3 * See https://wiki.hpdd.intel.com/display/PUB/Lustre+Support+Matrix
4 for currently supported client and server kernel versions.
5 * Server known to build on patched kernels:
6 2.6.32-431.29.2.el6 (RHEL6.5)
7 2.6.32-504.30.3.el6 (RHEL6.6)
8 2.6.32-573.18.1.el6 (RHEL6.7)
9 3.10.0-327.3.1.el7 (RHEL7.2)
10 3.0.101-0.47.71 (SLES11 SP3)
11 3.0.101-68 (SLES11 SP4)
12 vanilla linux 4.2.1 (ZFS only)
13 * Client known to build on unpatched kernels:
14 2.6.32-431.29.2.el6 (RHEL6.5)
15 2.6.32-504.30.3.el6 (RHEL6.6)
16 2.6.32-573.18.1.el6 (RHEL6.7)
17 3.10.0-327.3.1.el7 (RHEL7.2)
18 3.0.101-0.47.71 (SLES11 SP3)
19 3.0.101-68 (SLES11 SP4)
22 * Recommended e2fsprogs version: 1.42.13.wc4 or newer
23 * Recommended ZFS / SPL version: 0.6.4.2
24 * Tested with ZFS / SPL version: 0.6.4.2
25 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
26 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
27 stack overflow. For more information, please refer to bugzilla 17630.
28 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
29 the re-exporting nodes to be mounted with "32bitapi" mount option
31 --------------------------------------------------------------------------------
33 02-29-2016 Intel Corporation
35 * See https://wiki.hpdd.intel.com/display/PUB/Lustre+Support+Matrix
36 for currently supported client and server kernel versions.
37 * Server known to build on patched kernels:
38 2.6.32-431.29.2.el6 (RHEL6.5)
39 2.6.32-504.30.3.el6 (RHEL6.6)
40 2.6.32-573.12.1.el6 (RHEL6.7)
41 3.10.0-327.3.1.el7 (RHEL7.2)
42 3.0.101-0.47.71 (SLES11 SP3)
43 3.0.101-68 (SLES11 SP4)
44 vanilla linux 4.2.1 (ZFS only)
45 * Client known to build on unpatched kernels:
46 2.6.32-431.29.2.el6 (RHEL6.5)
47 2.6.32-504.30.3.el6 (RHEL6.6)
48 2.6.32-573.12.1.el6 (RHEL6.7)
49 3.10.0-327.3.1.el7 (RHEL7.2)
50 3.0.101-0.47.71 (SLES11 SP3)
51 3.0.101-68 (SLES11 SP4)
54 * Recommended e2fsprogs version: 1.42.13.wc4 or newer
55 * Recommended ZFS / SPL version: 0.6.4.2
56 * Tested with ZFS / SPL version: 0.6.4.2
57 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
58 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
59 stack overflow. For more information, please refer to bugzilla 17630.
60 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
61 the re-exporting nodes to be mounted with "32bitapi" mount option
63 --------------------------------------------------------------------------------
65 03-10-2015 Intel Corporation
67 * See https://wiki.hpdd.intel.com/display/PUB/Lustre+Support+Matrix
68 for currently supported client and server kernel versions.
69 * Server known to build on patched kernels:
70 2.6.32-431.29.2.el6 (RHEL6.5)
71 2.6.32-504.8.1.el6 (RHEL6.6)
72 3.0.101-0.46 (SLES11 SP3)
73 * Client known to build on unpatched kernels:
74 2.6.32-431.29.2.el6 (RHEL6.5)
75 2.6.32-504.8.1.el6 (RHEL6.6)
76 3.10.0-123.20.1.el7 (RHEL7)
77 3.0.101-0.46 (SLES11 SP3)
78 * Recommended e2fsprogs version: 1.42.9.wc1 or newer
79 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
80 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
81 stack overflow. For more information, please refer to bugzilla 17630.
82 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
83 the re-exporting nodes to be mounted with "32bitapi" mount option
85 Severity : enhancement
87 Description: control OST-index in IDIF via ROCOMPAT flag.
88 Details : Introduce new flag OBD_ROCOMPAT_IDX_IN_IDIF that is stored in the
89 last_rcvd file. For new formatted OST device, it will be auto set;
90 for the case of upgrading from old OST device, you can enable it
91 via the lproc interface osd-ldiskfs.index_in_idif. With such flag
92 enabled, for new created OST-object, its IDIF-in-LMA will contain
93 the OST-index; for the existing OST-object, the OSD will convert
94 old format IDIF as new format IDIF with OST-index stored in the
95 LMA EA when accessing such OST-object or via OI scrub. Once such
96 flag is enabled, it cannot be reverted back, so the system cannot
97 be downgraded to the orignal incompatible version.
99 --------------------------------------------------------------------------------
101 07-30-2014 Intel Corporation
103 * See https://wiki.hpdd.intel.com/display/PUB/Lustre+Support+Matrix
104 for currently supported client and server kernel versions.
105 * Server known to build on patched kernels:
106 2.6.32-431.20.3.el6 (RHEL6.5)
107 3.0.101-0.31 (SLES11 SP3)
108 * Client known to build on unpatched kernels:
109 2.6.32-431.20.3.el6 (RHEL6.5)
110 3.0.101-0.31 (SLES11 SP3)
113 * Recommended e2fsprogs version: 1.42.9.wc1 or newer
114 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
115 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
116 stack overflow. For more information, please refer to bugzilla 17630.
117 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
118 the re-exporting nodes to be mounted with "32bitapi" mount option
120 --------------------------------------------------------------------------------
122 10-31-2013 Intel Corporation
124 * See https://wiki.hpdd.intel.com/display/PUB/Lustre+Support+Matrix
125 * Server builds for kernels:
126 2.6.32-358.18.1.el6 (RHEL6.4)
127 * Client builds for unpatched kernels:
128 2.6.32-358.18.1.el6 (RHEL6.4)
129 3.0.92-0.8 (SLES11 SP3)
131 * Recommended e2fsprogs version: 1.42.7.wc2 or newer
132 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
133 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
134 stack overflow. For more information, please refer to bugzilla 17630.
135 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
136 the re-exporting nodes to be mounted with "32bitapi" mount option
138 --------------------------------------------------------------------------------
140 05-30-2013 Intel Corporation
142 * See https://wiki.hpdd.intel.com/display/PUB/Lustre+Support+Matrix
143 * Server builds for kernels:
144 2.6.32-358.6.2.el6 (RHEL6.4)
145 * Client builds for unpatched kernels:
146 2.6.32-358.6.2.el6 (RHEL6.4)
147 3.0.74-0.6.6 (SLES11 SP2)
149 * Recommended e2fsprogs version: 1.42.7.wc1 or newer
150 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
151 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
152 stack overflow. For more information, please refer to bugzilla 17630.
153 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
154 the re-exporting nodes to be mounted with "32bitapi" mount option
156 --------------------------------------------------------------------------------
158 9-30-2012 Whamcloud, Inc.
160 * See https://wiki.hpdd.intel.com/display/PUB/Lustre+Support+Matrix
161 * Server support for kernels:
162 2.6.32-279.5.1.el6 (RHEL6)
163 * Client support for unpatched kernels:
164 2.6.32-279.5.1.el6 (RHEL6)
166 2.6.18-238.19.1.el5 (RHEL5)
167 2.6.16.60-0.69.1 (SLES10)
168 * Recommended e2fsprogs version: 1.41.90.wc4
169 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
170 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
171 stack overflow. For more information, please refer to bugzilla 17630.
172 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
173 the re-exporting nodes to be mounted with "32bitapi" mount option
175 Severity : enhancement
177 Description: Update RHEL6.3 kernel to 2.6.32-279.5.1.el6.
179 Severity : enhancement
181 Description: Update RHEL6.3 kernel to 2.6.32-279.2.1.el6.
183 Severity : enhancement
185 Description: Update RHEL6.2 kernel to 2.6.32-220.17.1.el6
187 Severity : enhancement
189 Description: Update RHEL6.2 kernel to 2.6.32-220.13.1.el6
191 Severity : enhancement
193 Description: Update RHEL6.2 kernel to 2.6.32-220.7.1.el6
195 --------------------------------------------------------------------------------
197 04-30-2012 Whamcloud, Inc.
199 * See https://wiki.hpdd.intel.com/display/PUB/Lustre+Support+Matrix
200 * Server support for kernels:
201 2.6.32-220.4.2.el6 (RHEL6)
202 * Client support for unpatched kernels:
203 2.6.32-220.4.2.el6 (RHEL6)
205 2.6.18-238.19.1.el5 (RHEL5)
206 2.6.16.60-0.69.1 (SLES10)
207 * Recommended e2fsprogs version: 1.41.90.wc4
208 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
209 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
210 stack overflow. For more information, please refer to bugzilla 17630.
211 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
212 the re-exporting nodes to be mounted with "32bitapi" mount option
214 Severity : enhancement
216 Description: Update RHEL6.2 kernel to 2.6.32-220.4.2.el6
218 Severity : enhancement
220 Description: Update RHEL6.2 kernel to 2.6.32-220.4.1.el6
222 Severity : enhancement
224 Description: Update RHEL6.2 kernel to 2.6.32-220.el6
226 Severity : enhancement
228 Description: Update RHEL6.1 kernel to 2.6.32-131.17.1.el6
230 --------------------------------------------------------------------------------
232 09-30-2011 Whamcloud, Inc.
234 * Server support for kernels:
235 2.6.18-238.19.1.el5 (RHEL5)
236 2.6.32-131.6.1.el6 (RHEL6)
237 * Client support for unpatched kernels:
238 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
239 2.6.18-238.19.1.el5 (RHEL5)
240 2.6.32-131.6.1.el6 (RHEL6)
241 2.6.29.4-167.fc11 (FC11)
242 2.6.16.60-0.69.1 (SLES10)
244 * Recommended e2fsprogs version: 1.41.90.wc3
245 * NFS export disabled when stack size < 8192. Since the NFSv4 export of
246 Lustre filesystem with 4K stack may cause a stack overflow. For more
247 information, please refer to bugzilla 17630.
248 * NFSv4 reexport requires Lustre client on the re-exporting nodes to be
249 mounted with "32bitapi" mount option
251 Severity : enhancement
252 Description: Update RHEL6.1 kernel to 2.6.32-131.6.1.el6
254 Severity : enhancement
255 Description: Update RHEL5.6 kernel to 2.6.18-238.19.1.el5
257 Severity : enhancement
258 Description: Update RHEL5.6 kernel to 2.6.18-238.9.1.el5
261 Frequency : only with SLES10
263 Description: Use OFED "KMP" provided by Novell
264 Details : SLES10 SP3 ships with OFED in a separate "KMP" package.
265 Lustre is now built against this package. That means you need to
266 install the ofed-kmp package from Novell for the patchless client.
267 Note that the ofed-kmp that Novell ships may not exactly match the
268 kernel version but should still be compatible.
270 Severity : enhancement
272 Description: Update RHEL5.5 kernel to 2.6.18-194.17.1.el5
273 Update OEL5.5 kernel to 2.6.18-194.17.1.0.1.el5.
275 --------------------------------------------------------------------------------
277 2010-07-15 Oracle, Inc.
279 * Server support for kernels:
280 2.6.18-164.11.1.el5 (RHEL 5)
281 2.6.18-164.11.1.0.1.el5 (OEL 5)
282 * Client support for unpatched kernels:
283 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
284 2.6.18-164.11.1.el5 (RHEL 5),
285 2.6.18-164.11.1.0.1.el5 (OEL 5)
286 2.6.16.60-0.42.8 (SLES 10),
288 2.6.29.4-167.fc11 (FC11)
289 * Recommended e2fsprogs version: 1.41.10-sun2
291 Severity : enhancement
293 Description: Distribute lfs_migrate script with lustre
294 Details : The lfs_migrate mentioned in the manual is now included in the
295 lustre tree in lustre/scripts and will distributed in lustre RPMs.
297 Severity : enhancement
299 Description: Failover nid list added to import proc file.
301 Severity : enhancement
303 Description: "som_preview" mount option required to use SOM
304 Details : The clients will need to use som_preview mount option
308 Description: Lustre does not do 1MB IOs to HW RAID
309 Details : Bump MAX_PHYS/HW_SEGMENTS and SG_ALL to 256 in the RHEL5 kernel.
311 Severity : enhancement
313 Description: lfs quota output is not very convenient for awk/sed-parsing
314 Details : Some positions in lfs quota output table could be empty or
315 non-empty which made it hard to parse it with scripts, now
316 a dash is put instead of space where there is not supposed
321 Description: Update to OFED 1.5.1
322 Details : OFED 1.5.1 will be used in environments that don't use
323 vendor supplied OFED.
327 Description: Remove lnet. prefix from lctl params display
328 Details : Also renamed lnet.memused to lnet_memused to
333 Description: Use conf_param -d to remove permanant settings.
337 Description: Add list_param -R
338 Details : lctl list_param now supports recursive mode.
341 Bugzilla : 21829, 22312
342 Description: Fix llobdstat.
343 Details : 1. Now updating the write_bytes counter
344 2. Port script fixes from 1.8.
348 Description: decrease the usage of memory on clients.
349 Details : 1. On clients, recycle dentries and inodes unused.
350 2. Delete the code related to ll_deathrow(att 6215 in bug 1443). It
355 Description: llobdstat fix and enhancement
356 Details : add a counter to set a limit to how many samples will be returned
357 fix a wildcard in the path to limit to obdfilter stats only
361 Description: lfs check only allowed for root
362 Details : Code cleanups and added a regresion test of non-root use of
367 Description: MDS Oopses in ptlrpc_dispatch_difficult_reply()/__list_add()
368 Details : Fix hr->hr_index overload.
372 Description: Fix FIEMAP functionality.
376 Description: FC11 patchless client support added
380 Description: Update RHEL5.4 kernel to 2.6.18-164.11.1.el5 and
381 OEL5.4 kernel to 2.6.18-164.11.1.0.1.el5.
385 Description: New acc-sm script and various test-framework enhancments.
389 Description: Previously landed patch was reverted.
393 Description: Update SLES10 kernel to 2.6.16.60-0.42.8 (patchless)
397 Description: Add ldiskfs maxdirsize mount option.
401 Description: Fix mkfs.lustre support for 16TB LUNs.
405 Description: Print mmp_check_interval
406 Details : print mmp_check_interval and make it possible to abort
407 mount operation in case it takes too long.
411 Description: Add cascading_rw to lustre mpi tests.
415 Description: Store SOM data in EA.
416 Details : Also several SOM related cleanups.
420 Description: Add refcount to md_open_data
424 Description: Use refcount on lov_request_set to prevent memory corruption.
428 Description: Statahead: use parent dir's i_mutex for synchronization.
432 Description: loadgen improvements.
436 Description: Enable peer health detection in ksocklnd and ko2iblnd.
440 Description: Control DCACHE_LUSTRE_INVALID flag with MDS_INODE_LOOKCUP lock
444 Description: Fix race on obd_devcie:md_stats
448 Description: Fix SLES 11 client builds.
449 Details : Resolved uninitalized pointer gcc warnings.
453 Description: Allow non-root access for "lfs check".
454 Details : Added a check in obd_class_ioctl() for OBD_IOC_PING_TARGET.
458 Description: Clear imp_force_reconnect correctly in ptlrpc_connect_interpret()
460 Severity : enhancement
462 Description: Update kernel to RHEL5.4 2.6.18-164.6.1.el5 and
463 OEL5 2.6.18-164.6.1.0.1.el5(Both in-kernel OFED enabled).
467 Description: Sanity test_27p fails.
468 Details : OBD_FAIL_OST_ENOSPC has been moved to filter_statfs().
469 Sanity tests 27 were cleaned up and improved.
473 Description: Unify req format on client/servers
474 Details : Use new req_capsule API [almost] everywhere instead of old PTLRPC
475 buffers and swabbers approach..
480 Description: drop join file stuff.
482 Severity : enhancement
484 Description: Add 2.6.27 and SLES11 (patchless client) support
488 Description: Fix in ptlrpc_expire_one_request() to print the signed time
491 Severity : enhancement
493 Description: Build kernels (RHEL5, OEL5 and SLES10/11) using the vendor's own
496 Severity : enhancement
498 Description: Remove set_info(KEY_UNLINKED) from MDS/OSC
500 Severity : enhancement
502 Description: correctly handle big reply message.
503 Details : send LNet event if reply is bigger than buffer and adjust this buffer
508 Description: Drop unnecessary __GFP_NOMEMALLOC flag from filter_get_page()
510 Severity : enhancement
512 Description: Add support for OEL5.
516 Description: Conf-sanity.sh 50g test - deactivated OST should not cause a panic.
518 Severity : enhancement
520 Description: "weak-modules" support
521 Details : Implement "weak-modules" support which enables kernel modules
522 to be used with any kernel that implements the same kABI. In
523 order to achieve this modules are now installed in
524 /lib/modules/$(uname -r)/updates/kernel on all distributions.
528 Description: client could not reconnect to OST because of an active request.
529 Details : The client now retries to reconnect to the same server, if a
530 connect request failed with EBUSY or -EAGAIN. Test case 35b
531 conf-sanity.sh has been created.
535 Description: Don't allow make backward step on assignin osc next id.
536 Details : race between allocation next id and ll_sync thread can be cause
537 of set wrong osc next id and can be kill valid ost objects.
541 Description: Changes in raid5-large-io-rhel5.patch to calculate sectors properly
545 Description: Increase the default BLK_DEF_MAX_SECTORS value for RHEL5 and SLES11
549 Description: Error handling in osc_statfs_interpret() has been improved.
550 Details : Check in osc_statfs_interpret() for EBADR.
554 Description: Do not send statfs() requests to OSTs disabled by administrator.
555 Details : Check in lov_prep_statfs_set() for non-NULL ltd_exp.
557 Severity : enhancement
559 Description: Added sanity-quota test to for lfs(1) to work with numeric uid/gid.
561 Severity : enhancement
563 Description: Update kernel to RHEL5 2.6.18-128.7.1.el5.
565 Severity : enhancement
567 Description: Enhancement for lfs(1) command to use numeric uid/gid.
571 Description: Improved cleanup procedure for conf-sanity test_46a.
575 Description: Do not start bulk transfer, if client tried to reconnect.
576 Details : Do not start bulk transfer, if client tried to reconnect,
577 but failed due to active rpcs. Check for exp_abort_active_req
578 has been implemented.
582 Description: Prevent inconsistences between linux and lustre mount structures.
583 Details : Wait indefinitely in server_wait_finished() until mnt_count drops.
584 Make the sleep interruptible.
586 Severity : enhancement
588 Description: provide server to client comms path
589 Details : server to client communications path via new LDLM_SET_INFO rpc
593 Description: truncate starts GFP_FS allocation under transaction causing deadlock
594 Details : ldiskfs_truncate calls grab_cache_page which may start page
595 allocation under an open transaction. This may lead to
596 calling prune_icache with consequent lustre reentrance.
600 Description: send correctly lsm on open replay
601 Details : MDS is trust to LSM size on replay open, but client can set wrong size
606 Description: enable client interface failover
607 Details : When a child reconnects from another NID, properly update export
608 nid hash position and ldlm reverse import.
610 Severity : enhancement
612 Description: Communicate OST degraded/readonly state via statfs to MDS
613 Details : Flags in the statfs returned from OSTs indicate whether the
614 OST is in a degraded RAID state, or if the filesystem has
615 turned read-only after a filesystem error is detected.
620 Description: don't panic if EPROTO was hit when reading symlink
621 Details : correctly handling request reference in error cases.
623 Severity : enhancement
625 Description: Add LustreNetLink, a kernel-userspace communcation path.
627 Severity : enhancement
629 Description: Update kernel to SLES10 SP2 2.6.16.60-0.39.3.
634 Description: access to llog context before init.
635 Details : move handling CATALOGS file at osc layer and forbid access to llog
639 Frequency : rare, only when client is writing to OST on local node
641 Description: Avoid deadlock for local client writes
642 Details : Use new OBD_BRW_MEMALLOC flag to notify OST about writes in the
643 memory freeing context. This allows OST threads to set the
644 PF_MEMALLOC flag on task structures in order to allocate memory
645 from reserved pools and complete IO.
646 Use GFP_HIGHUSER for OST allocations for non-local client writes,
647 so that the OST threads generate memory pressure and allow
648 inactive pages to be reclaimed.
650 Severity : enhancement
652 Description: Update kernel to RHEL5.3 2.6.18-128.1.14.el5.
657 Description: lock ordering violation between &cli->cl_sem and _lprocfs_lock
658 Details : move ldlm namespace creation in setup phase to avoid grab
659 _lprocfs_lock with cli_sem held.
663 Description: Temporarily disable grant shrink.
664 Details : Disable the feature for debugging.
667 Frequency : only when formatting test filesystems
669 Description: Unable to run several mkfs.lustre on loop devices at the same time
670 Details : mkfs.lustre returns error 256 on the concurrent loop devices
671 formatting. The solution is to proper handle the error.
673 Severity : enhancement
675 Description: Update kernel to RHEL5.3 2.6.18-128.1.6.el5.
677 Severity : enhancement
679 Description: Update kernel to SLES10 SP2 2.6.16.60-0.37.
684 Description: resolve race between obd_disconnect and class_disconnect_exports
685 Details : if obd_disconnect will be called to already disconnected export he
686 forget release one reference and osc module can't unloaded.
688 Severity : enhancement
690 Description: Allow tuning service thread via /proc
691 Details : For each service a new
692 /proc/fs/lustre/{service}/*/thread_{min,max,started} entry is
693 created that can be used to set min/max thread counts, and get the
694 current number of running threads.
699 Descriptoin: don't return error if have partially created objects for file.
700 Details : lov_update_create_set uses set->set_success as index for created
701 objects, so if some requests will be failed, they will have hole
702 at end of array and qos_shrink_lsm for allocate correct lsm.
704 Severity : enhancement
706 Description: Update OFED support to 1.4.1
711 Description: Reduce small size read RPC
712 Details : Set read-ahead limit for every file and only do read-ahead when
713 available read-ahead pages are bigger than 1M to avoid small size
716 Severity : enhancement
718 Description: move AT tunable parameters for more consistent usage
719 Details : add AT tunables under /proc/sys/lustre, add to conf_param parsing
721 Severity : enhancement
723 Description: add lazystatfs mount option to allow statfs(2) to skip down OSTs
724 Details : allow skip disconnected ost for send statfs request and hide error
730 Description: fix racy locking of mballoc block bitmaps causing BUG
731 Details : The locking of the mballoc buddy bitmap and the in-memory
732 block bitmap was using two different spin locks in some
733 cases. This made it possible to incorrectly access the
734 mballoc bitmap while another process was modifying it,
735 causing a sanity assertion to fail. While no on-disk corruption
736 was reported, there was some risk of this happening.
739 Frequency : rare, on llog test 6
741 Descriptoin: don't allow connect to already connected import
742 Details : allowing connect to already connected import is hide connecting problem.
747 Descriptoin: allow kill process which wait statahead result
748 Details : for some reasons 'ls' can stick in waiting result from statahead,
749 in this case need way for kill this process.
751 Severity : enhancement
753 Description: Add state history info file, enhance import info file
754 Details : Track import connection state changes in a new osc/mdc proc file;
755 add overview-type data to the osc/mdc import proc file.
757 Severity : enhancement
759 Description: MDS create should not wait for statfs RPC while holding DLM lock.
761 Severity : enhancement
763 Description: Update to RHEL5U3 kernel-2.6.18-128.1.1.el5.
768 Descriptoin: OST grant too much space to client even there are not enough space.
769 Details : Client will shrink its grant cache to OST if there are no write
770 activity over 6 mins (GRANT_SHRINK_INTERVAL), and OST will retrieve
771 this grant cache if there are already not enough avaible space
772 (left_space < total_clients * 32M).
775 Frequency : start MDS on uncleanly shutdowned MDS device
777 Descriptoin: ll_sync thread stay in waiting mds<>ost recovery finished
778 Details : stay in waiting mds<>ost recovery finished produce random bugs
779 due race between two ll_sync thread for one lov target. send
780 ACTIVATE event only if connect realy finished and import have
784 Frequency : rare, connect and disconnect target at same time
786 Descriptoin: ASSERTION(atomic_read(&imp->imp_inflight) == 0
787 Details : don't call obd_disconnect under lov_lock. this long time
788 operation and can block ptlrpcd which answer to connect request.
793 Descriptoin: don't lose wakeup for imp_recovery_waitq
794 Details : recover_import_no_retry or invalidate_import and import_close can
795 both sleep on imp_recovery_waitq, but we was send only one wakeup
799 Frequency : always with long access acl
801 Descriptoin: mds can't pack reply with long acl.
802 Details : mds don't control size of acl but they limited by reint/getattr
805 Severity : enhancement
807 Description: Update to SLES10 kernel-2.6.16.60-0.33.
809 Severity : enhancement
811 Description: Update to RHEL5 kernel-2.6.18-92.1.22.el5.
814 Frequency : start MDS on uncleanly shutdowned MDS device
816 Descriptoin: aborting recovery hang on MDS
817 Details : don't throttle destroy RPCs for the MDT.
820 Frequency : on remount
822 Description: external journal device not working after the remount
823 Details : clear dev_rdonly flag for external journal devices in
829 Description: shutdown vs evict race
830 Details : client_disconnect_export vs connect request race.
831 if client will evicted at this time - we start invalidate
832 thread without referece to import and import can be freed
836 Frequency : rare, need acl's on inode.
838 Description: client can't handle ost additional correctly
839 Details : if ost was added after client connected to mds client can have
840 hit lnet_try_match_md ... to big messages to wide striped files.
841 in this case need teach client to handle config events about add
842 lov target and update client max ea size at that event.
844 Severity : enhancement
846 Description: Changelogs
847 Details : Changelogs are a lightweight mechanism to track filesystem
848 metadata and namespace changes. The changelog is recorded
849 permanently on the MDTs, and is periodically "consumed" / purged
850 when records are no longer needed.
852 Severity : enhancement
854 Description: compact fld format with extents
855 Details : Store range of seq rather than every seq in FLD. Seq
856 controller update FLD rather than clients. In Case of CMD, mdt0
857 has FLD, all other metadata server act as non persistent proxy
858 for FLD queries and cache fld entries in fld cache.
863 Description: don't skip ost target if they assigned to file
864 Details : Drop slow OSCs if we can, but not for requested start idx.
865 This means "if OSC is slow and it is not the requested
866 start OST, then it can be skipped, otherwise skip it only
867 if it is inactive/recovering/out-of-space.
871 Description: more cleanup in mds_lov
872 Details : not send LOV EA under replay, we can't know about they size at this
873 time. Don't allow client connect to mds before any ost connected,
874 for avoid problems with LOV EA size and returning EIO to client.
876 Severity : enhancement
878 Description: Interoperability at server side (Disk interoperability)
880 Severity : enhancement
882 Description: Update to RHEL5 kernel-2.6.18-92.1.17.el5.
884 Severity : enhancement
886 Description: Update to SLES10 SP2 kernel-2.6.16.60-0.31.
888 Severity : enhancement
890 Description: New client IO stack (CLIO).
892 Severity : enhancement
894 Description: Commit on sharing. Eliminate inter-client dependencies between
895 uncommitted transactions by doing transaction commits.
896 Thereby clients may recovery independently.
899 Frequency : Create a symlink file with a very long name
901 Description: ldlm_cancel_pack()) ASSERTION(max >= dlm->lock_count + count)
902 Details : If there is no extra space in the request for early cancels,
903 ldlm_req_handles_avail() returns 0 instead of a negative value.
905 Severity : enhancement
907 Description: Add /proc entry for import status
908 Details : The mdc, osc, and mgc import directories now have
909 an import directory that contains useful import data for debugging
912 Severity : enhancement
914 Description: Re-disable certain /proc logging
915 Details : Enable and disable client's offset_stats, extents_stats and
916 extents_stats_per_process stats logging on the fly.
919 Frequency : Only on FC kernels 2.6.22+
921 Description: oops in statahead
922 Details : Do not drop reference count for the dentry from VFS when lookup,
923 VFS will do that by itself.
925 Severity : enhancement
927 Description: Generic /proc file permissions
928 Details : Set /Proc file permissions in a more generic way to enable non-
929 root users operate on some /proc files.
933 Description: Hitting mdc_commit_close() ASSERTION
934 Details : Properly handle request reference release in
935 ll_release_openhandle().
939 Description: quota recovery deadlock during mds failover
940 Details : This patch includes att18982, att18236, att18237 in bz14840.
942 1. fix osts hang when mds does failover with quotaon
943 2. prevent watchdog storm when osts threads wait for the
948 Frequency : only patchless client
949 Description: add workaround for race between add/remove dentry from hash
951 Severity : enhancement
953 Description: Allow OST glimpses to return PW locks
957 Description: LBUG when llog conf file is full
958 Details : When llog bitmap is full, ENOSPC should be returned for plain log.
962 Description: Prevent import from entering FULL state when server in recovery
966 Description: service mount cannot take device name with ":"
967 Details : Only when device name contains ":/" will mount treat it as
973 Description: replace ptlrpcd with the statahead thread to interpret the async
974 statahead RPC callback
978 Frequency : on recovery
979 Description: I/O failures after umount during fail back
980 Details : if client reconnected to restarted server we need join to recovery
981 instead of find server handler is changed and process self
982 eviction with cancel all locks.
