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 2.6.32-504.16.2.el6 (RHEL6.6)
8 3.10.0-229.4.2.el7 (RHEL7.1)
9 3.0.101-0.47.52 (SLES11 SP3)
10 * Client known to build on unpatched kernels:
11 2.6.32-431.29.2.el6 (RHEL6.5)
12 2.6.32-504.16.2.el6 (RHEL6.6)
13 3.10.0-229.4.2.el7 (RHEL7.1)
14 3.0.101-0.47.52 (SLES11 SP3)
16 * Recommended e2fsprogs version: 1.42.9.wc1 or newer
17 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
18 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
19 stack overflow. For more information, please refer to bugzilla 17630.
20 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
21 the re-exporting nodes to be mounted with "32bitapi" mount option
23 --------------------------------------------------------------------------------
25 03-10-2015 Intel Corporation
27 * See https://wiki.hpdd.intel.com/display/PUB/Lustre+Support+Matrix
28 for currently supported client and server kernel versions.
29 * Server known to build on patched kernels:
30 2.6.32-431.29.2.el6 (RHEL6.5)
31 2.6.32-504.8.1.el6 (RHEL6.6)
32 3.0.101-0.46 (SLES11 SP3)
33 * Client known to build on unpatched kernels:
34 2.6.32-431.29.2.el6 (RHEL6.5)
35 2.6.32-504.8.1.el6 (RHEL6.6)
36 3.10.0-123.20.1.el7 (RHEL7)
37 3.0.101-0.46 (SLES11 SP3)
38 * Recommended e2fsprogs version: 1.42.9.wc1 or newer
39 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
40 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
41 stack overflow. For more information, please refer to bugzilla 17630.
42 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
43 the re-exporting nodes to be mounted with "32bitapi" mount option
45 Severity : enhancement
47 Description: control OST-index in IDIF via ROCOMPAT flag.
48 Details : Introduce new flag OBD_ROCOMPAT_IDX_IN_IDIF that is stored in the
49 last_rcvd file. For new formatted OST device, it will be auto set;
50 for the case of upgrading from old OST device, you can enable it
51 via the lproc interface osd-ldiskfs.index_in_idif. With such flag
52 enabled, for new created OST-object, its IDIF-in-LMA will contain
53 the OST-index; for the existing OST-object, the OSD will convert
54 old format IDIF as new format IDIF with OST-index stored in the
55 LMA EA when accessing such OST-object or via OI scrub. Once such
56 flag is enabled, it cannot be reverted back, so the system cannot
57 be downgraded to the orignal incompatible version.
59 --------------------------------------------------------------------------------
61 07-30-2014 Intel Corporation
63 * See https://wiki.hpdd.intel.com/display/PUB/Lustre+Support+Matrix
64 for currently supported client and server kernel versions.
65 * Server known to build on patched kernels:
66 2.6.32-431.20.3.el6 (RHEL6.5)
67 3.0.101-0.31 (SLES11 SP3)
68 * Client known to build on unpatched kernels:
69 2.6.32-431.20.3.el6 (RHEL6.5)
70 3.0.101-0.31 (SLES11 SP3)
73 * Recommended e2fsprogs version: 1.42.9.wc1 or newer
74 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
75 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
76 stack overflow. For more information, please refer to bugzilla 17630.
77 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
78 the re-exporting nodes to be mounted with "32bitapi" mount option
80 --------------------------------------------------------------------------------
82 10-31-2013 Intel Corporation
84 * See https://wiki.hpdd.intel.com/display/PUB/Lustre+Support+Matrix
85 * Server builds for kernels:
86 2.6.32-358.18.1.el6 (RHEL6.4)
87 * Client builds for unpatched kernels:
88 2.6.32-358.18.1.el6 (RHEL6.4)
89 3.0.92-0.8 (SLES11 SP3)
91 * Recommended e2fsprogs version: 1.42.7.wc2 or newer
92 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
93 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
94 stack overflow. For more information, please refer to bugzilla 17630.
95 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
96 the re-exporting nodes to be mounted with "32bitapi" mount option
98 --------------------------------------------------------------------------------
100 05-30-2013 Intel Corporation
102 * See https://wiki.hpdd.intel.com/display/PUB/Lustre+Support+Matrix
103 * Server builds for kernels:
104 2.6.32-358.6.2.el6 (RHEL6.4)
105 * Client builds for unpatched kernels:
106 2.6.32-358.6.2.el6 (RHEL6.4)
107 3.0.74-0.6.6 (SLES11 SP2)
109 * Recommended e2fsprogs version: 1.42.7.wc1 or newer
110 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
111 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
112 stack overflow. For more information, please refer to bugzilla 17630.
113 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
114 the re-exporting nodes to be mounted with "32bitapi" mount option
116 --------------------------------------------------------------------------------
118 9-30-2012 Whamcloud, Inc.
120 * See https://wiki.hpdd.intel.com/display/PUB/Lustre+Support+Matrix
121 * Server support for kernels:
122 2.6.32-279.5.1.el6 (RHEL6)
123 * Client support for unpatched kernels:
124 2.6.32-279.5.1.el6 (RHEL6)
126 2.6.18-238.19.1.el5 (RHEL5)
127 2.6.16.60-0.69.1 (SLES10)
128 * Recommended e2fsprogs version: 1.41.90.wc4
129 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
130 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
131 stack overflow. For more information, please refer to bugzilla 17630.
132 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
133 the re-exporting nodes to be mounted with "32bitapi" mount option
135 Severity : enhancement
137 Description: Update RHEL6.3 kernel to 2.6.32-279.5.1.el6.
139 Severity : enhancement
141 Description: Update RHEL6.3 kernel to 2.6.32-279.2.1.el6.
143 Severity : enhancement
145 Description: Update RHEL6.2 kernel to 2.6.32-220.17.1.el6
147 Severity : enhancement
149 Description: Update RHEL6.2 kernel to 2.6.32-220.13.1.el6
151 Severity : enhancement
153 Description: Update RHEL6.2 kernel to 2.6.32-220.7.1.el6
155 --------------------------------------------------------------------------------
157 04-30-2012 Whamcloud, Inc.
159 * See https://wiki.hpdd.intel.com/display/PUB/Lustre+Support+Matrix
160 * Server support for kernels:
161 2.6.32-220.4.2.el6 (RHEL6)
162 * Client support for unpatched kernels:
163 2.6.32-220.4.2.el6 (RHEL6)
165 2.6.18-238.19.1.el5 (RHEL5)
166 2.6.16.60-0.69.1 (SLES10)
167 * Recommended e2fsprogs version: 1.41.90.wc4
168 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
169 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
170 stack overflow. For more information, please refer to bugzilla 17630.
171 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
172 the re-exporting nodes to be mounted with "32bitapi" mount option
174 Severity : enhancement
176 Description: Update RHEL6.2 kernel to 2.6.32-220.4.2.el6
178 Severity : enhancement
180 Description: Update RHEL6.2 kernel to 2.6.32-220.4.1.el6
182 Severity : enhancement
184 Description: Update RHEL6.2 kernel to 2.6.32-220.el6
186 Severity : enhancement
188 Description: Update RHEL6.1 kernel to 2.6.32-131.17.1.el6
190 --------------------------------------------------------------------------------
192 09-30-2011 Whamcloud, Inc.
194 * Server support for kernels:
195 2.6.18-238.19.1.el5 (RHEL5)
196 2.6.32-131.6.1.el6 (RHEL6)
197 * Client support for unpatched kernels:
198 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
199 2.6.18-238.19.1.el5 (RHEL5)
200 2.6.32-131.6.1.el6 (RHEL6)
201 2.6.29.4-167.fc11 (FC11)
202 2.6.16.60-0.69.1 (SLES10)
204 * Recommended e2fsprogs version: 1.41.90.wc3
205 * NFS export disabled when stack size < 8192. Since the NFSv4 export of
206 Lustre filesystem with 4K stack may cause a stack overflow. For more
207 information, please refer to bugzilla 17630.
208 * NFSv4 reexport requires Lustre client on the re-exporting nodes to be
209 mounted with "32bitapi" mount option
211 Severity : enhancement
212 Description: Update RHEL6.1 kernel to 2.6.32-131.6.1.el6
214 Severity : enhancement
215 Description: Update RHEL5.6 kernel to 2.6.18-238.19.1.el5
217 Severity : enhancement
218 Description: Update RHEL5.6 kernel to 2.6.18-238.9.1.el5
221 Frequency : only with SLES10
223 Description: Use OFED "KMP" provided by Novell
224 Details : SLES10 SP3 ships with OFED in a separate "KMP" package.
225 Lustre is now built against this package. That means you need to
226 install the ofed-kmp package from Novell for the patchless client.
227 Note that the ofed-kmp that Novell ships may not exactly match the
228 kernel version but should still be compatible.
230 Severity : enhancement
232 Description: Update RHEL5.5 kernel to 2.6.18-194.17.1.el5
233 Update OEL5.5 kernel to 2.6.18-194.17.1.0.1.el5.
235 --------------------------------------------------------------------------------
237 2010-07-15 Oracle, Inc.
239 * Server support for kernels:
240 2.6.18-164.11.1.el5 (RHEL 5)
241 2.6.18-164.11.1.0.1.el5 (OEL 5)
242 * Client support for unpatched kernels:
243 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
244 2.6.18-164.11.1.el5 (RHEL 5),
245 2.6.18-164.11.1.0.1.el5 (OEL 5)
246 2.6.16.60-0.42.8 (SLES 10),
248 2.6.29.4-167.fc11 (FC11)
249 * Recommended e2fsprogs version: 1.41.10-sun2
251 Severity : enhancement
253 Description: Distribute lfs_migrate script with lustre
254 Details : The lfs_migrate mentioned in the manual is now included in the
255 lustre tree in lustre/scripts and will distributed in lustre RPMs.
257 Severity : enhancement
259 Description: Failover nid list added to import proc file.
261 Severity : enhancement
263 Description: "som_preview" mount option required to use SOM
264 Details : The clients will need to use som_preview mount option
268 Description: Lustre does not do 1MB IOs to HW RAID
269 Details : Bump MAX_PHYS/HW_SEGMENTS and SG_ALL to 256 in the RHEL5 kernel.
271 Severity : enhancement
273 Description: lfs quota output is not very convenient for awk/sed-parsing
274 Details : Some positions in lfs quota output table could be empty or
275 non-empty which made it hard to parse it with scripts, now
276 a dash is put instead of space where there is not supposed
281 Description: Update to OFED 1.5.1
282 Details : OFED 1.5.1 will be used in environments that don't use
283 vendor supplied OFED.
287 Description: Remove lnet. prefix from lctl params display
288 Details : Also renamed lnet.memused to lnet_memused to
293 Description: Use conf_param -d to remove permanant settings.
297 Description: Add list_param -R
298 Details : lctl list_param now supports recursive mode.
301 Bugzilla : 21829, 22312
302 Description: Fix llobdstat.
303 Details : 1. Now updating the write_bytes counter
304 2. Port script fixes from 1.8.
308 Description: decrease the usage of memory on clients.
309 Details : 1. On clients, recycle dentries and inodes unused.
310 2. Delete the code related to ll_deathrow(att 6215 in bug 1443). It
315 Description: llobdstat fix and enhancement
316 Details : add a counter to set a limit to how many samples will be returned
317 fix a wildcard in the path to limit to obdfilter stats only
321 Description: lfs check only allowed for root
322 Details : Code cleanups and added a regresion test of non-root use of
327 Description: MDS Oopses in ptlrpc_dispatch_difficult_reply()/__list_add()
328 Details : Fix hr->hr_index overload.
332 Description: Fix FIEMAP functionality.
336 Description: FC11 patchless client support added
340 Description: Update RHEL5.4 kernel to 2.6.18-164.11.1.el5 and
341 OEL5.4 kernel to 2.6.18-164.11.1.0.1.el5.
345 Description: New acc-sm script and various test-framework enhancments.
349 Description: Previously landed patch was reverted.
353 Description: Update SLES10 kernel to 2.6.16.60-0.42.8 (patchless)
357 Description: Add ldiskfs maxdirsize mount option.
361 Description: Fix mkfs.lustre support for 16TB LUNs.
365 Description: Print mmp_check_interval
366 Details : print mmp_check_interval and make it possible to abort
367 mount operation in case it takes too long.
371 Description: Add cascading_rw to lustre mpi tests.
375 Description: Store SOM data in EA.
376 Details : Also several SOM related cleanups.
380 Description: Add refcount to md_open_data
384 Description: Use refcount on lov_request_set to prevent memory corruption.
388 Description: Statahead: use parent dir's i_mutex for synchronization.
392 Description: loadgen improvements.
396 Description: Enable peer health detection in ksocklnd and ko2iblnd.
400 Description: Control DCACHE_LUSTRE_INVALID flag with MDS_INODE_LOOKCUP lock
404 Description: Fix race on obd_devcie:md_stats
408 Description: Fix SLES 11 client builds.
409 Details : Resolved uninitalized pointer gcc warnings.
413 Description: Allow non-root access for "lfs check".
414 Details : Added a check in obd_class_ioctl() for OBD_IOC_PING_TARGET.
418 Description: Clear imp_force_reconnect correctly in ptlrpc_connect_interpret()
420 Severity : enhancement
422 Description: Update kernel to RHEL5.4 2.6.18-164.6.1.el5 and
423 OEL5 2.6.18-164.6.1.0.1.el5(Both in-kernel OFED enabled).
427 Description: Sanity test_27p fails.
428 Details : OBD_FAIL_OST_ENOSPC has been moved to filter_statfs().
429 Sanity tests 27 were cleaned up and improved.
433 Description: Unify req format on client/servers
434 Details : Use new req_capsule API [almost] everywhere instead of old PTLRPC
435 buffers and swabbers approach..
440 Description: drop join file stuff.
442 Severity : enhancement
444 Description: Add 2.6.27 and SLES11 (patchless client) support
448 Description: Fix in ptlrpc_expire_one_request() to print the signed time
451 Severity : enhancement
453 Description: Build kernels (RHEL5, OEL5 and SLES10/11) using the vendor's own
456 Severity : enhancement
458 Description: Remove set_info(KEY_UNLINKED) from MDS/OSC
460 Severity : enhancement
462 Description: correctly handle big reply message.
463 Details : send LNet event if reply is bigger then buffer and adjust this buffer
468 Description: Drop unnecessary __GFP_NOMEMALLOC flag from filter_get_page()
470 Severity : enhancement
472 Description: Add support for OEL5.
476 Description: Conf-sanity.sh 50g test - deactivated OST should not cause a panic.
478 Severity : enhancement
480 Description: "weak-modules" support
481 Details : Implement "weak-modules" support which enables kernel modules
482 to be used with any kernel that implements the same kABI. In
483 order to achieve this modules are now installed in
484 /lib/modules/$(uname -r)/updates/kernel on all distributions.
488 Description: client could not reconnect to OST because of an active request.
489 Details : The client now retries to reconnect to the same server, if a
490 connect request failed with EBUSY or -EAGAIN. Test case 35b
491 conf-sanity.sh has been created.
495 Description: Don't allow make backward step on assignin osc next id.
496 Details : race between allocation next id and ll_sync thread can be cause
497 of set wrong osc next id and can be kill valid ost objects.
501 Description: Changes in raid5-large-io-rhel5.patch to calculate sectors properly
505 Description: Increase the default BLK_DEF_MAX_SECTORS value for RHEL5 and SLES11
509 Description: Error handling in osc_statfs_interpret() has been improved.
510 Details : Check in osc_statfs_interpret() for EBADR.
514 Description: Do not send statfs() requests to OSTs disabled by administrator.
515 Details : Check in lov_prep_statfs_set() for non-NULL ltd_exp.
517 Severity : enhancement
519 Description: Added sanity-quota test to for lfs(1) to work with numeric uid/gid.
521 Severity : enhancement
523 Description: Update kernel to RHEL5 2.6.18-128.7.1.el5.
525 Severity : enhancement
527 Description: Enhancement for lfs(1) command to use numeric uid/gid.
531 Description: Improved cleanup procedure for conf-sanity test_46a.
535 Description: Do not start bulk transfer, if client tried to reconnect.
536 Details : Do not start bulk transfer, if client tried to reconnect,
537 but failed due to active rpcs. Check for exp_abort_active_req
538 has been implemented.
542 Description: Prevent inconsistences between linux and lustre mount structures.
543 Details : Wait indefinitely in server_wait_finished() until mnt_count drops.
544 Make the sleep interruptible.
546 Severity : enhancement
548 Description: provide server to client comms path
549 Details : server to client communications path via new LDLM_SET_INFO rpc
553 Description: truncate starts GFP_FS allocation under transaction causing deadlock
554 Details : ldiskfs_truncate calls grab_cache_page which may start page
555 allocation under an open transaction. This may lead to
556 calling prune_icache with consequent lustre reentrance.
560 Description: send correctly lsm on open replay
561 Details : MDS is trust to LSM size on replay open, but client can set wrong size
566 Description: enable client interface failover
567 Details : When a child reconnects from another NID, properly update export
568 nid hash position and ldlm reverse import.
570 Severity : enhancement
572 Description: Communicate OST degraded/readonly state via statfs to MDS
573 Details : Flags in the statfs returned from OSTs indicate whether the
574 OST is in a degraded RAID state, or if the filesystem has
575 turned read-only after a filesystem error is detected.
580 Description: don't panic if EPROTO was hit when reading symlink
581 Details : correctly handling request reference in error cases.
583 Severity : enhancement
585 Description: Add LustreNetLink, a kernel-userspace communcation path.
587 Severity : enhancement
589 Description: Update kernel to SLES10 SP2 2.6.16.60-0.39.3.
594 Description: access to llog context before init.
595 Details : move handling CATALOGS file at osc layer and forbid access to llog
599 Frequency : rare, only when client is writing to OST on local node
601 Description: Avoid deadlock for local client writes
602 Details : Use new OBD_BRW_MEMALLOC flag to notify OST about writes in the
603 memory freeing context. This allows OST threads to set the
604 PF_MEMALLOC flag on task structures in order to allocate memory
605 from reserved pools and complete IO.
606 Use GFP_HIGHUSER for OST allocations for non-local client writes,
607 so that the OST threads generate memory pressure and allow
608 inactive pages to be reclaimed.
610 Severity : enhancement
612 Description: Update kernel to RHEL5.3 2.6.18-128.1.14.el5.
617 Description: lock ordering violation between &cli->cl_sem and _lprocfs_lock
618 Details : move ldlm namespace creation in setup phase to avoid grab
619 _lprocfs_lock with cli_sem held.
623 Description: Temporarily disable grant shrink.
624 Details : Disable the feature for debugging.
627 Frequency : only when formatting test filesystems
629 Description: Unable to run several mkfs.lustre on loop devices at the same time
630 Details : mkfs.lustre returns error 256 on the concurrent loop devices
631 formatting. The solution is to proper handle the error.
633 Severity : enhancement
635 Description: Update kernel to RHEL5.3 2.6.18-128.1.6.el5.
637 Severity : enhancement
639 Description: Update kernel to SLES10 SP2 2.6.16.60-0.37.
644 Description: resolve race between obd_disconnect and class_disconnect_exports
645 Details : if obd_disconnect will be called to already disconnected export he
646 forget release one reference and osc module can't unloaded.
648 Severity : enhancement
650 Description: Allow tuning service thread via /proc
651 Details : For each service a new
652 /proc/fs/lustre/{service}/*/thread_{min,max,started} entry is
653 created that can be used to set min/max thread counts, and get the
654 current number of running threads.
659 Descriptoin: don't return error if have partially created objects for file.
660 Details : lov_update_create_set uses set->set_success as index for created
661 objects, so if some requests will be failed, they will have hole
662 at end of array and qos_shrink_lsm for allocate correct lsm.
664 Severity : enhancement
666 Description: Update OFED support to 1.4.1
671 Description: Reduce small size read RPC
672 Details : Set read-ahead limit for every file and only do read-ahead when
673 available read-ahead pages are bigger than 1M to avoid small size
676 Severity : enhancement
678 Description: move AT tunable parameters for more consistent usage
679 Details : add AT tunables under /proc/sys/lustre, add to conf_param parsing
681 Severity : enhancement
683 Description: add lazystatfs mount option to allow statfs(2) to skip down OSTs
684 Details : allow skip disconnected ost for send statfs request and hide error
690 Description: fix racy locking of mballoc block bitmaps causing BUG
691 Details : The locking of the mballoc buddy bitmap and the in-memory
692 block bitmap was using two different spin locks in some
693 cases. This made it possible to incorrectly access the
694 mballoc bitmap while another process was modifying it,
695 causing a sanity assertion to fail. While no on-disk corruption
696 was reported, there was some risk of this happening.
699 Frequency : rare, on llog test 6
701 Descriptoin: don't allow connect to already connected import
702 Details : allowing connect to already connected import is hide connecting problem.
707 Descriptoin: allow kill process which wait statahead result
708 Details : for some reasons 'ls' can stick in waiting result from statahead,
709 in this case need way for kill this process.
711 Severity : enhancement
713 Description: Add state history info file, enhance import info file
714 Details : Track import connection state changes in a new osc/mdc proc file;
715 add overview-type data to the osc/mdc import proc file.
717 Severity : enhancement
719 Description: MDS create should not wait for statfs RPC while holding DLM lock.
721 Severity : enhancement
723 Description: Update to RHEL5U3 kernel-2.6.18-128.1.1.el5.
728 Descriptoin: OST grant too much space to client even there are not enough space.
729 Details : Client will shrink its grant cache to OST if there are no write
730 activity over 6 mins (GRANT_SHRINK_INTERVAL), and OST will retrieve
731 this grant cache if there are already not enough avaible space
732 (left_space < total_clients * 32M).
735 Frequency : start MDS on uncleanly shutdowned MDS device
737 Descriptoin: ll_sync thread stay in waiting mds<>ost recovery finished
738 Details : stay in waiting mds<>ost recovery finished produce random bugs
739 due race between two ll_sync thread for one lov target. send
740 ACTIVATE event only if connect realy finished and import have
744 Frequency : rare, connect and disconnect target at same time
746 Descriptoin: ASSERTION(atomic_read(&imp->imp_inflight) == 0
747 Details : don't call obd_disconnect under lov_lock. this long time
748 operation and can block ptlrpcd which answer to connect request.
753 Descriptoin: don't lose wakeup for imp_recovery_waitq
754 Details : recover_import_no_retry or invalidate_import and import_close can
755 both sleep on imp_recovery_waitq, but we was send only one wakeup
759 Frequency : always with long access acl
761 Descriptoin: mds can't pack reply with long acl.
762 Details : mds don't control size of acl but they limited by reint/getattr
765 Severity : enhancement
767 Description: Update to SLES10 kernel-2.6.16.60-0.33.
769 Severity : enhancement
771 Description: Update to RHEL5 kernel-2.6.18-92.1.22.el5.
774 Frequency : start MDS on uncleanly shutdowned MDS device
776 Descriptoin: aborting recovery hang on MDS
777 Details : don't throttle destroy RPCs for the MDT.
780 Frequency : on remount
782 Description: external journal device not working after the remount
783 Details : clear dev_rdonly flag for external journal devices in
789 Description: shutdown vs evict race
790 Details : client_disconnect_export vs connect request race.
791 if client will evicted at this time - we start invalidate
792 thread without referece to import and import can be freed
796 Frequency : rare, need acl's on inode.
798 Description: client can't handle ost additional correctly
799 Details : if ost was added after client connected to mds client can have
800 hit lnet_try_match_md ... to big messages to wide striped files.
801 in this case need teach client to handle config events about add
802 lov target and update client max ea size at that event.
804 Severity : enhancement
806 Description: Changelogs
807 Details : Changelogs are a lightweight mechanism to track filesystem
808 metadata and namespace changes. The changelog is recorded
809 permanently on the MDTs, and is periodically "consumed" / purged
810 when records are no longer needed.
812 Severity : enhancement
814 Description: compact fld format with extents
815 Details : Store range of seq rather than every seq in FLD. Seq
816 controller update FLD rather than clients. In Case of CMD, mdt0
817 has FLD, all other metadata server act as non persistent proxy
818 for FLD queries and cache fld entries in fld cache.
823 Description: don't skip ost target if they assigned to file
824 Details : Drop slow OSCs if we can, but not for requested start idx.
825 This means "if OSC is slow and it is not the requested
826 start OST, then it can be skipped, otherwise skip it only
827 if it is inactive/recovering/out-of-space.
831 Description: more cleanup in mds_lov
832 Details : not send LOV EA under replay, we can't know about they size at this
833 time. Don't allow client connect to mds before any ost connected,
834 for avoid problems with LOV EA size and returning EIO to client.
836 Severity : enhancement
838 Description: Interoperability at server side (Disk interoperability)
840 Severity : enhancement
842 Description: Update to RHEL5 kernel-2.6.18-92.1.17.el5.
844 Severity : enhancement
846 Description: Update to SLES10 SP2 kernel-2.6.16.60-0.31.
848 Severity : enhancement
850 Description: New client IO stack (CLIO).
852 Severity : enhancement
854 Description: Commit on sharing. Eliminate inter-client dependencies between
855 uncommitted transactions by doing transaction commits.
856 Thereby clients may recovery independently.
859 Frequency : Create a symlink file with a very long name
861 Description: ldlm_cancel_pack()) ASSERTION(max >= dlm->lock_count + count)
862 Details : If there is no extra space in the request for early cancels,
863 ldlm_req_handles_avail() returns 0 instead of a negative value.
865 Severity : enhancement
867 Description: Add /proc entry for import status
868 Details : The mdc, osc, and mgc import directories now have
869 an import directory that contains useful import data for debugging
872 Severity : enhancement
874 Description: Re-disable certain /proc logging
875 Details : Enable and disable client's offset_stats, extents_stats and
876 extents_stats_per_process stats logging on the fly.
879 Frequency : Only on FC kernels 2.6.22+
881 Description: oops in statahead
882 Details : Do not drop reference count for the dentry from VFS when lookup,
883 VFS will do that by itself.
885 Severity : enhancement
887 Description: Generic /proc file permissions
888 Details : Set /Proc file permissions in a more generic way to enable non-
889 root users operate on some /proc files.
893 Description: Hitting mdc_commit_close() ASSERTION
894 Details : Properly handle request reference release in
895 ll_release_openhandle().
899 Description: quota recovery deadlock during mds failover
900 Details : This patch includes att18982, att18236, att18237 in bz14840.
902 1. fix osts hang when mds does failover with quotaon
903 2. prevent watchdog storm when osts threads wait for the
908 Frequency : only patchless client
909 Description: add workaround for race between add/remove dentry from hash
911 Severity : enhancement
913 Description: Allow OST glimpses to return PW locks
917 Description: LBUG when llog conf file is full
918 Details : When llog bitmap is full, ENOSPC should be returned for plain log.
922 Description: Prevent import from entering FULL state when server in recovery
926 Description: service mount cannot take device name with ":"
927 Details : Only when device name contains ":/" will mount treat it as
933 Description: replace ptlrpcd with the statahead thread to interpret the async
934 statahead RPC callback
938 Frequency : on recovery
939 Description: I/O failures after umount during fail back
940 Details : if client reconnected to restarted server we need join to recovery
941 instead of find server handler is changed and process self
942 eviction with cancel all locks.
944 Severity : enhancement
946 Description: Update to RHEL5 kernel-2.6.18-92.1.10.el5.
948 Severity : enhancement
950 Description: Update to SLES10 SP2 kernel-2.6.16.60-0.27.
952 Severity : enhancement
954 Description: Upcall on Lustre log has been dumped
955 Details : Allow for a user mode script to be called once a Lustre log has
956 been dumped. It passes the filename of the dumped log to the
957 script, the location of the script can be specified via
958 /proc/sys/lnet/debug_log_upcall.
963 Description: avoid idr_remove called for id which is not allocated.
964 Details : Move assigment s_dev for clustered nfs to end of initialization,
965 to avoid problem with error handling.
970 Description: avoid Already found the key in hash [CONN_UNUSED_HASH] messages
971 Details : When connection is reused this not moved from CONN_UNUSED_HASH
972 into CONN_USED_HASH and this prodice warning when put connection
973 again in unused hash.
976 Severity : enhancement
978 Description: File striping can now be set to use an arbitrary pool of OSTs.
980 Severity : enhancement
982 Description: Export bytes_read/bytes_write count on OSC/OST.
986 Description: Early reply size mismatch, MGC loses connection
987 Details : Apply the MGS_CONNECT_SUPPORTED mask at reconnect time so
988 the connect flags are properly negotiated.
993 Description: quotas are not honored with O_DIRECT
994 Details : all writes with the flag O_DIRECT will use grants which leads to
995 this problem. Now using OBD_BRW_SYNC to guard this.
