more than 4GB of files on a single OST it
4646 will cause the OST to spin in an infinite loop. Release
4647 quota in < 4GB chunks, or use a 64-bit value for 1.4.7.1+.
4652 Description: statfs data retrieved from /proc may be stale or zero
4653 Details : When reading per-device statfs data from /proc, in the
4654 {kbytes,files}_{total,free,avail} files, it may appear
4655 as zero or be out of date.
4658 Frequency : systems with MD RAID1 external journal devices
4660 Description: lconf's call to blkid is confused by RAID1 journal devices
4661 Details : Use the "blkid -l" flag to locate the MD RAID device instead
4662 of returning all block devices that match the journal UUID.
4665 Frequency : always, for aggregate stripe size over 4GB
4667 Description: assertion fails when trying to use 4GB stripe size
4668 Details : Use "setstripe" to set stripe size over 4GB will fail the kernel,
4669 complaining "ASSERTION(lsm->lsm_xfersize != 0)"
4672 Frequency : always on ppc64
4674 Description: the first write on an ext3 filesystem with mballoc got stuck
4675 Details : ext3_mb_generate_buddy() uses find_next_bit() which does not
4676 perform endianness conversion.
4678 ------------------------------------------------------------------------------
4680 02-14-2006 Cluster File Systems, Inc. <info@clusterfs.com>
4682 * WIRE PROTOCOL CHANGE. This version of Lustre networking WILL NOT
4683 INTEROPERATE with older versions automatically. Please read the
4684 user documentation before upgrading any part of a live system.
4685 * WARNING: Lustre networking configuration changes are required with
4686 this release. See https://bugzilla.clusterfs.com/show_bug.cgi?id=10052
4689 * Support for kernels:
4690 2.6.9-22.0.2.EL (RHEL 4)
4691 2.6.5-7.244 (SLES 9)
4692 2.6.12.6 vanilla (kernel.org)
4695 Severity : enhancement
4696 Bugzilla : 7981/8208
4697 Description: Introduced Lustre Networking (LNET)
4698 Details : LNET is new networking infrastructure for Lustre, it includes
4699 a reorganized network configuration mode (see the user
4700 documentation for full details) as well as support for routing
4701 between different network fabrics. Lustre Networking Devices
4702 (LNDs) for the supported network fabrics have also been
4703 created for this new infrastructure.
4705 Severity : enhancement
4706 Description: Introduced Access control lists
4707 Details : clients can set ACLs on files and directories in order to have
4708 more fine-grained permissions than the standard Unix UGO+RWX.
4709 The MDS must be started with the "-o acl" mount option.
4711 Severity : enhancement
4712 Description: Introduced filesystem quotas
4713 Details : Administrators may now establish per-user quotas on the
4716 Severity : enhancement
4718 Description: Configuration change for the XT3
4719 The PTLLND is now used to run Lustre over Portals on the XT3
4720 The configure option(s) --with-cray-portals are no longer used.
4721 Rather --with-portals=<path-to-portals-includes> is used to
4722 enable building on the XT3. In addition to enable XT3 specific
4723 features the option --enable-cray-xt3 must be used.
4728 Description: Running on many-way SMP OSTs can trigger oops in llcd_send()
4729 Details : A race between allocating a new llcd and re-getting the llcd_lock
4730 allowed another thread to grab newly-allocated llcd.
4732 Severity : enhancement
4734 Description: 2.6 OST async journal commit and locking fix to improve performance
4735 Details : The filter_direct_io()+filter_commitrw_write() journal commits for
4736 2.6 kernels are now async as they already were in 2.4 kernels so
4737 that they can commit concurrently with the network bulk transfer.
4738 For block-allocated files the filter allocation semaphore is held
4739 to avoid filesystem fragmentation during allocation. BKL lock
4740 removed for 2.6 xattr operations where it is no longer needed.
4745 Description: lconf incorrectly determined whether two IP networks could talk
4746 Details : In some more complicated routing and multiple-network
4747 configurations, lconf will avoid trying to make a network
4748 connection to a disjoint part of the IP space. It was doing the
4749 math incorrectly for one set of cases.
4754 Description: Fix for potential infinite loop processing records in an llog.
4755 Details : If an llog record is corrupted/zeroed, it is possible to loop
4756 forever in llog_process(). Validate the llog record length
4757 and skip the remainder of the block on error.
4760 Frequency : occasional (liblustre only)
4762 Description: liblustre could not open files whose last component is a symlink
4763 Details : sysio_path_walk() would incorrectly pass the open intent to
4764 intermediate path components.
4767 Frequency : rare (liblustre only with non-standard tuning)
4768 Bugzilla : 7201 (7350)
4769 Description: Tuning the MDC DLM LRU size to zero triggers client LASSERT
4770 Details : llu_lookup_finish_locks() tries to set lock data on a lock
4771 after it has been released, only do this for referenced locks
4773 Severity : enhancement
4775 Description: specifying an (invalid) directory default stripe_size of -1
4776 would reset the directory default striping
4777 Details : stripe_size -1 was used internally to signal directory stripe
4778 removal, now use "all default" to signal dir stripe removal
4779 as a directory striping of "all default" is not useful
4782 Frequency : common for large clusters running liblustre clients
4784 Description: doing an ls when liblustre clients are running is slow
4785 Details : sending a glimpse AST to a liblustre client waits for every AST
4786 to time out, as liblustre clients will not respond. Since they
4787 cannot cache data we refresh the OST lock LVB from disk instead.
4789 Severity : enhancement
4791 Description: doing an ls at the same time as file IO can be slow
4792 Details : enqueue and other "small" requests can be blocked behind many
4793 large IO requests. Create a new OST IO portal for non-IO
4794 requests so they can be processed faster.
4797 Frequency : rare (only HPUX clients mounting unsupported re-exported NFS vol)
4799 Description: an HPUX NFS client would get -EACCESS when ftruncate()ing a newly
4800 created file with mode 000
4801 Details : the Linux NFS server relies on an MDS_OPEN_OWNEROVERRIDE hack to
4802 allow an ftruncate() as a non-root user to a file with mode 000.
4803 Lustre now respects this flag to disable mode checks when
4804 truncating a file owned by the user
4807 Frequency : liblustre-only, when liblustre client dies unexpectedly or becomes
4810 Description: Revoking locks from clients that went dead or catatonic might take
4812 Details : New lock flags FL_CANCEL_ON_BLOCK used by liblustre makes
4813 cancellation of such locks instant on servers without waiting for
4814 any reply from clients. Clients drops these locks when cancel
4815 notification from server is received without replying.
4818 Frequency : liblustre-only, when liblustre client dies or becomes busy
4820 Description: Doing ls on Linux clients can take a long time with active
4822 Details : Liblustre client cannot handle ASTs in timely manner, so avoid
4823 granting such locks to it in the first place if possible. Locks
4824 are taken by proxy on the OST during the read or write and
4825 dropped immediately afterward. Add connect flags handling, do
4826 not grant locks to liblustre clients for glimpse ASTs.
4828 Severity : enhancement
4830 Description: Improve read-ahead algorithm to avoid excessive IO for random reads
4831 Details : Existing read-ahead algorithm is tuned for the case of streamlined
4832 sequential reads and behaves badly with applications doing random
4833 reads. Improve it by reading ahead at least read region, and
4834 avoiding excessive large RPC for small reads.
4836 Severity : enhancement
4838 Description: Creating more than 1000 files for a single job may cause a load
4839 imbalance on the OSTs if there are also a large number of OSTs.
4840 Details : qos_prep_create() uses an OST index reseed value that is an
4841 even multiple of the number of available OSTs so that if the
4842 reseed happens in the middle of the object allocation it will
4843 still utilize the OSTs as uniformly as possible.
4848 Description: OST or MDS may oops in ping_evictor_main()
4849 Details : ping_evictor_main() drops obd_dev_lock if deleting a stale export
4850 but doesn't restart at beginning of obd_exports_timed list
4853 Severity : enhancement
4855 Description: improve by-nid export eviction on the MDS and OST
4856 Details : allow multiple exports with the same NID to be evicted at one
4857 time without re-searching the exports list.
4860 Frequency : rare, only with supplementary groups enabled on SMP 2.6 kernels
4862 Description: MDS may oops in groups_free()
4863 Details : in rare race conditions a newly allocated group_info struct is
4864 freed again, and this can be NULL. The 2.4 compatibility code
4865 for groups_free() checked for a NULL pointer, but 2.6 did not.
4868 Frequency : common for liblustre clients doing little filesystem IO
4869 Bugzilla : 9352, 7313
4870 Description: server may evict liblustre clients accessing contended locks
4871 Details : if a client is granted a lock or receives a completion AST
4872 with a blocking AST already set it would not reply to the AST
4873 for LDLM_FL_CANCEL_ON_BLOCK locks. It now replies to such ASTs.
4876 Frequency : lfs setstripe, only systems with more than 160 OSTs
4878 Description: unable to set striping with a starting offset beyond OST 160
4879 Details : llapi_create_file() incorrectly limited the starting stripe
4880 index to the maximum single-file stripe count.
4883 Frequency : LDAP users only
4885 Description: lconf did not handle in-kernel recovery with LDAP properly
4886 Details : lconf/LustreDB get_refs() is searching the wrong namespace
4888 Severity : enhancement
4890 Description: bind OST threads to NUMA nodes to improve performance
4891 Details : all OST threads are uniformly bound to CPUs on a single NUMA
4892 node and do their allocations there to localize memory access
4894 Severity : enhancement
4896 Description: llmount can determine client NID directly from Myrinet (GM)
4897 Details : the client NID code from gmnalnid was moved directly into
4898 llmount, removing the need to use this or specifying the
4899 client NID explicitly when mounting GM clients with zeroconf
4902 Frequency : if client is started with down MDS
4904 Description: if client is started with down MDS mount hangs in ptlrpc_queue_wait
4905 Details : Having an LWI_INTR() wait event (interruptible, but no timeout)
4906 will wait indefinitely in ptlrpc_queue_wait->l_wait_event() after
4907 ptlrpc_import_delayed_req() because we didn't check if the
4908 request was interrupted, and we also didn't break out of the
4909 event loop if there was no timeout
4914 Description: data loss during non-page-aligned writes to a single file from
4915 both multiple nodes and multiple threads on one node at same time
4916 Details : updates to KMS and lsm weren't protected by common lock. Resulting
4917 inconsistency led to false short-reads, that were cached and later
4918 used by ->prepare_write() to fill in partially written page,
4919 leading to data loss.
4922 Frequency : always, if lconf --abort_recovery used
4924 Description: lconf --abort_recovery fails with 'Operation not supported'
4925 Details : lconf was attempting to abort recovery on the MDT device and not
4928 Severity : enhancement
4930 Description: remove cleanup logs
4931 Details : replace lconf-generated cleanup logs with lustre internal
4932 cleanup routines. Eliminates the need for client-cleanup and
4935 Severity : enhancement
4937 Description: add support for EAs (user and system) on lustre filesystems
4938 Details : it is now possible to store extended attributes in the Lustre
4939 client filesystem, and with the user_xattr mount option it
4940 is possible to allow users to store EAs on their files also
4942 Severity : enhancement
4944 Description: Add possibility (config option) to show minimal available OST free
4946 Details : When compiled with --enable-mindf configure option, statfs(2)
4947 (and so, df) will return least minimal free space available from
4948 all OSTs as amount of free space on FS, instead of summary of
4949 free spaces of all OSTs.
4951 Severity : enhancement
4953 Description: do not expand extent locks acquired on OST-side
4954 Details : Modify ldlm_extent_policy() to not expand local locks, acquired
4955 by server: they are not cached anyway.
4958 Frequency : when mmap is used/binaries executed from Lustre
4960 Description: Unmmap pages before throwing them away from read cache.
4961 Details : llap_shrink cache now attempts to unmap pages before discarding
4962 them (if unmapping failed - do not discard). SLES9 kernel has
4963 extra checks that trigger if this unmapping is not done first.