984 Severity : enhancement
986 Description: Update to RHEL5 kernel-2.6.18-92.1.10.el5.
988 Severity : enhancement
990 Description: Update to SLES10 SP2 kernel-2.6.16.60-0.27.
992 Severity : enhancement
994 Description: Upcall on Lustre log has been dumped
995 Details : Allow for a user mode script to be called once a Lustre log has
996 been dumped. It passes the filename of the dumped log to the
997 script, the location of the script can be specified via
998 /proc/sys/lnet/debug_log_upcall.
1003 Description: avoid idr_remove called for id which is not allocated.
1004 Details : Move assigment s_dev for clustered nfs to end of initialization,
1005 to avoid problem with error handling.
1010 Description: avoid Already found the key in hash [CONN_UNUSED_HASH] messages
1011 Details : When connection is reused this not moved from CONN_UNUSED_HASH
1012 into CONN_USED_HASH and this prodice warning when put connection
1013 again in unused hash.
1016 Severity : enhancement
1018 Description: File striping can now be set to use an arbitrary pool of OSTs.
1020 Severity : enhancement
1022 Description: Export bytes_read/bytes_write count on OSC/OST.
1026 Description: Early reply size mismatch, MGC loses connection
1027 Details : Apply the MGS_CONNECT_SUPPORTED mask at reconnect time so
1028 the connect flags are properly negotiated.
1033 Description: quotas are not honored with O_DIRECT
1034 Details : all writes with the flag O_DIRECT will use grants which leads to
1035 this problem. Now using OBD_BRW_SYNC to guard this.
1039 Description: add quota statistics
1040 Details : 1. sort out quota proc entries and proc code.
1041 2. add quota statistics
1043 Severity : enhancement
1045 Description: enable quota support for HEAD.
1049 Description: Properly propagate oinfo flags from lov to osc for statfs
1050 Details : restore missing copy oi_flags to lov requests.
1052 Severity : enhancement
1054 Description: Add man pages for llobdstat(8), llstat(8), plot-llstat(8),
1055 : l_getgroups(8), lst(8), routerstat(8)
1056 Details : included man pages for llobdstat(8), llstat(8),
1057 : plot-llstat(8), l_getgroups(8), lst(8), routerstat(8)
1059 Severity : enhancement
1061 Description: configure's --enable-quota should check the
1062 : kernel .config for CONFIG_QUOTA
1063 Details : configure is terminated if --enable-quota is passed but
1064 : no quota support is in kernel
1068 Description: Remove portals compatibility
1069 Details : Remove portals compatibility, not interoperable with releases
1074 Description: Resolve device initialization race
1075 Details : Prevent proc handler from accessing devices added to the
1076 obd_devs array but yet be intialized.
1078 Severity : enhancement
1080 Description: Update to SLES10 SP2 kernel-2.6.16.60-0.23.
1082 Severity : enhancement
1084 Description: Update to RHEL5 kernel-2.6.18-92.1.6.el5.
1089 Description: Using wrong pointer in osc_brw_prep_request
1090 Details : Access to array[-1] can produce panic if kernel compiled with
1091 CONFIG_PAGE_ALLOC enabled
1095 Description: Client runs out of low memory
1096 Details : Consider only lowmem when counting initial number of llap pages
1100 Description: *optional* service tags registration
1101 Details : if the "service tags" package is installed on a Lustre node
1102 When the filesystem is mounted, a local-node service tag will
1103 be created. See http://inventory.sun.com/ for more information
1104 about the Service Tags asset management system.
1108 Description: Kernel BUG tries to release flock
1109 Details : Lustre does not destroy flock lock before last reference goes
1110 away. So always drop flock locks when client is evicted and
1111 perform unlock regardless of successfulness of speaking to MDS.
1115 Description: add refcount protection for osc callbacks, avoid panic on shutdown
1119 Description: sanity test 65a fails if stripecount of -1 is set
1120 Details : handle -1 striping on filesystem in ll_dirstripe_verify
1125 Description: ASSERTION(CheckWriteback(page,cmd)) failed
1126 Details : badly clear PG_Writeback bit in ll_ap_completion can produce false
1129 Severity : enhancement
1131 Description: Update to RHEL5 kernel-2.6.18-53.1.21.el5.
1135 Description: do not process already freed flock
1136 Details : flock can possibly be freed by another thread before it reaches
1137 to ldlm_flock_completion_ast.
1141 Description: LBUG during stress test
1142 Details : Need properly lock accesses the flock deadlock detection list.
1146 Description: oops in page fault handler
1147 Details : kernel page fault handler can return two special 'pages' in error
1148 case, don't try dereference NOPAGE_SIGBUS and NOPAGE_OMM.
1152 Description: timeout with invalidate import.
1153 Details : ptlrpcd_check call obd_zombie_impexp_cull and wait request which
1154 should be handled by ptlrpcd. This produce long age waiting and
1155 -ETIMEOUT ptlrpc_invalidate_import and as result LASSERT.
1157 Severity : enhancement
1159 Description: Update to RHEL5 kernel-2.6.18-53.1.19.el5.
1163 Description: enable MGS and MDT services start separately
1164 Details : add a 'nomgs' option in mount.lustre to enable start a MDT with
1165 a co-located MGS without starting the MGS, which is a complement
1166 to 'nosvc' mount option.
1170 Frequency : after recovery
1171 Description: precreate to many object's after del orphan.
1172 Details : del orphan st in oscc last_id == next_id and this triger growing
1173 count of precreated objects. Set flag LOW to skip increase count
1174 of precreated objects.
1178 Frequency : rare, on clear nid stats
1179 Description: ASSERTION(client_stat->nid_exp_ref_count == 0)
1180 Details : when clean nid stats sometimes try destroy live entry,
1181 and this produce panic in free.
1185 Description: Stack overflow during MDS log replay
1186 ease stack pressure by using a thread dealing llog_process.
1190 Description: wait until IO finished before start new when do lock cancel.
1191 Details : VM protocol want old IO finished before start new, in this case
1192 need wait until PG_writeback is cleared until check dirty flag and
1193 call writepages in lock cancel callback.
1195 Severity : enhancement
1197 Description: using special macro for print time and cleanup in includes.
1201 Description: mds_mfd_close() ASSERTION(rc == 0)
1202 Details : In mds_mfd_close(), we need protect inode's writecount change
1203 within its orphan write semaphore to prevent possible races.
1207 Description: Obsolete CURRENT_SECONDS and use cfs_time_current_sec() instead.
1211 Frequency : rare, on shutdown ost
1212 Description: don't hit live lock with umount ost.
1213 Details : shrink_dcache_parent can be in long loop with destroy dentries,
1214 use shrink_dcache_sb instead.
1218 Description: don't panic with use echo client
1219 Details : echo client pass NULL as client nid pointer and this produce null
1220 pointer dereference.
1224 Description: fix build on ppc32
1225 Details : compile code with -m64 flag produce wrong object file for ppc32.
1229 Description: add message levels for liblustreapi
1233 Description: fix for occasional failure case of -ENOSPC in recovery-small tests
1234 Details : Move the 'good_osts' check before the 'total_bavail' check. This
1235 will result in an -EAGAIN and in the exit call path we call
1236 alloc_rr() which will with increasing aggressiveness attempt to
1237 aquire precreated objects on the minimum number of required OSCs.
1241 Description: Use old size assignment to avoid deadlock
1242 Details : Reverts the changes in bugs 2369 and bug 14138 that introduced
1243 the scheduling while holding a spinlock. We do not need locking
1244 for size in ll_update_inode() because size is only updated from
1245 the MDS for directories or files without objects, so there is no
1246 other place to do the update, and concurrent access to such inodes
1247 are protected by the inode lock.
1251 Description: resolve "_IOWR redefined" build error on SLES10
1255 Description: dump the memory debugging after all modules are unloaded to
1256 suppress false negative in conf_sanity test 39
1258 Severity : enhancement
1260 Description: build kernel-ib packages for OFED 1.3 in our release cycle
1265 Description: fix SLES kernel versioning
1266 Details : the kernel version for our SLES 10 kernel did not include a "-"
1267 before the "smp" at the end. while this was not a problem in
1268 general, it did mean that software trying to use the kernel
1269 version to try to detect a vendor specific kernel would fail.
1270 this was most evident by the OFED build scripts.
1274 Description: Don't update lov_desc members until making sure they are valid
1275 Details : When updating lov_desc members via proc fs, need fix their
1276 validities before doing the real update.
1280 Description: don't put request into delay list while invalidate in flight.
1281 Details : ptlrpc_delay_request sometimes put in delay list while invalidate
1282 import in flight. this produce timeout for invalidate and sometimes
1283 can cause stale data.
1287 Frequency : on ppc only
1288 Description: not convert ost objects for directory because it's not exist.
1289 Details : ll_dir_getstripe assume dirrectory has ost objects but this wrong.
1293 Description: Add FMODE_EXEC file flag for SLES10 SP1 kernel.
1295 Severity : enhancement
1297 Description: Update to support 2.6.22.14 vanilla kernel.
1301 Frequency : rare, on recovery
1302 Description: read procfs can produce deadlock in some situation
1303 Details : Holding lprocfs lock which send rpc can produce block for destroy
1304 obd objects and this also block reconnect with -EALREADY. This
1305 isn't fix all lprocfs bugs - but make it rare.
1307 Severity : enhancement
1309 Description: Update kernel to RHEL5 2.6.18-53.1.14.el5.
1312 Frequency : frequent on X2 node
1314 Description: mdc_set_open_replay_data LBUG
1315 Details : Set replay data for requests that are eligible for replay.
1319 Description: lustre_mgs: operation 101 on unconnected MGS
1320 Details : When MGC is disconnected from MGS long enough, MGS will evict the
1321 MGC, and late on MGC cannot successfully connect to MGS and a lot
1322 of the error messages complaining that MGS is not connected.
1326 Frequency : on network error
1327 Description: panic with double free request if network error
1328 Details : mdc_finish_enqueue is finish request if any network error ocuring,
1329 but it's true only for synchronus enqueue, for async enqueue
1330 (via ptlrpcd) this incorrect and ptlrpcd want finish request
1333 Severity : enhancement
1335 Description: client-side metadata stat-ahead during readdir(directory readahead)
1336 Details : perform client-side metadata stat-ahead when the client detects
1337 readdir and sequential stat of dir entries therein
1340 Frequency : on start mds
1342 Description: Implement get_info(last_id) in obdfilter.
1345 Frequency : occasional
1347 Description: Correctly check stale fid, not start epoch if ost not support SOM
1348 Details : open with flag O_CREATE need set old fid in op_fid3 because
1349 op_fid2 was overwritten with new generated fid, but mds can answer
1350 with one of these two fids and both is not stale. Setattr
1351 incorrectly started an epoch and assume will be called
1352 done_writing, but without SOM done_writing ever being called.
1355 Frequency : rare, depends on device drivers and load
1357 Description: MDS or OSS nodes crash due to stack overflow
1358 Details : Code changes in 1.8.0 increased the stack usage of some functions.
1359 In some cases, in conjunction with device drivers that use a lot
1360 of stack the MDS (or possibly OSS) service threads could overflow
1361 the stack. One change which was identified to consume additional
1362 stack has been reworked to avoid the extra stack usage.
1365 Frequency : occasional
1367 Description: Do not fail import if osc_interpret_create gets -EAGAIN
1368 Details : If osc_interpret_create got -EAGAIN it immediately exits and
1369 wakeup oscc_waitq. After wakeup oscc_wait_for_objects call
1370 oscc_has_objects and see OSC has no objests and call
1371 oscc_internal_create to resend create request.
1373 Severity : enhancement
1375 Description: Update to SLES10 SP1 latest kernel-2.6.16.54-0.2.5.
1377 Severity : enhancement
1379 Description: Update to RHEL5 latest kernel-2.6.18-53.1.13.el5.
1382 Frequency : very rare
1384 Description: Fix replay if there is an un-replied request and open
1385 Details : In some cases, older replay request will revert the
1386 mcd->mcd_last_xid on MDS which is used to record the client's
1387 latest sent request.
1389 Severity : enhancement
1391 Description: Update to RHEL5 latest kernel-2.6.18-53.1.6.el5.
1393 Severity : enhancement
1395 Description: Add rhel5 support to HEAD.
1397 Serverity : enhancement
1399 Description: Update RHEL4 kernel to 2.6.9-67.0.4.
1404 Description: Don't allow skipping OSTs if index has been specified.
1405 Details : Don't allow skipping OSTs if index has been specified, make
1406 locking in internal create lots better.
1410 Description: LBUG in ptlrpc_check_set() bad phase ebc0de00
1411 Details : access to bitfield in structure is always rounded to long
1412 and this produce problem with not atomic change any bit.
1416 Description: Lustre make rpms failed.
1417 Details : Remove ldiskfs spec file to avoids rpmbuild be confused when
1418 builds Lustre rpms from tarball.
1421 Frequency : rare on shutdown ost
1423 Description: If llog cancel was not send before clean_exports phase, this can
1424 produce deadlock in llog code.
1425 Details : If llog thread has last reference to obd and call class_import_put
1426 this produce deadlock because llog_cleanup_commit_master wait when
1427 last llog_commit_thread exited, but this never success because was
1428 called from llog_commit_thread.
1432 Description: allow userland application know is lost one of stripes.
1433 Details : fill lvb_blocks with error code on ost and return it to
1434 application if error flag found.
1438 Description: NULL lov_tgts causing MDS oops
1439 Details : more safe checks for NULL lov_tgts for avoid oops.
1441 Severity : enhancement
1443 Description: Update to RHEL4 latest kernel-2.6.9-67.0.1.EL.
1447 Descriptoin: make lov_create() will not stuck in obd_statfs_rqset()
1448 Details : If an OST is down the MDS will hang indefinitely in
1449 obd_statfs_rqset() waiting for the statfs data. While for
1450 MDS QOS usage of statfs, it should not stuck in waiting.
1452 Severity : enhancement
1454 Description: remote_acl support
1455 Details : Support ACL-based permission check for remote user.
1456 Support setfacl/getfacl for remote user with the utils
1457 "lfs {l,r}{s,g}etfacl" which follow the same parameter format as
1458 the system "{s,g}etfacl" utils.
1460 Severity : enhancement
1462 Description: Update to RHEL4 U6 kernel-2.6.9-67.EL.
1464 Severity : enhancement
1466 Description: Update to RHEL5 latest kernel-2.6.18-53.1.4.el5.
1470 Description: make mgs_setparam() handle fsname containing dash
1471 Details : fsname containing a dash does not work with lctl conf_param
1473 Severity : enhancement
1475 Description: Update to SLES10 SP1 latest kernel-2.6.16.54-0.2.3.
1477 Severity : enhancement
1479 Description: Update to RHEL5 Update-1 kernel 2.6.18-53.el5.
1483 Frequency : rare, at shutdown
1484 Description: access already free / zero obd_namespace.
1485 Details : if client_disconnect_export was called without force flag set,
1486 and exist connect request in flight, this can produce access to
1487 NULL pointer (or already free pointer) when connect_interpret
1488 store ocd flags in obd_namespace.
1492 Frequency : only at startup
1493 Description: not alloc memory with spinlock held.
1494 Details : allocation memory with GFP_KERNEL can produce sleep deadlock,
1495 if any spinlock held.
1497 Severity : enhancement
1499 Description: make lustre randomly fail allocating memory
1500 Details : Make lustre randomly failed allocating memory for testing purpose.
1502 Severity : enhancement
1504 Description: lost problems with lov objid file
1505 Details : Fixes some scability and access to not inited memory problems
1506 in work with lov objdid file.
1511 Description: lfs find does not continue on file error
1512 Details : Continue other files processing when a file/dir is absent.
1516 Description: Inconsistent usage of lustre_pack_reply()
1517 Details : Standardize the usage of lustre_pack_reply() such that it
1518 always generate a CERROR on failure.
1521 Frequency : occasional
1523 Description: MDS hang or stay in waiting lock
1524 Details : If client receive lock with CBPENDING flag ldlm need send lock
1525 cancel as separate rpc, to avoid situation when cancel request
1526 can't processed due all i/o threads stay in wait lock.
1530 Description: Update to RHEL5 kernel 2.6.18-8.1.15.el5.
1534 Description: Update to SLES10 SP1 kernel 2.6.16.53-0.16
1538 Description: Update to SLES9 kernel-2.6.5-7.287.3.
1542 Description: Update to RHEL4 latest kernel.
1544 Severity : enhancement
1546 Description: Build SLES10 patchless client fails
1547 Details : The configure was broken by run ./configure with
1548 --with-linux-obj=.... argument for patchless client. When the
1549 configure use --with-linux-obj, the LINUXINCLUDE= -Iinclude
1550 can't search header adequately. Use absolute path such as
1551 -I($LINUX)/include instead.
1555 Description: interrupt oig_wait produce painc on resend.
1556 Details : brw_redo_request can be used for resend requests from ptlrpcd and
1557 private set, and this produce situation when rq_ptlrpcd_data not
1558 copyed to new allocated request and triggered LBUG on assert
1559 req->rq_ptlrpcd_data != NULL. But this member used only for wakeup
1560 ptlrpcd set if request is changed and can be safety changed to use
1565 Description: LASSERT_{REQ,REP}SWAB macros are buggy
1566 Details : If SWAB_PARANOIA is disabled, the LASSERT_REQSWAB and
1567 LASSERT_REPSWAB macros become no-ops, which is incorrect. Drop
1568 these macros and replace them with their difinitions instead.
1572 Description: Update kernel patches for SLES10 2.6.16.53-0.8.
1573 Details : Update which_patch & target file for SLES10 latest kernel.
1576 Description: Remove client patches from SLES 10 kernel.
1577 Details : This causes SLES 10 clients to behave as patchless clients
1578 even on a Lustre-patched (server) kernel.
1580 Severity : enhancement
1582 Description: self-adjustable client's lru lists
1583 Details : use adaptive algorithm for managing client cached locks lru
1584 lists according to current server load, other client's work
1585 pattern, memory activities, etc. Both, server and client
1586 side namespaces provide number of proc tunables for controlling
1589 Severity : enhancement
1591 Description: light-weight GSS support
1592 Details : Support krb5n and krb5a mode, which keep Kerberos 5 authentication
1593 and reduce performance overhead.
1595 Severity : enhancement
1597 Description: Linux keyring support
1598 Details : Support using service of Linux keyring for Lustre GSS internal
1599 context refresh/cache mechanism.
1603 Description: Fix errors in lfs documentation
1604 Details : Fixes man pages
1608 Description: don't use GFP_* in generic Lustre code.
1609 Details : Use cfs_alloc_* functions and CFS_* flags for code portability.
1613 Description: obdclass is limited by single OBD_ALLOC(idarray)
1614 Details : replace OBD_ALLOC/OBD_FREE with OBD_VMALLOC/OBD_VFREE
1618 Description: warnings with build patchless client with vanila 2.6.19 and up
1619 Details : change old ctl_table style and replace ctl_table/ctl_table_header
1620 with cfs_sysctl_table_t/cfs_sysctl_table_header_t
1624 Frequency : Only for SLES
1625 Description: sanity_quota fail test_1
1626 Details : There are multiple occurences of $TSTUSR in SLES's /etc/group
1627 file, which makes TSTID[2] inunique.
1631 Frequency : Only for SLES9
1632 Description: Kernel patches for SLES9 2.6.5-7.286 kernel
1633 Details : Update target/ChangeLog/which_patch .
1637 Frequency : Only for test_10 in sanity-quota.sh
1638 Description: a bug in quota test code
1639 Details : checking a test flag in a wrong place
1643 Frequency : Only for test_13 in sanity-quota.sh
1644 Description: a bug in quota test script
1645 Details : shouldn't check overfull allocated quota
1649 Description: jbd statistics
1650 Details : Port older jbd statistics patch for sles10
1652 Severity : enhancement
1654 Description: Add printing inode info into message about error in writepage.
1659 Description: reply_lock_interpret crash due to race with it and lock cancel.
1660 Details : Do not replay locks that are being cancelled. Do not reference
1661 locks by their address during replay, just by their handle.
1665 Frequency : When flocks are used.
1666 Description: assertion failure in ldlm_cli_enquque_fini for non NULL lock.
1667 Details : Flock locks might destroy just granted lock if it could be merged
1668 with another existing flock, this is done in completion handler,
1669 so teach ldlm_cli_enquque_fini that this is a valid case for
1675 Description: Oops in read and write path when failing to allocate lock.
1676 Details : Check if lock allocation failed and return error back.
1680 Description: lstripe command fails for valid OST index
1681 Details : The stripe offset is compared to lov->desc.ld_tgt_count
1682 instead of lov->desc.ld_active_tgt_count.
1686 Description: sanity.sh failed test 103
1687 Details : RHEL mis-interpret setfacl "-X" param, so we won't test setfacl
1692 Description: df doesn't work properly if diskfs blocksize != 4K
1693 Details : Choose biggest blocksize of OST's as the LOV's blocksize.
1697 Description: lfs find on -1 stripe looping in lsm_lmm_verify_common()
1698 Details : Avoid lov_verify_lmm_common() on directory with -1 stripe count.
1700 Severity : enhancement
1702 Description: Adaptive timeouts
1703 Details : RPC timeouts adapt to changing server load and network
1704 conditions to reduce resend attempts and improve recovery time.
1708 Description: llapi_file_create() does not allow some changes
1709 Details : add llapi_file_open() that allows specifying the mode and
1710 open flags, and also returns an open file handle.
1714 Description: merge and cleanup kernel patches.
1715 Details : Remove mnt_lustre_list in vfs_intent-2.6-rhel4.patch.
1719 Description: Add journal checksum support.(Kernel part)
1720 Details : The journal checksum feature adds two new flags i.e
1721 JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT and
1722 JBD2_FEATURE_COMPAT_CHECKSUM. JBD2_FEATURE_CHECKSUM flag
1723 indicates that the commit block contains the checksum for
1724 the blocks described by the descriptor blocks. Now commit
1725 record can be sent to disk without waiting for descriptor
1726 blocks to be written to disk. This behavior is controlled
1727 using JBD2_FEATURE_ASYNC_COMMIT flag.
1731 Description: OSS needs mutliple precreate threads
1732 Details : Add ability to start more than one create thread per OSS.
1736 Description: Sanity.sh test_65k failed: setstripe should have succeeded
1737 Details : execute lfs setstripe on client
1741 Description: mds_obd_create error creating tmp object
1742 Details : When the user sets quota on root, llog will be affected and can't
1743 create files and write files.
1746 Frequency : Always on ia64 patchless client, and possibly others.
1748 Description: Add EXPORT_SYMBOL check for node_to_cpumask symbol.
1749 Details : This allows the patchless client to be loaded on architectures
1750 without this export.
1754 Description: RedHat Update kernel for RHEL5
1755 Details : Add the kernel config file for RHEL5.
1759 Description: RedHat Update kernel for RHEL5
1760 Details : Modify the kernel config file more closer RHEL5.
1764 Description: Build failure against Centos5 (RHEL5)
1765 Details : Define PAGE_SIZE when it isn't present.
1769 Description: test_7 on sanity-quota.sh can't be used on separate machines
1770 Details : add facet to handle it
1774 Description: "ll_intent_file_open()) lock enqueue: err: -13" with nfs
1775 Details : with NFS, the anon dentry's parent was set to itself in
1776 d_alloc_anon(), so in MDS, we use rec->ur_fid1 to find the
1777 corresponding dentry other than use rec->ur_name.
1779 Severity : enhancement
1781 Description: lfs setstripe enhancement
1782 Details : Make lfs setstripe understand 'k', 'm' and 'g' for stripe size.
1786 Description: enable data checksumming by default
1787 Details : enable checksum by default, allow --disable-checksum
1788 configure option and "-o nochecksum" mount option. Checksums
1789 can also be disabled at runtime via $LPROC/osc/*/checksum_pages.
1793 Description: lustre support for RHEL5
1794 Details : Add support for RHEL5.
1798 Description: Client eviction due to failover config
1799 Details : after a connection loss, the lustre client should attempt to
1800 reconnect to the last active server first before trying the
1801 other potential connections.
1805 Description: when mds and osts use different quota unit(32bit and 64bit),
1806 quota will be released repeatly.
1807 Details : void sending multiple quota reqs to mds, which will keep the
1808 status between the reqs.
1812 Description: osts not allocated evenly to files
1813 Details : change the condition to increase offset_idx
1816 Frequency : Always for filesystems larger than 2TB on 32-bit systems.
1818 Description: Data corruption for OSTs that are formatted larger than 2TB
1820 Details : When generating the bio request for lustre file writes the
1821 sector number would overflow a temporary variable before being
1822 used for the IO. The data reads correctly from Lustre (which
1823 will overflow in a similar manner) but other file data or
1824 filesystem metadata may be corrupted in some cases.
1828 Description: Tune the kernel for good SCSI performance.
1829 Details : Set the value of /sys/block/{dev}/queue/max_sectors_kb
1830 to the value of /sys/block/{dev}/queue/max_hw_sectors_kb
1835 Description: rewrite ext2-derived code in obdclass/uuid.c
1836 Details : rewrite inherited code (uuid parsing code from ext2 utils)
1837 from scratch preserving functionality.
1841 Description: "lfs find -obd UUID" prints directories
1842 Details : "lfs find -obd UUID" will return all directory names instead
1843 of just file names. It is incorrect because the directories
1844 do not reside on the OSTs.
1848 Description: use i_size_read and i_size_write in 2.6 port
1849 Details : replace inode->i_size access with i_size_read/write()
1852 Frequency : when removing large files
1854 Description: scheduling issue during removal of large Lustre files
1855 Details : Don't take the BKL in fsfilt_ext3_setattr() for 2.6 kernels.
1856 It causes scheduling issues when removing large files (17TB in the
1860 Frequency : only with liblustre clients on XT3
1862 Description: evictions taking too long
1863 Details : allow llrd to evict clients directly on OSTs
1866 Frequency : only on ppc
1868 Description: /proc/fs/lustre/devices broken on ppc
1869 Details : The patch as applied to 1.6.2 doesn't look correct for all arches.
1870 We should make sure the type of 'index' is loff_t and then cast
1871 explicitly as needed below. Do not assign an explicitly cast
1875 Frequency : only for rhel5
1877 Description: Kernel patches update for RHEL5 2.6.18-8.1.10.el5.
1878 Details : Modify the target file & which_kernel.
1880 Severity : enhancement
1882 Description: omit set fsid for export NFS
1883 Details : fix set/restore device id for avoid EMFILE error and mark lustre
1884 fs as FS_REQUIRES_DEV for avoid problems with generate fsid.
1888 Frequency : Always, for kernels after 2.6.16
1889 Description: Fix warning idr_remove called for id=.. which is not allocated.
1890 Details : Last kernels save old s_dev before kill super and not allow
1891 to restore from callback - restore it before call kill_anon_super.
1895 Description: buffer overruns could theoretically occur
1896 Details : llapi_semantic_traverse() modifies the "path" argument by
1897 appending values to the end of the origin string, and a
1898 overrun may occur. Adding buffer overrun check in liblustreapi.
1902 Description: Fix error on 'ls .' at the top of the Lustre mount.
1903 Details : Don't revalidate dentry if it is a root dentry.
1907 Description: Kernel patches update for RHEL4 2.6.9-55.0.6.
1908 Details : Modify vm-tunables-rhel4.patch.
1912 Description: Kernel config for 2.6.18-vanilla.
1913 Details : Modify targets/2.6-vanilla.target.in.
1914 Add config file kernel-2.6.18-2.6-vanilla-i686.config.
1915 Add config file kernel-2.6.18-2.6-vanilla-i686-smp.config.
1916 Add config file kernel-2.6.18-2.6-vanilla-x86_64.config.
1917 Add config file kernel-2.6.18-2.6-vanilla-x86_64-smp.config.
1921 Description: improve handling recoverable errors
1922 Details : if request processig with error which can be recoverable on server
1923 request should be resend, otherwise page released from cache and
1928 Description: Kernel patches update for RHEL5 2.6.18-8.1.14.el5.
1929 Details : Modify target file & which_patch.
1930 A flaw was found in the IA32 system call emulation provided
1931 on AMD64 and Intel 64 platforms. An improperly validated 64-bit
1932 value could be stored in the %RAX register, which could trigger an
1933 out-of-bounds system call table access. An untrusted local user
1934 could exploit this flaw to run code in the kernel
1935 (ie a root privilege escalation). (CVE-2007-4573).
1939 Description: O_DIRECT bypasses client statistics.
1940 Details : When running with O_DIRECT I/O, neither the client rpc_stats nor
1941 read_ahead_stats were updated. copied the stats section from
1942 osc_send_oap_rpc() into async_internal().
1946 Description: Add jbd statistics patch for RHEL5 and 2.6.18-vanilla
1950 Description: change order of libsysio includes
1951 Details : '#include sysio.h' should always come before '#include xtio.h'
1955 Description: handle "serious error: objid * already exists" more gracefully
1956 Details : If LAST_ID value on disk is smaller than the objects existing in
1957 the O/0/d* directories, it indicates disk corruption and causes an
1958 LBUG(). If the object is 0-length, then we should use the existing
1959 object. This will help to avoid a full fsck in most cases.