999 Description: add quota statistics
1000 Details : 1. sort out quota proc entries and proc code.
1001 2. add quota statistics
1003 Severity : enhancement
1005 Description: enable quota support for HEAD.
1009 Description: Properly propagate oinfo flags from lov to osc for statfs
1010 Details : restore missing copy oi_flags to lov requests.
1012 Severity : enhancement
1014 Description: Add man pages for llobdstat(8), llstat(8), plot-llstat(8),
1015 : l_getgroups(8), lst(8), routerstat(8)
1016 Details : included man pages for llobdstat(8), llstat(8),
1017 : plot-llstat(8), l_getgroups(8), lst(8), routerstat(8)
1019 Severity : enhancement
1021 Description: configure's --enable-quota should check the
1022 : kernel .config for CONFIG_QUOTA
1023 Details : configure is terminated if --enable-quota is passed but
1024 : no quota support is in kernel
1028 Description: Remove portals compatibility
1029 Details : Remove portals compatibility, not interoperable with releases
1034 Description: Resolve device initialization race
1035 Details : Prevent proc handler from accessing devices added to the
1036 obd_devs array but yet be intialized.
1038 Severity : enhancement
1040 Description: Update to SLES10 SP2 kernel-2.6.16.60-0.23.
1042 Severity : enhancement
1044 Description: Update to RHEL5 kernel-2.6.18-92.1.6.el5.
1049 Description: Using wrong pointer in osc_brw_prep_request
1050 Details : Access to array[-1] can produce panic if kernel compiled with
1051 CONFIG_PAGE_ALLOC enabled
1055 Description: Client runs out of low memory
1056 Details : Consider only lowmem when counting initial number of llap pages
1060 Description: *optional* service tags registration
1061 Details : if the "service tags" package is installed on a Lustre node
1062 When the filesystem is mounted, a local-node service tag will
1063 be created. See http://inventory.sun.com/ for more information
1064 about the Service Tags asset management system.
1068 Description: Kernel BUG tries to release flock
1069 Details : Lustre does not destroy flock lock before last reference goes
1070 away. So always drop flock locks when client is evicted and
1071 perform unlock regardless of successfulness of speaking to MDS.
1075 Description: add refcount protection for osc callbacks, avoid panic on shutdown
1079 Description: sanity test 65a fails if stripecount of -1 is set
1080 Details : handle -1 striping on filesystem in ll_dirstripe_verify
1085 Description: ASSERTION(CheckWriteback(page,cmd)) failed
1086 Details : badly clear PG_Writeback bit in ll_ap_completion can produce false
1089 Severity : enhancement
1091 Description: Update to RHEL5 kernel-2.6.18-53.1.21.el5.
1095 Description: do not process already freed flock
1096 Details : flock can possibly be freed by another thread before it reaches
1097 to ldlm_flock_completion_ast.
1101 Description: LBUG during stress test
1102 Details : Need properly lock accesses the flock deadlock detection list.
1106 Description: oops in page fault handler
1107 Details : kernel page fault handler can return two special 'pages' in error
1108 case, don't try dereference NOPAGE_SIGBUS and NOPAGE_OMM.
1112 Description: timeout with invalidate import.
1113 Details : ptlrpcd_check call obd_zombie_impexp_cull and wait request which
1114 should be handled by ptlrpcd. This produce long age waiting and
1115 -ETIMEOUT ptlrpc_invalidate_import and as result LASSERT.
1117 Severity : enhancement
1119 Description: Update to RHEL5 kernel-2.6.18-53.1.19.el5.
1123 Description: enable MGS and MDT services start separately
1124 Details : add a 'nomgs' option in mount.lustre to enable start a MDT with
1125 a co-located MGS without starting the MGS, which is a complement
1126 to 'nosvc' mount option.
1130 Frequency : after recovery
1131 Description: precreate to many object's after del orphan.
1132 Details : del orphan st in oscc last_id == next_id and this triger growing
1133 count of precreated objects. Set flag LOW to skip increase count
1134 of precreated objects.
1138 Frequency : rare, on clear nid stats
1139 Description: ASSERTION(client_stat->nid_exp_ref_count == 0)
1140 Details : when clean nid stats sometimes try destroy live entry,
1141 and this produce panic in free.
1145 Description: Stack overflow during MDS log replay
1146 ease stack pressure by using a thread dealing llog_process.
1150 Description: wait until IO finished before start new when do lock cancel.
1151 Details : VM protocol want old IO finished before start new, in this case
1152 need wait until PG_writeback is cleared until check dirty flag and
1153 call writepages in lock cancel callback.
1155 Severity : enhancement
1157 Description: using special macro for print time and cleanup in includes.
1161 Description: mds_mfd_close() ASSERTION(rc == 0)
1162 Details : In mds_mfd_close(), we need protect inode's writecount change
1163 within its orphan write semaphore to prevent possible races.
1167 Description: Obsolete CURRENT_SECONDS and use cfs_time_current_sec() instead.
1171 Frequency : rare, on shutdown ost
1172 Description: don't hit live lock with umount ost.
1173 Details : shrink_dcache_parent can be in long loop with destroy dentries,
1174 use shrink_dcache_sb instead.
1178 Description: don't panic with use echo client
1179 Details : echo client pass NULL as client nid pointer and this produce null
1180 pointer dereference.
1184 Description: fix build on ppc32
1185 Details : compile code with -m64 flag produce wrong object file for ppc32.
1189 Description: add message levels for liblustreapi
1193 Description: fix for occasional failure case of -ENOSPC in recovery-small tests
1194 Details : Move the 'good_osts' check before the 'total_bavail' check. This
1195 will result in an -EAGAIN and in the exit call path we call
1196 alloc_rr() which will with increasing aggressiveness attempt to
1197 aquire precreated objects on the minimum number of required OSCs.
1201 Description: Use old size assignment to avoid deadlock
1202 Details : Reverts the changes in bugs 2369 and bug 14138 that introduced
1203 the scheduling while holding a spinlock. We do not need locking
1204 for size in ll_update_inode() because size is only updated from
1205 the MDS for directories or files without objects, so there is no
1206 other place to do the update, and concurrent access to such inodes
1207 are protected by the inode lock.
1211 Description: resolve "_IOWR redefined" build error on SLES10
1215 Description: dump the memory debugging after all modules are unloaded to
1216 suppress false negative in conf_sanity test 39
1218 Severity : enhancement
1220 Description: build kernel-ib packages for OFED 1.3 in our release cycle
1225 Description: fix SLES kernel versioning
1226 Details : the kernel version for our SLES 10 kernel did not include a "-"
1227 before the "smp" at the end. while this was not a problem in
1228 general, it did mean that software trying to use the kernel
1229 version to try to detect a vendor specific kernel would fail.
1230 this was most evident by the OFED build scripts.
1234 Description: Don't update lov_desc members until making sure they are valid
1235 Details : When updating lov_desc members via proc fs, need fix their
1236 validities before doing the real update.
1240 Description: don't put request into delay list while invalidate in flight.
1241 Details : ptlrpc_delay_request sometimes put in delay list while invalidate
1242 import in flight. this produce timeout for invalidate and sometimes
1243 can cause stale data.
1247 Frequency : on ppc only
1248 Description: not convert ost objects for directory because it's not exist.
1249 Details : ll_dir_getstripe assume dirrectory has ost objects but this wrong.
1253 Description: Add FMODE_EXEC file flag for SLES10 SP1 kernel.
1255 Severity : enhancement
1257 Description: Update to support 2.6.22.14 vanilla kernel.
1261 Frequency : rare, on recovery
1262 Description: read procfs can produce deadlock in some situation
1263 Details : Holding lprocfs lock which send rpc can produce block for destroy
1264 obd objects and this also block reconnect with -EALREADY. This
1265 isn't fix all lprocfs bugs - but make it rare.
1267 Severity : enhancement
1269 Description: Update kernel to RHEL5 2.6.18-53.1.14.el5.
1272 Frequency : frequent on X2 node
1274 Description: mdc_set_open_replay_data LBUG
1275 Details : Set replay data for requests that are eligible for replay.
1279 Description: lustre_mgs: operation 101 on unconnected MGS
1280 Details : When MGC is disconnected from MGS long enough, MGS will evict the
1281 MGC, and late on MGC cannot successfully connect to MGS and a lot
1282 of the error messages complaining that MGS is not connected.
1286 Frequency : on network error
1287 Description: panic with double free request if network error
1288 Details : mdc_finish_enqueue is finish request if any network error ocuring,
1289 but it's true only for synchronus enqueue, for async enqueue
1290 (via ptlrpcd) this incorrect and ptlrpcd want finish request
1293 Severity : enhancement
1295 Description: client-side metadata stat-ahead during readdir(directory readahead)
1296 Details : perform client-side metadata stat-ahead when the client detects
1297 readdir and sequential stat of dir entries therein
1300 Frequency : on start mds
1302 Description: Implement get_info(last_id) in obdfilter.
1305 Frequency : occasional
1307 Description: Correctly check stale fid, not start epoch if ost not support SOM
1308 Details : open with flag O_CREATE need set old fid in op_fid3 because
1309 op_fid2 was overwritten with new generated fid, but mds can answer
1310 with one of these two fids and both is not stale. Setattr
1311 incorrectly started an epoch and assume will be called
1312 done_writing, but without SOM done_writing ever being called.
1315 Frequency : rare, depends on device drivers and load
1317 Description: MDS or OSS nodes crash due to stack overflow
1318 Details : Code changes in 1.8.0 increased the stack usage of some functions.
1319 In some cases, in conjunction with device drivers that use a lot
1320 of stack the MDS (or possibly OSS) service threads could overflow
1321 the stack. One change which was identified to consume additional
1322 stack has been reworked to avoid the extra stack usage.
1325 Frequency : occasional
1327 Description: Do not fail import if osc_interpret_create gets -EAGAIN
1328 Details : If osc_interpret_create got -EAGAIN it immediately exits and
1329 wakeup oscc_waitq. After wakeup oscc_wait_for_objects call
1330 oscc_has_objects and see OSC has no objests and call
1331 oscc_internal_create to resend create request.
1333 Severity : enhancement
1335 Description: Update to SLES10 SP1 latest kernel-2.6.16.54-0.2.5.
1337 Severity : enhancement
1339 Description: Update to RHEL5 latest kernel-2.6.18-53.1.13.el5.
1342 Frequency : very rare
1344 Description: Fix replay if there is an un-replied request and open
1345 Details : In some cases, older replay request will revert the
1346 mcd->mcd_last_xid on MDS which is used to record the client's
1347 latest sent request.
1349 Severity : enhancement
1351 Description: Update to RHEL5 latest kernel-2.6.18-53.1.6.el5.
1353 Severity : enhancement
1355 Description: Add rhel5 support to HEAD.
1357 Serverity : enhancement
1359 Description: Update RHEL4 kernel to 2.6.9-67.0.4.
1364 Description: Don't allow skipping OSTs if index has been specified.
1365 Details : Don't allow skipping OSTs if index has been specified, make
1366 locking in internal create lots better.
1370 Description: LBUG in ptlrpc_check_set() bad phase ebc0de00
1371 Details : access to bitfield in structure is always rounded to long
1372 and this produce problem with not atomic change any bit.
1376 Description: Lustre make rpms failed.
1377 Details : Remove ldiskfs spec file to avoids rpmbuild be confused when
1378 builds Lustre rpms from tarball.
1381 Frequency : rare on shutdown ost
1383 Description: If llog cancel was not send before clean_exports phase, this can
1384 produce deadlock in llog code.
1385 Details : If llog thread has last reference to obd and call class_import_put
1386 this produce deadlock because llog_cleanup_commit_master wait when
1387 last llog_commit_thread exited, but this never success because was
1388 called from llog_commit_thread.
1392 Description: allow userland application know is lost one of stripes.
1393 Details : fill lvb_blocks with error code on ost and return it to
1394 application if error flag found.
1398 Description: NULL lov_tgts causing MDS oops
1399 Details : more safe checks for NULL lov_tgts for avoid oops.
1401 Severity : enhancement
1403 Description: Update to RHEL4 latest kernel-2.6.9-67.0.1.EL.
1407 Descriptoin: make lov_create() will not stuck in obd_statfs_rqset()
1408 Details : If an OST is down the MDS will hang indefinitely in
1409 obd_statfs_rqset() waiting for the statfs data. While for
1410 MDS QOS usage of statfs, it should not stuck in waiting.
1412 Severity : enhancement
1414 Description: remote_acl support
1415 Details : Support ACL-based permission check for remote user.
1416 Support setfacl/getfacl for remote user with the utils
1417 "lfs {l,r}{s,g}etfacl" which follow the same parameter format as
1418 the system "{s,g}etfacl" utils.
1420 Severity : enhancement
1422 Description: Update to RHEL4 U6 kernel-2.6.9-67.EL.
1424 Severity : enhancement
1426 Description: Update to RHEL5 latest kernel-2.6.18-53.1.4.el5.
1430 Description: make mgs_setparam() handle fsname containing dash
1431 Details : fsname containing a dash does not work with lctl conf_param
1433 Severity : enhancement
1435 Description: Update to SLES10 SP1 latest kernel-2.6.16.54-0.2.3.
1437 Severity : enhancement
1439 Description: Update to RHEL5 Update-1 kernel 2.6.18-53.el5.
1443 Frequency : rare, at shutdown
1444 Description: access already free / zero obd_namespace.
1445 Details : if client_disconnect_export was called without force flag set,
1446 and exist connect request in flight, this can produce access to
1447 NULL pointer (or already free pointer) when connect_interpret
1448 store ocd flags in obd_namespace.
1452 Frequency : only at startup
1453 Description: not alloc memory with spinlock held.
1454 Details : allocation memory with GFP_KERNEL can produce sleep deadlock,
1455 if any spinlock held.
1457 Severity : enhancement
1459 Description: make lustre randomly fail allocating memory
1460 Details : Make lustre randomly failed allocating memory for testing purpose.
1462 Severity : enhancement
1464 Description: lost problems with lov objid file
1465 Details : Fixes some scability and access to not inited memory problems
1466 in work with lov objdid file.
1471 Description: lfs find does not continue on file error
1472 Details : Continue other files processing when a file/dir is absent.
1476 Description: Inconsistent usage of lustre_pack_reply()
1477 Details : Standardize the usage of lustre_pack_reply() such that it
1478 always generate a CERROR on failure.
1481 Frequency : occasional
1483 Description: MDS hang or stay in waiting lock
1484 Details : If client receive lock with CBPENDING flag ldlm need send lock
1485 cancel as separate rpc, to avoid situation when cancel request
1486 can't processed due all i/o threads stay in wait lock.
1490 Description: Update to RHEL5 kernel 2.6.18-8.1.15.el5.
1494 Description: Update to SLES10 SP1 kernel 2.6.16.53-0.16
1498 Description: Update to SLES9 kernel-2.6.5-7.287.3.
1502 Description: Update to RHEL4 latest kernel.
1504 Severity : enhancement
1506 Description: Build SLES10 patchless client fails
1507 Details : The configure was broken by run ./configure with
1508 --with-linux-obj=.... argument for patchless client. When the
1509 configure use --with-linux-obj, the LINUXINCLUDE= -Iinclude
1510 can't search header adequately. Use absolute path such as
1511 -I($LINUX)/include instead.
1515 Description: interrupt oig_wait produce painc on resend.
1516 Details : brw_redo_request can be used for resend requests from ptlrpcd and
1517 private set, and this produce situation when rq_ptlrpcd_data not
1518 copyed to new allocated request and triggered LBUG on assert
1519 req->rq_ptlrpcd_data != NULL. But this member used only for wakeup
1520 ptlrpcd set if request is changed and can be safety changed to use
1525 Description: LASSERT_{REQ,REP}SWAB macros are buggy
1526 Details : If SWAB_PARANOIA is disabled, the LASSERT_REQSWAB and
1527 LASSERT_REPSWAB macros become no-ops, which is incorrect. Drop
1528 these macros and replace them with their difinitions instead.
1532 Description: Update kernel patches for SLES10 2.6.16.53-0.8.
1533 Details : Update which_patch & target file for SLES10 latest kernel.
1536 Description: Remove client patches from SLES 10 kernel.
1537 Details : This causes SLES 10 clients to behave as patchless clients
1538 even on a Lustre-patched (server) kernel.
1540 Severity : enhancement
1542 Description: self-adjustable client's lru lists
1543 Details : use adaptive algorithm for managing client cached locks lru
1544 lists according to current server load, other client's work
1545 pattern, memory activities, etc. Both, server and client
1546 side namespaces provide number of proc tunables for controlling
1549 Severity : enhancement
1551 Description: light-weight GSS support
1552 Details : Support krb5n and krb5a mode, which keep Kerberos 5 authentication
1553 and reduce performance overhead.
1555 Severity : enhancement
1557 Description: Linux keyring support
1558 Details : Support using service of Linux keyring for Lustre GSS internal
1559 context refresh/cache mechanism.
1563 Description: Fix errors in lfs documentation
1564 Details : Fixes man pages
1568 Description: don't use GFP_* in generic Lustre code.
1569 Details : Use cfs_alloc_* functions and CFS_* flags for code portability.
1573 Description: obdclass is limited by single OBD_ALLOC(idarray)
1574 Details : replace OBD_ALLOC/OBD_FREE with OBD_VMALLOC/OBD_VFREE
1578 Description: warnings with build patchless client with vanila 2.6.19 and up
1579 Details : change old ctl_table style and replace ctl_table/ctl_table_header
1580 with cfs_sysctl_table_t/cfs_sysctl_table_header_t
1584 Frequency : Only for SLES
1585 Description: sanity_quota fail test_1
1586 Details : There are multiple occurences of $TSTUSR in SLES's /etc/group
1587 file, which makes TSTID[2] inunique.
1591 Frequency : Only for SLES9
1592 Description: Kernel patches for SLES9 2.6.5-7.286 kernel
1593 Details : Update target/ChangeLog/which_patch .
1597 Frequency : Only for test_10 in sanity-quota.sh
1598 Description: a bug in quota test code
1599 Details : checking a test flag in a wrong place
1603 Frequency : Only for test_13 in sanity-quota.sh
1604 Description: a bug in quota test script
1605 Details : shouldn't check overfull allocated quota
1609 Description: jbd statistics
1610 Details : Port older jbd statistics patch for sles10
1612 Severity : enhancement
1614 Description: Add printing inode info into message about error in writepage.
1619 Description: reply_lock_interpret crash due to race with it and lock cancel.
1620 Details : Do not replay locks that are being cancelled. Do not reference
1621 locks by their address during replay, just by their handle.
1625 Frequency : When flocks are used.
1626 Description: assertion failure in ldlm_cli_enquque_fini for non NULL lock.
1627 Details : Flock locks might destroy just granted lock if it could be merged
1628 with another existing flock, this is done in completion handler,
1629 so teach ldlm_cli_enquque_fini that this is a valid case for
1635 Description: Oops in read and write path when failing to allocate lock.
1636 Details : Check if lock allocation failed and return error back.
1640 Description: lstripe command fails for valid OST index
1641 Details : The stripe offset is compared to lov->desc.ld_tgt_count
1642 instead of lov->desc.ld_active_tgt_count.
1646 Description: sanity.sh failed test 103
1647 Details : RHEL mis-interpret setfacl "-X" param, so we won't test setfacl
1652 Description: df doesn't work properly if diskfs blocksize != 4K
1653 Details : Choose biggest blocksize of OST's as the LOV's blocksize.
1657 Description: lfs find on -1 stripe looping in lsm_lmm_verify_common()
1658 Details : Avoid lov_verify_lmm_common() on directory with -1 stripe count.
1660 Severity : enhancement
1662 Description: Adaptive timeouts
1663 Details : RPC timeouts adapt to changing server load and network
1664 conditions to reduce resend attempts and improve recovery time.
1668 Description: llapi_file_create() does not allow some changes
1669 Details : add llapi_file_open() that allows specifying the mode and
1670 open flags, and also returns an open file handle.
1674 Description: merge and cleanup kernel patches.
1675 Details : Remove mnt_lustre_list in vfs_intent-2.6-rhel4.patch.
1679 Description: Add journal checksum support.(Kernel part)
1680 Details : The journal checksum feature adds two new flags i.e
1681 JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT and
1682 JBD2_FEATURE_COMPAT_CHECKSUM. JBD2_FEATURE_CHECKSUM flag
1683 indicates that the commit block contains the checksum for
1684 the blocks described by the descriptor blocks. Now commit
1685 record can be sent to disk without waiting for descriptor
1686 blocks to be written to disk. This behavior is controlled
1687 using JBD2_FEATURE_ASYNC_COMMIT flag.
1691 Description: OSS needs mutliple precreate threads
1692 Details : Add ability to start more than one create thread per OSS.
1696 Description: Sanity.sh test_65k failed: setstripe should have succeeded
1697 Details : execute lfs setstripe on client
1701 Description: mds_obd_create error creating tmp object
1702 Details : When the user sets quota on root, llog will be affected and can't
1703 create files and write files.
1706 Frequency : Always on ia64 patchless client, and possibly others.
1708 Description: Add EXPORT_SYMBOL check for node_to_cpumask symbol.
1709 Details : This allows the patchless client to be loaded on architectures
1710 without this export.
1714 Description: RedHat Update kernel for RHEL5
1715 Details : Add the kernel config file for RHEL5.
1719 Description: RedHat Update kernel for RHEL5
1720 Details : Modify the kernel config file more closer RHEL5.
1724 Description: Build failure against Centos5 (RHEL5)
1725 Details : Define PAGE_SIZE when it isn't present.
1729 Description: test_7 on sanity-quota.sh can't be used on separate machines
1730 Details : add facet to handle it
1734 Description: "ll_intent_file_open()) lock enqueue: err: -13" with nfs
1735 Details : with NFS, the anon dentry's parent was set to itself in
1736 d_alloc_anon(), so in MDS, we use rec->ur_fid1 to find the
1737 corresponding dentry other than use rec->ur_name.
1739 Severity : enhancement
1741 Description: lfs setstripe enhancement
1742 Details : Make lfs setstripe understand 'k', 'm' and 'g' for stripe size.
1746 Description: enable data checksumming by default
1747 Details : enable checksum by default, allow --disable-checksum
1748 configure option and "-o nochecksum" mount option. Checksums
1749 can also be disabled at runtime via $LPROC/osc/*/checksum_pages.
1753 Description: lustre support for RHEL5
1754 Details : Add support for RHEL5.
1758 Description: Client eviction due to failover config
1759 Details : after a connection loss, the lustre client should attempt to
1760 reconnect to the last active server first before trying the
1761 other potential connections.
1765 Description: when mds and osts use different quota unit(32bit and 64bit),
1766 quota will be released repeatly.
1767 Details : void sending multiple quota reqs to mds, which will keep the
1768 status between the reqs.
1772 Description: osts not allocated evenly to files
1773 Details : change the condition to increase offset_idx
1776 Frequency : Always for filesystems larger than 2TB on 32-bit systems.
1778 Description: Data corruption for OSTs that are formatted larger than 2TB
1780 Details : When generating the bio request for lustre file writes the
1781 sector number would overflow a temporary variable before being
1782 used for the IO. The data reads correctly from Lustre (which
1783 will overflow in a similar manner) but other file data or
1784 filesystem metadata may be corrupted in some cases.
1788 Description: Tune the kernel for good SCSI performance.
1789 Details : Set the value of /sys/block/{dev}/queue/max_sectors_kb
1790 to the value of /sys/block/{dev}/queue/max_hw_sectors_kb
1795 Description: rewrite ext2-derived code in obdclass/uuid.c
1796 Details : rewrite inherited code (uuid parsing code from ext2 utils)
1797 from scratch preserving functionality.
1801 Description: "lfs find -obd UUID" prints directories
1802 Details : "lfs find -obd UUID" will return all directory names instead
1803 of just file names. It is incorrect because the directories
1804 do not reside on the OSTs.
1808 Description: use i_size_read and i_size_write in 2.6 port
1809 Details : replace inode->i_size access with i_size_read/write()
1812 Frequency : when removing large files
1814 Description: scheduling issue during removal of large Lustre files
1815 Details : Don't take the BKL in fsfilt_ext3_setattr() for 2.6 kernels.
1816 It causes scheduling issues when removing large files (17TB in the
1820 Frequency : only with liblustre clients on XT3
1822 Description: evictions taking too long
1823 Details : allow llrd to evict clients directly on OSTs
1826 Frequency : only on ppc
1828 Description: /proc/fs/lustre/devices broken on ppc
1829 Details : The patch as applied to 1.6.2 doesn't look correct for all arches.
1830 We should make sure the type of 'index' is loff_t and then cast
1831 explicitly as needed below. Do not assign an explicitly cast
1835 Frequency : only for rhel5
1837 Description: Kernel patches update for RHEL5 2.6.18-8.1.10.el5.
1838 Details : Modify the target file & which_kernel.
1840 Severity : enhancement
1842 Description: omit set fsid for export NFS
1843 Details : fix set/restore device id for avoid EMFILE error and mark lustre
1844 fs as FS_REQUIRES_DEV for avoid problems with generate fsid.
1848 Frequency : Always, for kernels after 2.6.16
1849 Description: Fix warning idr_remove called for id=.. which is not allocated.
1850 Details : Last kernels save old s_dev before kill super and not allow
1851 to restore from callback - restore it before call kill_anon_super.
1855 Description: buffer overruns could theoretically occur
1856 Details : llapi_semantic_traverse() modifies the "path" argument by
1857 appending values to the end of the origin string, and a
1858 overrun may occur. Adding buffer overrun check in liblustreapi.
1862 Description: Fix error on 'ls .' at the top of the Lustre mount.
1863 Details : Don't revalidate dentry if it is a root dentry.
1867 Description: Kernel patches update for RHEL4 2.6.9-55.0.6.
1868 Details : Modify vm-tunables-rhel4.patch.
1872 Description: Kernel config for 2.6.18-vanilla.
1873 Details : Modify targets/2.6-vanilla.target.in.
1874 Add config file kernel-2.6.18-2.6-vanilla-i686.config.
1875 Add config file kernel-2.6.18-2.6-vanilla-i686-smp.config.
1876 Add config file kernel-2.6.18-2.6-vanilla-x86_64.config.
1877 Add config file kernel-2.6.18-2.6-vanilla-x86_64-smp.config.
1881 Description: improve handling recoverable errors
1882 Details : if request processig with error which can be recoverable on server
1883 request should be resend, otherwise page released from cache and
1888 Description: Kernel patches update for RHEL5 2.6.18-8.1.14.el5.
1889 Details : Modify target file & which_patch.
1890 A flaw was found in the IA32 system call emulation provided
1891 on AMD64 and Intel 64 platforms. An improperly validated 64-bit
1892 value could be stored in the %RAX register, which could trigger an
1893 out-of-bounds system call table access. An untrusted local user
1894 could exploit this flaw to run code in the kernel
1895 (ie a root privilege escalation). (CVE-2007-4573).
1899 Description: O_DIRECT bypasses client statistics.
1900 Details : When running with O_DIRECT I/O, neither the client rpc_stats nor
1901 read_ahead_stats were updated. copied the stats section from
1902 osc_send_oap_rpc() into async_internal().
1906 Description: Add jbd statistics patch for RHEL5 and 2.6.18-vanilla
1910 Description: change order of libsysio includes
1911 Details : '#include sysio.h' should always come before '#include xtio.h'
1915 Description: handle "serious error: objid * already exists" more gracefully
1916 Details : If LAST_ID value on disk is smaller than the objects existing in
1917 the O/0/d* directories, it indicates disk corruption and causes an
1918 LBUG(). If the object is 0-length, then we should use the existing
1919 object. This will help to avoid a full fsck in most cases.
1921 Severity : enhancement
1923 Description: adapt the lustre_config script to support the upgrade case
1924 Details : Add "-u" option for lustre_config script to support upgrading 1.4
1925 server targets to 1.6 in parallel.
1929 Description: To avoid grant space > avaible space when the disk is almost
1930 full. Without this patch you might see the error "grant XXXX >
1931 available" or some LBUG about grant, when the disk is almost
1933 Details : In filter_check_grant, for non_grant cache write, we should
1934 check the left space by if (*left > ungranted + bytes), instead
1935 of (*left > ungranted), because only we are sure the left space
1936 is enough for another "bytes", then the ungrant space should be
1937 increase. In client, we should update cl_avail_grant only there
1938 is OBD_MD_FLGRANT in the reply.
1942 Description: Update RHEL 4 kernel to fix local root privilege escalation.
1943 Details : Update to the latest RHEL 4 kernel to fix the vulnerability
1944 described in CVE-2007-4573. This problem could allow untrusted
1945 local users to gain root access.
1948 Frequency : when using O_DIRECT and quotas
1950 Description: Incorrect file ownership on O_DIRECT output files
1951 Details : block usage reported by 'lfs quota' does not take into account
1952 files that have been written with O_DIRECT.
1957 Description: touch file failed when fs is not full
1958 Details : OST in recovery should not be discarded by MDS in alloc_qos(),
1959 otherwise we can get ENOSP while fs is not full.
1963 Description: parallel lock callbacks
1964 Details : Instead of sending blocking and completion callbacks as separated
1965 requests, adding them to a set and sending in parallel.