4968 Description: lconf didn't resolve symlinks before checking to see whether a
4969 given mountpoint was already in use
4972 Frequency : when migrating failover services
4973 Bugzilla : 6395, 9514
4974 Description: When migrating a subset of services from a node (e.g. failback
4975 from a failover service node) the remaining services would
4976 time out and evict clients.
4977 Details : lconf --force (implied by --failover) sets the global obd_timeout
4978 to 5 seconds in order to quickly disconnect, but this caused
4979 other RPCs to time out too quickly. Do not change the global
4980 obd_timeout for force cleanup, only set it for DISCONNECT RPCs.
4982 Severity : enhancement
4983 Frequency : if MDS is started with down OST
4984 Bugzilla : 9439,5706
4985 Description: Allow startup/shutdown of an MDS without depending on the
4986 availability of the OSTs.
4987 Details : Asynchronously call mds_lov_synchronize during MDS startup.
4988 Add appropriate locking and lov-osc refcounts for safe
4989 cleaning. Add osc abort_inflight calls in case the
4990 synchronize never started.
4993 Frequency : occasional (Cray XT3 only)
4995 Description: root not authorized to access files in CRAY_PORTALS environment
4996 Details : The client process capabilities were not honoured on the MDS in
4997 a CRAY_PORTALS/CRAY_XT3 environment. If the file had previously
4998 been accessed by an authorized user then root was able to access
4999 the file on the local client also. The root user capabilities
5000 are now allowed on the MDS, as this environment has secure UID.
5003 Frequency : occasional
5005 Description: ldiskfs "too long searching" message happens too often
5006 Details : A debugging message (otherwise harmless) prints too often on
5007 the OST console. This has been reduced to only happen when
5008 there are fragmentation problems on the filesystem.
5013 Description: Division by zero in statfs when all OSCs are inactive
5014 Details : lov_get_stripecnt() returns zero due to incorrect order of checks,
5015 lov_statfs divides by value returned by lov_get_stripecnt().
5019 Bugzilla : 9489, 3273
5020 Description: First write from each client to each OST was only 4kB in size,
5021 to initialize client writeback cache, which caused sub-optimal
5022 RPCs and poor layout on disk for the first written file.
5023 Details : Clients now request an initial cache grant at (re)connect time
5024 and so that they can start streaming writes to the cache right
5025 away and always do full-sized RPCs if there is enough data.
5026 If the OST is rebooted the client also re-establishes its grant
5027 so that client cached writes will be honoured under the grant.
5032 Description: Slow ls (and stat(2) syscall) on files residing on IO-loaded OSTs
5033 Details : Now I/O RPCs go to different portal number and (presumably) fast
5034 lock requests (and glimses) and other RPCs get their own service
5035 threads pool that should be able to service those RPCs
5038 Severity : enhancement
5040 Description: Ability to exchange lustre version between client and servers and
5041 issue warnings at client side if client is too old. Also for
5042 liblustre clients there is ability to refuse connection of too old
5044 Details : New 'version' field is added to connect data structure that is
5045 filled with version info. That info is later checked by server and
5049 Frequency : rare, liblustre only.
5050 Bugzilla : 9296, 9581
5051 Description: Two simultaneous writes from liblustre at offset within same page
5052 might proceed at the same time overwriting eachother with stale
5054 Details : I/O lock withing llu_file_prwv was released too early, before data
5055 actually was hitting the wire. Extended lock-holding time until
5056 server acknowledges receiving data.
5059 Frequency : extremely rare. Never observed in practice.
5061 Description: avoid generating lustre_handle cookie of 0.
5062 Details : class_handle_hash() generates handle cookies by incrementing
5063 global counter, and can hit 0 occasionaly (this is unlikely, but
5064 not impossible, because initial value of cookie counter is
5065 selected randonly). Value of 0 is used as a sentinel meaning
5066 "unassigned handle" --- avoid it. Also coalesce two critical
5067 sections in this function into one.
5069 Severity : enhancement
5071 Description: allow liblustre clients to delegate truncate locking to OST
5072 Details : To avoid overhead of locking, liblustre client instructs OST to
5073 take extent lock in ost_punch() on client's behalf. New connection
5074 flag is added to handle backward compatibility.
5076 Severity : enhancement
5077 Bugzilla : 4928, 7341, 9758
5078 Description: allow number of OST service threads to be specified
5079 Details : a module parameter allows the number of OST service threads
5080 to be specified via "options ost ost_num_threads={N}" in the
5081 OSS's /etc/modules.conf or /etc/modprobe.conf.
5085 Bugzilla : 6146, 9635, 9895
5086 Description: servers crash with bad pointer in target_handle_connect()
5087 Details : In rare cases when a client is reconnecting it was possible that
5088 the connection request was the last reference for that export.
5089 We would temporarily drop the export reference and get a new
5090 one, but this may have been the last reference and the export
5091 was just destroyed. Get new reference before dropping old one.
5093 Severity : enhancement
5094 Frequency : if client is started with failover MDS
5096 Description: Allow multiple MDS hostnames in the mount command
5097 Details : Try to read the configuration from all specified MDS
5098 hostnames during a client mount in case the "primary"
5101 Severity : enhancement
5103 Description: Stop sending data to evicted clients as soon as possible.
5104 Details : Check if the client we are about to send or are sending data to
5105 was evicted already. (Check is done every second of waiting,
5106 for which l_wait_event interface was extended to allow checking
5107 of exit condition at specified intervals).
5110 Frequency : rare, normally only when NFS exporting is done from client
5112 Description: 'bad disk LOV MAGIC: 0x00000000' error when chown'ing files
5114 Details : Make mds_get_md() recognise empty md case and set lmm size to 0.
5117 Frequency : always, if srand() is called before liblustre initialization
5119 Description: Liblustre uses system PRNG disturbing its usage by user application
5120 Details : Introduce internal to lustre fast and high-quality PRNG for
5121 lustre usage and make liblustre and some other places in generic
5122 lustre code to use it.
5124 Severity : enhancement
5125 Bugzilla : 9477, 9557, 9870
5126 Description: Verify that the MDS configuration logs are updated when xml is
5127 Details : Check if the .xml configuration logs are newer than the config
5128 logs stored on the MDS and report an error if this is the case.
5129 Request --write-conf, or allow starting with --old_conf.
5131 Severity : enhancement
5133 Description: Handle symlinks in the path when checking if Lustre is mounted.
5134 Details : Resolve intermediate symlinks when checking if a client has
5135 mounted a filesystem to avoid duplicate client mounts.
5140 Description: lconf can hit an error exception but still return success.
5141 Details : The lconf command catches the Command error exception at the top
5142 level script context and will exit with the associated exit
5143 status, but doesn't ensure that this exit status is non-zero.
5148 Description: failure of ptlrpc thread startup can cause oops
5149 Details : Starting a ptlrpc service thread can fail if there are a large
5150 number of threads or the server memory is very fragmented.
5151 Handle this without oopsing.
5154 Frequency : always, only if liblustre and non-default acceptor port was used
5156 Description: liblustre cannot connect to servers with non-default acceptor port
5157 Details : tcpnal_set_default_params() was not called and was therefore
5158 ignoring the environment varaible TCPNAL_PORT, as well as other
5159 TCPNAL_ environment variables
5164 Description: two objects could be created on the same OST for a single file
5165 Details : If an OST is down, in some cases it was possible to create two
5166 objects on a single OST for a single file. No problems other
5167 than potential performance impact and spurious error messages.
5171 Bugzilla : 5681, 9562
5172 Description: Client may oops in ll_unhash_aliases
5173 Details : Client dcache may become inconsistent in race condition.
5174 In some cases "getcwd" can fail if the current directory is
5180 Description: Inode refcounting problems in NFS export code
5181 Details : link_raw functions used to call d_instantiate without obtaining
5182 extra inode reference first.
5186 Bugzilla : 9942, 9903
5187 Description: Referencing freed requests leading to crash, memleaks with NFS.
5188 Details : We used to require that call to ll_revalidate_it was always
5189 followed by ll_lookup_it. Also with revalidate_special() it is
5190 possible to call ll_revalidate_it() twice for the same dentry
5191 even if first occurence returned success. This fix changes semantic
5192 between DISP_ENQ_COMPLETE disposition flag to mean there is extra
5193 reference on a request referred from the intent.
5194 ll_intent_release() then releases such a request.
5197 Frequency : rare, normally benchmark loads only
5199 Description: unlinked inodes were kept in memory on the client
5200 Details : If a client is repeatedly creating and unlinking files it
5201 can accumulate a lot of stale inodes in the inode slab cache.
5202 If there is no other client load running this can cause the
5203 client node to run out of memory. Instead flush old inodes
5204 from client cache that have the same inode number as a new inode.
5207 Frequency : SLES9 2.6.5 kernel and long filenames only
5208 Bugzilla : 9969, 10379
5209 Description: utime reports stale NFS file handle
5210 Details : SLES9 uses out-of-dentry names in some cases, which confused
5211 the lustre dentry revalidation. Change it to always use the
5215 Frequency : rare, unless heavy write-truncate concurrency is continuous
5216 Bugzilla : 4180, 6984, 7171, 9963, 9331
5217 Description: OST becomes very slow and/or deadlocked during object unlink
5218 Details : filter_destroy() was holding onto the parent directory lock
5219 while truncating+unlinking objects. For very large objects this
5220 may block other threads for a long time and slow overall OST
5221 responsiveness. It may also be possible to get a lock ordering
5222 deadlock in this case, or run out of journal credits because of
5223 the combined truncate+unlink. Solution is to do object truncate
5224 first in one transaction without parent lock, and then do the
5225 final unlink in a new transaction with the parent lock. This
5226 reduces the lock hold time dramatically.
5229 Frequency : rare, 2.4 kernels only
5231 Description: MDS or OST cleanup may trip kernel BUG when dropping kernel lock
5232 Details : mds_cleanup() and filter_cleanup() need to drop the kernel lock
5233 before unmounting their filesystem in order to avoid deadlock.
5234 The kernel_locked() function in 2.4 kernels only checks whether
5235 the kernel lock is held, not whether it is this process that is
5236 holding it as 2.6 kernels do.
5241 Description: MDS or OST may oops/LBUG if a client is connecting multiple times
5242 Details : The client ptlrpc code may be trying to reconnect to a down
5243 server before a previous connection attempt has timed out.
5244 Increase the reconnect interval to be longer than the connection
5245 timeout interval to avoid sending duplicate connections to
5249 Frequency : echo_client brw_test command
5251 Description: fix echo_client to work with OST preallocated code
5252 Details : OST preallocation code (5137) didn't take echo_client IO path
5253 into account: echo_client calls filter methods outside of any
5254 OST thread and, hence, there is no per-thread preallocated
5255 pages and buffers to use. Solution: hijack pga pages for IO. As
5256 a byproduct, this avoids unnecessary data copying.
5260 Bugzilla : 3555, 5962, 6025, 6155, 6296, 9574
5261 Description: Client can oops in mdc_commit_close() after open replay
5262 Details : It was possible for the MDS to return an open request with no
5263 transaction number in mds_finish_transno() if the client was
5264 evicted, but without actually returning an error. Clients
5265 would later try to replay that open and may trip an assertion
5266 Simplify the client close codepath, and always return an error
5267 from the MDS in case the open is not successful.
5270 Frequency : rare, 2.6 OSTs only
5272 Description: OST may deadlock under high load on fragmented files
5273 Details : If there was a heavy load and highly-fragmented OST filesystems
5274 it was possible to have all the OST threads deadlock waiting on
5275 allocation of biovecs, because the biovecs were not released
5276 until the entire RPC IO was completed. Instead, release biovecs
5277 as soon as they are complete to ensure forward IO progress.
5279 Severity : enhancement
5281 Description: Support for specifying external journal device at mount
5282 Details : If an OST or MDS device is formatted with an external journal
5283 device, this device major/minor is stored in the ext3 superblock
5284 and may not be valid for failover. Allow detecting and
5285 specifying the external journal at mount time.