1961 Severity : enhancement
1963 Description: adapt the lustre_config script to support the upgrade case
1964 Details : Add "-u" option for lustre_config script to support upgrading 1.4
1965 server targets to 1.6 in parallel.
1969 Description: To avoid grant space > avaible space when the disk is almost
1970 full. Without this patch you might see the error "grant XXXX >
1971 available" or some LBUG about grant, when the disk is almost
1973 Details : In filter_check_grant, for non_grant cache write, we should
1974 check the left space by if (*left > ungranted + bytes), instead
1975 of (*left > ungranted), because only we are sure the left space
1976 is enough for another "bytes", then the ungrant space should be
1977 increase. In client, we should update cl_avail_grant only there
1978 is OBD_MD_FLGRANT in the reply.
1982 Description: Update RHEL 4 kernel to fix local root privilege escalation.
1983 Details : Update to the latest RHEL 4 kernel to fix the vulnerability
1984 described in CVE-2007-4573. This problem could allow untrusted
1985 local users to gain root access.
1988 Frequency : when using O_DIRECT and quotas
1990 Description: Incorrect file ownership on O_DIRECT output files
1991 Details : block usage reported by 'lfs quota' does not take into account
1992 files that have been written with O_DIRECT.
1997 Description: touch file failed when fs is not full
1998 Details : OST in recovery should not be discarded by MDS in alloc_qos(),
1999 otherwise we can get ENOSP while fs is not full.
2003 Description: parallel lock callbacks
2004 Details : Instead of sending blocking and completion callbacks as separated
2005 requests, adding them to a set and sending in parallel.
2008 Frequency : only for Cray XT3
2009 Bugzilla : 12829/13455
2010 Description: Changing primary group doesn't change the group lustre assigns to
2012 Details : When CRAY_XT3 is defined, the fsgid supplied by the client is
2013 overridden with the primary group provided by the group upcall,
2014 whereas the supplied fsgid can be trusted if it is in the list of
2015 supplementary groups returned by the group upcall.
2017 Severity : enhancement
2019 Description: Allow masking D_WARNING, D_ERROR messages from console
2020 Details : Console messages can now be disabled via lnet.printk.
2024 Description: User code with malformed file open parameter crashes client node
2025 Details : Before packing join_file req, all the related reference should be
2026 checked carefully in case some malformed flags cause fake
2027 join_file req on client.
2031 Description: LDLM_ENQUEUE races with LDLM_CP_CALLBACK
2032 Details : ldlm_completion_ast() assumes that a lock is granted when the req
2033 mode is equal to the granted mode. However, it should also check
2034 that LDLM_FL_CP_REQD is not set.
2038 Description: Heavy nfs access might result in deadlocks
2039 Details : After ELC code landed, it is now improper to enqueue any mds
2040 locks under och_sem, because enqueue might want to decide to
2041 cancel open locks for same inode we are holding och_sem for.
2045 Description: Client eviction while running blogbench
2046 Details : A lot of unlink operations with concurrent I/O can lead to a
2047 deadlock causing evictions. To address the problem, the number of
2048 oustanding OST_DESTROY requests is now throttled to
2049 max_rpcs_in_flight per OSC and LDLM_FL_DISCARD_DATA blocking
2050 callbacks are processed in priority.
2054 Description: enable ACLs on MDS by default
2055 Details : ACLs must be enabled on MDS by default.
2058 Frequency : PPC/PPC64 only
2060 Description: conflicts between asm-ppc64/types.h and lustre_types.h
2061 Details : fix duplicated definitions between asm-ppc64/types.h and
2062 lustre_types.h on PPC.
2065 Frequency : PPC/PPC64 only
2067 Description: asm-ppc/segment.h does not exist
2068 Details : fix compile issue on PPC.
2072 Description: better handle error messages in extents code
2075 Frequency : RHEL4 only
2077 Description: mkfs is very slow on IA64/RHEL4
2078 Details : A performance regression has been discovered in the MPT Fusion
2079 driver between versions 3.02.73rh and 3.02.99.00rh. As a
2080 consequence, we have downgraded the MPT Fusion driver in the RHEL4
2081 kernel from 3.02.99.00 to 3.02.73 until this problem is fixed.
2083 Severity : enhancement
2085 Description: SNMP support enhancement
2086 Details : Adding total number of sampled request for an MDS node in snmp
2089 Severity : enhancement
2091 Description: Optimize ldlm waiting list processing for PR extent locks
2092 Details : When processing waiting list for read extent lock and meeting read
2093 lock that is same or wider to it that is not contended, skip
2094 processing rest of the list and immediatelly return current
2095 status of conflictness, since we are guaranteed there are no
2096 conflicting locks in the rest of the list.
2100 Description: Time out and refuse to reconnect
2101 Details : When the failover node is the primary node, it is possible
2102 to have two identical connections in imp_conn_list. We must
2103 compare not conn's pointers but NIDs, otherwise we can defeat
2104 connection throttling.
2108 Description: port llog fixes from b1_6 into HEAD
2109 Details : Port llog reference couting and some llog cleanups from b1_6
2110 (bug 10800) into HEAD, for protect from panic and access to already
2111 free llog structures.
2115 Description: Detect stride IO mode in read-ahead
2116 Details : When a client does stride read, read-ahead should detect that and
2117 read-ahead pages according to the detected stride pattern.
2121 Description: data checksumming impacts single node performance
2122 Details : add support for several checksum algorithm. Currently, only CRC32
2123 and Adler-32 are supported. The checksum type can be changed on
2124 the fly via /proc/fs/lustre/osc/*/checksum_type.
2128 Description: use adler32 for page checksums
2129 Details : when available, use the Adler-32 algorithm instead of CRC32 for
2134 Description: build for x2 fails
2135 Details : fix compile issue on Cray systems.
2139 Description: Properly match for duplicate locks
2140 Details : Due to different lock order from skiplists code, we need to
2141 traverse entire list for now
2144 Frequency : only on PPC/SLES10
2146 Description: "BITS_PER_LONG is not 32 or 64" in linux/idr.h
2147 Details : On SLES10/PPC, fs.h includes idr.h which requires BITS_PER_LONG to
2148 be defined. Add a hack in mkfs_lustre.c to work around this compile
2153 Description: LASSERT on MDS when client holding flock lock dies
2154 Details : ldlm pool logic depends on number of granted locks equal to
2155 number of released locks which is not true for flock locks, so
2156 just exclude such locks from consideration.
2160 Description: MDS deadlock with many ll_sync_lov threads and I/O stalled
2161 Details : Use fsfilt_sync() for both the whole filesystem sync and
2162 individual file sync to eliminate dangerous inode locking
2163 with I_LOCK that can lead to a deadlock.
2167 Description: performance in 1.6.3
2168 Details : Force q->max_phys_segments to MAX_PHYS_SEGMENTS on SLES10 to be
2169 sure that 1MB requests are not fragmented by the block layer.
2171 Severity : enhancement
2173 Description: organize the server-side client stats on per-nid basis
2174 Details : Change the structure of stats under obdfilter and mds to
2182 The "uuid"s file would list the uuids of _active_ exports.
2183 And the clear entry is to clear all stats and stale nids.
2185 Severity : enhancement
2187 Description: eliminate client locks in face of contention
2188 Details : file contention detection and lockless i/o implementation
2189 for contended files.
2193 Description: Reinitialize optind to 0 so that interactive lfs works in all cases
2196 Frequency : very rare, if additional xattrs are used on kernels >= 2.6.12
2198 Description: MDS may lose file striping (and hence file data) in some cases
2199 Details : If there are additional extended attributes stored on the MDS,
2200 in particular ACLs, SELinux, or user attributes (if user_xattr
2201 is specified for the client mount options) then there is a risk
2202 of attribute loss. Additionally, the Lustre file striping
2203 needs to be larger than default (e.g. striped over all OSTs),
2204 and an additional attribute must be stored initially in the
2205 inode and then increase in size enough to be moved to the
2206 external attribute block (e.g. ACL growing in size) for file
2211 Description: skiplist implementation simplification
2212 Details : skiplists are used to group compatible locks on granted list
2213 that was implemented as tracking first and last lock of each
2214 lock group the patch changes that to using doubly linked lists
2218 Description: MDS LBUG: ASSERTION(!IS_ERR(dchild))
2219 Details : Change LASSERTs to client eviction (i.e. abort client's recovery)
2220 because LASSERT on both the data supplied by a client, and the
2221 data on disk is dangerous and incorrect.
2223 Severity : enhancement
2225 Description: Slow truncate/writes to huge files at high offsets.
2226 Details : Directly associate cached pages to lock that protect those pages,
2227 this allows us to quickly find what pages to write and remove
2228 once lock callback is received.
2232 Description: more ldlm soft lockups
2233 Details : In ldlm_resource_add_lock(), call to ldlm_resource_dump()
2234 starve other threads from the resource lock for a long time in
2235 case of long waiting queue, so change the debug level from
2236 D_OTHER to the less frequently used D_INFO.
2238 Severity : enhancement
2240 Description: add -gid, -group, -uid, -user options to lfs find
2244 Description: Hung threads in invalidate_inode_pages2_range
2245 Details : The direct IO path doesn't call check_rpcs to submit a new RPC
2246 once one is completed. As a result, some RPCs are stuck in the
2247 queue and are never sent.
2251 Description: filter threads hungs on waiting journal commit
2252 Details : Cleanup filter group llog code, then only filter group llog will
2253 be only created in the MDS/OST syncing process.
2257 Description: Procfs and llog threads access destoryed import sometimes.
2258 Details : Sync the import destoryed process with procfs and llog threads by
2259 the import refcount and semaphore.
2261 Severity : enhancement
2263 Description: openlock cache of b1_6 port to HEAD
2268 Description: kernel BUG at ldiskfs2_ext_new_extent_cb
2269 Details : If insertion of an extent fails, then discard the inode
2270 preallocation and free data blocks else it can lead to duplicate
2275 Description: don't always update ctime in ext3_xattr_set_handle()
2276 Details : Current xattr code updates inode ctime in ext3_xattr_set_handle.
2277 In some cases the ctime should not be updated, for example for
2278 2.0->1.8 compatibility it is necessary to delete an xattr and it
2279 should not update the ctime.
2283 Bugzilla : 15713/16362
2284 Description: Assertion in iopen_connect_dentry in 1.6.3
2285 Details : looking up an inode via iopen with the wrong generation number can
2286 populate the dcache with a disconneced dentry while the inode
2287 number is in the process of being reallocated. This causes an
2288 assertion failure in iopen since the inode's dentry list contains
2289 both a connected and disconnected dentry.
2293 Description: assertion failure in ldlm_handle2lock()
2294 Details : fix a race between class_handle_unhash() and class_handle2object()
2295 introduced in lustre 1.6.5 by bug 13622.
2300 Description: Kernel BUG: sd_iostats_bump: unexpected disk index
2301 Details : remove the limit of 256 scsi disks in the sd_iostat patch
2306 Description: oops in sd_iostats_seq_show()
2307 Details : unloading/reloading the scsi low level driver triggers a kernel
2308 bug when trying to access the sd iostat file.
2313 Description: Kernel panics during QLogic driver reload
2314 Details : REQ_BLOCK_PC requests are not handled properly in the sd iostat
2315 patch, causing memory corruption.
2320 Description: journal_dev option does not work in b1_6
2321 Details : pass mount option during pre-mount.
2323 Severity : enhancement
2325 Description: Add a FIEMAP(FIle Extent MAP) ioctl
2326 Details : FIEMAP ioctl will allow an application to efficiently fetch the
2327 extent information of a file. It can be used to map logical blocks
2328 in a file to physical blocks in the block device.
2332 Description: LDLM soft lockups - improvement
2333 Details : It is be possible to send the lock handle along with each read
2334 or write request because the client is already doing a lock match
2335 itself so there isn't any reason the OST should have to re-do that
2339 Frequency : only on Cray X2
2341 Description: X2 build failures
2342 Details : fix build failures on Cray X2.
2346 Description: xid & resent requests
2347 Details : Initialize RPC XID from clock at startup (randomly if clock is
2350 Severity : enhancement
2352 Description: Add lustre_start utility to start or stop multiple Lustre servers
2357 Description: Lustre GPF in {:ptlrpc:ptlrpc_server_free_request+373}
2358 Details : In case of memory pressure, list_del() can be called twice on
2359 req->rq_history_list, causing a kernel oops.
2363 Description: kptllnd_peer_check_sends()) ASSERTION(!in_interrupt()) failed
2364 Details : fix stack overflow in the distributed lock manager by defering
2365 export eviction after a failed AST to the elt thread instead of
2366 handling it in the dlm interpret routine.
2370 Description: Convert some comments to new format.
2371 Details : Update documenting comments to match doxygen conventions.
2375 Description: Grammar fixes.
2376 Details : A couple of trivial spelling fixes.
2380 Description: OSD_COUNTERS-mandatory
2381 Details : Make previously optional ->oti_{w,r}_locks sanity checks mandatory
2382 to simplify the code.
2386 Description: simplify cmm_device freeing logic.
2387 Details : Call cmm_device_free() in the failure path of cmm_device_alloc().
2391 Description: Add lockdep support to dt_object_operations locking interface.
2392 Details : Augment ->do_{read,write}_lock() prototypes with a `role'
2393 parameter indicating lock ordering. Update mdd code to use new
2398 Description: Introduce failloc constants for lockless IO tests.
2399 Details : Add two new failloc constants to test lockless IO. Only one of
2400 them in implemented---another is checked in yet to be landed
2405 Description: Add lockdep support for inode mutex.
2406 Details : Introduce and use new LOCK_INODE_MUTEX_PARENT() macro to be used
2407 in the situations where Lustre has to lock more than one inode
2412 Description: Add optional invariants checking support.
2413 Details : Add new LINVRNT() macro, optional on new --enable-invariants
2414 configure switch. This macro is to be used for consistency and
2415 sanity checks that are too expensive to be left in `production'
2420 Description: Zap lock->l_granted_mode with explicit LCK_MINMODE.
2421 Details : Use LCK_MINMODE rather than 0 to reset lock->l_granted_mode to
2426 Description: Add lockdep support for ldlm_lock and ldlm_resource.
2427 Details : Use spin_lock_nested() in (the only) situation where more than
2428 one ldlm_lock is locked simultaneously. Also, fix possible
2429 dead-lock in ldlm_lock_change_resource() by enforcing particular
2434 Description: Use struct ldlm_callback_suite in ldlm_lock_create().
2435 Details : Instead of specifying each ldlm_lock call-back through separate
2436 parameter, wrap them into struct ldlm_callback_suite.
2440 Description: Kill join_lru obd method and its callers.
2441 Details : CLIO uses lock weighting policy to keep locks over mmapped regions
2442 in memory---a requirement implemented through ->o_join_lru() obd
2443 method in HEAD. Remove this method and its users.
2447 Description: Add asynchronous ldlm ENQUEUE completion handler.
2448 Details : CLIO posts ENQUEUE requests asynchronously through ptlrpcd---a
2449 case that stock ldlm_completion_ast() cannot handle as it waits
2450 until lock is granted. Introduce new ldlm_completion_ast_async()
2451 for this. Also comment ldlm_completion_ast().
2455 Description: ldlm_error <-> errno conversion.
2456 Details : Add functions to map (rather arbitrary) between LDLM error codes
2457 and standard errno values. CLIO needs this to prevent LDLM specific
2458 constants from escaping ldlm and osc.
2462 Description: Kill unused ldlm_handle2lock_ns() function.
2463 Details : Kill unused ldlm_handle2lock_ns() function.
2467 Description: Add lu_ref support to ldlm_lock
2468 Details : lu_ref support for ldlm_lock and ldlm_resource. See lu_ref patch.
2469 lu_ref fields ->l_reference and ->lr_reference are added to ldlm_lock
2470 and ldlm_resource. LDLM interface has to be changed, because code that
2471 releases a reference on a lock, has to "know" what reference this is.
2472 In the most frequent case
2474 lock = ldlm_handle2lock(handle);
2476 LDLM_LOCK_PUT(lock);
2478 no changes are required. When any other reference (received _not_ from
2479 ldlm_handle2lock()) is released, LDLM_LOCK_RELEASE() has to be called
2480 instead of LDLM_LOCK_PUT().
2482 Arguably, changes are pervasive, and interface requires some discipline
2483 for proper use. On the other hand, it was very instrumental in finding
2484 a few leaked lock references.
2488 Description: Add ldlm_lock_addref_try().
2489 Details : Introduce ldlm_lock_addref_try() function (used by CLIO) that
2490 attempts to addref a lock that might be being canceled
2495 Description: Add ldlm_weigh_callback().
2496 Details : Add new ->l_weigh_ast() call-back to ldlm_lock. It is called
2497 by ldlm_cancel_shrink_policy() to estimate lock "value", instead of
2498 hard-coded `number of pages' logic.
2502 Description: Add lockdep annotations to llog code.
2503 Details : Use appropriately tagged _nested() locking calls in the places
2504 where llog takes more than one ->lgh_lock lock.
2508 Description: Add loi_kms_set().
2509 Details : Wrap kms updates into a helper function.
2513 Description: Constify instances of struct lsm_operations.
2514 Details : Constify instances of struct lsm_operations.
2518 Description: lu_conf support.
2519 Details : On a server, a file system object is uniquely identified
2520 by a fid, which is sufficient to locate and load all object
2521 state (inode). On a client, on the other hand, more data are
2522 necessary instantiate an object. Change lu_object_find() and
2523 friends to take additional `lu_conf' argument describing object.
2524 Typically this includes layout information.
2528 Description: lu_context fixes.
2529 Details : Introduce new lu_context functions that are needed on the client
2530 side, where some system threads (ptlrpcd) are shared by multiple
2531 modules, and so cannot be stopped during module shutdown.
2535 Description: Add start and stop methods to lu_device_type_operations.
2536 Details : Introduce two new methods in lu_device_type_operations, that are
2537 invoked when first instance of a given type is created and last one
2538 is destroyed respectively. This is need by CLIO.
2542 Description: Add lu_ref support to struct lu_device.
2543 Details : Add lu_ref support to lu_object and lu_device. lu_ref is used to
2544 track leaked references.
2548 Description: Introduce lu_kmem_descr.
2549 Details : lu_kmem_descr and its companion interface allow to create
2550 and destroy a number of kmem caches at once.
2554 Description: Fix lu_object finalization race.
2555 Details : Fix a race between lu_object_find() finding an object and its
2556 concurrent finalization. This race is (most likely) not possible
2557 on the server, but might happen on the client.
2561 Description: Introduce lu_ref interface.
2562 Details : lu_ref is a debugging module allowing to track references to
2563 a given object. It is quite cpu expensive, and has to be
2564 explicitly enabled with --enable-lu_ref. See usage description
2569 Description: Factor lu_site procfs stats into a separate function.
2570 Details : Separate lu_site stats printing code into a separate function
2571 that can be reused on a client.
2575 Description: Constify instances of struct {lu,dt,md}_device_operations.
2576 Details : Constify instances of struct {lu,dt,md}_device_operations.
2580 Description: Introduce struct md_site and move meta-data specific parts of
2581 struct lu_site here.
2582 Details : Move md-specific fields out of struct lu_site into special struct
2583 md_site, so that lu_site can be used on a client.
2587 Description: Kill mdd_lov_destroy().
2588 Details : Remove unused mdd code.
2592 Description: Add st_block checking to multistat.c.
2593 Details : Add st_block checking to multistat.c.
2597 Description: Add lu_ref support to struct obd_device.
2598 Details : Add lu_ref tracking to obd_device.
2602 Description: Kill obd_set_fail_loc().
2603 Details : Remove unused code.
2607 Description: Add special type for ptlrpc_request interpret functions.
2608 Details : Add lu_env parameter to ->rq_interpreter call-back. NULL is passed
2609 there. Actual usage will be in CLIO.
2613 Description: Replace RW_LOCK_UNLOCKED() macro with rwlock_init().
2614 Details : Replace RW_LOCK_UNLOCKED() with rwlock_init() as the former
2615 doesn't work with lockdep.
2619 Description: Add rwv.c test program.
2620 Details : New testing program exercising readv(2) and writev(2) (Qian).
2624 Description: Add sendfile.c test program.
2625 Details : New testing program exercising sendfile(2) (Jay).
2629 Description: Ratelimit a message that can be very frequent.
2630 Details : Ratelimit a memory allocation failure message that can
2635 Description: Use cdebug_show() in CDEBUG-style macros defined outside of libcfs.
2636 Details : Use cdebug_show() in CDEBUG-style macros defined outside of libcfs.
2640 Description: Liblustre build fixes.
2641 Details : Liblustre build fixes.
2645 Description: libcfs: add cfs_{need,cond}_resched() interface.
2646 Details : libcfs: add cfs_{need,cond}_resched() definition and
2647 implementations for Linux, NT, and liblustre.
2649 Severity : enhancement
2651 Description: More exported tunables for mballoc
2652 Details : Add support for tunable preallocation window and new tunables for
2653 large/small requests
2657 Description: Detect corruption of block bitmap and checking for preallocations
2658 Details : Checks validity of on-disk block bitmap. Also it does better
2659 checking of number of applied preallocations. When corruption is
2660 found, it turns filesystem readonly to prevent further corruptions.
2664 Description: (rw.c:1323:ll_read_ahead_pages()) ASSERTION(page_idx > ria->ria_stoff) failed
2665 Details : Once the unmatched stride IO mode is detected, shrink the stride-ahead
2666 window to 0. If it does hit cache miss, and read-pattern is still
2667 stride-io mode, does not reset the stride window, but also does not
2668 increase the stride window length in this case.
2672 Frequency : only for big-endian servers
2673 Description: Check if system is big-endian while mounting fs with extents feature
2674 Details : Mounting a filesystem with extents feature will fail on big-endian
2675 systems since ext3-based ldiskfs is not supported on big-endian
2676 systems. This can be over-riden with "bigendian_extents" mount option.
2678 Severity : enhancement
2680 Description: The root squash functionality
2681 Details : A security feature, which is to prevent users from being able
2682 to mount lustre on their desktop, run as root, and delete
2683 all of the files in the filesystem. The goal is accomplished by
2684 remapping user id (UID) and group id (GID) of the root user to
2685 a UID and GID specified by the system administartor via Lustre
2686 configuration management server (MGS). The functionality also
2687 allows to specify sets of clients for which the remapping does
2692 Description: Excessive recovery window
2693 Details : With AT enabled, the recovery window can be excessively long (6000+
2694 seconds). To address this problem, we no longer use
2695 OBD_RECOVERY_FACTOR when extending the recovery window (the connect
2696 timeout no longer depends on the service time, it is set to
2697 INITIAL_CONNECT_TIMEOUT now) and clients report the old service
2698 time via pb_service_time.
2702 Description: Watchdog triggered on MDS failover
2703 Details : enable OBD_CONNECT_MDT flag when connecting from the MDS so that
2704 the OSTs know that the MDS "UUID" can be reused for the same export
2705 from a different NID, so we do not need to wait for the export to be
2709 Frequency : rare, only if using MMP with Linux RAID
2711 Description: MMP doesn't work with Linux RAID
2712 Details : While using HA for Lustre servers with Linux RAID, it is possible
2713 that MMP will not detect multiple mounts. To make this work we
2714 need to unplug the device queue in RAID when the MMP block is being
2715 written. Also while reading the MMP block, we should read it from
2716 disk and not the cached one.
2718 Severity : enhancement
2720 Description: open file using fid
2721 Details : A file can be opened using just its fid, like
2722 <mntpt>/.lustre/fid/SEQ:OID:VER - this is needed for HSM and replication
2725 Frequency : Only in RHEL5 when mounting multiple ext3 filesystems
2728 Description: "kmem_cache_create: duplicate cache jbd_4k" error message
2729 Details : add proper locking for creation of jbd_4k slab cache
2733 Description: MMP check in ext3_remount() fails without displaying any error
2734 Details : When multiple mount protection fails during remount, proper error
2737 Severity : enhancement
2739 Description: Allow stripe size to be up to 4G-64k
2740 Details : Fix math logic to allow large stripe sizes.
2744 Description: add check for >8TB ldiskfs filesystems
2745 Details : ext3-based ldiskfs does not support greater than 8TB LUNs.
2746 Don't allow >8TB ldiskfs filesystems to be mounted without
2747 force_over_8tb mount option
2752 Descriptoin: OSS complains "too many blocks" and crashes
2753 Details : In filter_commitrw_write():
2754 1. lquota_chkquota();
2755 2. obdo_from_inode();
2756 3. lquota_pending_commit();
2757 "2." has changed uid and gid in this case, then caused
2758 lquota_chkquota() and lquota_pending_commit() to use
2759 different uid and gid. That is the root of the bug.
2763 Descriptoin: getstripe doesn't report stripe count = -1 sanely
2764 Details : Detect and print out the 'stripe count = -1' case properly.
2768 Description: Out or order replies might be lost on replay
2769 Details : In ptlrpc_retain_replayable_request if we cannot find retained
2770 request with tid smaller than one currently being added, add it
2771 to the start, not end of the list.
2773 --------------------------------------------------------------------------------
2775 2007-08-10 Cluster File Systems, Inc. <info@clusterfs.com>
2777 * Support for kernels:
2778 2.6.5-7.283 (SLES 9),
2779 2.6.9-55.EL (RHEL 4),
2780 2.6.16.46-0.14 (SLES 10),
2781 2.6.18.8 vanilla (kernel.org)
2782 * Client support for unpatched kernels:
2783 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
2784 2.6.16 - 2.6.22 vanilla (kernel.org)
2785 * Due to recently discovered recovery problems, we do not recommend
2786 using patchless RHEL 4 clients with this or any earlier release.
2787 * Recommended e2fsprogs version: 1.39.cfs8
2788 * Note that reiserfs quotas are disabled on SLES 10 in this kernel.
2789 * Starting with this release, the ldiskfs backing filesystem required
2790 by Lustre is now in its own package, lustre-ldiskfs. This package
2791 should be installed. It is versioned separately from Lustre and
2792 may be released separately in future.
2796 Description: block reactivating mgc import until all deactivates complete
2797 Details : Fix race when failing back MDT/MGS to itself (testing)
2800 Frequency : at statup only
2802 Description: mds_lov_synchronize race leads to various problems
2803 Details : simultaneous MDT->OST connections at startup can cause the
2804 sync to abort, leaving the OSC in a bad state.
2806 Severity : enhancement
2808 Description: add optional extra BUILD_VERSION info
2809 Details : add a new environment variable (namely LUSTRE_VERS) which allows
2810 to override the lustre version.
2813 Frequency : 2.6.18 servers only
2815 Description: ll_kern_mount() doesn't release the module reference
2816 Details : The ldiskfs module reference count never drops down to 0
2817 because ll_kern_mount() doesn't release the module reference.
2822 Description: server LBUG when using old ost_num_threads parameter
2823 Details : Accept the old ost_num_threads parameter but warn that it
2824 is deprecated, and fix an off-by-one error that caused an LBUG.
2829 Description: Transient SCSI error results in persistent IO issue
2830 Details : iobuf->dr_error is not reinitialized to 0 between two
2834 Frequency : sometimes when underlying device returns I/O errors
2836 Description: OSTs not going read-only during write failures
2837 Details : OSTs are not remounted read-only when the journal commit threads
2838 get I/O errors because fsfilt_ext3 calls journal_start/stop()
2839 instead of the ext3 wrappers.
2843 Description: poor connect scaling with increasing client count
2844 Details : Don't run filter_grant_sanity_check for more than 100 exports
2845 to improve scaling for large numbers of clients.
2848 Frequency : SLES10 only
2850 Description: sanity-quota.sh quotacheck failed: rc = -22
2851 Details : Quotas cannot be enabled on SLES10.
2854 Frequency : liblustre clients only
2856 Description: getdirentries does not give error when run on compute nodes
2857 Details : getdirentries does not fail when the size specified as an argument
2858 is too small to contain at least one entry
2860 Severity : enhancement
2862 Description: Add LNET router traceability for debug purposes
2863 Details : If a checksum failure occurs with a router as part of the
2864 IO path, the NID of the last router that forwarded the bulk data
2865 is printed so it can be identified.
2870 Description: OST "spontaneously" evicts client; client has imp_pingable == 0
2871 Details : Due to a race condition, liblustre clients were occasionally
2872 evicted incorrectly.
2874 Severity : enhancement
2876 Description: lfs setstripe use optional parameters instead of postional
2879 Severity : enhancement
2881 Description: Nanosecond timestamp support for ldiskfs
2882 Details : The on-disk ldiskfs filesystem has added support for nanosecond
2883 resolution timestamps. There is not yet support for this at
2884 the Lustre filesystem level.
2887 Frequency : during server recovery
2889 Description: MDS failing to send precreate requests due to OSCC_FLAG_RECOVERING
2890 Details : request with rq_no_resend flag not awake l_wait_event if they get
2894 Frequency : nfs export on patchless client
2896 Description: connectathon hang when test nfs export over patchless client
2897 Details : Disconnected dentry cannot be found with lookup, so we do not need
2898 to unhash it or make it invalid
2901 Description: fix llapi_lov_get_uuids() to allow many OSTs to be returned
2902 Details: : Change llapi_lov_get_uuids() to read the UUIDs from /proc instead
2903 of using an ioctl. This allows lfsck for > 160 OSTs to succeed.
