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 2.6.32-504.3.3.el6 (RHEL6.6)
8 3.0.101-0.40 (SLES11 SP3)
9 * Client known to build on unpatched kernels:
10 2.6.32-431.29.2.el6 (RHEL6.5)
11 2.6.32-504.3.3.el6 (RHEL6.6)
12 3.10.0-123.9.2.el7 (RHEL7)
13 3.0.101-0.40 (SLES11 SP3)
14 * Recommended e2fsprogs version: 1.42.9.wc1 or newer
15 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
16 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
17 stack overflow. For more information, please refer to bugzilla 17630.
18 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
19 the re-exporting nodes to be mounted with "32bitapi" mount option
21 --------------------------------------------------------------------------------
23 07-30-2014 Intel Corporation
25 * See https://wiki.hpdd.intel.com/display/PUB/Lustre+Support+Matrix
26 for currently supported client and server kernel versions.
27 * Server known to build on patched kernels:
28 2.6.32-431.20.3.el6 (RHEL6.5)
29 3.0.101-0.31 (SLES11 SP3)
30 * Client known to build on unpatched kernels:
31 2.6.32-431.20.3.el6 (RHEL6.5)
32 3.0.101-0.31 (SLES11 SP3)
35 * Recommended e2fsprogs version: 1.42.9.wc1 or newer
36 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
37 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
38 stack overflow. For more information, please refer to bugzilla 17630.
39 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
40 the re-exporting nodes to be mounted with "32bitapi" mount option
42 --------------------------------------------------------------------------------
44 10-31-2013 Intel Corporation
46 * See https://wiki.hpdd.intel.com/display/PUB/Lustre+Support+Matrix
47 * Server builds for kernels:
48 2.6.32-358.18.1.el6 (RHEL6.4)
49 * Client builds for unpatched kernels:
50 2.6.32-358.18.1.el6 (RHEL6.4)
51 3.0.92-0.8 (SLES11 SP3)
53 * Recommended e2fsprogs version: 1.42.7.wc2 or newer
54 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
55 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
56 stack overflow. For more information, please refer to bugzilla 17630.
57 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
58 the re-exporting nodes to be mounted with "32bitapi" mount option
60 --------------------------------------------------------------------------------
62 05-30-2013 Intel Corporation
64 * See https://wiki.hpdd.intel.com/display/PUB/Lustre+Support+Matrix
65 * Server builds for kernels:
66 2.6.32-358.6.2.el6 (RHEL6.4)
67 * Client builds for unpatched kernels:
68 2.6.32-358.6.2.el6 (RHEL6.4)
69 3.0.74-0.6.6 (SLES11 SP2)
71 * Recommended e2fsprogs version: 1.42.7.wc1 or newer
72 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
73 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
74 stack overflow. For more information, please refer to bugzilla 17630.
75 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
76 the re-exporting nodes to be mounted with "32bitapi" mount option
78 --------------------------------------------------------------------------------
80 9-30-2012 Whamcloud, Inc.
82 * See https://wiki.hpdd.intel.com/display/PUB/Lustre+Support+Matrix
83 * Server support for kernels:
84 2.6.32-279.5.1.el6 (RHEL6)
85 * Client support for unpatched kernels:
86 2.6.32-279.5.1.el6 (RHEL6)
88 2.6.18-238.19.1.el5 (RHEL5)
89 2.6.16.60-0.69.1 (SLES10)
90 * Recommended e2fsprogs version: 1.41.90.wc4
91 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
92 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
93 stack overflow. For more information, please refer to bugzilla 17630.
94 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
95 the re-exporting nodes to be mounted with "32bitapi" mount option
97 Severity : enhancement
99 Description: Update RHEL6.3 kernel to 2.6.32-279.5.1.el6.
101 Severity : enhancement
103 Description: Update RHEL6.3 kernel to 2.6.32-279.2.1.el6.
105 Severity : enhancement
107 Description: Update RHEL6.2 kernel to 2.6.32-220.17.1.el6
109 Severity : enhancement
111 Description: Update RHEL6.2 kernel to 2.6.32-220.13.1.el6
113 Severity : enhancement
115 Description: Update RHEL6.2 kernel to 2.6.32-220.7.1.el6
117 --------------------------------------------------------------------------------
119 04-30-2012 Whamcloud, Inc.
121 * See https://wiki.hpdd.intel.com/display/PUB/Lustre+Support+Matrix
122 * Server support for kernels:
123 2.6.32-220.4.2.el6 (RHEL6)
124 * Client support for unpatched kernels:
125 2.6.32-220.4.2.el6 (RHEL6)
127 2.6.18-238.19.1.el5 (RHEL5)
128 2.6.16.60-0.69.1 (SLES10)
129 * Recommended e2fsprogs version: 1.41.90.wc4
130 * NFS export disabled when stack size < 8192 (32-bit Lustre clients),
131 since the NFSv4 export of Lustre filesystem with 4K stack may cause a
132 stack overflow. For more information, please refer to bugzilla 17630.
133 * NFSv4 reexport to 32-bit NFS client nodes requires Lustre client on
134 the re-exporting nodes to be mounted with "32bitapi" mount option
136 Severity : enhancement
138 Description: Update RHEL6.2 kernel to 2.6.32-220.4.2.el6
140 Severity : enhancement
142 Description: Update RHEL6.2 kernel to 2.6.32-220.4.1.el6
144 Severity : enhancement
146 Description: Update RHEL6.2 kernel to 2.6.32-220.el6
148 Severity : enhancement
150 Description: Update RHEL6.1 kernel to 2.6.32-131.17.1.el6
152 --------------------------------------------------------------------------------
154 09-30-2011 Whamcloud, Inc.
156 * Server support for kernels:
157 2.6.18-238.19.1.el5 (RHEL5)
158 2.6.32-131.6.1.el6 (RHEL6)
159 * Client support for unpatched kernels:
160 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
161 2.6.18-238.19.1.el5 (RHEL5)
162 2.6.32-131.6.1.el6 (RHEL6)
163 2.6.29.4-167.fc11 (FC11)
164 2.6.16.60-0.69.1 (SLES10)
166 * Recommended e2fsprogs version: 1.41.90.wc3
167 * NFS export disabled when stack size < 8192. Since the NFSv4 export of
168 Lustre filesystem with 4K stack may cause a stack overflow. For more
169 information, please refer to bugzilla 17630.
170 * NFSv4 reexport requires Lustre client on the re-exporting nodes to be
171 mounted with "32bitapi" mount option
173 Severity : enhancement
174 Description: Update RHEL6.1 kernel to 2.6.32-131.6.1.el6
176 Severity : enhancement
177 Description: Update RHEL5.6 kernel to 2.6.18-238.19.1.el5
179 Severity : enhancement
180 Description: Update RHEL5.6 kernel to 2.6.18-238.9.1.el5
183 Frequency : only with SLES10
185 Description: Use OFED "KMP" provided by Novell
186 Details : SLES10 SP3 ships with OFED in a separate "KMP" package.
187 Lustre is now built against this package. That means you need to
188 install the ofed-kmp package from Novell for the patchless client.
189 Note that the ofed-kmp that Novell ships may not exactly match the
190 kernel version but should still be compatible.
192 Severity : enhancement
194 Description: Update RHEL5.5 kernel to 2.6.18-194.17.1.el5
195 Update OEL5.5 kernel to 2.6.18-194.17.1.0.1.el5.
197 --------------------------------------------------------------------------------
199 2010-07-15 Oracle, Inc.
201 * Server support for kernels:
202 2.6.18-164.11.1.el5 (RHEL 5)
203 2.6.18-164.11.1.0.1.el5 (OEL 5)
204 * Client support for unpatched kernels:
205 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
206 2.6.18-164.11.1.el5 (RHEL 5),
207 2.6.18-164.11.1.0.1.el5 (OEL 5)
208 2.6.16.60-0.42.8 (SLES 10),
210 2.6.29.4-167.fc11 (FC11)
211 * Recommended e2fsprogs version: 1.41.10-sun2
213 Severity : enhancement
215 Description: Distribute lfs_migrate script with lustre
216 Details : The lfs_migrate mentioned in the manual is now included in the
217 lustre tree in lustre/scripts and will distributed in lustre RPMs.
219 Severity : enhancement
221 Description: Failover nid list added to import proc file.
223 Severity : enhancement
225 Description: "som_preview" mount option required to use SOM
226 Details : The clients will need to use som_preview mount option
230 Description: Lustre does not do 1MB IOs to HW RAID
231 Details : Bump MAX_PHYS/HW_SEGMENTS and SG_ALL to 256 in the RHEL5 kernel.
233 Severity : enhancement
235 Description: lfs quota output is not very convenient for awk/sed-parsing
236 Details : Some positions in lfs quota output table could be empty or
237 non-empty which made it hard to parse it with scripts, now
238 a dash is put instead of space where there is not supposed
243 Description: Update to OFED 1.5.1
244 Details : OFED 1.5.1 will be used in environments that don't use
245 vendor supplied OFED.
249 Description: Remove lnet. prefix from lctl params display
250 Details : Also renamed lnet.memused to lnet_memused to
255 Description: Use conf_param -d to remove permanant settings.
259 Description: Add list_param -R
260 Details : lctl list_param now supports recursive mode.
263 Bugzilla : 21829, 22312
264 Description: Fix llobdstat.
265 Details : 1. Now updating the write_bytes counter
266 2. Port script fixes from 1.8.
270 Description: decrease the usage of memory on clients.
271 Details : 1. On clients, recycle dentries and inodes unused.
272 2. Delete the code related to ll_deathrow(att 6215 in bug 1443). It
277 Description: llobdstat fix and enhancement
278 Details : add a counter to set a limit to how many samples will be returned
279 fix a wildcard in the path to limit to obdfilter stats only
283 Description: lfs check only allowed for root
284 Details : Code cleanups and added a regresion test of non-root use of
289 Description: MDS Oopses in ptlrpc_dispatch_difficult_reply()/__list_add()
290 Details : Fix hr->hr_index overload.
294 Description: Fix FIEMAP functionality.
298 Description: FC11 patchless client support added
302 Description: Update RHEL5.4 kernel to 2.6.18-164.11.1.el5 and
303 OEL5.4 kernel to 2.6.18-164.11.1.0.1.el5.
307 Description: New acc-sm script and various test-framework enhancments.
311 Description: Previously landed patch was reverted.
315 Description: Update SLES10 kernel to 2.6.16.60-0.42.8 (patchless)
319 Description: Add ldiskfs maxdirsize mount option.
323 Description: Fix mkfs.lustre support for 16TB LUNs.
327 Description: Print mmp_check_interval
328 Details : print mmp_check_interval and make it possible to abort
329 mount operation in case it takes too long.
333 Description: Add cascading_rw to lustre mpi tests.
337 Description: Store SOM data in EA.
338 Details : Also several SOM related cleanups.
342 Description: Add refcount to md_open_data
346 Description: Use refcount on lov_request_set to prevent memory corruption.
350 Description: Statahead: use parent dir's i_mutex for synchronization.
354 Description: loadgen improvements.
358 Description: Enable peer health detection in ksocklnd and ko2iblnd.
362 Description: Control DCACHE_LUSTRE_INVALID flag with MDS_INODE_LOOKCUP lock
366 Description: Fix race on obd_devcie:md_stats
370 Description: Fix SLES 11 client builds.
371 Details : Resolved uninitalized pointer gcc warnings.
375 Description: Allow non-root access for "lfs check".
376 Details : Added a check in obd_class_ioctl() for OBD_IOC_PING_TARGET.
380 Description: Clear imp_force_reconnect correctly in ptlrpc_connect_interpret()
382 Severity : enhancement
384 Description: Update kernel to RHEL5.4 2.6.18-164.6.1.el5 and
385 OEL5 2.6.18-164.6.1.0.1.el5(Both in-kernel OFED enabled).
389 Description: Sanity test_27p fails.
390 Details : OBD_FAIL_OST_ENOSPC has been moved to filter_statfs().
391 Sanity tests 27 were cleaned up and improved.
395 Description: Unify req format on client/servers
396 Details : Use new req_capsule API [almost] everywhere instead of old PTLRPC
397 buffers and swabbers approach..