1968 Frequency : only for Cray XT3
1969 Bugzilla : 12829/13455
1970 Description: Changing primary group doesn't change the group lustre assigns to
1972 Details : When CRAY_XT3 is defined, the fsgid supplied by the client is
1973 overridden with the primary group provided by the group upcall,
1974 whereas the supplied fsgid can be trusted if it is in the list of
1975 supplementary groups returned by the group upcall.
1977 Severity : enhancement
1979 Description: Allow masking D_WARNING, D_ERROR messages from console
1980 Details : Console messages can now be disabled via lnet.printk.
1984 Description: User code with malformed file open parameter crashes client node
1985 Details : Before packing join_file req, all the related reference should be
1986 checked carefully in case some malformed flags cause fake
1987 join_file req on client.
1991 Description: LDLM_ENQUEUE races with LDLM_CP_CALLBACK
1992 Details : ldlm_completion_ast() assumes that a lock is granted when the req
1993 mode is equal to the granted mode. However, it should also check
1994 that LDLM_FL_CP_REQD is not set.
1998 Description: Heavy nfs access might result in deadlocks
1999 Details : After ELC code landed, it is now improper to enqueue any mds
2000 locks under och_sem, because enqueue might want to decide to
2001 cancel open locks for same inode we are holding och_sem for.
2005 Description: Client eviction while running blogbench
2006 Details : A lot of unlink operations with concurrent I/O can lead to a
2007 deadlock causing evictions. To address the problem, the number of
2008 oustanding OST_DESTROY requests is now throttled to
2009 max_rpcs_in_flight per OSC and LDLM_FL_DISCARD_DATA blocking
2010 callbacks are processed in priority.
2014 Description: enable ACLs on MDS by default
2015 Details : ACLs must be enabled on MDS by default.
2018 Frequency : PPC/PPC64 only
2020 Description: conflicts between asm-ppc64/types.h and lustre_types.h
2021 Details : fix duplicated definitions between asm-ppc64/types.h and
2022 lustre_types.h on PPC.
2025 Frequency : PPC/PPC64 only
2027 Description: asm-ppc/segment.h does not exist
2028 Details : fix compile issue on PPC.
2032 Description: better handle error messages in extents code
2035 Frequency : RHEL4 only
2037 Description: mkfs is very slow on IA64/RHEL4
2038 Details : A performance regression has been discovered in the MPT Fusion
2039 driver between versions 3.02.73rh and 3.02.99.00rh. As a
2040 consequence, we have downgraded the MPT Fusion driver in the RHEL4
2041 kernel from 3.02.99.00 to 3.02.73 until this problem is fixed.
2043 Severity : enhancement
2045 Description: SNMP support enhancement
2046 Details : Adding total number of sampled request for an MDS node in snmp
2049 Severity : enhancement
2051 Description: Optimize ldlm waiting list processing for PR extent locks
2052 Details : When processing waiting list for read extent lock and meeting read
2053 lock that is same or wider to it that is not contended, skip
2054 processing rest of the list and immediatelly return current
2055 status of conflictness, since we are guaranteed there are no
2056 conflicting locks in the rest of the list.
2060 Description: Time out and refuse to reconnect
2061 Details : When the failover node is the primary node, it is possible
2062 to have two identical connections in imp_conn_list. We must
2063 compare not conn's pointers but NIDs, otherwise we can defeat
2064 connection throttling.
2068 Description: port llog fixes from b1_6 into HEAD
2069 Details : Port llog reference couting and some llog cleanups from b1_6
2070 (bug 10800) into HEAD, for protect from panic and access to already
2071 free llog structures.
2075 Description: Detect stride IO mode in read-ahead
2076 Details : When a client does stride read, read-ahead should detect that and
2077 read-ahead pages according to the detected stride pattern.
2081 Description: data checksumming impacts single node performance
2082 Details : add support for several checksum algorithm. Currently, only CRC32
2083 and Adler-32 are supported. The checksum type can be changed on
2084 the fly via /proc/fs/lustre/osc/*/checksum_type.
2088 Description: use adler32 for page checksums
2089 Details : when available, use the Adler-32 algorithm instead of CRC32 for
2094 Description: build for x2 fails
2095 Details : fix compile issue on Cray systems.
2099 Description: Properly match for duplicate locks
2100 Details : Due to different lock order from skiplists code, we need to
2101 traverse entire list for now
2104 Frequency : only on PPC/SLES10
2106 Description: "BITS_PER_LONG is not 32 or 64" in linux/idr.h
2107 Details : On SLES10/PPC, fs.h includes idr.h which requires BITS_PER_LONG to
2108 be defined. Add a hack in mkfs_lustre.c to work around this compile
2113 Description: LASSERT on MDS when client holding flock lock dies
2114 Details : ldlm pool logic depends on number of granted locks equal to
2115 number of released locks which is not true for flock locks, so
2116 just exclude such locks from consideration.
2120 Description: MDS deadlock with many ll_sync_lov threads and I/O stalled
2121 Details : Use fsfilt_sync() for both the whole filesystem sync and
2122 individual file sync to eliminate dangerous inode locking
2123 with I_LOCK that can lead to a deadlock.
2127 Description: performance in 1.6.3
2128 Details : Force q->max_phys_segments to MAX_PHYS_SEGMENTS on SLES10 to be
2129 sure that 1MB requests are not fragmented by the block layer.
2131 Severity : enhancement
2133 Description: organize the server-side client stats on per-nid basis
2134 Details : Change the structure of stats under obdfilter and mds to
2142 The "uuid"s file would list the uuids of _active_ exports.
2143 And the clear entry is to clear all stats and stale nids.
2145 Severity : enhancement
2147 Description: eliminate client locks in face of contention
2148 Details : file contention detection and lockless i/o implementation
2149 for contended files.
2153 Description: Reinitialize optind to 0 so that interactive lfs works in all cases
2156 Frequency : very rare, if additional xattrs are used on kernels >= 2.6.12
2158 Description: MDS may lose file striping (and hence file data) in some cases
2159 Details : If there are additional extended attributes stored on the MDS,
2160 in particular ACLs, SELinux, or user attributes (if user_xattr
2161 is specified for the client mount options) then there is a risk
2162 of attribute loss. Additionally, the Lustre file striping
2163 needs to be larger than default (e.g. striped over all OSTs),
2164 and an additional attribute must be stored initially in the
2165 inode and then increase in size enough to be moved to the
2166 external attribute block (e.g. ACL growing in size) for file
2171 Description: skiplist implementation simplification
2172 Details : skiplists are used to group compatible locks on granted list
2173 that was implemented as tracking first and last lock of each
2174 lock group the patch changes that to using doubly linked lists
2178 Description: MDS LBUG: ASSERTION(!IS_ERR(dchild))
2179 Details : Change LASSERTs to client eviction (i.e. abort client's recovery)
2180 because LASSERT on both the data supplied by a client, and the
2181 data on disk is dangerous and incorrect.
2183 Severity : enhancement
2185 Description: Slow truncate/writes to huge files at high offsets.
2186 Details : Directly associate cached pages to lock that protect those pages,
2187 this allows us to quickly find what pages to write and remove
2188 once lock callback is received.
2192 Description: more ldlm soft lockups
2193 Details : In ldlm_resource_add_lock(), call to ldlm_resource_dump()
2194 starve other threads from the resource lock for a long time in
2195 case of long waiting queue, so change the debug level from
2196 D_OTHER to the less frequently used D_INFO.
2198 Severity : enhancement
2200 Description: add -gid, -group, -uid, -user options to lfs find
2204 Description: Hung threads in invalidate_inode_pages2_range
2205 Details : The direct IO path doesn't call check_rpcs to submit a new RPC
2206 once one is completed. As a result, some RPCs are stuck in the
2207 queue and are never sent.
2211 Description: filter threads hungs on waiting journal commit
2212 Details : Cleanup filter group llog code, then only filter group llog will
2213 be only created in the MDS/OST syncing process.
2217 Description: Procfs and llog threads access destoryed import sometimes.
2218 Details : Sync the import destoryed process with procfs and llog threads by
2219 the import refcount and semaphore.
2221 Severity : enhancement
2223 Description: openlock cache of b1_6 port to HEAD
2228 Description: kernel BUG at ldiskfs2_ext_new_extent_cb
2229 Details : If insertion of an extent fails, then discard the inode
2230 preallocation and free data blocks else it can lead to duplicate
2235 Description: don't always update ctime in ext3_xattr_set_handle()
2236 Details : Current xattr code updates inode ctime in ext3_xattr_set_handle.
2237 In some cases the ctime should not be updated, for example for
2238 2.0->1.8 compatibility it is necessary to delete an xattr and it
2239 should not update the ctime.
2243 Bugzilla : 15713/16362
2244 Description: Assertion in iopen_connect_dentry in 1.6.3
2245 Details : looking up an inode via iopen with the wrong generation number can
2246 populate the dcache with a disconneced dentry while the inode
2247 number is in the process of being reallocated. This causes an
2248 assertion failure in iopen since the inode's dentry list contains
2249 both a connected and disconnected dentry.
2253 Description: assertion failure in ldlm_handle2lock()
2254 Details : fix a race between class_handle_unhash() and class_handle2object()
2255 introduced in lustre 1.6.5 by bug 13622.
2260 Description: Kernel BUG: sd_iostats_bump: unexpected disk index
2261 Details : remove the limit of 256 scsi disks in the sd_iostat patch
2266 Description: oops in sd_iostats_seq_show()
2267 Details : unloading/reloading the scsi low level driver triggers a kernel
2268 bug when trying to access the sd iostat file.
2273 Description: Kernel panics during QLogic driver reload
2274 Details : REQ_BLOCK_PC requests are not handled properly in the sd iostat
2275 patch, causing memory corruption.
2280 Description: journal_dev option does not work in b1_6
2281 Details : pass mount option during pre-mount.
2283 Severity : enhancement
2285 Description: Add a FIEMAP(FIle Extent MAP) ioctl
2286 Details : FIEMAP ioctl will allow an application to efficiently fetch the
2287 extent information of a file. It can be used to map logical blocks
2288 in a file to physical blocks in the block device.
2292 Description: LDLM soft lockups - improvement
2293 Details : It is be possible to send the lock handle along with each read
2294 or write request because the client is already doing a lock match
2295 itself so there isn't any reason the OST should have to re-do that
2299 Frequency : only on Cray X2
2301 Description: X2 build failures
2302 Details : fix build failures on Cray X2.
2306 Description: xid & resent requests
2307 Details : Initialize RPC XID from clock at startup (randomly if clock is
2310 Severity : enhancement
2312 Description: Add lustre_start utility to start or stop multiple Lustre servers
2317 Description: Lustre GPF in {:ptlrpc:ptlrpc_server_free_request+373}
2318 Details : In case of memory pressure, list_del() can be called twice on
2319 req->rq_history_list, causing a kernel oops.
2323 Description: kptllnd_peer_check_sends()) ASSERTION(!in_interrupt()) failed
2324 Details : fix stack overflow in the distributed lock manager by defering
2325 export eviction after a failed AST to the elt thread instead of
2326 handling it in the dlm interpret routine.
2330 Description: Convert some comments to new format.
2331 Details : Update documenting comments to match doxygen conventions.
2335 Description: Grammar fixes.
2336 Details : A couple of trivial spelling fixes.
2340 Description: OSD_COUNTERS-mandatory
2341 Details : Make previously optional ->oti_{w,r}_locks sanity checks mandatory
2342 to simplify the code.
2346 Description: simplify cmm_device freeing logic.
2347 Details : Call cmm_device_free() in the failure path of cmm_device_alloc().
2351 Description: Add lockdep support to dt_object_operations locking interface.
2352 Details : Augment ->do_{read,write}_lock() prototypes with a `role'
2353 parameter indicating lock ordering. Update mdd code to use new
2358 Description: Introduce failloc constants for lockless IO tests.
2359 Details : Add two new failloc constants to test lockless IO. Only one of
2360 them in implemented---another is checked in yet to be landed
2365 Description: Add lockdep support for inode mutex.
2366 Details : Introduce and use new LOCK_INODE_MUTEX_PARENT() macro to be used
2367 in the situations where Lustre has to lock more than one inode
2372 Description: Add optional invariants checking support.
2373 Details : Add new LINVRNT() macro, optional on new --enable-invariants
2374 configure switch. This macro is to be used for consistency and
2375 sanity checks that are too expensive to be left in `production'
2380 Description: Zap lock->l_granted_mode with explicit LCK_MINMODE.
2381 Details : Use LCK_MINMODE rather than 0 to reset lock->l_granted_mode to
2386 Description: Add lockdep support for ldlm_lock and ldlm_resource.
2387 Details : Use spin_lock_nested() in (the only) situation where more than
2388 one ldlm_lock is locked simultaneously. Also, fix possible
2389 dead-lock in ldlm_lock_change_resource() by enforcing particular
2394 Description: Use struct ldlm_callback_suite in ldlm_lock_create().
2395 Details : Instead of specifying each ldlm_lock call-back through separate
2396 parameter, wrap them into struct ldlm_callback_suite.
2400 Description: Kill join_lru obd method and its callers.
2401 Details : CLIO uses lock weighting policy to keep locks over mmapped regions
2402 in memory---a requirement implemented through ->o_join_lru() obd
2403 method in HEAD. Remove this method and its users.
2407 Description: Add asynchronous ldlm ENQUEUE completion handler.
2408 Details : CLIO posts ENQUEUE requests asynchronously through ptlrpcd---a
2409 case that stock ldlm_completion_ast() cannot handle as it waits
2410 until lock is granted. Introduce new ldlm_completion_ast_async()
2411 for this. Also comment ldlm_completion_ast().
2415 Description: ldlm_error <-> errno conversion.
2416 Details : Add functions to map (rather arbitrary) between LDLM error codes
2417 and standard errno values. CLIO needs this to prevent LDLM specific
2418 constants from escaping ldlm and osc.
2422 Description: Kill unused ldlm_handle2lock_ns() function.
2423 Details : Kill unused ldlm_handle2lock_ns() function.
2427 Description: Add lu_ref support to ldlm_lock
2428 Details : lu_ref support for ldlm_lock and ldlm_resource. See lu_ref patch.
2429 lu_ref fields ->l_reference and ->lr_reference are added to ldlm_lock
2430 and ldlm_resource. LDLM interface has to be changed, because code that
2431 releases a reference on a lock, has to "know" what reference this is.
2432 In the most frequent case
2434 lock = ldlm_handle2lock(handle);
2436 LDLM_LOCK_PUT(lock);
2438 no changes are required. When any other reference (received _not_ from
2439 ldlm_handle2lock()) is released, LDLM_LOCK_RELEASE() has to be called
2440 instead of LDLM_LOCK_PUT().
2442 Arguably, changes are pervasive, and interface requires some discipline
2443 for proper use. On the other hand, it was very instrumental in finding
2444 a few leaked lock references.
2448 Description: Add ldlm_lock_addref_try().
2449 Details : Introduce ldlm_lock_addref_try() function (used by CLIO) that
2450 attempts to addref a lock that might be being canceled
2455 Description: Add ldlm_weigh_callback().
2456 Details : Add new ->l_weigh_ast() call-back to ldlm_lock. It is called
2457 by ldlm_cancel_shrink_policy() to estimate lock "value", instead of
2458 hard-coded `number of pages' logic.
2462 Description: Add lockdep annotations to llog code.
2463 Details : Use appropriately tagged _nested() locking calls in the places
2464 where llog takes more than one ->lgh_lock lock.
2468 Description: Add loi_kms_set().
2469 Details : Wrap kms updates into a helper function.
2473 Description: Constify instances of struct lsm_operations.
2474 Details : Constify instances of struct lsm_operations.
2478 Description: lu_conf support.
2479 Details : On a server, a file system object is uniquely identified
2480 by a fid, which is sufficient to locate and load all object
2481 state (inode). On a client, on the other hand, more data are
2482 necessary instantiate an object. Change lu_object_find() and
2483 friends to take additional `lu_conf' argument describing object.
2484 Typically this includes layout information.
2488 Description: lu_context fixes.
2489 Details : Introduce new lu_context functions that are needed on the client
2490 side, where some system threads (ptlrpcd) are shared by multiple
2491 modules, and so cannot be stopped during module shutdown.
2495 Description: Add start and stop methods to lu_device_type_operations.
2496 Details : Introduce two new methods in lu_device_type_operations, that are
2497 invoked when first instance of a given type is created and last one
2498 is destroyed respectively. This is need by CLIO.
2502 Description: Add lu_ref support to struct lu_device.
2503 Details : Add lu_ref support to lu_object and lu_device. lu_ref is used to
2504 track leaked references.
2508 Description: Introduce lu_kmem_descr.
2509 Details : lu_kmem_descr and its companion interface allow to create
2510 and destroy a number of kmem caches at once.
2514 Description: Fix lu_object finalization race.
2515 Details : Fix a race between lu_object_find() finding an object and its
2516 concurrent finalization. This race is (most likely) not possible
2517 on the server, but might happen on the client.
2521 Description: Introduce lu_ref interface.
2522 Details : lu_ref is a debugging module allowing to track references to
2523 a given object. It is quite cpu expensive, and has to be
2524 explicitly enabled with --enable-lu_ref. See usage description
2529 Description: Factor lu_site procfs stats into a separate function.
2530 Details : Separate lu_site stats printing code into a separate function
2531 that can be reused on a client.
2535 Description: Constify instances of struct {lu,dt,md}_device_operations.
2536 Details : Constify instances of struct {lu,dt,md}_device_operations.
2540 Description: Introduce struct md_site and move meta-data specific parts of
2541 struct lu_site here.
2542 Details : Move md-specific fields out of struct lu_site into special struct
2543 md_site, so that lu_site can be used on a client.
2547 Description: Kill mdd_lov_destroy().
2548 Details : Remove unused mdd code.
2552 Description: Add st_block checking to multistat.c.
2553 Details : Add st_block checking to multistat.c.
2557 Description: Add lu_ref support to struct obd_device.
2558 Details : Add lu_ref tracking to obd_device.
2562 Description: Kill obd_set_fail_loc().
2563 Details : Remove unused code.
2567 Description: Add special type for ptlrpc_request interpret functions.
2568 Details : Add lu_env parameter to ->rq_interpreter call-back. NULL is passed
2569 there. Actual usage will be in CLIO.
2573 Description: Replace RW_LOCK_UNLOCKED() macro with rwlock_init().
2574 Details : Replace RW_LOCK_UNLOCKED() with rwlock_init() as the former
2575 doesn't work with lockdep.
2579 Description: Add rwv.c test program.
2580 Details : New testing program exercising readv(2) and writev(2) (Qian).
2584 Description: Add sendfile.c test program.
2585 Details : New testing program exercising sendfile(2) (Jay).
2589 Description: Ratelimit a message that can be very frequent.
2590 Details : Ratelimit a memory allocation failure message that can
2595 Description: Use cdebug_show() in CDEBUG-style macros defined outside of libcfs.
2596 Details : Use cdebug_show() in CDEBUG-style macros defined outside of libcfs.
2600 Description: Liblustre build fixes.
2601 Details : Liblustre build fixes.
2605 Description: libcfs: add cfs_{need,cond}_resched() interface.
2606 Details : libcfs: add cfs_{need,cond}_resched() definition and
2607 implementations for Linux, NT, and liblustre.
2609 Severity : enhancement
2611 Description: More exported tunables for mballoc
2612 Details : Add support for tunable preallocation window and new tunables for
2613 large/small requests
2617 Description: Detect corruption of block bitmap and checking for preallocations
2618 Details : Checks validity of on-disk block bitmap. Also it does better
2619 checking of number of applied preallocations. When corruption is
2620 found, it turns filesystem readonly to prevent further corruptions.
2624 Description: (rw.c:1323:ll_read_ahead_pages()) ASSERTION(page_idx > ria->ria_stoff) failed
2625 Details : Once the unmatched stride IO mode is detected, shrink the stride-ahead
2626 window to 0. If it does hit cache miss, and read-pattern is still
2627 stride-io mode, does not reset the stride window, but also does not
2628 increase the stride window length in this case.
2632 Frequency : only for big-endian servers
2633 Description: Check if system is big-endian while mounting fs with extents feature
2634 Details : Mounting a filesystem with extents feature will fail on big-endian
2635 systems since ext3-based ldiskfs is not supported on big-endian
2636 systems. This can be over-riden with "bigendian_extents" mount option.
2638 Severity : enhancement
2640 Description: The root squash functionality
2641 Details : A security feature, which is to prevent users from being able
2642 to mount lustre on their desktop, run as root, and delete
2643 all of the files in the filesystem. The goal is accomplished by
2644 remapping user id (UID) and group id (GID) of the root user to
2645 a UID and GID specified by the system administartor via Lustre
2646 configuration management server (MGS). The functionality also
2647 allows to specify sets of clients for which the remapping does
2652 Description: Excessive recovery window
2653 Details : With AT enabled, the recovery window can be excessively long (6000+
2654 seconds). To address this problem, we no longer use
2655 OBD_RECOVERY_FACTOR when extending the recovery window (the connect
2656 timeout no longer depends on the service time, it is set to
2657 INITIAL_CONNECT_TIMEOUT now) and clients report the old service
2658 time via pb_service_time.
2662 Description: Watchdog triggered on MDS failover
2663 Details : enable OBD_CONNECT_MDT flag when connecting from the MDS so that
2664 the OSTs know that the MDS "UUID" can be reused for the same export
2665 from a different NID, so we do not need to wait for the export to be
2669 Frequency : rare, only if using MMP with Linux RAID
2671 Description: MMP doesn't work with Linux RAID
2672 Details : While using HA for Lustre servers with Linux RAID, it is possible
2673 that MMP will not detect multiple mounts. To make this work we
2674 need to unplug the device queue in RAID when the MMP block is being
2675 written. Also while reading the MMP block, we should read it from
2676 disk and not the cached one.
2678 Severity : enhancement
2680 Description: open file using fid
2681 Details : A file can be opened using just its fid, like
2682 <mntpt>/.lustre/fid/SEQ:OID:VER - this is needed for HSM and replication
2685 Frequency : Only in RHEL5 when mounting multiple ext3 filesystems
2688 Description: "kmem_cache_create: duplicate cache jbd_4k" error message
2689 Details : add proper locking for creation of jbd_4k slab cache
2693 Description: MMP check in ext3_remount() fails without displaying any error
2694 Details : When multiple mount protection fails during remount, proper error
2697 Severity : enhancement
2699 Description: Allow stripe size to be up to 4G-64k
2700 Details : Fix math logic to allow large stripe sizes.
2704 Description: add check for >8TB ldiskfs filesystems
2705 Details : ext3-based ldiskfs does not support greater than 8TB LUNs.
2706 Don't allow >8TB ldiskfs filesystems to be mounted without
2707 force_over_8tb mount option
2712 Descriptoin: OSS complains "too many blocks" and crashes
2713 Details : In filter_commitrw_write():
2714 1. lquota_chkquota();
2715 2. obdo_from_inode();
2716 3. lquota_pending_commit();
2717 "2." has changed uid and gid in this case, then caused
2718 lquota_chkquota() and lquota_pending_commit() to use
2719 different uid and gid. That is the root of the bug.
2723 Descriptoin: getstripe doesn't report stripe count = -1 sanely
2724 Details : Detect and print out the 'stripe count = -1' case properly.
2728 Description: Out or order replies might be lost on replay
2729 Details : In ptlrpc_retain_replayable_request if we cannot find retained
2730 request with tid smaller then one currently being added, add it
2731 to the start, not end of the list.
2733 --------------------------------------------------------------------------------
2735 2007-08-10 Cluster File Systems, Inc. <info@clusterfs.com>
2737 * Support for kernels:
2738 2.6.5-7.283 (SLES 9),
2739 2.6.9-55.EL (RHEL 4),
2740 2.6.16.46-0.14 (SLES 10),
2741 2.6.18.8 vanilla (kernel.org)
2742 * Client support for unpatched kernels:
2743 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
2744 2.6.16 - 2.6.22 vanilla (kernel.org)
2745 * Due to recently discovered recovery problems, we do not recommend
2746 using patchless RHEL 4 clients with this or any earlier release.
2747 * Recommended e2fsprogs version: 1.39.cfs8
2748 * Note that reiserfs quotas are disabled on SLES 10 in this kernel.
2749 * Starting with this release, the ldiskfs backing filesystem required
2750 by Lustre is now in its own package, lustre-ldiskfs. This package
2751 should be installed. It is versioned separately from Lustre and
2752 may be released separately in future.
2756 Description: block reactivating mgc import until all deactivates complete
2757 Details : Fix race when failing back MDT/MGS to itself (testing)
2760 Frequency : at statup only
2762 Description: mds_lov_synchronize race leads to various problems
2763 Details : simultaneous MDT->OST connections at startup can cause the
2764 sync to abort, leaving the OSC in a bad state.
2766 Severity : enhancement
2768 Description: add optional extra BUILD_VERSION info
2769 Details : add a new environment variable (namely LUSTRE_VERS) which allows
2770 to override the lustre version.
2773 Frequency : 2.6.18 servers only
2775 Description: ll_kern_mount() doesn't release the module reference
2776 Details : The ldiskfs module reference count never drops down to 0
2777 because ll_kern_mount() doesn't release the module reference.
2782 Description: server LBUG when using old ost_num_threads parameter
2783 Details : Accept the old ost_num_threads parameter but warn that it
2784 is deprecated, and fix an off-by-one error that caused an LBUG.
2789 Description: Transient SCSI error results in persistent IO issue
2790 Details : iobuf->dr_error is not reinitialized to 0 between two
2794 Frequency : sometimes when underlying device returns I/O errors
2796 Description: OSTs not going read-only during write failures
2797 Details : OSTs are not remounted read-only when the journal commit threads
2798 get I/O errors because fsfilt_ext3 calls journal_start/stop()
2799 instead of the ext3 wrappers.
2803 Description: poor connect scaling with increasing client count
2804 Details : Don't run filter_grant_sanity_check for more than 100 exports
2805 to improve scaling for large numbers of clients.
2808 Frequency : SLES10 only
2810 Description: sanity-quota.sh quotacheck failed: rc = -22
2811 Details : Quotas cannot be enabled on SLES10.
2814 Frequency : liblustre clients only
2816 Description: getdirentries does not give error when run on compute nodes
2817 Details : getdirentries does not fail when the size specified as an argument
2818 is too small to contain at least one entry
2820 Severity : enhancement
2822 Description: Add LNET router traceability for debug purposes
2823 Details : If a checksum failure occurs with a router as part of the
2824 IO path, the NID of the last router that forwarded the bulk data
2825 is printed so it can be identified.
2830 Description: OST "spontaneously" evicts client; client has imp_pingable == 0
2831 Details : Due to a race condition, liblustre clients were occasionally
2832 evicted incorrectly.
2834 Severity : enhancement
2836 Description: lfs setstripe use optional parameters instead of postional
2839 Severity : enhancement
2841 Description: Nanosecond timestamp support for ldiskfs
2842 Details : The on-disk ldiskfs filesystem has added support for nanosecond
2843 resolution timestamps. There is not yet support for this at
2844 the Lustre filesystem level.
2847 Frequency : during server recovery
2849 Description: MDS failing to send precreate requests due to OSCC_FLAG_RECOVERING
2850 Details : request with rq_no_resend flag not awake l_wait_event if they get
2854 Frequency : nfs export on patchless client
2856 Description: connectathon hang when test nfs export over patchless client
2857 Details : Disconnected dentry cannot be found with lookup, so we do not need
2858 to unhash it or make it invalid
2861 Description: fix llapi_lov_get_uuids() to allow many OSTs to be returned
2862 Details: : Change llapi_lov_get_uuids() to read the UUIDs from /proc instead
2863 of using an ioctl. This allows lfsck for > 160 OSTs to succeed.
2868 Description: open req refcounting wrong on reconnect
2869 Details : If reconnect happened between getting open reply from server and
2870 call to mdc_set_replay_data in ll_file_open, we will schedule
2871 replay for unreferenced request that we are about to free.
2872 Subsequent close will crash in variety of ways.
2873 Check that request is still eligible for replay in
2874 mdc_set_replay_data().
2879 Description: disable writes to filesystem when reading health_check file
2880 Details : the default for reading the health_check proc file has changed
2881 to NOT do a journal transaction and write to disk, because this
2882 can cause reads of the /proc file to hang and block HA state
2883 checking on a healthy but otherwise heavily loaded system. It
2884 is possible to return to the previous behaviour during configure
2885 with --enable-health-write.
2887 Severity : enhancement
2889 Description: 64-bit inode version
2890 Details: : Add a on-disk 64-bit inode version for ext3 to track changes made
2891 to the inode. This will be required for version-based recovery.
2896 Description: MDS fails to start if a duplicate client export is detected
2897 Details : in some rare cases it was possible for a client to connect to
2898 an MDS multiple times. Upon recovery the MDS would detect this
2899 and fail during startup. Handle this more gracefully.