2908 Description: open req refcounting wrong on reconnect
2909 Details : If reconnect happened between getting open reply from server and
2910 call to mdc_set_replay_data in ll_file_open, we will schedule
2911 replay for unreferenced request that we are about to free.
2912 Subsequent close will crash in variety of ways.
2913 Check that request is still eligible for replay in
2914 mdc_set_replay_data().
2919 Description: disable writes to filesystem when reading health_check file
2920 Details : the default for reading the health_check proc file has changed
2921 to NOT do a journal transaction and write to disk, because this
2922 can cause reads of the /proc file to hang and block HA state
2923 checking on a healthy but otherwise heavily loaded system. It
2924 is possible to return to the previous behaviour during configure
2925 with --enable-health-write.
2927 Severity : enhancement
2929 Description: 64-bit inode version
2930 Details: : Add a on-disk 64-bit inode version for ext3 to track changes made
2931 to the inode. This will be required for version-based recovery.
2936 Description: MDS fails to start if a duplicate client export is detected
2937 Details : in some rare cases it was possible for a client to connect to
2938 an MDS multiple times. Upon recovery the MDS would detect this
2939 and fail during startup. Handle this more gracefully.
2941 Severity : enhancement
2943 Description: Add -o localflock option to simulate old noflock
2945 Details : This will achieve local-only flock/fcntl locks
2951 Description: log_commit_thread vs filter_destroy race leads to crash
2952 Details : Take import reference before releasing llog record semaphore
2957 Description: Wrong request locking in request set processing
2958 Details : ptlrpc_check_set wrongly uses req->rq_lock for proctect add to
2959 imp_delayed_list, in this place should be used imp_lock.
2962 Frequency : when reconnection
2964 Description: Grant Leak when osc reconnect to OST
2965 Details : When osc reconnect ost, OST(filter) should check whether it
2966 should grant more space to client by comparing fed_grant and
2967 cl_avail_grant, and return the granted space to client instead
2968 of "new granted" space, because client will call osc_init_grant
2969 to update the client grant space info.
2972 Frequency : when client reconnect to OST
2974 Description: Grant Leak when osc do resend and replay bulk write
2975 Details : When osc reconnect to OST, OST(filter)should clear grant info of
2976 bulk write request, because the grant info while be sync between
2977 OSC and OST when reconnect, and we should ignore the grant info
2978 these of resend/replay write req.
2983 Description: Grant space more than available left space sometimes.
2984 Details : When then OST is about to be full, if two bulk writing from
2985 different clients came to OST. Accord the avaliable space of the
2986 OST, the first req should be permitted, and the second one
2987 should be denied by ENOSPC. But if the seconde arrived before
2988 the first one is committed. The OST might wrongly permit second
2989 writing, which will cause grant space > available space.
2992 Frequency : when client is evicted
2994 Description: Grant might be wrongly erased when osc is evicted by OST
2995 Details : when the import is evicted by server, it will fork another
2996 thread ptlrpc_invalidate_import_thread to invalidate the
2997 import, where the grant will be set to 0. While the original
2998 thread will update the grant it got when connecting. So if
2999 the former happened latter, the grant will be wrongly errased
3000 because of this race.
3005 Description: Checking Stale with correct fid
3006 Details : ll_revalidate_it should uses de_inode instead of op_data.fid2
3007 to check whether it is stale, because sometimes, we want the
3008 enqueue happened anyway, and op_data.fid2 will not be initialized.
3010 Severity : enhancement
3012 Description: update patchless client
3013 Details : Add support for patchless client with 2.6.20, 2.6.21 and RHEL 5
3016 Frequency : only with 2.4 kernel
3018 Description: random memory corruption
3019 Details : size of struct ll_inode_info is to big for union inode.u and this
3020 can be cause of random memory corruption.
3025 Description: Memory leak in recovery
3026 Details : Lov_mds_md was not free in an error handler in mds_create_object.
3027 It should also check obd_fail before fsfilt_start, otherwise if
3028 fsfilt_start return -EROFS,(failover mds during mds recovery).
3029 then the req will return with repmsg->transno = 0 and rc = EROFS.
3030 and we met hit the assert LASSERT(req->rq_reqmsg->transno ==
3031 req->rq_repmsg->transno) in ptlrpc_replay_interpret. Fcc should
3032 be freed no matter whether fsfilt_commit success or not.
3035 Frequency : only with huge count clients
3037 Description: Prevents from taking the superblock lock in llap_from_page for
3039 Details : using LL_ORIGIN_REMOVEPAGE origin flag instead of LL_ORIGIN_UNKNOW
3040 for llap_from_page call in ll_removepage prevents from taking the
3041 superblock lock for a soon died page.
3046 Description: Not check open intent error before release open handle
3047 Details : in some rare cases, the open intent error is not checked before
3048 release open handle, which may cause
3049 ASSERTION(open_req->rq_transno != 0), because it tries to release
3050 the failed open handle.
3055 Description: Set cat log bitmap only after create log success.
3056 Details : in some rare cases, the cat log bitmap is set too early. and it
3057 should be set only after create log success.
3061 Description: Accessing a block bevice can re-enable I/O when Lustre is
3062 tearing down a device.
3063 Details : dev_clear_rdonly(bdev) must be called in kill_bdev() instead of
3068 Description: service threads may hog cpus when there are a lot of requests
3069 Details : Insert cond_resched to give other threads a chance to use some CPU
3074 Description: the cat log was not initialized in recovery
3075 Details : When mds(mgs) do recovery, the tgt_count might be zero, so the
3076 unlink log on mds will not be initialized until mds post
3077 recovery. And also in mds post recovery, the unlink log will
3078 initialization will be done asynchronausly, so there will be race
3079 between add unlink log and unlink log initialization.
3083 Description: brw_stats were being printed incorrectly
3084 Details : brw_stats were being printed as log2 but all of them were not
3085 recorded as log2. Also remove some code duplication arising from
3086 filter_tally_{read,write}.
3090 Frequency : rare, only in recovery.
3091 Description: ASSERTION(req->rq_type != LI_POISON) failed
3092 Details : imp_lock should be held while iterating over imp_sending_list for
3093 prevent destroy request after get timeout in ptlrpc_queue_wait.
3097 Description: replay-single.sh test 52 fails
3098 Details : A lock's skiplist need to be cleanup when it being unlinked
3099 from its resource list.
3103 Description: Short directio read returns full requested size rather than
3105 Details : Direct I/O operations should return actual amount of bytes
3106 transferred rather than requested size.
3108 Severity : enhancement
3110 Description: metadata RPC reduction (e.g. for rm performance)
3111 Details : decrease the amount of synchronous RPC between clients and servers
3112 by canceling conflicing lock before the operation on the client
3113 and packing thier handles into the main operation RPC to server.
3115 Severity : enhancement
3117 Description: Async OSC create to avoid the blocking unnecessarily.
3118 Details : If an OST has no remaining object, system will block on the
3119 creation when it needs to create a new object on this OST. Now,
3120 ways use pre-created objects when available, instead of blocking on
3121 an empty osc while others are not empty. If we must block, we
3122 block for the shortest possible period of time.
3126 Description: improve handling recoverable errors
3127 Details : if request processig with error which can be recoverable on server
3128 request should be resend, otherwise page released from cache and
3131 Severity : enhancement
3133 Description: refine locking for avoid write wrong info into lov_objid file
3134 Details : fix possible races with add new target and write/update data in
3137 --------------------------------------------------------------------------------
3139 2007-05-03 Cluster File Systems, Inc. <info@clusterfs.com>
3144 Frequency : on some architectures
3146 Description: 1.6 client sometimes fails to mount from a 1.4 MDT
3147 Details : Uninitialized flags sometimes cause configuration commands to
3151 Frequency : patchless clients only
3153 Description: missing __iget() symbol export
3154 Details : The __iget() symbol export is missing. To avoid the need for
3155 this on patchless clients the deathrow inode reaper is turned
3156 off, and we depend on the VM to clean up old inodes. This
3157 dependency was during via the fix for bug 12181.
3159 --------------------------------------------------------------------------------
3161 2007-04-19 Cluster File Systems, Inc. <info@clusterfs.com>
3163 * CONFIGURATION CHANGE. This version of Lustre WILL NOT
3164 INTEROPERATE with older versions automatically. In many cases a
3165 special upgrade step is needed. Please read the
3166 user documentation before upgrading any part of a live system.
3167 * WIRE PROTOCOL CHANGE from previous 1.6 beta versions. This
3168 version will not interoperate with 1.6 betas before beta5 (1.5.95).
3169 * WARNING: Lustre configuration and startup changes are required with
3170 this release. See https://mail.clusterfs.com/wikis/lustre/MountConf
3174 Severity : enhancement
3176 Description: MountConf
3177 Details : Lustre configuration is now managed via mkfs and mount
3178 commands instead of lmc and lconf. New obd types (MGS, MGC)
3179 are added for dynamic configuration management. See
3180 https://mail.clusterfs.com/wikis/lustre/MountConf for
3183 Severity : enhancement
3185 Description: dynamic OST addition
3186 Details : OSTs can now be added to a live filesystem
3188 Severity : enhancement
3190 Description: startup order invariance
3191 Details : MDTs and OSTs can be started in any order. Clients only
3192 require the MDT to complete startup.
3194 Severity : enhancement
3196 Description: parallel, asynchronous orphan cleanup
3197 Details : orphan cleanup is now performed in separate threads for each
3198 OST, allowing parallel non-blocking operation.
3200 Severity : enhancement
3202 Description: optimized stripe assignment
3203 Details : stripe assignments are now made based on ost space available,
3204 ost previous usage, and OSS previous usage, in order to try
3205 to optimize storage space and networking resources.
3207 Severity : enhancement
3209 Description: Permanently set tunables
3210 Details : All writable /proc/fs/lustre tunables can now be permanently
3211 set on a per-server basis, at mkfs time or on a live system.
3213 Severity : enhancement
3215 Description: Lustre message v2
3216 Details : Add lustre message format v2.
3218 Severity : enhancement
3220 Description: client OST exclusion list
3221 Details : Clients can be started with a list of OSTs that should be
3222 declared "inactive" for known non-responsive OSTs.
3226 Description: SPEC SFS validation failure on NFS v2 over lustre.
3227 Details : Changes the blocksize for regular files to be 2x RPC size,
3228 and not depend on stripe size.
3230 Severity : enhancement
3232 Description: Multiple MD RPCs in flight.
3233 Details : Further unserialise some read-only MDS RPCs - learn about intents.
3234 To avoid overly-overloading MDS, introduce a limit on number of
3235 MDS RPCs in flight for a single client and add /proc controls
3236 to adjust this limit.
3238 Severity : enhancement
3240 Description: client read/write statistics
3241 Details : Add client read/write call usage stats for performance
3242 analysis of user processes.
3243 /proc/fs/lustre/llite/*/offset_stats shows non-sequential
3244 file access. extents_stats shows chunk size distribution.
3245 extents_stats_per_process show chunk size distribution per
3248 Severity : enhancement
3250 Description: mds statistics
3251 Details : Add detailed mds operations statistics in
3252 /proc/fs/lustre/mds/*/stats.
3256 Description: Failure of copying files with lustre special EAs.
3257 Details : Client side always return success for setxattr call for lustre
3258 special xattr (currently only "trusted.lov").
3263 Description: Refcount LNET uuids
3264 Details : The global LNET uuid list grew linearly with every startup;
3265 refcount repeated list entries instead of always adding to
3268 Severity : enhancement
3270 Description: Dynamic service threads
3271 Details : Within a small range, start extra service threads
3272 automatically when the request queue builds up.
3275 Frequency : mixed-endian client/server environments
3277 Description: mixed-endian crashes
3278 Details : The new msg_v2 system had some failures in mixed-endian
3281 Severity : enhancement
3283 Description: Easy OST removal
3284 Details : OSTs can be permanently deactivated with e.g. 'lctl
3285 conf_param lustre-OST0001.osc.active=0'
3287 Severity : enhancement
3289 Description: MGS proc entries
3290 Details : Added basic proc entries for the MGS showing what filesystems
3293 Severity : enhancement
3295 Description: provide MGS failover
3296 Details : Added config lock reacquisition after MGS server failover.
3298 Severity : enhancement
3300 Description: add Linux 2.4 support
3301 Details : Added support for RHEL 2.4.21 kernel for 1.6 servers and clients
3305 Description: a large application tries to do I/O to the same resource and dies
3306 in the middle of it.
3307 Details : Check the req->rq_arrival time after the call to
3308 ost_brw_lock_get(), but before we do anything about
3309 processing it & sending the BULK transfer request. This
3310 should help move old stale pending locks off the queue as
3311 quickly as obd_timeout.
3314 Frequency : when an incorrect nid is specified during startup
3316 Description: ptlrpc connect to non-existant node causes kernel crash
3317 Details : LNET can't be re-entered from an event callback, which
3318 happened when we expire a message after the export has been
3319 cleaned up. Instead, hand the zombie cleanup off to another
3322 Severity : enhancement
3324 Description: plain/inodebits lock performance improvement
3325 Details : Grouping plain/inodebits in granted list by their request modes
3326 and bits policy, thus improving the performance of search through
3330 Frequency : only if OST filesystem is corrupted
3332 Description: client incorrectly hits assertion in ptlrpc_replay_req()
3333 Details : for a short time RPCs with bulk IO are in the replay list,
3334 but replay of bulk IOs is unimplemented. If the OST filesystem
3335 is corrupted due to disk cache incoherency and then replay is
3336 started it is possible to trip an assertion. Avoid putting
3337 committed RPCs into the replay list at all to avoid this issue.
3340 Frequency : liblustre (e.g. catamount) on a large cluster with >= 8 OSTs/OSS
3342 Description: System hang on startup
3343 Details : This bug allowed the liblustre (e.g. catamount) client to
3344 return to the app before handling all startup RPCs. This
3345 could leave the node unresponsive to lustre network traffic
3346 and manifested as a server ptllnd timeout.
3348 Severity : enhancement
3350 Description: Add "/proc/sys/lustre/debug_peer_on_timeout"
3351 Details : liblustre envirable: LIBLUSTRE_DEBUG_PEER_ON_TIMEOUT
3352 boolean to control whether to print peer debug info when a
3353 client's RPC times out.
3356 Frequency : only for kernels with patches from Lustre below 1.4.3
3358 Description: Remove old rdonly API
3359 Details : Remove old rdonly API which unused from at least lustre 1.4.3
3362 Frequency : only for devices with external journals
3364 Description: Set external device read-only also
3365 Details : During a commanded failover stop, we set the disk device
3366 read-only while the server shuts down. We now also set any
3367 external journal device read-only at the same time.
3370 Frequency : when upgrading from 1.4 while trying to change parameters
3372 Description: The wrong (new) MDC name was used when setting parameters for
3373 upgraded MDT's. Also allows changing of OSC (and MDC)
3374 parameters if --writeconf is specified at tunefs upgrade time.
3377 Frequency : when setting specific ost indicies
3379 Description: QOS code breaks on skipped indicies
3380 Details : Add checks for missing OST indicies in the QOS code, so OSTs
3381 created with --index need not be sequential.
3383 Severity : enhancement
3385 Description: Add uninit_groups feature to ldiskfs2 to speed up e2fsck
3386 Details : The uninit_groups feature works in conjunction with the kernel
3387 filesystem code (ldiskfs2 only) and e2fsprogs-1.39-cfs6 to speed
3388 up the pass1 processing of e2fsck. This is a read-only feature
3389 in ldiskfs2 only, so older kernels and current ldiskfs cannot
3390 mount filesystems that have had this feature enabled.
3392 Severity : enhancement
3394 Description: Improve multi-block allocation algorithm to avoid fragmentation
3395 Details : The mballoc3 code (ldiskfs2 only) adds new mechanisms to improve
3396 allocation locality and avoid filesystem fragmentation.
3398 ------------------------------------------------------------------------------
3400 2007-04-01 Cluster File Systems, Inc. <info@clusterfs.com>
3402 * Support for kernels:
3403 2.4.21-47.0.1.EL (RHEL 3)
3404 2.6.5-7.283 (SLES 9)
3405 2.6.9-42.0.10.EL (RHEL 4)
3406 2.6.12.6 vanilla (kernel.org)
3407 2.6.16.27-0.9 (SLES 10)
3408 * Recommended e2fsprogs version: 1.39.cfs5
3410 * Note that reiserfs quotas are disabled on SLES 10 in this kernel
3414 Frequency : occasional, depends on client load and configuration
3415 Bugzilla : 12181, 12203
3416 Description: data loss for recently-modified files
3418 Details : In some cases it is possible that recently written or created
3419 files may not be written to disk in a timely manner (this should
3420 normally be within 30s unless client IO load is very high).
3421 The problem appears as zero-length files or files that are a
3422 multiple of 1MB in size after a client crash or client eviction
3423 that are missing data at the end of the file.
3425 This problem is more likely to be hit on clients where files are
3426 repeatedly created and unlinked in the same directory, clients
3427 have a large amount of RAM, have many CPUs, the filesystem has
3428 many OSTs, the clients are rebooted frequently, and/or the files
3429 are not accessed by other nodes after being written.
3431 The presence of the problem can be detected by looking at
3432 /proc/sys/fs/inode-state. If the first number (nr_inodes) is
3433 smaller than the second (nr_unused) then dirty files will not
3434 be flushed automatically to disk. "sync; sleep 10" should be
3435 run several times on the node before unmounting it to update
3436 Lustre (this is also safe to run on nodes without this problem).
3438 There is also a related kernel bug in the RHEL4 4 2.6.9 kernel
3439 that can cause this same problem, so customers using that kernel
3440 also need to update the kernel in addition to Lustre. In order
3441 to properly fix this bug, the RHEL3 2.4.21 kernel is also updated.
3443 It is normal that files written just before a client crash (less
3444 than 30s) may not yet have been flushed to disk, even for local
3448 Frequency : frequent on thin XT3 nodes
3450 Description: UUID collision on thin XT3 Linux nodes
3451 Details : UUIDs on Compute Node Linux XT3 nodes were not generated
3452 randomly, since we relied on an insufficiently-seeded PRNG.
3457 Description: OSS hangs after "All ost request buffers busy"
3458 Details : A deadlock between quota and journal operations caused OSS
3459 hangs after printing "All ost request buffers busy."
3462 Frequency : always on liblustre builds
3464 Description: Cleanup compiler warnings on liblustre
3467 Frequency : always on liblustre builds on XT3
3469 Description: LC_CONFIG_CDEBUG don't run while build liblustre on XT3.
3473 Description: Addition of EXT3_FEATURE_RO_COMPAT_DIR_NLINKS flag for
3474 > 32000 subdirectories
3475 Details : Add EXT3_FEATURE_RO_COMPAT_DIR_NLINK flag to
3476 EXT3_FEATURE_RO_COMPAT_SUPP. This flag will be set whenever
3477 subdirectory count crosses 32000. This will aid e2fsck to
3478 correctly handle more than 32000 subdirectories.
3481 Frequency : liblustre (e.g. catamount) on a large cluster with >= 8 OSTs/OSS
3483 Description: System hang on startup
3484 Details : This bug allowed the liblustre (e.g. catamount) client to
3485 return to the app before handling all startup RPCs. This
3486 could leave the node unresponsive to lustre network traffic
3487 and manifested as a server ptllnd timeout.
3489 Severity : enhancement
3491 Description: Add "/proc/sys/lustre/debug_peer_on_timeout"
3492 (liblustre envirable: LIBLUSTRE_DEBUG_PEER_ON_TIMEOUT)
3493 boolean to control whether to print peer debug info when a
3494 client's RPC times out.
3499 Description: make O_SYNC working on 2.6 kernels
3500 Details : 2.6 kernels use different method for mark pages for write,
3501 so need add a code to lustre for O_SYNC work.
3506 Description: Failure to close file and release space on NFS
3507 Details : Put inode details into lock acquired in ll_intent_file_open.
3508 Use mdc_intent_lock in ll_intent_open to properly
3509 detect all kind of errors unhandled by mdc_enqueue.
3514 Description: proc file read during shutdown sometimes raced obd removal,
3516 Details : Add lock to prevent obd access after proc file removal.
3519 Frequency : Only for files larger than 4GB on 32-bit clients.
3521 Description: improperly doing page alignment of locks
3522 Details : Modify lustre core code to use CFS_PAGE_* defines instead of
3523 PAGE_*. Make CFS_PAGE_MASK a 64-bit mask.
3528 Description: RPCs being resent when they shouldn't be
3529 Details : Some RPCs that should not be resent are being resent. This
3530 can cause inconsistencies in the RPC state machine. Do not
3531 resend such requests.
3534 Frequency : rare, only with NFS export
3536 Description: Crash on NFS re-export node
3537 Details : under very unusual load conditions an assertion is hit in
3538 ll_intent_file_open()
3541 Frequency : only if OST filesystem is corrupted
3543 Description: client incorrectly hits assertion in ptlrpc_replay_req()
3544 Details : for a short time RPCs with bulk IO are in the replay list,
3545 but replay of bulk IOs is unimplemented. If the OST filesystem
3546 is corrupted due to disk cache incoherency and then replay is
3547 started it is possible to trip an assertion. Avoid putting
3548 committed RPCs into the replay list at all to avoid this issue.
3553 Description: large O_DIRECT requests fail under memory pressure/fragmentation
3554 Details : Large single O_DIRECT read and write calls can fail to allocate
3555 a sufficiently large buffer to process the request. In case of
3556 allocation failure the allocation is retried with a smaller
3557 buffer and broken into smaller requests.
3559 Severity : enhancement
3561 Description: Add -o localflock option to simulate old noflock behaviour.
3562 Details : This will achieve local-only flock/fcntl locks coherentness.
3567 Description: versioning check is incomplete
3568 Details : Checking the version difference of client vs. server, report
3569 error if the gap is too big.
3574 Description: add support PG_writeback bit
3575 Details : add support for PG_writeback bit for Lustre, for more carefull
3576 work with page cache in 2.6 kernel. This also fix some deadlocks
3577 and remove hack for work O_SYNC with 2.6 kernel.
3579 Severity : enhancement
3581 Description: Add uninit_groups feature to ldiskfs2 to speed up e2fsck
3582 Details : The uninit_groups feature works in conjunction with the kernel
3583 filesystem code (ldiskfs2 only) and e2fsprogs-1.39-cfs6 to speed
3584 up the pass1 processing of e2fsck. This is a read-only feature
3585 in ldiskfs2 only, so older kernels and current ldiskfs cannot
3586 mount filesystems that have had this feature enabled.
3588 Severity : enhancement
3590 Description: Improve multi-block allocation algorithm to avoid fragmentation
3591 Details : The mballoc3 code (ldiskfs2 only) adds new mechanisms to improve
3592 allocation locality and avoid filesystem fragmentation.
3594 ------------------------------------------------------------------------------
3596 2007-02-09 Cluster File Systems, Inc. <info@clusterfs.com>
3598 * Support for kernels:
3599 2.6.9-42.0.3.EL (RHEL 4)
3600 2.6.5-7.276 (SLES 9)
3601 2.4.21-47.0.1.EL (RHEL 3)
3602 2.6.12.6 vanilla (kernel.org)
3603 2.6.16.21-0.8 (SLES10)
3604 * Recommended e2fsprogs version: 1.39.cfs2-0
3606 * The backwards-compatible /proc/sys/portals symlink has been removed
3607 in this release. Before upgrading, please ensure that you change
3608 any configuration scripts or /etc/sysctl.conf files that access
3609 /proc/sys/portals/* or sysctl portals.* to use the corresponding
3610 entry in /proc/sys/lnet or sysctl lnet.*. This change can be made
3611 in advance of the upgrade on any system running Lustre 1.4.6 or
3612 newer, since /proc/sys/lnet was added in that version.
3613 * Note that reiserfs quotas are disabled on SLES 10 in this kernel
3617 Frequency : only when quota is used
3619 Description: avoid scanning export list for quota master
3620 Details : Change the algorithms to avoid scanning export list in order
3621 to improve the efficiency.
3624 Frequency : MDS failover only, very rarely
3626 Description: "went back in time" messages on mds failover
3627 Details : The greatest transno may be lost when the current operation
3628 finishes with an error (transno==0) and the client's last_rcvd
3629 record is over-written. Save the greatest transno in the
3630 mds_last_transno for this case.
3633 Frequency : always for specific kernels and striping counts
3635 Description: client may get "Matching packet too big" without ACL support
3636 Details : Clients compiled without CONFIG_FS_POSIX_ACL get an error message
3637 when trying to access files in certain configurations. The
3638 clients should in fact be denied when mounting because they do
3639 not understand ACLs.
3642 Frequency : Cray XT3 with more than 4000 clients and multiple jobs
3644 Description: many clients connecting with IO in progress causes connect timeouts
3645 Details : Avoid synchronous journal commits to avoid delays caused by many
3646 clients connecting/disconnecting when bulk IO is in progress.
3647 Queue liblustre connect requests on OST_REQUEST_PORTAL instead of
3648 OST_IO_PORTAL to avoid delays behind potentially many pending
3652 Frequency : occasionally with multiple writers to a single file
3654 Description: shared writes to file may result in wrong size reported by stat()
3655 Details : Allow growing of kms when extent lock is cancelled
3658 Frequency : always with random mmap IO to multi-striped file
3660 Description: mmap write might be lost if we are writing to a 'hole' in stripe
3661 Details : Only if the hole is at the end of OST object so that kms is too
3662 small. Fix is to increase kms accordingly in ll_nopage.
3665 Frequency : rare, only if OST filesystem is inconsistent with MDS filesystem
3667 Description: writes to a missing object would leak memory on the OST
3668 Details : If there is an inconsistency between the MDS and OST filesystems,
3669 such that the MDS references an object that doesn't exist, writes
3670 to that object will leak memory due to incorrect cleanup in the
3671 error handling path, eventually running out of memory on the OST.
3676 Description: Creating too long symlink causes lustre errors
3677 Details : Check symlink and name lengths before sending requests to MDS.
3680 Frequency : only if flock is enabled (not on by default)
3682 Description: posix locks not released on fd closure on 2.6.9+
3683 Details : We failed to add posix locks to list of inode locks on 2.6.9+
3684 kernels, this caused such locks not to be released on fd close and
3685 then assertions on fs unmount about still used locks.
3688 Frequency : MDS failover only, very rarely
3690 Description: clients may get ASSERTION(granted_lock != NULL)
3691 Details : When request was taking a long time, and a client was resending
3692 a getattr by name lock request. The were multiple lock requests
3693 with the same client lock handle and
3694 mds_getattr_name->fixup_handle_for_resent_request found one of the
3695 lock handles but later failed with ASSERTION(granted_lock != NULL).
3700 Description: handle->h_buffer_credits > 0, assertion failure
3701 Details : h_buffer_credits is zero after truncate, causing assertion
3702 failure. This patch extends the transaction or creates a new
3706 Frequency : NFS re-export or patchless client
3707 Bugzilla : 11179, 10796
3708 Description: Crash on NFS re-export node (__d_move)
3709 Details : We do not want to hash the dentry if we don't have a lock.
3710 But if this dentry is later used in d_move, we'd hit uninitialised
3711 list head d_hash, so we just do this to init d_hash field but
3712 leave dentry unhashed.
3715 Frequency : NFS re-export or patchless client
3717 Description: NFS exports has problem with symbolic link
3718 Details : lustre client didn't properly install dentry when re-exported
3719 to NFS or running patchless client.
3722 Frequency : NFS re-export or patchless client
3724 Description: Various nfs/patchless fixes.
3725 Details : fixes reuse disconected alias for lookup process - this fixes
3726 warning "find_exported_dentry: npd != pd",
3727 fix permission error with open files at nfs.
3728 fix apply umask when do revalidate.
3731 Frequency : occasional
3733 Description: Crash on NFS re-export node
3734 Details : calling clear_page() on the wrong pointer triggered oops in
3735 generic_mapping_read().
3738 Frequency : rarely, using O_DIRECT IO
3740 Description: unaligned directio crashes client with LASSERT
3741 Details : check for unaligned buffers before trying any requests.
3744 Frequency : rarely, using CFS RAID5 patches in non-standard kernel series
3746 Description: stale data returned from RAID cache
3747 Details : If only a small amount of IO is done to the RAID device before
3748 reading it again it is possible to get stale data from the RAID
3749 cache instead of reading it from disk.
3752 Frequency : always for sles10 kernel
3754 Description: sles10 support
3755 Details : ll_follow_link: compile fixes and using of nd_set_link
3756 under newer kernels.
3759 Frequency : depends on arch, kernel and compiler version, always on sles10
3762 Description: recursive or deep enough symlinks cause stack overflow
3763 Details : getting rid of large stack-allocated variable in
3767 Frequency : depends on hardware
3769 Description: lustre write performance loss in the SLES10 kernel
3770 Details : the performance loss is caused by using of write barriers in the
3771 ext3 code. The SLES10 kernel turns barrier support on by
3772 default. The fix is to undo that change for ldiskfs.
