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 3.10.0-1062.9.1.el7 (RHEL7.7)
7 3.10.0-1127.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 4.4.120-92.70 (SLES12 SP2)
14 4.4.180-94.100 (SLES12 SP3)
15 4.4.0-131 (Ubuntu 16.04)
16 4.15.0-32 (Ubuntu 18.04)
17 vanilla linux 4.6.7 (ZFS only)
18 * ldiskfs needs an ldiskfs patch series for that kernel, ZFS does not
19 * Client primary unpatched kernels built and tested during release cycle:
20 3.10.0-1062.9.1.el7 (RHEL7.7)
21 3.10.0-1127.el7 (RHEL7.8)
22 4.12.14-95.45 (SLES12 SP4)
23 4.12.14-197.29 (SLES15 SP1)
24 4.15.0-48 (Ubuntu 18.04)
25 * Other clients known to build on unpatched kernels at some point (others may also work):
26 3.10.0-862.14.4.el7 (RHEL7.5)
27 3.10.0-957.27.2.el7 (RHEL7.6)
28 4.14.0-49.13.1.el7a (RHEL7.5)
29 4.14.0-115.2.2.el7a (RHEL7.6)
30 4.18.0-80.11.2.el8 (RHEL8.0)
31 4.18.0-147.8.1.el8 (RHEL8.1)
32 4.4.120-92.70 (SLES12 SP2)
33 4.4.180-94.100 (SLES12 SP3)
34 4.4.0-131 (Ubuntu 16.04)
35 vanilla linux 4.15.0 (ZFS only)
36 * Recommended e2fsprogs version: 1.45.2.wc1 or newer
37 * Recommended ZFS version: 0.8.3
38 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
39 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
40 stack overflow. For more information, please refer to bugzilla 17630.
41 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
42 the re-exporting nodes to be mounted with "32bitapi" mount option
43 * Note for Ubuntu 14.04.5:
44 The 4.4.0 series is the HWE Kernel derived from 16.04 (Xenial).
45 See: https://wiki.ubuntu.com/Kernel/LTSEnablementStack
47 --------------------------------------------------------------------------------
51 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
52 for currently supported client and server kernel versions.
53 * Server primary kernels built and tested during release cycle:
54 3.10.0-1062.1.1.el7 (RHEL7.7)
55 * Other server kernels known to build and work at some point (others may also work):
56 3.10.0-862.14.4.el7 (RHEL7.5)
57 3.10.0-957.27.2.el7 (RHEL7.6)
58 4.18.0-80.7.1.el8 (RHEL8.0)
59 4.4.120-92.70 (SLES12 SP2)
60 4.4.180-94.100 (SLES12 SP3)
61 4.4.0-131 (Ubuntu 16.04)
62 4.15.0-32 (Ubuntu 18.04)
63 vanilla linux 4.6.7 (ZFS only)
64 * ldiskfs needs an ldiskfs patch series for that kernel, ZFS does not
65 * Client primary unpatched kernels built and tested during release cycle:
66 3.10.0-1062.1.1.el7 (RHEL7.7)
67 4.12.14-95.29 (SLES12 SP4)
68 4.15.0-48 (Ubuntu 18.04)
69 * Other clients known to build on unpatched kernels at some point (others may also work):
70 3.10.0-862.14.4.el7 (RHEL7.5)
71 3.10.0-957.27.2.el7 (RHEL7.6)
72 4.14.0-49.13.1.el7a (RHEL7.5)
73 4.14.0-115.2.2.el7a (RHEL7.6)
74 4.18.0-80.7.1.el8 (RHEL8.0)
75 4.4.120-92.70 (SLES12 SP2)
76 4.4.180-94.100 (SLES12 SP3)
77 4.4.0-131 (Ubuntu 16.04)
78 vanilla linux 4.15.0 (ZFS only)
79 * Recommended e2fsprogs version: 1.45.2.wc1 or newer
80 * Recommended ZFS / SPL version: 0.7.13
81 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
82 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
83 stack overflow. For more information, please refer to bugzilla 17630.
84 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
85 the re-exporting nodes to be mounted with "32bitapi" mount option
86 * Note for Ubuntu 14.04.5:
87 The 4.4.0 series is the HWE Kernel derived from 16.04 (Xenial).
88 See: https://wiki.ubuntu.com/Kernel/LTSEnablementStack
90 --------------------------------------------------------------------------------
94 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
95 for currently supported client and server kernel versions.
96 * Server known to build on the following kernels (others may also work):
97 2.6.32-696.18.7.el6 (RHEL6.9)
98 3.10.0-862.14.4.el7 (RHEL7.5)
99 3.10.0-957.el7 (RHEL7.6)
100 4.4.120-92.70 (SLES12 SP2)
101 4.4.162-94.69 (SLES12 SP3)
102 4.4.0-131 (Ubuntu 16.04)
103 4.15.0-32 (Ubuntu 18.04)
104 vanilla linux 4.6.7 (ZFS only)
105 * ldiskfs needs an ldiskfs patch series for that kernel, ZFS does not
106 * Client known to build on unpatched kernels (others may also work):
107 2.6.32-696.18.7.el6 (RHEL6.9)
108 2.6.32-754.6.3.el6 (RHEL6.10)
109 3.10.0-862.14.4.el7 (RHEL7.5)
110 3.10.0-957.el7 (RHEL7.6)
111 4.14.0-49.13.1.el7a (RHEL7.5)
112 4.14.0-115.2.2.el7a (RHEL7.6)
113 4.4.120-92.70 (SLES12 SP2)
114 4.4.162-94.69 (SLES12 SP3)
115 4.4.0-131 (Ubuntu 16.04)
116 4.15.0-32 (Ubuntu 18.04)
117 vanilla linux 4.15.0 (ZFS only)
118 * Recommended e2fsprogs version: 1.44.3.wc1 or newer
119 * Recommended ZFS / SPL version: 0.7.9
120 * Tested with ZFS / SPL version: 0.7.9
121 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
122 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
123 stack overflow. For more information, please refer to bugzilla 17630.
124 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
125 the re-exporting nodes to be mounted with "32bitapi" mount option
126 * Note for Ubuntu 14.04.5:
127 The 4.4.0 series is the HWE Kernel derived from 16.04 (Xenial).
128 See: https://wiki.ubuntu.com/Kernel/LTSEnablementStack
130 --------------------------------------------------------------------------------
132 2018-04-03 Intel Corporation
134 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
135 for currently supported client and server kernel versions.
136 * Server known to build on patched kernels:
137 2.6.32-431.29.2.el6 (RHEL6.5)
138 2.6.32-504.30.3.el6 (RHEL6.6)
139 2.6.32-573.26.1.el6 (RHEL6.7)
140 2.6.32-642.15.1.el6 (RHEL6.8)
141 2.6.32-696.18.7.el6 (RHEL6.9)
142 3.10.0-693.21.1.el7 (RHEL7.4)
143 3.0.101-0.47.71 (SLES11 SP3)
144 3.0.101-107 (SLES11 SP4)
145 3.12.74-60.64.40 (SLES12 SP1)
146 4.4.114-92.64 (SLES12 SP2)
147 4.4.114-94.11 (SLES12 SP3)
148 3.13.0-101 (Ubuntu 14.04, ZFS only)
149 4.4.0-85.108 (Ubuntu 14.04.5 LTS)
150 4.4.0-116 (Ubuntu 16.04)
151 vanilla linux 4.6.7 (ZFS only)
152 * Client known to build on unpatched kernels:
153 2.6.32-431.29.2.el6 (RHEL6.5)
154 2.6.32-504.30.3.el6 (RHEL6.6)
155 2.6.32-573.26.1.el6 (RHEL6.7)
156 2.6.32-642.15.1.el6 (RHEL6.8)
157 2.6.32-696.18.7.el6 (RHEL6.9)
158 3.10.0-693.21.1.el7 (RHEL7.4)
159 3.0.101-0.47.71 (SLES11 SP3)
160 3.0.101-107 (SLES11 SP4)
161 3.12.74-60.64.40 (SLES12 SP1)
162 4.4.114-92.64 (SLES12 SP2)
163 4.4.114-94.11 (SLES12 SP3)
164 3.13.0-101 (Ubuntu 14.04)
165 4.4.0-85.108 (Ubuntu 14.04.5 LTS)
166 4.4.0-116 (Ubuntu 16.04)
168 * Recommended e2fsprogs version: 1.42.13.wc6 or newer
169 * Recommended ZFS / SPL version: 0.7.6
170 * Tested with ZFS / SPL version: 0.7.6
171 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
172 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
173 stack overflow. For more information, please refer to bugzilla 17630.
174 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
175 the re-exporting nodes to be mounted with "32bitapi" mount option
176 * Note for Ubuntu 14.04.5:
177 The 4.4.0 series is the HWE Kernel derived from 16.04 (Xenial).
178 See: https://wiki.ubuntu.com/Kernel/LTSEnablementStack
180 --------------------------------------------------------------------------------
182 2017-07-05 Intel Corporation
184 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
185 for currently supported client and server kernel versions.
186 * Server known to build on patched kernels:
187 2.6.32-431.29.2.el6 (RHEL6.5)
188 2.6.32-504.30.3.el6 (RHEL6.6)
189 2.6.32-573.26.1.el6 (RHEL6.7)
190 2.6.32-642.15.1.el6 (RHEL6.8)
191 2.6.32-696.3.1.el6 (RHEL6.9)
192 3.10.0-514.21.1.el7 (RHEL7.3)
193 3.0.101-0.47.71 (SLES11 SP3)
194 3.0.101-100 (SLES11 SP4)
195 3.12.74-60.64.40 (SLES12 SP1)
196 4.4.59-92.17 (SLES12 SP2)
197 vanilla linux 4.6.7 (ZFS only)
198 * Client known to build on unpatched kernels:
199 2.6.32-431.29.2.el6 (RHEL6.5)
200 2.6.32-504.30.3.el6 (RHEL6.6)
201 2.6.32-573.26.1.el6 (RHEL6.7)
202 2.6.32-642.15.1.el6 (RHEL6.8)
203 2.6.32-696.3.1.el6 (RHEL6.9)
204 3.10.0-514.21.1.el7 (RHEL7.3)
205 3.0.101-0.47.71 (SLES11 SP3)
206 3.0.101-100 (SLES11 SP4)
207 3.12.74-60.64.40 (SLES12 SP1)
208 4.4.59-92.17 (SLES12 SP2)
210 * Recommended e2fsprogs version: 1.42.13.wc4 or newer
211 * Recommended ZFS / SPL version: 0.6.5.9
212 * Tested with ZFS / SPL version: 0.6.5.9
213 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
214 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
215 stack overflow. For more information, please refer to bugzilla 17630.
216 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
217 the re-exporting nodes to be mounted with "32bitapi" mount option
219 --------------------------------------------------------------------------------
221 12-07-2016 Intel Corporation
223 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
224 for currently supported client and server kernel versions.
225 * Server known to build on patched kernels:
226 2.6.32-431.29.2.el6 (RHEL6.5)
227 2.6.32-504.30.3.el6 (RHEL6.6)
228 2.6.32-573.26.1.el6 (RHEL6.7)
229 2.6.32-642.11.1.el6 (RHEL6.8)
230 3.10.0-514.el7 (RHEL7.3)
231 3.0.101-0.47.71 (SLES11 SP3)
232 3.0.101-80 (SLES11 SP4)
233 3.12.62-60.64.8 (SLES12 SP1)
234 vanilla linux 4.5.7 (ZFS only)
235 * Client known to build on unpatched kernels:
236 2.6.32-431.29.2.el6 (RHEL6.5)
237 2.6.32-504.30.3.el6 (RHEL6.6)
238 2.6.32-573.26.1.el6 (RHEL6.7)
239 2.6.32-642.11.1.el6 (RHEL6.8)
240 3.10.0-514.el7 (RHEL7.3)
241 3.0.101-0.47.71 (SLES11 SP3)
242 3.0.101-80 (SLES11 SP4)
243 3.12.62-60.64.8 (SLES12 SP1)
245 * Recommended e2fsprogs version: 1.42.13.wc5 or newer
246 * Recommended ZFS / SPL version: 0.6.5.8
247 * Tested with ZFS / SPL version: 0.6.5.8
248 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
249 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
250 stack overflow. For more information, please refer to bugzilla 17630.
251 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
252 the re-exporting nodes to be mounted with "32bitapi" mount option
254 --------------------------------------------------------------------------------
256 02-29-2016 Intel Corporation
258 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
259 for currently supported client and server kernel versions.
260 * Server known to build on patched kernels:
261 2.6.32-431.29.2.el6 (RHEL6.5)
262 2.6.32-504.30.3.el6 (RHEL6.6)
263 2.6.32-573.12.1.el6 (RHEL6.7)
264 3.10.0-327.3.1.el7 (RHEL7.2)
265 3.0.101-0.47.71 (SLES11 SP3)
266 3.0.101-68 (SLES11 SP4)
267 vanilla linux 4.2.1 (ZFS only)
268 * Client known to build on unpatched kernels:
269 2.6.32-431.29.2.el6 (RHEL6.5)
270 2.6.32-504.30.3.el6 (RHEL6.6)
271 2.6.32-573.12.1.el6 (RHEL6.7)
272 3.10.0-327.3.1.el7 (RHEL7.2)
273 3.0.101-0.47.71 (SLES11 SP3)
274 3.0.101-68 (SLES11 SP4)
277 * Recommended e2fsprogs version: 1.42.13.wc4 or newer
278 * Recommended ZFS / SPL version: 0.6.4.2
279 * Tested with ZFS / SPL version: 0.6.4.2
280 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
281 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
282 stack overflow. For more information, please refer to bugzilla 17630.
283 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
284 the re-exporting nodes to be mounted with "32bitapi" mount option
286 --------------------------------------------------------------------------------
288 03-10-2015 Intel Corporation
290 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
291 for currently supported client and server kernel versions.
292 * Server known to build on patched kernels:
293 2.6.32-431.29.2.el6 (RHEL6.5)
294 2.6.32-504.8.1.el6 (RHEL6.6)
295 3.0.101-0.46 (SLES11 SP3)
296 * Client known to build on unpatched kernels:
297 2.6.32-431.29.2.el6 (RHEL6.5)
298 2.6.32-504.8.1.el6 (RHEL6.6)
299 3.10.0-123.20.1.el7 (RHEL7)
300 3.0.101-0.46 (SLES11 SP3)
301 * Recommended e2fsprogs version: 1.42.9.wc1 or newer
302 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
303 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
304 stack overflow. For more information, please refer to bugzilla 17630.
305 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
306 the re-exporting nodes to be mounted with "32bitapi" mount option
308 Severity : enhancement
310 Description: control OST-index in IDIF via ROCOMPAT flag.
311 Details : Introduce new flag OBD_ROCOMPAT_IDX_IN_IDIF that is stored in the
312 last_rcvd file. For new formatted OST device, it will be auto set;
313 for the case of upgrading from old OST device, you can enable it
314 via the lproc interface osd-ldiskfs.index_in_idif. With such flag
315 enabled, for new created OST-object, its IDIF-in-LMA will contain
316 the OST-index; for the existing OST-object, the OSD will convert
317 old format IDIF as new format IDIF with OST-index stored in the
318 LMA EA when accessing such OST-object or via OI scrub. Once such
319 flag is enabled, it cannot be reverted back, so the system cannot
320 be downgraded to the orignal incompatible version.
322 --------------------------------------------------------------------------------
324 07-30-2014 Intel Corporation
326 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
327 for currently supported client and server kernel versions.
328 * Server known to build on patched kernels:
329 2.6.32-431.20.3.el6 (RHEL6.5)
330 3.0.101-0.31 (SLES11 SP3)
331 * Client known to build on unpatched kernels:
332 2.6.32-431.20.3.el6 (RHEL6.5)
333 3.0.101-0.31 (SLES11 SP3)
336 * Recommended e2fsprogs version: 1.42.9.wc1 or newer
337 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
338 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
339 stack overflow. For more information, please refer to bugzilla 17630.
340 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
341 the re-exporting nodes to be mounted with "32bitapi" mount option
343 --------------------------------------------------------------------------------
345 10-31-2013 Intel Corporation
347 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
348 * Server builds for kernels:
349 2.6.32-358.18.1.el6 (RHEL6.4)
350 * Client builds for unpatched kernels:
351 2.6.32-358.18.1.el6 (RHEL6.4)
352 3.0.92-0.8 (SLES11 SP3)
354 * Recommended e2fsprogs version: 1.42.7.wc2 or newer
355 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
356 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
357 stack overflow. For more information, please refer to bugzilla 17630.