5290 Description: Mounting an MDS with pending unlinked files may cause oops
5291 Details : target_finish_recovery() calls mds_postrecov() which returned
5292 the number of orphans unlinked. mds_lov_connect->mds_postsetup()
5293 considers this an error and immediately begins cleaning up the
5294 lov, just after starting the mds_lov process
5296 Severity : enhancement
5298 Description: Implement 'lfs df' to report actual free space on per-OST basis
5299 Details : Add sub-command 'df' on 'lfs' to report the disk space usage of
5300 MDS/OSDs. Usage: lfs df [-i][-h]. Command Options: '-i' to report
5301 usage of objects; '-h' to report in human readable format.
5303 ------------------------------------------------------------------------------
5305 08-26-2005 Cluster File Systems, Inc. <info@clusterfs.com>
5312 Description: Mounting an ldiskfs file system with mballoc may crash OST node.
5313 Details : ldiskfs mballoc code may reference an uninitialized buddy struct
5314 at startup during orphan unlinking. Instead, skip buddy update
5315 before setup, as it will be regenerated after recovery is complete.
5320 Description: If an OST is inactive, its locks might reference stale inodes.
5321 Details : lov_change_cbdata() must iterate over all namespaces, even if
5322 they are inactive to clear inode references from the lock.
5324 Severity : enhancement
5325 Frequency : occasional, if non-standard max_dirty_mb used
5327 Description: Client will block write RPCs if not enough grant
5328 Details : If a client has max_dirty_mb smaller than max_rpcs_in_flight,
5329 then the client will block writes while waiting for another RPC
5330 to complete instead of consuming its dirty limit. With change
5331 we get improved performance when max_dirty_mb is small.
5333 Severity : enhancement
5334 Bugzilla : 3389, 6253
5335 Description: Add support for supplementary groups on the MDS.
5336 Details : The MDS has an upcall /proc/fs/lustre/mds/{mds}/group_upcall
5337 (set to /usr/sbin/l_getgroups if enabled) which will do MDS-side
5338 lookups for user supplementary groups into a cache.
5342 Description: O_CREAT|O_EXCL open flags in liblustre always return -EEXIST
5343 Details : Make libsysio to not enforce O_EXCL by clearing the flag,
5344 for liblustre O_EXCL is enforced by MDS.
5348 Description: readdir never returns NULL in liblustre.
5349 Details : Corrected llu_iop_getdirentries logic, to return offset of next
5350 dentry in struct dirent.
5354 Frequency : liblustre only, depends on application IO pattern
5355 Description: liblustre clients evicted if not contacting servers
5356 Details : Don't put liblustre clients into the ping_evictor list, so
5357 they will not be evicted by the pinger ever.
5359 Severity : enhancement
5361 Description: Add ability to evict clients by NID from MDS.
5362 Details : By echoing "nid:$NID" string into
5363 /proc/fs/lustre/mds/.../evict_client client with nid that equals to
5364 $NID would be instantly evicted from this MDS and from all active
5365 OSTs connected to it.
5369 Description: Do not query file size twice, somewhat slowing stat(2) calls.
5370 Details : lookup_it_finish() used to query file size from OSTs that was not
5375 Description: service threads change working directory to that of init
5376 Details : Starting lustre service threads may pin the working directory
5377 of the parent thread, making that filesystem busy. Threads
5378 now change to the working directory of init to avoid this.
5382 Frequency : during shutdown only
5383 Description: shutdown with a failed MDS or OST can cause unmount to hang
5384 Details : Don't resend DISCONNECT messages in ptlrpc_disconnect_import()
5389 Frequency : 2.6 only
5390 Description: chmod/chown may include an extra supplementary group
5391 Details : ll{,u}_mdc_pack_op_data() does not properly initialize the
5392 supplementary group and if none is specified this is used.
5395 Bugzilla : 5479 (6816)
5397 Description: Racing open + rm can assert client in mdc_set_open_replay_data()
5398 Details : If lookup is in progress on a file that is unlinked we might try
5399 to revalidate the inode and fail in revalidate after lookup is
5400 complete and ll_file_open() enqueues the open again but
5401 it_open_error() was not checking DISP_OPEN_OPEN errors correctly.
5404 Frequency : always, if lconf --abort_recovery used
5406 Description: lconf --abort_recovery fails with 'Operation not supported'
5407 Details : lconf was attempting to abort recovery on the MDT device and not
5410 ------------------------------------------------------------------------------
5412 2005-08-08 Cluster File Systems, Inc. <info@clusterfs.com>
5417 Frequency : rare (only unsupported configurations with a node running as an
5419 Bugzilla : 6514, 5137
5420 Description: Mounting a Lustre file system on a node running as an OST could
5422 Details : OSTs now preallocates memory needed to write out data at
5423 startup, instead of when needed, to avoid having to
5424 allocate memory in possibly low memory situations.
5425 Specifically, if the file system is mounted on on OST,
5426 memory pressure could force it to try to write out data,
5427 which it needed to allocate memory to do. Due to the low
5428 memory, it would be unable to do so and the node would
5429 become unresponsive.
5431 Severity : enhancement
5433 Description: Addition of lconf --service command line option
5434 Details : lconf now accepts a '--service <arg>' option, which is
5435 shorthand for 'lconf --group <arg> --select <arg>=<hostname>'
5437 Severity : enhancement
5439 Description: Failover mode is now the default for OSTs.
5440 Details : By default, OSTs will now run in failover mode. To return to
5441 the old behaviour, add '--failout' to the lmc line for OSTs.
5443 Severity : enhancement
5445 Description: Health checks are now provided for MDS and OSTs
5446 Details : Additional detailed health check information on MSD and OSTs
5447 is now provided through the procfs health_check value.
5450 Frequency : occasional, depends on IO load
5452 Description: Disk fragmentation on the OSTs could eventually cause slowdowns
5453 after numerous create/delete cycles
5454 Details : The ext3 inode allocation policy would not allocate new inodes
5455 very well on the OSTs because there are no new directories
5456 being created. Instead we look for groups with free space if
5457 the parent directories are nearly full.
5462 Description: Network or server problems during mount may cause partially
5463 mounted clients instead of returning an error.
5464 Details : The config llog parsing code may overwrite the error return
5465 code during mount error handling, returning success instead
5471 Description: MDS can fail to allocate large reply buffers
5472 Details : After long uptimes the MDS can fail to allocate large reply
5473 buffers (e.g. zconf client mount config records) due to memory
5474 fragmentation or consumption by the buffer cache. Preallocate
5475 some large reply buffers so that these replies can be sent even
5476 under memory pressure.
5480 Frequency : rare (liblustre)
5481 Description: fsx running with liblustre complained that using truncate() to
5482 extend the file doesn't work. This patch corrects that issue.
5483 Details : This is the liblustre equivalent of the fix for bug 6196. Fixes
5484 ATTR_SIZE and lsm use in llu_setattr_raw.
5488 Frequency : rare, only 2.6 kernels
5489 Description: Unusual file access patterns on the MDS may result in inode
5490 data being lost in very rare circumstances.
5491 Details : Bad interaction between the ea-in-inode patch and the "no-read"
5492 code in the 2.6 kernel caused the inode and/or EA data not to
5493 be read from disk, causing single-file corruption.
5497 Frequency : rare, only 2.6 filesystems using extents
5498 Description: Heavy concurrent write and delete load may cause data corruption.
5499 Details : It was possible under high-load situations to have an extent
5500 metadata block in the block device cache from a just-unlinked
5501 file overwrite a newly-allocated data block. We now unmap any
5502 metadata buffers that alias just-allocated data blocks.
5506 Frequency : filesystems with default stripe_count larger than 77
5507 Description: lconf+mke2fs fail when formatting filesystem with > 77 stripes
5508 Details : lconf specifies an inode size of 4096 bytes when the default
5509 stripe_count is larger than 77. This conflicts with the default
5510 inode density of 1 per 4096 bytes. Allocate smaller inodes in
5511 this case to avoid pinning too much memory for large EAs.
5513 ------------------------------------------------------------------------------
5515 2005-07-07 Cluster File Systems, Inc. <info@clusterfs.com>
5520 Frequency : rare (extremely heavy IO load with hundreds of clients)
5522 Description: Client is evicted, gets IO error writing to file
5523 Details : lock ordering changes for bug 5492 reintroduced bug 3267 and
5524 caused clients to be evicted for AST timeouts. The fixes in
5525 bug 5192 mean we no longer need to have such short AST timeouts
5526 so ldlm_timeout has been increased.
5529 Frequency : occasional during --force or --failover shutdown under load
5530 Bugzilla : 5949, 4834
5531 Description: Server oops/LBUG if stopped with --force or --failover under load
5532 Details : a collection of import/export refcount and cleanup ordering
5533 issues fixed for safer force cleanup
5536 Frequency : only filesystems larger than 120 OSTs
5537 Bugzilla : 5990, 6223
5538 Description: lfs getstripe would oops on a very large filesystem
5539 Details : lov_getconfig used kfree on vmalloc'd memory
5542 Frequency : only filesystems exporting via NFS to Solaris 10 clients
5543 Bugzilla : 6242, 6243
5544 Description: reading from files that had been truncated to a non-zero size
5545 but never opened returned no data
5546 Details : ll_file_read() reads zeros from no-object files to EOF
5551 Description: A bug in MDS/OSS recovery could cause the OSS to fail an assertion
5552 Details : There's little harm in aborting MDS/OSS recovery and letting it
5553 try again, so I removed the LASSERT and return an error instead.
5555 Severity : enhancement
5557 Description: New debugging infrastructure for tracking down data corruption
5558 Details : The I/O checksum code was replaced to: (a) control it at runtime,
5559 (b) cover more of the client-side code path, and (c) try to narrow
5560 down where problems occurred
5564 Bugzilla : 3819, 4364, 4397, 6313
5565 Description: Racing close and eviction MDS could cause assertion in mds_close
5566 Details : It was possible to get multiple mfd references during close and
5567 client eviction, leading to one thread referencing a freed mfd.
5569 Severity: : enhancement
5570 Bugzilla : 3262, 6359
5571 Description: Attempts to reconnect to servers are now more aggressive.
5572 Details : This builds on the enhanced upcall-less recovery that was added
5573 in 1.4.2. When trying to reconnect to servers, clients will
5574 now try each server in the failover group every 10 seconds. By
5575 default, clients would previously try one server every 25 seconds.
5580 Description: After recovery, certain operations trigger a failed
5581 assertion on a client.
5582 Details : Failing over an mds, using lconf -d --failover, while a
5583 client was doing a readdir() call would cause the client to
5584 LBUG after recovery completed and the readdir() was resent.
5586 Severity : enhancement
5588 Description: Default groups are now added by lconf
5589 Details : You can now run lconf --group <servicename> without having to
5590 manually add groups with lmc.
5593 Frequency : occasional
5595 Description: Nodes with an elan id of 0 trigger a failed assertion
5598 Frequency : always when accessing e.g. tty/console device nodes
5600 Description: tty and some other devices nodes cannot be used on lustre
5601 Details : file's private_data field is used by device data and lustre
5602 values in there got lost. New field was added to struct file to
5603 store fs-specific private data.
5606 Frequency : when exporting Lustre via NFS
5608 Description: NFSD failed occasionally when looking up a path component
5609 Details : NFSD is looking up ".." which was broken in ext3 directories
5610 that had grown large enough to become hashed.
5613 Frequency : Clusters with multiple interfaces not on the same subnet
5615 Description: Nodes will repeatedly try to reconnect to an interface which it
5616 cannot reach and report an error to the log.
5617 Details : Extra peer list entries will be created by lconf with some peers
5618 unreachable. lconf now validates the peer before adding it.
5621 Frequency : Only if a default stripe is set on the filesystem root.
5623 Description: Setting a default stripe on the filesystem root prevented the
5624 filesystem from being remounted.