3774 ------------------------------------------------------------------------------
3776 2006-12-09 Cluster File Systems, Inc. <info@clusterfs.com>
3778 * Support for kernels:
3779 2.6.9-42.0.3EL (RHEL 4)
3780 2.6.5-7.276 (SLES 9)
3781 2.4.21-47.0.1.EL (RHEL 3)
3782 2.6.12.6 vanilla (kernel.org)
3786 Frequency : quota enabled and large files being deleted
3788 Description: releasing more than 4GB of quota at once hangs OST
3789 Details : If a user deletes more than 4GB of files on a single OST it
3790 will cause the OST to spin in an infinite loop. Release
3791 quota in < 4GB chunks, or use a 64-bit value for 1.4.7.1+.
3796 Description: statfs data retrieved from /proc may be stale or zero
3797 Details : When reading per-device statfs data from /proc, in the
3798 {kbytes,files}_{total,free,avail} files, it may appear
3799 as zero or be out of date.
3802 Frequency : systems with MD RAID1 external journal devices
3804 Description: lconf's call to blkid is confused by RAID1 journal devices
3805 Details : Use the "blkid -l" flag to locate the MD RAID device instead
3806 of returning all block devices that match the journal UUID.
3809 Frequency : always, for aggregate stripe size over 4GB
3811 Description: "lfs setstripe" fails assertion when setting 4GB+ stripe width
3812 Details : Using "lfs setstripe" to set stripe size * stripe count over 4GB
3813 will fail the kernel with "ASSERTION(lsm->lsm_xfersize != 0)"
3816 Frequency : always if "lfs find" used on a local file/directory
3818 Description: "lfs find" segfaults if used on a local file/directory
3819 Details : The case where a directory component was not specified wasn't
3820 handled correctly. Handle this properly.
3823 Frequency : always on ppc64
3825 Description: the write to an ext3 filesystem mounted with mballoc got stuck
3826 Details : ext3_mb_generate_buddy() uses find_next_bit() which does not
3827 perform endianness conversion.
3830 Frequency : rarely (truncate to non-zero file size after write under load)
3831 Bugzilla : 10730, 10687
3832 Description: Files padded with zeros to next 4K multiple
3833 Details : With filesystems mounted using the "extents" option (2.6 kernels)
3834 it is possible that files that are truncated to a non-zero size
3835 immediately after being written are filled with zero bytes beyond
3836 the truncated size. No file data is lost.
3838 Severity : enhancement
3840 Description: Allow recovery/failover for liblustre clients.
3841 Details : liblustre clients were unaware of failover configurations until
3844 Severity : enhancement
3846 Description: user file locks should fail when not mounting with flock option
3847 Details : Set up an error-returning stub in ll_file_operations.lock field
3848 to prevent incorrect behaviour when client is mounted without
3849 flock option. Also, set up properly f_op->flock field for
3853 Frequency : always on ia64
3855 Description: "lfs df" loops on printing out MDS statfs information
3856 Details : The obd_ioctl_data was not initialized and in some systems
3857 this caused a failure during the ioctl that did not return
3858 an error. Initialize the struct and return an error on failure.
3861 Frequency : SLES 9 only
3863 Description: Error of copying files with lustre special EAs as root
3864 Details : Client side always return success for setxattr call for lustre
3865 special xattr (currently only "trusted.lov").
3868 Frequency : rarely on clusters with both ia64+i386 clients
3870 Description: ia64+i686 clients doing shared IO on the same file may LBUG
3871 Details : In rare cases when both ia64+i686 (or other mixed-PAGE_SIZE)
3872 clients are doing concurrent writes to the same file it is
3873 possible that the ia64 clients may LASSERT because the OST
3874 extent locks are not PAGE_SIZE aligned. Ensure that grown
3875 locks are always aligned on the request boundary.
3878 Frequency : specific use, occasional
3880 Description: Overwriting in use executable truncates on-disk binary image
3881 Details : If one node attempts to overwrite an executable in use by
3882 another node, we now correctly return ETXTBSY instead of
3883 truncating the file.
3885 Severity : enhancement
3887 Description: Async OSC create to avoid the blocking unnecessarily.
3888 Details : If an OST has no remaining object, system will block on the creating
3889 when need to create a new object on this OST. Now, ways use
3890 pre-created objects when available, instead of blocking on an
3891 empty osc while others are not empty. If we must block, we block
3892 for the shortest possible period of time.
3897 Description: chmod on Lustre root is propagated to other clients
3898 Details : Re-validate root's dentry in ll_lookup_it to avoid having it
3899 invalid by the follow_mount time.
3902 Frequency : liblustre clients only
3904 Description: Race in 'instant cancel' lock handling could lead to such locks
3905 never to be granted in case of SMP MDS
3906 Details : Do not destroy not yet granted but cbpending locks in
3910 Frequency : replay/resend of open
3912 Description: non null lock assetion failure in mds_intent_policy
3913 Details : Trying to replay/resend lockless open requests resulted in
3914 mds_open() returning 0 with no lock. Now it sets a flag if
3915 a lock is going to be returned.
3917 Severity : enhancement
3919 Description: Checksum enhancements
3920 Details : New checksum enhancements allow for resending RPCs that failed
3923 Severity : enhancement
3925 Description: Tunables on number of dirty pages in cacche
3926 Details : Allow to set limit on number of dirty pages cached.
3931 Description: client crash on unmount - lock still has references
3932 Details : In some error handling cases it was possible to leak a lock
3933 reference on a client while accessing a file. This was not
3934 harmful to the client during operation, but would cause the
3935 client to crash when the filesystem is unmounted.
3938 Frequency : specific case, rare
3940 Description: ETXTBSY on mds though file not in use by client
3941 Details : ETXTBSY is no longer incorrectly returned when attempting to
3942 chmod or chown a directory that the user previously tried to
3943 execute or a currently-executing binary.
3946 Frequency : extremely rare except on liblustre-based clients
3948 Description: Lustre space not freed when files are deleted
3949 Details : Clean up open-unlinked files after client eviction. Previously
3950 the unlink was skipped and the files remained as orphans.
3955 Description: OST failure "would be an LBUG" in waiting_locks_callback()
3956 Details : In some cases it was possible to send a blocking callback to
3957 a client doing a glimpse, even though that client didn't get
3958 a lock granted. When the glimpse lock is cancelled on the OST
3959 the freed lock is left on the waiting list and corrupted the list.
3962 Frequency : all core dumps
3964 Description: Broke core dumps to lustre
3965 Details : Negative dentry may be unhashed if parent does not have UPDATE
3966 lock, but some callers, e.g. do_coredump, expect dentry to be
3967 hashed after successful create, hash it in ll_create_it.
3969 ------------------------------------------------------------------------------
3971 2006-09-13 Cluster File Systems, Inc. <info@clusterfs.com>
3973 * Support for kernels:
3974 2.6.9-42.0.2.EL (RHEL 4)
3975 2.6.5-7.276 (SLES 9)
3976 2.4.21-40.EL (RHEL 3)
3977 2.6.12.6 vanilla (kernel.org)
3981 Frequency : always on RHEL 3
3983 Description: Number of open files grows over time
3984 Details : The number of open files grows over time, whether or not
3985 Lustre is started. This was due to a filp leak introduced
3986 by one of our kernel patches.
3988 ------------------------------------------------------------------------------
3990 08-20-2006 Cluster File Systems, Inc. <info@clusterfs.com>
3992 * Support for kernels:
3993 2.6.9-42.EL (RHEL 4)
3994 2.6.5-7.276 (SLES 9)
3995 2.4.21-40.EL (RHEL 3)
3996 2.6.12.6 vanilla (kernel.org)
4001 Bugzilla : 5719, 9635, 9792, 9684
4002 Description: OST (or MDS) trips assertions in (re)connection under heavy load
4003 Details : If a server is under heavy load and cannot reply to new
4004 connection requests before the client resends the (re)connect,
4005 the connection handling code can behave badly if two service
4006 threads are concurrently handing separate (re)connections from
4007 the same client. Add better locking to the connection handling
4008 code, and ensure that only a single connection will be processed
4009 for a given client UUID, even if the lock is dropped.
4011 Severity : enhancement
4013 Description: add TCP zero-copy support to kernel
4014 Details : Add support to the kernel TCP stack to allow zero-copy bulk
4015 sends if the hardware supports scatter-gather and checksumming.
4016 This allows socklnd to do client-write and server-read more
4017 efficiently and reduce CPU utilization from skbuf copying.
4020 Frequency : only if NFS exporting from client
4022 Description: NULL pointer deref in ll_iocontrol() if chattr mknod file
4023 Details : If setting attributes on a file created under NFS that had
4024 never been opened it would be possible to oops the client
4025 if the file had no objects.
4028 Frequency : always for liblustre
4030 Description: liblustre client does MDS+OSTs setattr RPC for each write
4031 Details : When doing a write from a liblustre client, the client
4032 incorrectly issued an RPC to the MDS and each OST the file was
4033 striped over in order to update the timestamps. When writing
4034 with small chunks and many clients this could overwhelm the MDS
4035 with RPCs. In all cases it would slow down the write because
4036 these RPCs are unnecessary.
4038 Severity : enhancement
4040 Description: allow number of MDS service threads to be changed at module load
4041 Details : It is now possible to change the number of MDS service threads
4042 running. Adding "options mds mds_num_threads={N}" to the MDS's
4043 /etc/modprobe.conf will set the number of threads for the next
4044 time Lustre is restarted (assuming the "mds" module is also
4045 reloaded at that time). The default number of threads will
4046 stay the same, 32 for most systems.
4051 Description: OST crash if filesystem is unformatted or corrupt
4052 Details : If an OST is started on a device that has never been formatted
4053 or if the filesystem is corrupt and cannot even mount then the
4054 error handling cleanup routines would dereference a NULL pointer.
4059 Description: NULL pointer deref in llap_from_page.
4060 Details : get_cache_page_nowait can return a page with NULL (or otherwise
4061 incorrect) mapping if the page was truncated/reclaimed while it was
4062 searched for. Check for this condition and skip such pages when
4063 doing readahead. Introduce extra check to llap_from_page() to
4064 verify page->mapping->host is non-NULL (so page is not anonymous).
4067 Frequency : Sometimes when using sys_sendfile
4069 Description: "page not covered by a lock" warnings from ll_readpage
4070 Details : sendfile called ll_readpage without right page locks present.
4071 Now we introduced ll_file_sendfile that does necessary locking
4072 around call to generic_file_sendfile() much like we do in
4076 Frequency : with certain MDS communication failures at client mount time
4078 Description: NULL pointer deref after failed client mount
4079 Details : a client connection request may delayed by the network layer
4080 and not be sent until after the PTLRPC layer has timed out the
4081 request. If the client fails the mount immediately it will try
4082 to clean up before the network times out the request. Add a
4083 reference from the request import to the obd device and delay
4084 the cleanup until the network drops the request.
4087 Frequency : occasionally during client (re)connect
4089 Description: assertion failure during client (re)connect
4090 Details : processing a client connection request may be delayed by the
4091 client or server longer than the client connect timeout. This
4092 causes the client to resend the connection request. If the
4093 original connection request is replied in this interval, the
4094 client may trip an assertion failure in ptlrpc_connect_interpret()
4095 which thought it would be the only running connect process.
4098 Frequency : only with obd_echo servers and clients that are rebooted
4100 Description: kernel BUG accessing uninitialized data structure
4101 Details : When running an obd_echo server it did not start the ping_evictor
4102 thread, and when a client was evicted an uninitialized data
4103 structure was accessed. Start the ping_evictor in the RPC
4104 service startup instead of the OBD startup.
4106 Severity : enhancement
4107 Bugzilla : 10193 (patchless)
4108 Description: Remove dependency on various unexported kernel interfaces.
4109 Details : No longer need reparent_to_init, exit_mm, exit_files,
4110 sock_getsockopt, filemap_populate, FMODE_EXEC, put_filp.
4113 Frequency : rare (only users of deprecated and unsupported LDAP config)
4115 Description: write_conf for zeroconf mount queried LDAP incorrectly for client
4116 Details : LDAP apparently contains 'lustreName' attributes instead of
4117 'name'. A simple remapping of the name is sufficient.
4120 Frequency : rare (only with non-default dump_on_timeout debug enabled)
4122 Description: waiting_locks_callback trips kernel BUG if client is evicted
4123 Details : Running with the dump_on_timeout debug flag turned on makes
4124 it possible that the waiting_locks_callback() can try to dump
4125 the Lustre kernel debug logs from an interrupt handler. Defer
4126 this log dumping to the expired_lock_main() thread.
4128 Severity : enhancement
4130 Description: Support NFS exporting on 2.6 kernels.
4131 Details : Implement non-rawops metadata methods for NFS server to use without
4132 changing NFS server code.
4135 Frequency : very rare (synthetic metadata workload only)
4137 Description: two racing renames might cause an MDS thread to deadlock
4138 Details : Running the "racer" program may cause one MDS thread to rename
4139 a file from being the source of a rename to being the target of
4140 a rename at exactly the same time that another thread is doing
4141 so, and the second thread has already enqueued these locks after
4142 doing a lookup of the target and is trying to relock them in
4143 order. Ensure that we don't try to re-lock the same resource.
4146 Frequency : only very large systems with liblustre clients
4148 Description: slow eviction of liblustre clients with the "evict_by_nid" RPC
4149 Details : Use asynchronous set_info RPCs to send the "evict_by_nid" to
4150 all OSTs in parallel. This allows the eviction of stale liblustre
4151 clients to proceed much faster than if they were done in series,
4152 and also offers similar improvements for other set_info RPCs.
4157 Description: excessive CPU usage during initial read phase on client
4158 Details : During the initial read phase on a client, it would agressively
4159 retry readahead on the file, consuming too much CPU and impacting
4160 performance (since 1.4.5.8). Improve the readahead algorithm
4161 to avoid this, and also improve some other common cases (read
4162 of small files in particular, where "small" is files smaller than
4163 /proc/fs/lustre/llite/*/max_read_ahead_whole_mb, 2MB by default).
4168 Description: MDS crash when receiving packet with unknown intent.
4169 Details : Do not LBUG in unknown intent case, just return -EFAULT
4171 Severity : enhancement
4172 Bugzilla : 9293, 9385
4173 Description: MDS RPCs are serialised on client. This is unnecessary for some.
4174 Details : Do not serialize getattr (non-intent version) and statfs.
4177 Frequency : occasional, when OST network is overloaded/intermittent
4179 Description: client evicted by OST after bulk IO timeout
4180 Details : If a client sends a bulk IO request (read or write) the OST
4181 may evict the client if it is unresposive to its data GET/PUT
4182 request. This is incorrect if the network is overloaded (takes
4183 too long to transfer the RPC data) or dropped the OST GET/PUT
4184 request. There is no need to evict the client at all, since
4185 the pinger and/or lock callbacks will handle this, and the
4186 client can restart the bulk request.
4189 Frequency : Always when mmapping file with no objects
4191 Description: client crashes when mmapping file with no objects
4192 Details : Check that we actually have objects in a file before doing any
4193 operations on objects in ll_vm_open, ll_vm_close and
4199 Description: Request leak when working with deleted CWD
4200 Details : Introduce advanced request refcount tracking for requests
4201 referenced from lustre intent.
4203 Severity : Enhancement
4205 Description: Cache open file handles on client.
4206 Details : MDS now will return special lock along with openhandle, if
4207 requested and client is allowed to hold openhandle, even if unused,
4208 until such a lock is revoked. Helps NFS a lot, since NFS is opening
4209 closing files for every read/write openration.
4211 Severity : Enhancement
4213 Description: Cache open negative dentries on client when possible.
4214 Details : Guard negative dentries with UPDATE lock on parent dir, drop
4215 negative dentries on lock revocation.
4220 Description: Remounting a client read-only wasn't possible with a zconf mount
4221 Details : It wasn't possible to remount a client read-only with llmount.
4223 Severity : enhancement
4224 Description: Include MPICH 1.2.6 Lustre ADIO interface patch
4225 Details : In lustre/contrib/ or /usr/share/lustre in RPM a patch for
4226 MPICH is included to add Lustre-specific ADIO interfaces.
4227 This is based closely on the UFS ADIO layer and only differs
4228 in file creation, in order to allow the OST striping to be set.
4229 This is user-contributed code and not supported by CFS.
4234 Description: extended inode attributes (immutable, append-only) work improperly
4235 when 2.4 and 2.6 kernels are used on client/server or vice versa
4236 Details : Introduce kernel-independent values for these flags.
4238 Severity : enhancement
4241 Description: Allow fractional MB tunings for lustre in /proc/ filesystem.
4242 Details : Many of the /proc/ tunables can only be tuned at a megabyte
4243 granularity. Now, Fractional MB granularity is be supported,
4244 this is very useful for low memory system.
4246 Severity : enhancement
4248 Description: Getattr by fid
4249 Details : Getting a file attributes by its fid, obtaining UPDATE|LOOKUP
4250 locks, avoids extra getattr rpc requests to MDS, allows '/' to
4251 have locks and avoids getattr rpc requests for it on every stat.
4254 Frequency : Always, for filesystems larger than 2TB
4256 Description: ldiskfs crash at mount for filesystem larger than 2TB with mballoc
4257 Details : Kenrel kmalloc limits allocations to 128kB and this prevents
4258 filesystems larger than 2TB to be mounted with mballoc enabled.
4261 Frequency : Always, for 32-bit kernel without CONFIG_LBD and filesystem > 2TB
4263 Description: ldiskfs crash at mount for filesystem larger than 2TB with mballoc
4264 Details : If a 32-bit kernel is compiled without CONFIG_LBD enabled and a
4265 filesystems larger than 2TB is mounted then the kernel will
4266 silently corrupt the start of the filesystem. CONFIG_LBD is
4267 enabled for all CFS-supported kernels, but the possibility of
4268 this happening with a modified kernel config exists.
4270 Severity : enhancement
4272 Description: add client O_DIRECT support for 2.6 kernels
4273 Details : It is now possible to do O_DIRECT reads and writes to files
4274 in the Lustre client mountpoint on 2.6 kernel clients.
4276 Severity : enhancement
4278 Description: parallel glimpse, setattr, statfs, punch, destroy requests
4279 Details : Sends glimpse, setattr, statfs, punch, destroy requests to OSTs in
4280 parallel, not waiting for response from every OST before sending
4281 a rpc to the next OST.
4286 Description: setattr vs write race when updating file timestamps
4287 Details : Client processes that update a file timestamp into the past
4288 right after writing to the file (e.g. tar) it is possible that
4289 the updated file modification time can be reset to the current
4290 time due to a race between processing the setattr and write RPC.
4292 Severity : enhancement
4294 Description: Bring 'lfs find' closer in line with regular Linux find.
4295 Details : lfs find util supports -atime, -mtime, -ctime, -maxdepth, -print,
4296 -print0 options and obtains all the needed info through the lustre
4299 Severity : enhancement
4301 Description: support up to 1024 configured devices on one node
4302 Details : change obd_dev array from statically allocated to dynamically
4303 allocated structs as they are first used to reduce memory usage
4308 Description: Flush dirty partially truncated pages during truncate
4309 Details : Immediatelly flush partially truncated pages in filter_setattr,
4310 this way we completely avoid having any pages in page cache on OST
4311 and can retire ugly workarounds during writes to flush such pages.
4316 Description: i_sem vs transaction deadlock in mds_obd_destroy during unlink.
4317 Details : protect inode from truncation within vfs_unlink() context
4318 just take a reference before calling vfs_unlink() and release it
4319 when parent's i_sem is free.
4324 Description: last_id value checked outside lock on OST caused LASSERT failure
4325 Details : If there were multiple MDS->OST object precreate requests in
4326 flight, it was possible that the OST's last object id was checked
4327 outside a lock and incorrectly tripped an assertion. Move checks
4328 inside locks, and discard old precreate requests.
4331 Frequency : always, if extents are used on OSTs
4333 Description: index ei_leaf_hi (48-bit extension) is not zeroed in extent index
4334 Details : OSTs using the extents format would not zero the high 16 bits of
4335 the index physical block number. This is not a problem for any
4336 OST filesystems smaller than 16TB, and no kernels support ext3
4337 filesystems larger than 16TB yet. This is fixed in 1.4.7 (all
4338 new/modified files) and can be fixed for existing filesystems
4339 with e2fsprogs-1.39-cfs1.
4344 Description: import connection selection may be incorrect if timer wraps
4345 Details : Using a 32-bit jiffies timer with HZ=1000 may cause backup
4346 import connections to be ignored if the 32-bit jiffies counter
4347 wraps. Use a 64-bit jiffies counter.
4350 Frequency : very large clusters immediately after boot
4352 Description: LNET request buffers exhausted under heavy short-term load
4353 Details : If a large number of client requests are generated on a service
4354 that has previously never seen so many requests it is possible
4355 that the request buffer growth cannot keep up with the spike in
4356 demand. Instead of dropping incoming requests, they are held in
4357 the LND until the RPC service can accept more requests.
4360 Frequency : Sometimes during replay
4362 Description: Assertion failure in ll_local_open after replay.
4363 Details : If replay happened on an open request reply before we were able
4364 to set replay handler, reply will become not swabbed tripping the
4365 assertion in ll_local_open. Now we set the handler right after
4366 recognising of open request
4369 Frequency : very rare
4371 Description: kernel reports "badness in vsnprintf"
4372 Details : Reading from the "recovery_status" /proc file in small chunks
4373 may cause a negative length in lprocfs_obd_rd_recovery_status()
4374 call to vsnprintf() (which is otherwise harmless). Exit early
4375 if there is no more space in the output buffer.
4377 Severity : enhancement
4379 Description: clear OBD RPC statistics by writing to them
4380 Details : It is now possible to clear the OBD RPC statistics by writing
4381 to the "stats" file.
4386 Description: Client mtime is not the same on different clients after utimes
4387 Details : In some cases, the client was using the utimes() syscall on
4388 a file cached on another node. The clients now validate the
4389 ctime from the MDS + OSTs to determine which one is right.
4394 Description: Inability to activate failout mode
4395 Details : lconf script incorrectly assumed that in pythong string's numeric
4396 value is used in comparisons.
4399 Frequency : always with multiple stripes per file
4401 Description: Inefficient object allocation for mutli-stripe files
4402 Details : When selecting which OSTs to stripe files over, for files with
4403 a stripe count that divides evenly into the number of OSTs,
4404 the MDS is always picking the same starting OST for each file.
4405 Return the OST selection heuristic to the original design.
4410 Description: mount failures may take full timeout to return an error
4411 Details : Under some heavy load conditions it is possible that a
4412 failed mount can wait for the full obd_timeout interval,
4413 possibly several minutes, before reporting an error.
4414 Instead return an error as soon as the status is known.
4416 Frequency : quota enabled and large files being deleted
4418 Description: releasing more than 4GB of quota at once hangs OST
4419 Details : If a user deletes more than 4GB of files on a single OST it
4420 will cause the OST to spin in an infinite loop. Release
4421 quota in < 4GB chunks, or use a 64-bit value for 1.4.7.1+.
4426 Description: statfs data retrieved from /proc may be stale or zero
4427 Details : When reading per-device statfs data from /proc, in the
4428 {kbytes,files}_{total,free,avail} files, it may appear
4429 as zero or be out of date.
4432 Frequency : systems with MD RAID1 external journal devices
4434 Description: lconf's call to blkid is confused by RAID1 journal devices
4435 Details : Use the "blkid -l" flag to locate the MD RAID device instead
4436 of returning all block devices that match the journal UUID.
4439 Frequency : always, for aggregate stripe size over 4GB
4441 Description: assertion fails when trying to use 4GB stripe size
4442 Details : Use "setstripe" to set stripe size over 4GB will fail the kernel,
4443 complaining "ASSERTION(lsm->lsm_xfersize != 0)"
4446 Frequency : always on ppc64
4448 Description: the first write on an ext3 filesystem with mballoc got stuck
4449 Details : ext3_mb_generate_buddy() uses find_next_bit() which does not
4450 perform endianness conversion.
4452 ------------------------------------------------------------------------------
4454 02-14-2006 Cluster File Systems, Inc. <info@clusterfs.com>
4456 * WIRE PROTOCOL CHANGE. This version of Lustre networking WILL NOT
4457 INTEROPERATE with older versions automatically. Please read the
4458 user documentation before upgrading any part of a live system.
4459 * WARNING: Lustre networking configuration changes are required with
4460 this release. See https://bugzilla.clusterfs.com/show_bug.cgi?id=10052
4463 * Support for kernels:
4464 2.6.9-22.0.2.EL (RHEL 4)
4465 2.6.5-7.244 (SLES 9)
4466 2.6.12.6 vanilla (kernel.org)
4469 Severity : enhancement
4470 Bugzilla : 7981/8208
4471 Description: Introduced Lustre Networking (LNET)
4472 Details : LNET is new networking infrastructure for Lustre, it includes
4473 a reorganized network configuration mode (see the user
4474 documentation for full details) as well as support for routing
4475 between different network fabrics. Lustre Networking Devices
4476 (LNDs) for the supported network fabrics have also been
4477 created for this new infrastructure.
4479 Severity : enhancement
4480 Description: Introduced Access control lists
4481 Details : clients can set ACLs on files and directories in order to have
4482 more fine-grained permissions than the standard Unix UGO+RWX.
4483 The MDS must be started with the "-o acl" mount option.
4485 Severity : enhancement
4486 Description: Introduced filesystem quotas
4487 Details : Administrators may now establish per-user quotas on the
4490 Severity : enhancement
4492 Description: Configuration change for the XT3
4493 The PTLLND is now used to run Lustre over Portals on the XT3
4494 The configure option(s) --with-cray-portals are no longer used.
4495 Rather --with-portals=<path-to-portals-includes> is used to
4496 enable building on the XT3. In addition to enable XT3 specific
4497 features the option --enable-cray-xt3 must be used.
4502 Description: Running on many-way SMP OSTs can trigger oops in llcd_send()
4503 Details : A race between allocating a new llcd and re-getting the llcd_lock
4504 allowed another thread to grab newly-allocated llcd.
4506 Severity : enhancement
4508 Description: 2.6 OST async journal commit and locking fix to improve performance
4509 Details : The filter_direct_io()+filter_commitrw_write() journal commits for
4510 2.6 kernels are now async as they already were in 2.4 kernels so
4511 that they can commit concurrently with the network bulk transfer.
4512 For block-allocated files the filter allocation semaphore is held
4513 to avoid filesystem fragmentation during allocation. BKL lock
4514 removed for 2.6 xattr operations where it is no longer needed.
4519 Description: lconf incorrectly determined whether two IP networks could talk
4520 Details : In some more complicated routing and multiple-network
4521 configurations, lconf will avoid trying to make a network
4522 connection to a disjoint part of the IP space. It was doing the
4523 math incorrectly for one set of cases.
4528 Description: Fix for potential infinite loop processing records in an llog.
4529 Details : If an llog record is corrupted/zeroed, it is possible to loop
4530 forever in llog_process(). Validate the llog record length
4531 and skip the remainder of the block on error.
4534 Frequency : occasional (liblustre only)
4536 Description: liblustre could not open files whose last component is a symlink
4537 Details : sysio_path_walk() would incorrectly pass the open intent to
4538 intermediate path components.
4541 Frequency : rare (liblustre only with non-standard tuning)
4542 Bugzilla : 7201 (7350)
4543 Description: Tuning the MDC DLM LRU size to zero triggers client LASSERT
4544 Details : llu_lookup_finish_locks() tries to set lock data on a lock
4545 after it has been released, only do this for referenced locks
4547 Severity : enhancement
4549 Description: specifying an (invalid) directory default stripe_size of -1
4550 would reset the directory default striping
4551 Details : stripe_size -1 was used internally to signal directory stripe
4552 removal, now use "all default" to signal dir stripe removal
4553 as a directory striping of "all default" is not useful
4556 Frequency : common for large clusters running liblustre clients
4558 Description: doing an ls when liblustre clients are running is slow
4559 Details : sending a glimpse AST to a liblustre client waits for every AST
4560 to time out, as liblustre clients will not respond. Since they
4561 cannot cache data we refresh the OST lock LVB from disk instead.
4563 Severity : enhancement
4565 Description: doing an ls at the same time as file IO can be slow
4566 Details : enqueue and other "small" requests can be blocked behind many
4567 large IO requests. Create a new OST IO portal for non-IO
4568 requests so they can be processed faster.
4571 Frequency : rare (only HPUX clients mounting unsupported re-exported NFS vol)
4573 Description: an HPUX NFS client would get -EACCESS when ftruncate()ing a newly
4574 created file with mode 000
4575 Details : the Linux NFS server relies on an MDS_OPEN_OWNEROVERRIDE hack to
4576 allow an ftruncate() as a non-root user to a file with mode 000.
4577 Lustre now respects this flag to disable mode checks when
4578 truncating a file owned by the user
4581 Frequency : liblustre-only, when liblustre client dies unexpectedly or becomes
4584 Description: Revoking locks from clients that went dead or catatonic might take
4586 Details : New lock flags FL_CANCEL_ON_BLOCK used by liblustre makes
4587 cancellation of such locks instant on servers without waiting for
4588 any reply from clients. Clients drops these locks when cancel
4589 notification from server is received without replying.