358 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
359 the re-exporting nodes to be mounted with "32bitapi" mount option
361 --------------------------------------------------------------------------------
363 05-30-2013 Intel Corporation
365 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
366 * Server builds for kernels:
367 2.6.32-358.6.2.el6 (RHEL6.4)
368 * Client builds for unpatched kernels:
369 2.6.32-358.6.2.el6 (RHEL6.4)
370 3.0.74-0.6.6 (SLES11 SP2)
372 * Recommended e2fsprogs version: 1.42.7.wc1 or newer
373 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
374 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
375 stack overflow. For more information, please refer to bugzilla 17630.
376 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
377 the re-exporting nodes to be mounted with "32bitapi" mount option
379 --------------------------------------------------------------------------------
381 9-30-2012 Whamcloud, Inc.
383 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
384 * Server support for kernels:
385 2.6.32-279.5.1.el6 (RHEL6)
386 * Client support for unpatched kernels:
387 2.6.32-279.5.1.el6 (RHEL6)
389 2.6.18-238.19.1.el5 (RHEL5)
390 2.6.16.60-0.69.1 (SLES10)
391 * Recommended e2fsprogs version: 1.41.90.wc4
392 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
393 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
394 stack overflow. For more information, please refer to bugzilla 17630.
395 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
396 the re-exporting nodes to be mounted with "32bitapi" mount option
398 Severity : enhancement
400 Description: Update RHEL6.3 kernel to 2.6.32-279.5.1.el6.
402 Severity : enhancement
404 Description: Update RHEL6.3 kernel to 2.6.32-279.2.1.el6.
406 Severity : enhancement
408 Description: Update RHEL6.2 kernel to 2.6.32-220.17.1.el6
410 Severity : enhancement
412 Description: Update RHEL6.2 kernel to 2.6.32-220.13.1.el6
414 Severity : enhancement
416 Description: Update RHEL6.2 kernel to 2.6.32-220.7.1.el6
418 --------------------------------------------------------------------------------
420 04-30-2012 Whamcloud, Inc.
422 * See https://wiki.whamcloud.com/display/PUB/Lustre+Support+Matrix
423 * Server support for kernels:
424 2.6.32-220.4.2.el6 (RHEL6)
425 * Client support for unpatched kernels:
426 2.6.32-220.4.2.el6 (RHEL6)
428 2.6.18-238.19.1.el5 (RHEL5)
429 2.6.16.60-0.69.1 (SLES10)
430 * Recommended e2fsprogs version: 1.41.90.wc4
431 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
432 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
433 stack overflow. For more information, please refer to bugzilla 17630.
434 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
435 the re-exporting nodes to be mounted with "32bitapi" mount option
437 Severity : enhancement
439 Description: Update RHEL6.2 kernel to 2.6.32-220.4.2.el6
441 Severity : enhancement
443 Description: Update RHEL6.2 kernel to 2.6.32-220.4.1.el6
445 Severity : enhancement
447 Description: Update RHEL6.2 kernel to 2.6.32-220.el6
449 Severity : enhancement
451 Description: Update RHEL6.1 kernel to 2.6.32-131.17.1.el6
453 --------------------------------------------------------------------------------
455 09-30-2011 Whamcloud, Inc.
457 * Server support for kernels:
458 2.6.18-238.19.1.el5 (RHEL5)
459 2.6.32-131.6.1.el6 (RHEL6)
460 * Client support for unpatched kernels:
461 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
462 2.6.18-238.19.1.el5 (RHEL5)
463 2.6.32-131.6.1.el6 (RHEL6)
464 2.6.29.4-167.fc11 (FC11)
465 2.6.16.60-0.69.1 (SLES10)
467 * Recommended e2fsprogs version: 1.41.90.wc3
468 * NFS export disabled when stack size < 8192. Since the NFSv4 export of
469 Lustre filesystem with 4K stack may cause a stack overflow. For more
470 information, please refer to bugzilla 17630.
471 * NFSv4 reexport requires Lustre client on the re-exporting nodes to be
472 mounted with "32bitapi" mount option
474 Severity : enhancement
475 Description: Update RHEL6.1 kernel to 2.6.32-131.6.1.el6
477 Severity : enhancement
478 Description: Update RHEL5.6 kernel to 2.6.18-238.19.1.el5
480 Severity : enhancement
481 Description: Update RHEL5.6 kernel to 2.6.18-238.9.1.el5
484 Frequency : only with SLES10
486 Description: Use OFED "KMP" provided by Novell
487 Details : SLES10 SP3 ships with OFED in a separate "KMP" package.
488 Lustre is now built against this package. That means you need to
489 install the ofed-kmp package from Novell for the patchless client.
490 Note that the ofed-kmp that Novell ships may not exactly match the
491 kernel version but should still be compatible.
493 Severity : enhancement
495 Description: Update RHEL5.5 kernel to 2.6.18-194.17.1.el5
496 Update OEL5.5 kernel to 2.6.18-194.17.1.0.1.el5.
498 --------------------------------------------------------------------------------
500 2010-07-15 Oracle, Inc.
502 * Server support for kernels:
503 2.6.18-164.11.1.el5 (RHEL 5)
504 2.6.18-164.11.1.0.1.el5 (OEL 5)
505 * Client support for unpatched kernels:
506 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
507 2.6.18-164.11.1.el5 (RHEL 5),
508 2.6.18-164.11.1.0.1.el5 (OEL 5)
509 2.6.16.60-0.42.8 (SLES 10),
511 2.6.29.4-167.fc11 (FC11)
512 * Recommended e2fsprogs version: 1.41.10-sun2
514 Severity : enhancement
516 Description: Distribute lfs_migrate script with lustre
517 Details : The lfs_migrate mentioned in the manual is now included in the
518 lustre tree in lustre/scripts and will distributed in lustre RPMs.
520 Severity : enhancement
522 Description: Failover nid list added to import proc file.
524 Severity : enhancement
526 Description: "som_preview" mount option required to use SOM
527 Details : The clients will need to use som_preview mount option
531 Description: Lustre does not do 1MB IOs to HW RAID
532 Details : Bump MAX_PHYS/HW_SEGMENTS and SG_ALL to 256 in the RHEL5 kernel.
534 Severity : enhancement
536 Description: lfs quota output is not very convenient for awk/sed-parsing
537 Details : Some positions in lfs quota output table could be empty or
538 non-empty which made it hard to parse it with scripts, now
539 a dash is put instead of space where there is not supposed
544 Description: Update to OFED 1.5.1
545 Details : OFED 1.5.1 will be used in environments that don't use
546 vendor supplied OFED.
550 Description: Remove lnet. prefix from lctl params display
551 Details : Also renamed lnet.memused to lnet_memused to
556 Description: Use conf_param -d to remove permanant settings.
560 Description: Add list_param -R
561 Details : lctl list_param now supports recursive mode.
564 Bugzilla : 21829, 22312
565 Description: Fix llobdstat.
566 Details : 1. Now updating the write_bytes counter
567 2. Port script fixes from 1.8.
571 Description: decrease the usage of memory on clients.
572 Details : 1. On clients, recycle dentries and inodes unused.
573 2. Delete the code related to ll_deathrow(att 6215 in bug 1443). It
578 Description: llobdstat fix and enhancement
579 Details : add a counter to set a limit to how many samples will be returned
580 fix a wildcard in the path to limit to obdfilter stats only
584 Description: lfs check only allowed for root
585 Details : Code cleanups and added a regresion test of non-root use of
590 Description: MDS Oopses in ptlrpc_dispatch_difficult_reply()/__list_add()
591 Details : Fix hr->hr_index overload.
595 Description: Fix FIEMAP functionality.
599 Description: FC11 patchless client support added
603 Description: Update RHEL5.4 kernel to 2.6.18-164.11.1.el5 and
604 OEL5.4 kernel to 2.6.18-164.11.1.0.1.el5.
608 Description: New acc-sm script and various test-framework enhancments.
612 Description: Previously landed patch was reverted.
616 Description: Update SLES10 kernel to 2.6.16.60-0.42.8 (patchless)
620 Description: Add ldiskfs maxdirsize mount option.
624 Description: Fix mkfs.lustre support for 16TB LUNs.
628 Description: Print mmp_check_interval
629 Details : print mmp_check_interval and make it possible to abort
630 mount operation in case it takes too long.
634 Description: Add cascading_rw to lustre mpi tests.
638 Description: Store SOM data in EA.
639 Details : Also several SOM related cleanups.
643 Description: Add refcount to md_open_data
647 Description: Use refcount on lov_request_set to prevent memory corruption.
651 Description: Statahead: use parent dir's i_mutex for synchronization.
655 Description: loadgen improvements.
659 Description: Enable peer health detection in ksocklnd and ko2iblnd.
663 Description: Control DCACHE_LUSTRE_INVALID flag with MDS_INODE_LOOKCUP lock
667 Description: Fix race on obd_devcie:md_stats
671 Description: Fix SLES 11 client builds.
672 Details : Resolved uninitalized pointer gcc warnings.
676 Description: Allow non-root access for "lfs check".
677 Details : Added a check in obd_class_ioctl() for OBD_IOC_PING_TARGET.
681 Description: Clear imp_force_reconnect correctly in ptlrpc_connect_interpret()
683 Severity : enhancement
685 Description: Update kernel to RHEL5.4 2.6.18-164.6.1.el5 and
686 OEL5 2.6.18-164.6.1.0.1.el5(Both in-kernel OFED enabled).
690 Description: Sanity test_27p fails.
691 Details : OBD_FAIL_OST_ENOSPC has been moved to filter_statfs().
692 Sanity tests 27 were cleaned up and improved.
696 Description: Unify req format on client/servers
697 Details : Use new req_capsule API [almost] everywhere instead of old PTLRPC
698 buffers and swabbers approach..
703 Description: drop join file stuff.
705 Severity : enhancement
707 Description: Add 2.6.27 and SLES11 (patchless client) support
711 Description: Fix in ptlrpc_expire_one_request() to print the signed time
714 Severity : enhancement
716 Description: Build kernels (RHEL5, OEL5 and SLES10/11) using the vendor's own
719 Severity : enhancement
721 Description: Remove set_info(KEY_UNLINKED) from MDS/OSC
723 Severity : enhancement
725 Description: correctly handle big reply message.
726 Details : send LNet event if reply is bigger than buffer and adjust this buffer
731 Description: Drop unnecessary __GFP_NOMEMALLOC flag from filter_get_page()
733 Severity : enhancement
735 Description: Add support for OEL5.
739 Description: Conf-sanity.sh 50g test - deactivated OST should not cause a panic.
741 Severity : enhancement
743 Description: "weak-modules" support
744 Details : Implement "weak-modules" support which enables kernel modules
745 to be used with any kernel that implements the same kABI. In
746 order to achieve this modules are now installed in
747 /lib/modules/$(uname -r)/updates/kernel on all distributions.
751 Description: client could not reconnect to OST because of an active request.
752 Details : The client now retries to reconnect to the same server, if a
753 connect request failed with EBUSY or -EAGAIN. Test case 35b
754 conf-sanity.sh has been created.
758 Description: Don't allow make backward step on assignin osc next id.
759 Details : race between allocation next id and ll_sync thread can be cause
760 of set wrong osc next id and can be kill valid ost objects.
764 Description: Changes in raid5-large-io-rhel5.patch to calculate sectors properly
768 Description: Increase the default BLK_DEF_MAX_SECTORS value for RHEL5 and SLES11
772 Description: Error handling in osc_statfs_interpret() has been improved.
773 Details : Check in osc_statfs_interpret() for EBADR.
777 Description: Do not send statfs() requests to OSTs disabled by administrator.
778 Details : Check in lov_prep_statfs_set() for non-NULL ltd_exp.
780 Severity : enhancement
782 Description: Added sanity-quota test to for lfs(1) to work with numeric uid/gid.
784 Severity : enhancement
786 Description: Update kernel to RHEL5 2.6.18-128.7.1.el5.
788 Severity : enhancement
790 Description: Enhancement for lfs(1) command to use numeric uid/gid.
794 Description: Improved cleanup procedure for conf-sanity test_46a.
798 Description: Do not start bulk transfer, if client tried to reconnect.
799 Details : Do not start bulk transfer, if client tried to reconnect,
800 but failed due to active rpcs. Check for exp_abort_active_req
801 has been implemented.
805 Description: Prevent inconsistences between linux and lustre mount structures.
806 Details : Wait indefinitely in server_wait_finished() until mnt_count drops.
807 Make the sleep interruptible.
809 Severity : enhancement
811 Description: provide server to client comms path
812 Details : server to client communications path via new LDLM_SET_INFO rpc
816 Description: truncate starts GFP_FS allocation under transaction causing deadlock
817 Details : ldiskfs_truncate calls grab_cache_page which may start page
818 allocation under an open transaction. This may lead to
819 calling prune_icache with consequent lustre reentrance.
823 Description: send correctly lsm on open replay
824 Details : MDS is trust to LSM size on replay open, but client can set wrong size
829 Description: enable client interface failover
830 Details : When a child reconnects from another NID, properly update export
831 nid hash position and ldlm reverse import.
833 Severity : enhancement
835 Description: Communicate OST degraded/readonly state via statfs to MDS
836 Details : Flags in the statfs returned from OSTs indicate whether the
837 OST is in a degraded RAID state, or if the filesystem has
838 turned read-only after a filesystem error is detected.
843 Description: don't panic if EPROTO was hit when reading symlink
844 Details : correctly handling request reference in error cases.
846 Severity : enhancement
848 Description: Add LustreNetLink, a kernel-userspace communcation path.
850 Severity : enhancement
852 Description: Update kernel to SLES10 SP2 2.6.16.60-0.39.3.
857 Description: access to llog context before init.
858 Details : move handling CATALOGS file at osc layer and forbid access to llog
862 Frequency : rare, only when client is writing to OST on local node
864 Description: Avoid deadlock for local client writes
865 Details : Use new OBD_BRW_MEMALLOC flag to notify OST about writes in the
866 memory freeing context. This allows OST threads to set the
867 PF_MEMALLOC flag on task structures in order to allocate memory
868 from reserved pools and complete IO.
869 Use GFP_HIGHUSER for OST allocations for non-local client writes,
870 so that the OST threads generate memory pressure and allow
871 inactive pages to be reclaimed.
873 Severity : enhancement
875 Description: Update kernel to RHEL5.3 2.6.18-128.1.14.el5.
880 Description: lock ordering violation between &cli->cl_sem and _lprocfs_lock
881 Details : move ldlm namespace creation in setup phase to avoid grab
882 _lprocfs_lock with cli_sem held.
886 Description: Temporarily disable grant shrink.
887 Details : Disable the feature for debugging.
890 Frequency : only when formatting test filesystems
892 Description: Unable to run several mkfs.lustre on loop devices at the same time
893 Details : mkfs.lustre returns error 256 on the concurrent loop devices
894 formatting. The solution is to proper handle the error.
896 Severity : enhancement
898 Description: Update kernel to RHEL5.3 2.6.18-128.1.6.el5.
900 Severity : enhancement
902 Description: Update kernel to SLES10 SP2 2.6.16.60-0.37.
907 Description: resolve race between obd_disconnect and class_disconnect_exports
908 Details : if obd_disconnect will be called to already disconnected export he
909 forget release one reference and osc module can't unloaded.
911 Severity : enhancement
913 Description: Allow tuning service thread via /proc
914 Details : For each service a new
915 /proc/fs/lustre/{service}/*/thread_{min,max,started} entry is
916 created that can be used to set min/max thread counts, and get the
917 current number of running threads.
922 Descriptoin: don't return error if have partially created objects for file.
923 Details : lov_update_create_set uses set->set_success as index for created
924 objects, so if some requests will be failed, they will have hole
925 at end of array and qos_shrink_lsm for allocate correct lsm.
927 Severity : enhancement
929 Description: Update OFED support to 1.4.1
934 Description: Reduce small size read RPC
935 Details : Set read-ahead limit for every file and only do read-ahead when
936 available read-ahead pages are bigger than 1M to avoid small size
939 Severity : enhancement
941 Description: move AT tunable parameters for more consistent usage
942 Details : add AT tunables under /proc/sys/lustre, add to conf_param parsing
944 Severity : enhancement
946 Description: add lazystatfs mount option to allow statfs(2) to skip down OSTs
947 Details : allow skip disconnected ost for send statfs request and hide error
953 Description: fix racy locking of mballoc block bitmaps causing BUG
954 Details : The locking of the mballoc buddy bitmap and the in-memory
955 block bitmap was using two different spin locks in some
956 cases. This made it possible to incorrectly access the
957 mballoc bitmap while another process was modifying it,
958 causing a sanity assertion to fail. While no on-disk corruption
959 was reported, there was some risk of this happening.