2901 Severity : enhancement
2903 Description: Add -o localflock option to simulate old noflock
2905 Details : This will achieve local-only flock/fcntl locks
2911 Description: log_commit_thread vs filter_destroy race leads to crash
2912 Details : Take import reference before releasing llog record semaphore
2917 Description: Wrong request locking in request set processing
2918 Details : ptlrpc_check_set wrongly uses req->rq_lock for proctect add to
2919 imp_delayed_list, in this place should be used imp_lock.
2922 Frequency : when reconnection
2924 Description: Grant Leak when osc reconnect to OST
2925 Details : When osc reconnect ost, OST(filter) should check whether it
2926 should grant more space to client by comparing fed_grant and
2927 cl_avail_grant, and return the granted space to client instead
2928 of "new granted" space, because client will call osc_init_grant
2929 to update the client grant space info.
2932 Frequency : when client reconnect to OST
2934 Description: Grant Leak when osc do resend and replay bulk write
2935 Details : When osc reconnect to OST, OST(filter)should clear grant info of
2936 bulk write request, because the grant info while be sync between
2937 OSC and OST when reconnect, and we should ignore the grant info
2938 these of resend/replay write req.
2943 Description: Grant space more than available left space sometimes.
2944 Details : When then OST is about to be full, if two bulk writing from
2945 different clients came to OST. Accord the avaliable space of the
2946 OST, the first req should be permitted, and the second one
2947 should be denied by ENOSPC. But if the seconde arrived before
2948 the first one is commited. The OST might wrongly permit second
2949 writing, which will cause grant space > available space.
2952 Frequency : when client is evicted
2954 Description: Grant might be wrongly erased when osc is evicted by OST
2955 Details : when the import is evicted by server, it will fork another
2956 thread ptlrpc_invalidate_import_thread to invalidate the
2957 import, where the grant will be set to 0. While the original
2958 thread will update the grant it got when connecting. So if
2959 the former happened latter, the grant will be wrongly errased
2960 because of this race.
2965 Description: Checking Stale with correct fid
2966 Details : ll_revalidate_it should uses de_inode instead of op_data.fid2
2967 to check whether it is stale, because sometimes, we want the
2968 enqueue happened anyway, and op_data.fid2 will not be initialized.
2970 Severity : enhancement
2972 Description: update patchless client
2973 Details : Add support for patchless client with 2.6.20, 2.6.21 and RHEL 5
2976 Frequency : only with 2.4 kernel
2978 Description: random memory corruption
2979 Details : size of struct ll_inode_info is to big for union inode.u and this
2980 can be cause of random memory corruption.
2985 Description: Memory leak in recovery
2986 Details : Lov_mds_md was not free in an error handler in mds_create_object.
2987 It should also check obd_fail before fsfilt_start, otherwise if
2988 fsfilt_start return -EROFS,(failover mds during mds recovery).
2989 then the req will return with repmsg->transno = 0 and rc = EROFS.
2990 and we met hit the assert LASSERT(req->rq_reqmsg->transno ==
2991 req->rq_repmsg->transno) in ptlrpc_replay_interpret. Fcc should
2992 be freed no matter whether fsfilt_commit success or not.
2995 Frequency : only with huge count clients
2997 Description: Prevents from taking the superblock lock in llap_from_page for
2999 Details : using LL_ORIGIN_REMOVEPAGE origin flag instead of LL_ORIGIN_UNKNOW
3000 for llap_from_page call in ll_removepage prevents from taking the
3001 superblock lock for a soon died page.
3006 Description: Not check open intent error before release open handle
3007 Details : in some rare cases, the open intent error is not checked before
3008 release open handle, which may cause
3009 ASSERTION(open_req->rq_transno != 0), because it tries to release
3010 the failed open handle.
3015 Description: Set cat log bitmap only after create log success.
3016 Details : in some rare cases, the cat log bitmap is set too early. and it
3017 should be set only after create log success.
3021 Description: Accessing a block bevice can re-enable I/O when Lustre is
3022 tearing down a device.
3023 Details : dev_clear_rdonly(bdev) must be called in kill_bdev() instead of
3028 Description: service threads may hog cpus when there are a lot of requests
3029 Details : Insert cond_resched to give other threads a chance to use some CPU
3034 Description: the cat log was not initialized in recovery
3035 Details : When mds(mgs) do recovery, the tgt_count might be zero, so the
3036 unlink log on mds will not be initialized until mds post
3037 recovery. And also in mds post recovery, the unlink log will
3038 initialization will be done asynchronausly, so there will be race
3039 between add unlink log and unlink log initialization.
3043 Description: brw_stats were being printed incorrectly
3044 Details : brw_stats were being printed as log2 but all of them were not
3045 recorded as log2. Also remove some code duplication arising from
3046 filter_tally_{read,write}.
3050 Frequency : rare, only in recovery.
3051 Description: ASSERTION(req->rq_type != LI_POISON) failed
3052 Details : imp_lock should be held while iterating over imp_sending_list for
3053 prevent destroy request after get timeout in ptlrpc_queue_wait.
3057 Description: replay-single.sh test 52 fails
3058 Details : A lock's skiplist need to be cleanup when it being unlinked
3059 from its resource list.
3063 Description: Short directio read returns full requested size rather than
3065 Details : Direct I/O operations should return actual amount of bytes
3066 transferred rather than requested size.
3068 Severity : enhancement
3070 Description: metadata RPC reduction (e.g. for rm performance)
3071 Details : decrease the amount of synchronous RPC between clients and servers
3072 by canceling conflicing lock before the operation on the client
3073 and packing thier handles into the main operation RPC to server.
3075 Severity : enhancement
3077 Description: Async OSC create to avoid the blocking unnecessarily.
3078 Details : If an OST has no remaining object, system will block on the
3079 creation when it needs to create a new object on this OST. Now,
3080 ways use pre-created objects when available, instead of blocking on
3081 an empty osc while others are not empty. If we must block, we
3082 block for the shortest possible period of time.
3086 Description: improve handling recoverable errors
3087 Details : if request processig with error which can be recoverable on server
3088 request should be resend, otherwise page released from cache and
3091 Severity : enhancement
3093 Description: refine locking for avoid write wrong info into lov_objid file
3094 Details : fix possible races with add new target and write/update data in
3097 --------------------------------------------------------------------------------
3099 2007-05-03 Cluster File Systems, Inc. <info@clusterfs.com>
3104 Frequency : on some architectures
3106 Description: 1.6 client sometimes fails to mount from a 1.4 MDT
3107 Details : Uninitialized flags sometimes cause configuration commands to
3111 Frequency : patchless clients only
3113 Description: missing __iget() symbol export
3114 Details : The __iget() symbol export is missing. To avoid the need for
3115 this on patchless clients the deathrow inode reaper is turned
3116 off, and we depend on the VM to clean up old inodes. This
3117 dependency was during via the fix for bug 12181.
3119 --------------------------------------------------------------------------------
3121 2007-04-19 Cluster File Systems, Inc. <info@clusterfs.com>
3123 * CONFIGURATION CHANGE. This version of Lustre WILL NOT
3124 INTEROPERATE with older versions automatically. In many cases a
3125 special upgrade step is needed. Please read the
3126 user documentation before upgrading any part of a live system.
3127 * WIRE PROTOCOL CHANGE from previous 1.6 beta versions. This
3128 version will not interoperate with 1.6 betas before beta5 (1.5.95).
3129 * WARNING: Lustre configuration and startup changes are required with
3130 this release. See https://mail.clusterfs.com/wikis/lustre/MountConf
3134 Severity : enhancement
3136 Description: MountConf
3137 Details : Lustre configuration is now managed via mkfs and mount
3138 commands instead of lmc and lconf. New obd types (MGS, MGC)
3139 are added for dynamic configuration management. See
3140 https://mail.clusterfs.com/wikis/lustre/MountConf for
3143 Severity : enhancement
3145 Description: dynamic OST addition
3146 Details : OSTs can now be added to a live filesystem
3148 Severity : enhancement
3150 Description: startup order invariance
3151 Details : MDTs and OSTs can be started in any order. Clients only
3152 require the MDT to complete startup.
3154 Severity : enhancement
3156 Description: parallel, asynchronous orphan cleanup
3157 Details : orphan cleanup is now performed in separate threads for each
3158 OST, allowing parallel non-blocking operation.
3160 Severity : enhancement
3162 Description: optimized stripe assignment
3163 Details : stripe assignments are now made based on ost space available,
3164 ost previous usage, and OSS previous usage, in order to try
3165 to optimize storage space and networking resources.
3167 Severity : enhancement
3169 Description: Permanently set tunables
3170 Details : All writable /proc/fs/lustre tunables can now be permanently
3171 set on a per-server basis, at mkfs time or on a live system.
3173 Severity : enhancement
3175 Description: Lustre message v2
3176 Details : Add lustre message format v2.
3178 Severity : enhancement
3180 Description: client OST exclusion list
3181 Details : Clients can be started with a list of OSTs that should be
3182 declared "inactive" for known non-responsive OSTs.
3186 Description: SPEC SFS validation failure on NFS v2 over lustre.
3187 Details : Changes the blocksize for regular files to be 2x RPC size,
3188 and not depend on stripe size.
3190 Severity : enhancement
3192 Description: Multiple MD RPCs in flight.
3193 Details : Further unserialise some read-only MDS RPCs - learn about intents.
3194 To avoid overly-overloading MDS, introduce a limit on number of
3195 MDS RPCs in flight for a single client and add /proc controls
3196 to adjust this limit.
3198 Severity : enhancement
3200 Description: client read/write statistics
3201 Details : Add client read/write call usage stats for performance
3202 analysis of user processes.
3203 /proc/fs/lustre/llite/*/offset_stats shows non-sequential
3204 file access. extents_stats shows chunk size distribution.
3205 extents_stats_per_process show chunk size distribution per
3208 Severity : enhancement
3210 Description: mds statistics
3211 Details : Add detailed mds operations statistics in
3212 /proc/fs/lustre/mds/*/stats.
3216 Description: Failure of copying files with lustre special EAs.
3217 Details : Client side always return success for setxattr call for lustre
3218 special xattr (currently only "trusted.lov").
3223 Description: Refcount LNET uuids
3224 Details : The global LNET uuid list grew linearly with every startup;
3225 refcount repeated list entries instead of always adding to
3228 Severity : enhancement
3230 Description: Dynamic service threads
3231 Details : Within a small range, start extra service threads
3232 automatically when the request queue builds up.
3235 Frequency : mixed-endian client/server environments
3237 Description: mixed-endian crashes
3238 Details : The new msg_v2 system had some failures in mixed-endian
3241 Severity : enhancement
3243 Description: Easy OST removal
3244 Details : OSTs can be permanently deactivated with e.g. 'lctl
3245 conf_param lustre-OST0001.osc.active=0'
3247 Severity : enhancement
3249 Description: MGS proc entries
3250 Details : Added basic proc entries for the MGS showing what filesystems
3253 Severity : enhancement
3255 Description: provide MGS failover
3256 Details : Added config lock reacquisition after MGS server failover.
3258 Severity : enhancement
3260 Description: add Linux 2.4 support
3261 Details : Added support for RHEL 2.4.21 kernel for 1.6 servers and clients
3265 Description: a large application tries to do I/O to the same resource and dies
3266 in the middle of it.
3267 Details : Check the req->rq_arrival time after the call to
3268 ost_brw_lock_get(), but before we do anything about
3269 processing it & sending the BULK transfer request. This
3270 should help move old stale pending locks off the queue as
3271 quickly as obd_timeout.
3274 Frequency : when an incorrect nid is specified during startup
3276 Description: ptlrpc connect to non-existant node causes kernel crash
3277 Details : LNET can't be re-entered from an event callback, which
3278 happened when we expire a message after the export has been
3279 cleaned up. Instead, hand the zombie cleanup off to another
3282 Severity : enhancement
3284 Description: plain/inodebits lock performance improvement
3285 Details : Grouping plain/inodebits in granted list by their request modes
3286 and bits policy, thus improving the performance of search through
3290 Frequency : only if OST filesystem is corrupted
3292 Description: client incorrectly hits assertion in ptlrpc_replay_req()
3293 Details : for a short time RPCs with bulk IO are in the replay list,
3294 but replay of bulk IOs is unimplemented. If the OST filesystem
3295 is corrupted due to disk cache incoherency and then replay is
3296 started it is possible to trip an assertion. Avoid putting
3297 committed RPCs into the replay list at all to avoid this issue.
3300 Frequency : liblustre (e.g. catamount) on a large cluster with >= 8 OSTs/OSS
3302 Description: System hang on startup
3303 Details : This bug allowed the liblustre (e.g. catamount) client to
3304 return to the app before handling all startup RPCs. This
3305 could leave the node unresponsive to lustre network traffic
3306 and manifested as a server ptllnd timeout.
3308 Severity : enhancement
3310 Description: Add "/proc/sys/lustre/debug_peer_on_timeout"
3311 Details : liblustre envirable: LIBLUSTRE_DEBUG_PEER_ON_TIMEOUT
3312 boolean to control whether to print peer debug info when a
3313 client's RPC times out.
3316 Frequency : only for kernels with patches from Lustre below 1.4.3
3318 Description: Remove old rdonly API
3319 Details : Remove old rdonly API which unused from at least lustre 1.4.3
3322 Frequency : only for devices with external journals
3324 Description: Set external device read-only also
3325 Details : During a commanded failover stop, we set the disk device
3326 read-only while the server shuts down. We now also set any
3327 external journal device read-only at the same time.
3330 Frequency : when upgrading from 1.4 while trying to change parameters
3332 Description: The wrong (new) MDC name was used when setting parameters for
3333 upgraded MDT's. Also allows changing of OSC (and MDC)
3334 parameters if --writeconf is specified at tunefs upgrade time.
3337 Frequency : when setting specific ost indicies
3339 Description: QOS code breaks on skipped indicies
3340 Details : Add checks for missing OST indicies in the QOS code, so OSTs
3341 created with --index need not be sequential.
3343 Severity : enhancement
3345 Description: Add uninit_groups feature to ldiskfs2 to speed up e2fsck
3346 Details : The uninit_groups feature works in conjunction with the kernel
3347 filesystem code (ldiskfs2 only) and e2fsprogs-1.39-cfs6 to speed
3348 up the pass1 processing of e2fsck. This is a read-only feature
3349 in ldiskfs2 only, so older kernels and current ldiskfs cannot
3350 mount filesystems that have had this feature enabled.
3352 Severity : enhancement
3354 Description: Improve multi-block allocation algorithm to avoid fragmentation
3355 Details : The mballoc3 code (ldiskfs2 only) adds new mechanisms to improve
3356 allocation locality and avoid filesystem fragmentation.
3358 ------------------------------------------------------------------------------
3360 2007-04-01 Cluster File Systems, Inc. <info@clusterfs.com>
3362 * Support for kernels:
3363 2.4.21-47.0.1.EL (RHEL 3)
3364 2.6.5-7.283 (SLES 9)
3365 2.6.9-42.0.10.EL (RHEL 4)
3366 2.6.12.6 vanilla (kernel.org)
3367 2.6.16.27-0.9 (SLES 10)
3368 * Recommended e2fsprogs version: 1.39.cfs5
3370 * Note that reiserfs quotas are disabled on SLES 10 in this kernel
3374 Frequency : occasional, depends on client load and configuration
3375 Bugzilla : 12181, 12203
3376 Description: data loss for recently-modified files
3378 Details : In some cases it is possible that recently written or created
3379 files may not be written to disk in a timely manner (this should
3380 normally be within 30s unless client IO load is very high).
3381 The problem appears as zero-length files or files that are a
3382 multiple of 1MB in size after a client crash or client eviction
3383 that are missing data at the end of the file.
3385 This problem is more likely to be hit on clients where files are
3386 repeatedly created and unlinked in the same directory, clients
3387 have a large amount of RAM, have many CPUs, the filesystem has
3388 many OSTs, the clients are rebooted frequently, and/or the files
3389 are not accessed by other nodes after being written.
3391 The presence of the problem can be detected by looking at
3392 /proc/sys/fs/inode-state. If the first number (nr_inodes) is
3393 smaller than the second (nr_unused) then dirty files will not
3394 be flushed automatically to disk. "sync; sleep 10" should be
3395 run several times on the node before unmounting it to update
3396 Lustre (this is also safe to run on nodes without this problem).
3398 There is also a related kernel bug in the RHEL4 4 2.6.9 kernel
3399 that can cause this same problem, so customers using that kernel
3400 also need to update the kernel in addition to Lustre. In order
3401 to properly fix this bug, the RHEL3 2.4.21 kernel is also updated.
3403 It is normal that files written just before a client crash (less
3404 than 30s) may not yet have been flushed to disk, even for local
3408 Frequency : frequent on thin XT3 nodes
3410 Description: UUID collision on thin XT3 Linux nodes
3411 Details : UUIDs on Compute Node Linux XT3 nodes were not generated
3412 randomly, since we relied on an insufficiently-seeded PRNG.
3417 Description: OSS hangs after "All ost request buffers busy"
3418 Details : A deadlock between quota and journal operations caused OSS
3419 hangs after printing "All ost request buffers busy."
3422 Frequency : always on liblustre builds
3424 Description: Cleanup compiler warnings on liblustre
3427 Frequency : always on liblustre builds on XT3
3429 Description: LC_CONFIG_CDEBUG don't run while build liblustre on XT3.
3433 Description: Addition of EXT3_FEATURE_RO_COMPAT_DIR_NLINKS flag for
3434 > 32000 subdirectories
3435 Details : Add EXT3_FEATURE_RO_COMPAT_DIR_NLINK flag to
3436 EXT3_FEATURE_RO_COMPAT_SUPP. This flag will be set whenever
3437 subdirectory count crosses 32000. This will aid e2fsck to
3438 correctly handle more than 32000 subdirectories.
3441 Frequency : liblustre (e.g. catamount) on a large cluster with >= 8 OSTs/OSS
3443 Description: System hang on startup
3444 Details : This bug allowed the liblustre (e.g. catamount) client to
3445 return to the app before handling all startup RPCs. This
3446 could leave the node unresponsive to lustre network traffic
3447 and manifested as a server ptllnd timeout.
3449 Severity : enhancement
3451 Description: Add "/proc/sys/lustre/debug_peer_on_timeout"
3452 (liblustre envirable: LIBLUSTRE_DEBUG_PEER_ON_TIMEOUT)
3453 boolean to control whether to print peer debug info when a
3454 client's RPC times out.
3459 Description: make O_SYNC working on 2.6 kernels
3460 Details : 2.6 kernels use different method for mark pages for write,
3461 so need add a code to lustre for O_SYNC work.
3466 Description: Failure to close file and release space on NFS
3467 Details : Put inode details into lock acquired in ll_intent_file_open.
3468 Use mdc_intent_lock in ll_intent_open to properly
3469 detect all kind of errors unhandled by mdc_enqueue.
3474 Description: proc file read during shutdown sometimes raced obd removal,
3476 Details : Add lock to prevent obd access after proc file removal.
3479 Frequency : Only for files larger than 4GB on 32-bit clients.
3481 Description: improperly doing page alignment of locks
3482 Details : Modify lustre core code to use CFS_PAGE_* defines instead of
3483 PAGE_*. Make CFS_PAGE_MASK a 64-bit mask.
3488 Description: RPCs being resent when they shouldn't be
3489 Details : Some RPCs that should not be resent are being resent. This
3490 can cause inconsistencies in the RPC state machine. Do not
3491 resend such requests.
3494 Frequency : rare, only with NFS export
3496 Description: Crash on NFS re-export node
3497 Details : under very unusual load conditions an assertion is hit in
3498 ll_intent_file_open()
3501 Frequency : only if OST filesystem is corrupted
3503 Description: client incorrectly hits assertion in ptlrpc_replay_req()
3504 Details : for a short time RPCs with bulk IO are in the replay list,
3505 but replay of bulk IOs is unimplemented. If the OST filesystem
3506 is corrupted due to disk cache incoherency and then replay is
3507 started it is possible to trip an assertion. Avoid putting
3508 committed RPCs into the replay list at all to avoid this issue.
3513 Description: large O_DIRECT requests fail under memory pressure/fragmentation
3514 Details : Large single O_DIRECT read and write calls can fail to allocate
3515 a sufficiently large buffer to process the request. In case of
3516 allocation failure the allocation is retried with a smaller
3517 buffer and broken into smaller requests.
3519 Severity : enhancement
3521 Description: Add -o localflock option to simulate old noflock behaviour.
3522 Details : This will achieve local-only flock/fcntl locks coherentness.
3527 Description: versioning check is incomplete
3528 Details : Checking the version difference of client vs. server, report
3529 error if the gap is too big.
3534 Description: add support PG_writeback bit
3535 Details : add support for PG_writeback bit for Lustre, for more carefull
3536 work with page cache in 2.6 kernel. This also fix some deadlocks
3537 and remove hack for work O_SYNC with 2.6 kernel.
3539 Severity : enhancement
3541 Description: Add uninit_groups feature to ldiskfs2 to speed up e2fsck
3542 Details : The uninit_groups feature works in conjunction with the kernel
3543 filesystem code (ldiskfs2 only) and e2fsprogs-1.39-cfs6 to speed
3544 up the pass1 processing of e2fsck. This is a read-only feature
3545 in ldiskfs2 only, so older kernels and current ldiskfs cannot
3546 mount filesystems that have had this feature enabled.
3548 Severity : enhancement
3550 Description: Improve multi-block allocation algorithm to avoid fragmentation
3551 Details : The mballoc3 code (ldiskfs2 only) adds new mechanisms to improve
3552 allocation locality and avoid filesystem fragmentation.
3554 ------------------------------------------------------------------------------
3556 2007-02-09 Cluster File Systems, Inc. <info@clusterfs.com>
3558 * Support for kernels:
3559 2.6.9-42.0.3.EL (RHEL 4)
3560 2.6.5-7.276 (SLES 9)
3561 2.4.21-47.0.1.EL (RHEL 3)
3562 2.6.12.6 vanilla (kernel.org)
3563 2.6.16.21-0.8 (SLES10)
3564 * Recommended e2fsprogs version: 1.39.cfs2-0
3566 * The backwards-compatible /proc/sys/portals symlink has been removed
3567 in this release. Before upgrading, please ensure that you change
3568 any configuration scripts or /etc/sysctl.conf files that access
3569 /proc/sys/portals/* or sysctl portals.* to use the corresponding
3570 entry in /proc/sys/lnet or sysctl lnet.*. This change can be made
3571 in advance of the upgrade on any system running Lustre 1.4.6 or
3572 newer, since /proc/sys/lnet was added in that version.
3573 * Note that reiserfs quotas are disabled on SLES 10 in this kernel
3577 Frequency : only when quota is used
3579 Description: avoid scanning export list for quota master
3580 Details : Change the algorithms to avoid scanning export list in order
3581 to improve the efficiency.
3584 Frequency : MDS failover only, very rarely
3586 Description: "went back in time" messages on mds failover
3587 Details : The greatest transno may be lost when the current operation
3588 finishes with an error (transno==0) and the client's last_rcvd
3589 record is over-written. Save the greatest transno in the
3590 mds_last_transno for this case.
3593 Frequency : always for specific kernels and striping counts
3595 Description: client may get "Matching packet too big" without ACL support
3596 Details : Clients compiled without CONFIG_FS_POSIX_ACL get an error message
3597 when trying to access files in certain configurations. The
3598 clients should in fact be denied when mounting because they do
3599 not understand ACLs.
3602 Frequency : Cray XT3 with more than 4000 clients and multiple jobs
3604 Description: many clients connecting with IO in progress causes connect timeouts
3605 Details : Avoid synchronous journal commits to avoid delays caused by many
3606 clients connecting/disconnecting when bulk IO is in progress.
3607 Queue liblustre connect requests on OST_REQUEST_PORTAL instead of
3608 OST_IO_PORTAL to avoid delays behind potentially many pending
3612 Frequency : occasionally with multiple writers to a single file
3614 Description: shared writes to file may result in wrong size reported by stat()
3615 Details : Allow growing of kms when extent lock is cancelled
3618 Frequency : always with random mmap IO to multi-striped file
3620 Description: mmap write might be lost if we are writing to a 'hole' in stripe
3621 Details : Only if the hole is at the end of OST object so that kms is too
3622 small. Fix is to increase kms accordingly in ll_nopage.
3625 Frequency : rare, only if OST filesystem is inconsistent with MDS filesystem
3627 Description: writes to a missing object would leak memory on the OST
3628 Details : If there is an inconsistency between the MDS and OST filesystems,
3629 such that the MDS references an object that doesn't exist, writes
3630 to that object will leak memory due to incorrect cleanup in the
3631 error handling path, eventually running out of memory on the OST.
3636 Description: Creating too long symlink causes lustre errors
3637 Details : Check symlink and name lengths before sending requests to MDS.
3640 Frequency : only if flock is enabled (not on by default)
3642 Description: posix locks not released on fd closure on 2.6.9+
3643 Details : We failed to add posix locks to list of inode locks on 2.6.9+
3644 kernels, this caused such locks not to be released on fd close and
3645 then assertions on fs unmount about still used locks.
3648 Frequency : MDS failover only, very rarely
3650 Description: clients may get ASSERTION(granted_lock != NULL)
3651 Details : When request was taking a long time, and a client was resending
3652 a getattr by name lock request. The were multiple lock requests
3653 with the same client lock handle and
3654 mds_getattr_name->fixup_handle_for_resent_request found one of the
3655 lock handles but later failed with ASSERTION(granted_lock != NULL).
3660 Description: handle->h_buffer_credits > 0, assertion failure
3661 Details : h_buffer_credits is zero after truncate, causing assertion
3662 failure. This patch extends the transaction or creates a new
3666 Frequency : NFS re-export or patchless client
3667 Bugzilla : 11179, 10796
3668 Description: Crash on NFS re-export node (__d_move)
3669 Details : We do not want to hash the dentry if we don't have a lock.
3670 But if this dentry is later used in d_move, we'd hit uninitialised
3671 list head d_hash, so we just do this to init d_hash field but
3672 leave dentry unhashed.
3675 Frequency : NFS re-export or patchless client
3677 Description: NFS exports has problem with symbolic link
3678 Details : lustre client didn't properly install dentry when re-exported
3679 to NFS or running patchless client.
3682 Frequency : NFS re-export or patchless client
3684 Description: Various nfs/patchless fixes.
3685 Details : fixes reuse disconected alias for lookup process - this fixes
3686 warning "find_exported_dentry: npd != pd",
3687 fix permission error with open files at nfs.
3688 fix apply umask when do revalidate.
3691 Frequency : occasional
3693 Description: Crash on NFS re-export node
3694 Details : calling clear_page() on the wrong pointer triggered oops in
3695 generic_mapping_read().
3698 Frequency : rarely, using O_DIRECT IO
3700 Description: unaligned directio crashes client with LASSERT
3701 Details : check for unaligned buffers before trying any requests.
3704 Frequency : rarely, using CFS RAID5 patches in non-standard kernel series
3706 Description: stale data returned from RAID cache
3707 Details : If only a small amount of IO is done to the RAID device before
3708 reading it again it is possible to get stale data from the RAID
3709 cache instead of reading it from disk.
3712 Frequency : always for sles10 kernel
3714 Description: sles10 support
3715 Details : ll_follow_link: compile fixes and using of nd_set_link
3716 under newer kernels.
3719 Frequency : depends on arch, kernel and compiler version, always on sles10
3722 Description: recursive or deep enough symlinks cause stack overflow
3723 Details : getting rid of large stack-allocated variable in
3727 Frequency : depends on hardware
3729 Description: lustre write performance loss in the SLES10 kernel
3730 Details : the performance loss is caused by using of write barriers in the
3731 ext3 code. The SLES10 kernel turns barrier support on by
3732 default. The fix is to undo that change for ldiskfs.
3734 ------------------------------------------------------------------------------
3736 2006-12-09 Cluster File Systems, Inc. <info@clusterfs.com>
3738 * Support for kernels:
3739 2.6.9-42.0.3EL (RHEL 4)
3740 2.6.5-7.276 (SLES 9)
3741 2.4.21-47.0.1.EL (RHEL 3)
3742 2.6.12.6 vanilla (kernel.org)
3746 Frequency : quota enabled and large files being deleted
3748 Description: releasing more than 4GB of quota at once hangs OST
3749 Details : If a user deletes more than 4GB of files on a single OST it
3750 will cause the OST to spin in an infinite loop. Release
3751 quota in < 4GB chunks, or use a 64-bit value for 1.4.7.1+.
3756 Description: statfs data retrieved from /proc may be stale or zero
3757 Details : When reading per-device statfs data from /proc, in the
3758 {kbytes,files}_{total,free,avail} files, it may appear
3759 as zero or be out of date.
3762 Frequency : systems with MD RAID1 external journal devices
3764 Description: lconf's call to blkid is confused by RAID1 journal devices
3765 Details : Use the "blkid -l" flag to locate the MD RAID device instead
3766 of returning all block devices that match the journal UUID.