402 Description: drop join file stuff.
404 Severity : enhancement
406 Description: Add 2.6.27 and SLES11 (patchless client) support
410 Description: Fix in ptlrpc_expire_one_request() to print the signed time
413 Severity : enhancement
415 Description: Build kernels (RHEL5, OEL5 and SLES10/11) using the vendor's own
418 Severity : enhancement
420 Description: Remove set_info(KEY_UNLINKED) from MDS/OSC
422 Severity : enhancement
424 Description: correctly handle big reply message.
425 Details : send LNet event if reply is bigger then buffer and adjust this buffer
430 Description: Drop unnecessary __GFP_NOMEMALLOC flag from filter_get_page()
432 Severity : enhancement
434 Description: Add support for OEL5.
438 Description: Conf-sanity.sh 50g test - deactivated OST should not cause a panic.
440 Severity : enhancement
442 Description: "weak-modules" support
443 Details : Implement "weak-modules" support which enables kernel modules
444 to be used with any kernel that implements the same kABI. In
445 order to achieve this modules are now installed in
446 /lib/modules/$(uname -r)/updates/kernel on all distributions.
450 Description: client could not reconnect to OST because of an active request.
451 Details : The client now retries to reconnect to the same server, if a
452 connect request failed with EBUSY or -EAGAIN. Test case 35b
453 conf-sanity.sh has been created.
457 Description: Don't allow make backward step on assignin osc next id.
458 Details : race between allocation next id and ll_sync thread can be cause
459 of set wrong osc next id and can be kill valid ost objects.
463 Description: Changes in raid5-large-io-rhel5.patch to calculate sectors properly
467 Description: Increase the default BLK_DEF_MAX_SECTORS value for RHEL5 and SLES11
471 Description: Error handling in osc_statfs_interpret() has been improved.
472 Details : Check in osc_statfs_interpret() for EBADR.
476 Description: Do not send statfs() requests to OSTs disabled by administrator.
477 Details : Check in lov_prep_statfs_set() for non-NULL ltd_exp.
479 Severity : enhancement
481 Description: Added sanity-quota test to for lfs(1) to work with numeric uid/gid.
483 Severity : enhancement
485 Description: Update kernel to RHEL5 2.6.18-128.7.1.el5.
487 Severity : enhancement
489 Description: Enhancement for lfs(1) command to use numeric uid/gid.
493 Description: Improved cleanup procedure for conf-sanity test_46a.
497 Description: Do not start bulk transfer, if client tried to reconnect.
498 Details : Do not start bulk transfer, if client tried to reconnect,
499 but failed due to active rpcs. Check for exp_abort_active_req
500 has been implemented.
504 Description: Prevent inconsistences between linux and lustre mount structures.
505 Details : Wait indefinitely in server_wait_finished() until mnt_count drops.
506 Make the sleep interruptible.
508 Severity : enhancement
510 Description: provide server to client comms path
511 Details : server to client communications path via new LDLM_SET_INFO rpc
515 Description: truncate starts GFP_FS allocation under transaction causing deadlock
516 Details : ldiskfs_truncate calls grab_cache_page which may start page
517 allocation under an open transaction. This may lead to
518 calling prune_icache with consequent lustre reentrance.
522 Description: send correctly lsm on open replay
523 Details : MDS is trust to LSM size on replay open, but client can set wrong size
528 Description: enable client interface failover
529 Details : When a child reconnects from another NID, properly update export
530 nid hash position and ldlm reverse import.
532 Severity : enhancement
534 Description: Communicate OST degraded/readonly state via statfs to MDS
535 Details : Flags in the statfs returned from OSTs indicate whether the
536 OST is in a degraded RAID state, or if the filesystem has
537 turned read-only after a filesystem error is detected.
542 Description: don't panic if EPROTO was hit when reading symlink
543 Details : correctly handling request reference in error cases.
545 Severity : enhancement
547 Description: Add LustreNetLink, a kernel-userspace communcation path.
549 Severity : enhancement
551 Description: Update kernel to SLES10 SP2 2.6.16.60-0.39.3.
556 Description: access to llog context before init.
557 Details : move handling CATALOGS file at osc layer and forbid access to llog
561 Frequency : rare, only when client is writing to OST on local node
563 Description: Avoid deadlock for local client writes
564 Details : Use new OBD_BRW_MEMALLOC flag to notify OST about writes in the
565 memory freeing context. This allows OST threads to set the
566 PF_MEMALLOC flag on task structures in order to allocate memory
567 from reserved pools and complete IO.
568 Use GFP_HIGHUSER for OST allocations for non-local client writes,
569 so that the OST threads generate memory pressure and allow
570 inactive pages to be reclaimed.
572 Severity : enhancement
574 Description: Update kernel to RHEL5.3 2.6.18-128.1.14.el5.
579 Description: lock ordering violation between &cli->cl_sem and _lprocfs_lock
580 Details : move ldlm namespace creation in setup phase to avoid grab
581 _lprocfs_lock with cli_sem held.
585 Description: Temporarily disable grant shrink.
586 Details : Disable the feature for debugging.
589 Frequency : only when formatting test filesystems
591 Description: Unable to run several mkfs.lustre on loop devices at the same time
592 Details : mkfs.lustre returns error 256 on the concurrent loop devices
593 formatting. The solution is to proper handle the error.
595 Severity : enhancement
597 Description: Update kernel to RHEL5.3 2.6.18-128.1.6.el5.
599 Severity : enhancement
601 Description: Update kernel to SLES10 SP2 2.6.16.60-0.37.
606 Description: resolve race between obd_disconnect and class_disconnect_exports
607 Details : if obd_disconnect will be called to already disconnected export he
608 forget release one reference and osc module can't unloaded.
610 Severity : enhancement
612 Description: Allow tuning service thread via /proc
613 Details : For each service a new
614 /proc/fs/lustre/{service}/*/thread_{min,max,started} entry is
615 created that can be used to set min/max thread counts, and get the
616 current number of running threads.
621 Descriptoin: don't return error if have partially created objects for file.
622 Details : lov_update_create_set uses set->set_success as index for created
623 objects, so if some requests will be failed, they will have hole
624 at end of array and qos_shrink_lsm for allocate correct lsm.
626 Severity : enhancement
628 Description: Update OFED support to 1.4.1
633 Description: Reduce small size read RPC
634 Details : Set read-ahead limit for every file and only do read-ahead when
635 available read-ahead pages are bigger than 1M to avoid small size
638 Severity : enhancement
640 Description: move AT tunable parameters for more consistent usage
641 Details : add AT tunables under /proc/sys/lustre, add to conf_param parsing
643 Severity : enhancement
645 Description: add lazystatfs mount option to allow statfs(2) to skip down OSTs
646 Details : allow skip disconnected ost for send statfs request and hide error
652 Description: fix racy locking of mballoc block bitmaps causing BUG
653 Details : The locking of the mballoc buddy bitmap and the in-memory
654 block bitmap was using two different spin locks in some
655 cases. This made it possible to incorrectly access the
656 mballoc bitmap while another process was modifying it,
657 causing a sanity assertion to fail. While no on-disk corruption
658 was reported, there was some risk of this happening.
661 Frequency : rare, on llog test 6
663 Descriptoin: don't allow connect to already connected import
664 Details : allowing connect to already connected import is hide connecting problem.
669 Descriptoin: allow kill process which wait statahead result
670 Details : for some reasons 'ls' can stick in waiting result from statahead,
671 in this case need way for kill this process.
673 Severity : enhancement
675 Description: Add state history info file, enhance import info file
676 Details : Track import connection state changes in a new osc/mdc proc file;
677 add overview-type data to the osc/mdc import proc file.
679 Severity : enhancement
681 Description: MDS create should not wait for statfs RPC while holding DLM lock.
683 Severity : enhancement
685 Description: Update to RHEL5U3 kernel-2.6.18-128.1.1.el5.
690 Descriptoin: OST grant too much space to client even there are not enough space.
691 Details : Client will shrink its grant cache to OST if there are no write
692 activity over 6 mins (GRANT_SHRINK_INTERVAL), and OST will retrieve
693 this grant cache if there are already not enough avaible space
694 (left_space < total_clients * 32M).
697 Frequency : start MDS on uncleanly shutdowned MDS device
699 Descriptoin: ll_sync thread stay in waiting mds<>ost recovery finished
700 Details : stay in waiting mds<>ost recovery finished produce random bugs
701 due race between two ll_sync thread for one lov target. send
702 ACTIVATE event only if connect realy finished and import have
706 Frequency : rare, connect and disconnect target at same time
708 Descriptoin: ASSERTION(atomic_read(&imp->imp_inflight) == 0
709 Details : don't call obd_disconnect under lov_lock. this long time
710 operation and can block ptlrpcd which answer to connect request.
715 Descriptoin: don't lose wakeup for imp_recovery_waitq
716 Details : recover_import_no_retry or invalidate_import and import_close can
717 both sleep on imp_recovery_waitq, but we was send only one wakeup
721 Frequency : always with long access acl
723 Descriptoin: mds can't pack reply with long acl.
724 Details : mds don't control size of acl but they limited by reint/getattr
727 Severity : enhancement
729 Description: Update to SLES10 kernel-2.6.16.60-0.33.
731 Severity : enhancement
733 Description: Update to RHEL5 kernel-2.6.18-92.1.22.el5.
736 Frequency : start MDS on uncleanly shutdowned MDS device
738 Descriptoin: aborting recovery hang on MDS
739 Details : don't throttle destroy RPCs for the MDT.
742 Frequency : on remount
744 Description: external journal device not working after the remount
745 Details : clear dev_rdonly flag for external journal devices in
751 Description: shutdown vs evict race
752 Details : client_disconnect_export vs connect request race.
753 if client will evicted at this time - we start invalidate
754 thread without referece to import and import can be freed
758 Frequency : rare, need acl's on inode.
760 Description: client can't handle ost additional correctly
761 Details : if ost was added after client connected to mds client can have
762 hit lnet_try_match_md ... to big messages to wide striped files.
763 in this case need teach client to handle config events about add
764 lov target and update client max ea size at that event.
766 Severity : enhancement
768 Description: Changelogs
769 Details : Changelogs are a lightweight mechanism to track filesystem
770 metadata and namespace changes. The changelog is recorded
771 permanently on the MDTs, and is periodically "consumed" / purged
772 when records are no longer needed.
774 Severity : enhancement
776 Description: compact fld format with extents
777 Details : Store range of seq rather than every seq in FLD. Seq
778 controller update FLD rather than clients. In Case of CMD, mdt0
779 has FLD, all other metadata server act as non persistent proxy
780 for FLD queries and cache fld entries in fld cache.
785 Description: don't skip ost target if they assigned to file
786 Details : Drop slow OSCs if we can, but not for requested start idx.
787 This means "if OSC is slow and it is not the requested
788 start OST, then it can be skipped, otherwise skip it only
789 if it is inactive/recovering/out-of-space.
793 Description: more cleanup in mds_lov
794 Details : not send LOV EA under replay, we can't know about they size at this
795 time. Don't allow client connect to mds before any ost connected,
796 for avoid problems with LOV EA size and returning EIO to client.