962 Frequency : rare, on llog test 6
964 Descriptoin: don't allow connect to already connected import
965 Details : allowing connect to already connected import is hide connecting problem.
970 Descriptoin: allow kill process which wait statahead result
971 Details : for some reasons 'ls' can stick in waiting result from statahead,
972 in this case need way for kill this process.
974 Severity : enhancement
976 Description: Add state history info file, enhance import info file
977 Details : Track import connection state changes in a new osc/mdc proc file;
978 add overview-type data to the osc/mdc import proc file.
980 Severity : enhancement
982 Description: MDS create should not wait for statfs RPC while holding DLM lock.
984 Severity : enhancement
986 Description: Update to RHEL5U3 kernel-2.6.18-128.1.1.el5.
991 Descriptoin: OST grant too much space to client even there are not enough space.
992 Details : Client will shrink its grant cache to OST if there are no write
993 activity over 6 mins (GRANT_SHRINK_INTERVAL), and OST will retrieve
994 this grant cache if there are already not enough avaible space
995 (left_space < total_clients * 32M).
998 Frequency : start MDS on uncleanly shutdowned MDS device
1000 Descriptoin: ll_sync thread stay in waiting mds<>ost recovery finished
1001 Details : stay in waiting mds<>ost recovery finished produce random bugs
1002 due race between two ll_sync thread for one lov target. send
1003 ACTIVATE event only if connect realy finished and import have
1007 Frequency : rare, connect and disconnect target at same time
1009 Descriptoin: ASSERTION(atomic_read(&imp->imp_inflight) == 0
1010 Details : don't call obd_disconnect under lov_lock. this long time
1011 operation and can block ptlrpcd which answer to connect request.
1016 Descriptoin: don't lose wakeup for imp_recovery_waitq
1017 Details : recover_import_no_retry or invalidate_import and import_close can
1018 both sleep on imp_recovery_waitq, but we was send only one wakeup
1022 Frequency : always with long access acl
1024 Descriptoin: mds can't pack reply with long acl.
1025 Details : mds don't control size of acl but they limited by reint/getattr
1028 Severity : enhancement
1030 Description: Update to SLES10 kernel-2.6.16.60-0.33.
1032 Severity : enhancement
1034 Description: Update to RHEL5 kernel-2.6.18-92.1.22.el5.
1037 Frequency : start MDS on uncleanly shutdowned MDS device
1039 Descriptoin: aborting recovery hang on MDS
1040 Details : don't throttle destroy RPCs for the MDT.
1043 Frequency : on remount
1045 Description: external journal device not working after the remount
1046 Details : clear dev_rdonly flag for external journal devices in
1052 Description: shutdown vs evict race
1053 Details : client_disconnect_export vs connect request race.
1054 if client will evicted at this time - we start invalidate
1055 thread without referece to import and import can be freed
1059 Frequency : rare, need acl's on inode.
1061 Description: client can't handle ost additional correctly
1062 Details : if ost was added after client connected to mds client can have
1063 hit lnet_try_match_md ... to big messages to wide striped files.
1064 in this case need teach client to handle config events about add
1065 lov target and update client max ea size at that event.
1067 Severity : enhancement
1069 Description: Changelogs
1070 Details : Changelogs are a lightweight mechanism to track filesystem
1071 metadata and namespace changes. The changelog is recorded
1072 permanently on the MDTs, and is periodically "consumed" / purged
1073 when records are no longer needed.
1075 Severity : enhancement
1077 Description: compact fld format with extents
1078 Details : Store range of seq rather than every seq in FLD. Seq
1079 controller update FLD rather than clients. In Case of CMD, mdt0
1080 has FLD, all other metadata server act as non persistent proxy
1081 for FLD queries and cache fld entries in fld cache.
1086 Description: don't skip ost target if they assigned to file
1087 Details : Drop slow OSCs if we can, but not for requested start idx.
1088 This means "if OSC is slow and it is not the requested
1089 start OST, then it can be skipped, otherwise skip it only
1090 if it is inactive/recovering/out-of-space.
1094 Description: more cleanup in mds_lov
1095 Details : not send LOV EA under replay, we can't know about they size at this
1096 time. Don't allow client connect to mds before any ost connected,
1097 for avoid problems with LOV EA size and returning EIO to client.
1099 Severity : enhancement
1101 Description: Interoperability at server side (Disk interoperability)
1103 Severity : enhancement
1105 Description: Update to RHEL5 kernel-2.6.18-92.1.17.el5.
1107 Severity : enhancement
1109 Description: Update to SLES10 SP2 kernel-2.6.16.60-0.31.
1111 Severity : enhancement
1113 Description: New client IO stack (CLIO).
1115 Severity : enhancement
1117 Description: Commit on sharing. Eliminate inter-client dependencies between
1118 uncommitted transactions by doing transaction commits.
1119 Thereby clients may recovery independently.
1122 Frequency : Create a symlink file with a very long name
1124 Description: ldlm_cancel_pack()) ASSERTION(max >= dlm->lock_count + count)
1125 Details : If there is no extra space in the request for early cancels,
1126 ldlm_req_handles_avail() returns 0 instead of a negative value.
1128 Severity : enhancement
1130 Description: Add /proc entry for import status
1131 Details : The mdc, osc, and mgc import directories now have
1132 an import directory that contains useful import data for debugging
1133 connection problems.
1135 Severity : enhancement
1137 Description: Re-disable certain /proc logging
1138 Details : Enable and disable client's offset_stats, extents_stats and
1139 extents_stats_per_process stats logging on the fly.
1142 Frequency : Only on FC kernels 2.6.22+
1144 Description: oops in statahead
1145 Details : Do not drop reference count for the dentry from VFS when lookup,
1146 VFS will do that by itself.
1148 Severity : enhancement
1150 Description: Generic /proc file permissions
1151 Details : Set /Proc file permissions in a more generic way to enable non-
1152 root users operate on some /proc files.
1156 Description: Hitting mdc_commit_close() ASSERTION
1157 Details : Properly handle request reference release in
1158 ll_release_openhandle().
1162 Description: quota recovery deadlock during mds failover
1163 Details : This patch includes att18982, att18236, att18237 in bz14840.
1165 1. fix osts hang when mds does failover with quotaon
1166 2. prevent watchdog storm when osts threads wait for the
1171 Frequency : only patchless client
1172 Description: add workaround for race between add/remove dentry from hash
1174 Severity : enhancement
1176 Description: Allow OST glimpses to return PW locks
1180 Description: LBUG when llog conf file is full
1181 Details : When llog bitmap is full, ENOSPC should be returned for plain log.
1185 Description: Prevent import from entering FULL state when server in recovery
1189 Description: service mount cannot take device name with ":"
1190 Details : Only when device name contains ":/" will mount treat it as
1196 Description: replace ptlrpcd with the statahead thread to interpret the async
1197 statahead RPC callback
1201 Frequency : on recovery
1202 Description: I/O failures after umount during fail back
1203 Details : if client reconnected to restarted server we need join to recovery
1204 instead of find server handler is changed and process self
1205 eviction with cancel all locks.
1207 Severity : enhancement
1209 Description: Update to RHEL5 kernel-2.6.18-92.1.10.el5.
1211 Severity : enhancement
1213 Description: Update to SLES10 SP2 kernel-2.6.16.60-0.27.
1215 Severity : enhancement
1217 Description: Upcall on Lustre log has been dumped
1218 Details : Allow for a user mode script to be called once a Lustre log has
1219 been dumped. It passes the filename of the dumped log to the
1220 script, the location of the script can be specified via
1221 /proc/sys/lnet/debug_log_upcall.
1226 Description: avoid idr_remove called for id which is not allocated.
1227 Details : Move assigment s_dev for clustered nfs to end of initialization,
1228 to avoid problem with error handling.
1233 Description: avoid Already found the key in hash [CONN_UNUSED_HASH] messages
1234 Details : When connection is reused this not moved from CONN_UNUSED_HASH
1235 into CONN_USED_HASH and this prodice warning when put connection
1236 again in unused hash.
1239 Severity : enhancement
1241 Description: File striping can now be set to use an arbitrary pool of OSTs.
1243 Severity : enhancement
1245 Description: Export bytes_read/bytes_write count on OSC/OST.
1249 Description: Early reply size mismatch, MGC loses connection
1250 Details : Apply the MGS_CONNECT_SUPPORTED mask at reconnect time so
1251 the connect flags are properly negotiated.
1256 Description: quotas are not honored with O_DIRECT
1257 Details : all writes with the flag O_DIRECT will use grants which leads to
1258 this problem. Now using OBD_BRW_SYNC to guard this.
1262 Description: add quota statistics
1263 Details : 1. sort out quota proc entries and proc code.
1264 2. add quota statistics
1266 Severity : enhancement
1268 Description: enable quota support for HEAD.
1272 Description: Properly propagate oinfo flags from lov to osc for statfs
1273 Details : restore missing copy oi_flags to lov requests.
1275 Severity : enhancement
1277 Description: Add man pages for llobdstat(8), llstat(8), plot-llstat(8),
1278 : l_getgroups(8), lst(8), routerstat(8)
1279 Details : included man pages for llobdstat(8), llstat(8),
1280 : plot-llstat(8), l_getgroups(8), lst(8), routerstat(8)
1282 Severity : enhancement
1284 Description: configure's --enable-quota should check the
1285 : kernel .config for CONFIG_QUOTA
1286 Details : configure is terminated if --enable-quota is passed but
1287 : no quota support is in kernel
1291 Description: Remove portals compatibility
1292 Details : Remove portals compatibility, not interoperable with releases
1297 Description: Resolve device initialization race
1298 Details : Prevent proc handler from accessing devices added to the
1299 obd_devs array but yet be intialized.
1301 Severity : enhancement
1303 Description: Update to SLES10 SP2 kernel-2.6.16.60-0.23.
1305 Severity : enhancement
1307 Description: Update to RHEL5 kernel-2.6.18-92.1.6.el5.
1312 Description: Using wrong pointer in osc_brw_prep_request
1313 Details : Access to array[-1] can produce panic if kernel compiled with
1314 CONFIG_PAGE_ALLOC enabled
1318 Description: Client runs out of low memory
1319 Details : Consider only lowmem when counting initial number of llap pages
1323 Description: *optional* service tags registration
1324 Details : if the "service tags" package is installed on a Lustre node
1325 When the filesystem is mounted, a local-node service tag will
1326 be created. See http://inventory.sun.com/ for more information
1327 about the Service Tags asset management system.
1331 Description: Kernel BUG tries to release flock
1332 Details : Lustre does not destroy flock lock before last reference goes
1333 away. So always drop flock locks when client is evicted and
1334 perform unlock regardless of successfulness of speaking to MDS.
1338 Description: add refcount protection for osc callbacks, avoid panic on shutdown
1342 Description: sanity test 65a fails if stripecount of -1 is set
1343 Details : handle -1 striping on filesystem in ll_dirstripe_verify
1348 Description: ASSERTION(CheckWriteback(page,cmd)) failed
1349 Details : badly clear PG_Writeback bit in ll_ap_completion can produce false
1352 Severity : enhancement
1354 Description: Update to RHEL5 kernel-2.6.18-53.1.21.el5.
1358 Description: do not process already freed flock
1359 Details : flock can possibly be freed by another thread before it reaches
1360 to ldlm_flock_completion_ast.
1364 Description: LBUG during stress test
1365 Details : Need properly lock accesses the flock deadlock detection list.
1369 Description: oops in page fault handler
1370 Details : kernel page fault handler can return two special 'pages' in error
1371 case, don't try dereference NOPAGE_SIGBUS and NOPAGE_OMM.
1375 Description: timeout with invalidate import.
1376 Details : ptlrpcd_check call obd_zombie_impexp_cull and wait request which
1377 should be handled by ptlrpcd. This produce long age waiting and
1378 -ETIMEOUT ptlrpc_invalidate_import and as result LASSERT.
1380 Severity : enhancement
1382 Description: Update to RHEL5 kernel-2.6.18-53.1.19.el5.
1386 Description: enable MGS and MDT services start separately
1387 Details : add a 'nomgs' option in mount.lustre to enable start a MDT with
1388 a co-located MGS without starting the MGS, which is a complement
1389 to 'nosvc' mount option.
1393 Frequency : after recovery
1394 Description: precreate to many object's after del orphan.
1395 Details : del orphan st in oscc last_id == next_id and this triger growing
1396 count of precreated objects. Set flag LOW to skip increase count
1397 of precreated objects.
1401 Frequency : rare, on clear nid stats
1402 Description: ASSERTION(client_stat->nid_exp_ref_count == 0)
1403 Details : when clean nid stats sometimes try destroy live entry,
1404 and this produce panic in free.
1408 Description: Stack overflow during MDS log replay
1409 ease stack pressure by using a thread dealing llog_process.
1413 Description: wait until IO finished before start new when do lock cancel.
1414 Details : VM protocol want old IO finished before start new, in this case
1415 need wait until PG_writeback is cleared until check dirty flag and
1416 call writepages in lock cancel callback.
1418 Severity : enhancement
1420 Description: using special macro for print time and cleanup in includes.
1424 Description: mds_mfd_close() ASSERTION(rc == 0)
1425 Details : In mds_mfd_close(), we need protect inode's writecount change
1426 within its orphan write semaphore to prevent possible races.
1430 Description: Obsolete CURRENT_SECONDS and use cfs_time_current_sec() instead.
1434 Frequency : rare, on shutdown ost
1435 Description: don't hit live lock with umount ost.
1436 Details : shrink_dcache_parent can be in long loop with destroy dentries,
1437 use shrink_dcache_sb instead.
1441 Description: don't panic with use echo client
1442 Details : echo client pass NULL as client nid pointer and this produce null
1443 pointer dereference.
1447 Description: fix build on ppc32
1448 Details : compile code with -m64 flag produce wrong object file for ppc32.
1452 Description: add message levels for liblustreapi
1456 Description: fix for occasional failure case of -ENOSPC in recovery-small tests
1457 Details : Move the 'good_osts' check before the 'total_bavail' check. This
1458 will result in an -EAGAIN and in the exit call path we call
1459 alloc_rr() which will with increasing aggressiveness attempt to
1460 aquire precreated objects on the minimum number of required OSCs.
1464 Description: Use old size assignment to avoid deadlock
1465 Details : Reverts the changes in bugs 2369 and bug 14138 that introduced
1466 the scheduling while holding a spinlock. We do not need locking
1467 for size in ll_update_inode() because size is only updated from
1468 the MDS for directories or files without objects, so there is no
1469 other place to do the update, and concurrent access to such inodes
1470 are protected by the inode lock.
1474 Description: resolve "_IOWR redefined" build error on SLES10
1478 Description: dump the memory debugging after all modules are unloaded to
1479 suppress false negative in conf_sanity test 39
1481 Severity : enhancement
1483 Description: build kernel-ib packages for OFED 1.3 in our release cycle
1488 Description: fix SLES kernel versioning
1489 Details : the kernel version for our SLES 10 kernel did not include a "-"
1490 before the "smp" at the end. while this was not a problem in
1491 general, it did mean that software trying to use the kernel
1492 version to try to detect a vendor specific kernel would fail.
1493 this was most evident by the OFED build scripts.
1497 Description: Don't update lov_desc members until making sure they are valid
1498 Details : When updating lov_desc members via proc fs, need fix their
1499 validities before doing the real update.
1503 Description: don't put request into delay list while invalidate in flight.
1504 Details : ptlrpc_delay_request sometimes put in delay list while invalidate
1505 import in flight. this produce timeout for invalidate and sometimes
1506 can cause stale data.
1510 Frequency : on ppc only
1511 Description: not convert ost objects for directory because it's not exist.
1512 Details : ll_dir_getstripe assume dirrectory has ost objects but this wrong.
1516 Description: Add FMODE_EXEC file flag for SLES10 SP1 kernel.
1518 Severity : enhancement
1520 Description: Update to support 2.6.22.14 vanilla kernel.
1524 Frequency : rare, on recovery
1525 Description: read procfs can produce deadlock in some situation
1526 Details : Holding lprocfs lock which send rpc can produce block for destroy
1527 obd objects and this also block reconnect with -EALREADY. This
1528 isn't fix all lprocfs bugs - but make it rare.
1530 Severity : enhancement
1532 Description: Update kernel to RHEL5 2.6.18-53.1.14.el5.
1535 Frequency : frequent on X2 node
1537 Description: mdc_set_open_replay_data LBUG
1538 Details : Set replay data for requests that are eligible for replay.
1542 Description: lustre_mgs: operation 101 on unconnected MGS
1543 Details : When MGC is disconnected from MGS long enough, MGS will evict the
1544 MGC, and late on MGC cannot successfully connect to MGS and a lot
1545 of the error messages complaining that MGS is not connected.