3769 Frequency : always, for aggregate stripe size over 4GB
3771 Description: "lfs setstripe" fails assertion when setting 4GB+ stripe width
3772 Details : Using "lfs setstripe" to set stripe size * stripe count over 4GB
3773 will fail the kernel with "ASSERTION(lsm->lsm_xfersize != 0)"
3776 Frequency : always if "lfs find" used on a local file/directory
3778 Description: "lfs find" segfaults if used on a local file/directory
3779 Details : The case where a directory component was not specified wasn't
3780 handled correctly. Handle this properly.
3783 Frequency : always on ppc64
3785 Description: the write to an ext3 filesystem mounted with mballoc got stuck
3786 Details : ext3_mb_generate_buddy() uses find_next_bit() which does not
3787 perform endianness conversion.
3790 Frequency : rarely (truncate to non-zero file size after write under load)
3791 Bugzilla : 10730, 10687
3792 Description: Files padded with zeros to next 4K multiple
3793 Details : With filesystems mounted using the "extents" option (2.6 kernels)
3794 it is possible that files that are truncated to a non-zero size
3795 immediately after being written are filled with zero bytes beyond
3796 the truncated size. No file data is lost.
3798 Severity : enhancement
3800 Description: Allow recovery/failover for liblustre clients.
3801 Details : liblustre clients were unaware of failover configurations until
3804 Severity : enhancement
3806 Description: user file locks should fail when not mounting with flock option
3807 Details : Set up an error-returning stub in ll_file_operations.lock field
3808 to prevent incorrect behaviour when client is mounted without
3809 flock option. Also, set up properly f_op->flock field for
3813 Frequency : always on ia64
3815 Description: "lfs df" loops on printing out MDS statfs information
3816 Details : The obd_ioctl_data was not initialized and in some systems
3817 this caused a failure during the ioctl that did not return
3818 an error. Initialize the struct and return an error on failure.
3821 Frequency : SLES 9 only
3823 Description: Error of copying files with lustre special EAs as root
3824 Details : Client side always return success for setxattr call for lustre
3825 special xattr (currently only "trusted.lov").
3828 Frequency : rarely on clusters with both ia64+i386 clients
3830 Description: ia64+i686 clients doing shared IO on the same file may LBUG
3831 Details : In rare cases when both ia64+i686 (or other mixed-PAGE_SIZE)
3832 clients are doing concurrent writes to the same file it is
3833 possible that the ia64 clients may LASSERT because the OST
3834 extent locks are not PAGE_SIZE aligned. Ensure that grown
3835 locks are always aligned on the request boundary.
3838 Frequency : specific use, occasional
3840 Description: Overwriting in use executable truncates on-disk binary image
3841 Details : If one node attempts to overwrite an executable in use by
3842 another node, we now correctly return ETXTBSY instead of
3843 truncating the file.
3845 Severity : enhancement
3847 Description: Async OSC create to avoid the blocking unnecessarily.
3848 Details : If an OST has no remaining object, system will block on the creating
3849 when need to create a new object on this OST. Now, ways use
3850 pre-created objects when available, instead of blocking on an
3851 empty osc while others are not empty. If we must block, we block
3852 for the shortest possible period of time.
3857 Description: chmod on Lustre root is propagated to other clients
3858 Details : Re-validate root's dentry in ll_lookup_it to avoid having it
3859 invalid by the follow_mount time.
3862 Frequency : liblustre clients only
3864 Description: Race in 'instant cancel' lock handling could lead to such locks
3865 never to be granted in case of SMP MDS
3866 Details : Do not destroy not yet granted but cbpending locks in
3870 Frequency : replay/resend of open
3872 Description: non null lock assetion failure in mds_intent_policy
3873 Details : Trying to replay/resend lockless open requests resulted in
3874 mds_open() returning 0 with no lock. Now it sets a flag if
3875 a lock is going to be returned.
3877 Severity : enhancement
3879 Description: Checksum enhancements
3880 Details : New checksum enhancements allow for resending RPCs that failed
3883 Severity : enhancement
3885 Description: Tunables on number of dirty pages in cacche
3886 Details : Allow to set limit on number of dirty pages cached.
3891 Description: client crash on unmount - lock still has references
3892 Details : In some error handling cases it was possible to leak a lock
3893 reference on a client while accessing a file. This was not
3894 harmful to the client during operation, but would cause the
3895 client to crash when the filesystem is unmounted.
3898 Frequency : specific case, rare
3900 Description: ETXTBSY on mds though file not in use by client
3901 Details : ETXTBSY is no longer incorrectly returned when attempting to
3902 chmod or chown a directory that the user previously tried to
3903 execute or a currently-executing binary.
3906 Frequency : extremely rare except on liblustre-based clients
3908 Description: Lustre space not freed when files are deleted
3909 Details : Clean up open-unlinked files after client eviction. Previously
3910 the unlink was skipped and the files remained as orphans.
3915 Description: OST failure "would be an LBUG" in waiting_locks_callback()
3916 Details : In some cases it was possible to send a blocking callback to
3917 a client doing a glimpse, even though that client didn't get
3918 a lock granted. When the glimpse lock is cancelled on the OST
3919 the freed lock is left on the waiting list and corrupted the list.
3922 Frequency : all core dumps
3924 Description: Broke core dumps to lustre
3925 Details : Negative dentry may be unhashed if parent does not have UPDATE
3926 lock, but some callers, e.g. do_coredump, expect dentry to be
3927 hashed after successful create, hash it in ll_create_it.
3929 ------------------------------------------------------------------------------
3931 2006-09-13 Cluster File Systems, Inc. <info@clusterfs.com>
3933 * Support for kernels:
3934 2.6.9-42.0.2.EL (RHEL 4)
3935 2.6.5-7.276 (SLES 9)
3936 2.4.21-40.EL (RHEL 3)
3937 2.6.12.6 vanilla (kernel.org)
3941 Frequency : always on RHEL 3
3943 Description: Number of open files grows over time
3944 Details : The number of open files grows over time, whether or not
3945 Lustre is started. This was due to a filp leak introduced
3946 by one of our kernel patches.
3948 ------------------------------------------------------------------------------
3950 08-20-2006 Cluster File Systems, Inc. <info@clusterfs.com>
3952 * Support for kernels:
3953 2.6.9-42.EL (RHEL 4)
3954 2.6.5-7.276 (SLES 9)
3955 2.4.21-40.EL (RHEL 3)
3956 2.6.12.6 vanilla (kernel.org)
3961 Bugzilla : 5719, 9635, 9792, 9684
3962 Description: OST (or MDS) trips assertions in (re)connection under heavy load
3963 Details : If a server is under heavy load and cannot reply to new
3964 connection requests before the client resends the (re)connect,
3965 the connection handling code can behave badly if two service
3966 threads are concurrently handing separate (re)connections from
3967 the same client. Add better locking to the connection handling
3968 code, and ensure that only a single connection will be processed
3969 for a given client UUID, even if the lock is dropped.
3971 Severity : enhancement
3973 Description: add TCP zero-copy support to kernel
3974 Details : Add support to the kernel TCP stack to allow zero-copy bulk
3975 sends if the hardware supports scatter-gather and checksumming.
3976 This allows socklnd to do client-write and server-read more
3977 efficiently and reduce CPU utilization from skbuf copying.
3980 Frequency : only if NFS exporting from client
3982 Description: NULL pointer deref in ll_iocontrol() if chattr mknod file
3983 Details : If setting attributes on a file created under NFS that had
3984 never been opened it would be possible to oops the client
3985 if the file had no objects.
3988 Frequency : always for liblustre
3990 Description: liblustre client does MDS+OSTs setattr RPC for each write
3991 Details : When doing a write from a liblustre client, the client
3992 incorrectly issued an RPC to the MDS and each OST the file was
3993 striped over in order to update the timestamps. When writing
3994 with small chunks and many clients this could overwhelm the MDS
3995 with RPCs. In all cases it would slow down the write because
3996 these RPCs are unnecessary.
3998 Severity : enhancement
4000 Description: allow number of MDS service threads to be changed at module load
4001 Details : It is now possible to change the number of MDS service threads
4002 running. Adding "options mds mds_num_threads={N}" to the MDS's
4003 /etc/modprobe.conf will set the number of threads for the next
4004 time Lustre is restarted (assuming the "mds" module is also
4005 reloaded at that time). The default number of threads will
4006 stay the same, 32 for most systems.
4011 Description: OST crash if filesystem is unformatted or corrupt
4012 Details : If an OST is started on a device that has never been formatted
4013 or if the filesystem is corrupt and cannot even mount then the
4014 error handling cleanup routines would dereference a NULL pointer.
4019 Description: NULL pointer deref in llap_from_page.
4020 Details : get_cache_page_nowait can return a page with NULL (or otherwise
4021 incorrect) mapping if the page was truncated/reclaimed while it was
4022 searched for. Check for this condition and skip such pages when
4023 doing readahead. Introduce extra check to llap_from_page() to
4024 verify page->mapping->host is non-NULL (so page is not anonymous).
4027 Frequency : Sometimes when using sys_sendfile
4029 Description: "page not covered by a lock" warnings from ll_readpage
4030 Details : sendfile called ll_readpage without right page locks present.
4031 Now we introduced ll_file_sendfile that does necessary locking
4032 around call to generic_file_sendfile() much like we do in
4036 Frequency : with certain MDS communication failures at client mount time
4038 Description: NULL pointer deref after failed client mount
4039 Details : a client connection request may delayed by the network layer
4040 and not be sent until after the PTLRPC layer has timed out the
4041 request. If the client fails the mount immediately it will try
4042 to clean up before the network times out the request. Add a
4043 reference from the request import to the obd device and delay
4044 the cleanup until the network drops the request.
4047 Frequency : occasionally during client (re)connect
4049 Description: assertion failure during client (re)connect
4050 Details : processing a client connection request may be delayed by the
4051 client or server longer than the client connect timeout. This
4052 causes the client to resend the connection request. If the
4053 original connection request is replied in this interval, the
4054 client may trip an assertion failure in ptlrpc_connect_interpret()
4055 which thought it would be the only running connect process.
4058 Frequency : only with obd_echo servers and clients that are rebooted
4060 Description: kernel BUG accessing uninitialized data structure
4061 Details : When running an obd_echo server it did not start the ping_evictor
4062 thread, and when a client was evicted an uninitialized data
4063 structure was accessed. Start the ping_evictor in the RPC
4064 service startup instead of the OBD startup.
4066 Severity : enhancement
4067 Bugzilla : 10193 (patchless)
4068 Description: Remove dependency on various unexported kernel interfaces.
4069 Details : No longer need reparent_to_init, exit_mm, exit_files,
4070 sock_getsockopt, filemap_populate, FMODE_EXEC, put_filp.
4073 Frequency : rare (only users of deprecated and unsupported LDAP config)
4075 Description: write_conf for zeroconf mount queried LDAP incorrectly for client
4076 Details : LDAP apparently contains 'lustreName' attributes instead of
4077 'name'. A simple remapping of the name is sufficient.
4080 Frequency : rare (only with non-default dump_on_timeout debug enabled)
4082 Description: waiting_locks_callback trips kernel BUG if client is evicted
4083 Details : Running with the dump_on_timeout debug flag turned on makes
4084 it possible that the waiting_locks_callback() can try to dump
4085 the Lustre kernel debug logs from an interrupt handler. Defer
4086 this log dumping to the expired_lock_main() thread.
4088 Severity : enhancement
4090 Description: Support NFS exporting on 2.6 kernels.
4091 Details : Implement non-rawops metadata methods for NFS server to use without
4092 changing NFS server code.
4095 Frequency : very rare (synthetic metadata workload only)
4097 Description: two racing renames might cause an MDS thread to deadlock
4098 Details : Running the "racer" program may cause one MDS thread to rename
4099 a file from being the source of a rename to being the target of
4100 a rename at exactly the same time that another thread is doing
4101 so, and the second thread has already enqueued these locks after
4102 doing a lookup of the target and is trying to relock them in
4103 order. Ensure that we don't try to re-lock the same resource.
4106 Frequency : only very large systems with liblustre clients
4108 Description: slow eviction of liblustre clients with the "evict_by_nid" RPC
4109 Details : Use asynchronous set_info RPCs to send the "evict_by_nid" to
4110 all OSTs in parallel. This allows the eviction of stale liblustre
4111 clients to proceed much faster than if they were done in series,
4112 and also offers similar improvements for other set_info RPCs.
4117 Description: excessive CPU usage during initial read phase on client
4118 Details : During the initial read phase on a client, it would agressively
4119 retry readahead on the file, consuming too much CPU and impacting
4120 performance (since 1.4.5.8). Improve the readahead algorithm
4121 to avoid this, and also improve some other common cases (read
4122 of small files in particular, where "small" is files smaller than
4123 /proc/fs/lustre/llite/*/max_read_ahead_whole_mb, 2MB by default).
4128 Description: MDS crash when receiving packet with unknown intent.
4129 Details : Do not LBUG in unknown intent case, just return -EFAULT
4131 Severity : enhancement
4132 Bugzilla : 9293, 9385
4133 Description: MDS RPCs are serialised on client. This is unnecessary for some.
4134 Details : Do not serialize getattr (non-intent version) and statfs.
4137 Frequency : occasional, when OST network is overloaded/intermittent
4139 Description: client evicted by OST after bulk IO timeout
4140 Details : If a client sends a bulk IO request (read or write) the OST
4141 may evict the client if it is unresposive to its data GET/PUT
4142 request. This is incorrect if the network is overloaded (takes
4143 too long to transfer the RPC data) or dropped the OST GET/PUT
4144 request. There is no need to evict the client at all, since
4145 the pinger and/or lock callbacks will handle this, and the
4146 client can restart the bulk request.
4149 Frequency : Always when mmapping file with no objects
4151 Description: client crashes when mmapping file with no objects
4152 Details : Check that we actually have objects in a file before doing any
4153 operations on objects in ll_vm_open, ll_vm_close and
4159 Description: Request leak when working with deleted CWD
4160 Details : Introduce advanced request refcount tracking for requests
4161 referenced from lustre intent.
4163 Severity : Enhancement
4165 Description: Cache open file handles on client.
4166 Details : MDS now will return special lock along with openhandle, if
4167 requested and client is allowed to hold openhandle, even if unused,
4168 until such a lock is revoked. Helps NFS a lot, since NFS is opening
4169 closing files for every read/write openration.
4171 Severity : Enhancement
4173 Description: Cache open negative dentries on client when possible.
4174 Details : Guard negative dentries with UPDATE lock on parent dir, drop
4175 negative dentries on lock revocation.
4180 Description: Remounting a client read-only wasn't possible with a zconf mount
4181 Details : It wasn't possible to remount a client read-only with llmount.
4183 Severity : enhancement
4184 Description: Include MPICH 1.2.6 Lustre ADIO interface patch
4185 Details : In lustre/contrib/ or /usr/share/lustre in RPM a patch for
4186 MPICH is included to add Lustre-specific ADIO interfaces.
4187 This is based closely on the UFS ADIO layer and only differs
4188 in file creation, in order to allow the OST striping to be set.
4189 This is user-contributed code and not supported by CFS.
4194 Description: extended inode attributes (immutable, append-only) work improperly
4195 when 2.4 and 2.6 kernels are used on client/server or vice versa
4196 Details : Introduce kernel-independent values for these flags.
4198 Severity : enhancement
4201 Description: Allow fractional MB tunings for lustre in /proc/ filesystem.
4202 Details : Many of the /proc/ tunables can only be tuned at a megabyte
4203 granularity. Now, Fractional MB granularity is be supported,
4204 this is very useful for low memory system.
4206 Severity : enhancement
4208 Description: Getattr by fid
4209 Details : Getting a file attributes by its fid, obtaining UPDATE|LOOKUP
4210 locks, avoids extra getattr rpc requests to MDS, allows '/' to
4211 have locks and avoids getattr rpc requests for it on every stat.
4214 Frequency : Always, for filesystems larger than 2TB
4216 Description: ldiskfs crash at mount for filesystem larger than 2TB with mballoc
4217 Details : Kenrel kmalloc limits allocations to 128kB and this prevents
4218 filesystems larger than 2TB to be mounted with mballoc enabled.
4221 Frequency : Always, for 32-bit kernel without CONFIG_LBD and filesystem > 2TB
4223 Description: ldiskfs crash at mount for filesystem larger than 2TB with mballoc
4224 Details : If a 32-bit kernel is compiled without CONFIG_LBD enabled and a
4225 filesystems larger than 2TB is mounted then the kernel will
4226 silently corrupt the start of the filesystem. CONFIG_LBD is
4227 enabled for all CFS-supported kernels, but the possibility of
4228 this happening with a modified kernel config exists.
4230 Severity : enhancement
4232 Description: add client O_DIRECT support for 2.6 kernels
4233 Details : It is now possible to do O_DIRECT reads and writes to files
4234 in the Lustre client mountpoint on 2.6 kernel clients.
4236 Severity : enhancement
4238 Description: parallel glimpse, setattr, statfs, punch, destroy requests
4239 Details : Sends glimpse, setattr, statfs, punch, destroy requests to OSTs in
4240 parallel, not waiting for response from every OST before sending
4241 a rpc to the next OST.
4246 Description: setattr vs write race when updating file timestamps
4247 Details : Client processes that update a file timestamp into the past
4248 right after writing to the file (e.g. tar) it is possible that
4249 the updated file modification time can be reset to the current
4250 time due to a race between processing the setattr and write RPC.
4252 Severity : enhancement
4254 Description: Bring 'lfs find' closer in line with regular Linux find.
4255 Details : lfs find util supports -atime, -mtime, -ctime, -maxdepth, -print,
4256 -print0 options and obtains all the needed info through the lustre
4259 Severity : enhancement
4261 Description: support up to 1024 configured devices on one node
4262 Details : change obd_dev array from statically allocated to dynamically
4263 allocated structs as they are first used to reduce memory usage
4268 Description: Flush dirty partially truncated pages during truncate
4269 Details : Immediatelly flush partially truncated pages in filter_setattr,
4270 this way we completely avoid having any pages in page cache on OST
4271 and can retire ugly workarounds during writes to flush such pages.
4276 Description: i_sem vs transaction deadlock in mds_obd_destroy during unlink.
4277 Details : protect inode from truncation within vfs_unlink() context
4278 just take a reference before calling vfs_unlink() and release it
4279 when parent's i_sem is free.
4284 Description: last_id value checked outside lock on OST caused LASSERT failure
4285 Details : If there were multiple MDS->OST object precreate requests in
4286 flight, it was possible that the OST's last object id was checked
4287 outside a lock and incorrectly tripped an assertion. Move checks
4288 inside locks, and discard old precreate requests.
4291 Frequency : always, if extents are used on OSTs
4293 Description: index ei_leaf_hi (48-bit extension) is not zeroed in extent index
4294 Details : OSTs using the extents format would not zero the high 16 bits of
4295 the index physical block number. This is not a problem for any
4296 OST filesystems smaller than 16TB, and no kernels support ext3
4297 filesystems larger than 16TB yet. This is fixed in 1.4.7 (all
4298 new/modified files) and can be fixed for existing filesystems
4299 with e2fsprogs-1.39-cfs1.
4304 Description: import connection selection may be incorrect if timer wraps
4305 Details : Using a 32-bit jiffies timer with HZ=1000 may cause backup
4306 import connections to be ignored if the 32-bit jiffies counter
4307 wraps. Use a 64-bit jiffies counter.
4310 Frequency : very large clusters immediately after boot
4312 Description: LNET request buffers exhausted under heavy short-term load
4313 Details : If a large number of client requests are generated on a service
4314 that has previously never seen so many requests it is possible
4315 that the request buffer growth cannot keep up with the spike in
4316 demand. Instead of dropping incoming requests, they are held in
4317 the LND until the RPC service can accept more requests.
4320 Frequency : Sometimes during replay
4322 Description: Assertion failure in ll_local_open after replay.
4323 Details : If replay happened on an open request reply before we were able
4324 to set replay handler, reply will become not swabbed tripping the
4325 assertion in ll_local_open. Now we set the handler right after
4326 recognising of open request
4329 Frequency : very rare
4331 Description: kernel reports "badness in vsnprintf"
4332 Details : Reading from the "recovery_status" /proc file in small chunks
4333 may cause a negative length in lprocfs_obd_rd_recovery_status()
4334 call to vsnprintf() (which is otherwise harmless). Exit early
4335 if there is no more space in the output buffer.
4337 Severity : enhancement
4339 Description: clear OBD RPC statistics by writing to them
4340 Details : It is now possible to clear the OBD RPC statistics by writing
4341 to the "stats" file.
4346 Description: Client mtime is not the same on different clients after utimes
4347 Details : In some cases, the client was using the utimes() syscall on
4348 a file cached on another node. The clients now validate the
4349 ctime from the MDS + OSTs to determine which one is right.
4354 Description: Inability to activate failout mode
4355 Details : lconf script incorrectly assumed that in pythong string's numeric
4356 value is used in comparisons.
4359 Frequency : always with multiple stripes per file
4361 Description: Inefficient object allocation for mutli-stripe files
4362 Details : When selecting which OSTs to stripe files over, for files with
4363 a stripe count that divides evenly into the number of OSTs,
4364 the MDS is always picking the same starting OST for each file.
4365 Return the OST selection heuristic to the original design.
4370 Description: mount failures may take full timeout to return an error
4371 Details : Under some heavy load conditions it is possible that a
4372 failed mount can wait for the full obd_timeout interval,
4373 possibly several minutes, before reporting an error.
4374 Instead return an error as soon as the status is known.
4376 Frequency : quota enabled and large files being deleted
4378 Description: releasing more than 4GB of quota at once hangs OST
4379 Details : If a user deletes more than 4GB of files on a single OST it
4380 will cause the OST to spin in an infinite loop. Release
4381 quota in < 4GB chunks, or use a 64-bit value for 1.4.7.1+.
4386 Description: statfs data retrieved from /proc may be stale or zero
4387 Details : When reading per-device statfs data from /proc, in the
4388 {kbytes,files}_{total,free,avail} files, it may appear
4389 as zero or be out of date.
4392 Frequency : systems with MD RAID1 external journal devices
4394 Description: lconf's call to blkid is confused by RAID1 journal devices
4395 Details : Use the "blkid -l" flag to locate the MD RAID device instead
4396 of returning all block devices that match the journal UUID.
4399 Frequency : always, for aggregate stripe size over 4GB
4401 Description: assertion fails when trying to use 4GB stripe size
4402 Details : Use "setstripe" to set stripe size over 4GB will fail the kernel,
4403 complaining "ASSERTION(lsm->lsm_xfersize != 0)"
4406 Frequency : always on ppc64
4408 Description: the first write on an ext3 filesystem with mballoc got stuck
4409 Details : ext3_mb_generate_buddy() uses find_next_bit() which does not
4410 perform endianness conversion.
4412 ------------------------------------------------------------------------------
4414 02-14-2006 Cluster File Systems, Inc. <info@clusterfs.com>
4416 * WIRE PROTOCOL CHANGE. This version of Lustre networking WILL NOT
4417 INTEROPERATE with older versions automatically. Please read the
4418 user documentation before upgrading any part of a live system.
4419 * WARNING: Lustre networking configuration changes are required with
4420 this release. See https://bugzilla.clusterfs.com/show_bug.cgi?id=10052
4423 * Support for kernels:
4424 2.6.9-22.0.2.EL (RHEL 4)
4425 2.6.5-7.244 (SLES 9)
4426 2.6.12.6 vanilla (kernel.org)
4429 Severity : enhancement
4430 Bugzilla : 7981/8208
4431 Description: Introduced Lustre Networking (LNET)
4432 Details : LNET is new networking infrastructure for Lustre, it includes
4433 a reorganized network configuration mode (see the user
4434 documentation for full details) as well as support for routing
4435 between different network fabrics. Lustre Networking Devices
4436 (LNDs) for the supported network fabrics have also been
4437 created for this new infrastructure.
4439 Severity : enhancement
4440 Description: Introduced Access control lists
4441 Details : clients can set ACLs on files and directories in order to have
4442 more fine-grained permissions than the standard Unix UGO+RWX.
4443 The MDS must be started with the "-o acl" mount option.
4445 Severity : enhancement
4446 Description: Introduced filesystem quotas
4447 Details : Administrators may now establish per-user quotas on the
4450 Severity : enhancement
4452 Description: Configuration change for the XT3
4453 The PTLLND is now used to run Lustre over Portals on the XT3
4454 The configure option(s) --with-cray-portals are no longer used.
4455 Rather --with-portals=<path-to-portals-includes> is used to
4456 enable building on the XT3. In addition to enable XT3 specific
4457 features the option --enable-cray-xt3 must be used.
4462 Description: Running on many-way SMP OSTs can trigger oops in llcd_send()
4463 Details : A race between allocating a new llcd and re-getting the llcd_lock
4464 allowed another thread to grab newly-allocated llcd.
4466 Severity : enhancement
4468 Description: 2.6 OST async journal commit and locking fix to improve performance
4469 Details : The filter_direct_io()+filter_commitrw_write() journal commits for
4470 2.6 kernels are now async as they already were in 2.4 kernels so
4471 that they can commit concurrently with the network bulk transfer.
4472 For block-allocated files the filter allocation semaphore is held
4473 to avoid filesystem fragmentation during allocation. BKL lock
4474 removed for 2.6 xattr operations where it is no longer needed.
4479 Description: lconf incorrectly determined whether two IP networks could talk
4480 Details : In some more complicated routing and multiple-network
4481 configurations, lconf will avoid trying to make a network
4482 connection to a disjoint part of the IP space. It was doing the
4483 math incorrectly for one set of cases.
4488 Description: Fix for potential infinite loop processing records in an llog.
4489 Details : If an llog record is corrupted/zeroed, it is possible to loop
4490 forever in llog_process(). Validate the llog record length
4491 and skip the remainder of the block on error.
4494 Frequency : occasional (liblustre only)
4496 Description: liblustre could not open files whose last component is a symlink
4497 Details : sysio_path_walk() would incorrectly pass the open intent to
4498 intermediate path components.
4501 Frequency : rare (liblustre only with non-standard tuning)
4502 Bugzilla : 7201 (7350)
4503 Description: Tuning the MDC DLM LRU size to zero triggers client LASSERT
4504 Details : llu_lookup_finish_locks() tries to set lock data on a lock
4505 after it has been released, only do this for referenced locks
4507 Severity : enhancement
4509 Description: specifying an (invalid) directory default stripe_size of -1
4510 would reset the directory default striping
4511 Details : stripe_size -1 was used internally to signal directory stripe
4512 removal, now use "all default" to signal dir stripe removal
4513 as a directory striping of "all default" is not useful
4516 Frequency : common for large clusters running liblustre clients
4518 Description: doing an ls when liblustre clients are running is slow
4519 Details : sending a glimpse AST to a liblustre client waits for every AST
4520 to time out, as liblustre clients will not respond. Since they
4521 cannot cache data we refresh the OST lock LVB from disk instead.
4523 Severity : enhancement
4525 Description: doing an ls at the same time as file IO can be slow
4526 Details : enqueue and other "small" requests can be blocked behind many
4527 large IO requests. Create a new OST IO portal for non-IO
4528 requests so they can be processed faster.
4531 Frequency : rare (only HPUX clients mounting unsupported re-exported NFS vol)
4533 Description: an HPUX NFS client would get -EACCESS when ftruncate()ing a newly
4534 created file with mode 000
4535 Details : the Linux NFS server relies on an MDS_OPEN_OWNEROVERRIDE hack to
4536 allow an ftruncate() as a non-root user to a file with mode 000.
4537 Lustre now respects this flag to disable mode checks when
4538 truncating a file owned by the user
4541 Frequency : liblustre-only, when liblustre client dies unexpectedly or becomes
4544 Description: Revoking locks from clients that went dead or catatonic might take
4546 Details : New lock flags FL_CANCEL_ON_BLOCK used by liblustre makes
4547 cancellation of such locks instant on servers without waiting for
4548 any reply from clients. Clients drops these locks when cancel
4549 notification from server is received without replying.
4552 Frequency : liblustre-only, when liblustre client dies or becomes busy
4554 Description: Doing ls on Linux clients can take a long time with active
4556 Details : Liblustre client cannot handle ASTs in timely manner, so avoid
4557 granting such locks to it in the first place if possible. Locks
4558 are taken by proxy on the OST during the read or write and
4559 dropped immediately afterward. Add connect flags handling, do
4560 not grant locks to liblustre clients for glimpse ASTs.
4562 Severity : enhancement
4564 Description: Improve read-ahead algorithm to avoid excessive IO for random reads
4565 Details : Existing read-ahead algorithm is tuned for the case of streamlined
4566 sequential reads and behaves badly with applications doing random
4567 reads. Improve it by reading ahead at least read region, and
4568 avoiding excessive large RPC for small reads.
4570 Severity : enhancement
4572 Description: Creating more than 1000 files for a single job may cause a load
4573 imbalance on the OSTs if there are also a large number of OSTs.