798 Severity : enhancement
800 Description: Interoperability at server side (Disk interoperability)
802 Severity : enhancement
804 Description: Update to RHEL5 kernel-2.6.18-92.1.17.el5.
806 Severity : enhancement
808 Description: Update to SLES10 SP2 kernel-2.6.16.60-0.31.
810 Severity : enhancement
812 Description: New client IO stack (CLIO).
814 Severity : enhancement
816 Description: Commit on sharing. Eliminate inter-client dependencies between
817 uncommitted transactions by doing transaction commits.
818 Thereby clients may recovery independently.
821 Frequency : Create a symlink file with a very long name
823 Description: ldlm_cancel_pack()) ASSERTION(max >= dlm->lock_count + count)
824 Details : If there is no extra space in the request for early cancels,
825 ldlm_req_handles_avail() returns 0 instead of a negative value.
827 Severity : enhancement
829 Description: Add /proc entry for import status
830 Details : The mdc, osc, and mgc import directories now have
831 an import directory that contains useful import data for debugging
834 Severity : enhancement
836 Description: Re-disable certain /proc logging
837 Details : Enable and disable client's offset_stats, extents_stats and
838 extents_stats_per_process stats logging on the fly.
841 Frequency : Only on FC kernels 2.6.22+
843 Description: oops in statahead
844 Details : Do not drop reference count for the dentry from VFS when lookup,
845 VFS will do that by itself.
847 Severity : enhancement
849 Description: Generic /proc file permissions
850 Details : Set /Proc file permissions in a more generic way to enable non-
851 root users operate on some /proc files.
855 Description: Hitting mdc_commit_close() ASSERTION
856 Details : Properly handle request reference release in
857 ll_release_openhandle().
861 Description: quota recovery deadlock during mds failover
862 Details : This patch includes att18982, att18236, att18237 in bz14840.
864 1. fix osts hang when mds does failover with quotaon
865 2. prevent watchdog storm when osts threads wait for the
870 Frequency : only patchless client
871 Description: add workaround for race between add/remove dentry from hash
873 Severity : enhancement
875 Description: Allow OST glimpses to return PW locks
879 Description: LBUG when llog conf file is full
880 Details : When llog bitmap is full, ENOSPC should be returned for plain log.
884 Description: Prevent import from entering FULL state when server in recovery
888 Description: service mount cannot take device name with ":"
889 Details : Only when device name contains ":/" will mount treat it as
895 Description: replace ptlrpcd with the statahead thread to interpret the async
896 statahead RPC callback
900 Frequency : on recovery
901 Description: I/O failures after umount during fail back
902 Details : if client reconnected to restarted server we need join to recovery
903 instead of find server handler is changed and process self
904 eviction with cancel all locks.
906 Severity : enhancement
908 Description: Update to RHEL5 kernel-2.6.18-92.1.10.el5.
910 Severity : enhancement
912 Description: Update to SLES10 SP2 kernel-2.6.16.60-0.27.
914 Severity : enhancement
916 Description: Upcall on Lustre log has been dumped
917 Details : Allow for a user mode script to be called once a Lustre log has
918 been dumped. It passes the filename of the dumped log to the
919 script, the location of the script can be specified via
920 /proc/sys/lnet/debug_log_upcall.
925 Description: avoid idr_remove called for id which is not allocated.
926 Details : Move assigment s_dev for clustered nfs to end of initialization,
927 to avoid problem with error handling.
932 Description: avoid Already found the key in hash [CONN_UNUSED_HASH] messages
933 Details : When connection is reused this not moved from CONN_UNUSED_HASH
934 into CONN_USED_HASH and this prodice warning when put connection
935 again in unused hash.
938 Severity : enhancement
940 Description: File striping can now be set to use an arbitrary pool of OSTs.
942 Severity : enhancement
944 Description: Export bytes_read/bytes_write count on OSC/OST.
948 Description: Early reply size mismatch, MGC loses connection
949 Details : Apply the MGS_CONNECT_SUPPORTED mask at reconnect time so
950 the connect flags are properly negotiated.
955 Description: quotas are not honored with O_DIRECT
956 Details : all writes with the flag O_DIRECT will use grants which leads to
957 this problem. Now using OBD_BRW_SYNC to guard this.
961 Description: add quota statistics
962 Details : 1. sort out quota proc entries and proc code.
963 2. add quota statistics
965 Severity : enhancement
967 Description: enable quota support for HEAD.
971 Description: Properly propagate oinfo flags from lov to osc for statfs
972 Details : restore missing copy oi_flags to lov requests.
974 Severity : enhancement
976 Description: Add man pages for llobdstat(8), llstat(8), plot-llstat(8),
977 : l_getgroups(8), lst(8), routerstat(8)
978 Details : included man pages for llobdstat(8), llstat(8),
979 : plot-llstat(8), l_getgroups(8), lst(8), routerstat(8)
981 Severity : enhancement
983 Description: configure's --enable-quota should check the
984 : kernel .config for CONFIG_QUOTA
985 Details : configure is terminated if --enable-quota is passed but
986 : no quota support is in kernel
990 Description: Remove portals compatibility
991 Details : Remove portals compatibility, not interoperable with releases
996 Description: Resolve device initialization race
997 Details : Prevent proc handler from accessing devices added to the
998 obd_devs array but yet be intialized.
1000 Severity : enhancement
1002 Description: Update to SLES10 SP2 kernel-2.6.16.60-0.23.
1004 Severity : enhancement
1006 Description: Update to RHEL5 kernel-2.6.18-92.1.6.el5.
1011 Description: Using wrong pointer in osc_brw_prep_request
1012 Details : Access to array[-1] can produce panic if kernel compiled with
1013 CONFIG_PAGE_ALLOC enabled
1017 Description: Client runs out of low memory
1018 Details : Consider only lowmem when counting initial number of llap pages
1022 Description: *optional* service tags registration
1023 Details : if the "service tags" package is installed on a Lustre node
1024 When the filesystem is mounted, a local-node service tag will
1025 be created. See http://inventory.sun.com/ for more information
1026 about the Service Tags asset management system.
1030 Description: Kernel BUG tries to release flock
1031 Details : Lustre does not destroy flock lock before last reference goes
1032 away. So always drop flock locks when client is evicted and
1033 perform unlock regardless of successfulness of speaking to MDS.
1037 Description: add refcount protection for osc callbacks, avoid panic on shutdown
1041 Description: sanity test 65a fails if stripecount of -1 is set
1042 Details : handle -1 striping on filesystem in ll_dirstripe_verify
1047 Description: ASSERTION(CheckWriteback(page,cmd)) failed
1048 Details : badly clear PG_Writeback bit in ll_ap_completion can produce false
1051 Severity : enhancement
1053 Description: Update to RHEL5 kernel-2.6.18-53.1.21.el5.
1057 Description: do not process already freed flock
1058 Details : flock can possibly be freed by another thread before it reaches
1059 to ldlm_flock_completion_ast.
1063 Description: LBUG during stress test
1064 Details : Need properly lock accesses the flock deadlock detection list.
1068 Description: oops in page fault handler
1069 Details : kernel page fault handler can return two special 'pages' in error
1070 case, don't try dereference NOPAGE_SIGBUS and NOPAGE_OMM.
1074 Description: timeout with invalidate import.
1075 Details : ptlrpcd_check call obd_zombie_impexp_cull and wait request which
1076 should be handled by ptlrpcd. This produce long age waiting and
1077 -ETIMEOUT ptlrpc_invalidate_import and as result LASSERT.
1079 Severity : enhancement
1081 Description: Update to RHEL5 kernel-2.6.18-53.1.19.el5.
1085 Description: enable MGS and MDT services start separately
1086 Details : add a 'nomgs' option in mount.lustre to enable start a MDT with
1087 a co-located MGS without starting the MGS, which is a complement
1088 to 'nosvc' mount option.
1092 Frequency : after recovery
1093 Description: precreate to many object's after del orphan.
1094 Details : del orphan st in oscc last_id == next_id and this triger growing
1095 count of precreated objects. Set flag LOW to skip increase count
1096 of precreated objects.
1100 Frequency : rare, on clear nid stats
1101 Description: ASSERTION(client_stat->nid_exp_ref_count == 0)
1102 Details : when clean nid stats sometimes try destroy live entry,
1103 and this produce panic in free.
1107 Description: Stack overflow during MDS log replay
1108 ease stack pressure by using a thread dealing llog_process.
1112 Description: wait until IO finished before start new when do lock cancel.
1113 Details : VM protocol want old IO finished before start new, in this case
1114 need wait until PG_writeback is cleared until check dirty flag and
1115 call writepages in lock cancel callback.
1117 Severity : enhancement
1119 Description: using special macro for print time and cleanup in includes.
1123 Description: mds_mfd_close() ASSERTION(rc == 0)
1124 Details : In mds_mfd_close(), we need protect inode's writecount change
1125 within its orphan write semaphore to prevent possible races.
1129 Description: Obsolete CURRENT_SECONDS and use cfs_time_current_sec() instead.
1133 Frequency : rare, on shutdown ost
1134 Description: don't hit live lock with umount ost.
1135 Details : shrink_dcache_parent can be in long loop with destroy dentries,
1136 use shrink_dcache_sb instead.
1140 Description: don't panic with use echo client
1141 Details : echo client pass NULL as client nid pointer and this produce null
1142 pointer dereference.
1146 Description: fix build on ppc32
1147 Details : compile code with -m64 flag produce wrong object file for ppc32.
1151 Description: add message levels for liblustreapi
1155 Description: fix for occasional failure case of -ENOSPC in recovery-small tests
1156 Details : Move the 'good_osts' check before the 'total_bavail' check. This
1157 will result in an -EAGAIN and in the exit call path we call
1158 alloc_rr() which will with increasing aggressiveness attempt to
1159 aquire precreated objects on the minimum number of required OSCs.
1163 Description: Use old size assignment to avoid deadlock
1164 Details : Reverts the changes in bugs 2369 and bug 14138 that introduced
1165 the scheduling while holding a spinlock. We do not need locking
1166 for size in ll_update_inode() because size is only updated from
1167 the MDS for directories or files without objects, so there is no
1168 other place to do the update, and concurrent access to such inodes
1169 are protected by the inode lock.
1173 Description: resolve "_IOWR redefined" build error on SLES10
1177 Description: dump the memory debugging after all modules are unloaded to
1178 suppress false negative in conf_sanity test 39
1180 Severity : enhancement
1182 Description: build kernel-ib packages for OFED 1.3 in our release cycle
1187 Description: fix SLES kernel versioning
1188 Details : the kernel version for our SLES 10 kernel did not include a "-"
1189 before the "smp" at the end. while this was not a problem in
1190 general, it did mean that software trying to use the kernel
1191 version to try to detect a vendor specific kernel would fail.
1192 this was most evident by the OFED build scripts.
1196 Description: Don't update lov_desc members until making sure they are valid
1197 Details : When updating lov_desc members via proc fs, need fix their
1198 validities before doing the real update.
1202 Description: don't put request into delay list while invalidate in flight.
1203 Details : ptlrpc_delay_request sometimes put in delay list while invalidate
1204 import in flight. this produce timeout for invalidate and sometimes
1205 can cause stale data.
1209 Frequency : on ppc only
1210 Description: not convert ost objects for directory because it's not exist.
1211 Details : ll_dir_getstripe assume dirrectory has ost objects but this wrong.
1215 Description: Add FMODE_EXEC file flag for SLES10 SP1 kernel.
1217 Severity : enhancement
1219 Description: Update to support 2.6.22.14 vanilla kernel.
1223 Frequency : rare, on recovery
1224 Description: read procfs can produce deadlock in some situation
1225 Details : Holding lprocfs lock which send rpc can produce block for destroy
1226 obd objects and this also block reconnect with -EALREADY. This
1227 isn't fix all lprocfs bugs - but make it rare.