1549 Frequency : on network error
1550 Description: panic with double free request if network error
1551 Details : mdc_finish_enqueue is finish request if any network error ocuring,
1552 but it's true only for synchronus enqueue, for async enqueue
1553 (via ptlrpcd) this incorrect and ptlrpcd want finish request
1556 Severity : enhancement
1558 Description: client-side metadata stat-ahead during readdir(directory readahead)
1559 Details : perform client-side metadata stat-ahead when the client detects
1560 readdir and sequential stat of dir entries therein
1563 Frequency : on start mds
1565 Description: Implement get_info(last_id) in obdfilter.
1568 Frequency : occasional
1570 Description: Correctly check stale fid, not start epoch if ost not support SOM
1571 Details : open with flag O_CREATE need set old fid in op_fid3 because
1572 op_fid2 was overwritten with new generated fid, but mds can answer
1573 with one of these two fids and both is not stale. Setattr
1574 incorrectly started an epoch and assume will be called
1575 done_writing, but without SOM done_writing ever being called.
1578 Frequency : rare, depends on device drivers and load
1580 Description: MDS or OSS nodes crash due to stack overflow
1581 Details : Code changes in 1.8.0 increased the stack usage of some functions.
1582 In some cases, in conjunction with device drivers that use a lot
1583 of stack the MDS (or possibly OSS) service threads could overflow
1584 the stack. One change which was identified to consume additional
1585 stack has been reworked to avoid the extra stack usage.
1588 Frequency : occasional
1590 Description: Do not fail import if osc_interpret_create gets -EAGAIN
1591 Details : If osc_interpret_create got -EAGAIN it immediately exits and
1592 wakeup oscc_waitq. After wakeup oscc_wait_for_objects call
1593 oscc_has_objects and see OSC has no objests and call
1594 oscc_internal_create to resend create request.
1596 Severity : enhancement
1598 Description: Update to SLES10 SP1 latest kernel-2.6.16.54-0.2.5.
1600 Severity : enhancement
1602 Description: Update to RHEL5 latest kernel-2.6.18-53.1.13.el5.
1605 Frequency : very rare
1607 Description: Fix replay if there is an un-replied request and open
1608 Details : In some cases, older replay request will revert the
1609 mcd->mcd_last_xid on MDS which is used to record the client's
1610 latest sent request.
1612 Severity : enhancement
1614 Description: Update to RHEL5 latest kernel-2.6.18-53.1.6.el5.
1616 Severity : enhancement
1618 Description: Add rhel5 support to HEAD.
1620 Serverity : enhancement
1622 Description: Update RHEL4 kernel to 2.6.9-67.0.4.
1627 Description: Don't allow skipping OSTs if index has been specified.
1628 Details : Don't allow skipping OSTs if index has been specified, make
1629 locking in internal create lots better.
1633 Description: LBUG in ptlrpc_check_set() bad phase ebc0de00
1634 Details : access to bitfield in structure is always rounded to long
1635 and this produce problem with not atomic change any bit.
1639 Description: Lustre make rpms failed.
1640 Details : Remove ldiskfs spec file to avoids rpmbuild be confused when
1641 builds Lustre rpms from tarball.
1644 Frequency : rare on shutdown ost
1646 Description: If llog cancel was not send before clean_exports phase, this can
1647 produce deadlock in llog code.
1648 Details : If llog thread has last reference to obd and call class_import_put
1649 this produce deadlock because llog_cleanup_commit_master wait when
1650 last llog_commit_thread exited, but this never success because was
1651 called from llog_commit_thread.
1655 Description: allow userland application know is lost one of stripes.
1656 Details : fill lvb_blocks with error code on ost and return it to
1657 application if error flag found.
1661 Description: NULL lov_tgts causing MDS oops
1662 Details : more safe checks for NULL lov_tgts for avoid oops.
1664 Severity : enhancement
1666 Description: Update to RHEL4 latest kernel-2.6.9-67.0.1.EL.
1670 Descriptoin: make lov_create() will not stuck in obd_statfs_rqset()
1671 Details : If an OST is down the MDS will hang indefinitely in
1672 obd_statfs_rqset() waiting for the statfs data. While for
1673 MDS QOS usage of statfs, it should not stuck in waiting.
1675 Severity : enhancement
1677 Description: remote_acl support
1678 Details : Support ACL-based permission check for remote user.
1679 Support setfacl/getfacl for remote user with the utils
1680 "lfs {l,r}{s,g}etfacl" which follow the same parameter format as
1681 the system "{s,g}etfacl" utils.
1683 Severity : enhancement
1685 Description: Update to RHEL4 U6 kernel-2.6.9-67.EL.
1687 Severity : enhancement
1689 Description: Update to RHEL5 latest kernel-2.6.18-53.1.4.el5.
1693 Description: make mgs_setparam() handle fsname containing dash
1694 Details : fsname containing a dash does not work with lctl conf_param
1696 Severity : enhancement
1698 Description: Update to SLES10 SP1 latest kernel-2.6.16.54-0.2.3.
1700 Severity : enhancement
1702 Description: Update to RHEL5 Update-1 kernel 2.6.18-53.el5.
1706 Frequency : rare, at shutdown
1707 Description: access already free / zero obd_namespace.
1708 Details : if client_disconnect_export was called without force flag set,
1709 and exist connect request in flight, this can produce access to
1710 NULL pointer (or already free pointer) when connect_interpret
1711 store ocd flags in obd_namespace.
1715 Frequency : only at startup
1716 Description: not alloc memory with spinlock held.
1717 Details : allocation memory with GFP_KERNEL can produce sleep deadlock,
1718 if any spinlock held.
1720 Severity : enhancement
1722 Description: make lustre randomly fail allocating memory
1723 Details : Make lustre randomly failed allocating memory for testing purpose.
1725 Severity : enhancement
1727 Description: lost problems with lov objid file
1728 Details : Fixes some scability and access to not inited memory problems
1729 in work with lov objdid file.
1734 Description: lfs find does not continue on file error
1735 Details : Continue other files processing when a file/dir is absent.
1739 Description: Inconsistent usage of lustre_pack_reply()
1740 Details : Standardize the usage of lustre_pack_reply() such that it
1741 always generate a CERROR on failure.
1744 Frequency : occasional
1746 Description: MDS hang or stay in waiting lock
1747 Details : If client receive lock with CBPENDING flag ldlm need send lock
1748 cancel as separate rpc, to avoid situation when cancel request
1749 can't processed due all i/o threads stay in wait lock.
1753 Description: Update to RHEL5 kernel 2.6.18-8.1.15.el5.
1757 Description: Update to SLES10 SP1 kernel 2.6.16.53-0.16
1761 Description: Update to SLES9 kernel-2.6.5-7.287.3.
1765 Description: Update to RHEL4 latest kernel.
1767 Severity : enhancement
1769 Description: Build SLES10 patchless client fails
1770 Details : The configure was broken by run ./configure with
1771 --with-linux-obj=.... argument for patchless client. When the
1772 configure use --with-linux-obj, the LINUXINCLUDE= -Iinclude
1773 can't search header adequately. Use absolute path such as
1774 -I($LINUX)/include instead.
1778 Description: interrupt oig_wait produce painc on resend.
1779 Details : brw_redo_request can be used for resend requests from ptlrpcd and
1780 private set, and this produce situation when rq_ptlrpcd_data not
1781 copyed to new allocated request and triggered LBUG on assert
1782 req->rq_ptlrpcd_data != NULL. But this member used only for wakeup
1783 ptlrpcd set if request is changed and can be safety changed to use
1788 Description: LASSERT_{REQ,REP}SWAB macros are buggy
1789 Details : If SWAB_PARANOIA is disabled, the LASSERT_REQSWAB and
1790 LASSERT_REPSWAB macros become no-ops, which is incorrect. Drop
1791 these macros and replace them with their difinitions instead.
1795 Description: Update kernel patches for SLES10 2.6.16.53-0.8.
1796 Details : Update which_patch & target file for SLES10 latest kernel.
1799 Description: Remove client patches from SLES 10 kernel.
1800 Details : This causes SLES 10 clients to behave as patchless clients
1801 even on a Lustre-patched (server) kernel.
1803 Severity : enhancement
1805 Description: self-adjustable client's lru lists
1806 Details : use adaptive algorithm for managing client cached locks lru
1807 lists according to current server load, other client's work
1808 pattern, memory activities, etc. Both, server and client
1809 side namespaces provide number of proc tunables for controlling
1812 Severity : enhancement
1814 Description: light-weight GSS support
1815 Details : Support krb5n and krb5a mode, which keep Kerberos 5 authentication
1816 and reduce performance overhead.
1818 Severity : enhancement
1820 Description: Linux keyring support
1821 Details : Support using service of Linux keyring for Lustre GSS internal
1822 context refresh/cache mechanism.
1826 Description: Fix errors in lfs documentation
1827 Details : Fixes man pages
1831 Description: don't use GFP_* in generic Lustre code.
1832 Details : Use cfs_alloc_* functions and CFS_* flags for code portability.
1836 Description: obdclass is limited by single OBD_ALLOC(idarray)
1837 Details : replace OBD_ALLOC/OBD_FREE with OBD_VMALLOC/OBD_VFREE
1841 Description: warnings with build patchless client with vanila 2.6.19 and up
1842 Details : change old ctl_table style and replace ctl_table/ctl_table_header
1843 with cfs_sysctl_table_t/cfs_sysctl_table_header_t
1847 Frequency : Only for SLES
1848 Description: sanity_quota fail test_1
1849 Details : There are multiple occurences of $TSTUSR in SLES's /etc/group
1850 file, which makes TSTID[2] inunique.
1854 Frequency : Only for SLES9
1855 Description: Kernel patches for SLES9 2.6.5-7.286 kernel
1856 Details : Update target/ChangeLog/which_patch .
1860 Frequency : Only for test_10 in sanity-quota.sh
1861 Description: a bug in quota test code
1862 Details : checking a test flag in a wrong place
1866 Frequency : Only for test_13 in sanity-quota.sh
1867 Description: a bug in quota test script
1868 Details : shouldn't check overfull allocated quota
1872 Description: jbd statistics
1873 Details : Port older jbd statistics patch for sles10
1875 Severity : enhancement
1877 Description: Add printing inode info into message about error in writepage.
1882 Description: reply_lock_interpret crash due to race with it and lock cancel.
1883 Details : Do not replay locks that are being cancelled. Do not reference
1884 locks by their address during replay, just by their handle.
1888 Frequency : When flocks are used.
1889 Description: assertion failure in ldlm_cli_enquque_fini for non NULL lock.
1890 Details : Flock locks might destroy just granted lock if it could be merged
1891 with another existing flock, this is done in completion handler,
1892 so teach ldlm_cli_enquque_fini that this is a valid case for
1898 Description: Oops in read and write path when failing to allocate lock.
1899 Details : Check if lock allocation failed and return error back.
1903 Description: lstripe command fails for valid OST index
1904 Details : The stripe offset is compared to lov->desc.ld_tgt_count
1905 instead of lov->desc.ld_active_tgt_count.
1909 Description: sanity.sh failed test 103
1910 Details : RHEL mis-interpret setfacl "-X" param, so we won't test setfacl
1915 Description: df doesn't work properly if diskfs blocksize != 4K
1916 Details : Choose biggest blocksize of OST's as the LOV's blocksize.
1920 Description: lfs find on -1 stripe looping in lsm_lmm_verify_common()
1921 Details : Avoid lov_verify_lmm_common() on directory with -1 stripe count.
1923 Severity : enhancement
1925 Description: Adaptive timeouts
1926 Details : RPC timeouts adapt to changing server load and network
1927 conditions to reduce resend attempts and improve recovery time.
1931 Description: llapi_file_create() does not allow some changes
1932 Details : add llapi_file_open() that allows specifying the mode and
1933 open flags, and also returns an open file handle.
1937 Description: merge and cleanup kernel patches.
1938 Details : Remove mnt_lustre_list in vfs_intent-2.6-rhel4.patch.
1942 Description: Add journal checksum support.(Kernel part)
1943 Details : The journal checksum feature adds two new flags i.e
1944 JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT and
1945 JBD2_FEATURE_COMPAT_CHECKSUM. JBD2_FEATURE_CHECKSUM flag
1946 indicates that the commit block contains the checksum for
1947 the blocks described by the descriptor blocks. Now commit
1948 record can be sent to disk without waiting for descriptor
1949 blocks to be written to disk. This behavior is controlled
1950 using JBD2_FEATURE_ASYNC_COMMIT flag.
1954 Description: OSS needs mutliple precreate threads
1955 Details : Add ability to start more than one create thread per OSS.
1959 Description: Sanity.sh test_65k failed: setstripe should have succeeded
1960 Details : execute lfs setstripe on client
1964 Description: mds_obd_create error creating tmp object
1965 Details : When the user sets quota on root, llog will be affected and can't
1966 create files and write files.
1969 Frequency : Always on ia64 patchless client, and possibly others.
1971 Description: Add EXPORT_SYMBOL check for node_to_cpumask symbol.
1972 Details : This allows the patchless client to be loaded on architectures
1973 without this export.
1977 Description: RedHat Update kernel for RHEL5
1978 Details : Add the kernel config file for RHEL5.
1982 Description: RedHat Update kernel for RHEL5
1983 Details : Modify the kernel config file more closer RHEL5.
1987 Description: Build failure against Centos5 (RHEL5)
1988 Details : Define PAGE_SIZE when it isn't present.
1992 Description: test_7 on sanity-quota.sh can't be used on separate machines
1993 Details : add facet to handle it
1997 Description: "ll_intent_file_open()) lock enqueue: err: -13" with nfs
1998 Details : with NFS, the anon dentry's parent was set to itself in
1999 d_alloc_anon(), so in MDS, we use rec->ur_fid1 to find the
2000 corresponding dentry other than use rec->ur_name.
2002 Severity : enhancement
2004 Description: lfs setstripe enhancement
2005 Details : Make lfs setstripe understand 'k', 'm' and 'g' for stripe size.
2009 Description: enable data checksumming by default
2010 Details : enable checksum by default, allow --disable-checksum
2011 configure option and "-o nochecksum" mount option. Checksums
2012 can also be disabled at runtime via $LPROC/osc/*/checksum_pages.
2016 Description: lustre support for RHEL5
2017 Details : Add support for RHEL5.
2021 Description: Client eviction due to failover config
2022 Details : after a connection loss, the lustre client should attempt to
2023 reconnect to the last active server first before trying the
2024 other potential connections.
2028 Description: when mds and osts use different quota unit(32bit and 64bit),
2029 quota will be released repeatly.
2030 Details : void sending multiple quota reqs to mds, which will keep the
2031 status between the reqs.
2035 Description: osts not allocated evenly to files
2036 Details : change the condition to increase offset_idx
2039 Frequency : Always for filesystems larger than 2TB on 32-bit systems.
2041 Description: Data corruption for OSTs that are formatted larger than 2TB
2043 Details : When generating the bio request for lustre file writes the
2044 sector number would overflow a temporary variable before being
2045 used for the IO. The data reads correctly from Lustre (which
2046 will overflow in a similar manner) but other file data or
2047 filesystem metadata may be corrupted in some cases.
2051 Description: Tune the kernel for good SCSI performance.
2052 Details : Set the value of /sys/block/{dev}/queue/max_sectors_kb
2053 to the value of /sys/block/{dev}/queue/max_hw_sectors_kb
2058 Description: rewrite ext2-derived code in obdclass/uuid.c
2059 Details : rewrite inherited code (uuid parsing code from ext2 utils)
2060 from scratch preserving functionality.
2064 Description: "lfs find -obd UUID" prints directories
2065 Details : "lfs find -obd UUID" will return all directory names instead
2066 of just file names. It is incorrect because the directories
2067 do not reside on the OSTs.
2071 Description: use i_size_read and i_size_write in 2.6 port
2072 Details : replace inode->i_size access with i_size_read/write()
2075 Frequency : when removing large files
2077 Description: scheduling issue during removal of large Lustre files
2078 Details : Don't take the BKL in fsfilt_ext3_setattr() for 2.6 kernels.
2079 It causes scheduling issues when removing large files (17TB in the
2083 Frequency : only with liblustre clients on XT3
2085 Description: evictions taking too long
2086 Details : allow llrd to evict clients directly on OSTs
2089 Frequency : only on ppc
2091 Description: /proc/fs/lustre/devices broken on ppc
2092 Details : The patch as applied to 1.6.2 doesn't look correct for all arches.
2093 We should make sure the type of 'index' is loff_t and then cast
2094 explicitly as needed below. Do not assign an explicitly cast
2098 Frequency : only for rhel5
2100 Description: Kernel patches update for RHEL5 2.6.18-8.1.10.el5.
2101 Details : Modify the target file & which_kernel.
2103 Severity : enhancement
2105 Description: omit set fsid for export NFS
2106 Details : fix set/restore device id for avoid EMFILE error and mark lustre
2107 fs as FS_REQUIRES_DEV for avoid problems with generate fsid.