4592 Frequency : liblustre-only, when liblustre client dies or becomes busy
4594 Description: Doing ls on Linux clients can take a long time with active
4596 Details : Liblustre client cannot handle ASTs in timely manner, so avoid
4597 granting such locks to it in the first place if possible. Locks
4598 are taken by proxy on the OST during the read or write and
4599 dropped immediately afterward. Add connect flags handling, do
4600 not grant locks to liblustre clients for glimpse ASTs.
4602 Severity : enhancement
4604 Description: Improve read-ahead algorithm to avoid excessive IO for random reads
4605 Details : Existing read-ahead algorithm is tuned for the case of streamlined
4606 sequential reads and behaves badly with applications doing random
4607 reads. Improve it by reading ahead at least read region, and
4608 avoiding excessive large RPC for small reads.
4610 Severity : enhancement
4612 Description: Creating more than 1000 files for a single job may cause a load
4613 imbalance on the OSTs if there are also a large number of OSTs.
4614 Details : qos_prep_create() uses an OST index reseed value that is an
4615 even multiple of the number of available OSTs so that if the
4616 reseed happens in the middle of the object allocation it will
4617 still utilize the OSTs as uniformly as possible.
4622 Description: OST or MDS may oops in ping_evictor_main()
4623 Details : ping_evictor_main() drops obd_dev_lock if deleting a stale export
4624 but doesn't restart at beginning of obd_exports_timed list
4627 Severity : enhancement
4629 Description: improve by-nid export eviction on the MDS and OST
4630 Details : allow multiple exports with the same NID to be evicted at one
4631 time without re-searching the exports list.
4634 Frequency : rare, only with supplementary groups enabled on SMP 2.6 kernels
4636 Description: MDS may oops in groups_free()
4637 Details : in rare race conditions a newly allocated group_info struct is
4638 freed again, and this can be NULL. The 2.4 compatibility code
4639 for groups_free() checked for a NULL pointer, but 2.6 did not.
4642 Frequency : common for liblustre clients doing little filesystem IO
4643 Bugzilla : 9352, 7313
4644 Description: server may evict liblustre clients accessing contended locks
4645 Details : if a client is granted a lock or receives a completion AST
4646 with a blocking AST already set it would not reply to the AST
4647 for LDLM_FL_CANCEL_ON_BLOCK locks. It now replies to such ASTs.
4650 Frequency : lfs setstripe, only systems with more than 160 OSTs
4652 Description: unable to set striping with a starting offset beyond OST 160
4653 Details : llapi_create_file() incorrectly limited the starting stripe
4654 index to the maximum single-file stripe count.
4657 Frequency : LDAP users only
4659 Description: lconf did not handle in-kernel recovery with LDAP properly
4660 Details : lconf/LustreDB get_refs() is searching the wrong namespace
4662 Severity : enhancement
4664 Description: bind OST threads to NUMA nodes to improve performance
4665 Details : all OST threads are uniformly bound to CPUs on a single NUMA
4666 node and do their allocations there to localize memory access
4668 Severity : enhancement
4670 Description: llmount can determine client NID directly from Myrinet (GM)
4671 Details : the client NID code from gmnalnid was moved directly into
4672 llmount, removing the need to use this or specifying the
4673 client NID explicitly when mounting GM clients with zeroconf
4676 Frequency : if client is started with down MDS
4678 Description: if client is started with down MDS mount hangs in ptlrpc_queue_wait
4679 Details : Having an LWI_INTR() wait event (interruptible, but no timeout)
4680 will wait indefinitely in ptlrpc_queue_wait->l_wait_event() after
4681 ptlrpc_import_delayed_req() because we didn't check if the
4682 request was interrupted, and we also didn't break out of the
4683 event loop if there was no timeout
4688 Description: data loss during non-page-aligned writes to a single file from
4689 both multiple nodes and multiple threads on one node at same time
4690 Details : updates to KMS and lsm weren't protected by common lock. Resulting
4691 inconsistency led to false short-reads, that were cached and later
4692 used by ->prepare_write() to fill in partially written page,
4693 leading to data loss.
4696 Frequency : always, if lconf --abort_recovery used
4698 Description: lconf --abort_recovery fails with 'Operation not supported'
4699 Details : lconf was attempting to abort recovery on the MDT device and not
4702 Severity : enhancement
4704 Description: remove cleanup logs
4705 Details : replace lconf-generated cleanup logs with lustre internal
4706 cleanup routines. Eliminates the need for client-cleanup and
4709 Severity : enhancement
4711 Description: add support for EAs (user and system) on lustre filesystems
4712 Details : it is now possible to store extended attributes in the Lustre
4713 client filesystem, and with the user_xattr mount option it
4714 is possible to allow users to store EAs on their files also
4716 Severity : enhancement
4718 Description: Add possibility (config option) to show minimal available OST free
4720 Details : When compiled with --enable-mindf configure option, statfs(2)
4721 (and so, df) will return least minimal free space available from
4722 all OSTs as amount of free space on FS, instead of summary of
4723 free spaces of all OSTs.
4725 Severity : enhancement
4727 Description: do not expand extent locks acquired on OST-side
4728 Details : Modify ldlm_extent_policy() to not expand local locks, acquired
4729 by server: they are not cached anyway.
4732 Frequency : when mmap is used/binaries executed from Lustre
4734 Description: Unmmap pages before throwing them away from read cache.
4735 Details : llap_shrink cache now attempts to unmap pages before discarding
4736 them (if unmapping failed - do not discard). SLES9 kernel has
4737 extra checks that trigger if this unmapping is not done first.
4742 Description: lconf didn't resolve symlinks before checking to see whether a
4743 given mountpoint was already in use
4746 Frequency : when migrating failover services
4747 Bugzilla : 6395, 9514
4748 Description: When migrating a subset of services from a node (e.g. failback
4749 from a failover service node) the remaining services would
4750 time out and evict clients.
4751 Details : lconf --force (implied by --failover) sets the global obd_timeout
4752 to 5 seconds in order to quickly disconnect, but this caused
4753 other RPCs to time out too quickly. Do not change the global
4754 obd_timeout for force cleanup, only set it for DISCONNECT RPCs.
4756 Severity : enhancement
4757 Frequency : if MDS is started with down OST
4758 Bugzilla : 9439,5706
4759 Description: Allow startup/shutdown of an MDS without depending on the
4760 availability of the OSTs.
4761 Details : Asynchronously call mds_lov_synchronize during MDS startup.
4762 Add appropriate locking and lov-osc refcounts for safe
4763 cleaning. Add osc abort_inflight calls in case the
4764 synchronize never started.
4767 Frequency : occasional (Cray XT3 only)
4769 Description: root not authorized to access files in CRAY_PORTALS environment
4770 Details : The client process capabilities were not honoured on the MDS in
4771 a CRAY_PORTALS/CRAY_XT3 environment. If the file had previously
4772 been accessed by an authorized user then root was able to access
4773 the file on the local client also. The root user capabilities
4774 are now allowed on the MDS, as this environment has secure UID.
4777 Frequency : occasional
4779 Description: ldiskfs "too long searching" message happens too often
4780 Details : A debugging message (otherwise harmless) prints too often on
4781 the OST console. This has been reduced to only happen when
4782 there are fragmentation problems on the filesystem.
4787 Description: Division by zero in statfs when all OSCs are inactive
4788 Details : lov_get_stripecnt() returns zero due to incorrect order of checks,
4789 lov_statfs divides by value returned by lov_get_stripecnt().
4793 Bugzilla : 9489, 3273
4794 Description: First write from each client to each OST was only 4kB in size,
4795 to initialize client writeback cache, which caused sub-optimal
4796 RPCs and poor layout on disk for the first written file.
4797 Details : Clients now request an initial cache grant at (re)connect time
4798 and so that they can start streaming writes to the cache right
4799 away and always do full-sized RPCs if there is enough data.
4800 If the OST is rebooted the client also re-establishes its grant
4801 so that client cached writes will be honoured under the grant.
4806 Description: Slow ls (and stat(2) syscall) on files residing on IO-loaded OSTs
4807 Details : Now I/O RPCs go to different portal number and (presumably) fast
4808 lock requests (and glimses) and other RPCs get their own service
4809 threads pool that should be able to service those RPCs
4812 Severity : enhancement
4814 Description: Ability to exchange lustre version between client and servers and
4815 issue warnings at client side if client is too old. Also for
4816 liblustre clients there is ability to refuse connection of too old
4818 Details : New 'version' field is added to connect data structure that is
4819 filled with version info. That info is later checked by server and
4823 Frequency : rare, liblustre only.
4824 Bugzilla : 9296, 9581
4825 Description: Two simultaneous writes from liblustre at offset within same page
4826 might proceed at the same time overwriting eachother with stale
4828 Details : I/O lock withing llu_file_prwv was released too early, before data
4829 actually was hitting the wire. Extended lock-holding time until
4830 server acknowledges receiving data.
4833 Frequency : extremely rare. Never observed in practice.
4835 Description: avoid generating lustre_handle cookie of 0.
4836 Details : class_handle_hash() generates handle cookies by incrementing
4837 global counter, and can hit 0 occasionaly (this is unlikely, but
4838 not impossible, because initial value of cookie counter is
4839 selected randonly). Value of 0 is used as a sentinel meaning
4840 "unassigned handle" --- avoid it. Also coalesce two critical
4841 sections in this function into one.
4843 Severity : enhancement
4845 Description: allow liblustre clients to delegate truncate locking to OST
4846 Details : To avoid overhead of locking, liblustre client instructs OST to
4847 take extent lock in ost_punch() on client's behalf. New connection
4848 flag is added to handle backward compatibility.
4850 Severity : enhancement
4851 Bugzilla : 4928, 7341, 9758
4852 Description: allow number of OST service threads to be specified
4853 Details : a module parameter allows the number of OST service threads
4854 to be specified via "options ost ost_num_threads={N}" in the
4855 OSS's /etc/modules.conf or /etc/modprobe.conf.
4859 Bugzilla : 6146, 9635, 9895
4860 Description: servers crash with bad pointer in target_handle_connect()
4861 Details : In rare cases when a client is reconnecting it was possible that
4862 the connection request was the last reference for that export.
4863 We would temporarily drop the export reference and get a new
4864 one, but this may have been the last reference and the export
4865 was just destroyed. Get new reference before dropping old one.
4867 Severity : enhancement
4868 Frequency : if client is started with failover MDS
4870 Description: Allow multiple MDS hostnames in the mount command
4871 Details : Try to read the configuration from all specified MDS
4872 hostnames during a client mount in case the "primary"
4875 Severity : enhancement
4877 Description: Stop sending data to evicted clients as soon as possible.
4878 Details : Check if the client we are about to send or are sending data to
4879 was evicted already. (Check is done every second of waiting,
4880 for which l_wait_event interface was extended to allow checking
4881 of exit condition at specified intervals).
4884 Frequency : rare, normally only when NFS exporting is done from client
4886 Description: 'bad disk LOV MAGIC: 0x00000000' error when chown'ing files
4888 Details : Make mds_get_md() recognise empty md case and set lmm size to 0.
4891 Frequency : always, if srand() is called before liblustre initialization
4893 Description: Liblustre uses system PRNG disturbing its usage by user application
4894 Details : Introduce internal to lustre fast and high-quality PRNG for
4895 lustre usage and make liblustre and some other places in generic
4896 lustre code to use it.
4898 Severity : enhancement
4899 Bugzilla : 9477, 9557, 9870
4900 Description: Verify that the MDS configuration logs are updated when xml is
4901 Details : Check if the .xml configuration logs are newer than the config
4902 logs stored on the MDS and report an error if this is the case.
4903 Request --write-conf, or allow starting with --old_conf.
4905 Severity : enhancement
4907 Description: Handle symlinks in the path when checking if Lustre is mounted.
4908 Details : Resolve intermediate symlinks when checking if a client has
4909 mounted a filesystem to avoid duplicate client mounts.
4914 Description: lconf can hit an error exception but still return success.
4915 Details : The lconf command catches the Command error exception at the top
4916 level script context and will exit with the associated exit
4917 status, but doesn't ensure that this exit status is non-zero.
4922 Description: failure of ptlrpc thread startup can cause oops
4923 Details : Starting a ptlrpc service thread can fail if there are a large
4924 number of threads or the server memory is very fragmented.
4925 Handle this without oopsing.
4928 Frequency : always, only if liblustre and non-default acceptor port was used
4930 Description: liblustre cannot connect to servers with non-default acceptor port
4931 Details : tcpnal_set_default_params() was not called and was therefore
4932 ignoring the environment varaible TCPNAL_PORT, as well as other
4933 TCPNAL_ environment variables
4938 Description: two objects could be created on the same OST for a single file
4939 Details : If an OST is down, in some cases it was possible to create two
4940 objects on a single OST for a single file. No problems other
4941 than potential performance impact and spurious error messages.
4945 Bugzilla : 5681, 9562
4946 Description: Client may oops in ll_unhash_aliases
4947 Details : Client dcache may become inconsistent in race condition.
4948 In some cases "getcwd" can fail if the current directory is
4954 Description: Inode refcounting problems in NFS export code
4955 Details : link_raw functions used to call d_instantiate without obtaining
4956 extra inode reference first.
4960 Bugzilla : 9942, 9903
4961 Description: Referencing freed requests leading to crash, memleaks with NFS.
4962 Details : We used to require that call to ll_revalidate_it was always
4963 followed by ll_lookup_it. Also with revalidate_special() it is
4964 possible to call ll_revalidate_it() twice for the same dentry
4965 even if first occurence returned success. This fix changes semantic
4966 between DISP_ENQ_COMPLETE disposition flag to mean there is extra
4967 reference on a request referred from the intent.
4968 ll_intent_release() then releases such a request.
4971 Frequency : rare, normally benchmark loads only
4973 Description: unlinked inodes were kept in memory on the client
4974 Details : If a client is repeatedly creating and unlinking files it
4975 can accumulate a lot of stale inodes in the inode slab cache.
4976 If there is no other client load running this can cause the
4977 client node to run out of memory. Instead flush old inodes
4978 from client cache that have the same inode number as a new inode.
4981 Frequency : SLES9 2.6.5 kernel and long filenames only
4982 Bugzilla : 9969, 10379
4983 Description: utime reports stale NFS file handle
4984 Details : SLES9 uses out-of-dentry names in some cases, which confused
4985 the lustre dentry revalidation. Change it to always use the
4989 Frequency : rare, unless heavy write-truncate concurrency is continuous
4990 Bugzilla : 4180, 6984, 7171, 9963, 9331
4991 Description: OST becomes very slow and/or deadlocked during object unlink
4992 Details : filter_destroy() was holding onto the parent directory lock
4993 while truncating+unlinking objects. For very large objects this
4994 may block other threads for a long time and slow overall OST
4995 responsiveness. It may also be possible to get a lock ordering
4996 deadlock in this case, or run out of journal credits because of
4997 the combined truncate+unlink. Solution is to do object truncate
4998 first in one transaction without parent lock, and then do the
4999 final unlink in a new transaction with the parent lock. This
5000 reduces the lock hold time dramatically.
5003 Frequency : rare, 2.4 kernels only
5005 Description: MDS or OST cleanup may trip kernel BUG when dropping kernel lock
5006 Details : mds_cleanup() and filter_cleanup() need to drop the kernel lock
5007 before unmounting their filesystem in order to avoid deadlock.
5008 The kernel_locked() function in 2.4 kernels only checks whether
5009 the kernel lock is held, not whether it is this process that is
5010 holding it as 2.6 kernels do.
5015 Description: MDS or OST may oops/LBUG if a client is connecting multiple times
5016 Details : The client ptlrpc code may be trying to reconnect to a down
5017 server before a previous connection attempt has timed out.
5018 Increase the reconnect interval to be longer than the connection
5019 timeout interval to avoid sending duplicate connections to
5023 Frequency : echo_client brw_test command
5025 Description: fix echo_client to work with OST preallocated code
5026 Details : OST preallocation code (5137) didn't take echo_client IO path
5027 into account: echo_client calls filter methods outside of any
5028 OST thread and, hence, there is no per-thread preallocated
5029 pages and buffers to use. Solution: hijack pga pages for IO. As
5030 a byproduct, this avoids unnecessary data copying.
5034 Bugzilla : 3555, 5962, 6025, 6155, 6296, 9574
5035 Description: Client can oops in mdc_commit_close() after open replay
5036 Details : It was possible for the MDS to return an open request with no
5037 transaction number in mds_finish_transno() if the client was
5038 evicted, but without actually returning an error. Clients
5039 would later try to replay that open and may trip an assertion
5040 Simplify the client close codepath, and always return an error
5041 from the MDS in case the open is not successful.
5044 Frequency : rare, 2.6 OSTs only
5046 Description: OST may deadlock under high load on fragmented files
5047 Details : If there was a heavy load and highly-fragmented OST filesystems
5048 it was possible to have all the OST threads deadlock waiting on
5049 allocation of biovecs, because the biovecs were not released
5050 until the entire RPC IO was completed. Instead, release biovecs
5051 as soon as they are complete to ensure forward IO progress.
5053 Severity : enhancement
5055 Description: Support for specifying external journal device at mount
5056 Details : If an OST or MDS device is formatted with an external journal
5057 device, this device major/minor is stored in the ext3 superblock
5058 and may not be valid for failover. Allow detecting and
5059 specifying the external journal at mount time.
5064 Description: Mounting an MDS with pending unlinked files may cause oops
5065 Details : target_finish_recovery() calls mds_postrecov() which returned
5066 the number of orphans unlinked. mds_lov_connect->mds_postsetup()
5067 considers this an error and immediately begins cleaning up the
5068 lov, just after starting the mds_lov process
5070 Severity : enhancement
5072 Description: Implement 'lfs df' to report actual free space on per-OST basis
5073 Details : Add sub-command 'df' on 'lfs' to report the disk space usage of
5074 MDS/OSDs. Usage: lfs df [-i][-h]. Command Options: '-i' to report
5075 usage of objects; '-h' to report in human readable format.
5077 ------------------------------------------------------------------------------
5079 08-26-2005 Cluster File Systems, Inc. <info@clusterfs.com>
5086 Description: Mounting an ldiskfs file system with mballoc may crash OST node.
5087 Details : ldiskfs mballoc code may reference an uninitialized buddy struct
5088 at startup during orphan unlinking. Instead, skip buddy update
5089 before setup, as it will be regenerated after recovery is complete.
5094 Description: If an OST is inactive, its locks might reference stale inodes.
5095 Details : lov_change_cbdata() must iterate over all namespaces, even if
5096 they are inactive to clear inode references from the lock.
5098 Severity : enhancement
5099 Frequency : occasional, if non-standard max_dirty_mb used
5101 Description: Client will block write RPCs if not enough grant
5102 Details : If a client has max_dirty_mb smaller than max_rpcs_in_flight,
5103 then the client will block writes while waiting for another RPC
5104 to complete instead of consuming its dirty limit. With change
5105 we get improved performance when max_dirty_mb is small.
5107 Severity : enhancement
5108 Bugzilla : 3389, 6253
5109 Description: Add support for supplementary groups on the MDS.
5110 Details : The MDS has an upcall /proc/fs/lustre/mds/{mds}/group_upcall
5111 (set to /usr/sbin/l_getgroups if enabled) which will do MDS-side
5112 lookups for user supplementary groups into a cache.
5116 Description: O_CREAT|O_EXCL open flags in liblustre always return -EEXIST
5117 Details : Make libsysio to not enforce O_EXCL by clearing the flag,
5118 for liblustre O_EXCL is enforced by MDS.
5122 Description: readdir never returns NULL in liblustre.
5123 Details : Corrected llu_iop_getdirentries logic, to return offset of next
5124 dentry in struct dirent.
5128 Frequency : liblustre only, depends on application IO pattern
5129 Description: liblustre clients evicted if not contacting servers
5130 Details : Don't put liblustre clients into the ping_evictor list, so
5131 they will not be evicted by the pinger ever.
5133 Severity : enhancement
5135 Description: Add ability to evict clients by NID from MDS.
5136 Details : By echoing "nid:$NID" string into
5137 /proc/fs/lustre/mds/.../evict_client client with nid that equals to
5138 $NID would be instantly evicted from this MDS and from all active
5139 OSTs connected to it.
5143 Description: Do not query file size twice, somewhat slowing stat(2) calls.
5144 Details : lookup_it_finish() used to query file size from OSTs that was not
5149 Description: service threads change working directory to that of init
5150 Details : Starting lustre service threads may pin the working directory
5151 of the parent thread, making that filesystem busy. Threads
5152 now change to the working directory of init to avoid this.
5156 Frequency : during shutdown only
5157 Description: shutdown with a failed MDS or OST can cause unmount to hang
5158 Details : Don't resend DISCONNECT messages in ptlrpc_disconnect_import()
5163 Frequency : 2.6 only
5164 Description: chmod/chown may include an extra supplementary group
5165 Details : ll{,u}_mdc_pack_op_data() does not properly initialize the
5166 supplementary group and if none is specified this is used.
5169 Bugzilla : 5479 (6816)
5171 Description: Racing open + rm can assert client in mdc_set_open_replay_data()
5172 Details : If lookup is in progress on a file that is unlinked we might try
5173 to revalidate the inode and fail in revalidate after lookup is
5174 complete and ll_file_open() enqueues the open again but
5175 it_open_error() was not checking DISP_OPEN_OPEN errors correctly.
5178 Frequency : always, if lconf --abort_recovery used
5180 Description: lconf --abort_recovery fails with 'Operation not supported'
5181 Details : lconf was attempting to abort recovery on the MDT device and not
5184 ------------------------------------------------------------------------------
5186 2005-08-08 Cluster File Systems, Inc. <info@clusterfs.com>
5191 Frequency : rare (only unsupported configurations with a node running as an
5193 Bugzilla : 6514, 5137
5194 Description: Mounting a Lustre file system on a node running as an OST could
5196 Details : OSTs now preallocates memory needed to write out data at
5197 startup, instead of when needed, to avoid having to
5198 allocate memory in possibly low memory situations.
5199 Specifically, if the file system is mounted on on OST,
5200 memory pressure could force it to try to write out data,
5201 which it needed to allocate memory to do. Due to the low
5202 memory, it would be unable to do so and the node would
5203 become unresponsive.
5205 Severity : enhancement
5207 Description: Addition of lconf --service command line option
5208 Details : lconf now accepts a '--service <arg>' option, which is
5209 shorthand for 'lconf --group <arg> --select <arg>=<hostname>'
5211 Severity : enhancement
5213 Description: Failover mode is now the default for OSTs.
5214 Details : By default, OSTs will now run in failover mode. To return to
5215 the old behaviour, add '--failout' to the lmc line for OSTs.
5217 Severity : enhancement
5219 Description: Health checks are now provided for MDS and OSTs
5220 Details : Additional detailed health check information on MSD and OSTs
5221 is now provided through the procfs health_check value.
5224 Frequency : occasional, depends on IO load
5226 Description: Disk fragmentation on the OSTs could eventually cause slowdowns
5227 after numerous create/delete cycles
5228 Details : The ext3 inode allocation policy would not allocate new inodes
5229 very well on the OSTs because there are no new directories
5230 being created. Instead we look for groups with free space if
5231 the parent directories are nearly full.
5236 Description: Network or server problems during mount may cause partially
5237 mounted clients instead of returning an error.
5238 Details : The config llog parsing code may overwrite the error return
5239 code during mount error handling, returning success instead
5245 Description: MDS can fail to allocate large reply buffers
5246 Details : After long uptimes the MDS can fail to allocate large reply
5247 buffers (e.g. zconf client mount config records) due to memory
5248 fragmentation or consumption by the buffer cache. Preallocate
5249 some large reply buffers so that these replies can be sent even
5250 under memory pressure.
5254 Frequency : rare (liblustre)
5255 Description: fsx running with liblustre complained that using truncate() to
5256 extend the file doesn't work. This patch corrects that issue.
5257 Details : This is the liblustre equivalent of the fix for bug 6196. Fixes
5258 ATTR_SIZE and lsm use in llu_setattr_raw.
5262 Frequency : rare, only 2.6 kernels
5263 Description: Unusual file access patterns on the MDS may result in inode
5264 data being lost in very rare circumstances.
5265 Details : Bad interaction between the ea-in-inode patch and the "no-read"
5266 code in the 2.6 kernel caused the inode and/or EA data not to
5267 be read from disk, causing single-file corruption.
5271 Frequency : rare, only 2.6 filesystems using extents
5272 Description: Heavy concurrent write and delete load may cause data corruption.
5273 Details : It was possible under high-load situations to have an extent
5274 metadata block in the block device cache from a just-unlinked
5275 file overwrite a newly-allocated data block. We now unmap any
5276 metadata buffers that alias just-allocated data blocks.
5280 Frequency : filesystems with default stripe_count larger than 77
5281 Description: lconf+mke2fs fail when formatting filesystem with > 77 stripes
5282 Details : lconf specifies an inode size of 4096 bytes when the default
5283 stripe_count is larger than 77. This conflicts with the default
5284 inode density of 1 per 4096 bytes. Allocate smaller inodes in
5285 this case to avoid pinning too much memory for large EAs.
5287 ------------------------------------------------------------------------------
5289 2005-07-07 Cluster File Systems, Inc. <info@clusterfs.com>
5294 Frequency : rare (extremely heavy IO load with hundreds of clients)
5296 Description: Client is evicted, gets IO error writing to file
5297 Details : lock ordering changes for bug 5492 reintroduced bug 3267 and
5298 caused clients to be evicted for AST timeouts. The fixes in
5299 bug 5192 mean we no longer need to have such short AST timeouts
5300 so ldlm_timeout has been increased.
5303 Frequency : occasional during --force or --failover shutdown under load
5304 Bugzilla : 5949, 4834
5305 Description: Server oops/LBUG if stopped with --force or --failover under load
5306 Details : a collection of import/export refcount and cleanup ordering
5307 issues fixed for safer force cleanup
5310 Frequency : only filesystems larger than 120 OSTs
5311 Bugzilla : 5990, 6223
5312 Description: lfs getstripe would oops on a very large filesystem
5313 Details : lov_getconfig used kfree on vmalloc'd memory
5316 Frequency : only filesystems exporting via NFS to Solaris 10 clients
5317 Bugzilla : 6242, 6243
5318 Description: reading from files that had been truncated to a non-zero size
5319 but never opened returned no data
5320 Details : ll_file_read() reads zeros from no-object files to EOF
5325 Description: A bug in MDS/OSS recovery could cause the OSS to fail an assertion
5326 Details : There's little harm in aborting MDS/OSS recovery and letting it
5327 try again, so I removed the LASSERT and return an error instead.
5329 Severity : enhancement
5331 Description: New debugging infrastructure for tracking down data corruption
5332 Details : The I/O checksum code was replaced to: (a) control it at runtime,
5333 (b) cover more of the client-side code path, and (c) try to narrow
5334 down where problems occurred
5338 Bugzilla : 3819, 4364, 4397, 6313
5339 Description: Racing close and eviction MDS could cause assertion in mds_close
5340 Details : It was possible to get multiple mfd references during close and
5341 client eviction, leading to one thread referencing a freed mfd.
5343 Severity: : enhancement
5344 Bugzilla : 3262, 6359
5345 Description: Attempts to reconnect to servers are now more aggressive.
5346 Details : This builds on the enhanced upcall-less recovery that was added
5347 in 1.4.2. When trying to reconnect to servers, clients will
5348 now try each server in the failover group every 10 seconds. By
5349 default, clients would previously try one server every 25 seconds.
5354 Description: After recovery, certain operations trigger a failed
5355 assertion on a client.
5356 Details : Failing over an mds, using lconf -d --failover, while a
5357 client was doing a readdir() call would cause the client to
5358 LBUG after recovery completed and the readdir() was resent.
5360 Severity : enhancement
5362 Description: Default groups are now added by lconf
5363 Details : You can now run lconf --group <servicename> without having to
5364 manually add groups with lmc.
5367 Frequency : occasional
5369 Description: Nodes with an elan id of 0 trigger a failed assertion
5372 Frequency : always when accessing e.g. tty/console device nodes
5374 Description: tty and some other devices nodes cannot be used on lustre
5375 Details : file's private_data field is used by device data and lustre
5376 values in there got lost. New field was added to struct file to
5377 store fs-specific private data.
5380 Frequency : when exporting Lustre via NFS
5382 Description: NFSD failed occasionally when looking up a path component
5383 Details : NFSD is looking up ".." which was broken in ext3 directories
5384 that had grown large enough to become hashed.
5387 Frequency : Clusters with multiple interfaces not on the same subnet
5389 Description: Nodes will repeatedly try to reconnect to an interface which it
5390 cannot reach and report an error to the log.
5391 Details : Extra peer list entries will be created by lconf with some peers
5392 unreachable. lconf now validates the peer before adding it.
5395 Frequency : Only if a default stripe is set on the filesystem root.
5397 Description: Setting a default stripe on the filesystem root prevented the
5398 filesystem from being remounted.
5399 Details : The client was sending extra request flags in the root getattr
5400 request and did not allocate a reply buffer for the dir EA.
5403 Frequency : occasional, higher if lots of files are accessed by one client
5404 Bugzilla : 6159, 6097
5405 Description: Client trips assertion regarding lsm mismatch/magic
5406 Details : While revalidating inodes the VFS looks up inodes with ifind()
5407 and in rare cases can find an inode that is being freed.
5408 The ll_test_inode() code will free the lsm during ifind()
5409 when it finds an existing inode and then the VFS later attaches
5410 this free lsm to a new inode.