4574 Details : qos_prep_create() uses an OST index reseed value that is an
4575 even multiple of the number of available OSTs so that if the
4576 reseed happens in the middle of the object allocation it will
4577 still utilize the OSTs as uniformly as possible.
4582 Description: OST or MDS may oops in ping_evictor_main()
4583 Details : ping_evictor_main() drops obd_dev_lock if deleting a stale export
4584 but doesn't restart at beginning of obd_exports_timed list
4587 Severity : enhancement
4589 Description: improve by-nid export eviction on the MDS and OST
4590 Details : allow multiple exports with the same NID to be evicted at one
4591 time without re-searching the exports list.
4594 Frequency : rare, only with supplementary groups enabled on SMP 2.6 kernels
4596 Description: MDS may oops in groups_free()
4597 Details : in rare race conditions a newly allocated group_info struct is
4598 freed again, and this can be NULL. The 2.4 compatibility code
4599 for groups_free() checked for a NULL pointer, but 2.6 did not.
4602 Frequency : common for liblustre clients doing little filesystem IO
4603 Bugzilla : 9352, 7313
4604 Description: server may evict liblustre clients accessing contended locks
4605 Details : if a client is granted a lock or receives a completion AST
4606 with a blocking AST already set it would not reply to the AST
4607 for LDLM_FL_CANCEL_ON_BLOCK locks. It now replies to such ASTs.
4610 Frequency : lfs setstripe, only systems with more than 160 OSTs
4612 Description: unable to set striping with a starting offset beyond OST 160
4613 Details : llapi_create_file() incorrectly limited the starting stripe
4614 index to the maximum single-file stripe count.
4617 Frequency : LDAP users only
4619 Description: lconf did not handle in-kernel recovery with LDAP properly
4620 Details : lconf/LustreDB get_refs() is searching the wrong namespace
4622 Severity : enhancement
4624 Description: bind OST threads to NUMA nodes to improve performance
4625 Details : all OST threads are uniformly bound to CPUs on a single NUMA
4626 node and do their allocations there to localize memory access
4628 Severity : enhancement
4630 Description: llmount can determine client NID directly from Myrinet (GM)
4631 Details : the client NID code from gmnalnid was moved directly into
4632 llmount, removing the need to use this or specifying the
4633 client NID explicitly when mounting GM clients with zeroconf
4636 Frequency : if client is started with down MDS
4638 Description: if client is started with down MDS mount hangs in ptlrpc_queue_wait
4639 Details : Having an LWI_INTR() wait event (interruptible, but no timeout)
4640 will wait indefinitely in ptlrpc_queue_wait->l_wait_event() after
4641 ptlrpc_import_delayed_req() because we didn't check if the
4642 request was interrupted, and we also didn't break out of the
4643 event loop if there was no timeout
4648 Description: data loss during non-page-aligned writes to a single file from
4649 both multiple nodes and multiple threads on one node at same time
4650 Details : updates to KMS and lsm weren't protected by common lock. Resulting
4651 inconsistency led to false short-reads, that were cached and later
4652 used by ->prepare_write() to fill in partially written page,
4653 leading to data loss.
4656 Frequency : always, if lconf --abort_recovery used
4658 Description: lconf --abort_recovery fails with 'Operation not supported'
4659 Details : lconf was attempting to abort recovery on the MDT device and not
4662 Severity : enhancement
4664 Description: remove cleanup logs
4665 Details : replace lconf-generated cleanup logs with lustre internal
4666 cleanup routines. Eliminates the need for client-cleanup and
4669 Severity : enhancement
4671 Description: add support for EAs (user and system) on lustre filesystems
4672 Details : it is now possible to store extended attributes in the Lustre
4673 client filesystem, and with the user_xattr mount option it
4674 is possible to allow users to store EAs on their files also
4676 Severity : enhancement
4678 Description: Add possibility (config option) to show minimal available OST free
4680 Details : When compiled with --enable-mindf configure option, statfs(2)
4681 (and so, df) will return least minimal free space available from
4682 all OSTs as amount of free space on FS, instead of summary of
4683 free spaces of all OSTs.
4685 Severity : enhancement
4687 Description: do not expand extent locks acquired on OST-side
4688 Details : Modify ldlm_extent_policy() to not expand local locks, acquired
4689 by server: they are not cached anyway.
4692 Frequency : when mmap is used/binaries executed from Lustre
4694 Description: Unmmap pages before throwing them away from read cache.
4695 Details : llap_shrink cache now attempts to unmap pages before discarding
4696 them (if unmapping failed - do not discard). SLES9 kernel has
4697 extra checks that trigger if this unmapping is not done first.
4702 Description: lconf didn't resolve symlinks before checking to see whether a
4703 given mountpoint was already in use
4706 Frequency : when migrating failover services
4707 Bugzilla : 6395, 9514
4708 Description: When migrating a subset of services from a node (e.g. failback
4709 from a failover service node) the remaining services would
4710 time out and evict clients.
4711 Details : lconf --force (implied by --failover) sets the global obd_timeout
4712 to 5 seconds in order to quickly disconnect, but this caused
4713 other RPCs to time out too quickly. Do not change the global
4714 obd_timeout for force cleanup, only set it for DISCONNECT RPCs.
4716 Severity : enhancement
4717 Frequency : if MDS is started with down OST
4718 Bugzilla : 9439,5706
4719 Description: Allow startup/shutdown of an MDS without depending on the
4720 availability of the OSTs.
4721 Details : Asynchronously call mds_lov_synchronize during MDS startup.
4722 Add appropriate locking and lov-osc refcounts for safe
4723 cleaning. Add osc abort_inflight calls in case the
4724 synchronize never started.
4727 Frequency : occasional (Cray XT3 only)
4729 Description: root not authorized to access files in CRAY_PORTALS environment
4730 Details : The client process capabilities were not honoured on the MDS in
4731 a CRAY_PORTALS/CRAY_XT3 environment. If the file had previously
4732 been accessed by an authorized user then root was able to access
4733 the file on the local client also. The root user capabilities
4734 are now allowed on the MDS, as this environment has secure UID.
4737 Frequency : occasional
4739 Description: ldiskfs "too long searching" message happens too often
4740 Details : A debugging message (otherwise harmless) prints too often on
4741 the OST console. This has been reduced to only happen when
4742 there are fragmentation problems on the filesystem.
4747 Description: Division by zero in statfs when all OSCs are inactive
4748 Details : lov_get_stripecnt() returns zero due to incorrect order of checks,
4749 lov_statfs divides by value returned by lov_get_stripecnt().
4753 Bugzilla : 9489, 3273
4754 Description: First write from each client to each OST was only 4kB in size,
4755 to initialize client writeback cache, which caused sub-optimal
4756 RPCs and poor layout on disk for the first written file.
4757 Details : Clients now request an initial cache grant at (re)connect time
4758 and so that they can start streaming writes to the cache right
4759 away and always do full-sized RPCs if there is enough data.
4760 If the OST is rebooted the client also re-establishes its grant
4761 so that client cached writes will be honoured under the grant.
4766 Description: Slow ls (and stat(2) syscall) on files residing on IO-loaded OSTs
4767 Details : Now I/O RPCs go to different portal number and (presumably) fast
4768 lock requests (and glimses) and other RPCs get their own service
4769 threads pool that should be able to service those RPCs
4772 Severity : enhancement
4774 Description: Ability to exchange lustre version between client and servers and
4775 issue warnings at client side if client is too old. Also for
4776 liblustre clients there is ability to refuse connection of too old
4778 Details : New 'version' field is added to connect data structure that is
4779 filled with version info. That info is later checked by server and
4783 Frequency : rare, liblustre only.
4784 Bugzilla : 9296, 9581
4785 Description: Two simultaneous writes from liblustre at offset within same page
4786 might proceed at the same time overwriting eachother with stale
4788 Details : I/O lock withing llu_file_prwv was released too early, before data
4789 actually was hitting the wire. Extended lock-holding time until
4790 server acknowledges receiving data.
4793 Frequency : extremely rare. Never observed in practice.
4795 Description: avoid generating lustre_handle cookie of 0.
4796 Details : class_handle_hash() generates handle cookies by incrementing
4797 global counter, and can hit 0 occasionaly (this is unlikely, but
4798 not impossible, because initial value of cookie counter is
4799 selected randonly). Value of 0 is used as a sentinel meaning
4800 "unassigned handle" --- avoid it. Also coalesce two critical
4801 sections in this function into one.
4803 Severity : enhancement
4805 Description: allow liblustre clients to delegate truncate locking to OST
4806 Details : To avoid overhead of locking, liblustre client instructs OST to
4807 take extent lock in ost_punch() on client's behalf. New connection
4808 flag is added to handle backward compatibility.
4810 Severity : enhancement
4811 Bugzilla : 4928, 7341, 9758
4812 Description: allow number of OST service threads to be specified
4813 Details : a module parameter allows the number of OST service threads
4814 to be specified via "options ost ost_num_threads={N}" in the
4815 OSS's /etc/modules.conf or /etc/modprobe.conf.
4819 Bugzilla : 6146, 9635, 9895
4820 Description: servers crash with bad pointer in target_handle_connect()
4821 Details : In rare cases when a client is reconnecting it was possible that
4822 the connection request was the last reference for that export.
4823 We would temporarily drop the export reference and get a new
4824 one, but this may have been the last reference and the export
4825 was just destroyed. Get new reference before dropping old one.
4827 Severity : enhancement
4828 Frequency : if client is started with failover MDS
4830 Description: Allow multiple MDS hostnames in the mount command
4831 Details : Try to read the configuration from all specified MDS
4832 hostnames during a client mount in case the "primary"
4835 Severity : enhancement
4837 Description: Stop sending data to evicted clients as soon as possible.
4838 Details : Check if the client we are about to send or are sending data to
4839 was evicted already. (Check is done every second of waiting,
4840 for which l_wait_event interface was extended to allow checking
4841 of exit condition at specified intervals).
4844 Frequency : rare, normally only when NFS exporting is done from client
4846 Description: 'bad disk LOV MAGIC: 0x00000000' error when chown'ing files
4848 Details : Make mds_get_md() recognise empty md case and set lmm size to 0.
4851 Frequency : always, if srand() is called before liblustre initialization
4853 Description: Liblustre uses system PRNG disturbing its usage by user application
4854 Details : Introduce internal to lustre fast and high-quality PRNG for
4855 lustre usage and make liblustre and some other places in generic
4856 lustre code to use it.
4858 Severity : enhancement
4859 Bugzilla : 9477, 9557, 9870
4860 Description: Verify that the MDS configuration logs are updated when xml is
4861 Details : Check if the .xml configuration logs are newer than the config
4862 logs stored on the MDS and report an error if this is the case.
4863 Request --write-conf, or allow starting with --old_conf.
4865 Severity : enhancement
4867 Description: Handle symlinks in the path when checking if Lustre is mounted.
4868 Details : Resolve intermediate symlinks when checking if a client has
4869 mounted a filesystem to avoid duplicate client mounts.
4874 Description: lconf can hit an error exception but still return success.
4875 Details : The lconf command catches the Command error exception at the top
4876 level script context and will exit with the associated exit
4877 status, but doesn't ensure that this exit status is non-zero.
4882 Description: failure of ptlrpc thread startup can cause oops
4883 Details : Starting a ptlrpc service thread can fail if there are a large
4884 number of threads or the server memory is very fragmented.
4885 Handle this without oopsing.
4888 Frequency : always, only if liblustre and non-default acceptor port was used
4890 Description: liblustre cannot connect to servers with non-default acceptor port
4891 Details : tcpnal_set_default_params() was not called and was therefore
4892 ignoring the environment varaible TCPNAL_PORT, as well as other
4893 TCPNAL_ environment variables
4898 Description: two objects could be created on the same OST for a single file
4899 Details : If an OST is down, in some cases it was possible to create two
4900 objects on a single OST for a single file. No problems other
4901 than potential performance impact and spurious error messages.
4905 Bugzilla : 5681, 9562
4906 Description: Client may oops in ll_unhash_aliases
4907 Details : Client dcache may become inconsistent in race condition.
4908 In some cases "getcwd" can fail if the current directory is
4914 Description: Inode refcounting problems in NFS export code
4915 Details : link_raw functions used to call d_instantiate without obtaining
4916 extra inode reference first.
4920 Bugzilla : 9942, 9903
4921 Description: Referencing freed requests leading to crash, memleaks with NFS.
4922 Details : We used to require that call to ll_revalidate_it was always
4923 followed by ll_lookup_it. Also with revalidate_special() it is
4924 possible to call ll_revalidate_it() twice for the same dentry
4925 even if first occurence returned success. This fix changes semantic
4926 between DISP_ENQ_COMPLETE disposition flag to mean there is extra
4927 reference on a request referred from the intent.
4928 ll_intent_release() then releases such a request.
4931 Frequency : rare, normally benchmark loads only
4933 Description: unlinked inodes were kept in memory on the client
4934 Details : If a client is repeatedly creating and unlinking files it
4935 can accumulate a lot of stale inodes in the inode slab cache.
4936 If there is no other client load running this can cause the
4937 client node to run out of memory. Instead flush old inodes
4938 from client cache that have the same inode number as a new inode.
4941 Frequency : SLES9 2.6.5 kernel and long filenames only
4942 Bugzilla : 9969, 10379
4943 Description: utime reports stale NFS file handle
4944 Details : SLES9 uses out-of-dentry names in some cases, which confused
4945 the lustre dentry revalidation. Change it to always use the
4949 Frequency : rare, unless heavy write-truncate concurrency is continuous
4950 Bugzilla : 4180, 6984, 7171, 9963, 9331
4951 Description: OST becomes very slow and/or deadlocked during object unlink
4952 Details : filter_destroy() was holding onto the parent directory lock
4953 while truncating+unlinking objects. For very large objects this
4954 may block other threads for a long time and slow overall OST
4955 responsiveness. It may also be possible to get a lock ordering
4956 deadlock in this case, or run out of journal credits because of
4957 the combined truncate+unlink. Solution is to do object truncate
4958 first in one transaction without parent lock, and then do the
4959 final unlink in a new transaction with the parent lock. This
4960 reduces the lock hold time dramatically.
4963 Frequency : rare, 2.4 kernels only
4965 Description: MDS or OST cleanup may trip kernel BUG when dropping kernel lock
4966 Details : mds_cleanup() and filter_cleanup() need to drop the kernel lock
4967 before unmounting their filesystem in order to avoid deadlock.
4968 The kernel_locked() function in 2.4 kernels only checks whether
4969 the kernel lock is held, not whether it is this process that is
4970 holding it as 2.6 kernels do.
4975 Description: MDS or OST may oops/LBUG if a client is connecting multiple times
4976 Details : The client ptlrpc code may be trying to reconnect to a down
4977 server before a previous connection attempt has timed out.
4978 Increase the reconnect interval to be longer than the connection
4979 timeout interval to avoid sending duplicate connections to
4983 Frequency : echo_client brw_test command
4985 Description: fix echo_client to work with OST preallocated code
4986 Details : OST preallocation code (5137) didn't take echo_client IO path
4987 into account: echo_client calls filter methods outside of any
4988 OST thread and, hence, there is no per-thread preallocated
4989 pages and buffers to use. Solution: hijack pga pages for IO. As
4990 a byproduct, this avoids unnecessary data copying.
4994 Bugzilla : 3555, 5962, 6025, 6155, 6296, 9574
4995 Description: Client can oops in mdc_commit_close() after open replay
4996 Details : It was possible for the MDS to return an open request with no
4997 transaction number in mds_finish_transno() if the client was
4998 evicted, but without actually returning an error. Clients
4999 would later try to replay that open and may trip an assertion
5000 Simplify the client close codepath, and always return an error
5001 from the MDS in case the open is not successful.
5004 Frequency : rare, 2.6 OSTs only
5006 Description: OST may deadlock under high load on fragmented files
5007 Details : If there was a heavy load and highly-fragmented OST filesystems
5008 it was possible to have all the OST threads deadlock waiting on
5009 allocation of biovecs, because the biovecs were not released
5010 until the entire RPC IO was completed. Instead, release biovecs
5011 as soon as they are complete to ensure forward IO progress.
5013 Severity : enhancement
5015 Description: Support for specifying external journal device at mount
5016 Details : If an OST or MDS device is formatted with an external journal
5017 device, this device major/minor is stored in the ext3 superblock
5018 and may not be valid for failover. Allow detecting and
5019 specifying the external journal at mount time.
5024 Description: Mounting an MDS with pending unlinked files may cause oops
5025 Details : target_finish_recovery() calls mds_postrecov() which returned
5026 the number of orphans unlinked. mds_lov_connect->mds_postsetup()
5027 considers this an error and immediately begins cleaning up the
5028 lov, just after starting the mds_lov process
5030 Severity : enhancement
5032 Description: Implement 'lfs df' to report actual free space on per-OST basis
5033 Details : Add sub-command 'df' on 'lfs' to report the disk space usage of
5034 MDS/OSDs. Usage: lfs df [-i][-h]. Command Options: '-i' to report
5035 usage of objects; '-h' to report in human readable format.
5037 ------------------------------------------------------------------------------
5039 08-26-2005 Cluster File Systems, Inc. <info@clusterfs.com>
5046 Description: Mounting an ldiskfs file system with mballoc may crash OST node.
5047 Details : ldiskfs mballoc code may reference an uninitialized buddy struct
5048 at startup during orphan unlinking. Instead, skip buddy update
5049 before setup, as it will be regenerated after recovery is complete.
5054 Description: If an OST is inactive, its locks might reference stale inodes.
5055 Details : lov_change_cbdata() must iterate over all namespaces, even if
5056 they are inactive to clear inode references from the lock.
5058 Severity : enhancement
5059 Frequency : occasional, if non-standard max_dirty_mb used
5061 Description: Client will block write RPCs if not enough grant
5062 Details : If a client has max_dirty_mb smaller than max_rpcs_in_flight,
5063 then the client will block writes while waiting for another RPC
5064 to complete instead of consuming its dirty limit. With change
5065 we get improved performance when max_dirty_mb is small.
5067 Severity : enhancement
5068 Bugzilla : 3389, 6253
5069 Description: Add support for supplementary groups on the MDS.
5070 Details : The MDS has an upcall /proc/fs/lustre/mds/{mds}/group_upcall
5071 (set to /usr/sbin/l_getgroups if enabled) which will do MDS-side
5072 lookups for user supplementary groups into a cache.
5076 Description: O_CREAT|O_EXCL open flags in liblustre always return -EEXIST
5077 Details : Make libsysio to not enforce O_EXCL by clearing the flag,
5078 for liblustre O_EXCL is enforced by MDS.
5082 Description: readdir never returns NULL in liblustre.
5083 Details : Corrected llu_iop_getdirentries logic, to return offset of next
5084 dentry in struct dirent.
5088 Frequency : liblustre only, depends on application IO pattern
5089 Description: liblustre clients evicted if not contacting servers
5090 Details : Don't put liblustre clients into the ping_evictor list, so
5091 they will not be evicted by the pinger ever.
5093 Severity : enhancement
5095 Description: Add ability to evict clients by NID from MDS.
5096 Details : By echoing "nid:$NID" string into
5097 /proc/fs/lustre/mds/.../evict_client client with nid that equals to
5098 $NID would be instantly evicted from this MDS and from all active
5099 OSTs connected to it.
5103 Description: Do not query file size twice, somewhat slowing stat(2) calls.
5104 Details : lookup_it_finish() used to query file size from OSTs that was not
5109 Description: service threads change working directory to that of init
5110 Details : Starting lustre service threads may pin the working directory
5111 of the parent thread, making that filesystem busy. Threads
5112 now change to the working directory of init to avoid this.
5116 Frequency : during shutdown only
5117 Description: shutdown with a failed MDS or OST can cause unmount to hang
5118 Details : Don't resend DISCONNECT messages in ptlrpc_disconnect_import()
5123 Frequency : 2.6 only
5124 Description: chmod/chown may include an extra supplementary group
5125 Details : ll{,u}_mdc_pack_op_data() does not properly initialize the
5126 supplementary group and if none is specified this is used.
5129 Bugzilla : 5479 (6816)
5131 Description: Racing open + rm can assert client in mdc_set_open_replay_data()
5132 Details : If lookup is in progress on a file that is unlinked we might try
5133 to revalidate the inode and fail in revalidate after lookup is
5134 complete and ll_file_open() enqueues the open again but
5135 it_open_error() was not checking DISP_OPEN_OPEN errors correctly.
5138 Frequency : always, if lconf --abort_recovery used
5140 Description: lconf --abort_recovery fails with 'Operation not supported'
5141 Details : lconf was attempting to abort recovery on the MDT device and not
5144 ------------------------------------------------------------------------------
5146 2005-08-08 Cluster File Systems, Inc. <info@clusterfs.com>
5151 Frequency : rare (only unsupported configurations with a node running as an
5153 Bugzilla : 6514, 5137
5154 Description: Mounting a Lustre file system on a node running as an OST could
5156 Details : OSTs now preallocates memory needed to write out data at
5157 startup, instead of when needed, to avoid having to
5158 allocate memory in possibly low memory situations.
5159 Specifically, if the file system is mounted on on OST,
5160 memory pressure could force it to try to write out data,
5161 which it needed to allocate memory to do. Due to the low
5162 memory, it would be unable to do so and the node would
5163 become unresponsive.
5165 Severity : enhancement
5167 Description: Addition of lconf --service command line option
5168 Details : lconf now accepts a '--service <arg>' option, which is
5169 shorthand for 'lconf --group <arg> --select <arg>=<hostname>'
5171 Severity : enhancement
5173 Description: Failover mode is now the default for OSTs.
5174 Details : By default, OSTs will now run in failover mode. To return to
5175 the old behaviour, add '--failout' to the lmc line for OSTs.
5177 Severity : enhancement
5179 Description: Health checks are now provided for MDS and OSTs
5180 Details : Additional detailed health check information on MSD and OSTs
5181 is now provided through the procfs health_check value.
5184 Frequency : occasional, depends on IO load
5186 Description: Disk fragmentation on the OSTs could eventually cause slowdowns
5187 after numerous create/delete cycles
5188 Details : The ext3 inode allocation policy would not allocate new inodes
5189 very well on the OSTs because there are no new directories
5190 being created. Instead we look for groups with free space if
5191 the parent directories are nearly full.
5196 Description: Network or server problems during mount may cause partially
5197 mounted clients instead of returning an error.
5198 Details : The config llog parsing code may overwrite the error return
5199 code during mount error handling, returning success instead
5205 Description: MDS can fail to allocate large reply buffers
5206 Details : After long uptimes the MDS can fail to allocate large reply
5207 buffers (e.g. zconf client mount config records) due to memory
5208 fragmentation or consumption by the buffer cache. Preallocate
5209 some large reply buffers so that these replies can be sent even
5210 under memory pressure.
5214 Frequency : rare (liblustre)
5215 Description: fsx running with liblustre complained that using truncate() to
5216 extend the file doesn't work. This patch corrects that issue.
5217 Details : This is the liblustre equivalent of the fix for bug 6196. Fixes
5218 ATTR_SIZE and lsm use in llu_setattr_raw.
5222 Frequency : rare, only 2.6 kernels
5223 Description: Unusual file access patterns on the MDS may result in inode
5224 data being lost in very rare circumstances.
5225 Details : Bad interaction between the ea-in-inode patch and the "no-read"
5226 code in the 2.6 kernel caused the inode and/or EA data not to
5227 be read from disk, causing single-file corruption.
5231 Frequency : rare, only 2.6 filesystems using extents
5232 Description: Heavy concurrent write and delete load may cause data corruption.
5233 Details : It was possible under high-load situations to have an extent
5234 metadata block in the block device cache from a just-unlinked
5235 file overwrite a newly-allocated data block. We now unmap any
5236 metadata buffers that alias just-allocated data blocks.
5240 Frequency : filesystems with default stripe_count larger than 77
5241 Description: lconf+mke2fs fail when formatting filesystem with > 77 stripes
5242 Details : lconf specifies an inode size of 4096 bytes when the default
5243 stripe_count is larger than 77. This conflicts with the default
5244 inode density of 1 per 4096 bytes. Allocate smaller inodes in
5245 this case to avoid pinning too much memory for large EAs.
5247 ------------------------------------------------------------------------------
5249 2005-07-07 Cluster File Systems, Inc. <info@clusterfs.com>
5254 Frequency : rare (extremely heavy IO load with hundreds of clients)
5256 Description: Client is evicted, gets IO error writing to file
5257 Details : lock ordering changes for bug 5492 reintroduced bug 3267 and
5258 caused clients to be evicted for AST timeouts. The fixes in
5259 bug 5192 mean we no longer need to have such short AST timeouts
5260 so ldlm_timeout has been increased.
5263 Frequency : occasional during --force or --failover shutdown under load
5264 Bugzilla : 5949, 4834
5265 Description: Server oops/LBUG if stopped with --force or --failover under load
5266 Details : a collection of import/export refcount and cleanup ordering
5267 issues fixed for safer force cleanup
5270 Frequency : only filesystems larger than 120 OSTs
5271 Bugzilla : 5990, 6223
5272 Description: lfs getstripe would oops on a very large filesystem
5273 Details : lov_getconfig used kfree on vmalloc'd memory
5276 Frequency : only filesystems exporting via NFS to Solaris 10 clients
5277 Bugzilla : 6242, 6243
5278 Description: reading from files that had been truncated to a non-zero size
5279 but never opened returned no data
5280 Details : ll_file_read() reads zeros from no-object files to EOF
5285 Description: A bug in MDS/OSS recovery could cause the OSS to fail an assertion
5286 Details : There's little harm in aborting MDS/OSS recovery and letting it
5287 try again, so I removed the LASSERT and return an error instead.
5289 Severity : enhancement
5291 Description: New debugging infrastructure for tracking down data corruption
5292 Details : The I/O checksum code was replaced to: (a) control it at runtime,
5293 (b) cover more of the client-side code path, and (c) try to narrow
5294 down where problems occurred
5298 Bugzilla : 3819, 4364, 4397, 6313
5299 Description: Racing close and eviction MDS could cause assertion in mds_close
5300 Details : It was possible to get multiple mfd references during close and
5301 client eviction, leading to one thread referencing a freed mfd.
5303 Severity: : enhancement
5304 Bugzilla : 3262, 6359
5305 Description: Attempts to reconnect to servers are now more aggressive.
5306 Details : This builds on the enhanced upcall-less recovery that was added
5307 in 1.4.2. When trying to reconnect to servers, clients will
5308 now try each server in the failover group every 10 seconds. By
5309 default, clients would previously try one server every 25 seconds.
5314 Description: After recovery, certain operations trigger a failed
5315 assertion on a client.
5316 Details : Failing over an mds, using lconf -d --failover, while a
5317 client was doing a readdir() call would cause the client to
5318 LBUG after recovery completed and the readdir() was resent.
5320 Severity : enhancement
5322 Description: Default groups are now added by lconf
5323 Details : You can now run lconf --group <servicename> without having to
5324 manually add groups with lmc.
5327 Frequency : occasional
5329 Description: Nodes with an elan id of 0 trigger a failed assertion
5332 Frequency : always when accessing e.g. tty/console device nodes
5334 Description: tty and some other devices nodes cannot be used on lustre
5335 Details : file's private_data field is used by device data and lustre
5336 values in there got lost. New field was added to struct file to
5337 store fs-specific private data.
5340 Frequency : when exporting Lustre via NFS
5342 Description: NFSD failed occasionally when looking up a path component
5343 Details : NFSD is looking up ".." which was broken in ext3 directories
5344 that had grown large enough to become hashed.
5347 Frequency : Clusters with multiple interfaces not on the same subnet
5349 Description: Nodes will repeatedly try to reconnect to an interface which it
5350 cannot reach and report an error to the log.
5351 Details : Extra peer list entries will be created by lconf with some peers
5352 unreachable. lconf now validates the peer before adding it.
5355 Frequency : Only if a default stripe is set on the filesystem root.
5357 Description: Setting a default stripe on the filesystem root prevented the
5358 filesystem from being remounted.
5359 Details : The client was sending extra request flags in the root getattr
5360 request and did not allocate a reply buffer for the dir EA.
5363 Frequency : occasional, higher if lots of files are accessed by one client
5364 Bugzilla : 6159, 6097
5365 Description: Client trips assertion regarding lsm mismatch/magic
5366 Details : While revalidating inodes the VFS looks up inodes with ifind()
5367 and in rare cases can find an inode that is being freed.
5368 The ll_test_inode() code will free the lsm during ifind()
5369 when it finds an existing inode and then the VFS later attaches
5370 this free lsm to a new inode.
5374 Bugzilla : 6422, 7030
5375 Description: MDS deadlock between mkdir and client eviction
5376 Details : Creating a new file via mkdir or mknod (starting a transaction
5377 and getting the ns lock) can deadlock with client eviction
5378 (gets ns lock and trying to finish a synchronous transaction).
5381 Frequency : occasional
5382 Description: While starting a server, the fsfilt_ext3 module could not be
5384 Details : CFS's improved ext3 filesystem is named ldiskfs for 2.6
5385 kernels. Previously, lconf would still use the ext3 name
5386 when trying to load modules. Now, it will correctly use
5387 ext3 on 2.4 and ldiskfs on 2.6.