1229 Severity : enhancement
1231 Description: Update kernel to RHEL5 2.6.18-53.1.14.el5.
1234 Frequency : frequent on X2 node
1236 Description: mdc_set_open_replay_data LBUG
1237 Details : Set replay data for requests that are eligible for replay.
1241 Description: lustre_mgs: operation 101 on unconnected MGS
1242 Details : When MGC is disconnected from MGS long enough, MGS will evict the
1243 MGC, and late on MGC cannot successfully connect to MGS and a lot
1244 of the error messages complaining that MGS is not connected.
1248 Frequency : on network error
1249 Description: panic with double free request if network error
1250 Details : mdc_finish_enqueue is finish request if any network error ocuring,
1251 but it's true only for synchronus enqueue, for async enqueue
1252 (via ptlrpcd) this incorrect and ptlrpcd want finish request
1255 Severity : enhancement
1257 Description: client-side metadata stat-ahead during readdir(directory readahead)
1258 Details : perform client-side metadata stat-ahead when the client detects
1259 readdir and sequential stat of dir entries therein
1262 Frequency : on start mds
1264 Description: Implement get_info(last_id) in obdfilter.
1267 Frequency : occasional
1269 Description: Correctly check stale fid, not start epoch if ost not support SOM
1270 Details : open with flag O_CREATE need set old fid in op_fid3 because
1271 op_fid2 was overwritten with new generated fid, but mds can answer
1272 with one of these two fids and both is not stale. Setattr
1273 incorrectly started an epoch and assume will be called
1274 done_writing, but without SOM done_writing ever being called.
1277 Frequency : rare, depends on device drivers and load
1279 Description: MDS or OSS nodes crash due to stack overflow
1280 Details : Code changes in 1.8.0 increased the stack usage of some functions.
1281 In some cases, in conjunction with device drivers that use a lot
1282 of stack the MDS (or possibly OSS) service threads could overflow
1283 the stack. One change which was identified to consume additional
1284 stack has been reworked to avoid the extra stack usage.
1287 Frequency : occasional
1289 Description: Do not fail import if osc_interpret_create gets -EAGAIN
1290 Details : If osc_interpret_create got -EAGAIN it immediately exits and
1291 wakeup oscc_waitq. After wakeup oscc_wait_for_objects call
1292 oscc_has_objects and see OSC has no objests and call
1293 oscc_internal_create to resend create request.
1295 Severity : enhancement
1297 Description: Update to SLES10 SP1 latest kernel-2.6.16.54-0.2.5.
1299 Severity : enhancement
1301 Description: Update to RHEL5 latest kernel-2.6.18-53.1.13.el5.
1304 Frequency : very rare
1306 Description: Fix replay if there is an un-replied request and open
1307 Details : In some cases, older replay request will revert the
1308 mcd->mcd_last_xid on MDS which is used to record the client's
1309 latest sent request.
1311 Severity : enhancement
1313 Description: Update to RHEL5 latest kernel-2.6.18-53.1.6.el5.
1315 Severity : enhancement
1317 Description: Add rhel5 support to HEAD.
1319 Serverity : enhancement
1321 Description: Update RHEL4 kernel to 2.6.9-67.0.4.
1326 Description: Don't allow skipping OSTs if index has been specified.
1327 Details : Don't allow skipping OSTs if index has been specified, make
1328 locking in internal create lots better.
1332 Description: LBUG in ptlrpc_check_set() bad phase ebc0de00
1333 Details : access to bitfield in structure is always rounded to long
1334 and this produce problem with not atomic change any bit.
1338 Description: Lustre make rpms failed.
1339 Details : Remove ldiskfs spec file to avoids rpmbuild be confused when
1340 builds Lustre rpms from tarball.
1343 Frequency : rare on shutdown ost
1345 Description: If llog cancel was not send before clean_exports phase, this can
1346 produce deadlock in llog code.
1347 Details : If llog thread has last reference to obd and call class_import_put
1348 this produce deadlock because llog_cleanup_commit_master wait when
1349 last llog_commit_thread exited, but this never success because was
1350 called from llog_commit_thread.
1354 Description: allow userland application know is lost one of stripes.
1355 Details : fill lvb_blocks with error code on ost and return it to
1356 application if error flag found.
1360 Description: NULL lov_tgts causing MDS oops
1361 Details : more safe checks for NULL lov_tgts for avoid oops.
1363 Severity : enhancement
1365 Description: Update to RHEL4 latest kernel-2.6.9-67.0.1.EL.
1369 Descriptoin: make lov_create() will not stuck in obd_statfs_rqset()
1370 Details : If an OST is down the MDS will hang indefinitely in
1371 obd_statfs_rqset() waiting for the statfs data. While for
1372 MDS QOS usage of statfs, it should not stuck in waiting.
1374 Severity : enhancement
1376 Description: remote_acl support
1377 Details : Support ACL-based permission check for remote user.
1378 Support setfacl/getfacl for remote user with the utils
1379 "lfs {l,r}{s,g}etfacl" which follow the same parameter format as
1380 the system "{s,g}etfacl" utils.
1382 Severity : enhancement
1384 Description: Update to RHEL4 U6 kernel-2.6.9-67.EL.
1386 Severity : enhancement
1388 Description: Update to RHEL5 latest kernel-2.6.18-53.1.4.el5.
1392 Description: make mgs_setparam() handle fsname containing dash
1393 Details : fsname containing a dash does not work with lctl conf_param
1395 Severity : enhancement
1397 Description: Update to SLES10 SP1 latest kernel-2.6.16.54-0.2.3.
1399 Severity : enhancement
1401 Description: Update to RHEL5 Update-1 kernel 2.6.18-53.el5.
1405 Frequency : rare, at shutdown
1406 Description: access already free / zero obd_namespace.
1407 Details : if client_disconnect_export was called without force flag set,
1408 and exist connect request in flight, this can produce access to
1409 NULL pointer (or already free pointer) when connect_interpret
1410 store ocd flags in obd_namespace.
1414 Frequency : only at startup
1415 Description: not alloc memory with spinlock held.
1416 Details : allocation memory with GFP_KERNEL can produce sleep deadlock,
1417 if any spinlock held.
1419 Severity : enhancement
1421 Description: make lustre randomly fail allocating memory
1422 Details : Make lustre randomly failed allocating memory for testing purpose.
1424 Severity : enhancement
1426 Description: lost problems with lov objid file
1427 Details : Fixes some scability and access to not inited memory problems
1428 in work with lov objdid file.
1433 Description: lfs find does not continue on file error
1434 Details : Continue other files processing when a file/dir is absent.
1438 Description: Inconsistent usage of lustre_pack_reply()
1439 Details : Standardize the usage of lustre_pack_reply() such that it
1440 always generate a CERROR on failure.
1443 Frequency : occasional
1445 Description: MDS hang or stay in waiting lock
1446 Details : If client receive lock with CBPENDING flag ldlm need send lock
1447 cancel as separate rpc, to avoid situation when cancel request
1448 can't processed due all i/o threads stay in wait lock.
1452 Description: Update to RHEL5 kernel 2.6.18-8.1.15.el5.
1456 Description: Update to SLES10 SP1 kernel 2.6.16.53-0.16
1460 Description: Update to SLES9 kernel-2.6.5-7.287.3.
1464 Description: Update to RHEL4 latest kernel.
1466 Severity : enhancement
1468 Description: Build SLES10 patchless client fails
1469 Details : The configure was broken by run ./configure with
1470 --with-linux-obj=.... argument for patchless client. When the
1471 configure use --with-linux-obj, the LINUXINCLUDE= -Iinclude
1472 can't search header adequately. Use absolute path such as
1473 -I($LINUX)/include instead.
1477 Description: interrupt oig_wait produce painc on resend.
1478 Details : brw_redo_request can be used for resend requests from ptlrpcd and
1479 private set, and this produce situation when rq_ptlrpcd_data not
1480 copyed to new allocated request and triggered LBUG on assert
1481 req->rq_ptlrpcd_data != NULL. But this member used only for wakeup
1482 ptlrpcd set if request is changed and can be safety changed to use
1487 Description: LASSERT_{REQ,REP}SWAB macros are buggy
1488 Details : If SWAB_PARANOIA is disabled, the LASSERT_REQSWAB and
1489 LASSERT_REPSWAB macros become no-ops, which is incorrect. Drop
1490 these macros and replace them with their difinitions instead.
1494 Description: Update kernel patches for SLES10 2.6.16.53-0.8.
1495 Details : Update which_patch & target file for SLES10 latest kernel.
1498 Description: Remove client patches from SLES 10 kernel.
1499 Details : This causes SLES 10 clients to behave as patchless clients
1500 even on a Lustre-patched (server) kernel.
1502 Severity : enhancement
1504 Description: self-adjustable client's lru lists
1505 Details : use adaptive algorithm for managing client cached locks lru
1506 lists according to current server load, other client's work
1507 pattern, memory activities, etc. Both, server and client
1508 side namespaces provide number of proc tunables for controlling
1511 Severity : enhancement
1513 Description: light-weight GSS support
1514 Details : Support krb5n and krb5a mode, which keep Kerberos 5 authentication
1515 and reduce performance overhead.
1517 Severity : enhancement
1519 Description: Linux keyring support
1520 Details : Support using service of Linux keyring for Lustre GSS internal
1521 context refresh/cache mechanism.
1525 Description: Fix errors in lfs documentation
1526 Details : Fixes man pages
1530 Description: don't use GFP_* in generic Lustre code.
1531 Details : Use cfs_alloc_* functions and CFS_* flags for code portability.
1535 Description: obdclass is limited by single OBD_ALLOC(idarray)
1536 Details : replace OBD_ALLOC/OBD_FREE with OBD_VMALLOC/OBD_VFREE
1540 Description: warnings with build patchless client with vanila 2.6.19 and up
1541 Details : change old ctl_table style and replace ctl_table/ctl_table_header
1542 with cfs_sysctl_table_t/cfs_sysctl_table_header_t
1546 Frequency : Only for SLES
1547 Description: sanity_quota fail test_1
1548 Details : There are multiple occurences of $TSTUSR in SLES's /etc/group
1549 file, which makes TSTID[2] inunique.
1553 Frequency : Only for SLES9
1554 Description: Kernel patches for SLES9 2.6.5-7.286 kernel
1555 Details : Update target/ChangeLog/which_patch .
1559 Frequency : Only for test_10 in sanity-quota.sh
1560 Description: a bug in quota test code
1561 Details : checking a test flag in a wrong place
1565 Frequency : Only for test_13 in sanity-quota.sh
1566 Description: a bug in quota test script
1567 Details : shouldn't check overfull allocated quota
1571 Description: jbd statistics
1572 Details : Port older jbd statistics patch for sles10
1574 Severity : enhancement
1576 Description: Add printing inode info into message about error in writepage.
1581 Description: reply_lock_interpret crash due to race with it and lock cancel.
1582 Details : Do not replay locks that are being cancelled. Do not reference
1583 locks by their address during replay, just by their handle.
1587 Frequency : When flocks are used.
1588 Description: assertion failure in ldlm_cli_enquque_fini for non NULL lock.
1589 Details : Flock locks might destroy just granted lock if it could be merged
1590 with another existing flock, this is done in completion handler,
1591 so teach ldlm_cli_enquque_fini that this is a valid case for
1597 Description: Oops in read and write path when failing to allocate lock.
1598 Details : Check if lock allocation failed and return error back.
1602 Description: lstripe command fails for valid OST index
1603 Details : The stripe offset is compared to lov->desc.ld_tgt_count
1604 instead of lov->desc.ld_active_tgt_count.