5414 Bugzilla : 6422, 7030
5415 Description: MDS deadlock between mkdir and client eviction
5416 Details : Creating a new file via mkdir or mknod (starting a transaction
5417 and getting the ns lock) can deadlock with client eviction
5418 (gets ns lock and trying to finish a synchronous transaction).
5421 Frequency : occasional
5422 Description: While starting a server, the fsfilt_ext3 module could not be
5424 Details : CFS's improved ext3 filesystem is named ldiskfs for 2.6
5425 kernels. Previously, lconf would still use the ext3 name
5426 when trying to load modules. Now, it will correctly use
5427 ext3 on 2.4 and ldiskfs on 2.6.
5429 Severity : enhancement
5430 Description: The default stripe count has been changed to 1
5431 Details : The interpretation of the default stripe count (0, to lfs
5432 or lmc) has been changed to mean striping across a single
5433 OST, rather than all available. For general usage we have
5434 found a stripe count of 1 or 2 works best.
5436 Severity : enhancement
5437 Description: Add support for compiling against Cray portals.
5438 Details : Conditional compiling for some areas that are different
5442 Frequency : occasional
5443 Bugzilla : 6409, 6834
5444 Description: Creating files with an explicit stripe count may lead to
5445 a failed assertion on the MDS
5446 Details : If some OSTs are full or unavailable, creating files may
5447 trigger a failed assertion on the MDS. Now, Lustre will
5448 try to use other servers or return an error to the
5452 Frequency : occasional
5454 Description: Multiple concurrent overlapping read+write on multiple SMP nodes
5455 caused lock timeout during readahead (since 1.4.2).
5456 Details : Processes doing readahead might match a lock that hasn't been
5457 granted yet if there are overlapping and conflicting lock
5458 requests. The readahead process waits on ungranted lock
5459 (original lock is CBPENDING), while OST waits for that process
5460 to cancel CBPENDING read lock and eventually evicts client.
5462 Severity : enhancement
5464 Description: Initial enabling of flock support for clients
5465 Details : Implements fcntl advisory locking and file status functions.
5466 This feature is provided as an optional mount flag (default
5467 off), and is NOT CURRENTLY SUPPORTED. Not all types of record
5468 locking are implemented yet, and those that are are not guaranteed
5469 to be completely correct in production environments.
5470 mount -t lustre -o [flock|noflock] ...
5473 Frequency : occasional
5475 Description: OSTs running 2.4 kernels but with extents enabled might trip an
5476 assertion in the ext3 JBD (journaling) layer.
5477 Details : The b_committed_data struct is protected by the big kernel lock
5478 in 2.4 kernels, serializing journal_commit_transaction() and
5479 ext3_get_block_handle->ext3_new_block->find_next_usable_block()
5480 access to this struct. In 2.6 kernels there is finer grained
5481 locking to improve SMP performance of the JBD layer.
5485 Description: Changes the "SCSI I/O Stats" kernel patch to default to "enabled"
5490 Description: merge and cleanup kernel patches.
5493 -----------------------------------------------------------------------------
5495 2005-05-05 Cluster File Systems, Inc. <info@clusterfs.com>
5497 NOTE: Lustre 1.4.2 uses an incompatible network protocol than previous
5498 versions of Lustre. Please update all servers and clients to
5499 version 1.4.2 or later at the same time. You must also run
5500 "lconf --write-conf {config}.xml" on the MDS while it is stopped
5501 to update the configuration logs.
5503 - fix for HPUX NFS client breakage when NFS exporting Lustre (5781)
5504 - mdc_enqueue does not need max_mds_easize request buffer on send (5707)
5505 - swab llog records of type '0' so we get proper header size/idx (5861)
5506 - send llog cancel req to DLM cancel portal instead of cb portal (5515)
5507 - fix rename of one directory over another leaking an inode (5953)
5508 - avoid SetPageDirty on 2.6 (5981)
5509 - don't re-add just-being-destroyed locks to the waiting list (5653)
5510 - when creating new directories, inherit the parent's custom
5511 striping settings if present parent (3048)
5512 - flush buffers from cache before direct IO in 2.6 obdfilter (4982)
5513 - don't hold i_size_sem in ll_nopage() and ll_ap_refresh_count (6077)
5514 - don't hold client locks on temporary worklist from l_lru (5666)
5515 - handle IO errors in 2.6 obdfilter bio completion routine (6046)
5516 - automatically evict dead clients (5921)
5517 - Update file size properly in create+truncate+fstat case (6196)
5518 - Do not unhash mountpoint dentries, do not allow removal of
5520 - Avoid lock ordering deadlock issue with write/truncate (6203,5654)
5521 - reserve enough journal credits in fsfilt_start_log for setattr (4554)
5522 - ldlm_enqueue freed-export error path would always LBUG (6149,6184)
5523 - don't reference lr_lvb_data until after we hold lr_lvb_sem (6170)
5524 - don't overwrite last_rcvd if there is a *_client_add() error (6086)
5525 - Correctly handle reads of files with no objects (6243)
5526 - lctl recover will also mark a device active if deactivate used (5933)
5528 - by default create 1 inode per 4kB space on MDS, per 16kB on OSTs
5529 - allow --write-conf on an MDS with different nettype than client (5619)
5530 - don't write config llogs to MDS for mounts not from that MDS (5617)
5531 - lconf should create multiple TCP connections from a client (5201)
5532 - init scripts are now turned off by default; run chkconfig --on
5533 lustre and chkconfig --on lustrefs to use them
5534 - upcalls are no longer needed for clients to recover to failover
5536 - add --abort-recovery option to lconf to abort recovery on device
5538 - add support for an arbitrary number of OSTs (3026)
5539 - Quota support protocol changes.
5540 - forward compatibility changes to wire structs (6007)
5541 - rmmod NALs that might be loaded because of /etc/modules.conf (6133)
5542 - support for mountfsoptions and clientoptions to the Lustre LDAP (5873)
5543 - improved "lustre status" script
5544 - initialize blocksize for non-regular files (6062)
5545 - added --disable-server and --disable-client configure options (5782)
5546 - introduce a lookup cache for lconf to avoid repeated DB scans (6204)
5547 - Vanilla 2.4.29 support
5548 - increase maximum number of obd devices to 520 (6242)
5549 - remove the tcp-zero-copy patch from the suse-2.4 series (5902)
5550 - Quadrics Elan drivers are now included for the RHEL 3 2.4.21 and
5551 SLES 9 2.6.5 kernels
5552 - limit stripes per file to 160 (the maximum EA size) (6093)
5554 2005-03-22 Cluster File Systems, Inc. <info@clusterfs.com>
5557 - don't LASSERT in ll_release on NULL lld with NFS export (4655, 5760)
5558 - hold NS lock when calling handle_ast_error->del_waiting_lock (5746)
5559 - fix setattr mtime regression from lovcleanup merge (4829, 5669)
5560 - workaround for 2.6 crash in ll_unhash_aliases (5687, 5210)
5561 - small ext3 extents cleanups and fixes (5733)
5562 - improved mballoc code, several small races and bugs fixed (5733, 5638)
5563 - kernel version 43 - fix remove_suid bugs in both 2.4 and 2.6 (5695)
5564 - avoid needless client->OST connect, fix handle mismatch (5317)
5565 - fix DLM error path that led to out-of-sync client, long delays (5779)
5566 - support common vfs-enforced mount options (nodev,nosuid,noexec) (5637)
5567 - fix several locking issues related to i_size (5492,5624,5654,5672)
5568 - don't move pending lock onto export if it is already evicted (5683)
5569 - fix kernel oops when creating .foo in unlinked directory (5548)
5570 - fix deadlock in obdfilter statistics vs. object create (5811)
5571 - use time_{before,after} to avoid timer jiffies wrap (5882)
5572 - shutdown --force/--failover stability (3607,3651,4797,5203,4834)
5573 - Do not leak request if server was not able to process it (5154)
5574 - If mds_open unable to find parent dir, make that negative lookup(5154)
5575 - don't create new directories with extent-mapping (5909, 5936)
5577 - fix lustre/lustrefs init scripts for SuSE (patch from Scali, 5702)
5578 - don't hold the pinger_sem in ptlrpc_pinger_sending_on_import
5579 - change obd_increase_kms to obd_adjust_kms (up or down) (5654)
5580 - lconf, lmc search both /usr/lib and /usr/lib64 for Python libs (5800)
5581 - support for RHEL4 kernel on i686 (5773)
5582 - provide error messages when incompatible logs are encountered (5898)
5584 2005-02-18 Cluster File Systems, Inc. <info@clusterfs.com>
5585 * version 1.4.0.10 (1.4.1 release candidate 1)
5587 - don't keep a lock reference when lock is not granted (4238)
5588 - unsafe list practices (rarely) led to infinite eviction loop (4908)
5589 - add per-fs limit of Lustre pages in page cache, avoid OOM (4699)
5590 - drop import inflight refcount on signal_completed_replay error (5255)
5591 - unlock page after async write error during send (3677)
5592 - handle missing objects in filter_preprw_read properly (5265)
5593 - no transno return for symlink open, don't save no-trasno open (3440)
5594 - don't try to complete elan receive that already failed (4012)
5595 - free RPC server reply state on error (5406)
5596 - clean up thread from ptlrpc_start_thread() on error (5160)
5597 - readahead could read extra page into cache that wasn't ejected (5388)
5598 - prevent races in class_attach/setup/cleanup/detach (5260)
5599 - don't dereference de->d_inode after l_dput of de (5458)
5600 - use "int" for stripe value returned from lock_to_stripe (5544)
5601 - mballoc allocation and error-checking fixes in 2.6 (5504)
5602 - block device patches to fix I/O request sizes in 2.6 (5482)
5603 - look up hostnames for IB nals (5602)
5604 - 2.6 changed lock ordering of 2 semaphores, caused deadlock (5654)
5605 - don't start multiple acceptors for the same port (5277)
5606 - fix incorrect LASSERT in mds_getattr_name (5635)
5607 - export a proc file for general "ping" checking (5628)
5608 - fix "lfs check" to not block when the MDS is down (5628)
5610 - service request history (4965)
5611 - put {ll,lov,osc}_async_page structs in a single slab (4699)
5612 - create an "evict_client" /proc entry on OSTs, like the MDS has
5613 - fix mount usage message, return errors per mount(8) (5168)
5614 - change grep [] to grep "[]" in tests so they work in more UMLs
5615 - fix ppc64/x86_64 spec to use %{_libdir} instead of /usr/lib (5389)
5616 - remove ancient LOV_MAGIC_V0 EA support (5047)
5617 - add "disk I/Os in flight" and "I/O req time" stats in obdfilter
5618 - align r/w RPCs to PTLRPC_MAX_BRW_SIZE boundary for performance (3451)
5619 - allow readahead allocations to fail when low on memory (5383)
5620 - mmap locking landed again, after considerable improvement (2828)
5621 - add get_hostaddr() to lustreDB.py for LDAP support (5459)
5623 2004-11-23 Cluster File Systems, Inc. <info@clusterfs.com>
5626 - send OST transaction number in read/write reply to free req (4966)
5627 - don't ASSERT in ptl_send_rpc() if we run out of memory (5119)
5628 - lock /proc/sys/portals/routes internal state, avoiding oops (4827)
5629 - the watchdog thread now runs as interruptible (5246)
5630 - flock/lockf fixes (but it's still disabled, pending 5135)
5631 - don't use EXT3 constants in llite code (5094)
5632 - memory shortage at startup could cause assertion (5176)
5634 - reorganization of lov code
5635 - single portals codebase
5637 - add extents/mballoc support (5025)
5638 - direct I/O reads in the obdfilter (4048)
5639 - kernel patches from LNXI for 2.6 (bluesmoke, perfctr, mtd, kexec)
5641 tbd Cluster File Systems, Inc. <info@clusterfs.com>
5644 - send OST transaction number in read/write reply to free req (4966)
5645 - don't ASSERT in ptl_send_rpc() if we run out of memory (5119)
5646 - lock /proc/sys/portals/routes internal state, avoiding oops (4827)
5647 - the watchdog thread now runs as interruptible (5246)
5648 - handle missing objects in filter_preprw_read properly (5265)
5649 - unsafe list practices (rarely) led to infinite eviction loop (4908)
5650 - drop import inflight refcount on signal_completed_replay error (5255)
5651 - unlock page after async write error during send (3677)
5652 - return original error code on reconstructed replies (3761)
5653 - no transno return for symlink open, don't save no-trasno open (3440)
5655 - add pid to ldlm debugging output (4922)
5656 - bump the watchdog timeouts -- we can't handle 30sec yet
5657 - extra debugging for orphan dentry/inode bug (5259)
5659 2004-11-16 Cluster File Systems, Inc. <info@clusterfs.com>
5662 - fix TCP_NODELAY bug, which caused extreme perf regression (5134)
5663 - allocate qswnal tx descriptors singly to avoid fragmentation (4504)
5664 - don't LBUG on obdo_alloc() failure, use OBD_SLAB_ALLOC() (4800)
5665 - fix NULL dereference in /proc/sys/portals/routes (4827)
5666 - allow failed mdc_close() operations to be interrupted (4561)
5667 - stop precreate on OST before MDS would time out on it (4778)
5668 - don't send partial-page writes before EOF from client (4410)
5669 - discard client grant for sub-page writes on large-page clients (4520)
5670 - don't free dentries not owned by NFS code, check generation (4806)
5671 - fix lsm leak if mds_create_objects() fails (4801)
5672 - limit debug_daemon file size, always print CERROR messages (4789)
5673 - use transno after validating reply (3892)
5674 - process timed out requests if import state changes (3754)
5675 - update mtime on OST during writes, return in glimpse (4829)
5676 - add mkfsoptions to LDAP (4679)
5677 - use ->max_readahead method instead of zapping global ra (5039)
5678 - don't interrupt __l_wait_event() during strace
5680 - add software watchdogs to catch hung threads quickly (4941)
5681 - make lustrefs init script start after nfs is mounted
5682 - fix CWARN/ERROR duplication (4930)
5683 - return async write errors to application if possible (2248)
5684 - add /proc/sys/portal/memused (bytes allocated by PORTALS_ALLOC)
5685 - print NAL number in %x format (4645)
5686 - update barely-supported suse-2.4.21-171 series (4842)
5687 - support for sles 9 %post scripts
5688 - support for building 2.6 kernel-source packages
5689 - support for sles km_* packages
5691 2004-10-07 Cluster File Systems, Inc. <info@clusterfs.com>
5694 - ignore -ENOENT errors in osc_destroy (3639)
5695 - notify osc create thread that OSC is being cleaned up (4600)
5696 - add nettype argument for llmount in #5d in conf-sanity.sh (3936)
5697 - reconstruct ost_handle() like mds_handle() (4657)
5698 - create a new thread to do import eviction to avoid deadlock (3969)
5699 - let lconf resolve symlinked-to devices (4629)
5700 - don't unlink "objects" from directory with default EA (4554)
5701 - hold socknal file ref over connect in case target is down (4394)
5702 - allow more than 32000 subdirectories in a single directory (3244)
5703 - fix blocks count for O_DIRECT writes (3751)
5704 - OST returns ENOSPC from object create when no space left (4539)
5705 - don't send truncate RPC if file size isn't changing (4410)
5706 - limit OSC precreate to 1/2 of value OST considers bogus (4778)
5707 - bind to privileged port in socknal and tcpnal (3689)
5709 - rate limit CERROR/CWARN console message to avoid overload (4519)
5710 - GETFILEINFO dir ioctl returns LOV EA + MDS stat in 1 call (3327)
5711 - basic mmap support (3918)
5712 - kernel patch series update from b1_4 (4711)
5714 2004-09-16 Cluster File Systems, Inc. <info@clusterfs.com>
5717 - avoid crash during MDS cleanup with OST shut down (2775)
5718 - fix loi_list_lock/oig_lock inversion on interrupted IO (4136)
5719 - don't use bad inodes on the MDS (3744)
5720 - dynamic object preallocation to improve recovery speed (4236)
5721 - don't hold spinlock over lock dumping or change debug flags (4401)
5722 - don't zero obd_dev when it is force cleaned (3651)
5723 - print grants to console if they go negative (4431)
5724 - "lctl deactivate" will stop automatic recovery attempts (3406)
5725 - look for existing locks in ldlm_handle_enqueue() (3764)
5726 - don't resolve lock handle twice in recovery avoiding race (4401)
5727 - revalidate should check working dir is a directory (4134)
5729 - don't always mark "slow" obdfilter messages as errors (4418)
5731 2004-08-24 Cluster File Systems, Inc. <info@clusterfs.com>
5734 - don't close LustreDB during write_conf until it is done (3860)
5735 - fix typo in lconf for_each_profile (3821)
5736 - allow dumping logs from multiple threads at one time (3820)
5737 - don't allow multiple threads in OSC recovery (3812)
5738 - fix debug_size parameters (3864)
5739 - fix mds_postrecov to initialize import for llog ctxt (3121)
5740 - replace config semaphore with spinlock (3306)
5741 - be sure to send a reply for a CANCEL rpc with bad export (3863)
5742 - don't allow enqueue to complete on a destroyed export (3822)
5743 - down write_lock before checking llog header bitmap (3825)
5744 - recover from lock replay timeout (3764)
5745 - up llog sem before sending rpc (3652)
5746 - reduce ns lock hold times when setting kms (3267)
5747 - change a dlm LBUG to LASSERTF, to maybe learn something (4228)
5748 - fix NULL deref and obd_dev leak on setup error (3312)
5749 - replace some LBUG about llog ops with error handling (3841)
5750 - don't match INVALID dentries from d_lookup and spin (3784)
5751 - hold dcache_lock while marking dentries INVALID and hashing (4255)
5752 - fix invalid assertion in ptlrpc_set_wait (3880)
5754 - add libwrap support for the TCP acceptor (3996)
5755 - add /proc/sys/portals/routes for non-root route listing (3994)
5756 - allow setting MDS UUID in .xml (2580)
5757 - print the stack of a process that LBUGs (4228)
5759 2004-07-14 Cluster File Systems, Inc. <info@clusterfs.com>
5762 - don't cleanup request in ll_file_open() on failed MDS open (3430)
5763 - make sure to unset replay flag from failed open requests (3440)
5764 - if default stripe count is 0, use OST count for inode size (3636)
5765 - update parent mtime/ctime on client for create/unlink (2611)
5766 - drop dentry ref in ext3_add_link from open_connect_dentry (3266)
5767 - free recovery state on server during a forced cleanup (3571)
5768 - unregister_reply for resent reqs (3063)
5769 - loop back devices mounting and status check on 2.6 (3563)
5770 - fix resource-creation race that can provoke i_size == 0 (3513)
5771 - don't try to use bad inodes returned from MDS/OST fs lookup (3688)
5772 - more debugging for page-accounting assertion (3746)
5773 - return -ENOENT instead of asserting if ost getattr+unlink race (3558)
5774 - avoid deadlock after precreation failure (3758)
5775 - fix race and lock order deadlock in orphan handling (3450, 3750)
5776 - add validity checks when grabbing inodes from l_ast_data (3599)
5778 - add /proc/.../recovery_status to obdfilter (3428)
5779 - lightweight CDEBUG infrastructure, debug daemon (3668)
5780 - change default OSC RPC parameters to be better on small clusters
5781 - turn off OST read cache for files smaller than 32MB
5782 - install man pages and include them in rpms (3100)
5783 - add new init script for (un)mounting lustre filesystems (2593)
5784 - run chkconfig in %post for init scripts (3701)
5785 - drop scimac NAL (unmaintained)
5787 2004-06-17 Cluster File Systems, Inc. <info@clusterfs.com>
5790 - clean kiobufs before and after use (3485)
5791 - strip trailing '/'s before comparing paths with /proc/mounts (3486)
5792 - remove assertions to work around "in-flight rpcs" recovery bug (3063)
5793 - change init script to fail more clearly if not run as root (1528)
5794 - allow clients to reconnect during replay (1742)
5795 - fix ns_lock/i_sem lock ordering deadlock for kms update (3477)
5796 - don't do DNS lookups on NIDs too small for IP addresses (3442)
5797 - re-awaken ptlrpcd if new requests arrive during check_set (3554)
5798 - fix cond_resched (3554)
5799 - only evict unfinished clients after recovery (3515)
5800 - allow bulk resend, prevent data loss (3570)
5801 - dynamic ptlrpc request buffer allocation (2102)
5802 - don't allow unlinking open directory if it isn't empty (2904)
5803 - set MDS/OST threads to umask 0 to not clobber client modes (3359)
5804 - remove extraneous obd dereference causing LASSERT failure (3334)
5805 - don't use get_cycles() when creating temp. files on the mds (3156)
5806 - hold i_sem when setting i_size in ll_extent_lock() (3564)
5807 - handle EEXIST for set-stripe, set proper directory name (3336)
5809 - servers can dump a log evicting a client - lustre.dump_on_timeout=1
5810 - fix ksocknal_fmb_callback() error messages (2918)
5812 2004-05-27 Cluster File Systems, Inc. <info@clusterfs.com>
5815 - don't copy lvb into (possibly NULL) reply on error (2983)
5816 - don't deref dentry after dput, don't free lvb on error (2922)
5817 - use the kms to determine writeback rpc length (2947)
5818 - increment oti_logcookies when osc is inactive (2948)
5819 - update client's i_blocks count via lvb messages (2543)
5820 - handle intent open/close of special files properly (1557)
5821 - mount MDS with errors=remount-ro, like obdfilter (2009)
5822 - initialize lock handle to avoid ASSERT on error cleanup (3057)
5823 - don't use cancelling-locks' kms values (2947)
5824 - use highest lock extent for kms, not last one (2925)
5825 - don't dereference ERR_PTR() dentry in error handling path (3107)
5826 - fix thread race in portals_debug_dumplog() (3122)
5827 - create lprocfs device entries at setup instead of at attach (1519)
5828 - common AST error handler, don't evict client on completion race (3145)
5829 - zero nameidata in detach_mnt in 2.6 (3118)
5830 - verify d_inode after revalidate_special is valid in 2.6 (3116)
5831 - use lustre_put_super() to handle zconf unmounts in 2.6 (3064)
5832 - initialize RPC timeout timer earlier for 2.6 (3219)
5833 - don't dereference NULL reply buffer if mdc_close was never sent (2410)
5834 - print nal/nid for unknown nid (3258)
5835 - additional checks for oscc recovery before doing precreate (3284)
5836 - fix ll_extent_lock() error return code for 64-bit systems (3043)
5837 - don't crash in mdc_close for bad permissions on open (3285)
5838 - zero i_rdev for non-device files (3147)
5839 - clear page->private before handing to FS, better assertion (3119)
5840 - tune the read pipeline (3236)
5841 - fix incorrect decref of invalidated dentry (2350)
5842 - provide read-ahead stats and refine rpc in flight stats (3328)
5843 - don't hold journal transaction open across create RPC (3313)
5844 - update atime on MDS at close time (3265)
5845 - close LDAP connection when recovering to avoid server load (3315)
5846 - update iopen-2.6 patch with fixes from 2399,2517,2904 (3301)
5847 - don't leak open file on MDS after open resend (3325)
5848 - serialize filter_precreate and filter_destroy_precreated (3329)
5849 - loop device shouldn't call sync_dev() for nul device (3092)
5850 - clear page cache after eviction (2766)
5851 - resynchronize MDS->OST in background (2824)
5852 - refuse to mount the same filesystem twice on same mountpoint (3394)
5853 - allow llmount to create routes for mounting behind routers (3320)
5854 - push lock cancellation to blocking thread for glimpse ASTs (3409)
5855 - don't call osc_set_data_with_check() for TEST_LOCK matches (3159)
5856 - fix rare problem with rename on htree directories (3417)
5858 - allow default OST striping configuration per directory (1414)
5859 - fix compilation for qswnal for 2.6 kernels (3125)
5860 - increase maximum number of MDS request buffers for large systems
5861 - change liblustreapi to be useful for external progs like lfsck (3098)
5862 - increase local configuration timeout for slow disks (3353)
5863 - allow configuring ldlm AST timeout - lustre.ldlm_timeout=<seconds>
5865 2004-03-22 Cluster File Systems, Inc. <info@clusterfs.com>
5868 - fixes for glimpse AST timeouts / incorrectly 0-sized files (2818)
5869 - don't overwrite extent policy data in reply if lock was blocked (2901)
5870 - drop filter export grants atomically with removal from device (2663)
5871 - del obd_self_export from work_list in class_disconnect_exports (2908)
5872 - don't LBUG if MDS recovery times out during orphan cleanup (2530)
5873 - swab reply message in mdc_close, other PPC fixes (2464)
5874 - fix destroying of named logs (2325)
5875 - overwrite old logs when running lconf --write_conf (2264)
5876 - bump LLOG_CHUNKSIZE to 8k to allow for larger clusters (2306)
5877 - fix race in target_handle_connect (2898)
5878 - mds_reint_create() should take same inode create lock (2926)
5879 - correct journal credits calculated for CANCEL_UNLINK_LOG (2931)
5880 - don't close files for self_export to avoid uninitialized obd (2936)
5881 - allow MDS with the same name as client node (2939)
5882 - hold dentry reference for closed log files for unlink (2325)
5883 - reserve space for all logs during transactions (2059)
5884 - don't evict page beyond end of stripe extent (2925)
5885 - don't oops on a deleted current working directory (2399)
5886 - handle hard links to targets without a parent properly (2517)
5887 - don't dereference NULL lock when racing during eviction (2867)
5888 - don't grow lock extents when lots of conflicting locks (2919)
5890 2004-03-04 Cluster File Systems, Inc. <info@clusterfs.com>
5893 - account for cache space usage on clients to avoid data loss (974)
5894 - lfsck support in lustre kernel code (2349)
5895 - reduce journal credits needed for BRW writes (2370)
5896 - orphan handling to avoid losing space on client/server crashes
5897 - ptlrpcd can be blocked, stopping ALL progress (2477)
5898 - use lock value blocks to assist in proper KMS, faster stat (1021)
5899 - takes i_sem instead of DLM locks internally on obdfilter (2720)
5900 - recovery for initial connections (2355)
5901 - fixes for mds_cleanup_orphans (1934)
5902 - abort_recovery crashes MDS in b_eq (mds_unlink_orphan) (2584)
5903 - block all file creations until orphan recovery completes (1901)
5904 - client remove rq_connection from request struct (2423)
5905 - conf-sanity test_5, proper cleanup in umount log not availale (2640)
5906 - recovery timer race (2670)
5907 - mdc_close recovey bug (2532)
5908 - ptlrpc cleanup bug (2710)
5909 - mds timeout on local locks (2588)
5910 - namespace lock held during RPCs (2431)
5911 - handle interrupted sync write properly (2503)
5912 - don't try to handle a message that hasn't been replied to (2699)
5913 - client assert failure during cleanup after abort recovery (2701)
5914 - leak mdc device after failed mount (2712)
5915 - ptlrpc_check_set allows timedout requests to complete (2714)
5916 - wait for inflight reqs when ptlrpcd finishes (2710)
5917 - make sure unregistered services are removed from the srv_list
5918 - reset bulk XID's when resending them (caught by 1138 test)
5919 - unregister_bulk after timeout
5920 - fix lconf error (2694)
5921 - handle write after unfinished setstripe, stripe-only getstripe (2388)
5922 - readahead locks pages, leaves pending causing memory pressure (2673)
5923 - increase OST request buffers to 4096 on large machines (2729)
5924 - fix up permission of existing directories in simple_mkdir (2661)
5925 - init deleted item, add assertions ptlrpc_abort_inflight() (2725)
5926 - don't assign transno to errored transactions (2742)
5927 - don't delete objects on OST if given a bogus objid from MDS (2751)
5928 - handle large client PAGE_SIZE readdir on small PAGE_SIZE MDS (2777)
5929 - if rq_no_resend, then timeout request after recovery (2432)
5930 - fix MDS llog_logid record size, 64-bit array alignment (2733)
5931 - don't call usermode_helper from ptlrpcd, DEFAULT upcall (2773)
5932 - put magic in mount.lustre data, check for bad/NULL mount data (2529)
5933 - MDS recovery shouldn't delete objects that it has given out (2730)
5934 - if enqueue arrives after completion, don't clobber LVB (2819)
5935 - don't unlock pages twice when trigger_group_io returns error (2814)
5936 - don't deref NULL rq_repmsg if ldlm_handle_enqueue failed (2822)
5937 - don't write pages to disk if there was an error (1450)
5938 - don't ping imports that have recovery disabled (2676)
5939 - take buffered bytes into account when balancing socknal conn (2817)
5940 - hold a DLM lock over readdir always, use truncate_inode_pages (2706)
5941 - reconnect unlink llog connection after MDS reconnects to OST (2816)
5942 - remove little-endian swabbing of llog records (1987)
5943 - set/limit i_blksize to LL_MAX_BLKSIZE on client (2884)
5944 - retry reposting request buffers if they fail (1191)
5945 - grow extent at grant time to avoid granting a revoked lock (2809)