5625 Details : The client was sending extra request flags in the root getattr
5626 request and did not allocate a reply buffer for the dir EA.
5629 Frequency : occasional, higher if lots of files are accessed by one client
5630 Bugzilla : 6159, 6097
5631 Description: Client trips assertion regarding lsm mismatch/magic
5632 Details : While revalidating inodes the VFS looks up inodes with ifind()
5633 and in rare cases can find an inode that is being freed.
5634 The ll_test_inode() code will free the lsm during ifind()
5635 when it finds an existing inode and then the VFS later attaches
5636 this free lsm to a new inode.
5640 Bugzilla : 6422, 7030
5641 Description: MDS deadlock between mkdir and client eviction
5642 Details : Creating a new file via mkdir or mknod (starting a transaction
5643 and getting the ns lock) can deadlock with client eviction
5644 (gets ns lock and trying to finish a synchronous transaction).
5647 Frequency : occasional
5648 Description: While starting a server, the fsfilt_ext3 module could not be
5650 Details : CFS's improved ext3 filesystem is named ldiskfs for 2.6
5651 kernels. Previously, lconf would still use the ext3 name
5652 when trying to load modules. Now, it will correctly use
5653 ext3 on 2.4 and ldiskfs on 2.6.
5655 Severity : enhancement
5656 Description: The default stripe count has been changed to 1
5657 Details : The interpretation of the default stripe count (0, to lfs
5658 or lmc) has been changed to mean striping across a single
5659 OST, rather than all available. For general usage we have
5660 found a stripe count of 1 or 2 works best.
5662 Severity : enhancement
5663 Description: Add support for compiling against Cray portals.
5664 Details : Conditional compiling for some areas that are different
5668 Frequency : occasional
5669 Bugzilla : 6409, 6834
5670 Description: Creating files with an explicit stripe count may lead to
5671 a failed assertion on the MDS
5672 Details : If some OSTs are full or unavailable, creating files may
5673 trigger a failed assertion on the MDS. Now, Lustre will
5674 try to use other servers or return an error to the
5678 Frequency : occasional
5680 Description: Multiple concurrent overlapping read+write on multiple SMP nodes
5681 caused lock timeout during readahead (since 1.4.2).
5682 Details : Processes doing readahead might match a lock that hasn't been
5683 granted yet if there are overlapping and conflicting lock
5684 requests. The readahead process waits on ungranted lock
5685 (original lock is CBPENDING), while OST waits for that process
5686 to cancel CBPENDING read lock and eventually evicts client.
5688 Severity : enhancement
5690 Description: Initial enabling of flock support for clients
5691 Details : Implements fcntl advisory locking and file status functions.
5692 This feature is provided as an optional mount flag (default
5693 off), and is NOT CURRENTLY SUPPORTED. Not all types of record
5694 locking are implemented yet, and those that are are not guaranteed
5695 to be completely correct in production environments.
5696 mount -t lustre -o [flock|noflock] ...
5699 Frequency : occasional
5701 Description: OSTs running 2.4 kernels but with extents enabled might trip an
5702 assertion in the ext3 JBD (journaling) layer.
5703 Details : The b_committed_data struct is protected by the big kernel lock
5704 in 2.4 kernels, serializing journal_commit_transaction() and
5705 ext3_get_block_handle->ext3_new_block->find_next_usable_block()
5706 access to this struct. In 2.6 kernels there is finer grained
5707 locking to improve SMP performance of the JBD layer.
5711 Description: Changes the "SCSI I/O Stats" kernel patch to default to "enabled"
5716 Description: merge and cleanup kernel patches.
5719 -----------------------------------------------------------------------------
5721 2005-05-05 Cluster File Systems, Inc. <info@clusterfs.com>
5723 NOTE: Lustre 1.4.2 uses an incompatible network protocol than previous
5724 versions of Lustre. Please update all servers and clients to
5725 version 1.4.2 or later at the same time. You must also run
5726 "lconf --write-conf {config}.xml" on the MDS while it is stopped
5727 to update the configuration logs.
5729 - fix for HPUX NFS client breakage when NFS exporting Lustre (5781)
5730 - mdc_enqueue does not need max_mds_easize request buffer on send (5707)
5731 - swab llog records of type '0' so we get proper header size/idx (5861)
5732 - send llog cancel req to DLM cancel portal instead of cb portal (5515)
5733 - fix rename of one directory over another leaking an inode (5953)
5734 - avoid SetPageDirty on 2.6 (5981)
5735 - don't re-add just-being-destroyed locks to the waiting list (5653)
5736 - when creating new directories, inherit the parent's custom
5737 striping settings if present parent (3048)
5738 - flush buffers from cache before direct IO in 2.6 obdfilter (4982)
5739 - don't hold i_size_sem in ll_nopage() and ll_ap_refresh_count (6077)
5740 - don't hold client locks on temporary worklist from l_lru (5666)
5741 - handle IO errors in 2.6 obdfilter bio completion routine (6046)
5742 - automatically evict dead clients (5921)
5743 - Update file size properly in create+truncate+fstat case (6196)
5744 - Do not unhash mountpoint dentries, do not allow removal of
5746 - Avoid lock ordering deadlock issue with write/truncate (6203,5654)
5747 - reserve enough journal credits in fsfilt_start_log for setattr (4554)
5748 - ldlm_enqueue freed-export error path would always LBUG (6149,6184)
5749 - don't reference lr_lvb_data until after we hold lr_lvb_sem (6170)
5750 - don't overwrite last_rcvd if there is a *_client_add() error (6086)
5751 - Correctly handle reads of files with no objects (6243)
5752 - lctl recover will also mark a device active if deactivate used (5933)
5754 - by default create 1 inode per 4kB space on MDS, per 16kB on OSTs
5755 - allow --write-conf on an MDS with different nettype than client (5619)
5756 - don't write config llogs to MDS for mounts not from that MDS (5617)
5757 - lconf should create multiple TCP connections from a client (5201)
5758 - init scripts are now turned off by default; run chkconfig --on
5759 lustre and chkconfig --on lustrefs to use them
5760 - upcalls are no longer needed for clients to recover to failover
5762 - add --abort-recovery option to lconf to abort recovery on device
5764 - add support for an arbitrary number of OSTs (3026)
5765 - Quota support protocol changes.
5766 - forward compatibility changes to wire structs (6007)
5767 - rmmod NALs that might be loaded because of /etc/modules.conf (6133)
5768 - support for mountfsoptions and clientoptions to the Lustre LDAP (5873)
5769 - improved "lustre status" script
5770 - initialize blocksize for non-regular files (6062)
5771 - added --disable-server and --disable-client configure options (5782)
5772 - introduce a lookup cache for lconf to avoid repeated DB scans (6204)
5773 - Vanilla 2.4.29 support
5774 - increase maximum number of obd devices to 520 (6242)
5775 - remove the tcp-zero-copy patch from the suse-2.4 series (5902)
5776 - Quadrics Elan drivers are now included for the RHEL 3 2.4.21 and
5777 SLES 9 2.6.5 kernels
5778 - limit stripes per file to 160 (the maximum EA size) (6093)
5780 2005-03-22 Cluster File Systems, Inc. <info@clusterfs.com>
5783 - don't LASSERT in ll_release on NULL lld with NFS export (4655, 5760)
5784 - hold NS lock when calling handle_ast_error->del_waiting_lock (5746)
5785 - fix setattr mtime regression from lovcleanup merge (4829, 5669)
5786 - workaround for 2.6 crash in ll_unhash_aliases (5687, 5210)
5787 - small ext3 extents cleanups and fixes (5733)
5788 - improved mballoc code, several small races and bugs fixed (5733, 5638)
5789 - kernel version 43 - fix remove_suid bugs in both 2.4 and 2.6 (5695)
5790 - avoid needless client->OST connect, fix handle mismatch (5317)
5791 - fix DLM error path that led to out-of-sync client, long delays (5779)
5792 - support common vfs-enforced mount options (nodev,nosuid,noexec) (5637)
5793 - fix several locking issues related to i_size (5492,5624,5654,5672)
5794 - don't move pending lock onto export if it is already evicted (5683)
5795 - fix kernel oops when creating .foo in unlinked directory (5548)
5796 - fix deadlock in obdfilter statistics vs. object create (5811)
5797 - use time_{before,after} to avoid timer jiffies wrap (5882)
5798 - shutdown --force/--failover stability (3607,3651,4797,5203,4834)
5799 - Do not leak request if server was not able to process it (5154)
5800 - If mds_open unable to find parent dir, make that negative lookup(5154)
5801 - don't create new directories with extent-mapping (5909, 5936)
5803 - fix lustre/lustrefs init scripts for SuSE (patch from Scali, 5702)
5804 - don't hold the pinger_sem in ptlrpc_pinger_sending_on_import
5805 - change obd_increase_kms to obd_adjust_kms (up or down) (5654)
5806 - lconf, lmc search both /usr/lib and /usr/lib64 for Python libs (5800)
5807 - support for RHEL4 kernel on i686 (5773)
5808 - provide error messages when incompatible logs are encountered (5898)
5810 2005-02-18 Cluster File Systems, Inc. <info@clusterfs.com>
5811 * version 1.4.0.10 (1.4.1 release candidate 1)
5813 - don't keep a lock reference when lock is not granted (4238)
5814 - unsafe list practices (rarely) led to infinite eviction loop (4908)
5815 - add per-fs limit of Lustre pages in page cache, avoid OOM (4699)
5816 - drop import inflight refcount on signal_completed_replay error (5255)
5817 - unlock page after async write error during send (3677)
5818 - handle missing objects in filter_preprw_read properly (5265)
5819 - no transno return for symlink open, don't save no-trasno open (3440)
5820 - don't try to complete elan receive that already failed (4012)
5821 - free RPC server reply state on error (5406)
5822 - clean up thread from ptlrpc_start_thread() on error (5160)
5823 - readahead could read extra page into cache that wasn't ejected (5388)
5824 - prevent races in class_attach/setup/cleanup/detach (5260)
5825 - don't dereference de->d_inode after l_dput of de (5458)
5826 - use "int" for stripe value returned from lock_to_stripe (5544)
5827 - mballoc allocation and error-checking fixes in 2.6 (5504)
5828 - block device patches to fix I/O request sizes in 2.6 (5482)
5829 - look up hostnames for IB nals (5602)
5830 - 2.6 changed lock ordering of 2 semaphores, caused deadlock (5654)
5831 - don't start multiple acceptors for the same port (5277)
5832 - fix incorrect LASSERT in mds_getattr_name (5635)
5833 - export a proc file for general "ping" checking (5628)
5834 - fix "lfs check" to not block when the MDS is down (5628)
5836 - service request history (4965)
5837 - put {ll,lov,osc}_async_page structs in a single slab (4699)
5838 - create an "evict_client" /proc entry on OSTs, like the MDS has
5839 - fix mount usage message, return errors per mount(8) (5168)
5840 - change grep [] to grep "[]" in tests so they work in more UMLs
5841 - fix ppc64/x86_64 spec to use %{_libdir} instead of /usr/lib (5389)