5389 Severity : enhancement
5390 Description: The default stripe count has been changed to 1
5391 Details : The interpretation of the default stripe count (0, to lfs
5392 or lmc) has been changed to mean striping across a single
5393 OST, rather than all available. For general usage we have
5394 found a stripe count of 1 or 2 works best.
5396 Severity : enhancement
5397 Description: Add support for compiling against Cray portals.
5398 Details : Conditional compiling for some areas that are different
5402 Frequency : occasional
5403 Bugzilla : 6409, 6834
5404 Description: Creating files with an explicit stripe count may lead to
5405 a failed assertion on the MDS
5406 Details : If some OSTs are full or unavailable, creating files may
5407 trigger a failed assertion on the MDS. Now, Lustre will
5408 try to use other servers or return an error to the
5412 Frequency : occasional
5414 Description: Multiple concurrent overlapping read+write on multiple SMP nodes
5415 caused lock timeout during readahead (since 1.4.2).
5416 Details : Processes doing readahead might match a lock that hasn't been
5417 granted yet if there are overlapping and conflicting lock
5418 requests. The readahead process waits on ungranted lock
5419 (original lock is CBPENDING), while OST waits for that process
5420 to cancel CBPENDING read lock and eventually evicts client.
5422 Severity : enhancement
5424 Description: Initial enabling of flock support for clients
5425 Details : Implements fcntl advisory locking and file status functions.
5426 This feature is provided as an optional mount flag (default
5427 off), and is NOT CURRENTLY SUPPORTED. Not all types of record
5428 locking are implemented yet, and those that are are not guaranteed
5429 to be completely correct in production environments.
5430 mount -t lustre -o [flock|noflock] ...
5433 Frequency : occasional
5435 Description: OSTs running 2.4 kernels but with extents enabled might trip an
5436 assertion in the ext3 JBD (journaling) layer.
5437 Details : The b_committed_data struct is protected by the big kernel lock
5438 in 2.4 kernels, serializing journal_commit_transaction() and
5439 ext3_get_block_handle->ext3_new_block->find_next_usable_block()
5440 access to this struct. In 2.6 kernels there is finer grained
5441 locking to improve SMP performance of the JBD layer.
5445 Description: Changes the "SCSI I/O Stats" kernel patch to default to "enabled"
5450 Description: merge and cleanup kernel patches.
5453 -----------------------------------------------------------------------------
5455 2005-05-05 Cluster File Systems, Inc. <info@clusterfs.com>
5457 NOTE: Lustre 1.4.2 uses an incompatible network protocol than previous
5458 versions of Lustre. Please update all servers and clients to
5459 version 1.4.2 or later at the same time. You must also run
5460 "lconf --write-conf {config}.xml" on the MDS while it is stopped
5461 to update the configuration logs.
5463 - fix for HPUX NFS client breakage when NFS exporting Lustre (5781)
5464 - mdc_enqueue does not need max_mds_easize request buffer on send (5707)
5465 - swab llog records of type '0' so we get proper header size/idx (5861)
5466 - send llog cancel req to DLM cancel portal instead of cb portal (5515)
5467 - fix rename of one directory over another leaking an inode (5953)
5468 - avoid SetPageDirty on 2.6 (5981)
5469 - don't re-add just-being-destroyed locks to the waiting list (5653)
5470 - when creating new directories, inherit the parent's custom
5471 striping settings if present parent (3048)
5472 - flush buffers from cache before direct IO in 2.6 obdfilter (4982)
5473 - don't hold i_size_sem in ll_nopage() and ll_ap_refresh_count (6077)
5474 - don't hold client locks on temporary worklist from l_lru (5666)
5475 - handle IO errors in 2.6 obdfilter bio completion routine (6046)
5476 - automatically evict dead clients (5921)
5477 - Update file size properly in create+truncate+fstat case (6196)
5478 - Do not unhash mountpoint dentries, do not allow removal of
5480 - Avoid lock ordering deadlock issue with write/truncate (6203,5654)
5481 - reserve enough journal credits in fsfilt_start_log for setattr (4554)
5482 - ldlm_enqueue freed-export error path would always LBUG (6149,6184)
5483 - don't reference lr_lvb_data until after we hold lr_lvb_sem (6170)
5484 - don't overwrite last_rcvd if there is a *_client_add() error (6086)
5485 - Correctly handle reads of files with no objects (6243)
5486 - lctl recover will also mark a device active if deactivate used (5933)
5488 - by default create 1 inode per 4kB space on MDS, per 16kB on OSTs
5489 - allow --write-conf on an MDS with different nettype than client (5619)
5490 - don't write config llogs to MDS for mounts not from that MDS (5617)
5491 - lconf should create multiple TCP connections from a client (5201)
5492 - init scripts are now turned off by default; run chkconfig --on
5493 lustre and chkconfig --on lustrefs to use them
5494 - upcalls are no longer needed for clients to recover to failover
5496 - add --abort-recovery option to lconf to abort recovery on device
5498 - add support for an arbitrary number of OSTs (3026)
5499 - Quota support protocol changes.
5500 - forward compatibility changes to wire structs (6007)
5501 - rmmod NALs that might be loaded because of /etc/modules.conf (6133)
5502 - support for mountfsoptions and clientoptions to the Lustre LDAP (5873)
5503 - improved "lustre status" script
5504 - initialize blocksize for non-regular files (6062)
5505 - added --disable-server and --disable-client configure options (5782)
5506 - introduce a lookup cache for lconf to avoid repeated DB scans (6204)
5507 - Vanilla 2.4.29 support
5508 - increase maximum number of obd devices to 520 (6242)
5509 - remove the tcp-zero-copy patch from the suse-2.4 series (5902)
5510 - Quadrics Elan drivers are now included for the RHEL 3 2.4.21 and
5511 SLES 9 2.6.5 kernels
5512 - limit stripes per file to 160 (the maximum EA size) (6093)
5514 2005-03-22 Cluster File Systems, Inc. <info@clusterfs.com>
5517 - don't LASSERT in ll_release on NULL lld with NFS export (4655, 5760)
5518 - hold NS lock when calling handle_ast_error->del_waiting_lock (5746)
5519 - fix setattr mtime regression from lovcleanup merge (4829, 5669)
5520 - workaround for 2.6 crash in ll_unhash_aliases (5687, 5210)
5521 - small ext3 extents cleanups and fixes (5733)
5522 - improved mballoc code, several small races and bugs fixed (5733, 5638)
5523 - kernel version 43 - fix remove_suid bugs in both 2.4 and 2.6 (5695)
5524 - avoid needless client->OST connect, fix handle mismatch (5317)
5525 - fix DLM error path that led to out-of-sync client, long delays (5779)
5526 - support common vfs-enforced mount options (nodev,nosuid,noexec) (5637)
5527 - fix several locking issues related to i_size (5492,5624,5654,5672)
5528 - don't move pending lock onto export if it is already evicted (5683)
5529 - fix kernel oops when creating .foo in unlinked directory (5548)
5530 - fix deadlock in obdfilter statistics vs. object create (5811)
5531 - use time_{before,after} to avoid timer jiffies wrap (5882)
5532 - shutdown --force/--failover stability (3607,3651,4797,5203,4834)
5533 - Do not leak request if server was not able to process it (5154)
5534 - If mds_open unable to find parent dir, make that negative lookup(5154)
5535 - don't create new directories with extent-mapping (5909, 5936)
5537 - fix lustre/lustrefs init scripts for SuSE (patch from Scali, 5702)
5538 - don't hold the pinger_sem in ptlrpc_pinger_sending_on_import
5539 - change obd_increase_kms to obd_adjust_kms (up or down) (5654)
5540 - lconf, lmc search both /usr/lib and /usr/lib64 for Python libs (5800)
5541 - support for RHEL4 kernel on i686 (5773)
5542 - provide error messages when incompatible logs are encountered (5898)
5544 2005-02-18 Cluster File Systems, Inc. <info@clusterfs.com>
5545 * version 1.4.0.10 (1.4.1 release candidate 1)
5547 - don't keep a lock reference when lock is not granted (4238)
5548 - unsafe list practices (rarely) led to infinite eviction loop (4908)
5549 - add per-fs limit of Lustre pages in page cache, avoid OOM (4699)
5550 - drop import inflight refcount on signal_completed_replay error (5255)
5551 - unlock page after async write error during send (3677)
5552 - handle missing objects in filter_preprw_read properly (5265)
5553 - no transno return for symlink open, don't save no-trasno open (3440)
5554 - don't try to complete elan receive that already failed (4012)
5555 - free RPC server reply state on error (5406)
5556 - clean up thread from ptlrpc_start_thread() on error (5160)
5557 - readahead could read extra page into cache that wasn't ejected (5388)
5558 - prevent races in class_attach/setup/cleanup/detach (5260)
5559 - don't dereference de->d_inode after l_dput of de (5458)
5560 - use "int" for stripe value returned from lock_to_stripe (5544)
5561 - mballoc allocation and error-checking fixes in 2.6 (5504)
5562 - block device patches to fix I/O request sizes in 2.6 (5482)
5563 - look up hostnames for IB nals (5602)
5564 - 2.6 changed lock ordering of 2 semaphores, caused deadlock (5654)
5565 - don't start multiple acceptors for the same port (5277)
5566 - fix incorrect LASSERT in mds_getattr_name (5635)
5567 - export a proc file for general "ping" checking (5628)
5568 - fix "lfs check" to not block when the MDS is down (5628)
5570 - service request history (4965)
5571 - put {ll,lov,osc}_async_page structs in a single slab (4699)
5572 - create an "evict_client" /proc entry on OSTs, like the MDS has
5573 - fix mount usage message, return errors per mount(8) (5168)
5574 - change grep [] to grep "[]" in tests so they work in more UMLs
5575 - fix ppc64/x86_64 spec to use %{_libdir} instead of /usr/lib (5389)
5576 - remove ancient LOV_MAGIC_V0 EA support (5047)
5577 - add "disk I/Os in flight" and "I/O req time" stats in obdfilter
5578 - align r/w RPCs to PTLRPC_MAX_BRW_SIZE boundary for performance (3451)
5579 - allow readahead allocations to fail when low on memory (5383)
5580 - mmap locking landed again, after considerable improvement (2828)
5581 - add get_hostaddr() to lustreDB.py for LDAP support (5459)
5583 2004-11-23 Cluster File Systems, Inc. <info@clusterfs.com>
5586 - send OST transaction number in read/write reply to free req (4966)
5587 - don't ASSERT in ptl_send_rpc() if we run out of memory (5119)
5588 - lock /proc/sys/portals/routes internal state, avoiding oops (4827)
5589 - the watchdog thread now runs as interruptible (5246)
5590 - flock/lockf fixes (but it's still disabled, pending 5135)
5591 - don't use EXT3 constants in llite code (5094)
5592 - memory shortage at startup could cause assertion (5176)
5594 - reorganization of lov code
5595 - single portals codebase
5597 - add extents/mballoc support (5025)
5598 - direct I/O reads in the obdfilter (4048)
5599 - kernel patches from LNXI for 2.6 (bluesmoke, perfctr, mtd, kexec)
5601 tbd Cluster File Systems, Inc. <info@clusterfs.com>
5604 - send OST transaction number in read/write reply to free req (4966)
5605 - don't ASSERT in ptl_send_rpc() if we run out of memory (5119)
5606 - lock /proc/sys/portals/routes internal state, avoiding oops (4827)
5607 - the watchdog thread now runs as interruptible (5246)
5608 - handle missing objects in filter_preprw_read properly (5265)
5609 - unsafe list practices (rarely) led to infinite eviction loop (4908)
5610 - drop import inflight refcount on signal_completed_replay error (5255)
5611 - unlock page after async write error during send (3677)
5612 - return original error code on reconstructed replies (3761)
5613 - no transno return for symlink open, don't save no-trasno open (3440)
5615 - add pid to ldlm debugging output (4922)
5616 - bump the watchdog timeouts -- we can't handle 30sec yet
5617 - extra debugging for orphan dentry/inode bug (5259)
5619 2004-11-16 Cluster File Systems, Inc. <info@clusterfs.com>
5622 - fix TCP_NODELAY bug, which caused extreme perf regression (5134)
5623 - allocate qswnal tx descriptors singly to avoid fragmentation (4504)
5624 - don't LBUG on obdo_alloc() failure, use OBD_SLAB_ALLOC() (4800)
5625 - fix NULL dereference in /proc/sys/portals/routes (4827)
5626 - allow failed mdc_close() operations to be interrupted (4561)
5627 - stop precreate on OST before MDS would time out on it (4778)
5628 - don't send partial-page writes before EOF from client (4410)
5629 - discard client grant for sub-page writes on large-page clients (4520)
5630 - don't free dentries not owned by NFS code, check generation (4806)
5631 - fix lsm leak if mds_create_objects() fails (4801)
5632 - limit debug_daemon file size, always print CERROR messages (4789)
5633 - use transno after validating reply (3892)
5634 - process timed out requests if import state changes (3754)
5635 - update mtime on OST during writes, return in glimpse (4829)
5636 - add mkfsoptions to LDAP (4679)
5637 - use ->max_readahead method instead of zapping global ra (5039)
5638 - don't interrupt __l_wait_event() during strace
5640 - add software watchdogs to catch hung threads quickly (4941)
5641 - make lustrefs init script start after nfs is mounted
5642 - fix CWARN/ERROR duplication (4930)
5643 - return async write errors to application if possible (2248)
5644 - add /proc/sys/portal/memused (bytes allocated by PORTALS_ALLOC)
5645 - print NAL number in %x format (4645)
5646 - update barely-supported suse-2.4.21-171 series (4842)
5647 - support for sles 9 %post scripts
5648 - support for building 2.6 kernel-source packages
5649 - support for sles km_* packages
5651 2004-10-07 Cluster File Systems, Inc. <info@clusterfs.com>
5654 - ignore -ENOENT errors in osc_destroy (3639)
5655 - notify osc create thread that OSC is being cleaned up (4600)
5656 - add nettype argument for llmount in #5d in conf-sanity.sh (3936)
5657 - reconstruct ost_handle() like mds_handle() (4657)
5658 - create a new thread to do import eviction to avoid deadlock (3969)
5659 - let lconf resolve symlinked-to devices (4629)
5660 - don't unlink "objects" from directory with default EA (4554)
5661 - hold socknal file ref over connect in case target is down (4394)
5662 - allow more than 32000 subdirectories in a single directory (3244)
5663 - fix blocks count for O_DIRECT writes (3751)
5664 - OST returns ENOSPC from object create when no space left (4539)
5665 - don't send truncate RPC if file size isn't changing (4410)
5666 - limit OSC precreate to 1/2 of value OST considers bogus (4778)
5667 - bind to privileged port in socknal and tcpnal (3689)
5669 - rate limit CERROR/CWARN console message to avoid overload (4519)
5670 - GETFILEINFO dir ioctl returns LOV EA + MDS stat in 1 call (3327)
5671 - basic mmap support (3918)
5672 - kernel patch series update from b1_4 (4711)
5674 2004-09-16 Cluster File Systems, Inc. <info@clusterfs.com>
5677 - avoid crash during MDS cleanup with OST shut down (2775)
5678 - fix loi_list_lock/oig_lock inversion on interrupted IO (4136)
5679 - don't use bad inodes on the MDS (3744)
5680 - dynamic object preallocation to improve recovery speed (4236)
5681 - don't hold spinlock over lock dumping or change debug flags (4401)
5682 - don't zero obd_dev when it is force cleaned (3651)
5683 - print grants to console if they go negative (4431)
5684 - "lctl deactivate" will stop automatic recovery attempts (3406)
5685 - look for existing locks in ldlm_handle_enqueue() (3764)
5686 - don't resolve lock handle twice in recovery avoiding race (4401)
5687 - revalidate should check working dir is a directory (4134)
5689 - don't always mark "slow" obdfilter messages as errors (4418)
5691 2004-08-24 Cluster File Systems, Inc. <info@clusterfs.com>
5694 - don't close LustreDB during write_conf until it is done (3860)
5695 - fix typo in lconf for_each_profile (3821)
5696 - allow dumping logs from multiple threads at one time (3820)
5697 - don't allow multiple threads in OSC recovery (3812)
5698 - fix debug_size parameters (3864)
5699 - fix mds_postrecov to initialize import for llog ctxt (3121)
5700 - replace config semaphore with spinlock (3306)
5701 - be sure to send a reply for a CANCEL rpc with bad export (3863)
5702 - don't allow enqueue to complete on a destroyed export (3822)
5703 - down write_lock before checking llog header bitmap (3825)
5704 - recover from lock replay timeout (3764)
5705 - up llog sem before sending rpc (3652)
5706 - reduce ns lock hold times when setting kms (3267)
5707 - change a dlm LBUG to LASSERTF, to maybe learn something (4228)
5708 - fix NULL deref and obd_dev leak on setup error (3312)
5709 - replace some LBUG about llog ops with error handling (3841)
5710 - don't match INVALID dentries from d_lookup and spin (3784)
5711 - hold dcache_lock while marking dentries INVALID and hashing (4255)
5712 - fix invalid assertion in ptlrpc_set_wait (3880)
5714 - add libwrap support for the TCP acceptor (3996)
5715 - add /proc/sys/portals/routes for non-root route listing (3994)
5716 - allow setting MDS UUID in .xml (2580)
5717 - print the stack of a process that LBUGs (4228)
5719 2004-07-14 Cluster File Systems, Inc. <info@clusterfs.com>
5722 - don't cleanup request in ll_file_open() on failed MDS open (3430)
5723 - make sure to unset replay flag from failed open requests (3440)
5724 - if default stripe count is 0, use OST count for inode size (3636)
5725 - update parent mtime/ctime on client for create/unlink (2611)
5726 - drop dentry ref in ext3_add_link from open_connect_dentry (3266)
5727 - free recovery state on server during a forced cleanup (3571)
5728 - unregister_reply for resent reqs (3063)
5729 - loop back devices mounting and status check on 2.6 (3563)
5730 - fix resource-creation race that can provoke i_size == 0 (3513)
5731 - don't try to use bad inodes returned from MDS/OST fs lookup (3688)
5732 - more debugging for page-accounting assertion (3746)
5733 - return -ENOENT instead of asserting if ost getattr+unlink race (3558)
5734 - avoid deadlock after precreation failure (3758)
5735 - fix race and lock order deadlock in orphan handling (3450, 3750)
5736 - add validity checks when grabbing inodes from l_ast_data (3599)
5738 - add /proc/.../recovery_status to obdfilter (3428)
5739 - lightweight CDEBUG infrastructure, debug daemon (3668)
5740 - change default OSC RPC parameters to be better on small clusters
5741 - turn off OST read cache for files smaller than 32MB
5742 - install man pages and include them in rpms (3100)
5743 - add new init script for (un)mounting lustre filesystems (2593)
5744 - run chkconfig in %post for init scripts (3701)
5745 - drop scimac NAL (unmaintained)
5747 2004-06-17 Cluster File Systems, Inc. <info@clusterfs.com>
5750 - clean kiobufs before and after use (3485)
5751 - strip trailing '/'s before comparing paths with /proc/mounts (3486)
5752 - remove assertions to work around "in-flight rpcs" recovery bug (3063)
5753 - change init script to fail more clearly if not run as root (1528)
5754 - allow clients to reconnect during replay (1742)
5755 - fix ns_lock/i_sem lock ordering deadlock for kms update (3477)
5756 - don't do DNS lookups on NIDs too small for IP addresses (3442)
5757 - re-awaken ptlrpcd if new requests arrive during check_set (3554)
5758 - fix cond_resched (3554)
5759 - only evict unfinished clients after recovery (3515)
5760 - allow bulk resend, prevent data loss (3570)
5761 - dynamic ptlrpc request buffer allocation (2102)
5762 - don't allow unlinking open directory if it isn't empty (2904)
5763 - set MDS/OST threads to umask 0 to not clobber client modes (3359)
5764 - remove extraneous obd dereference causing LASSERT failure (3334)
5765 - don't use get_cycles() when creating temp. files on the mds (3156)
5766 - hold i_sem when setting i_size in ll_extent_lock() (3564)
5767 - handle EEXIST for set-stripe, set proper directory name (3336)
5769 - servers can dump a log evicting a client - lustre.dump_on_timeout=1
5770 - fix ksocknal_fmb_callback() error messages (2918)
5772 2004-05-27 Cluster File Systems, Inc. <info@clusterfs.com>
5775 - don't copy lvb into (possibly NULL) reply on error (2983)
5776 - don't deref dentry after dput, don't free lvb on error (2922)
5777 - use the kms to determine writeback rpc length (2947)
5778 - increment oti_logcookies when osc is inactive (2948)
5779 - update client's i_blocks count via lvb messages (2543)
5780 - handle intent open/close of special files properly (1557)
5781 - mount MDS with errors=remount-ro, like obdfilter (2009)
5782 - initialize lock handle to avoid ASSERT on error cleanup (3057)
5783 - don't use cancelling-locks' kms values (2947)
5784 - use highest lock extent for kms, not last one (2925)
5785 - don't dereference ERR_PTR() dentry in error handling path (3107)
5786 - fix thread race in portals_debug_dumplog() (3122)
5787 - create lprocfs device entries at setup instead of at attach (1519)
5788 - common AST error handler, don't evict client on completion race (3145)
5789 - zero nameidata in detach_mnt in 2.6 (3118)
5790 - verify d_inode after revalidate_special is valid in 2.6 (3116)
5791 - use lustre_put_super() to handle zconf unmounts in 2.6 (3064)
5792 - initialize RPC timeout timer earlier for 2.6 (3219)
5793 - don't dereference NULL reply buffer if mdc_close was never sent (2410)
5794 - print nal/nid for unknown nid (3258)
5795 - additional checks for oscc recovery before doing precreate (3284)
5796 - fix ll_extent_lock() error return code for 64-bit systems (3043)
5797 - don't crash in mdc_close for bad permissions on open (3285)
5798 - zero i_rdev for non-device files (3147)
5799 - clear page->private before handing to FS, better assertion (3119)
5800 - tune the read pipeline (3236)
5801 - fix incorrect decref of invalidated dentry (2350)
5802 - provide read-ahead stats and refine rpc in flight stats (3328)
5803 - don't hold journal transaction open across create RPC (3313)
5804 - update atime on MDS at close time (3265)
5805 - close LDAP connection when recovering to avoid server load (3315)
5806 - update iopen-2.6 patch with fixes from 2399,2517,2904 (3301)
5807 - don't leak open file on MDS after open resend (3325)
5808 - serialize filter_precreate and filter_destroy_precreated (3329)
5809 - loop device shouldn't call sync_dev() for nul device (3092)
5810 - clear page cache after eviction (2766)
5811 - resynchronize MDS->OST in background (2824)
5812 - refuse to mount the same filesystem twice on same mountpoint (3394)
5813 - allow llmount to create routes for mounting behind routers (3320)
5814 - push lock cancellation to blocking thread for glimpse ASTs (3409)
5815 - don't call osc_set_data_with_check() for TEST_LOCK matches (3159)
5816 - fix rare problem with rename on htree directories (3417)
5818 - allow default OST striping configuration per directory (1414)
5819 - fix compilation for qswnal for 2.6 kernels (3125)
5820 - increase maximum number of MDS request buffers for large systems
5821 - change liblustreapi to be useful for external progs like lfsck (3098)
5822 - increase local configuration timeout for slow disks (3353)
5823 - allow configuring ldlm AST timeout - lustre.ldlm_timeout=<seconds>
5825 2004-03-22 Cluster File Systems, Inc. <info@clusterfs.com>
5828 - fixes for glimpse AST timeouts / incorrectly 0-sized files (2818)
5829 - don't overwrite extent policy data in reply if lock was blocked (2901)
5830 - drop filter export grants atomically with removal from device (2663)
5831 - del obd_self_export from work_list in class_disconnect_exports (2908)
5832 - don't LBUG if MDS recovery times out during orphan cleanup (2530)
5833 - swab reply message in mdc_close, other PPC fixes (2464)
5834 - fix destroying of named logs (2325)
5835 - overwrite old logs when running lconf --write_conf (2264)
5836 - bump LLOG_CHUNKSIZE to 8k to allow for larger clusters (2306)
5837 - fix race in target_handle_connect (2898)
5838 - mds_reint_create() should take same inode create lock (2926)
5839 - correct journal credits calculated for CANCEL_UNLINK_LOG (2931)
5840 - don't close files for self_export to avoid uninitialized obd (2936)
5841 - allow MDS with the same name as client node (2939)
5842 - hold dentry reference for closed log files for unlink (2325)
5843 - reserve space for all logs during transactions (2059)
5844 - don't evict page beyond end of stripe extent (2925)
5845 - don't oops on a deleted current working directory (2399)
5846 - handle hard links to targets without a parent properly (2517)
5847 - don't dereference NULL lock when racing during eviction (2867)
5848 - don't grow lock extents when lots of conflicting locks (2919)
5850 2004-03-04 Cluster File Systems, Inc. <info@clusterfs.com>
5853 - account for cache space usage on clients to avoid data loss (974)
5854 - lfsck support in lustre kernel code (2349)
5855 - reduce journal credits needed for BRW writes (2370)
5856 - orphan handling to avoid losing space on client/server crashes
5857 - ptlrpcd can be blocked, stopping ALL progress (2477)
5858 - use lock value blocks to assist in proper KMS, faster stat (1021)
5859 - takes i_sem instead of DLM locks internally on obdfilter (2720)
5860 - recovery for initial connections (2355)
5861 - fixes for mds_cleanup_orphans (1934)
5862 - abort_recovery crashes MDS in b_eq (mds_unlink_orphan) (2584)
5863 - block all file creations until orphan recovery completes (1901)
5864 - client remove rq_connection from request struct (2423)
5865 - conf-sanity test_5, proper cleanup in umount log not availale (2640)
5866 - recovery timer race (2670)
5867 - mdc_close recovey bug (2532)
5868 - ptlrpc cleanup bug (2710)
5869 - mds timeout on local locks (2588)
5870 - namespace lock held during RPCs (2431)
5871 - handle interrupted sync write properly (2503)
5872 - don't try to handle a message that hasn't been replied to (2699)
5873 - client assert failure during cleanup after abort recovery (2701)
5874 - leak mdc device after failed mount (2712)
5875 - ptlrpc_check_set allows timedout requests to complete (2714)
5876 - wait for inflight reqs when ptlrpcd finishes (2710)
5877 - make sure unregistered services are removed from the srv_list
5878 - reset bulk XID's when resending them (caught by 1138 test)
5879 - unregister_bulk after timeout
5880 - fix lconf error (2694)
5881 - handle write after unfinished setstripe, stripe-only getstripe (2388)
5882 - readahead locks pages, leaves pending causing memory pressure (2673)
5883 - increase OST request buffers to 4096 on large machines (2729)
5884 - fix up permission of existing directories in simple_mkdir (2661)
5885 - init deleted item, add assertions ptlrpc_abort_inflight() (2725)
5886 - don't assign transno to errored transactions (2742)
5887 - don't delete objects on OST if given a bogus objid from MDS (2751)
5888 - handle large client PAGE_SIZE readdir on small PAGE_SIZE MDS (2777)
5889 - if rq_no_resend, then timeout request after recovery (2432)
5890 - fix MDS llog_logid record size, 64-bit array alignment (2733)
5891 - don't call usermode_helper from ptlrpcd, DEFAULT upcall (2773)
5892 - put magic in mount.lustre data, check for bad/NULL mount data (2529)
5893 - MDS recovery shouldn't delete objects that it has given out (2730)
5894 - if enqueue arrives after completion, don't clobber LVB (2819)
5895 - don't unlock pages twice when trigger_group_io returns error (2814)
5896 - don't deref NULL rq_repmsg if ldlm_handle_enqueue failed (2822)
5897 - don't write pages to disk if there was an error (1450)
5898 - don't ping imports that have recovery disabled (2676)
5899 - take buffered bytes into account when balancing socknal conn (2817)
5900 - hold a DLM lock over readdir always, use truncate_inode_pages (2706)
5901 - reconnect unlink llog connection after MDS reconnects to OST (2816)
5902 - remove little-endian swabbing of llog records (1987)
5903 - set/limit i_blksize to LL_MAX_BLKSIZE on client (2884)
5904 - retry reposting request buffers if they fail (1191)
5905 - grow extent at grant time to avoid granting a revoked lock (2809)
5906 - lock revoke doesn't evict page if covered by a second lock (2765)
5907 - disable VM readahead to avoid reading outside lock extents (2805)
5909 - return LL_SUPER_MAGIC from statfs for the filesystem type (1972)
5910 - updated kernel patches for hp-2.4.20 kernel (2681)