2111 Frequency : Always, for kernels after 2.6.16
2112 Description: Fix warning idr_remove called for id=.. which is not allocated.
2113 Details : Last kernels save old s_dev before kill super and not allow
2114 to restore from callback - restore it before call kill_anon_super.
2118 Description: buffer overruns could theoretically occur
2119 Details : llapi_semantic_traverse() modifies the "path" argument by
2120 appending values to the end of the origin string, and a
2121 overrun may occur. Adding buffer overrun check in liblustreapi.
2125 Description: Fix error on 'ls .' at the top of the Lustre mount.
2126 Details : Don't revalidate dentry if it is a root dentry.
2130 Description: Kernel patches update for RHEL4 2.6.9-55.0.6.
2131 Details : Modify vm-tunables-rhel4.patch.
2135 Description: Kernel config for 2.6.18-vanilla.
2136 Details : Modify targets/2.6-vanilla.target.in.
2137 Add config file kernel-2.6.18-2.6-vanilla-i686.config.
2138 Add config file kernel-2.6.18-2.6-vanilla-i686-smp.config.
2139 Add config file kernel-2.6.18-2.6-vanilla-x86_64.config.
2140 Add config file kernel-2.6.18-2.6-vanilla-x86_64-smp.config.
2144 Description: improve handling recoverable errors
2145 Details : if request processig with error which can be recoverable on server
2146 request should be resend, otherwise page released from cache and
2151 Description: Kernel patches update for RHEL5 2.6.18-8.1.14.el5.
2152 Details : Modify target file & which_patch.
2153 A flaw was found in the IA32 system call emulation provided
2154 on AMD64 and Intel 64 platforms. An improperly validated 64-bit
2155 value could be stored in the %RAX register, which could trigger an
2156 out-of-bounds system call table access. An untrusted local user
2157 could exploit this flaw to run code in the kernel
2158 (ie a root privilege escalation). (CVE-2007-4573).
2162 Description: O_DIRECT bypasses client statistics.
2163 Details : When running with O_DIRECT I/O, neither the client rpc_stats nor
2164 read_ahead_stats were updated. copied the stats section from
2165 osc_send_oap_rpc() into async_internal().
2169 Description: Add jbd statistics patch for RHEL5 and 2.6.18-vanilla
2173 Description: change order of libsysio includes
2174 Details : '#include sysio.h' should always come before '#include xtio.h'
2178 Description: handle "serious error: objid * already exists" more gracefully
2179 Details : If LAST_ID value on disk is smaller than the objects existing in
2180 the O/0/d* directories, it indicates disk corruption and causes an
2181 LBUG(). If the object is 0-length, then we should use the existing
2182 object. This will help to avoid a full fsck in most cases.
2184 Severity : enhancement
2186 Description: adapt the lustre_config script to support the upgrade case
2187 Details : Add "-u" option for lustre_config script to support upgrading 1.4
2188 server targets to 1.6 in parallel.
2192 Description: To avoid grant space > avaible space when the disk is almost
2193 full. Without this patch you might see the error "grant XXXX >
2194 available" or some LBUG about grant, when the disk is almost
2196 Details : In filter_check_grant, for non_grant cache write, we should
2197 check the left space by if (*left > ungranted + bytes), instead
2198 of (*left > ungranted), because only we are sure the left space
2199 is enough for another "bytes", then the ungrant space should be
2200 increase. In client, we should update cl_avail_grant only there
2201 is OBD_MD_FLGRANT in the reply.
2205 Description: Update RHEL 4 kernel to fix local root privilege escalation.
2206 Details : Update to the latest RHEL 4 kernel to fix the vulnerability
2207 described in CVE-2007-4573. This problem could allow untrusted
2208 local users to gain root access.
2211 Frequency : when using O_DIRECT and quotas
2213 Description: Incorrect file ownership on O_DIRECT output files
2214 Details : block usage reported by 'lfs quota' does not take into account
2215 files that have been written with O_DIRECT.
2220 Description: touch file failed when fs is not full
2221 Details : OST in recovery should not be discarded by MDS in alloc_qos(),
2222 otherwise we can get ENOSP while fs is not full.
2226 Description: parallel lock callbacks
2227 Details : Instead of sending blocking and completion callbacks as separated
2228 requests, adding them to a set and sending in parallel.
2231 Frequency : only for Cray XT3
2232 Bugzilla : 12829/13455
2233 Description: Changing primary group doesn't change the group lustre assigns to
2235 Details : When CRAY_XT3 is defined, the fsgid supplied by the client is
2236 overridden with the primary group provided by the group upcall,
2237 whereas the supplied fsgid can be trusted if it is in the list of
2238 supplementary groups returned by the group upcall.
2240 Severity : enhancement
2242 Description: Allow masking D_WARNING, D_ERROR messages from console
2243 Details : Console messages can now be disabled via lnet.printk.
2247 Description: User code with malformed file open parameter crashes client node
2248 Details : Before packing join_file req, all the related reference should be
2249 checked carefully in case some malformed flags cause fake
2250 join_file req on client.
2254 Description: LDLM_ENQUEUE races with LDLM_CP_CALLBACK
2255 Details : ldlm_completion_ast() assumes that a lock is granted when the req
2256 mode is equal to the granted mode. However, it should also check
2257 that LDLM_FL_CP_REQD is not set.
2261 Description: Heavy nfs access might result in deadlocks
2262 Details : After ELC code landed, it is now improper to enqueue any mds
2263 locks under och_sem, because enqueue might want to decide to
2264 cancel open locks for same inode we are holding och_sem for.
2268 Description: Client eviction while running blogbench
2269 Details : A lot of unlink operations with concurrent I/O can lead to a
2270 deadlock causing evictions. To address the problem, the number of
2271 oustanding OST_DESTROY requests is now throttled to
2272 max_rpcs_in_flight per OSC and LDLM_FL_DISCARD_DATA blocking
2273 callbacks are processed in priority.
2277 Description: enable ACLs on MDS by default
2278 Details : ACLs must be enabled on MDS by default.
2281 Frequency : PPC/PPC64 only
2283 Description: conflicts between asm-ppc64/types.h and lustre_types.h
2284 Details : fix duplicated definitions between asm-ppc64/types.h and
2285 lustre_types.h on PPC.
2288 Frequency : PPC/PPC64 only
2290 Description: asm-ppc/segment.h does not exist
2291 Details : fix compile issue on PPC.
2295 Description: better handle error messages in extents code
2298 Frequency : RHEL4 only
2300 Description: mkfs is very slow on IA64/RHEL4
2301 Details : A performance regression has been discovered in the MPT Fusion
2302 driver between versions 3.02.73rh and 3.02.99.00rh. As a
2303 consequence, we have downgraded the MPT Fusion driver in the RHEL4
2304 kernel from 3.02.99.00 to 3.02.73 until this problem is fixed.
2306 Severity : enhancement
2308 Description: SNMP support enhancement
2309 Details : Adding total number of sampled request for an MDS node in snmp
2312 Severity : enhancement
2314 Description: Optimize ldlm waiting list processing for PR extent locks
2315 Details : When processing waiting list for read extent lock and meeting read
2316 lock that is same or wider to it that is not contended, skip
2317 processing rest of the list and immediatelly return current
2318 status of conflictness, since we are guaranteed there are no
2319 conflicting locks in the rest of the list.
2323 Description: Time out and refuse to reconnect
2324 Details : When the failover node is the primary node, it is possible
2325 to have two identical connections in imp_conn_list. We must
2326 compare not conn's pointers but NIDs, otherwise we can defeat
2327 connection throttling.
2331 Description: port llog fixes from b1_6 into HEAD
2332 Details : Port llog reference couting and some llog cleanups from b1_6
2333 (bug 10800) into HEAD, for protect from panic and access to already
2334 free llog structures.
2338 Description: Detect stride IO mode in read-ahead
2339 Details : When a client does stride read, read-ahead should detect that and
2340 read-ahead pages according to the detected stride pattern.
2344 Description: data checksumming impacts single node performance
2345 Details : add support for several checksum algorithm. Currently, only CRC32
2346 and Adler-32 are supported. The checksum type can be changed on
2347 the fly via /proc/fs/lustre/osc/*/checksum_type.
2351 Description: use adler32 for page checksums
2352 Details : when available, use the Adler-32 algorithm instead of CRC32 for
2357 Description: build for x2 fails
2358 Details : fix compile issue on Cray systems.
2362 Description: Properly match for duplicate locks
2363 Details : Due to different lock order from skiplists code, we need to
2364 traverse entire list for now
2367 Frequency : only on PPC/SLES10
2369 Description: "BITS_PER_LONG is not 32 or 64" in linux/idr.h
2370 Details : On SLES10/PPC, fs.h includes idr.h which requires BITS_PER_LONG to
2371 be defined. Add a hack in mkfs_lustre.c to work around this compile
2376 Description: LASSERT on MDS when client holding flock lock dies
2377 Details : ldlm pool logic depends on number of granted locks equal to
2378 number of released locks which is not true for flock locks, so
2379 just exclude such locks from consideration.
2383 Description: MDS deadlock with many ll_sync_lov threads and I/O stalled
2384 Details : Use fsfilt_sync() for both the whole filesystem sync and
2385 individual file sync to eliminate dangerous inode locking
2386 with I_LOCK that can lead to a deadlock.
2390 Description: performance in 1.6.3
2391 Details : Force q->max_phys_segments to MAX_PHYS_SEGMENTS on SLES10 to be
2392 sure that 1MB requests are not fragmented by the block layer.
2394 Severity : enhancement
2396 Description: organize the server-side client stats on per-nid basis
2397 Details : Change the structure of stats under obdfilter and mds to
2405 The "uuid"s file would list the uuids of _active_ exports.
2406 And the clear entry is to clear all stats and stale nids.
2408 Severity : enhancement
2410 Description: eliminate client locks in face of contention
2411 Details : file contention detection and lockless i/o implementation
2412 for contended files.
2416 Description: Reinitialize optind to 0 so that interactive lfs works in all cases
2419 Frequency : very rare, if additional xattrs are used on kernels >= 2.6.12
2421 Description: MDS may lose file striping (and hence file data) in some cases
2422 Details : If there are additional extended attributes stored on the MDS,
2423 in particular ACLs, SELinux, or user attributes (if user_xattr
2424 is specified for the client mount options) then there is a risk
2425 of attribute loss. Additionally, the Lustre file striping
2426 needs to be larger than default (e.g. striped over all OSTs),
2427 and an additional attribute must be stored initially in the
2428 inode and then increase in size enough to be moved to the
2429 external attribute block (e.g. ACL growing in size) for file
2434 Description: skiplist implementation simplification
2435 Details : skiplists are used to group compatible locks on granted list
2436 that was implemented as tracking first and last lock of each
2437 lock group the patch changes that to using doubly linked lists
2441 Description: MDS LBUG: ASSERTION(!IS_ERR(dchild))
2442 Details : Change LASSERTs to client eviction (i.e. abort client's recovery)
2443 because LASSERT on both the data supplied by a client, and the
2444 data on disk is dangerous and incorrect.
2446 Severity : enhancement
2448 Description: Slow truncate/writes to huge files at high offsets.
2449 Details : Directly associate cached pages to lock that protect those pages,
2450 this allows us to quickly find what pages to write and remove
2451 once lock callback is received.
2455 Description: more ldlm soft lockups
2456 Details : In ldlm_resource_add_lock(), call to ldlm_resource_dump()
2457 starve other threads from the resource lock for a long time in
2458 case of long waiting queue, so change the debug level from
2459 D_OTHER to the less frequently used D_INFO.
2461 Severity : enhancement
2463 Description: add -gid, -group, -uid, -user options to lfs find
2467 Description: Hung threads in invalidate_inode_pages2_range
2468 Details : The direct IO path doesn't call check_rpcs to submit a new RPC
2469 once one is completed. As a result, some RPCs are stuck in the
2470 queue and are never sent.
2474 Description: filter threads hungs on waiting journal commit
2475 Details : Cleanup filter group llog code, then only filter group llog will
2476 be only created in the MDS/OST syncing process.
2480 Description: Procfs and llog threads access destoryed import sometimes.
2481 Details : Sync the import destoryed process with procfs and llog threads by
2482 the import refcount and semaphore.
2484 Severity : enhancement
2486 Description: openlock cache of b1_6 port to HEAD
2491 Description: kernel BUG at ldiskfs2_ext_new_extent_cb
2492 Details : If insertion of an extent fails, then discard the inode
2493 preallocation and free data blocks else it can lead to duplicate
2498 Description: don't always update ctime in ext3_xattr_set_handle()
2499 Details : Current xattr code updates inode ctime in ext3_xattr_set_handle.
2500 In some cases the ctime should not be updated, for example for
2501 2.0->1.8 compatibility it is necessary to delete an xattr and it
2502 should not update the ctime.
2506 Bugzilla : 15713/16362
2507 Description: Assertion in iopen_connect_dentry in 1.6.3
2508 Details : looking up an inode via iopen with the wrong generation number can
2509 populate the dcache with a disconneced dentry while the inode
2510 number is in the process of being reallocated. This causes an
2511 assertion failure in iopen since the inode's dentry list contains
2512 both a connected and disconnected dentry.
2516 Description: assertion failure in ldlm_handle2lock()
2517 Details : fix a race between class_handle_unhash() and class_handle2object()
2518 introduced in lustre 1.6.5 by bug 13622.
2523 Description: Kernel BUG: sd_iostats_bump: unexpected disk index
2524 Details : remove the limit of 256 scsi disks in the sd_iostat patch
2529 Description: oops in sd_iostats_seq_show()
2530 Details : unloading/reloading the scsi low level driver triggers a kernel
2531 bug when trying to access the sd iostat file.
2536 Description: Kernel panics during QLogic driver reload
2537 Details : REQ_BLOCK_PC requests are not handled properly in the sd iostat
2538 patch, causing memory corruption.
2543 Description: journal_dev option does not work in b1_6
2544 Details : pass mount option during pre-mount.
2546 Severity : enhancement
2548 Description: Add a FIEMAP(FIle Extent MAP) ioctl
2549 Details : FIEMAP ioctl will allow an application to efficiently fetch the
2550 extent information of a file. It can be used to map logical blocks
2551 in a file to physical blocks in the block device.
2555 Description: LDLM soft lockups - improvement
2556 Details : It is be possible to send the lock handle along with each read
2557 or write request because the client is already doing a lock match
2558 itself so there isn't any reason the OST should have to re-do that
2562 Frequency : only on Cray X2
2564 Description: X2 build failures
2565 Details : fix build failures on Cray X2.
2569 Description: xid & resent requests
2570 Details : Initialize RPC XID from clock at startup (randomly if clock is
2573 Severity : enhancement
2575 Description: Add lustre_start utility to start or stop multiple Lustre servers
2580 Description: Lustre GPF in {:ptlrpc:ptlrpc_server_free_request+373}
2581 Details : In case of memory pressure, list_del() can be called twice on
2582 req->rq_history_list, causing a kernel oops.
2586 Description: kptllnd_peer_check_sends()) ASSERTION(!in_interrupt()) failed
2587 Details : fix stack overflow in the distributed lock manager by defering
2588 export eviction after a failed AST to the elt thread instead of
2589 handling it in the dlm interpret routine.
2593 Description: Convert some comments to new format.
2594 Details : Update documenting comments to match doxygen conventions.
2598 Description: Grammar fixes.
2599 Details : A couple of trivial spelling fixes.
2603 Description: OSD_COUNTERS-mandatory
2604 Details : Make previously optional ->oti_{w,r}_locks sanity checks mandatory
2605 to simplify the code.