5946 - lock revoke doesn't evict page if covered by a second lock (2765)
5947 - disable VM readahead to avoid reading outside lock extents (2805)
5949 - return LL_SUPER_MAGIC from statfs for the filesystem type (1972)
5950 - updated kernel patches for hp-2.4.20 kernel (2681)
5952 2004-02-07 Cluster File Systems, Inc. <info@clusterfs.com>
5955 - fix truncated write corruption (2366)
5956 - fix for failed assertion in iopen_connect_dentry (1792,2517)
5958 - don't flag the ptlrpcd thread with PF_MEMALLOC (2636)
5959 - ensure len(uuid) < 37 in lmc (1171)
5960 - fix ia64 OOPS in llog_test (2255)
5961 - zero end of page at obdfilter for partial page writes (2648)
5962 - don't leave stale dentries around after renames (bug 2428)
5963 - fix timeouts when evicting a client with a single lock held (2642)
5964 - set deadline for the initial HELLO message to drain (2634)
5965 - print out dotted-quad IP addresses in the socknal (2302)
5967 - additional debugging for MDS client eviction problem (2443)
5968 - fix mkfsoptions support for osts (2603, 2604)
5970 2004-01-27 Cluster File Systems, Inc. <info@clusterfs.com>
5973 - add series for the vanilla 2.6.0 kernel
5974 - add series for the vanilla 2.4.24 kernel
5975 - add series for a cray x86/64 UL kernel drop
5976 - fix xattr patches for the vanilla 2.4.19 series
5978 - generate true UUIDs in lmc (1171)
5979 - have portals stack dumping break in UML (2466)
5980 - avoid bad dchild deref; avoid inum lock w/o creation (2362)
5981 - allocate with _NOFS in ldlm to avoid deadlock (1933)
5982 - wake callback waiting threads on client eviction (2460)
5983 - Add --ptldebug and --subsystem to lmc (1719)
5984 - update assertion to allow safe interrupt allocation
5985 - set rq_no_resend for cancel requests (2432)
5986 - recalculate ptlrpcd timeout after resend (2494)
5987 - call vfs_rmdir when removing pending directories (2368)
5988 - fix renaming a file to itself (2429)
5989 - lmc creates a default one-stripe lov (2454)
5990 - expand procfs space to handle large clusters (2326)
5991 - increase UML stack to avoid overflow
5992 - update lconf's list of debug and subsystem masks
5993 - fix lfs find --obd (2510)
5994 - /proc tunable for disabling filter read caching (2591)
5995 - stop rpm packages from altering slapd.conf (2301)
5996 - disable nagle in the socknal under 0conf (2578)
5997 - choose mds inode size based on stripe count (2572)
5998 - fix kernel-source rpm problems (2516)
6000 - add --disable-doc to avoid pdf generation (2421)
6001 - update documentation, tests, type-os, comments
6002 - avoid format warnings on ia64
6003 - remove the TOE NAL
6004 - tiny code cleanups by removing unused fields
6006 2004-01-07 Cluster File Systems, Inc. <info@clusterfs.com>
6009 - fix obvious semaphore misuse in as-yet-unused setattr path (2348)
6010 - remove the most blatant lies from BUILDING file (2371)
6011 - change default debug level to reasonable production setting
6012 - reduce client side cache size to reduce cache flush time
6013 - reduce max RPCs in flight to avoid unnecessary file fragmentation
6014 - make TCP zerocopy and pinger support enabled by default (2476)
6015 - sync writes completed after process exits caused crashes (2319)
6016 - maintain correct mount count on the MDS (2356)
6017 - backout 1557, because 2316 wasn't really fixed
6018 - better file I/O statistics gathering in /proc
6019 - don't take unnecessary, deadlock-inducing bug in readpage (2383)
6020 - another kernel patch to fix zero-copy TCP function export
6021 - don't take duplicate lock when processing re-sent getattr (2420)
6022 - lctl uses obd_self_export instead of creating new conn (2353)
6023 - MDS/OST recovery case which requires object creation asserted (2425)
6024 - move lfs from /usr/sbin to /usr/bin in packages
6025 - fix race between mds_client_add and mds_client_free (2417)
6026 - use kmalloc instead of slabs in portals (2430)
6027 - don't create duplicate records when a failover MDS is present (2442)
6028 - remove unnecessary mount age check (2332)
6029 - don't remove directory inodes from locks prematurely (2451)
6030 - don't break if MDS service name is the same as hostname (2103)
6031 - fix races in client write RPC generation when cache full (2482)
6033 2003-12-13 Cluster File Systems, Inc. <info@clusterfs.com>
6036 - remove now-unused request->rq_obd (278)
6037 - if an allocation fails, print out how much memory we've used (1933)
6038 - use PORTAL_SLAB_ALLOC for structures, to get GFP_MEMALLOC (1933)
6039 - add the "configurable stack size" patch to most series files (1256)
6040 - ability to write large log records, for 100+ OST configs (2306)
6041 - fix NULL deref when filter_prep fails (2314)
6042 - fix operator precedence error in filter_sync
6043 - dynamic allocation of socknal TX descriptors (2315)
6044 - fix a missed case in the GFP_MEMALLOC patch, can cause deadlock (2310)
6045 - fix gcc 2.96 compilation problem in xattr kernel patch (2294)
6046 - ensure that CWARN messages in Portals always get to the syslog
6047 - __init/__exit are not for prototype decls (ldlm_init/exit)
6048 - x86-64 compile warning fixes
6049 - fix gateway LMC keyword conflict (2318)
6050 - fix MDS lock inversions in getattr/reint paths (1844)
6051 - fix a rare lock re-ordering bug, which caused deadlock (2322)
6052 - fix i_sem/journal inversion in fsfilt_ext3_write_record (2306)
6053 - DLM race condition prevented some lock evictions (2328)
6054 - ENOMEM detection and retry on socknal sends (2230)
6055 - use GFP_NOFS throughout Lustre, to combat ENOMEM (2230)
6056 - move osc_rpcd into ptlrpc, for use in MDC and others (2329)
6057 - protect MDS inode fsdata with stronger locking; fixes assertion (2313)
6058 - better error messages when a client is rejected during recovery (1505)
6059 - avoid cancelling locks which were never granted, after failure (2330)
6060 - fix i_sem/journal inversion in mds_client_add (2333)
6061 - fix truncate/getattr lock cycle deadlock (2334)
6062 - use rpcd to send close; allows resend after timeout, avoid leak (1897)
6063 - fix two rare exit paths which could leak an l_lock() ref (2321)
6064 - fencepost error in MDS/OST orphan recovery (2226)
6065 - make log record alignment 8 bytes (1988)
6066 - lstripe now fails when requested offset > ost_count (2237)
6067 - ensure that all kernel series have a complete list.h (1607)
6068 - fix crashes in special-file operations (2316)
6069 - lctl create/brw OID mismatch, caused by obsolete filter loop (2339)
6071 - allow configurable automake binary, for testing new versions
6072 - small update to the lfs documentation
6074 2003-12-03 Cluster File Systems, Inc. <info@clusterfs.com>
6076 * fix negative export reference count in fsfilt_sync (2312)
6078 2003-12-01 Cluster File Systems, Inc. <info@clusterfs.com>
6079 * release candidate 0.9.1
6081 - orphans are moved into the PENDING directory for possible recovery
6082 - replayed opens now open by fid for orphan/rename safety (1042)
6083 - last close of an orphan inode generates a transno (683)
6084 - chdir() and mount() now pin the directory entry (1020)
6085 - avoid CERROR in normal ll_setattr_raw() error case (1500)
6086 - discard very old requests without processing them (1502)
6087 - remove some common, well-understood CERRORs (1505)
6088 - require O_DIRECT I/O to be page-sized to workaround IA64 crash (1609)
6089 - clear "grant" flags in OST replies until OST grant code lands (1644)
6090 - fix read performance by not clobbering i_blksize on client (1598)
6091 - fix __ldlm_handle2lock oops by not dereferencing lock after PUT (1625)
6092 - make LRU size a /proc tunable, clears locks when reduced (707)
6093 - fix some lprocfs rot that prevented ptlbd from loading (1732)
6094 - server locks take references on exports now (1558)
6095 - build fixes for 2.4.20-rh trees (1663)
6096 - return an error from lov_create if all OSCs are inactive (1751)
6097 - fix import levels when a reconnect happens without a timeout (1597)
6098 - exit early from mds_open if we get a lookup error (1749)
6099 - partial page read at EOF wouldn't wait for disk before sending (1642)
6100 - avoid NULL deref in obdfilter when reading page past EOF (1592)
6101 - avoid LASSERT in ll_intent_lock if server failed very early (1090)
6102 - fix LBUG in ll_it_open_error with rc = -2 (1861)
6103 - write/truncate lock inversion (1639)
6104 - Don't auto-load obdclass, portals modules during cleanup (1495)
6105 - fix timestamps from jumping to "now" (1763)
6106 - extra journal assertions (1648)
6107 - add an extra multiunlink test (1771)
6108 - fix read_record/write_record API (1776)
6109 - fix leak of offset_extent, possible incorrect i_size later (1772)
6110 - fix lasserts in mis-matched transnos during open-unlink testing (1541)
6111 - Debugging for the kqswnal_get_idle_tx problems (1820)
6112 - Allow recovery to be attempted multiple times (1536)
6113 - Write out MDS last_rcvd file after it is first created (1600)
6114 - Fix tx_descriptor leak in failed transmit situations (1827)
6115 - ext3 journaling fixes for assertion failure after IO error (1871)
6116 - class_export_put() on freed export after completion AST error (1896)
6117 - Fix revalidate looping in VFS (1322)
6118 - Don't access a freed export during MDS_REINT timeout (1521)
6119 - Add open-unlink recovery support on the MDS (1673,1764)
6120 - Return an error if no MDS data was read from last_rcvd (1946)
6121 - Fix for lookup "." or ".." crash on error (1932,1931,1935)
6122 - Don't setup a disk device that doesn't match exported UUID (317)
6123 - Reduce bulk RPC timeout to avoid cascading client/OST failures (1845)
6124 - avoid committing NULL handle in force close
6125 - local.sh is now a one-stripe LOV configuration
6126 - POSIX utime.4 -EPERM on FIFO not owned by user (56)
6127 - fix ext3 htree duplicate directory entry corruption (1516)
6128 - POSIX creat.13, fstat.1, open.18, stat.3 new file atime/mtime (2020)
6129 - update to new LOV EA format (2097)
6130 - interoperability for different PAGE_SIZE/wordsize (686,1821,1343,2042)
6132 2003-06-15 Phil Schwan <phil@clusterfs.com>
6135 - imports and exports cleanup too early, need refcounts (349, 879, 1045)
6136 - per-import/export recovery handling (958, 931, 959)
6137 - multiple last-rcvd slots, for serving multiple FSes (949)
6138 - connections are again shared between multiple imp/exports (963, 964)
6139 - "umount -f" would hang if any requests needed to be sent (393, 978)
6140 - avoid pinning large req buffer by copying for queued messages (989)
6141 - add "uuid" to "lctl device" command to help upcalls (991)
6142 - "open" RPCs with transnos would confuse recovery counters (1037)
6143 - do proper endian conversion of all wire messages (288, 340, 891)
6144 - remove OST bulk get LBUGs, fix ost_brw_write cleanup (1126)
6145 - call waiting locks callback from LDLM recovery thread (1127, 1151)
6146 - fix ptlrpc_connection leak in target_handle_connect (1174)
6147 - fix import refcounting bug in OST and MDS cleanup (1134)
6148 - if an invalid-at-open-time OSC returned before close(), LBUG (1150)
6149 - fix very unlikely obd_types race condition (501)
6150 - remove osc_open hack for echo_client (1187)
6151 - we leaked exports/dlmimps for forcibly disconnected clients (1143)
6152 - a failure in read_inode2 leads to deadlock (1139)
6153 - cancel ack-locks as soon as transaction is committed (1072)
6154 - fix major leaks and crashes in the bulk I/O path (937, 1057)
6155 - make sure to commitrw after any preprw to avoid deadlock (1162)
6156 - failing to execute a file in a lustre FS would lock inode (1203)
6157 - small DEBUG_REQ fix to avoid dereferencing a NULL (1227)
6158 - don't ASSERT while cleaning up an incompletely-setup obd (1248)
6159 - obd_uuid2tgt would walk off the end of the list (1255)
6160 - on IA64 the osc would give portals incorrect bulk size (1258)
6161 - fix debug daemon ioctl interface; allows daemon on ia64 (1274)
6162 - fix lock inversion caused by new llite matching code (1282)
6163 - limit the number of dirty pages on a client to 10MB (1286)
6164 - timed out locks were not being corrected cancelled (1289)
6165 - fix O_DIRECT above 4GB on IA-32 (1292)
6166 * major user-visible changes
6167 - fail out/fail over policy now controlled by the upcall (993)
6169 - add OBD_PING to check server availability and failure (954)
6170 - lustre messages are now sent in sending host order (288, 340, 891)
6171 - add eadatalen to MDS getattr reply (340)
6172 - OST read replies may contain second buffer, with per-page status (593)
6174 2003-03-11 Phil Schwan <phil@clusterfs.com>
6177 - LDLM_DEBUG macro fix, for gcc 3.2 (850)
6178 - failed open()s could cause deadlock; fixed (867, 869)
6179 - stop cancelling OST locks when files are closed (481)
6180 - overlapping XID spaces caused network corruption (851, 853)
6181 - fix unsafe fsfilt counter arithmetic; change to atomic_t
6182 - setattr_raw added, to do single-RPC, server-side setattrs
6183 - lmc/lconf syntax change for OST UUIDs
6184 - fix crashy race condition between ptlrpc_free_req and osc_close
6185 - don't use request in mdc_enqueue if we hit a timeout (889)
6186 - don't set the inode i_size for regular files from the MDS (896)
6187 - handle out of order completion AST (842)
6188 - don't LBUG if a lock request times out after receiving AST (913)
6189 - avoid d_rehash race in ll_find_alias by rehashing inside dcache_lock
6190 - if a bad lock AST arrives, send an error instead of dropping entirely
6191 - return 0 from revalidate2 if ll_intent_lock returns -EINTR (912)
6192 - fix leak in bulk IO when only partially completed (899, 900, 926)
6193 - fix O_DIRECT for ia64 (55)
6194 - (almost) eliminate Lustre-kernel-thread effects on load average (722)
6195 - C-z after timeout could hang a process forever; fixed (977)
6197 - client-side I/O cache (678, 924, 929, 941, 970)
6199 - READPAGE and SETATTRs which don't take server-side locks get
6202 2003-02-11 Phil Schwan <phil@clusterfs.com>
6205 - Fix ldlm_lock_match on the MDS to avoid matching remote locks (592)
6206 - Fix fsfilt_extN_readpage() to read a full page of directory
6207 entries, or fake the remainder if PAGE_SIZE != blocksize (500)
6208 - Avoid extra mdc_getattr() in ll_intent_lock when possible (534, 604)
6209 - Fix imbalanced LOV object allocation and out-of-bound access (469)
6210 - Most intent operations were removed, in favour of a new RPC mode
6211 that does a single RPC to the server and bypasses most of the VFS
6212 - All LDLM resource ID arrays were removed in favour of ldlm_res_id
6213 - Aggressively cancel local locks on DLM servers
6214 - mds_reint_unlink sends EA to the client if it's the last nlink.
6215 client uses that EA to unlink OST objects.
6216 - mds_reint_{rename,unlink,link} were rewritten to take ordered locks
6217 - recursive symlinks were fixed (439)
6218 - fixed NULL deref in DEBUG_REQ
6219 - filter_update_lastobjid no longer calls sync, which annoyed extN
6220 - fixed multi-client small-writes to a single file problem (445)
6221 - fixed mtime updates during file writes (607)
6222 - fixed vector writes on obdfilter causing problems when ENOSPC (670)
6223 - fixed bug in obd_brw_read/write() (under guise of testing 367)
6224 - fixed Linux OST size reporting problem (444, 656)
6225 - OST now updates object mtime with writes or setattr (607, 619)
6226 - client verifies file size before zeroing page past EOF (445)
6227 - OST now writes last allocated objid to disk with allocation (108)
6228 - LOV on echo now works (409)
6230 - mds_reint_unlink sends a new buffer, with the EA included. this
6231 buffer is only valid if body->valid & OBD_MD_FLEASIZE, which is only
6232 set if a regular file was being unlinked, and it was the last link
6233 - use PtlGet from the target for bulk writes (315)
6234 - OST now updates object mtime with writes or setattr (607, 619)
6235 - LDLM now has a grant-time callback to revalidate locked items, if
6237 - Many MDS operations were reorganized to combat race conditions
6239 - Merge b_intel branch (updated lprocfs code) - now at /proc/fs/lustre
6240 - configure check to avoid gcc version 2.96 20000731-2.96-98) (606)
6242 2003-01-06 Andreas Dilger <adilger@clusterfs.com>
6245 - Fully reactivate OST imports after reconnection (512, others)
6246 - Make sure client sees our -ENOTCONN from mds_handle (513 - partial)
6247 - More graceful error handling for truncating on dead OST (515)
6248 - Don't error out unless we're actually accessing dead stripes (474)
6249 - Fix garbage sizes when stripes are missing (410)
6250 - LRU counters were broken, causing constant lock purge (433, 432)
6251 - garbage on read from stripes with failed OSTs (441)
6252 - mark OSCs as active before reconnecting during recovery (438)
6253 - lov_enqueue and lov_cancel need to handle inactive OSTs (403)
6254 - lfind did not preserve OST order in output (443)
6255 - symlinks cause hung clients, incorrect data (439)
6256 - stop dereferencing request after dropping refcount (457)
6257 - don't LASSERT(spin_is_locked) on non-SMP (455)
6258 - fixes for many rename() bugs
6259 - fstat didn't correctly synchronize attributes (399)
6260 - server must handle lock cancellation during blocking AST prep (487)
6261 - bulk descriptors were free()d too soon (511)
6262 - fix paths in lconf, which would load incorrect modules (451, 507)
6263 - fix confusing lconf 'host not found' error message (386)
6264 - fix lock order deadlock on OST (O/R i_sem before journal ops, 478)
6265 - fix race condition in mdc_blocking_ast() for inode access (526)
6266 - fix lov_unpackmd() unpacking wrong number of stripes (537)
6267 - fix lov_set_osc_active() marking wrong OSC inactive (440)
6268 - fix bad lstripe lov_unpackmd() assertion (fix layering too) (527)
6269 - fix multiple writes of stripe MD to MDS (358, maybe 519)
6270 - fix lstripe in several ways (kernel side) (527)
6271 - fix request leak in ldlm_cli_enqueue (262)
6272 - incorrect OSC was marked inactive after OST failure
6273 - call mds_fs_cleanup before unmounting filesystem (524)
6274 - fix races between taking ns_lock and ldlm_lock_change_resource
6275 - fix races updating LOV export open file list
6276 - fix lov_enqueue error path, avoid decref-ing bad lock handle
6277 - fix recovery NULL deref in ldlm_cli_cancel_unused
6278 - fix some DLM races by using new hash table for lock handles (419)
6279 - permit the client to specify desired inodes, at replay
6280 - duplicate requests when we queue them for replay reintegration
6281 - fix last_rcvd offset calculation
6282 - sync after each recovered transaction, so we always make progress
6283 - never, not always, ERESTART requests without transnos
6284 - store the lov_desc in the MDS, so we don't depend on getlovinfo to
6286 - skip replay if the MDS says that the client is already connected
6287 - don't check for a recovery-enabled export to match lctl's UUID
6288 - don't INC_USE_COUNT for phantom exports
6289 - don't crash when cleaning up phantom exports (567)
6290 - don't double-finish or set replay data for errored mdc_open requests
6291 - abort requests when they time out, so we don't get old replies
6292 - send/receive replies for AST messages again
6293 - if the client says that it doesn't have the lock, cancel it on the
6295 - if we timeout during I/O, don't try to cancel an in-use lock; instead
6296 mark it as destroyed, it will all work out when decref is called
6297 - fix module use counts (22, 581)
6299 - ASTs now expect a reply (server cancels lock on error reply)
6301 2002-12-02 Andreas Dilger <adilger@clusterfs.com>
6304 - fix many simultaneous client startup (392)
6305 - fix dentry->d_it clobbering
6306 - credentials weren't being shipped for readdir/getattr operations
6307 - remove invalid assertions triggered during some concurrent MD
6309 - proper Lustre versions added (336, 389)
6310 - fix memory leak for create error case (398)
6311 - fix LOV locking bug that would get cli/srv out of sync
6312 - fix echo client over LOV (409)
6313 - fix dbench 2, extN refcount problem (170, 258, 356, 418)
6314 - fix double-O_EXCL intent crash (424)
6315 - avoid sending multiple lock CANCELs (352)
6317 - MDS can do multi-client recovery (modulo bugs in new code)
6319 - many updates, edits, cleanups
6321 2002-11-18 Phil Schwan <phil@clusterfs.com>
6324 - fix null d_it dereference (346)
6325 - fix full OST/dbench hang (333)
6326 - fix permission problem with file removal (286)
6327 - fix removal of OSCs from LOV when they fail
6328 - fix NULL deref during bulk timeout (214)
6329 - fix problems related to multiple filesystems on one MDS (241)
6330 - fixed serious subtle metadata locking bugs
6331 - free locks on clients when inodes are removed due to memory
6333 - fix inode pointer in lock data (285)
6334 - partial support for multiple MDS on a single host (241)
6335 - data locks weren't cancelled at clear_inode time (290, 311)
6336 - intent locks could lead to unbounded lock growth (205)
6337 - added a maximum lock count, an LRU list, and a flusher
6338 - fix multiple rename (365)
6339 - properly abstracted the echo client
6340 - OSC locked 1 byte too many; fixed
6341 - rewrote brw callback code:
6342 - fixed recovery bugs related to LOVs (306)
6343 - fixed too-many-pages-in-one-write crash (191)
6344 - fixed (again) crash in sync_io_timeout (214)
6345 - probably fixed callback-related race (385)
6347 - Add capability to MDS protocol
6348 - LDLM cancellations and callbacks on different portals
6350 2002-10-28 Andreas Dilger <adilger@clusterfs.com>
6353 - limit client IOV size to PTL_MD_MAX_IOV (611336, 191)
6354 - defer open object destruction to close time (601981, 138)
6355 - open/close OST file handle in obdo (OBD_MD_FLHANDLE) (601981, 138)
6356 - move LDLM_ENQUEUE/CONVERT back to MDS portal (625069)
6357 - abstract ll_lookup2, fix ll_revalidate2 to use abstraction (256)
6358 - don't call obd_setattr in ll_file_release for destroyed objects
6359 * protocol change to lustre_msg: move |version| and add |flags|
6360 * protocol change to osc_punch: "start" in "o_size", "end" in "o_blocks"
6361 * lock replay: for LDLM_FL_REPLAY trust client to do right thing
6362 * added replay of create, unlink, link and rename operations during
6363 MDS failover; recovery should be much more robust now
6364 * remove failed OSCs from LOVs (only lov_create uses this so far)
6365 * the lustre-HOWTO was brought (more) up to date (582544)
6367 2002-10-23 Phil Schwan <phil@clusterfs.com>
6370 - in-use dentries weren't being reused properly (617851)
6371 - prevent multiple LDLM setup (599178)
6372 - fix LOV size calculations for truncate (617853)
6373 - fix client handling of MDS intent errors (POSIX)
6374 - fix permission bug in lovstripe.c test (624321)
6375 - fix MDS thread deadlock - move LDLM handler to DLM portal (625069)
6376 - truncate past end of file could corrupt data
6377 - proper cleanup after timeouts, crashes, etc (592524, 550815)
6378 - a race in recovery could return ETIMEDOUT to apps (623947)
6379 - building outside the source directory was fixed
6380 * the lustre-HOWTO was brought (more) up to date (582544)
6381 * major progress was made on recovery functionality
6383 2002-10-10 Phil Schwan <phil@clusterfs.com>
6386 - recovery deadlock fix
6387 - rm -rf causes LBUG fix (617817)
6388 - file open by multiple tasks fix (618962)
6389 - directory permissions bugs (602707 and 620007)
6390 - journal_stop fixed with locking (611313)
6391 - O_APPEND failures resolved (618273, perhaps 614459)
6392 - lconf PATH fix (619770)
6393 - IA64 build fix (621450)
6394 - RPC buffer sizes scale with amount of memory
6396 2002-10-01 Phil Schwan <phil@clusterfs.com>
6399 - locks would be cancelled without throwing away data pages,
6400 resulting in inconsistent data (605627)
6401 - inode attributes were not always being refreshed (605627, 612449)
6402 - lconf now continues to cleanup after lctl reports an error
6403 - MDS now enforces user permissions (602707)
6404 - lprocfs cleanup fixed, but not yet enabled (614157)
6405 - fixed infinite server hang, should a client not respond to an AST
6406 - avoid going into recovery if user calls readlink() with a buffer
6407 that's too small (613941)
6408 - AST RPCs no longer require replies (614867) -- this may be changed
6409 - don't crash server if client sends an IOV that's too big (611336)
6410 - fixed lock conversion deadlock (611892)
6411 - fixed the following of symlinks (614622)
6412 * recovery: the server can remove locks from a client that dies, other
6413 clients can make progress
6414 * more extN patch fixes
6415 * compile-time configurable ptlrpc buffer allocations
6417 - collaborative read cache document
6418 - Lustre Lite Performance CDR document-in-progress
6420 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
6423 - fix typo in patch-2.4.18
6425 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
6428 - clear ptlrpc request each time in handle_incoming_request()
6429 - unlink of files now destroys the object on the OST
6431 2002-09-19 Peter Braam <braam@clusterfs.com>
6433 * add hard link support
6434 * change obdfile creation method
6435 * kernel patch changed
6437 2002-09-19 Peter Braam <braam@clusterfs.com>
6440 - stack overflow bug in extN fixed
6442 2002-09-18 Andreas Dilger <adilger@clusterfs.com>
6444 * documentation updates
6445 - add man pages for config tools
6446 - update tests/README to describe testing with new config tools
6447 - finish metadata API descriptions
6448 * bug fixes and cleanups
6449 - statfs workaround for 16TB limit
6450 - LOV stripe allocation improved, can stripe on subset of OSTs
6451 - LOV file size/IO offset was wrong for files > 4GB in size
6452 - object EA data was being dropped, caused files to be unreadable
6453 - memory overflow with non-LOV OST caused memory corruption
6454 - fixed regression tests to work with new config tools, obdfilter
6455 - fixed bug when directory size became larger than 1 block
6456 - fixed bug (for single client case) when PWD was deleted
6457 - invalidate local directory pages when doing intent-based ops
6458 - avoid LDLM oops when lock callback contained bad data
6460 2002-09-09 Andreas Dilger <adilger@clusterfs.com>
6462 * documentation updates
6463 * bug fixes and cleanups
6464 - configuration tools
6467 - 64-bit compile warnings
6468 - 64-bit internal statfs data
6470 * test_brw on persistent OST devices
6472 * lprocfs (disabled)
6474 2002-09-04 Andreas Dilger <adilger@clusterfs.com>
6476 * documentation updates
6477 * bug fixes and cleanups
6478 * configuration tools
6480 2002-08-30 Peter J. Braam <braam@clusterfs.com>
6483 * many small fixes to 0_5_4
6484 * io/network handling
6485 * thinkos in MDS operations
6487 2002-08-24 Peter J. Braam <braam@clusterfs.com>
6490 * crucial basic fixes to 0.5.3
6491 * IOR, Iozone work over Elan
6494 2002-08-07 Phil Schwan <phil@clusterfs.com>
6495 * version 0_5_3, our first alpha
6496 * we use the new Portals iovs
6497 * documentation updates
6498 * bug fixes and cleanups
6499 * small changes in the DLM wire protocol
6501 2002-07-25 Peter J. Braam <braam@clusterfs.com>
6502 * version 0_5_1 with some initial stability,
6503 * locking on MD and file I/O.
6504 * documentation updates
6505 * several bug fixes since 0.5.0
6506 * small changes in wire protocol
6508 2002-07-18 Phil Schwan <phil@clusterfs.com>
6510 * delivered as Lustre Light Alpha
6511 * fixed a crash after handling invalid MDS requests
6512 * fixed directory pages for architectures with non-4k pages sizes
6514 2002-07-11 Andreas Dilger <adilger@clusterfs.com>
6515 * release version v0_4_4
6516 * Moves TCP acceptor to be on port 2432 (unused Coda port) instead
6518 * Fixes a number of interruption problems with OST operations.
6519 * Update documentation for portals header changes
6520 * Move all wire protocol structs/defines to lustre_idl.h
6521 * Fixes symlink length bug.
6522 * Add tcpdump to repository.
6524 2002-07-05 Andreas Dilger <adilger@clusterfs.com>
6525 * release version v0_4_3
6526 * Fixes statfs for inodes on extN.
6527 * Fixes bug in runtests which would delete /etc/hosts.
6528 * Use 64-bit object IDs wherever possible (not into VFS though)
6529 Remove ost_get_info, which is unused by lustre, and out of date.
6531 2002-07-03 Peter Braam <braam@clusterfs.com>
6532 * release version v0_4_2 Fixes a lookup error (type not passed)
6533 * move forward to head of Portals
6534 * move forward to latest Lustre kernel
6536 2002-06-25 Peter Braam <braam@clusterfs.com>
6537 * release version v0_4_1. Hopefully stable on single node use.