5912 2004-02-07 Cluster File Systems, Inc. <info@clusterfs.com>
5915 - fix truncated write corruption (2366)
5916 - fix for failed assertion in iopen_connect_dentry (1792,2517)
5918 - don't flag the ptlrpcd thread with PF_MEMALLOC (2636)
5919 - ensure len(uuid) < 37 in lmc (1171)
5920 - fix ia64 OOPS in llog_test (2255)
5921 - zero end of page at obdfilter for partial page writes (2648)
5922 - don't leave stale dentries around after renames (bug 2428)
5923 - fix timeouts when evicting a client with a single lock held (2642)
5924 - set deadline for the initial HELLO message to drain (2634)
5925 - print out dotted-quad IP addresses in the socknal (2302)
5927 - additional debugging for MDS client eviction problem (2443)
5928 - fix mkfsoptions support for osts (2603, 2604)
5930 2004-01-27 Cluster File Systems, Inc. <info@clusterfs.com>
5933 - add series for the vanilla 2.6.0 kernel
5934 - add series for the vanilla 2.4.24 kernel
5935 - add series for a cray x86/64 UL kernel drop
5936 - fix xattr patches for the vanilla 2.4.19 series
5938 - generate true UUIDs in lmc (1171)
5939 - have portals stack dumping break in UML (2466)
5940 - avoid bad dchild deref; avoid inum lock w/o creation (2362)
5941 - allocate with _NOFS in ldlm to avoid deadlock (1933)
5942 - wake callback waiting threads on client eviction (2460)
5943 - Add --ptldebug and --subsystem to lmc (1719)
5944 - update assertion to allow safe interrupt allocation
5945 - set rq_no_resend for cancel requests (2432)
5946 - recalculate ptlrpcd timeout after resend (2494)
5947 - call vfs_rmdir when removing pending directories (2368)
5948 - fix renaming a file to itself (2429)
5949 - lmc creates a default one-stripe lov (2454)
5950 - expand procfs space to handle large clusters (2326)
5951 - increase UML stack to avoid overflow
5952 - update lconf's list of debug and subsystem masks
5953 - fix lfs find --obd (2510)
5954 - /proc tunable for disabling filter read caching (2591)
5955 - stop rpm packages from altering slapd.conf (2301)
5956 - disable nagle in the socknal under 0conf (2578)
5957 - choose mds inode size based on stripe count (2572)
5958 - fix kernel-source rpm problems (2516)
5960 - add --disable-doc to avoid pdf generation (2421)
5961 - update documentation, tests, type-os, comments
5962 - avoid format warnings on ia64
5963 - remove the TOE NAL
5964 - tiny code cleanups by removing unused fields
5966 2004-01-07 Cluster File Systems, Inc. <info@clusterfs.com>
5969 - fix obvious semaphore misuse in as-yet-unused setattr path (2348)
5970 - remove the most blatant lies from BUILDING file (2371)
5971 - change default debug level to reasonable production setting
5972 - reduce client side cache size to reduce cache flush time
5973 - reduce max RPCs in flight to avoid unnecessary file fragmentation
5974 - make TCP zerocopy and pinger support enabled by default (2476)
5975 - sync writes completed after process exits caused crashes (2319)
5976 - maintain correct mount count on the MDS (2356)
5977 - backout 1557, because 2316 wasn't really fixed
5978 - better file I/O statistics gathering in /proc
5979 - don't take unnecessary, deadlock-inducing bug in readpage (2383)
5980 - another kernel patch to fix zero-copy TCP function export
5981 - don't take duplicate lock when processing re-sent getattr (2420)
5982 - lctl uses obd_self_export instead of creating new conn (2353)
5983 - MDS/OST recovery case which requires object creation asserted (2425)
5984 - move lfs from /usr/sbin to /usr/bin in packages
5985 - fix race between mds_client_add and mds_client_free (2417)
5986 - use kmalloc instead of slabs in portals (2430)
5987 - don't create duplicate records when a failover MDS is present (2442)
5988 - remove unnecessary mount age check (2332)
5989 - don't remove directory inodes from locks prematurely (2451)
5990 - don't break if MDS service name is the same as hostname (2103)
5991 - fix races in client write RPC generation when cache full (2482)
5993 2003-12-13 Cluster File Systems, Inc. <info@clusterfs.com>
5996 - remove now-unused request->rq_obd (278)
5997 - if an allocation fails, print out how much memory we've used (1933)
5998 - use PORTAL_SLAB_ALLOC for structures, to get GFP_MEMALLOC (1933)
5999 - add the "configurable stack size" patch to most series files (1256)
6000 - ability to write large log records, for 100+ OST configs (2306)
6001 - fix NULL deref when filter_prep fails (2314)
6002 - fix operator precedence error in filter_sync
6003 - dynamic allocation of socknal TX descriptors (2315)
6004 - fix a missed case in the GFP_MEMALLOC patch, can cause deadlock (2310)
6005 - fix gcc 2.96 compilation problem in xattr kernel patch (2294)
6006 - ensure that CWARN messages in Portals always get to the syslog
6007 - __init/__exit are not for prototype decls (ldlm_init/exit)
6008 - x86-64 compile warning fixes
6009 - fix gateway LMC keyword conflict (2318)
6010 - fix MDS lock inversions in getattr/reint paths (1844)
6011 - fix a rare lock re-ordering bug, which caused deadlock (2322)
6012 - fix i_sem/journal inversion in fsfilt_ext3_write_record (2306)
6013 - DLM race condition prevented some lock evictions (2328)
6014 - ENOMEM detection and retry on socknal sends (2230)
6015 - use GFP_NOFS throughout Lustre, to combat ENOMEM (2230)
6016 - move osc_rpcd into ptlrpc, for use in MDC and others (2329)
6017 - protect MDS inode fsdata with stronger locking; fixes assertion (2313)
6018 - better error messages when a client is rejected during recovery (1505)
6019 - avoid cancelling locks which were never granted, after failure (2330)
6020 - fix i_sem/journal inversion in mds_client_add (2333)
6021 - fix truncate/getattr lock cycle deadlock (2334)
6022 - use rpcd to send close; allows resend after timeout, avoid leak (1897)
6023 - fix two rare exit paths which could leak an l_lock() ref (2321)
6024 - fencepost error in MDS/OST orphan recovery (2226)
6025 - make log record alignment 8 bytes (1988)
6026 - lstripe now fails when requested offset > ost_count (2237)
6027 - ensure that all kernel series have a complete list.h (1607)
6028 - fix crashes in special-file operations (2316)
6029 - lctl create/brw OID mismatch, caused by obsolete filter loop (2339)
6031 - allow configurable automake binary, for testing new versions
6032 - small update to the lfs documentation
6034 2003-12-03 Cluster File Systems, Inc. <info@clusterfs.com>
6036 * fix negative export reference count in fsfilt_sync (2312)
6038 2003-12-01 Cluster File Systems, Inc. <info@clusterfs.com>
6039 * release candidate 0.9.1
6041 - orphans are moved into the PENDING directory for possible recovery
6042 - replayed opens now open by fid for orphan/rename safety (1042)
6043 - last close of an orphan inode generates a transno (683)
6044 - chdir() and mount() now pin the directory entry (1020)
6045 - avoid CERROR in normal ll_setattr_raw() error case (1500)
6046 - discard very old requests without processing them (1502)
6047 - remove some common, well-understood CERRORs (1505)
6048 - require O_DIRECT I/O to be page-sized to workaround IA64 crash (1609)
6049 - clear "grant" flags in OST replies until OST grant code lands (1644)
6050 - fix read performance by not clobbering i_blksize on client (1598)
6051 - fix __ldlm_handle2lock oops by not dereferencing lock after PUT (1625)
6052 - make LRU size a /proc tunable, clears locks when reduced (707)
6053 - fix some lprocfs rot that prevented ptlbd from loading (1732)
6054 - server locks take references on exports now (1558)
6055 - build fixes for 2.4.20-rh trees (1663)
6056 - return an error from lov_create if all OSCs are inactive (1751)
6057 - fix import levels when a reconnect happens without a timeout (1597)
6058 - exit early from mds_open if we get a lookup error (1749)
6059 - partial page read at EOF wouldn't wait for disk before sending (1642)
6060 - avoid NULL deref in obdfilter when reading page past EOF (1592)
6061 - avoid LASSERT in ll_intent_lock if server failed very early (1090)
6062 - fix LBUG in ll_it_open_error with rc = -2 (1861)
6063 - write/truncate lock inversion (1639)
6064 - Don't auto-load obdclass, portals modules during cleanup (1495)
6065 - fix timestamps from jumping to "now" (1763)
6066 - extra journal assertions (1648)
6067 - add an extra multiunlink test (1771)
6068 - fix read_record/write_record API (1776)
6069 - fix leak of offset_extent, possible incorrect i_size later (1772)
6070 - fix lasserts in mis-matched transnos during open-unlink testing (1541)
6071 - Debugging for the kqswnal_get_idle_tx problems (1820)
6072 - Allow recovery to be attempted multiple times (1536)
6073 - Write out MDS last_rcvd file after it is first created (1600)
6074 - Fix tx_descriptor leak in failed transmit situations (1827)
6075 - ext3 journaling fixes for assertion failure after IO error (1871)
6076 - class_export_put() on freed export after completion AST error (1896)
6077 - Fix revalidate looping in VFS (1322)
6078 - Don't access a freed export during MDS_REINT timeout (1521)
6079 - Add open-unlink recovery support on the MDS (1673,1764)
6080 - Return an error if no MDS data was read from last_rcvd (1946)
6081 - Fix for lookup "." or ".." crash on error (1932,1931,1935)
6082 - Don't setup a disk device that doesn't match exported UUID (317)
6083 - Reduce bulk RPC timeout to avoid cascading client/OST failures (1845)
6084 - avoid committing NULL handle in force close
6085 - local.sh is now a one-stripe LOV configuration
6086 - POSIX utime.4 -EPERM on FIFO not owned by user (56)
6087 - fix ext3 htree duplicate directory entry corruption (1516)
6088 - POSIX creat.13, fstat.1, open.18, stat.3 new file atime/mtime (2020)
6089 - update to new LOV EA format (2097)
6090 - interoperability for different PAGE_SIZE/wordsize (686,1821,1343,2042)
6092 2003-06-15 Phil Schwan <phil@clusterfs.com>
6095 - imports and exports cleanup too early, need refcounts (349, 879, 1045)
6096 - per-import/export recovery handling (958, 931, 959)
6097 - multiple last-rcvd slots, for serving multiple FSes (949)
6098 - connections are again shared between multiple imp/exports (963, 964)
6099 - "umount -f" would hang if any requests needed to be sent (393, 978)
6100 - avoid pinning large req buffer by copying for queued messages (989)
6101 - add "uuid" to "lctl device" command to help upcalls (991)
6102 - "open" RPCs with transnos would confuse recovery counters (1037)
6103 - do proper endian conversion of all wire messages (288, 340, 891)
6104 - remove OST bulk get LBUGs, fix ost_brw_write cleanup (1126)
6105 - call waiting locks callback from LDLM recovery thread (1127, 1151)
6106 - fix ptlrpc_connection leak in target_handle_connect (1174)
6107 - fix import refcounting bug in OST and MDS cleanup (1134)
6108 - if an invalid-at-open-time OSC returned before close(), LBUG (1150)
6109 - fix very unlikely obd_types race condition (501)
6110 - remove osc_open hack for echo_client (1187)
6111 - we leaked exports/dlmimps for forcibly disconnected clients (1143)
6112 - a failure in read_inode2 leads to deadlock (1139)
6113 - cancel ack-locks as soon as transaction is committed (1072)
6114 - fix major leaks and crashes in the bulk I/O path (937, 1057)
6115 - make sure to commitrw after any preprw to avoid deadlock (1162)
6116 - failing to execute a file in a lustre FS would lock inode (1203)
6117 - small DEBUG_REQ fix to avoid dereferencing a NULL (1227)
6118 - don't ASSERT while cleaning up an incompletely-setup obd (1248)
6119 - obd_uuid2tgt would walk off the end of the list (1255)
6120 - on IA64 the osc would give portals incorrect bulk size (1258)
6121 - fix debug daemon ioctl interface; allows daemon on ia64 (1274)
6122 - fix lock inversion caused by new llite matching code (1282)
6123 - limit the number of dirty pages on a client to 10MB (1286)
6124 - timed out locks were not being corrected cancelled (1289)
6125 - fix O_DIRECT above 4GB on IA-32 (1292)
6126 * major user-visible changes
6127 - fail out/fail over policy now controlled by the upcall (993)
6129 - add OBD_PING to check server availability and failure (954)
6130 - lustre messages are now sent in sending host order (288, 340, 891)
6131 - add eadatalen to MDS getattr reply (340)
6132 - OST read replies may contain second buffer, with per-page status (593)
6134 2003-03-11 Phil Schwan <phil@clusterfs.com>
6137 - LDLM_DEBUG macro fix, for gcc 3.2 (850)
6138 - failed open()s could cause deadlock; fixed (867, 869)
6139 - stop cancelling OST locks when files are closed (481)
6140 - overlapping XID spaces caused network corruption (851, 853)
6141 - fix unsafe fsfilt counter arithmetic; change to atomic_t
6142 - setattr_raw added, to do single-RPC, server-side setattrs
6143 - lmc/lconf syntax change for OST UUIDs
6144 - fix crashy race condition between ptlrpc_free_req and osc_close
6145 - don't use request in mdc_enqueue if we hit a timeout (889)
6146 - don't set the inode i_size for regular files from the MDS (896)
6147 - handle out of order completion AST (842)
6148 - don't LBUG if a lock request times out after receiving AST (913)
6149 - avoid d_rehash race in ll_find_alias by rehashing inside dcache_lock
6150 - if a bad lock AST arrives, send an error instead of dropping entirely
6151 - return 0 from revalidate2 if ll_intent_lock returns -EINTR (912)
6152 - fix leak in bulk IO when only partially completed (899, 900, 926)
6153 - fix O_DIRECT for ia64 (55)
6154 - (almost) eliminate Lustre-kernel-thread effects on load average (722)
6155 - C-z after timeout could hang a process forever; fixed (977)
6157 - client-side I/O cache (678, 924, 929, 941, 970)
6159 - READPAGE and SETATTRs which don't take server-side locks get
6162 2003-02-11 Phil Schwan <phil@clusterfs.com>
6165 - Fix ldlm_lock_match on the MDS to avoid matching remote locks (592)
6166 - Fix fsfilt_extN_readpage() to read a full page of directory
6167 entries, or fake the remainder if PAGE_SIZE != blocksize (500)
6168 - Avoid extra mdc_getattr() in ll_intent_lock when possible (534, 604)
6169 - Fix imbalanced LOV object allocation and out-of-bound access (469)
6170 - Most intent operations were removed, in favour of a new RPC mode
6171 that does a single RPC to the server and bypasses most of the VFS
6172 - All LDLM resource ID arrays were removed in favour of ldlm_res_id
6173 - Aggressively cancel local locks on DLM servers
6174 - mds_reint_unlink sends EA to the client if it's the last nlink.
6175 client uses that EA to unlink OST objects.
6176 - mds_reint_{rename,unlink,link} were rewritten to take ordered locks
6177 - recursive symlinks were fixed (439)
6178 - fixed NULL deref in DEBUG_REQ
6179 - filter_update_lastobjid no longer calls sync, which annoyed extN
6180 - fixed multi-client small-writes to a single file problem (445)
6181 - fixed mtime updates during file writes (607)
6182 - fixed vector writes on obdfilter causing problems when ENOSPC (670)
6183 - fixed bug in obd_brw_read/write() (under guise of testing 367)
6184 - fixed Linux OST size reporting problem (444, 656)
6185 - OST now updates object mtime with writes or setattr (607, 619)
6186 - client verifies file size before zeroing page past EOF (445)
6187 - OST now writes last allocated objid to disk with allocation (108)
6188 - LOV on echo now works (409)
6190 - mds_reint_unlink sends a new buffer, with the EA included. this
6191 buffer is only valid if body->valid & OBD_MD_FLEASIZE, which is only
6192 set if a regular file was being unlinked, and it was the last link
6193 - use PtlGet from the target for bulk writes (315)
6194 - OST now updates object mtime with writes or setattr (607, 619)
6195 - LDLM now has a grant-time callback to revalidate locked items, if
6197 - Many MDS operations were reorganized to combat race conditions
6199 - Merge b_intel branch (updated lprocfs code) - now at /proc/fs/lustre
6200 - configure check to avoid gcc version 2.96 20000731-2.96-98) (606)
6202 2003-01-06 Andreas Dilger <adilger@clusterfs.com>
6205 - Fully reactivate OST imports after reconnection (512, others)
6206 - Make sure client sees our -ENOTCONN from mds_handle (513 - partial)
6207 - More graceful error handling for truncating on dead OST (515)
6208 - Don't error out unless we're actually accessing dead stripes (474)
6209 - Fix garbage sizes when stripes are missing (410)
6210 - LRU counters were broken, causing constant lock purge (433, 432)
6211 - garbage on read from stripes with failed OSTs (441)
6212 - mark OSCs as active before reconnecting during recovery (438)
6213 - lov_enqueue and lov_cancel need to handle inactive OSTs (403)
6214 - lfind did not preserve OST order in output (443)
6215 - symlinks cause hung clients, incorrect data (439)
6216 - stop dereferencing request after dropping refcount (457)
6217 - don't LASSERT(spin_is_locked) on non-SMP (455)
6218 - fixes for many rename() bugs
6219 - fstat didn't correctly synchronize attributes (399)
6220 - server must handle lock cancellation during blocking AST prep (487)
6221 - bulk descriptors were free()d too soon (511)
6222 - fix paths in lconf, which would load incorrect modules (451, 507)
6223 - fix confusing lconf 'host not found' error message (386)
6224 - fix lock order deadlock on OST (O/R i_sem before journal ops, 478)
6225 - fix race condition in mdc_blocking_ast() for inode access (526)
6226 - fix lov_unpackmd() unpacking wrong number of stripes (537)
6227 - fix lov_set_osc_active() marking wrong OSC inactive (440)
6228 - fix bad lstripe lov_unpackmd() assertion (fix layering too) (527)
6229 - fix multiple writes of stripe MD to MDS (358, maybe 519)
6230 - fix lstripe in several ways (kernel side) (527)
6231 - fix request leak in ldlm_cli_enqueue (262)
6232 - incorrect OSC was marked inactive after OST failure
6233 - call mds_fs_cleanup before unmounting filesystem (524)
6234 - fix races between taking ns_lock and ldlm_lock_change_resource
6235 - fix races updating LOV export open file list
6236 - fix lov_enqueue error path, avoid decref-ing bad lock handle
6237 - fix recovery NULL deref in ldlm_cli_cancel_unused
6238 - fix some DLM races by using new hash table for lock handles (419)
6239 - permit the client to specify desired inodes, at replay
6240 - duplicate requests when we queue them for replay reintegration
6241 - fix last_rcvd offset calculation
6242 - sync after each recovered transaction, so we always make progress
6243 - never, not always, ERESTART requests without transnos
6244 - store the lov_desc in the MDS, so we don't depend on getlovinfo to
6246 - skip replay if the MDS says that the client is already connected
6247 - don't check for a recovery-enabled export to match lctl's UUID
6248 - don't INC_USE_COUNT for phantom exports
6249 - don't crash when cleaning up phantom exports (567)
6250 - don't double-finish or set replay data for errored mdc_open requests
6251 - abort requests when they time out, so we don't get old replies
6252 - send/receive replies for AST messages again
6253 - if the client says that it doesn't have the lock, cancel it on the
6255 - if we timeout during I/O, don't try to cancel an in-use lock; instead
6256 mark it as destroyed, it will all work out when decref is called
6257 - fix module use counts (22, 581)
6259 - ASTs now expect a reply (server cancels lock on error reply)
6261 2002-12-02 Andreas Dilger <adilger@clusterfs.com>
6264 - fix many simultaneous client startup (392)
6265 - fix dentry->d_it clobbering
6266 - credentials weren't being shipped for readdir/getattr operations
6267 - remove invalid assertions triggered during some concurrent MD
6269 - proper Lustre versions added (336, 389)
6270 - fix memory leak for create error case (398)
6271 - fix LOV locking bug that would get cli/srv out of sync
6272 - fix echo client over LOV (409)
6273 - fix dbench 2, extN refcount problem (170, 258, 356, 418)
6274 - fix double-O_EXCL intent crash (424)
6275 - avoid sending multiple lock CANCELs (352)
6277 - MDS can do multi-client recovery (modulo bugs in new code)
6279 - many updates, edits, cleanups
6281 2002-11-18 Phil Schwan <phil@clusterfs.com>
6284 - fix null d_it dereference (346)
6285 - fix full OST/dbench hang (333)
6286 - fix permission problem with file removal (286)
6287 - fix removal of OSCs from LOV when they fail
6288 - fix NULL deref during bulk timeout (214)
6289 - fix problems related to multiple filesystems on one MDS (241)
6290 - fixed serious subtle metadata locking bugs
6291 - free locks on clients when inodes are removed due to memory
6293 - fix inode pointer in lock data (285)
6294 - partial support for multiple MDS on a single host (241)
6295 - data locks weren't cancelled at clear_inode time (290, 311)
6296 - intent locks could lead to unbounded lock growth (205)
6297 - added a maximum lock count, an LRU list, and a flusher
6298 - fix multiple rename (365)
6299 - properly abstracted the echo client
6300 - OSC locked 1 byte too many; fixed
6301 - rewrote brw callback code:
6302 - fixed recovery bugs related to LOVs (306)
6303 - fixed too-many-pages-in-one-write crash (191)
6304 - fixed (again) crash in sync_io_timeout (214)
6305 - probably fixed callback-related race (385)
6307 - Add capability to MDS protocol
6308 - LDLM cancellations and callbacks on different portals
6310 2002-10-28 Andreas Dilger <adilger@clusterfs.com>
6313 - limit client IOV size to PTL_MD_MAX_IOV (611336, 191)
6314 - defer open object destruction to close time (601981, 138)
6315 - open/close OST file handle in obdo (OBD_MD_FLHANDLE) (601981, 138)
6316 - move LDLM_ENQUEUE/CONVERT back to MDS portal (625069)
6317 - abstract ll_lookup2, fix ll_revalidate2 to use abstraction (256)
6318 - don't call obd_setattr in ll_file_release for destroyed objects
6319 * protocol change to lustre_msg: move |version| and add |flags|
6320 * protocol change to osc_punch: "start" in "o_size", "end" in "o_blocks"
6321 * lock replay: for LDLM_FL_REPLAY trust client to do right thing
6322 * added replay of create, unlink, link and rename operations during
6323 MDS failover; recovery should be much more robust now
6324 * remove failed OSCs from LOVs (only lov_create uses this so far)
6325 * the lustre-HOWTO was brought (more) up to date (582544)
6327 2002-10-23 Phil Schwan <phil@clusterfs.com>
6330 - in-use dentries weren't being reused properly (617851)
6331 - prevent multiple LDLM setup (599178)
6332 - fix LOV size calculations for truncate (617853)
6333 - fix client handling of MDS intent errors (POSIX)
6334 - fix permission bug in lovstripe.c test (624321)
6335 - fix MDS thread deadlock - move LDLM handler to DLM portal (625069)
6336 - truncate past end of file could corrupt data
6337 - proper cleanup after timeouts, crashes, etc (592524, 550815)
6338 - a race in recovery could return ETIMEDOUT to apps (623947)
6339 - building outside the source directory was fixed
6340 * the lustre-HOWTO was brought (more) up to date (582544)
6341 * major progress was made on recovery functionality
6343 2002-10-10 Phil Schwan <phil@clusterfs.com>
6346 - recovery deadlock fix
6347 - rm -rf causes LBUG fix (617817)
6348 - file open by multiple tasks fix (618962)
6349 - directory permissions bugs (602707 and 620007)
6350 - journal_stop fixed with locking (611313)
6351 - O_APPEND failures resolved (618273, perhaps 614459)
6352 - lconf PATH fix (619770)
6353 - IA64 build fix (621450)
6354 - RPC buffer sizes scale with amount of memory
6356 2002-10-01 Phil Schwan <phil@clusterfs.com>
6359 - locks would be cancelled without throwing away data pages,
6360 resulting in inconsistent data (605627)
6361 - inode attributes were not always being refreshed (605627, 612449)
6362 - lconf now continues to cleanup after lctl reports an error
6363 - MDS now enforces user permissions (602707)
6364 - lprocfs cleanup fixed, but not yet enabled (614157)
6365 - fixed infinite server hang, should a client not respond to an AST
6366 - avoid going into recovery if user calls readlink() with a buffer
6367 that's too small (613941)
6368 - AST RPCs no longer require replies (614867) -- this may be changed
6369 - don't crash server if client sends an IOV that's too big (611336)
6370 - fixed lock conversion deadlock (611892)
6371 - fixed the following of symlinks (614622)
6372 * recovery: the server can remove locks from a client that dies, other
6373 clients can make progress
6374 * more extN patch fixes
6375 * compile-time configurable ptlrpc buffer allocations
6377 - collaborative read cache document
6378 - Lustre Lite Performance CDR document-in-progress
6380 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
6383 - fix typo in patch-2.4.18
6385 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
6388 - clear ptlrpc request each time in handle_incoming_request()
6389 - unlink of files now destroys the object on the OST
6391 2002-09-19 Peter Braam <braam@clusterfs.com>
6393 * add hard link support
6394 * change obdfile creation method
6395 * kernel patch changed
6397 2002-09-19 Peter Braam <braam@clusterfs.com>
6400 - stack overflow bug in extN fixed
6402 2002-09-18 Andreas Dilger <adilger@clusterfs.com>
6404 * documentation updates
6405 - add man pages for config tools
6406 - update tests/README to describe testing with new config tools
6407 - finish metadata API descriptions
6408 * bug fixes and cleanups
6409 - statfs workaround for 16TB limit
6410 - LOV stripe allocation improved, can stripe on subset of OSTs
6411 - LOV file size/IO offset was wrong for files > 4GB in size
6412 - object EA data was being dropped, caused files to be unreadable
6413 - memory overflow with non-LOV OST caused memory corruption
6414 - fixed regression tests to work with new config tools, obdfilter
6415 - fixed bug when directory size became larger than 1 block
6416 - fixed bug (for single client case) when PWD was deleted
6417 - invalidate local directory pages when doing intent-based ops
6418 - avoid LDLM oops when lock callback contained bad data
6420 2002-09-09 Andreas Dilger <adilger@clusterfs.com>
6422 * documentation updates
6423 * bug fixes and cleanups
6424 - configuration tools
6427 - 64-bit compile warnings
6428 - 64-bit internal statfs data
6430 * test_brw on persistent OST devices
6432 * lprocfs (disabled)
6434 2002-09-04 Andreas Dilger <adilger@clusterfs.com>
6436 * documentation updates
6437 * bug fixes and cleanups
6438 * configuration tools
6440 2002-08-30 Peter J. Braam <braam@clusterfs.com>
6443 * many small fixes to 0_5_4
6444 * io/network handling
6445 * thinkos in MDS operations
6447 2002-08-24 Peter J. Braam <braam@clusterfs.com>
6450 * crucial basic fixes to 0.5.3
6451 * IOR, Iozone work over Elan
6454 2002-08-07 Phil Schwan <phil@clusterfs.com>
6455 * version 0_5_3, our first alpha
6456 * we use the new Portals iovs
6457 * documentation updates
6458 * bug fixes and cleanups
6459 * small changes in the DLM wire protocol
6461 2002-07-25 Peter J. Braam <braam@clusterfs.com>
6462 * version 0_5_1 with some initial stability,
6463 * locking on MD and file I/O.
6464 * documentation updates
6465 * several bug fixes since 0.5.0
6466 * small changes in wire protocol
6468 2002-07-18 Phil Schwan <phil@clusterfs.com>
6470 * delivered as Lustre Light Alpha
6471 * fixed a crash after handling invalid MDS requests
6472 * fixed directory pages for architectures with non-4k pages sizes
6474 2002-07-11 Andreas Dilger <adilger@clusterfs.com>
6475 * release version v0_4_4
6476 * Moves TCP acceptor to be on port 2432 (unused Coda port) instead
6478 * Fixes a number of interruption problems with OST operations.
6479 * Update documentation for portals header changes
6480 * Move all wire protocol structs/defines to lustre_idl.h
6481 * Fixes symlink length bug.
6482 * Add tcpdump to repository.
6484 2002-07-05 Andreas Dilger <adilger@clusterfs.com>
6485 * release version v0_4_3
6486 * Fixes statfs for inodes on extN.
6487 * Fixes bug in runtests which would delete /etc/hosts.
6488 * Use 64-bit object IDs wherever possible (not into VFS though)
6489 Remove ost_get_info, which is unused by lustre, and out of date.
6491 2002-07-03 Peter Braam <braam@clusterfs.com>
6492 * release version v0_4_2 Fixes a lookup error (type not passed)
6493 * move forward to head of Portals
6494 * move forward to latest Lustre kernel
6496 2002-06-25 Peter Braam <braam@clusterfs.com>
6497 * release version v0_4_1. Hopefully stable on single node use.