2609 Description: simplify cmm_device freeing logic.
2610 Details : Call cmm_device_free() in the failure path of cmm_device_alloc().
2614 Description: Add lockdep support to dt_object_operations locking interface.
2615 Details : Augment ->do_{read,write}_lock() prototypes with a `role'
2616 parameter indicating lock ordering. Update mdd code to use new
2621 Description: Introduce failloc constants for lockless IO tests.
2622 Details : Add two new failloc constants to test lockless IO. Only one of
2623 them in implemented---another is checked in yet to be landed
2628 Description: Add lockdep support for inode mutex.
2629 Details : Introduce and use new LOCK_INODE_MUTEX_PARENT() macro to be used
2630 in the situations where Lustre has to lock more than one inode
2635 Description: Add optional invariants checking support.
2636 Details : Add new LINVRNT() macro, optional on new --enable-invariants
2637 configure switch. This macro is to be used for consistency and
2638 sanity checks that are too expensive to be left in `production'
2643 Description: Zap lock->l_granted_mode with explicit LCK_MINMODE.
2644 Details : Use LCK_MINMODE rather than 0 to reset lock->l_granted_mode to
2649 Description: Add lockdep support for ldlm_lock and ldlm_resource.
2650 Details : Use spin_lock_nested() in (the only) situation where more than
2651 one ldlm_lock is locked simultaneously. Also, fix possible
2652 dead-lock in ldlm_lock_change_resource() by enforcing particular
2657 Description: Use struct ldlm_callback_suite in ldlm_lock_create().
2658 Details : Instead of specifying each ldlm_lock call-back through separate
2659 parameter, wrap them into struct ldlm_callback_suite.
2663 Description: Kill join_lru obd method and its callers.
2664 Details : CLIO uses lock weighting policy to keep locks over mmapped regions
2665 in memory---a requirement implemented through ->o_join_lru() obd
2666 method in HEAD. Remove this method and its users.
2670 Description: Add asynchronous ldlm ENQUEUE completion handler.
2671 Details : CLIO posts ENQUEUE requests asynchronously through ptlrpcd---a
2672 case that stock ldlm_completion_ast() cannot handle as it waits
2673 until lock is granted. Introduce new ldlm_completion_ast_async()
2674 for this. Also comment ldlm_completion_ast().
2678 Description: ldlm_error <-> errno conversion.
2679 Details : Add functions to map (rather arbitrary) between LDLM error codes
2680 and standard errno values. CLIO needs this to prevent LDLM specific
2681 constants from escaping ldlm and osc.
2685 Description: Kill unused ldlm_handle2lock_ns() function.
2686 Details : Kill unused ldlm_handle2lock_ns() function.
2690 Description: Add lu_ref support to ldlm_lock
2691 Details : lu_ref support for ldlm_lock and ldlm_resource. See lu_ref patch.
2692 lu_ref fields ->l_reference and ->lr_reference are added to ldlm_lock
2693 and ldlm_resource. LDLM interface has to be changed, because code that
2694 releases a reference on a lock, has to "know" what reference this is.
2695 In the most frequent case
2697 lock = ldlm_handle2lock(handle);
2699 LDLM_LOCK_PUT(lock);
2701 no changes are required. When any other reference (received _not_ from
2702 ldlm_handle2lock()) is released, LDLM_LOCK_RELEASE() has to be called
2703 instead of LDLM_LOCK_PUT().
2705 Arguably, changes are pervasive, and interface requires some discipline
2706 for proper use. On the other hand, it was very instrumental in finding
2707 a few leaked lock references.
2711 Description: Add ldlm_lock_addref_try().
2712 Details : Introduce ldlm_lock_addref_try() function (used by CLIO) that
2713 attempts to addref a lock that might be being canceled
2718 Description: Add ldlm_weigh_callback().
2719 Details : Add new ->l_weigh_ast() call-back to ldlm_lock. It is called
2720 by ldlm_cancel_shrink_policy() to estimate lock "value", instead of
2721 hard-coded `number of pages' logic.
2725 Description: Add lockdep annotations to llog code.
2726 Details : Use appropriately tagged _nested() locking calls in the places
2727 where llog takes more than one ->lgh_lock lock.
2731 Description: Add loi_kms_set().
2732 Details : Wrap kms updates into a helper function.
2736 Description: Constify instances of struct lsm_operations.
2737 Details : Constify instances of struct lsm_operations.
2741 Description: lu_conf support.
2742 Details : On a server, a file system object is uniquely identified
2743 by a fid, which is sufficient to locate and load all object
2744 state (inode). On a client, on the other hand, more data are
2745 necessary instantiate an object. Change lu_object_find() and
2746 friends to take additional `lu_conf' argument describing object.
2747 Typically this includes layout information.
2751 Description: lu_context fixes.
2752 Details : Introduce new lu_context functions that are needed on the client
2753 side, where some system threads (ptlrpcd) are shared by multiple
2754 modules, and so cannot be stopped during module shutdown.
2758 Description: Add start and stop methods to lu_device_type_operations.
2759 Details : Introduce two new methods in lu_device_type_operations, that are
2760 invoked when first instance of a given type is created and last one
2761 is destroyed respectively. This is need by CLIO.
2765 Description: Add lu_ref support to struct lu_device.
2766 Details : Add lu_ref support to lu_object and lu_device. lu_ref is used to
2767 track leaked references.
2771 Description: Introduce lu_kmem_descr.
2772 Details : lu_kmem_descr and its companion interface allow to create
2773 and destroy a number of kmem caches at once.
2777 Description: Fix lu_object finalization race.
2778 Details : Fix a race between lu_object_find() finding an object and its
2779 concurrent finalization. This race is (most likely) not possible
2780 on the server, but might happen on the client.
2784 Description: Introduce lu_ref interface.
2785 Details : lu_ref is a debugging module allowing to track references to
2786 a given object. It is quite cpu expensive, and has to be
2787 explicitly enabled with --enable-lu_ref. See usage description
2792 Description: Factor lu_site procfs stats into a separate function.
2793 Details : Separate lu_site stats printing code into a separate function
2794 that can be reused on a client.
2798 Description: Constify instances of struct {lu,dt,md}_device_operations.
2799 Details : Constify instances of struct {lu,dt,md}_device_operations.
2803 Description: Introduce struct md_site and move meta-data specific parts of
2804 struct lu_site here.
2805 Details : Move md-specific fields out of struct lu_site into special struct
2806 md_site, so that lu_site can be used on a client.
2810 Description: Kill mdd_lov_destroy().
2811 Details : Remove unused mdd code.
2815 Description: Add st_block checking to multistat.c.
2816 Details : Add st_block checking to multistat.c.
2820 Description: Add lu_ref support to struct obd_device.
2821 Details : Add lu_ref tracking to obd_device.
2825 Description: Kill obd_set_fail_loc().
2826 Details : Remove unused code.
2830 Description: Add special type for ptlrpc_request interpret functions.
2831 Details : Add lu_env parameter to ->rq_interpreter call-back. NULL is passed
2832 there. Actual usage will be in CLIO.
2836 Description: Replace RW_LOCK_UNLOCKED() macro with rwlock_init().
2837 Details : Replace RW_LOCK_UNLOCKED() with rwlock_init() as the former
2838 doesn't work with lockdep.
2842 Description: Add rwv.c test program.
2843 Details : New testing program exercising readv(2) and writev(2) (Qian).
2847 Description: Add sendfile.c test program.
2848 Details : New testing program exercising sendfile(2) (Jay).
2852 Description: Ratelimit a message that can be very frequent.
2853 Details : Ratelimit a memory allocation failure message that can
2858 Description: Use cdebug_show() in CDEBUG-style macros defined outside of libcfs.
2859 Details : Use cdebug_show() in CDEBUG-style macros defined outside of libcfs.
2863 Description: Liblustre build fixes.
2864 Details : Liblustre build fixes.
2868 Description: libcfs: add cfs_{need,cond}_resched() interface.
2869 Details : libcfs: add cfs_{need,cond}_resched() definition and
2870 implementations for Linux, NT, and liblustre.
2872 Severity : enhancement
2874 Description: More exported tunables for mballoc
2875 Details : Add support for tunable preallocation window and new tunables for
2876 large/small requests
2880 Description: Detect corruption of block bitmap and checking for preallocations
2881 Details : Checks validity of on-disk block bitmap. Also it does better
2882 checking of number of applied preallocations. When corruption is
2883 found, it turns filesystem readonly to prevent further corruptions.
2887 Description: (rw.c:1323:ll_read_ahead_pages()) ASSERTION(page_idx > ria->ria_stoff) failed
2888 Details : Once the unmatched stride IO mode is detected, shrink the stride-ahead
2889 window to 0. If it does hit cache miss, and read-pattern is still
2890 stride-io mode, does not reset the stride window, but also does not
2891 increase the stride window length in this case.
2895 Frequency : only for big-endian servers
2896 Description: Check if system is big-endian while mounting fs with extents feature
2897 Details : Mounting a filesystem with extents feature will fail on big-endian
2898 systems since ext3-based ldiskfs is not supported on big-endian
2899 systems. This can be over-riden with "bigendian_extents" mount option.
2901 Severity : enhancement
2903 Description: The root squash functionality
2904 Details : A security feature, which is to prevent users from being able
2905 to mount lustre on their desktop, run as root, and delete
2906 all of the files in the filesystem. The goal is accomplished by
2907 remapping user id (UID) and group id (GID) of the root user to
2908 a UID and GID specified by the system administartor via Lustre
2909 configuration management server (MGS). The functionality also
2910 allows to specify sets of clients for which the remapping does
2915 Description: Excessive recovery window
2916 Details : With AT enabled, the recovery window can be excessively long (6000+
2917 seconds). To address this problem, we no longer use
2918 OBD_RECOVERY_FACTOR when extending the recovery window (the connect
2919 timeout no longer depends on the service time, it is set to
2920 INITIAL_CONNECT_TIMEOUT now) and clients report the old service
2921 time via pb_service_time.
2925 Description: Watchdog triggered on MDS failover
2926 Details : enable OBD_CONNECT_MDT flag when connecting from the MDS so that
2927 the OSTs know that the MDS "UUID" can be reused for the same export
2928 from a different NID, so we do not need to wait for the export to be
2932 Frequency : rare, only if using MMP with Linux RAID
2934 Description: MMP doesn't work with Linux RAID
2935 Details : While using HA for Lustre servers with Linux RAID, it is possible
2936 that MMP will not detect multiple mounts. To make this work we
2937 need to unplug the device queue in RAID when the MMP block is being
2938 written. Also while reading the MMP block, we should read it from
2939 disk and not the cached one.
2941 Severity : enhancement
2943 Description: open file using fid
2944 Details : A file can be opened using just its fid, like
2945 <mntpt>/.lustre/fid/SEQ:OID:VER - this is needed for HSM and replication
2948 Frequency : Only in RHEL5 when mounting multiple ext3 filesystems
2951 Description: "kmem_cache_create: duplicate cache jbd_4k" error message
2952 Details : add proper locking for creation of jbd_4k slab cache
2956 Description: MMP check in ext3_remount() fails without displaying any error
2957 Details : When multiple mount protection fails during remount, proper error
2960 Severity : enhancement
2962 Description: Allow stripe size to be up to 4G-64k
2963 Details : Fix math logic to allow large stripe sizes.
2967 Description: add check for >8TB ldiskfs filesystems
2968 Details : ext3-based ldiskfs does not support greater than 8TB LUNs.
2969 Don't allow >8TB ldiskfs filesystems to be mounted without
2970 force_over_8tb mount option
2975 Descriptoin: OSS complains "too many blocks" and crashes
2976 Details : In filter_commitrw_write():
2977 1. lquota_chkquota();
2978 2. obdo_from_inode();
2979 3. lquota_pending_commit();
2980 "2." has changed uid and gid in this case, then caused
2981 lquota_chkquota() and lquota_pending_commit() to use
2982 different uid and gid. That is the root of the bug.
2986 Descriptoin: getstripe doesn't report stripe count = -1 sanely
2987 Details : Detect and print out the 'stripe count = -1' case properly.
2991 Description: Out or order replies might be lost on replay
2992 Details : In ptlrpc_retain_replayable_request if we cannot find retained
2993 request with tid smaller than one currently being added, add it
2994 to the start, not end of the list.
2996 --------------------------------------------------------------------------------
2998 2007-08-10 Cluster File Systems, Inc. <info@clusterfs.com>
3000 * Support for kernels:
3001 2.6.5-7.283 (SLES 9),
3002 2.6.9-55.EL (RHEL 4),
3003 2.6.16.46-0.14 (SLES 10),
3004 2.6.18.8 vanilla (kernel.org)
3005 * Client support for unpatched kernels:
3006 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
3007 2.6.16 - 2.6.22 vanilla (kernel.org)
3008 * Due to recently discovered recovery problems, we do not recommend
3009 using patchless RHEL 4 clients with this or any earlier release.
3010 * Recommended e2fsprogs version: 1.39.cfs8
3011 * Note that reiserfs quotas are disabled on SLES 10 in this kernel.
3012 * Starting with this release, the ldiskfs backing filesystem required
3013 by Lustre is now in its own package, lustre-ldiskfs. This package
3014 should be installed. It is versioned separately from Lustre and
3015 may be released separately in future.
3019 Description: block reactivating mgc import until all deactivates complete
3020 Details : Fix race when failing back MDT/MGS to itself (testing)
3023 Frequency : at statup only
3025 Description: mds_lov_synchronize race leads to various problems
3026 Details : simultaneous MDT->OST connections at startup can cause the
3027 sync to abort, leaving the OSC in a bad state.
3029 Severity : enhancement
3031 Description: add optional extra BUILD_VERSION info
3032 Details : add a new environment variable (namely LUSTRE_VERS) which allows
3033 to override the lustre version.
3036 Frequency : 2.6.18 servers only
3038 Description: ll_kern_mount() doesn't release the module reference
3039 Details : The ldiskfs module reference count never drops down to 0
3040 because ll_kern_mount() doesn't release the module reference.
3045 Description: server LBUG when using old ost_num_threads parameter
3046 Details : Accept the old ost_num_threads parameter but warn that it
3047 is deprecated, and fix an off-by-one error that caused an LBUG.
3052 Description: Transient SCSI error results in persistent IO issue
3053 Details : iobuf->dr_error is not reinitialized to 0 between two
3057 Frequency : sometimes when underlying device returns I/O errors
3059 Description: OSTs not going read-only during write failures
3060 Details : OSTs are not remounted read-only when the journal commit threads
3061 get I/O errors because fsfilt_ext3 calls journal_start/stop()
3062 instead of the ext3 wrappers.
3066 Description: poor connect scaling with increasing client count
3067 Details : Don't run filter_grant_sanity_check for more than 100 exports
3068 to improve scaling for large numbers of clients.
3071 Frequency : SLES10 only
3073 Description: sanity-quota.sh quotacheck failed: rc = -22
3074 Details : Quotas cannot be enabled on SLES10.
3077 Frequency : liblustre clients only
3079 Description: getdirentries does not give error when run on compute nodes
3080 Details : getdirentries does not fail when the size specified as an argument
3081 is too small to contain at least one entry
3083 Severity : enhancement
3085 Description: Add LNET router traceability for debug purposes
3086 Details : If a checksum failure occurs with a router as part of the
3087 IO path, the NID of the last router that forwarded the bulk data
3088 is printed so it can be identified.
3093 Description: OST "spontaneously" evicts client; client has imp_pingable == 0
3094 Details : Due to a race condition, liblustre clients were occasionally
3095 evicted incorrectly.
3097 Severity : enhancement
3099 Description: lfs setstripe use optional parameters instead of postional
3102 Severity : enhancement
3104 Description: Nanosecond timestamp support for ldiskfs
3105 Details : The on-disk ldiskfs filesystem has added support for nanosecond
3106 resolution timestamps. There is not yet support for this at
3107 the Lustre filesystem level.
3110 Frequency : during server recovery
3112 Description: MDS failing to send precreate requests due to OSCC_FLAG_RECOVERING
3113 Details : request with rq_no_resend flag not awake l_wait_event if they get
3117 Frequency : nfs export on patchless client
3119 Description: connectathon hang when test nfs export over patchless client
3120 Details : Disconnected dentry cannot be found with lookup, so we do not need
3121 to unhash it or make it invalid
3124 Description: fix llapi_lov_get_uuids() to allow many OSTs to be returned
3125 Details: : Change llapi_lov_get_uuids() to read the UUIDs from /proc instead
3126 of using an ioctl. This allows lfsck for > 160 OSTs to succeed.
3131 Description: open req refcounting wrong on reconnect
3132 Details : If reconnect happened between getting open reply from server and
3133 call to mdc_set_replay_data in ll_file_open, we will schedule
3134 replay for unreferenced request that we are about to free.
3135 Subsequent close will crash in variety of ways.
3136 Check that request is still eligible for replay in
3137 mdc_set_replay_data().
3142 Description: disable writes to filesystem when reading health_check file
3143 Details : the default for reading the health_check proc file has changed
3144 to NOT do a journal transaction and write to disk, because this
3145 can cause reads of the /proc file to hang and block HA state
3146 checking on a healthy but otherwise heavily loaded system. It
3147 is possible to return to the previous behaviour during configure
3148 with --enable-health-write.