1608 Description: sanity.sh failed test 103
1609 Details : RHEL mis-interpret setfacl "-X" param, so we won't test setfacl
1614 Description: df doesn't work properly if diskfs blocksize != 4K
1615 Details : Choose biggest blocksize of OST's as the LOV's blocksize.
1619 Description: lfs find on -1 stripe looping in lsm_lmm_verify_common()
1620 Details : Avoid lov_verify_lmm_common() on directory with -1 stripe count.
1622 Severity : enhancement
1624 Description: Adaptive timeouts
1625 Details : RPC timeouts adapt to changing server load and network
1626 conditions to reduce resend attempts and improve recovery time.
1630 Description: llapi_file_create() does not allow some changes
1631 Details : add llapi_file_open() that allows specifying the mode and
1632 open flags, and also returns an open file handle.
1636 Description: merge and cleanup kernel patches.
1637 Details : Remove mnt_lustre_list in vfs_intent-2.6-rhel4.patch.
1641 Description: Add journal checksum support.(Kernel part)
1642 Details : The journal checksum feature adds two new flags i.e
1643 JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT and
1644 JBD2_FEATURE_COMPAT_CHECKSUM. JBD2_FEATURE_CHECKSUM flag
1645 indicates that the commit block contains the checksum for
1646 the blocks described by the descriptor blocks. Now commit
1647 record can be sent to disk without waiting for descriptor
1648 blocks to be written to disk. This behavior is controlled
1649 using JBD2_FEATURE_ASYNC_COMMIT flag.
1653 Description: OSS needs mutliple precreate threads
1654 Details : Add ability to start more than one create thread per OSS.
1658 Description: Sanity.sh test_65k failed: setstripe should have succeeded
1659 Details : execute lfs setstripe on client
1663 Description: mds_obd_create error creating tmp object
1664 Details : When the user sets quota on root, llog will be affected and can't
1665 create files and write files.
1668 Frequency : Always on ia64 patchless client, and possibly others.
1670 Description: Add EXPORT_SYMBOL check for node_to_cpumask symbol.
1671 Details : This allows the patchless client to be loaded on architectures
1672 without this export.
1676 Description: RedHat Update kernel for RHEL5
1677 Details : Add the kernel config file for RHEL5.
1681 Description: RedHat Update kernel for RHEL5
1682 Details : Modify the kernel config file more closer RHEL5.
1686 Description: Build failure against Centos5 (RHEL5)
1687 Details : Define PAGE_SIZE when it isn't present.
1691 Description: test_7 on sanity-quota.sh can't be used on separate machines
1692 Details : add facet to handle it
1696 Description: "ll_intent_file_open()) lock enqueue: err: -13" with nfs
1697 Details : with NFS, the anon dentry's parent was set to itself in
1698 d_alloc_anon(), so in MDS, we use rec->ur_fid1 to find the
1699 corresponding dentry other than use rec->ur_name.
1701 Severity : enhancement
1703 Description: lfs setstripe enhancement
1704 Details : Make lfs setstripe understand 'k', 'm' and 'g' for stripe size.
1708 Description: enable data checksumming by default
1709 Details : enable checksum by default, allow --disable-checksum
1710 configure option and "-o nochecksum" mount option. Checksums
1711 can also be disabled at runtime via $LPROC/osc/*/checksum_pages.
1715 Description: lustre support for RHEL5
1716 Details : Add support for RHEL5.
1720 Description: Client eviction due to failover config
1721 Details : after a connection loss, the lustre client should attempt to
1722 reconnect to the last active server first before trying the
1723 other potential connections.
1727 Description: when mds and osts use different quota unit(32bit and 64bit),
1728 quota will be released repeatly.
1729 Details : void sending multiple quota reqs to mds, which will keep the
1730 status between the reqs.
1734 Description: osts not allocated evenly to files
1735 Details : change the condition to increase offset_idx
1738 Frequency : Always for filesystems larger than 2TB on 32-bit systems.
1740 Description: Data corruption for OSTs that are formatted larger than 2TB
1742 Details : When generating the bio request for lustre file writes the
1743 sector number would overflow a temporary variable before being
1744 used for the IO. The data reads correctly from Lustre (which
1745 will overflow in a similar manner) but other file data or
1746 filesystem metadata may be corrupted in some cases.
1750 Description: Tune the kernel for good SCSI performance.
1751 Details : Set the value of /sys/block/{dev}/queue/max_sectors_kb
1752 to the value of /sys/block/{dev}/queue/max_hw_sectors_kb
1757 Description: rewrite ext2-derived code in obdclass/uuid.c
1758 Details : rewrite inherited code (uuid parsing code from ext2 utils)
1759 from scratch preserving functionality.
1763 Description: "lfs find -obd UUID" prints directories
1764 Details : "lfs find -obd UUID" will return all directory names instead
1765 of just file names. It is incorrect because the directories
1766 do not reside on the OSTs.
1770 Description: use i_size_read and i_size_write in 2.6 port
1771 Details : replace inode->i_size access with i_size_read/write()
1774 Frequency : when removing large files
1776 Description: scheduling issue during removal of large Lustre files
1777 Details : Don't take the BKL in fsfilt_ext3_setattr() for 2.6 kernels.
1778 It causes scheduling issues when removing large files (17TB in the
1782 Frequency : only with liblustre clients on XT3
1784 Description: evictions taking too long
1785 Details : allow llrd to evict clients directly on OSTs
1788 Frequency : only on ppc
1790 Description: /proc/fs/lustre/devices broken on ppc
1791 Details : The patch as applied to 1.6.2 doesn't look correct for all arches.
1792 We should make sure the type of 'index' is loff_t and then cast
1793 explicitly as needed below. Do not assign an explicitly cast
1797 Frequency : only for rhel5
1799 Description: Kernel patches update for RHEL5 2.6.18-8.1.10.el5.
1800 Details : Modify the target file & which_kernel.
1802 Severity : enhancement
1804 Description: omit set fsid for export NFS
1805 Details : fix set/restore device id for avoid EMFILE error and mark lustre
1806 fs as FS_REQUIRES_DEV for avoid problems with generate fsid.
1810 Frequency : Always, for kernels after 2.6.16
1811 Description: Fix warning idr_remove called for id=.. which is not allocated.
1812 Details : Last kernels save old s_dev before kill super and not allow
1813 to restore from callback - restore it before call kill_anon_super.
1817 Description: buffer overruns could theoretically occur
1818 Details : llapi_semantic_traverse() modifies the "path" argument by
1819 appending values to the end of the origin string, and a
1820 overrun may occur. Adding buffer overrun check in liblustreapi.
1824 Description: Fix error on 'ls .' at the top of the Lustre mount.
1825 Details : Don't revalidate dentry if it is a root dentry.
1829 Description: Kernel patches update for RHEL4 2.6.9-55.0.6.
1830 Details : Modify vm-tunables-rhel4.patch.
1834 Description: Kernel config for 2.6.18-vanilla.
1835 Details : Modify targets/2.6-vanilla.target.in.
1836 Add config file kernel-2.6.18-2.6-vanilla-i686.config.
1837 Add config file kernel-2.6.18-2.6-vanilla-i686-smp.config.
1838 Add config file kernel-2.6.18-2.6-vanilla-x86_64.config.
1839 Add config file kernel-2.6.18-2.6-vanilla-x86_64-smp.config.
1843 Description: improve handling recoverable errors
1844 Details : if request processig with error which can be recoverable on server
1845 request should be resend, otherwise page released from cache and
1850 Description: Kernel patches update for RHEL5 2.6.18-8.1.14.el5.
1851 Details : Modify target file & which_patch.
1852 A flaw was found in the IA32 system call emulation provided
1853 on AMD64 and Intel 64 platforms. An improperly validated 64-bit
1854 value could be stored in the %RAX register, which could trigger an
1855 out-of-bounds system call table access. An untrusted local user
1856 could exploit this flaw to run code in the kernel
1857 (ie a root privilege escalation). (CVE-2007-4573).
1861 Description: O_DIRECT bypasses client statistics.
1862 Details : When running with O_DIRECT I/O, neither the client rpc_stats nor
1863 read_ahead_stats were updated. copied the stats section from
1864 osc_send_oap_rpc() into async_internal().
1868 Description: Add jbd statistics patch for RHEL5 and 2.6.18-vanilla
1872 Description: change order of libsysio includes
1873 Details : '#include sysio.h' should always come before '#include xtio.h'
1877 Description: handle "serious error: objid * already exists" more gracefully
1878 Details : If LAST_ID value on disk is smaller than the objects existing in
1879 the O/0/d* directories, it indicates disk corruption and causes an
1880 LBUG(). If the object is 0-length, then we should use the existing
1881 object. This will help to avoid a full fsck in most cases.
1883 Severity : enhancement
1885 Description: adapt the lustre_config script to support the upgrade case
1886 Details : Add "-u" option for lustre_config script to support upgrading 1.4
1887 server targets to 1.6 in parallel.
1891 Description: To avoid grant space > avaible space when the disk is almost
1892 full. Without this patch you might see the error "grant XXXX >
1893 available" or some LBUG about grant, when the disk is almost
1895 Details : In filter_check_grant, for non_grant cache write, we should
1896 check the left space by if (*left > ungranted + bytes), instead
1897 of (*left > ungranted), because only we are sure the left space
1898 is enough for another "bytes", then the ungrant space should be
1899 increase. In client, we should update cl_avail_grant only there
1900 is OBD_MD_FLGRANT in the reply.
1904 Description: Update RHEL 4 kernel to fix local root privilege escalation.
1905 Details : Update to the latest RHEL 4 kernel to fix the vulnerability
1906 described in CVE-2007-4573. This problem could allow untrusted
1907 local users to gain root access.
1910 Frequency : when using O_DIRECT and quotas
1912 Description: Incorrect file ownership on O_DIRECT output files
1913 Details : block usage reported by 'lfs quota' does not take into account
1914 files that have been written with O_DIRECT.
1919 Description: touch file failed when fs is not full
1920 Details : OST in recovery should not be discarded by MDS in alloc_qos(),
1921 otherwise we can get ENOSP while fs is not full.
1925 Description: parallel lock callbacks
1926 Details : Instead of sending blocking and completion callbacks as separated
1927 requests, adding them to a set and sending in parallel.
1930 Frequency : only for Cray XT3
1931 Bugzilla : 12829/13455
1932 Description: Changing primary group doesn't change the group lustre assigns to
1934 Details : When CRAY_XT3 is defined, the fsgid supplied by the client is
1935 overridden with the primary group provided by the group upcall,
1936 whereas the supplied fsgid can be trusted if it is in the list of
1937 supplementary groups returned by the group upcall.
1939 Severity : enhancement
1941 Description: Allow masking D_WARNING, D_ERROR messages from console
1942 Details : Console messages can now be disabled via lnet.printk.
1946 Description: User code with malformed file open parameter crashes client node
1947 Details : Before packing join_file req, all the related reference should be
1948 checked carefully in case some malformed flags cause fake
1949 join_file req on client.
1953 Description: LDLM_ENQUEUE races with LDLM_CP_CALLBACK
1954 Details : ldlm_completion_ast() assumes that a lock is granted when the req
1955 mode is equal to the granted mode. However, it should also check
1956 that LDLM_FL_CP_REQD is not set.
1960 Description: Heavy nfs access might result in deadlocks
1961 Details : After ELC code landed, it is now improper to enqueue any mds
1962 locks under och_sem, because enqueue might want to decide to
1963 cancel open locks for same inode we are holding och_sem for.