5842 - remove ancient LOV_MAGIC_V0 EA support (5047)
5843 - add "disk I/Os in flight" and "I/O req time" stats in obdfilter
5844 - align r/w RPCs to PTLRPC_MAX_BRW_SIZE boundary for performance (3451)
5845 - allow readahead allocations to fail when low on memory (5383)
5846 - mmap locking landed again, after considerable improvement (2828)
5847 - add get_hostaddr() to lustreDB.py for LDAP support (5459)
5849 2004-11-23 Cluster File Systems, Inc. <info@clusterfs.com>
5852 - send OST transaction number in read/write reply to free req (4966)
5853 - don't ASSERT in ptl_send_rpc() if we run out of memory (5119)
5854 - lock /proc/sys/portals/routes internal state, avoiding oops (4827)
5855 - the watchdog thread now runs as interruptible (5246)
5856 - flock/lockf fixes (but it's still disabled, pending 5135)
5857 - don't use EXT3 constants in llite code (5094)
5858 - memory shortage at startup could cause assertion (5176)
5860 - reorganization of lov code
5861 - single portals codebase
5863 - add extents/mballoc support (5025)
5864 - direct I/O reads in the obdfilter (4048)
5865 - kernel patches from LNXI for 2.6 (bluesmoke, perfctr, mtd, kexec)
5867 tbd Cluster File Systems, Inc. <info@clusterfs.com>
5870 - send OST transaction number in read/write reply to free req (4966)
5871 - don't ASSERT in ptl_send_rpc() if we run out of memory (5119)
5872 - lock /proc/sys/portals/routes internal state, avoiding oops (4827)
5873 - the watchdog thread now runs as interruptible (5246)
5874 - handle missing objects in filter_preprw_read properly (5265)
5875 - unsafe list practices (rarely) led to infinite eviction loop (4908)
5876 - drop import inflight refcount on signal_completed_replay error (5255)
5877 - unlock page after async write error during send (3677)
5878 - return original error code on reconstructed replies (3761)
5879 - no transno return for symlink open, don't save no-trasno open (3440)
5881 - add pid to ldlm debugging output (4922)
5882 - bump the watchdog timeouts -- we can't handle 30sec yet
5883 - extra debugging for orphan dentry/inode bug (5259)
5885 2004-11-16 Cluster File Systems, Inc. <info@clusterfs.com>
5888 - fix TCP_NODELAY bug, which caused extreme perf regression (5134)
5889 - allocate qswnal tx descriptors singly to avoid fragmentation (4504)
5890 - don't LBUG on obdo_alloc() failure, use OBD_SLAB_ALLOC() (4800)
5891 - fix NULL dereference in /proc/sys/portals/routes (4827)
5892 - allow failed mdc_close() operations to be interrupted (4561)
5893 - stop precreate on OST before MDS would time out on it (4778)
5894 - don't send partial-page writes before EOF from client (4410)
5895 - discard client grant for sub-page writes on large-page clients (4520)
5896 - don't free dentries not owned by NFS code, check generation (4806)
5897 - fix lsm leak if mds_create_objects() fails (4801)
5898 - limit debug_daemon file size, always print CERROR messages (4789)
5899 - use transno after validating reply (3892)
5900 - process timed out requests if import state changes (3754)
5901 - update mtime on OST during writes, return in glimpse (4829)
5902 - add mkfsoptions to LDAP (4679)
5903 - use ->max_readahead method instead of zapping global ra (5039)
5904 - don't interrupt __l_wait_event() during strace
5906 - add software watchdogs to catch hung threads quickly (4941)
5907 - make lustrefs init script start after nfs is mounted
5908 - fix CWARN/ERROR duplication (4930)
5909 - return async write errors to application if possible (2248)
5910 - add /proc/sys/portal/memused (bytes allocated by PORTALS_ALLOC)
5911 - print NAL number in %x format (4645)
5912 - update barely-supported suse-2.4.21-171 series (4842)
5913 - support for sles 9 %post scripts
5914 - support for building 2.6 kernel-source packages
5915 - support for sles km_* packages
5917 2004-10-07 Cluster File Systems, Inc. <info@clusterfs.com>
5920 - ignore -ENOENT errors in osc_destroy (3639)
5921 - notify osc create thread that OSC is being cleaned up (4600)
5922 - add nettype argument for llmount in #5d in conf-sanity.sh (3936)
5923 - reconstruct ost_handle() like mds_handle() (4657)
5924 - create a new thread to do import eviction to avoid deadlock (3969)
5925 - let lconf resolve symlinked-to devices (4629)
5926 - don't unlink "objects" from directory with default EA (4554)
5927 - hold socknal file ref over connect in case target is down (4394)
5928 - allow more than 32000 subdirectories in a single directory (3244)
5929 - fix blocks count for O_DIRECT writes (3751)
5930 - OST returns ENOSPC from object create when no space left (4539)
5931 - don't send truncate RPC if file size isn't changing (4410)
5932 - limit OSC precreate to 1/2 of value OST considers bogus (4778)
5933 - bind to privileged port in socknal and tcpnal (3689)
5935 - rate limit CERROR/CWARN console message to avoid overload (4519)
5936 - GETFILEINFO dir ioctl returns LOV EA + MDS stat in 1 call (3327)
5937 - basic mmap support (3918)
5938 - kernel patch series update from b1_4 (4711)
5940 2004-09-16 Cluster File Systems, Inc. <info@clusterfs.com>
5943 - avoid crash during MDS cleanup with OST shut down (2775)
5944 - fix loi_list_lock/oig_lock inversion on interrupted IO (4136)
5945 - don't use bad inodes on the MDS (3744)
5946 - dynamic object preallocation to improve recovery speed (4236)
5947 - don't hold spinlock over lock dumping or change debug flags (4401)
5948 - don't zero obd_dev when it is force cleaned (3651)
5949 - print grants to console if they go negative (4431)
5950 - "lctl deactivate" will stop automatic recovery attempts (3406)
5951 - look for existing locks in ldlm_handle_enqueue() (3764)
5952 - don't resolve lock handle twice in recovery avoiding race (4401)
5953 - revalidate should check working dir is a directory (4134)
5955 - don't always mark "slow" obdfilter messages as errors (4418)
5957 2004-08-24 Cluster File Systems, Inc. <info@clusterfs.com>
5960 - don't close LustreDB during write_conf until it is done (3860)
5961 - fix typo in lconf for_each_profile (3821)
5962 - allow dumping logs from multiple threads at one time (3820)
5963 - don't allow multiple threads in OSC recovery (3812)
5964 - fix debug_size parameters (3864)
5965 - fix mds_postrecov to initialize import for llog ctxt (3121)
5966 - replace config semaphore with spinlock (3306)
5967 - be sure to send a reply for a CANCEL rpc with bad export (3863)
5968 - don't allow enqueue to complete on a destroyed export (3822)
5969 - down write_lock before checking llog header bitmap (3825)
5970 - recover from lock replay timeout (3764)
5971 - up llog sem before sending rpc (3652)
5972 - reduce ns lock hold times when setting kms (3267)
5973 - change a dlm LBUG to LASSERTF, to maybe learn something (4228)
5974 - fix NULL deref and obd_dev leak on setup error (3312)
5975 - replace some LBUG about llog ops with error handling (3841)
5976 - don't match INVALID dentries from d_lookup and spin (3784)
5977 - hold dcache_lock while marking dentries INVALID and hashing (4255)
5978 - fix invalid assertion in ptlrpc_set_wait (3880)
5980 - add libwrap support for the TCP acceptor (3996)
5981 - add /proc/sys/portals/routes for non-root route listing (3994)
5982 - allow setting MDS UUID in .xml (2580)
5983 - print the stack of a process that LBUGs (4228)
5985 2004-07-14 Cluster File Systems, Inc. <info@clusterfs.com>
5988 - don't cleanup request in ll_file_open() on failed MDS open (3430)
5989 - make sure to unset replay flag from failed open requests (3440)
5990 - if default stripe count is 0, use OST count for inode size (3636)
5991 - update parent mtime/ctime on client for create/unlink (2611)
5992 - drop dentry ref in ext3_add_link from open_connect_dentry (3266)
5993 - free recovery state on server during a forced cleanup (3571)
5994 - unregister_reply for resent reqs (3063)
5995 - loop back devices mounting and status check on 2.6 (3563)
5996 - fix resource-creation race that can provoke i_size == 0 (3513)
5997 - don't try to use bad inodes returned from MDS/OST fs lookup (3688)
5998 - more debugging for page-accounting assertion (3746)
5999 - return -ENOENT instead of asserting if ost getattr+unlink race (3558)
6000 - avoid deadlock after precreation failure (3758)
6001 - fix race and lock order deadlock in orphan handling (3450, 3750)
6002 - add validity checks when grabbing inodes from l_ast_data (3599)
6004 - add /proc/.../recovery_status to obdfilter (3428)
6005 - lightweight CDEBUG infrastructure, debug daemon (3668)
6006 - change default OSC RPC parameters to be better on small clusters
6007 - turn off OST read cache for files smaller than 32MB
6008 - install man pages and include them in rpms (3100)
6009 - add new init script for (un)mounting lustre filesystems (2593)
6010 - run chkconfig in %post for init scripts (3701)
6011 - drop scimac NAL (unmaintained)
6013 2004-06-17 Cluster File Systems, Inc. <info@clusterfs.com>
6016 - clean kiobufs before and after use (3485)
6017 - strip trailing '/'s before comparing paths with /proc/mounts (3486)
6018 - remove assertions to work around "in-flight rpcs" recovery bug (3063)
6019 - change init script to fail more clearly if not run as root (1528)
6020 - allow clients to reconnect during replay (1742)
6021 - fix ns_lock/i_sem lock ordering deadlock for kms update (3477)
6022 - don't do DNS lookups on NIDs too small for IP addresses (3442)
6023 - re-awaken ptlrpcd if new requests arrive during check_set (3554)
6024 - fix cond_resched (3554)
6025 - only evict unfinished clients after recovery (3515)
6026 - allow bulk resend, prevent data loss (3570)
6027 - dynamic ptlrpc request buffer allocation (2102)
6028 - don't allow unlinking open directory if it isn't empty (2904)
6029 - set MDS/OST threads to umask 0 to not clobber client modes (3359)
6030 - remove extraneous obd dereference causing LASSERT failure (3334)
6031 - don't use get_cycles() when creating temp. files on the mds (3156)
6032 - hold i_sem when setting i_size in ll_extent_lock() (3564)
6033 - handle EEXIST for set-stripe, set proper directory name (3336)
6035 - servers can dump a log evicting a client - lustre.dump_on_timeout=1
6036 - fix ksocknal_fmb_callback() error messages (2918)
6038 2004-05-27 Cluster File Systems, Inc. <info@clusterfs.com>
6041 - don't copy lvb into (possibly NULL) reply on error (2983)
6042 - don't deref dentry after dput, don't free lvb on error (2922)
6043 - use the kms to determine writeback rpc length (2947)
6044 - increment oti_logcookies when osc is inactive (2948)
6045 - update client's i_blocks count via lvb messages (2543)
6046 - handle intent open/close of special files properly (1557)
6047 - mount MDS with errors=remount-ro, like obdfilter (2009)
6048 - initialize lock handle to avoid ASSERT on error cleanup (3057)
6049 - don't use cancelling-locks' kms values (2947)
6050 - use highest lock extent for kms, not last one (2925)
6051 - don't dereference ERR_PTR() dentry in error handling path (3107)