3150 Severity : enhancement
3152 Description: 64-bit inode version
3153 Details: : Add a on-disk 64-bit inode version for ext3 to track changes made
3154 to the inode. This will be required for version-based recovery.
3159 Description: MDS fails to start if a duplicate client export is detected
3160 Details : in some rare cases it was possible for a client to connect to
3161 an MDS multiple times. Upon recovery the MDS would detect this
3162 and fail during startup. Handle this more gracefully.
3164 Severity : enhancement
3166 Description: Add -o localflock option to simulate old noflock
3168 Details : This will achieve local-only flock/fcntl locks
3174 Description: log_commit_thread vs filter_destroy race leads to crash
3175 Details : Take import reference before releasing llog record semaphore
3180 Description: Wrong request locking in request set processing
3181 Details : ptlrpc_check_set wrongly uses req->rq_lock for proctect add to
3182 imp_delayed_list, in this place should be used imp_lock.
3185 Frequency : when reconnection
3187 Description: Grant Leak when osc reconnect to OST
3188 Details : When osc reconnect ost, OST(filter) should check whether it
3189 should grant more space to client by comparing fed_grant and
3190 cl_avail_grant, and return the granted space to client instead
3191 of "new granted" space, because client will call osc_init_grant
3192 to update the client grant space info.
3195 Frequency : when client reconnect to OST
3197 Description: Grant Leak when osc do resend and replay bulk write
3198 Details : When osc reconnect to OST, OST(filter)should clear grant info of
3199 bulk write request, because the grant info while be sync between
3200 OSC and OST when reconnect, and we should ignore the grant info
3201 these of resend/replay write req.
3206 Description: Grant space more than available left space sometimes.
3207 Details : When then OST is about to be full, if two bulk writing from
3208 different clients came to OST. Accord the avaliable space of the
3209 OST, the first req should be permitted, and the second one
3210 should be denied by ENOSPC. But if the seconde arrived before
3211 the first one is committed. The OST might wrongly permit second
3212 writing, which will cause grant space > available space.
3215 Frequency : when client is evicted
3217 Description: Grant might be wrongly erased when osc is evicted by OST
3218 Details : when the import is evicted by server, it will fork another
3219 thread ptlrpc_invalidate_import_thread to invalidate the
3220 import, where the grant will be set to 0. While the original
3221 thread will update the grant it got when connecting. So if
3222 the former happened latter, the grant will be wrongly errased
3223 because of this race.
3228 Description: Checking Stale with correct fid
3229 Details : ll_revalidate_it should uses de_inode instead of op_data.fid2
3230 to check whether it is stale, because sometimes, we want the
3231 enqueue happened anyway, and op_data.fid2 will not be initialized.
3233 Severity : enhancement
3235 Description: update patchless client
3236 Details : Add support for patchless client with 2.6.20, 2.6.21 and RHEL 5
3239 Frequency : only with 2.4 kernel
3241 Description: random memory corruption
3242 Details : size of struct ll_inode_info is to big for union inode.u and this
3243 can be cause of random memory corruption.
3248 Description: Memory leak in recovery
3249 Details : Lov_mds_md was not free in an error handler in mds_create_object.
3250 It should also check obd_fail before fsfilt_start, otherwise if
3251 fsfilt_start return -EROFS,(failover mds during mds recovery).
3252 then the req will return with repmsg->transno = 0 and rc = EROFS.
3253 and we met hit the assert LASSERT(req->rq_reqmsg->transno ==
3254 req->rq_repmsg->transno) in ptlrpc_replay_interpret. Fcc should
3255 be freed no matter whether fsfilt_commit success or not.
3258 Frequency : only with huge count clients
3260 Description: Prevents from taking the superblock lock in llap_from_page for
3262 Details : using LL_ORIGIN_REMOVEPAGE origin flag instead of LL_ORIGIN_UNKNOW
3263 for llap_from_page call in ll_removepage prevents from taking the
3264 superblock lock for a soon died page.
3269 Description: Not check open intent error before release open handle
3270 Details : in some rare cases, the open intent error is not checked before
3271 release open handle, which may cause
3272 ASSERTION(open_req->rq_transno != 0), because it tries to release
3273 the failed open handle.
3278 Description: Set cat log bitmap only after create log success.
3279 Details : in some rare cases, the cat log bitmap is set too early. and it
3280 should be set only after create log success.
3284 Description: Accessing a block bevice can re-enable I/O when Lustre is
3285 tearing down a device.
3286 Details : dev_clear_rdonly(bdev) must be called in kill_bdev() instead of
3291 Description: service threads may hog cpus when there are a lot of requests
3292 Details : Insert cond_resched to give other threads a chance to use some CPU
3297 Description: the cat log was not initialized in recovery
3298 Details : When mds(mgs) do recovery, the tgt_count might be zero, so the
3299 unlink log on mds will not be initialized until mds post
3300 recovery. And also in mds post recovery, the unlink log will
3301 initialization will be done asynchronausly, so there will be race
3302 between add unlink log and unlink log initialization.
3306 Description: brw_stats were being printed incorrectly
3307 Details : brw_stats were being printed as log2 but all of them were not
3308 recorded as log2. Also remove some code duplication arising from
3309 filter_tally_{read,write}.
3313 Frequency : rare, only in recovery.
3314 Description: ASSERTION(req->rq_type != LI_POISON) failed
3315 Details : imp_lock should be held while iterating over imp_sending_list for
3316 prevent destroy request after get timeout in ptlrpc_queue_wait.
3320 Description: replay-single.sh test 52 fails
3321 Details : A lock's skiplist need to be cleanup when it being unlinked
3322 from its resource list.
3326 Description: Short directio read returns full requested size rather than
3328 Details : Direct I/O operations should return actual amount of bytes
3329 transferred rather than requested size.
3331 Severity : enhancement
3333 Description: metadata RPC reduction (e.g. for rm performance)
3334 Details : decrease the amount of synchronous RPC between clients and servers
3335 by canceling conflicing lock before the operation on the client
3336 and packing thier handles into the main operation RPC to server.
3338 Severity : enhancement
3340 Description: Async OSC create to avoid the blocking unnecessarily.
3341 Details : If an OST has no remaining object, system will block on the
3342 creation when it needs to create a new object on this OST. Now,
3343 ways use pre-created objects when available, instead of blocking on
3344 an empty osc while others are not empty. If we must block, we
3345 block for the shortest possible period of time.
3349 Description: improve handling recoverable errors
3350 Details : if request processig with error which can be recoverable on server
3351 request should be resend, otherwise page released from cache and
3354 Severity : enhancement
3356 Description: refine locking for avoid write wrong info into lov_objid file
3357 Details : fix possible races with add new target and write/update data in
3360 --------------------------------------------------------------------------------
3362 2007-05-03 Cluster File Systems, Inc. <info@clusterfs.com>
3367 Frequency : on some architectures
3369 Description: 1.6 client sometimes fails to mount from a 1.4 MDT
3370 Details : Uninitialized flags sometimes cause configuration commands to
3374 Frequency : patchless clients only
3376 Description: missing __iget() symbol export
3377 Details : The __iget() symbol export is missing. To avoid the need for
3378 this on patchless clients the deathrow inode reaper is turned
3379 off, and we depend on the VM to clean up old inodes. This
3380 dependency was during via the fix for bug 12181.
3382 --------------------------------------------------------------------------------
3384 2007-04-19 Cluster File Systems, Inc. <info@clusterfs.com>
3386 * CONFIGURATION CHANGE. This version of Lustre WILL NOT
3387 INTEROPERATE with older versions automatically. In many cases a
3388 special upgrade step is needed. Please read the
3389 user documentation before upgrading any part of a live system.
3390 * WIRE PROTOCOL CHANGE from previous 1.6 beta versions. This
3391 version will not interoperate with 1.6 betas before beta5 (1.5.95).
3392 * WARNING: Lustre configuration and startup changes are required with
3393 this release. See https://mail.clusterfs.com/wikis/lustre/MountConf
3397 Severity : enhancement
3399 Description: MountConf
3400 Details : Lustre configuration is now managed via mkfs and mount
3401 commands instead of lmc and lconf. New obd types (MGS, MGC)
3402 are added for dynamic configuration management. See
3403 https://mail.clusterfs.com/wikis/lustre/MountConf for
3406 Severity : enhancement
3408 Description: dynamic OST addition
3409 Details : OSTs can now be added to a live filesystem
3411 Severity : enhancement
3413 Description: startup order invariance
3414 Details : MDTs and OSTs can be started in any order. Clients only
3415 require the MDT to complete startup.
3417 Severity : enhancement
3419 Description: parallel, asynchronous orphan cleanup
3420 Details : orphan cleanup is now performed in separate threads for each
3421 OST, allowing parallel non-blocking operation.
3423 Severity : enhancement
3425 Description: optimized stripe assignment
3426 Details : stripe assignments are now made based on ost space available,
3427 ost previous usage, and OSS previous usage, in order to try
3428 to optimize storage space and networking resources.
3430 Severity : enhancement
3432 Description: Permanently set tunables
3433 Details : All writable /proc/fs/lustre tunables can now be permanently
3434 set on a per-server basis, at mkfs time or on a live system.
3436 Severity : enhancement
3438 Description: Lustre message v2
3439 Details : Add lustre message format v2.
3441 Severity : enhancement
3443 Description: client OST exclusion list
3444 Details : Clients can be started with a list of OSTs that should be
3445 declared "inactive" for known non-responsive OSTs.
3449 Description: SPEC SFS validation failure on NFS v2 over lustre.
3450 Details : Changes the blocksize for regular files to be 2x RPC size,
3451 and not depend on stripe size.
3453 Severity : enhancement
3455 Description: Multiple MD RPCs in flight.
3456 Details : Further unserialise some read-only MDS RPCs - learn about intents.
3457 To avoid overly-overloading MDS, introduce a limit on number of
3458 MDS RPCs in flight for a single client and add /proc controls
3459 to adjust this limit.
3461 Severity : enhancement
3463 Description: client read/write statistics
3464 Details : Add client read/write call usage stats for performance
3465 analysis of user processes.
3466 /proc/fs/lustre/llite/*/offset_stats shows non-sequential
3467 file access. extents_stats shows chunk size distribution.
3468 extents_stats_per_process show chunk size distribution per
3471 Severity : enhancement
3473 Description: mds statistics
3474 Details : Add detailed mds operations statistics in
3475 /proc/fs/lustre/mds/*/stats.
3479 Description: Failure of copying files with lustre special EAs.
3480 Details : Client side always return success for setxattr call for lustre
3481 special xattr (currently only "trusted.lov").
3486 Description: Refcount LNET uuids
3487 Details : The global LNET uuid list grew linearly with every startup;
3488 refcount repeated list entries instead of always adding to
3491 Severity : enhancement
3493 Description: Dynamic service threads
3494 Details : Within a small range, start extra service threads
3495 automatically when the request queue builds up.
3498 Frequency : mixed-endian client/server environments
3500 Description: mixed-endian crashes
3501 Details : The new msg_v2 system had some failures in mixed-endian
3504 Severity : enhancement
3506 Description: Easy OST removal
3507 Details : OSTs can be permanently deactivated with e.g. 'lctl
3508 conf_param lustre-OST0001.osc.active=0'
3510 Severity : enhancement
3512 Description: MGS proc entries
3513 Details : Added basic proc entries for the MGS showing what filesystems
3516 Severity : enhancement
3518 Description: provide MGS failover
3519 Details : Added config lock reacquisition after MGS server failover.
3521 Severity : enhancement
3523 Description: add Linux 2.4 support
3524 Details : Added support for RHEL 2.4.21 kernel for 1.6 servers and clients
3528 Description: a large application tries to do I/O to the same resource and dies
3529 in the middle of it.
3530 Details : Check the req->rq_arrival time after the call to
3531 ost_brw_lock_get(), but before we do anything about
3532 processing it & sending the BULK transfer request. This
3533 should help move old stale pending locks off the queue as
3534 quickly as obd_timeout.
3537 Frequency : when an incorrect nid is specified during startup
3539 Description: ptlrpc connect to non-existant node causes kernel crash
3540 Details : LNET can't be re-entered from an event callback, which
3541 happened when we expire a message after the export has been
3542 cleaned up. Instead, hand the zombie cleanup off to another
3545 Severity : enhancement
3547 Description: plain/inodebits lock performance improvement
3548 Details : Grouping plain/inodebits in granted list by their request modes
3549 and bits policy, thus improving the performance of search through
3553 Frequency : only if OST filesystem is corrupted
3555 Description: client incorrectly hits assertion in ptlrpc_replay_req()
3556 Details : for a short time RPCs with bulk IO are in the replay list,
3557 but replay of bulk IOs is unimplemented. If the OST filesystem
3558 is corrupted due to disk cache incoherency and then replay is
3559 started it is possible to trip an assertion. Avoid putting
3560 committed RPCs into the replay list at all to avoid this issue.
3563 Frequency : liblustre (e.g. catamount) on a large cluster with >= 8 OSTs/OSS
3565 Description: System hang on startup
3566 Details : This bug allowed the liblustre (e.g. catamount) client to
3567 return to the app before handling all startup RPCs. This
3568 could leave the node unresponsive to lustre network traffic
3569 and manifested as a server ptllnd timeout.
3571 Severity : enhancement
3573 Description: Add "/proc/sys/lustre/debug_peer_on_timeout"
3574 Details : liblustre envirable: LIBLUSTRE_DEBUG_PEER_ON_TIMEOUT
3575 boolean to control whether to print peer debug info when a
3576 client's RPC times out.
3579 Frequency : only for kernels with patches from Lustre below 1.4.3
3581 Description: Remove old rdonly API
3582 Details : Remove old rdonly API which unused from at least lustre 1.4.3
3585 Frequency : only for devices with external journals
3587 Description: Set external device read-only also
3588 Details : During a commanded failover stop, we set the disk device
3589 read-only while the server shuts down. We now also set any
3590 external journal device read-only at the same time.
3593 Frequency : when upgrading from 1.4 while trying to change parameters
3595 Description: The wrong (new) MDC name was used when setting parameters for
3596 upgraded MDT's. Also allows changing of OSC (and MDC)
3597 parameters if --writeconf is specified at tunefs upgrade time.
3600 Frequency : when setting specific ost indicies
3602 Description: QOS code breaks on skipped indicies
3603 Details : Add checks for missing OST indicies in the QOS code, so OSTs
3604 created with --index need not be sequential.
3606 Severity : enhancement
3608 Description: Add uninit_groups feature to ldiskfs2 to speed up e2fsck
3609 Details : The uninit_groups feature works in conjunction with the kernel
3610 filesystem code (ldiskfs2 only) and e2fsprogs-1.39-cfs6 to speed
3611 up the pass1 processing of e2fsck. This is a read-only feature
3612 in ldiskfs2 only, so older kernels and current ldiskfs cannot
3613 mount filesystems that have had this feature enabled.
3615 Severity : enhancement
3617 Description: Improve multi-block allocation algorithm to avoid fragmentation
3618 Details : The mballoc3 code (ldiskfs2 only) adds new mechanisms to improve
3619 allocation locality and avoid filesystem fragmentation.
3621 ------------------------------------------------------------------------------
3623 2007-04-01 Cluster File Systems, Inc. <info@clusterfs.com>
3625 * Support for kernels:
3626 2.4.21-47.0.1.EL (RHEL 3)
3627 2.6.5-7.283 (SLES 9)
3628 2.6.9-42.0.10.EL (RHEL 4)
3629 2.6.12.6 vanilla (kernel.org)
3630 2.6.16.27-0.9 (SLES 10)
3631 * Recommended e2fsprogs version: 1.39.cfs5
3633 * Note that reiserfs quotas are disabled on SLES 10 in this kernel
3637 Frequency : occasional, depends on client load and configuration
3638 Bugzilla : 12181, 12203
3639 Description: data loss for recently-modified files
3641 Details : In some cases it is possible that recently written or created
3642 files may not be written to disk in a timely manner (this should
3643 normally be within 30s unless client IO load is very high).
3644 The problem appears as zero-length files or files that are a
3645 multiple of 1MB in size after a client crash or client eviction
3646 that are missing data at the end of the file.
3648 This problem is more likely to be hit on clients where files are
3649 repeatedly created and unlinked in the same directory, clients
3650 have a large amount of RAM, have many CPUs, the filesystem has
3651 many OSTs, the clients are rebooted frequently, and/or the files
3652 are not accessed by other nodes after being written.