1967 Description: Client eviction while running blogbench
1968 Details : A lot of unlink operations with concurrent I/O can lead to a
1969 deadlock causing evictions. To address the problem, the number of
1970 oustanding OST_DESTROY requests is now throttled to
1971 max_rpcs_in_flight per OSC and LDLM_FL_DISCARD_DATA blocking
1972 callbacks are processed in priority.
1976 Description: enable ACLs on MDS by default
1977 Details : ACLs must be enabled on MDS by default.
1980 Frequency : PPC/PPC64 only
1982 Description: conflicts between asm-ppc64/types.h and lustre_types.h
1983 Details : fix duplicated definitions between asm-ppc64/types.h and
1984 lustre_types.h on PPC.
1987 Frequency : PPC/PPC64 only
1989 Description: asm-ppc/segment.h does not exist
1990 Details : fix compile issue on PPC.
1994 Description: better handle error messages in extents code
1997 Frequency : RHEL4 only
1999 Description: mkfs is very slow on IA64/RHEL4
2000 Details : A performance regression has been discovered in the MPT Fusion
2001 driver between versions 3.02.73rh and 3.02.99.00rh. As a
2002 consequence, we have downgraded the MPT Fusion driver in the RHEL4
2003 kernel from 3.02.99.00 to 3.02.73 until this problem is fixed.
2005 Severity : enhancement
2007 Description: SNMP support enhancement
2008 Details : Adding total number of sampled request for an MDS node in snmp
2011 Severity : enhancement
2013 Description: Optimize ldlm waiting list processing for PR extent locks
2014 Details : When processing waiting list for read extent lock and meeting read
2015 lock that is same or wider to it that is not contended, skip
2016 processing rest of the list and immediatelly return current
2017 status of conflictness, since we are guaranteed there are no
2018 conflicting locks in the rest of the list.
2022 Description: Time out and refuse to reconnect
2023 Details : When the failover node is the primary node, it is possible
2024 to have two identical connections in imp_conn_list. We must
2025 compare not conn's pointers but NIDs, otherwise we can defeat
2026 connection throttling.
2030 Description: port llog fixes from b1_6 into HEAD
2031 Details : Port llog reference couting and some llog cleanups from b1_6
2032 (bug 10800) into HEAD, for protect from panic and access to already
2033 free llog structures.
2037 Description: Detect stride IO mode in read-ahead
2038 Details : When a client does stride read, read-ahead should detect that and
2039 read-ahead pages according to the detected stride pattern.
2043 Description: data checksumming impacts single node performance
2044 Details : add support for several checksum algorithm. Currently, only CRC32
2045 and Adler-32 are supported. The checksum type can be changed on
2046 the fly via /proc/fs/lustre/osc/*/checksum_type.
2050 Description: use adler32 for page checksums
2051 Details : when available, use the Adler-32 algorithm instead of CRC32 for
2056 Description: build for x2 fails
2057 Details : fix compile issue on Cray systems.
2061 Description: Properly match for duplicate locks
2062 Details : Due to different lock order from skiplists code, we need to
2063 traverse entire list for now
2066 Frequency : only on PPC/SLES10
2068 Description: "BITS_PER_LONG is not 32 or 64" in linux/idr.h
2069 Details : On SLES10/PPC, fs.h includes idr.h which requires BITS_PER_LONG to
2070 be defined. Add a hack in mkfs_lustre.c to work around this compile
2075 Description: LASSERT on MDS when client holding flock lock dies
2076 Details : ldlm pool logic depends on number of granted locks equal to
2077 number of released locks which is not true for flock locks, so
2078 just exclude such locks from consideration.
2082 Description: MDS deadlock with many ll_sync_lov threads and I/O stalled
2083 Details : Use fsfilt_sync() for both the whole filesystem sync and
2084 individual file sync to eliminate dangerous inode locking
2085 with I_LOCK that can lead to a deadlock.
2089 Description: performance in 1.6.3
2090 Details : Force q->max_phys_segments to MAX_PHYS_SEGMENTS on SLES10 to be
2091 sure that 1MB requests are not fragmented by the block layer.
2093 Severity : enhancement
2095 Description: organize the server-side client stats on per-nid basis
2096 Details : Change the structure of stats under obdfilter and mds to
2104 The "uuid"s file would list the uuids of _active_ exports.
2105 And the clear entry is to clear all stats and stale nids.
2107 Severity : enhancement
2109 Description: eliminate client locks in face of contention
2110 Details : file contention detection and lockless i/o implementation
2111 for contended files.
2115 Description: Reinitialize optind to 0 so that interactive lfs works in all cases
2118 Frequency : very rare, if additional xattrs are used on kernels >= 2.6.12
2120 Description: MDS may lose file striping (and hence file data) in some cases
2121 Details : If there are additional extended attributes stored on the MDS,
2122 in particular ACLs, SELinux, or user attributes (if user_xattr
2123 is specified for the client mount options) then there is a risk
2124 of attribute loss. Additionally, the Lustre file striping
2125 needs to be larger than default (e.g. striped over all OSTs),
2126 and an additional attribute must be stored initially in the
2127 inode and then increase in size enough to be moved to the
2128 external attribute block (e.g. ACL growing in size) for file
2133 Description: skiplist implementation simplification
2134 Details : skiplists are used to group compatible locks on granted list
2135 that was implemented as tracking first and last lock of each
2136 lock group the patch changes that to using doubly linked lists
2140 Description: MDS LBUG: ASSERTION(!IS_ERR(dchild))
2141 Details : Change LASSERTs to client eviction (i.e. abort client's recovery)
2142 because LASSERT on both the data supplied by a client, and the
2143 data on disk is dangerous and incorrect.
2145 Severity : enhancement
2147 Description: Slow truncate/writes to huge files at high offsets.
2148 Details : Directly associate cached pages to lock that protect those pages,
2149 this allows us to quickly find what pages to write and remove
2150 once lock callback is received.
2154 Description: more ldlm soft lockups
2155 Details : In ldlm_resource_add_lock(), call to ldlm_resource_dump()
2156 starve other threads from the resource lock for a long time in
2157 case of long waiting queue, so change the debug level from
2158 D_OTHER to the less frequently used D_INFO.
2160 Severity : enhancement
2162 Description: add -gid, -group, -uid, -user options to lfs find
2166 Description: Hung threads in invalidate_inode_pages2_range
2167 Details : The direct IO path doesn't call check_rpcs to submit a new RPC
2168 once one is completed. As a result, some RPCs are stuck in the
2169 queue and are never sent.
2173 Description: filter threads hungs on waiting journal commit
2174 Details : Cleanup filter group llog code, then only filter group llog will
2175 be only created in the MDS/OST syncing process.
2179 Description: Procfs and llog threads access destoryed import sometimes.
2180 Details : Sync the import destoryed process with procfs and llog threads by
2181 the import refcount and semaphore.
2183 Severity : enhancement
2185 Description: openlock cache of b1_6 port to HEAD
2190 Description: kernel BUG at ldiskfs2_ext_new_extent_cb
2191 Details : If insertion of an extent fails, then discard the inode
2192 preallocation and free data blocks else it can lead to duplicate
2197 Description: don't always update ctime in ext3_xattr_set_handle()
2198 Details : Current xattr code updates inode ctime in ext3_xattr_set_handle.
2199 In some cases the ctime should not be updated, for example for
2200 2.0->1.8 compatibility it is necessary to delete an xattr and it
2201 should not update the ctime.
2205 Bugzilla : 15713/16362
2206 Description: Assertion in iopen_connect_dentry in 1.6.3
2207 Details : looking up an inode via iopen with the wrong generation number can
2208 populate the dcache with a disconneced dentry while the inode
2209 number is in the process of being reallocated. This causes an
2210 assertion failure in iopen since the inode's dentry list contains
2211 both a connected and disconnected dentry.
2215 Description: assertion failure in ldlm_handle2lock()
2216 Details : fix a race between class_handle_unhash() and class_handle2object()
2217 introduced in lustre 1.6.5 by bug 13622.
2222 Description: Kernel BUG: sd_iostats_bump: unexpected disk index
2223 Details : remove the limit of 256 scsi disks in the sd_iostat patch
2228 Description: oops in sd_iostats_seq_show()
2229 Details : unloading/reloading the scsi low level driver triggers a kernel
2230 bug when trying to access the sd iostat file.
2235 Description: Kernel panics during QLogic driver reload
2236 Details : REQ_BLOCK_PC requests are not handled properly in the sd iostat
2237 patch, causing memory corruption.
2242 Description: journal_dev option does not work in b1_6
2243 Details : pass mount option during pre-mount.
2245 Severity : enhancement
2247 Description: Add a FIEMAP(FIle Extent MAP) ioctl
2248 Details : FIEMAP ioctl will allow an application to efficiently fetch the
2249 extent information of a file. It can be used to map logical blocks
2250 in a file to physical blocks in the block device.
2254 Description: LDLM soft lockups - improvement
2255 Details : It is be possible to send the lock handle along with each read
2256 or write request because the client is already doing a lock match
2257 itself so there isn't any reason the OST should have to re-do that
2261 Frequency : only on Cray X2
2263 Description: X2 build failures
2264 Details : fix build failures on Cray X2.
2268 Description: xid & resent requests
2269 Details : Initialize RPC XID from clock at startup (randomly if clock is
2272 Severity : enhancement
2274 Description: Add lustre_start utility to start or stop multiple Lustre servers
2279 Description: Lustre GPF in {:ptlrpc:ptlrpc_server_free_request+373}
2280 Details : In case of memory pressure, list_del() can be called twice on
2281 req->rq_history_list, causing a kernel oops.
2285 Description: kptllnd_peer_check_sends()) ASSERTION(!in_interrupt()) failed
2286 Details : fix stack overflow in the distributed lock manager by defering
2287 export eviction after a failed AST to the elt thread instead of
2288 handling it in the dlm interpret routine.
2292 Description: Convert some comments to new format.
2293 Details : Update documenting comments to match doxygen conventions.
2297 Description: Grammar fixes.
2298 Details : A couple of trivial spelling fixes.
2302 Description: OSD_COUNTERS-mandatory
2303 Details : Make previously optional ->oti_{w,r}_locks sanity checks mandatory
2304 to simplify the code.
2308 Description: simplify cmm_device freeing logic.
2309 Details : Call cmm_device_free() in the failure path of cmm_device_alloc().
2313 Description: Add lockdep support to dt_object_operations locking interface.
2314 Details : Augment ->do_{read,write}_lock() prototypes with a `role'
2315 parameter indicating lock ordering. Update mdd code to use new
2320 Description: Introduce failloc constants for lockless IO tests.
2321 Details : Add two new failloc constants to test lockless IO. Only one of
2322 them in implemented---another is checked in yet to be landed
2327 Description: Add lockdep support for inode mutex.
2328 Details : Introduce and use new LOCK_INODE_MUTEX_PARENT() macro to be used
2329 in the situations where Lustre has to lock more than one inode
2334 Description: Add optional invariants checking support.
2335 Details : Add new LINVRNT() macro, optional on new --enable-invariants
2336 configure switch. This macro is to be used for consistency and
2337 sanity checks that are too expensive to be left in `production'
2342 Description: Zap lock->l_granted_mode with explicit LCK_MINMODE.
2343 Details : Use LCK_MINMODE rather than 0 to reset lock->l_granted_mode to
2348 Description: Add lockdep support for ldlm_lock and ldlm_resource.
2349 Details : Use spin_lock_nested() in (the only) situation where more than
2350 one ldlm_lock is locked simultaneously. Also, fix possible
2351 dead-lock in ldlm_lock_change_resource() by enforcing particular
2356 Description: Use struct ldlm_callback_suite in ldlm_lock_create().
2357 Details : Instead of specifying each ldlm_lock call-back through separate
2358 parameter, wrap them into struct ldlm_callback_suite.