6052 - fix thread race in portals_debug_dumplog() (3122)
6053 - create lprocfs device entries at setup instead of at attach (1519)
6054 - common AST error handler, don't evict client on completion race (3145)
6055 - zero nameidata in detach_mnt in 2.6 (3118)
6056 - verify d_inode after revalidate_special is valid in 2.6 (3116)
6057 - use lustre_put_super() to handle zconf unmounts in 2.6 (3064)
6058 - initialize RPC timeout timer earlier for 2.6 (3219)
6059 - don't dereference NULL reply buffer if mdc_close was never sent (2410)
6060 - print nal/nid for unknown nid (3258)
6061 - additional checks for oscc recovery before doing precreate (3284)
6062 - fix ll_extent_lock() error return code for 64-bit systems (3043)
6063 - don't crash in mdc_close for bad permissions on open (3285)
6064 - zero i_rdev for non-device files (3147)
6065 - clear page->private before handing to FS, better assertion (3119)
6066 - tune the read pipeline (3236)
6067 - fix incorrect decref of invalidated dentry (2350)
6068 - provide read-ahead stats and refine rpc in flight stats (3328)
6069 - don't hold journal transaction open across create RPC (3313)
6070 - update atime on MDS at close time (3265)
6071 - close LDAP connection when recovering to avoid server load (3315)
6072 - update iopen-2.6 patch with fixes from 2399,2517,2904 (3301)
6073 - don't leak open file on MDS after open resend (3325)
6074 - serialize filter_precreate and filter_destroy_precreated (3329)
6075 - loop device shouldn't call sync_dev() for nul device (3092)
6076 - clear page cache after eviction (2766)
6077 - resynchronize MDS->OST in background (2824)
6078 - refuse to mount the same filesystem twice on same mountpoint (3394)
6079 - allow llmount to create routes for mounting behind routers (3320)
6080 - push lock cancellation to blocking thread for glimpse ASTs (3409)
6081 - don't call osc_set_data_with_check() for TEST_LOCK matches (3159)
6082 - fix rare problem with rename on htree directories (3417)
6084 - allow default OST striping configuration per directory (1414)
6085 - fix compilation for qswnal for 2.6 kernels (3125)
6086 - increase maximum number of MDS request buffers for large systems
6087 - change liblustreapi to be useful for external progs like lfsck (3098)
6088 - increase local configuration timeout for slow disks (3353)
6089 - allow configuring ldlm AST timeout - lustre.ldlm_timeout=<seconds>
6091 2004-03-22 Cluster File Systems, Inc. <info@clusterfs.com>
6094 - fixes for glimpse AST timeouts / incorrectly 0-sized files (2818)
6095 - don't overwrite extent policy data in reply if lock was blocked (2901)
6096 - drop filter export grants atomically with removal from device (2663)
6097 - del obd_self_export from work_list in class_disconnect_exports (2908)
6098 - don't LBUG if MDS recovery times out during orphan cleanup (2530)
6099 - swab reply message in mdc_close, other PPC fixes (2464)
6100 - fix destroying of named logs (2325)
6101 - overwrite old logs when running lconf --write_conf (2264)
6102 - bump LLOG_CHUNKSIZE to 8k to allow for larger clusters (2306)
6103 - fix race in target_handle_connect (2898)
6104 - mds_reint_create() should take same inode create lock (2926)
6105 - correct journal credits calculated for CANCEL_UNLINK_LOG (2931)
6106 - don't close files for self_export to avoid uninitialized obd (2936)
6107 - allow MDS with the same name as client node (2939)
6108 - hold dentry reference for closed log files for unlink (2325)
6109 - reserve space for all logs during transactions (2059)
6110 - don't evict page beyond end of stripe extent (2925)
6111 - don't oops on a deleted current working directory (2399)
6112 - handle hard links to targets without a parent properly (2517)
6113 - don't dereference NULL lock when racing during eviction (2867)
6114 - don't grow lock extents when lots of conflicting locks (2919)
6116 2004-03-04 Cluster File Systems, Inc. <info@clusterfs.com>
6119 - account for cache space usage on clients to avoid data loss (974)
6120 - lfsck support in lustre kernel code (2349)
6121 - reduce journal credits needed for BRW writes (2370)
6122 - orphan handling to avoid losing space on client/server crashes
6123 - ptlrpcd can be blocked, stopping ALL progress (2477)
6124 - use lock value blocks to assist in proper KMS, faster stat (1021)
6125 - takes i_sem instead of DLM locks internally on obdfilter (2720)
6126 - recovery for initial connections (2355)
6127 - fixes for mds_cleanup_orphans (1934)
6128 - abort_recovery crashes MDS in b_eq (mds_unlink_orphan) (2584)
6129 - block all file creations until orphan recovery completes (1901)
6130 - client remove rq_connection from request struct (2423)
6131 - conf-sanity test_5, proper cleanup in umount log not availale (2640)
6132 - recovery timer race (2670)
6133 - mdc_close recovey bug (2532)
6134 - ptlrpc cleanup bug (2710)
6135 - mds timeout on local locks (2588)
6136 - namespace lock held during RPCs (2431)
6137 - handle interrupted sync write properly (2503)
6138 - don't try to handle a message that hasn't been replied to (2699)
6139 - client assert failure during cleanup after abort recovery (2701)
6140 - leak mdc device after failed mount (2712)
6141 - ptlrpc_check_set allows timedout requests to complete (2714)
6142 - wait for inflight reqs when ptlrpcd finishes (2710)
6143 - make sure unregistered services are removed from the srv_list
6144 - reset bulk XID's when resending them (caught by 1138 test)
6145 - unregister_bulk after timeout
6146 - fix lconf error (2694)
6147 - handle write after unfinished setstripe, stripe-only getstripe (2388)
6148 - readahead locks pages, leaves pending causing memory pressure (2673)
6149 - increase OST request buffers to 4096 on large machines (2729)
6150 - fix up permission of existing directories in simple_mkdir (2661)
6151 - init deleted item, add assertions ptlrpc_abort_inflight() (2725)
6152 - don't assign transno to errored transactions (2742)
6153 - don't delete objects on OST if given a bogus objid from MDS (2751)
6154 - handle large client PAGE_SIZE readdir on small PAGE_SIZE MDS (2777)
6155 - if rq_no_resend, then timeout request after recovery (2432)
6156 - fix MDS llog_logid record size, 64-bit array alignment (2733)
6157 - don't call usermode_helper from ptlrpcd, DEFAULT upcall (2773)
6158 - put magic in mount.lustre data, check for bad/NULL mount data (2529)
6159 - MDS recovery shouldn't delete objects that it has given out (2730)
6160 - if enqueue arrives after completion, don't clobber LVB (2819)
6161 - don't unlock pages twice when trigger_group_io returns error (2814)
6162 - don't deref NULL rq_repmsg if ldlm_handle_enqueue failed (2822)
6163 - don't write pages to disk if there was an error (1450)
6164 - don't ping imports that have recovery disabled (2676)
6165 - take buffered bytes into account when balancing socknal conn (2817)
6166 - hold a DLM lock over readdir always, use truncate_inode_pages (2706)
6167 - reconnect unlink llog connection after MDS reconnects to OST (2816)
6168 - remove little-endian swabbing of llog records (1987)
6169 - set/limit i_blksize to LL_MAX_BLKSIZE on client (2884)
6170 - retry reposting request buffers if they fail (1191)
6171 - grow extent at grant time to avoid granting a revoked lock (2809)
6172 - lock revoke doesn't evict page if covered by a second lock (2765)
6173 - disable VM readahead to avoid reading outside lock extents (2805)
6175 - return LL_SUPER_MAGIC from statfs for the filesystem type (1972)
6176 - updated kernel patches for hp-2.4.20 kernel (2681)
6178 2004-02-07 Cluster File Systems, Inc. <info@clusterfs.com>
6181 - fix truncated write corruption (2366)
6182 - fix for failed assertion in iopen_connect_dentry (1792,2517)
6184 - don't flag the ptlrpcd thread with PF_MEMALLOC (2636)
6185 - ensure len(uuid) < 37 in lmc (1171)
6186 - fix ia64 OOPS in llog_test (2255)
6187 - zero end of page at obdfilter for partial page writes (2648)
6188 - don't leave stale dentries around after renames (bug 2428)
6189 - fix timeouts when evicting a client with a single lock held (2642)
6190 - set deadline for the initial HELLO message to drain (2634)
6191 - print out dotted-quad IP addresses in the socknal (2302)
6193 - additional debugging for MDS client eviction problem (2443)
6194 - fix mkfsoptions support for osts (2603, 2604)
6196 2004-01-27 Cluster File Systems, Inc. <info@clusterfs.com>
6199 - add series for the vanilla 2.6.0 kernel
6200 - add series for the vanilla 2.4.24 kernel
6201 - add series for a cray x86/64 UL kernel drop
6202 - fix xattr patches for the vanilla 2.4.19 series
6204 - generate true UUIDs in lmc (1171)
6205 - have portals stack dumping break in UML (2466)
6206 - avoid bad dchild deref; avoid inum lock w/o creation (2362)
6207 - allocate with _NOFS in ldlm to avoid deadlock (1933)
6208 - wake callback waiting threads on client eviction (2460)
6209 - Add --ptldebug and --subsystem to lmc (1719)
6210 - update assertion to allow safe interrupt allocation
6211 - set rq_no_resend for cancel requests (2432)
6212 - recalculate ptlrpcd timeout after resend (2494)
6213 - call vfs_rmdir when removing pending directories (2368)
6214 - fix renaming a file to itself (2429)
6215 - lmc creates a default one-stripe lov (2454)
6216 - expand procfs space to handle large clusters (2326)
6217 - increase UML stack to avoid overflow
6218 - update lconf's list of debug and subsystem masks
6219 - fix lfs find --obd (2510)
6220 - /proc tunable for disabling filter read caching (2591)
6221 - stop rpm packages from altering slapd.conf (2301)
6222 - disable nagle in the socknal under 0conf (2578)
6223 - choose mds inode size based on stripe count (2572)
6224 - fix kernel-source rpm problems (2516)
6226 - add --disable-doc to avoid pdf generation (2421)
6227 - update documentation, tests, type-os, comments
6228 - avoid format warnings on ia64
6229 - remove the TOE NAL
6230 - tiny code cleanups by removing unused fields
6232 2004-01-07 Cluster File Systems, Inc. <info@clusterfs.com>
6235 - fix obvious semaphore misuse in as-yet-unused setattr path (2348)
6236 - remove the most blatant lies from BUILDING file (2371)
6237 - change default debug level to reasonable production setting
6238 - reduce client side cache size to reduce cache flush time
6239 - reduce max RPCs in flight to avoid unnecessary file fragmentation
6240 - make TCP zerocopy and pinger support enabled by default (2476)
6241 - sync writes completed after process exits caused crashes (2319)
6242 - maintain correct mount count on the MDS (2356)
6243 - backout 1557, because 2316 wasn't really fixed
6244 - better file I/O statistics gathering in /proc
6245 - don't take unnecessary, deadlock-inducing bug in readpage (2383)
6246 - another kernel patch to fix zero-copy TCP function export
6247 - don't take duplicate lock when processing re-sent getattr (2420)
6248 - lctl uses obd_self_export instead of creating new conn (2353)
6249 - MDS/OST recovery case which requires object creation asserted (2425)