3654 The presence of the problem can be detected by looking at
3655 /proc/sys/fs/inode-state. If the first number (nr_inodes) is
3656 smaller than the second (nr_unused) then dirty files will not
3657 be flushed automatically to disk. "sync; sleep 10" should be
3658 run several times on the node before unmounting it to update
3659 Lustre (this is also safe to run on nodes without this problem).
3661 There is also a related kernel bug in the RHEL4 4 2.6.9 kernel
3662 that can cause this same problem, so customers using that kernel
3663 also need to update the kernel in addition to Lustre. In order
3664 to properly fix this bug, the RHEL3 2.4.21 kernel is also updated.
3666 It is normal that files written just before a client crash (less
3667 than 30s) may not yet have been flushed to disk, even for local
3671 Frequency : frequent on thin XT3 nodes
3673 Description: UUID collision on thin XT3 Linux nodes
3674 Details : UUIDs on Compute Node Linux XT3 nodes were not generated
3675 randomly, since we relied on an insufficiently-seeded PRNG.
3680 Description: OSS hangs after "All ost request buffers busy"
3681 Details : A deadlock between quota and journal operations caused OSS
3682 hangs after printing "All ost request buffers busy."
3685 Frequency : always on liblustre builds
3687 Description: Cleanup compiler warnings on liblustre
3690 Frequency : always on liblustre builds on XT3
3692 Description: LC_CONFIG_CDEBUG don't run while build liblustre on XT3.
3696 Description: Addition of EXT3_FEATURE_RO_COMPAT_DIR_NLINKS flag for
3697 > 32000 subdirectories
3698 Details : Add EXT3_FEATURE_RO_COMPAT_DIR_NLINK flag to
3699 EXT3_FEATURE_RO_COMPAT_SUPP. This flag will be set whenever
3700 subdirectory count crosses 32000. This will aid e2fsck to
3701 correctly handle more than 32000 subdirectories.
3704 Frequency : liblustre (e.g. catamount) on a large cluster with >= 8 OSTs/OSS
3706 Description: System hang on startup
3707 Details : This bug allowed the liblustre (e.g. catamount) client to
3708 return to the app before handling all startup RPCs. This
3709 could leave the node unresponsive to lustre network traffic
3710 and manifested as a server ptllnd timeout.
3712 Severity : enhancement
3714 Description: Add "/proc/sys/lustre/debug_peer_on_timeout"
3715 (liblustre envirable: LIBLUSTRE_DEBUG_PEER_ON_TIMEOUT)
3716 boolean to control whether to print peer debug info when a
3717 client's RPC times out.
3722 Description: make O_SYNC working on 2.6 kernels
3723 Details : 2.6 kernels use different method for mark pages for write,
3724 so need add a code to lustre for O_SYNC work.
3729 Description: Failure to close file and release space on NFS
3730 Details : Put inode details into lock acquired in ll_intent_file_open.
3731 Use mdc_intent_lock in ll_intent_open to properly
3732 detect all kind of errors unhandled by mdc_enqueue.
3737 Description: proc file read during shutdown sometimes raced obd removal,
3739 Details : Add lock to prevent obd access after proc file removal.
3742 Frequency : Only for files larger than 4GB on 32-bit clients.
3744 Description: improperly doing page alignment of locks
3745 Details : Modify lustre core code to use CFS_PAGE_* defines instead of
3746 PAGE_*. Make CFS_PAGE_MASK a 64-bit mask.
3751 Description: RPCs being resent when they shouldn't be
3752 Details : Some RPCs that should not be resent are being resent. This
3753 can cause inconsistencies in the RPC state machine. Do not
3754 resend such requests.
3757 Frequency : rare, only with NFS export
3759 Description: Crash on NFS re-export node
3760 Details : under very unusual load conditions an assertion is hit in
3761 ll_intent_file_open()
3764 Frequency : only if OST filesystem is corrupted
3766 Description: client incorrectly hits assertion in ptlrpc_replay_req()
3767 Details : for a short time RPCs with bulk IO are in the replay list,
3768 but replay of bulk IOs is unimplemented. If the OST filesystem
3769 is corrupted due to disk cache incoherency and then replay is
3770 started it is possible to trip an assertion. Avoid putting
3771 committed RPCs into the replay list at all to avoid this issue.
3776 Description: large O_DIRECT requests fail under memory pressure/fragmentation
3777 Details : Large single O_DIRECT read and write calls can fail to allocate
3778 a sufficiently large buffer to process the request. In case of
3779 allocation failure the allocation is retried with a smaller
3780 buffer and broken into smaller requests.
3782 Severity : enhancement
3784 Description: Add -o localflock option to simulate old noflock behaviour.
3785 Details : This will achieve local-only flock/fcntl locks coherentness.
3790 Description: versioning check is incomplete
3791 Details : Checking the version difference of client vs. server, report
3792 error if the gap is too big.
3797 Description: add support PG_writeback bit
3798 Details : add support for PG_writeback bit for Lustre, for more carefull
3799 work with page cache in 2.6 kernel. This also fix some deadlocks
3800 and remove hack for work O_SYNC with 2.6 kernel.
3802 Severity : enhancement
3804 Description: Add uninit_groups feature to ldiskfs2 to speed up e2fsck
3805 Details : The uninit_groups feature works in conjunction with the kernel
3806 filesystem code (ldiskfs2 only) and e2fsprogs-1.39-cfs6 to speed
3807 up the pass1 processing of e2fsck. This is a read-only feature
3808 in ldiskfs2 only, so older kernels and current ldiskfs cannot
3809 mount filesystems that have had this feature enabled.
3811 Severity : enhancement
3813 Description: Improve multi-block allocation algorithm to avoid fragmentation
3814 Details : The mballoc3 code (ldiskfs2 only) adds new mechanisms to improve
3815 allocation locality and avoid filesystem fragmentation.
3817 ------------------------------------------------------------------------------
3819 2007-02-09 Cluster File Systems, Inc. <info@clusterfs.com>
3821 * Support for kernels:
3822 2.6.9-42.0.3.EL (RHEL 4)
3823 2.6.5-7.276 (SLES 9)
3824 2.4.21-47.0.1.EL (RHEL 3)
3825 2.6.12.6 vanilla (kernel.org)
3826 2.6.16.21-0.8 (SLES10)
3827 * Recommended e2fsprogs version: 1.39.cfs2-0
3829 * The backwards-compatible /proc/sys/portals symlink has been removed
3830 in this release. Before upgrading, please ensure that you change
3831 any configuration scripts or /etc/sysctl.conf files that access
3832 /proc/sys/portals/* or sysctl portals.* to use the corresponding
3833 entry in /proc/sys/lnet or sysctl lnet.*. This change can be made
3834 in advance of the upgrade on any system running Lustre 1.4.6 or
3835 newer, since /proc/sys/lnet was added in that version.
3836 * Note that reiserfs quotas are disabled on SLES 10 in this kernel
3840 Frequency : only when quota is used
3842 Description: avoid scanning export list for quota master
3843 Details : Change the algorithms to avoid scanning export list in order
3844 to improve the efficiency.
3847 Frequency : MDS failover only, very rarely
3849 Description: "went back in time" messages on mds failover
3850 Details : The greatest transno may be lost when the current operation
3851 finishes with an error (transno==0) and the client's last_rcvd
3852 record is over-written. Save the greatest transno in the
3853 mds_last_transno for this case.
3856 Frequency : always for specific kernels and striping counts
3858 Description: client may get "Matching packet too big" without ACL support
3859 Details : Clients compiled without CONFIG_FS_POSIX_ACL get an error message
3860 when trying to access files in certain configurations. The
3861 clients should in fact be denied when mounting because they do
3862 not understand ACLs.
3865 Frequency : Cray XT3 with more than 4000 clients and multiple jobs
3867 Description: many clients connecting with IO in progress causes connect timeouts
3868 Details : Avoid synchronous journal commits to avoid delays caused by many
3869 clients connecting/disconnecting when bulk IO is in progress.
3870 Queue liblustre connect requests on OST_REQUEST_PORTAL instead of
3871 OST_IO_PORTAL to avoid delays behind potentially many pending
3875 Frequency : occasionally with multiple writers to a single file
3877 Description: shared writes to file may result in wrong size reported by stat()
3878 Details : Allow growing of kms when extent lock is cancelled
3881 Frequency : always with random mmap IO to multi-striped file
3883 Description: mmap write might be lost if we are writing to a 'hole' in stripe
3884 Details : Only if the hole is at the end of OST object so that kms is too
3885 small. Fix is to increase kms accordingly in ll_nopage.
3888 Frequency : rare, only if OST filesystem is inconsistent with MDS filesystem
3890 Description: writes to a missing object would leak memory on the OST
3891 Details : If there is an inconsistency between the MDS and OST filesystems,
3892 such that the MDS references an object that doesn't exist, writes
3893 to that object will leak memory due to incorrect cleanup in the
3894 error handling path, eventually running out of memory on the OST.
3899 Description: Creating too long symlink causes lustre errors
3900 Details : Check symlink and name lengths before sending requests to MDS.
3903 Frequency : only if flock is enabled (not on by default)
3905 Description: posix locks not released on fd closure on 2.6.9+
3906 Details : We failed to add posix locks to list of inode locks on 2.6.9+
3907 kernels, this caused such locks not to be released on fd close and
3908 then assertions on fs unmount about still used locks.
3911 Frequency : MDS failover only, very rarely
3913 Description: clients may get ASSERTION(granted_lock != NULL)
3914 Details : When request was taking a long time, and a client was resending
3915 a getattr by name lock request. The were multiple lock requests
3916 with the same client lock handle and
3917 mds_getattr_name->fixup_handle_for_resent_request found one of the
3918 lock handles but later failed with ASSERTION(granted_lock != NULL).
3923 Description: handle->h_buffer_credits > 0, assertion failure
3924 Details : h_buffer_credits is zero after truncate, causing assertion
3925 failure. This patch extends the transaction or creates a new
3929 Frequency : NFS re-export or patchless client
3930 Bugzilla : 11179, 10796
3931 Description: Crash on NFS re-export node (__d_move)
3932 Details : We do not want to hash the dentry if we don't have a lock.
3933 But if this dentry is later used in d_move, we'd hit uninitialised
3934 list head d_hash, so we just do this to init d_hash field but
3935 leave dentry unhashed.
3938 Frequency : NFS re-export or patchless client
3940 Description: NFS exports has problem with symbolic link
3941 Details : lustre client didn't properly install dentry when re-exported
3942 to NFS or running patchless client.
3945 Frequency : NFS re-export or patchless client
3947 Description: Various nfs/patchless fixes.
3948 Details : fixes reuse disconected alias for lookup process - this fixes
3949 warning "find_exported_dentry: npd != pd",
3950 fix permission error with open files at nfs.
3951 fix apply umask when do revalidate.
3954 Frequency : occasional
3956 Description: Crash on NFS re-export node
3957 Details : calling clear_page() on the wrong pointer triggered oops in
3958 generic_mapping_read().
3961 Frequency : rarely, using O_DIRECT IO
3963 Description: unaligned directio crashes client with LASSERT
3964 Details : check for unaligned buffers before trying any requests.
3967 Frequency : rarely, using CFS RAID5 patches in non-standard kernel series
3969 Description: stale data returned from RAID cache
3970 Details : If only a small amount of IO is done to the RAID device before
3971 reading it again it is possible to get stale data from the RAID
3972 cache instead of reading it from disk.
3975 Frequency : always for sles10 kernel
3977 Description: sles10 support
3978 Details : ll_follow_link: compile fixes and using of nd_set_link
3979 under newer kernels.
3982 Frequency : depends on arch, kernel and compiler version, always on sles10
3985 Description: recursive or deep enough symlinks cause stack overflow
3986 Details : getting rid of large stack-allocated variable in
3990 Frequency : depends on hardware
3992 Description: lustre write performance loss in the SLES10 kernel
3993 Details : the performance loss is caused by using of write barriers in the
3994 ext3 code. The SLES10 kernel turns barrier support on by
3995 default. The fix is to undo that change for ldiskfs.
3997 ------------------------------------------------------------------------------
3999 2006-12-09 Cluster File Systems, Inc. <info@clusterfs.com>
4001 * Support for kernels:
4002 2.6.9-42.0.3EL (RHEL 4)
4003 2.6.5-7.276 (SLES 9)
4004 2.4.21-47.0.1.EL (RHEL 3)
4005 2.6.12.6 vanilla (kernel.org)
4009 Frequency : quota enabled and large files being deleted
4011 Description: releasing more than 4GB of quota at once hangs OST
4012 Details : If a user deletes more than 4GB of files on a single OST it
4013 will cause the OST to spin in an infinite loop. Release
4014 quota in < 4GB chunks, or use a 64-bit value for 1.4.7.1+.
4019 Description: statfs data retrieved from /proc may be stale or zero
4020 Details : When reading per-device statfs data from /proc, in the
4021 {kbytes,files}_{total,free,avail} files, it may appear
4022 as zero or be out of date.
4025 Frequency : systems with MD RAID1 external journal devices
4027 Description: lconf's call to blkid is confused by RAID1 journal devices
4028 Details : Use the "blkid -l" flag to locate the MD RAID device instead
4029 of returning all block devices that match the journal UUID.
4032 Frequency : always, for aggregate stripe size over 4GB
4034 Description: "lfs setstripe" fails assertion when setting 4GB+ stripe width
4035 Details : Using "lfs setstripe" to set stripe size * stripe count over 4GB
4036 will fail the kernel with "ASSERTION(lsm->lsm_xfersize != 0)"
4039 Frequency : always if "lfs find" used on a local file/directory
4041 Description: "lfs find" segfaults if used on a local file/directory
4042 Details : The case where a directory component was not specified wasn't
4043 handled correctly. Handle this properly.
4046 Frequency : always on ppc64
4048 Description: the write to an ext3 filesystem mounted with mballoc got stuck
4049 Details : ext3_mb_generate_buddy() uses find_next_bit() which does not
4050 perform endianness conversion.
4053 Frequency : rarely (truncate to non-zero file size after write under load)
4054 Bugzilla : 10730, 10687
4055 Description: Files padded with zeros to next 4K multiple
4056 Details : With filesystems mounted using the "extents" option (2.6 kernels)
4057 it is possible that files that are truncated to a non-zero size
4058 immediately after being written are filled with zero bytes beyond
4059 the truncated size. No file data is lost.
4061 Severity : enhancement
4063 Description: Allow recovery/failover for liblustre clients.
4064 Details : liblustre clients were unaware of failover configurations until
4067 Severity : enhancement
4069 Description: user file locks should fail when not mounting with flock option
4070 Details : Set up an error-returning stub in ll_file_operations.lock field
4071 to prevent incorrect behaviour when client is mounted without
4072 flock option. Also, set up properly f_op->flock field for
4076 Frequency : always on ia64
4078 Description: "lfs df" loops on printing out MDS statfs information
4079 Details : The obd_ioctl_data was not initialized and in some systems
4080 this caused a failure during the ioctl that did not return
4081 an error. Initialize the struct and return an error on failure.
4084 Frequency : SLES 9 only
4086 Description: Error of copying files with lustre special EAs as root
4087 Details : Client side always return success for setxattr call for lustre
4088 special xattr (currently only "trusted.lov").
4091 Frequency : rarely on clusters with both ia64+i386 clients
4093 Description: ia64+i686 clients doing shared IO on the same file may LBUG
4094 Details : In rare cases when both ia64+i686 (or other mixed-PAGE_SIZE)
4095 clients are doing concurrent writes to the same file it is
4096 possible that the ia64 clients may LASSERT because the OST
4097 extent locks are not PAGE_SIZE aligned. Ensure that grown
4098 locks are always aligned on the request boundary.
4101 Frequency : specific use, occasional
4103 Description: Overwriting in use executable truncates on-disk binary image
4104 Details : If one node attempts to overwrite an executable in use by
4105 another node, we now correctly return ETXTBSY instead of
4106 truncating the file.
4108 Severity : enhancement
4110 Description: Async OSC create to avoid the blocking unnecessarily.
4111 Details : If an OST has no remaining object, system will block on the creating
4112 when need to create a new object on this OST. Now, ways use
4113 pre-created objects when available, instead of blocking on an
4114 empty osc while others are not empty. If we must block, we block
4115 for the shortest possible period of time.
4120 Description: chmod on Lustre root is propagated to other clients
4121 Details : Re-validate root's dentry in ll_lookup_it to avoid having it
4122 invalid by the follow_mount time.
4125 Frequency : liblustre clients only
4127 Description: Race in 'instant cancel' lock handling could lead to such locks
4128 never to be granted in case of SMP MDS
4129 Details : Do not destroy not yet granted but cbpending locks in
4133 Frequency : replay/resend of open
4135 Description: non null lock assetion failure in mds_intent_policy
4136 Details : Trying to replay/resend lockless open requests resulted in
4137 mds_open() returning 0 with no lock. Now it sets a flag if
4138 a lock is going to be returned.