2362 Description: Kill join_lru obd method and its callers.
2363 Details : CLIO uses lock weighting policy to keep locks over mmapped regions
2364 in memory---a requirement implemented through ->o_join_lru() obd
2365 method in HEAD. Remove this method and its users.
2369 Description: Add asynchronous ldlm ENQUEUE completion handler.
2370 Details : CLIO posts ENQUEUE requests asynchronously through ptlrpcd---a
2371 case that stock ldlm_completion_ast() cannot handle as it waits
2372 until lock is granted. Introduce new ldlm_completion_ast_async()
2373 for this. Also comment ldlm_completion_ast().
2377 Description: ldlm_error <-> errno conversion.
2378 Details : Add functions to map (rather arbitrary) between LDLM error codes
2379 and standard errno values. CLIO needs this to prevent LDLM specific
2380 constants from escaping ldlm and osc.
2384 Description: Kill unused ldlm_handle2lock_ns() function.
2385 Details : Kill unused ldlm_handle2lock_ns() function.
2389 Description: Add lu_ref support to ldlm_lock
2390 Details : lu_ref support for ldlm_lock and ldlm_resource. See lu_ref patch.
2391 lu_ref fields ->l_reference and ->lr_reference are added to ldlm_lock
2392 and ldlm_resource. LDLM interface has to be changed, because code that
2393 releases a reference on a lock, has to "know" what reference this is.
2394 In the most frequent case
2396 lock = ldlm_handle2lock(handle);
2398 LDLM_LOCK_PUT(lock);
2400 no changes are required. When any other reference (received _not_ from
2401 ldlm_handle2lock()) is released, LDLM_LOCK_RELEASE() has to be called
2402 instead of LDLM_LOCK_PUT().
2404 Arguably, changes are pervasive, and interface requires some discipline
2405 for proper use. On the other hand, it was very instrumental in finding
2406 a few leaked lock references.
2410 Description: Add ldlm_lock_addref_try().
2411 Details : Introduce ldlm_lock_addref_try() function (used by CLIO) that
2412 attempts to addref a lock that might be being canceled
2417 Description: Add ldlm_weigh_callback().
2418 Details : Add new ->l_weigh_ast() call-back to ldlm_lock. It is called
2419 by ldlm_cancel_shrink_policy() to estimate lock "value", instead of
2420 hard-coded `number of pages' logic.
2424 Description: Add lockdep annotations to llog code.
2425 Details : Use appropriately tagged _nested() locking calls in the places
2426 where llog takes more than one ->lgh_lock lock.
2430 Description: Add loi_kms_set().
2431 Details : Wrap kms updates into a helper function.
2435 Description: Constify instances of struct lsm_operations.
2436 Details : Constify instances of struct lsm_operations.
2440 Description: lu_conf support.
2441 Details : On a server, a file system object is uniquely identified
2442 by a fid, which is sufficient to locate and load all object
2443 state (inode). On a client, on the other hand, more data are
2444 necessary instantiate an object. Change lu_object_find() and
2445 friends to take additional `lu_conf' argument describing object.
2446 Typically this includes layout information.
2450 Description: lu_context fixes.
2451 Details : Introduce new lu_context functions that are needed on the client
2452 side, where some system threads (ptlrpcd) are shared by multiple
2453 modules, and so cannot be stopped during module shutdown.
2457 Description: Add start and stop methods to lu_device_type_operations.
2458 Details : Introduce two new methods in lu_device_type_operations, that are
2459 invoked when first instance of a given type is created and last one
2460 is destroyed respectively. This is need by CLIO.
2464 Description: Add lu_ref support to struct lu_device.
2465 Details : Add lu_ref support to lu_object and lu_device. lu_ref is used to
2466 track leaked references.
2470 Description: Introduce lu_kmem_descr.
2471 Details : lu_kmem_descr and its companion interface allow to create
2472 and destroy a number of kmem caches at once.
2476 Description: Fix lu_object finalization race.
2477 Details : Fix a race between lu_object_find() finding an object and its
2478 concurrent finalization. This race is (most likely) not possible
2479 on the server, but might happen on the client.
2483 Description: Introduce lu_ref interface.
2484 Details : lu_ref is a debugging module allowing to track references to
2485 a given object. It is quite cpu expensive, and has to be
2486 explicitly enabled with --enable-lu_ref. See usage description
2491 Description: Factor lu_site procfs stats into a separate function.
2492 Details : Separate lu_site stats printing code into a separate function
2493 that can be reused on a client.
2497 Description: Constify instances of struct {lu,dt,md}_device_operations.
2498 Details : Constify instances of struct {lu,dt,md}_device_operations.
2502 Description: Introduce struct md_site and move meta-data specific parts of
2503 struct lu_site here.
2504 Details : Move md-specific fields out of struct lu_site into special struct
2505 md_site, so that lu_site can be used on a client.
2509 Description: Kill mdd_lov_destroy().
2510 Details : Remove unused mdd code.
2514 Description: Add st_block checking to multistat.c.
2515 Details : Add st_block checking to multistat.c.
2519 Description: Add lu_ref support to struct obd_device.
2520 Details : Add lu_ref tracking to obd_device.
2524 Description: Kill obd_set_fail_loc().
2525 Details : Remove unused code.
2529 Description: Add special type for ptlrpc_request interpret functions.
2530 Details : Add lu_env parameter to ->rq_interpreter call-back. NULL is passed
2531 there. Actual usage will be in CLIO.
2535 Description: Replace RW_LOCK_UNLOCKED() macro with rwlock_init().
2536 Details : Replace RW_LOCK_UNLOCKED() with rwlock_init() as the former
2537 doesn't work with lockdep.
2541 Description: Add rwv.c test program.
2542 Details : New testing program exercising readv(2) and writev(2) (Qian).
2546 Description: Add sendfile.c test program.
2547 Details : New testing program exercising sendfile(2) (Jay).
2551 Description: Ratelimit a message that can be very frequent.
2552 Details : Ratelimit a memory allocation failure message that can
2557 Description: Use cdebug_show() in CDEBUG-style macros defined outside of libcfs.
2558 Details : Use cdebug_show() in CDEBUG-style macros defined outside of libcfs.
2562 Description: Liblustre build fixes.
2563 Details : Liblustre build fixes.
2567 Description: libcfs: add cfs_{need,cond}_resched() interface.
2568 Details : libcfs: add cfs_{need,cond}_resched() definition and
2569 implementations for Linux, NT, and liblustre.
2571 Severity : enhancement
2573 Description: More exported tunables for mballoc
2574 Details : Add support for tunable preallocation window and new tunables for
2575 large/small requests
2579 Description: Detect corruption of block bitmap and checking for preallocations
2580 Details : Checks validity of on-disk block bitmap. Also it does better
2581 checking of number of applied preallocations. When corruption is
2582 found, it turns filesystem readonly to prevent further corruptions.
2586 Description: (rw.c:1323:ll_read_ahead_pages()) ASSERTION(page_idx > ria->ria_stoff) failed
2587 Details : Once the unmatched stride IO mode is detected, shrink the stride-ahead
2588 window to 0. If it does hit cache miss, and read-pattern is still
2589 stride-io mode, does not reset the stride window, but also does not
2590 increase the stride window length in this case.
2594 Frequency : only for big-endian servers
2595 Description: Check if system is big-endian while mounting fs with extents feature
2596 Details : Mounting a filesystem with extents feature will fail on big-endian
2597 systems since ext3-based ldiskfs is not supported on big-endian
2598 systems. This can be over-riden with "bigendian_extents" mount option.
2600 Severity : enhancement
2602 Description: The root squash functionality
2603 Details : A security feature, which is to prevent users from being able
2604 to mount lustre on their desktop, run as root, and delete
2605 all of the files in the filesystem. The goal is accomplished by
2606 remapping user id (UID) and group id (GID) of the root user to
2607 a UID and GID specified by the system administartor via Lustre
2608 configuration management server (MGS). The functionality also
2609 allows to specify sets of clients for which the remapping does
2614 Description: Excessive recovery window
2615 Details : With AT enabled, the recovery window can be excessively long (6000+
2616 seconds). To address this problem, we no longer use
2617 OBD_RECOVERY_FACTOR when extending the recovery window (the connect
2618 timeout no longer depends on the service time, it is set to
2619 INITIAL_CONNECT_TIMEOUT now) and clients report the old service
2620 time via pb_service_time.
2624 Description: Watchdog triggered on MDS failover
2625 Details : enable OBD_CONNECT_MDT flag when connecting from the MDS so that
2626 the OSTs know that the MDS "UUID" can be reused for the same export
2627 from a different NID, so we do not need to wait for the export to be
2631 Frequency : rare, only if using MMP with Linux RAID
2633 Description: MMP doesn't work with Linux RAID
2634 Details : While using HA for Lustre servers with Linux RAID, it is possible
2635 that MMP will not detect multiple mounts. To make this work we
2636 need to unplug the device queue in RAID when the MMP block is being
2637 written. Also while reading the MMP block, we should read it from
2638 disk and not the cached one.
2640 Severity : enhancement
2642 Description: open file using fid
2643 Details : A file can be opened using just its fid, like
2644 <mntpt>/.lustre/fid/SEQ:OID:VER - this is needed for HSM and replication
2647 Frequency : Only in RHEL5 when mounting multiple ext3 filesystems
2650 Description: "kmem_cache_create: duplicate cache jbd_4k" error message
2651 Details : add proper locking for creation of jbd_4k slab cache
2655 Description: MMP check in ext3_remount() fails without displaying any error
2656 Details : When multiple mount protection fails during remount, proper error
2659 Severity : enhancement
2661 Description: Allow stripe size to be up to 4G-64k
2662 Details : Fix math logic to allow large stripe sizes.
2666 Description: add check for >8TB ldiskfs filesystems
2667 Details : ext3-based ldiskfs does not support greater than 8TB LUNs.
2668 Don't allow >8TB ldiskfs filesystems to be mounted without
2669 force_over_8tb mount option
2674 Descriptoin: OSS complains "too many blocks" and crashes
2675 Details : In filter_commitrw_write():
2676 1. lquota_chkquota();
2677 2. obdo_from_inode();
2678 3. lquota_pending_commit();
2679 "2." has changed uid and gid in this case, then caused
2680 lquota_chkquota() and lquota_pending_commit() to use
2681 different uid and gid. That is the root of the bug.
2685 Descriptoin: getstripe doesn't report stripe count = -1 sanely
2686 Details : Detect and print out the 'stripe count = -1' case properly.
2690 Description: Out or order replies might be lost on replay
2691 Details : In ptlrpc_retain_replayable_request if we cannot find retained
2692 request with tid smaller then one currently being added, add it
2693 to the start, not end of the list.
2695 --------------------------------------------------------------------------------
2697 2007-08-10 Cluster File Systems, Inc. <info@clusterfs.com>
2699 * Support for kernels:
2700 2.6.5-7.283 (SLES 9),
2701 2.6.9-55.EL (RHEL 4),
2702 2.6.16.46-0.14 (SLES 10),
2703 2.6.18.8 vanilla (kernel.org)
2704 * Client support for unpatched kernels:
2705 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
2706 2.6.16 - 2.6.22 vanilla (kernel.org)
2707 * Due to recently discovered recovery problems, we do not recommend
2708 using patchless RHEL 4 clients with this or any earlier release.
2709 * Recommended e2fsprogs version: 1.39.cfs8
2710 * Note that reiserfs quotas are disabled on SLES 10 in this kernel.