6250 - move lfs from /usr/sbin to /usr/bin in packages
6251 - fix race between mds_client_add and mds_client_free (2417)
6252 - use kmalloc instead of slabs in portals (2430)
6253 - don't create duplicate records when a failover MDS is present (2442)
6254 - remove unnecessary mount age check (2332)
6255 - don't remove directory inodes from locks prematurely (2451)
6256 - don't break if MDS service name is the same as hostname (2103)
6257 - fix races in client write RPC generation when cache full (2482)
6259 2003-12-13 Cluster File Systems, Inc. <info@clusterfs.com>
6262 - remove now-unused request->rq_obd (278)
6263 - if an allocation fails, print out how much memory we've used (1933)
6264 - use PORTAL_SLAB_ALLOC for structures, to get GFP_MEMALLOC (1933)
6265 - add the "configurable stack size" patch to most series files (1256)
6266 - ability to write large log records, for 100+ OST configs (2306)
6267 - fix NULL deref when filter_prep fails (2314)
6268 - fix operator precedence error in filter_sync
6269 - dynamic allocation of socknal TX descriptors (2315)
6270 - fix a missed case in the GFP_MEMALLOC patch, can cause deadlock (2310)
6271 - fix gcc 2.96 compilation problem in xattr kernel patch (2294)
6272 - ensure that CWARN messages in Portals always get to the syslog
6273 - __init/__exit are not for prototype decls (ldlm_init/exit)
6274 - x86-64 compile warning fixes
6275 - fix gateway LMC keyword conflict (2318)
6276 - fix MDS lock inversions in getattr/reint paths (1844)
6277 - fix a rare lock re-ordering bug, which caused deadlock (2322)
6278 - fix i_sem/journal inversion in fsfilt_ext3_write_record (2306)
6279 - DLM race condition prevented some lock evictions (2328)
6280 - ENOMEM detection and retry on socknal sends (2230)
6281 - use GFP_NOFS throughout Lustre, to combat ENOMEM (2230)
6282 - move osc_rpcd into ptlrpc, for use in MDC and others (2329)
6283 - protect MDS inode fsdata with stronger locking; fixes assertion (2313)
6284 - better error messages when a client is rejected during recovery (1505)
6285 - avoid cancelling locks which were never granted, after failure (2330)
6286 - fix i_sem/journal inversion in mds_client_add (2333)
6287 - fix truncate/getattr lock cycle deadlock (2334)
6288 - use rpcd to send close; allows resend after timeout, avoid leak (1897)
6289 - fix two rare exit paths which could leak an l_lock() ref (2321)
6290 - fencepost error in MDS/OST orphan recovery (2226)
6291 - make log record alignment 8 bytes (1988)
6292 - lstripe now fails when requested offset > ost_count (2237)
6293 - ensure that all kernel series have a complete list.h (1607)
6294 - fix crashes in special-file operations (2316)
6295 - lctl create/brw OID mismatch, caused by obsolete filter loop (2339)
6297 - allow configurable automake binary, for testing new versions
6298 - small update to the lfs documentation
6300 2003-12-03 Cluster File Systems, Inc. <info@clusterfs.com>
6302 * fix negative export reference count in fsfilt_sync (2312)
6304 2003-12-01 Cluster File Systems, Inc. <info@clusterfs.com>
6305 * release candidate 0.9.1
6307 - orphans are moved into the PENDING directory for possible recovery
6308 - replayed opens now open by fid for orphan/rename safety (1042)
6309 - last close of an orphan inode generates a transno (683)
6310 - chdir() and mount() now pin the directory entry (1020)
6311 - avoid CERROR in normal ll_setattr_raw() error case (1500)
6312 - discard very old requests without processing them (1502)
6313 - remove some common, well-understood CERRORs (1505)
6314 - require O_DIRECT I/O to be page-sized to workaround IA64 crash (1609)
6315 - clear "grant" flags in OST replies until OST grant code lands (1644)
6316 - fix read performance by not clobbering i_blksize on client (1598)
6317 - fix __ldlm_handle2lock oops by not dereferencing lock after PUT (1625)
6318 - make LRU size a /proc tunable, clears locks when reduced (707)
6319 - fix some lprocfs rot that prevented ptlbd from loading (1732)
6320 - server locks take references on exports now (1558)
6321 - build fixes for 2.4.20-rh trees (1663)
6322 - return an error from lov_create if all OSCs are inactive (1751)
6323 - fix import levels when a reconnect happens without a timeout (1597)
6324 - exit early from mds_open if we get a lookup error (1749)
6325 - partial page read at EOF wouldn't wait for disk before sending (1642)
6326 - avoid NULL deref in obdfilter when reading page past EOF (1592)
6327 - avoid LASSERT in ll_intent_lock if server failed very early (1090)
6328 - fix LBUG in ll_it_open_error with rc = -2 (1861)
6329 - write/truncate lock inversion (1639)
6330 - Don't auto-load obdclass, portals modules during cleanup (1495)
6331 - fix timestamps from jumping to "now" (1763)
6332 - extra journal assertions (1648)
6333 - add an extra multiunlink test (1771)
6334 - fix read_record/write_record API (1776)
6335 - fix leak of offset_extent, possible incorrect i_size later (1772)
6336 - fix lasserts in mis-matched transnos during open-unlink testing (1541)
6337 - Debugging for the kqswnal_get_idle_tx problems (1820)
6338 - Allow recovery to be attempted multiple times (1536)
6339 - Write out MDS last_rcvd file after it is first created (1600)
6340 - Fix tx_descriptor leak in failed transmit situations (1827)
6341 - ext3 journaling fixes for assertion failure after IO error (1871)
6342 - class_export_put() on freed export after completion AST error (1896)
6343 - Fix revalidate looping in VFS (1322)
6344 - Don't access a freed export during MDS_REINT timeout (1521)
6345 - Add open-unlink recovery support on the MDS (1673,1764)
6346 - Return an error if no MDS data was read from last_rcvd (1946)
6347 - Fix for lookup "." or ".." crash on error (1932,1931,1935)
6348 - Don't setup a disk device that doesn't match exported UUID (317)
6349 - Reduce bulk RPC timeout to avoid cascading client/OST failures (1845)
6350 - avoid committing NULL handle in force close
6351 - local.sh is now a one-stripe LOV configuration
6352 - POSIX utime.4 -EPERM on FIFO not owned by user (56)
6353 - fix ext3 htree duplicate directory entry corruption (1516)
6354 - POSIX creat.13, fstat.1, open.18, stat.3 new file atime/mtime (2020)
6355 - update to new LOV EA format (2097)
6356 - interoperability for different PAGE_SIZE/wordsize (686,1821,1343,2042)
6358 2003-06-15 Phil Schwan <phil@clusterfs.com>
6361 - imports and exports cleanup too early, need refcounts (349, 879, 1045)
6362 - per-import/export recovery handling (958, 931, 959)
6363 - multiple last-rcvd slots, for serving multiple FSes (949)
6364 - connections are again shared between multiple imp/exports (963, 964)
6365 - "umount -f" would hang if any requests needed to be sent (393, 978)
6366 - avoid pinning large req buffer by copying for queued messages (989)
6367 - add "uuid" to "lctl device" command to help upcalls (991)
6368 - "open" RPCs with transnos would confuse recovery counters (1037)
6369 - do proper endian conversion of all wire messages (288, 340, 891)
6370 - remove OST bulk get LBUGs, fix ost_brw_write cleanup (1126)
6371 - call waiting locks callback from LDLM recovery thread (1127, 1151)
6372 - fix ptlrpc_connection leak in target_handle_connect (1174)
6373 - fix import refcounting bug in OST and MDS cleanup (1134)
6374 - if an invalid-at-open-time OSC returned before close(), LBUG (1150)
6375 - fix very unlikely obd_types race condition (501)
6376 - remove osc_open hack for echo_client (1187)
6377 - we leaked exports/dlmimps for forcibly disconnected clients (1143)
6378 - a failure in read_inode2 leads to deadlock (1139)
6379 - cancel ack-locks as soon as transaction is committed (1072)
6380 - fix major leaks and crashes in the bulk I/O path (937, 1057)
6381 - make sure to commitrw after any preprw to avoid deadlock (1162)
6382 - failing to execute a file in a lustre FS would lock inode (1203)
6383 - small DEBUG_REQ fix to avoid dereferencing a NULL (1227)
6384 - don't ASSERT while cleaning up an incompletely-setup obd (1248)
6385 - obd_uuid2tgt would walk off the end of the list (1255)
6386 - on IA64 the osc would give portals incorrect bulk size (1258)
6387 - fix debug daemon ioctl interface; allows daemon on ia64 (1274)
6388 - fix lock inversion caused by new llite matching code (1282)
6389 - limit the number of dirty pages on a client to 10MB (1286)
6390 - timed out locks were not being corrected cancelled (1289)
6391 - fix O_DIRECT above 4GB on IA-32 (1292)
6392 * major user-visible changes
6393 - fail out/fail over policy now controlled by the upcall (993)
6395 - add OBD_PING to check server availability and failure (954)
6396 - lustre messages are now sent in sending host order (288, 340, 891)
6397 - add eadatalen to MDS getattr reply (340)
6398 - OST read replies may contain second buffer, with per-page status (593)
6400 2003-03-11 Phil Schwan <phil@clusterfs.com>
6403 - LDLM_DEBUG macro fix, for gcc 3.2 (850)
6404 - failed open()s could cause deadlock; fixed (867, 869)
6405 - stop cancelling OST locks when files are closed (481)
6406 - overlapping XID spaces caused network corruption (851, 853)
6407 - fix unsafe fsfilt counter arithmetic; change to atomic_t
6408 - setattr_raw added, to do single-RPC, server-side setattrs
6409 - lmc/lconf syntax change for OST UUIDs
6410 - fix crashy race condition between ptlrpc_free_req and osc_close
6411 - don't use request in mdc_enqueue if we hit a timeout (889)
6412 - don't set the inode i_size for regular files from the MDS (896)
6413 - handle out of order completion AST (842)
6414 - don't LBUG if a lock request times out after receiving AST (913)
6415 - avoid d_rehash race in ll_find_alias by rehashing inside dcache_lock
6416 - if a bad lock AST arrives, send an error instead of dropping entirely
6417 - return 0 from revalidate2 if ll_intent_lock returns -EINTR (912)
6418 - fix leak in bulk IO when only partially completed (899, 900, 926)
6419 - fix O_DIRECT for ia64 (55)
6420 - (almost) eliminate Lustre-kernel-thread effects on load average (722)
6421 - C-z after timeout could hang a process forever; fixed (977)
6423 - client-side I/O cache (678, 924, 929, 941, 970)
6425 - READPAGE and SETATTRs which don't take server-side locks get
6428 2003-02-11 Phil Schwan <phil@clusterfs.com>
6431 - Fix ldlm_lock_match on the MDS to avoid matching remote locks (592)
6432 - Fix fsfilt_extN_readpage() to read a full page of directory
6433 entries, or fake the remainder if PAGE_SIZE != blocksize (500)
6434 - Avoid extra mdc_getattr() in ll_intent_lock when possible (534, 604)
6435 - Fix imbalanced LOV object allocation and out-of-bound access (469)
6436 - Most intent operations were removed, in favour of a new RPC mode
6437 that does a single RPC to the server and bypasses most of the VFS
6438 - All LDLM resource ID arrays were removed in favour of ldlm_res_id
6439 - Aggressively cancel local locks on DLM servers
6440 - mds_reint_unlink sends EA to the client if it's the last nlink.
6441 client uses that EA to unlink OST objects.
6442 - mds_reint_{rename,unlink,link} were rewritten to take ordered locks
6443 - recursive symlinks were fixed (439)
6444 - fixed NULL deref in DEBUG_REQ
6445 - filter_update_lastobjid no longer calls sync, which annoyed extN
6446 - fixed multi-client small-writes to a single file problem (445)
6447 - fixed mtime updates during file writes (607)
6448 - fixed vector writes on obdfilter causing problems when ENOSPC (670)
6449 - fixed bug in obd_brw_read/write() (under guise of testing 367)
6450 - fixed Linux OST size reporting problem (444, 656)
6451 - OST now updates object mtime with writes or setattr (607, 619)
6452 - client verifies file size before zeroing page past EOF (445)
6453 - OST now writes last allocated objid to disk with allocation (108)
6454 - LOV on echo now works (409)
6456 - mds_reint_unlink sends a new buffer, with the EA included. this
6457 buffer is only valid if body->valid & OBD_MD_FLEASIZE, which is only
6458 set if a regular file was being unlinked, and it was the last link
6459 - use PtlGet from the target for bulk writes (315)
6460 - OST now updates object mtime with writes or setattr (607, 619)
6461 - LDLM now has a grant-time callback to revalidate locked items, if
6463 - Many MDS operations were reorganized to combat race conditions
6465 - Merge b_intel branch (updated lprocfs code) - now at /proc/fs/lustre
6466 - configure check to avoid gcc version 2.96 20000731-2.96-98) (606)
6468 2003-01-06 Andreas Dilger <adilger@clusterfs.com>
6471 - Fully reactivate OST imports after reconnection (512, others)
6472 - Make sure client sees our -ENOTCONN from mds_handle (513 - partial)
6473 - More graceful error handling for truncating on dead OST (515)
6474 - Don't error out unless we're actually accessing dead stripes (474)
6475 - Fix garbage sizes when stripes are missing (410)
6476 - LRU counters were broken, causing constant lock purge (433, 432)
6477 - garbage on read from stripes with failed OSTs (441)
6478 - mark OSCs as active before reconnecting during recovery (438)
6479 - lov_enqueue and lov_cancel need to handle inactive OSTs (403)
6480 - lfind did not preserve OST order in output (443)
6481 - symlinks cause hung clients, incorrect data (439)
6482 - stop dereferencing request after dropping refcount (457)
6483 - don't LASSERT(spin_is_locked) on non-SMP (455)
6484 - fixes for many rename() bugs
6485 - fstat didn't correctly synchronize attributes (399)
6486 - server must handle lock cancellation during blocking AST prep (487)
6487 - bulk descriptors were free()d too soon (511)
6488 - fix paths in lconf, which would load incorrect modules (451, 507)
6489 - fix confusing lconf 'host not found' error message (386)
6490 - fix lock order deadlock on OST (O/R i_sem before journal ops, 478)
6491 - fix race condition in mdc_blocking_ast() for inode access (526)
6492 - fix lov_unpackmd() unpacking wrong number of stripes (537)
6493 - fix lov_set_osc_active() marking wrong OSC inactive (440)
6494 - fix bad lstripe lov_unpackmd() assertion (fix layering too) (527)
6495 - fix multiple writes of stripe MD to MDS (358, maybe 519)
6496 - fix lstripe in several ways (kernel side) (527)
6497 - fix request leak in ldlm_cli_enqueue (262)
6498 - incorrect OSC was marked inactive after OST failure
6499 - call mds_fs_cleanup before unmounting filesystem (524)
6500 - fix races between taking ns_lock and ldlm_lock_change_resource
6501 - fix races updating LOV export open file list
6502 - fix lov_enqueue error path, avoid decref-ing bad lock handle
6503 - fix recovery NULL deref in ldlm_cli_cancel_unused
6504 - fix some DLM races by using new hash table for lock handles (419)
6505 - permit the client to specify desired inodes, at replay
6506 - duplicate requests when we queue them for replay reintegration
6507 - fix last_rcvd offset calculation
6508 - sync after each recovered transaction, so we always make progress
6509 - never, not always, ERESTART requests without transnos
6510 - store the lov_desc in the MDS, so we don't depend on getlovinfo to
6512 - skip replay if the MDS says that the client is already connected
6513 - don't check for a recovery-enabled export to match lctl's UUID
6514 - don't INC_USE_COUNT for phantom exports
6515 - don't crash when cleaning up phantom exports (567)
6516 - don't double-finish or set replay data for errored mdc_open requests
6517 - abort requests when they time out, so we don't get old replies
6518 - send/receive replies for AST messages again
6519 - if the client says that it doesn't have the lock, cancel it on the
6521 - if we timeout during I/O, don't try to cancel an in-use lock; instead
6522 mark it as destroyed, it will all work out when decref is called
6523 - fix module use counts (22, 581)
6525 - ASTs now expect a reply (server cancels lock on error reply)
6527 2002-12-02 Andreas Dilger <adilger@clusterfs.com>
6530 - fix many simultaneous client startup (392)
6531 - fix dentry->d_it clobbering
6532 - credentials weren't being shipped for readdir/getattr operations
6533 - remove invalid assertions triggered during some concurrent MD
6535 - proper Lustre versions added (336, 389)
6536 - fix memory leak for create error case (398)
6537 - fix LOV locking bug that would get cli/srv out of sync
6538 - fix echo client over LOV (409)
6539 - fix dbench 2, extN refcount problem (170, 258, 356, 418)
6540 - fix double-O_EXCL intent crash (424)
6541 - avoid sending multiple lock CANCELs (352)
6543 - MDS can do multi-client recovery (modulo bugs in new code)
6545 - many updates, edits, cleanups
6547 2002-11-18 Phil Schwan <phil@clusterfs.com>
6550 - fix null d_it dereference (346)
6551 - fix full OST/dbench hang (333)
6552 - fix permission problem with file removal (286)
6553 - fix removal of OSCs from LOV when they fail
6554 - fix NULL deref during bulk timeout (214)
6555 - fix problems related to multiple filesystems on one MDS (241)
6556 - fixed serious subtle metadata locking bugs
6557 - free locks on clients when inodes are removed due to memory
6559 - fix inode pointer in lock data (285)
6560 - partial support for multiple MDS on a single host (241)
6561 - data locks weren't cancelled at clear_inode time (290, 311)
6562 - intent locks could lead to unbounded lock growth (205)
6563 - added a maximum lock count, an LRU list, and a flusher
6564 - fix multiple rename (365)
6565 - properly abstracted the echo client
6566 - OSC locked 1 byte too many; fixed
6567 - rewrote brw callback code:
6568 - fixed recovery bugs related to LOVs (306)
6569 - fixed too-many-pages-in-one-write crash (191)
6570 - fixed (again) crash in sync_io_timeout (214)
6571 - probably fixed callback-related race (385)
6573 - Add capability to MDS protocol
6574 - LDLM cancellations and callbacks on different portals
6576 2002-10-28 Andreas Dilger <adilger@clusterfs.com>
6579 - limit client IOV size to PTL_MD_MAX_IOV (611336, 191)
6580 - defer open object destruction to close time (601981, 138)
6581 - open/close OST file handle in obdo (OBD_MD_FLHANDLE) (601981, 138)
6582 - move LDLM_ENQUEUE/CONVERT back to MDS portal (625069)
6583 - abstract ll_lookup2, fix ll_revalidate2 to use abstraction (256)
6584 - don't call obd_setattr in ll_file_release for destroyed objects
6585 * protocol change to lustre_msg: move |version| and add |flags|
6586 * protocol change to osc_punch: "start" in "o_size", "end" in "o_blocks"
6587 * lock replay: for LDLM_FL_REPLAY trust client to do right thing
6588 * added replay of create, unlink, link and rename operations during
6589 MDS failover; recovery should be much more robust now
6590 * remove failed OSCs from LOVs (only lov_create uses this so far)
6591 * the lustre-HOWTO was brought (more) up to date (582544)
6593 2002-10-23 Phil Schwan <phil@clusterfs.com>
6596 - in-use dentries weren't being reused properly (617851)
6597 - prevent multiple LDLM setup (599178)
6598 - fix LOV size calculations for truncate (617853)
6599 - fix client handling of MDS intent errors (POSIX)
6600 - fix permission bug in lovstripe.c test (624321)
6601 - fix MDS thread deadlock - move LDLM handler to DLM portal (625069)
6602 - truncate past end of file could corrupt data
6603 - proper cleanup after timeouts, crashes, etc (592524, 550815)
6604 - a race in recovery could return ETIMEDOUT to apps (623947)
6605 - building outside the source directory was fixed
6606 * the lustre-HOWTO was brought (more) up to date (582544)
6607 * major progress was made on recovery functionality
6609 2002-10-10 Phil Schwan <phil@clusterfs.com>
6612 - recovery deadlock fix
6613 - rm -rf causes LBUG fix (617817)
6614 - file open by multiple tasks fix (618962)
6615 - directory permissions bugs (602707 and 620007)
6616 - journal_stop fixed with locking (611313)
6617 - O_APPEND failures resolved (618273, perhaps 614459)
6618 - lconf PATH fix (619770)
6619 - IA64 build fix (621450)
6620 - RPC buffer sizes scale with amount of memory
6622 2002-10-01 Phil Schwan <phil@clusterfs.com>
6625 - locks would be cancelled without throwing away data pages,
6626 resulting in inconsistent data (605627)
6627 - inode attributes were not always being refreshed (605627, 612449)
6628 - lconf now continues to cleanup after lctl reports an error
6629 - MDS now enforces user permissions (602707)
6630 - lprocfs cleanup fixed, but not yet enabled (614157)
6631 - fixed infinite server hang, should a client not respond to an AST
6632 - avoid going into recovery if user calls readlink() with a buffer
6633 that's too small (613941)
6634 - AST RPCs no longer require replies (614867) -- this may be changed
6635 - don't crash server if client sends an IOV that's too big (611336)
6636 - fixed lock conversion deadlock (611892)
6637 - fixed the following of symlinks (614622)
6638 * recovery: the server can remove locks from a client that dies, other
6639 clients can make progress
6640 * more extN patch fixes
6641 * compile-time configurable ptlrpc buffer allocations
6643 - collaborative read cache document
6644 - Lustre Lite Performance CDR document-in-progress
6646 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
6649 - fix typo in patch-2.4.18
6651 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
6654 - clear ptlrpc request each time in handle_incoming_request()
6655 - unlink of files now destroys the object on the OST
6657 2002-09-19 Peter Braam <braam@clusterfs.com>
6659 * add hard link support
6660 * change obdfile creation method
6661 * kernel patch changed
6663 2002-09-19 Peter Braam <braam@clusterfs.com>
6666 - stack overflow bug in extN fixed
6668 2002-09-18 Andreas Dilger <adilger@clusterfs.com>
6670 * documentation updates
6671 - add man pages for config tools
6672 - update tests/README to describe testing with new config tools
6673 - finish metadata API descriptions
6674 * bug fixes and cleanups
6675 - statfs workaround for 16TB limit
6676 - LOV stripe allocation improved, can stripe on subset of OSTs
6677 - LOV file size/IO offset was wrong for files > 4GB in size
6678 - object EA data was being dropped, caused files to be unreadable
6679 - memory overflow with non-LOV OST caused memory corruption
6680 - fixed regression tests to work with new config tools, obdfilter
6681 - fixed bug when directory size became larger than 1 block
6682 - fixed bug (for single client case) when PWD was deleted
6683 - invalidate local directory pages when doing intent-based ops
6684 - avoid LDLM oops when lock callback contained bad data
6686 2002-09-09 Andreas Dilger <adilger@clusterfs.com>
6688 * documentation updates
6689 * bug fixes and cleanups
6690 - configuration tools
6693 - 64-bit compile warnings
6694 - 64-bit internal statfs data
6696 * test_brw on persistent OST devices
6698 * lprocfs (disabled)
6700 2002-09-04 Andreas Dilger <adilger@clusterfs.com>
6702 * documentation updates
6703 * bug fixes and cleanups
6704 * configuration tools
6706 2002-08-30 Peter J. Braam <braam@clusterfs.com>
6709 * many small fixes to 0_5_4
6710 * io/network handling
6711 * thinkos in MDS operations
6713 2002-08-24 Peter J. Braam <braam@clusterfs.com>
6716 * crucial basic fixes to 0.5.3
6717 * IOR, Iozone work over Elan
6720 2002-08-07 Phil Schwan <phil@clusterfs.com>
6721 * version 0_5_3, our first alpha
6722 * we use the new Portals iovs
6723 * documentation updates
6724 * bug fixes and cleanups
6725 * small changes in the DLM wire protocol
6727 2002-07-25 Peter J. Braam <braam@clusterfs.com>
6728 * version 0_5_1 with some initial stability,
6729 * locking on MD and file I/O.
6730 * documentation updates
6731 * several bug fixes since 0.5.0
6732 * small changes in wire protocol
6734 2002-07-18 Phil Schwan <phil@clusterfs.com>
6736 * delivered as Lustre Light Alpha
6737 * fixed a crash after handling invalid MDS requests
6738 * fixed directory pages for architectures with non-4k pages sizes
6740 2002-07-11 Andreas Dilger <adilger@clusterfs.com>
6741 * release version v0_4_4
6742 * Moves TCP acceptor to be on port 2432 (unused Coda port) instead
6744 * Fixes a number of interruption problems with OST operations.
6745 * Update documentation for portals header changes
6746 * Move all wire protocol structs/defines to lustre_idl.h
6747 * Fixes symlink length bug.
6748 * Add tcpdump to repository.
6750 2002-07-05 Andreas Dilger <adilger@clusterfs.com>
6751 * release version v0_4_3
6752 * Fixes statfs for inodes on extN.
6753 * Fixes bug in runtests which would delete /etc/hosts.
6754 * Use 64-bit object IDs wherever possible (not into VFS though)
6755 Remove ost_get_info, which is unused by lustre, and out of date.
6757 2002-07-03 Peter Braam <braam@clusterfs.com>
6758 * release version v0_4_2 Fixes a lookup error (type not passed)
6759 * move forward to head of Portals
6760 * move forward to latest Lustre kernel
6762 2002-06-25 Peter Braam <braam@clusterfs.com>
6763 * release version v0_4_1. Hopefully stable on single node use.