2711 * Starting with this release, the ldiskfs backing filesystem required
2712 by Lustre is now in its own package, lustre-ldiskfs. This package
2713 should be installed. It is versioned separately from Lustre and
2714 may be released separately in future.
2718 Description: block reactivating mgc import until all deactivates complete
2719 Details : Fix race when failing back MDT/MGS to itself (testing)
2722 Frequency : at statup only
2724 Description: mds_lov_synchronize race leads to various problems
2725 Details : simultaneous MDT->OST connections at startup can cause the
2726 sync to abort, leaving the OSC in a bad state.
2728 Severity : enhancement
2730 Description: add optional extra BUILD_VERSION info
2731 Details : add a new environment variable (namely LUSTRE_VERS) which allows
2732 to override the lustre version.
2735 Frequency : 2.6.18 servers only
2737 Description: ll_kern_mount() doesn't release the module reference
2738 Details : The ldiskfs module reference count never drops down to 0
2739 because ll_kern_mount() doesn't release the module reference.
2744 Description: server LBUG when using old ost_num_threads parameter
2745 Details : Accept the old ost_num_threads parameter but warn that it
2746 is deprecated, and fix an off-by-one error that caused an LBUG.
2751 Description: Transient SCSI error results in persistent IO issue
2752 Details : iobuf->dr_error is not reinitialized to 0 between two
2756 Frequency : sometimes when underlying device returns I/O errors
2758 Description: OSTs not going read-only during write failures
2759 Details : OSTs are not remounted read-only when the journal commit threads
2760 get I/O errors because fsfilt_ext3 calls journal_start/stop()
2761 instead of the ext3 wrappers.
2765 Description: poor connect scaling with increasing client count
2766 Details : Don't run filter_grant_sanity_check for more than 100 exports
2767 to improve scaling for large numbers of clients.
2770 Frequency : SLES10 only
2772 Description: sanity-quota.sh quotacheck failed: rc = -22
2773 Details : Quotas cannot be enabled on SLES10.
2776 Frequency : liblustre clients only
2778 Description: getdirentries does not give error when run on compute nodes
2779 Details : getdirentries does not fail when the size specified as an argument
2780 is too small to contain at least one entry
2782 Severity : enhancement
2784 Description: Add LNET router traceability for debug purposes
2785 Details : If a checksum failure occurs with a router as part of the
2786 IO path, the NID of the last router that forwarded the bulk data
2787 is printed so it can be identified.
2792 Description: OST "spontaneously" evicts client; client has imp_pingable == 0
2793 Details : Due to a race condition, liblustre clients were occasionally
2794 evicted incorrectly.
2796 Severity : enhancement
2798 Description: lfs setstripe use optional parameters instead of postional
2801 Severity : enhancement
2803 Description: Nanosecond timestamp support for ldiskfs
2804 Details : The on-disk ldiskfs filesystem has added support for nanosecond
2805 resolution timestamps. There is not yet support for this at
2806 the Lustre filesystem level.
2809 Frequency : during server recovery
2811 Description: MDS failing to send precreate requests due to OSCC_FLAG_RECOVERING
2812 Details : request with rq_no_resend flag not awake l_wait_event if they get
2816 Frequency : nfs export on patchless client
2818 Description: connectathon hang when test nfs export over patchless client
2819 Details : Disconnected dentry cannot be found with lookup, so we do not need
2820 to unhash it or make it invalid
2823 Description: fix llapi_lov_get_uuids() to allow many OSTs to be returned
2824 Details: : Change llapi_lov_get_uuids() to read the UUIDs from /proc instead
2825 of using an ioctl. This allows lfsck for > 160 OSTs to succeed.
2830 Description: open req refcounting wrong on reconnect
2831 Details : If reconnect happened between getting open reply from server and
2832 call to mdc_set_replay_data in ll_file_open, we will schedule
2833 replay for unreferenced request that we are about to free.
2834 Subsequent close will crash in variety of ways.
2835 Check that request is still eligible for replay in
2836 mdc_set_replay_data().
2841 Description: disable writes to filesystem when reading health_check file
2842 Details : the default for reading the health_check proc file has changed
2843 to NOT do a journal transaction and write to disk, because this
2844 can cause reads of the /proc file to hang and block HA state
2845 checking on a healthy but otherwise heavily loaded system. It
2846 is possible to return to the previous behaviour during configure
2847 with --enable-health-write.
2849 Severity : enhancement
2851 Description: 64-bit inode version
2852 Details: : Add a on-disk 64-bit inode version for ext3 to track changes made
2853 to the inode. This will be required for version-based recovery.
2858 Description: MDS fails to start if a duplicate client export is detected
2859 Details : in some rare cases it was possible for a client to connect to
2860 an MDS multiple times. Upon recovery the MDS would detect this
2861 and fail during startup. Handle this more gracefully.
2863 Severity : enhancement
2865 Description: Add -o localflock option to simulate old noflock
2867 Details : This will achieve local-only flock/fcntl locks
2873 Description: log_commit_thread vs filter_destroy race leads to crash
2874 Details : Take import reference before releasing llog record semaphore
2879 Description: Wrong request locking in request set processing
2880 Details : ptlrpc_check_set wrongly uses req->rq_lock for proctect add to
2881 imp_delayed_list, in this place should be used imp_lock.
2884 Frequency : when reconnection
2886 Description: Grant Leak when osc reconnect to OST
2887 Details : When osc reconnect ost, OST(filter) should check whether it
2888 should grant more space to client by comparing fed_grant and
2889 cl_avail_grant, and return the granted space to client instead
2890 of "new granted" space, because client will call osc_init_grant
2891 to update the client grant space info.
2894 Frequency : when client reconnect to OST
2896 Description: Grant Leak when osc do resend and replay bulk write
2897 Details : When osc reconnect to OST, OST(filter)should clear grant info of
2898 bulk write request, because the grant info while be sync between
2899 OSC and OST when reconnect, and we should ignore the grant info
2900 these of resend/replay write req.
2905 Description: Grant space more than avaiable left space sometimes.
2906 Details : When then OST is about to be full, if two bulk writing from
2907 different clients came to OST. Accord the avaliable space of the
2908 OST, the first req should be permitted, and the second one
2909 should be denied by ENOSPC. But if the seconde arrived before
2910 the first one is commited. The OST might wrongly permit second
2911 writing, which will cause grant space > avaiable space.
2914 Frequency : when client is evicted
2916 Description: Grant might be wrongly erased when osc is evicted by OST
2917 Details : when the import is evicted by server, it will fork another
2918 thread ptlrpc_invalidate_import_thread to invalidate the
2919 import, where the grant will be set to 0. While the original
2920 thread will update the grant it got when connecting. So if
2921 the former happened latter, the grant will be wrongly errased
2922 because of this race.
2927 Description: Checking Stale with correct fid
2928 Details : ll_revalidate_it should uses de_inode instead of op_data.fid2
2929 to check whether it is stale, because sometimes, we want the
2930 enqueue happened anyway, and op_data.fid2 will not be initialized.
2932 Severity : enhancement
2934 Description: update patchless client
2935 Details : Add support for patchless client with 2.6.20, 2.6.21 and RHEL 5
2938 Frequency : only with 2.4 kernel
2940 Description: random memory corruption
2941 Details : size of struct ll_inode_info is to big for union inode.u and this
2942 can be cause of random memory corruption.
2947 Description: Memory leak in recovery
2948 Details : Lov_mds_md was not free in an error handler in mds_create_object.
2949 It should also check obd_fail before fsfilt_start, otherwise if
2950 fsfilt_start return -EROFS,(failover mds during mds recovery).
2951 then the req will return with repmsg->transno = 0 and rc = EROFS.
2952 and we met hit the assert LASSERT(req->rq_reqmsg->transno ==
2953 req->rq_repmsg->transno) in ptlrpc_replay_interpret. Fcc should
2954 be freed no matter whether fsfilt_commit success or not.
2957 Frequency : only with huge count clients
2959 Description: Prevents from taking the superblock lock in llap_from_page for
2961 Details : using LL_ORIGIN_REMOVEPAGE origin flag instead of LL_ORIGIN_UNKNOW
2962 for llap_from_page call in ll_removepage prevents from taking the
2963 superblock lock for a soon died page.
2968 Description: Not check open intent error before release open handle
2969 Details : in some rare cases, the open intent error is not checked before
2970 release open handle, which may cause
2971 ASSERTION(open_req->rq_transno != 0), because it tries to release
2972 the failed open handle.
2977 Description: Set cat log bitmap only after create log success.
2978 Details : in some rare cases, the cat log bitmap is set too early. and it
2979 should be set only after create log success.
2983 Description: Accessing a block bevice can re-enable I/O when Lustre is
2984 tearing down a device.
2985 Details : dev_clear_rdonly(bdev) must be called in kill_bdev() instead of
2990 Description: service threads may hog cpus when there are a lot of requests
2991 Details : Insert cond_resched to give other threads a chance to use some CPU
2996 Description: the cat log was not initialized in recovery
2997 Details : When mds(mgs) do recovery, the tgt_count might be zero, so the
2998 unlink log on mds will not be initialized until mds post
2999 recovery. And also in mds post recovery, the unlink log will
3000 initialization will be done asynchronausly, so there will be race
3001 between add unlink log and unlink log initialization.
3005 Description: brw_stats were being printed incorrectly
3006 Details : brw_stats were being printed as log2 but all of them were not
3007 recorded as log2. Also remove some code duplication arising from
3008 filter_tally_{read,write}.
3012 Frequency : rare, only in recovery.
3013 Description: ASSERTION(req->rq_type != LI_POISON) failed
3014 Details : imp_lock should be held while iterating over imp_sending_list for
3015 prevent destroy request after get timeout in ptlrpc_queue_wait.
3019 Description: replay-single.sh test 52 fails
3020 Details : A lock's skiplist need to be cleanup when it being unlinked
3021 from its resource list.
3025 Description: Short directio read returns full requested size rather than
3027 Details : Direct I/O operations should return actual amount of bytes
3028 transferred rather than requested size.
3030 Severity : enhancement
3032 Description: metadata RPC reduction (e.g. for rm performance)
3033 Details : decrease the amount of synchronous RPC between clients and servers
3034 by canceling conflicing lock before the operation on the client
3035 and packing thier handles into the main operation RPC to server.
3037 Severity : enhancement
3039 Description: Async OSC create to avoid the blocking unnecessarily.
3040 Details : If a OST has no remain object, system will block on the creating
3041 when need to create a new object on this OST. Now, ways use
3042 pre-created objects when available, instead of blocking on an
3043 empty osc while others are not empty. If we must block, we block
3044 for the shortest possible period of time.
3048 Description: improve handling recoverable errors
3049 Details : if request processig with error which can be recoverable on server
3050 request should be resend, otherwise page released from cache and
3053 Severity : enhancement
3055 Description: refine locking for avoid write wrong info into lov_objid file
3056 Details : fix possible races with add new target and write/update data in
3059 --------------------------------------------------------------------------------
3061 2007-05-03 Cluster File Systems, Inc. <info@clusterfs.com>
3066 Frequency : on some architectures
3068 Description: 1.6 client sometimes fails to mount from a 1.4 MDT
3069 Details : Uninitialized flags sometimes cause configuration commands to
3073 Frequency : patchless clients only
3075 Description: missing __iget() symbol export
3076 Details : The __iget() symbol export is missing. To avoid the need for
3077 this on patchless clients the deathrow inode reaper is turned
3078 off, and we depend on the VM to clean up old inodes. This
3079 dependency was during via the fix for bug 12181.
3081 --------------------------------------------------------------------------------
3083 2007-04-19 Cluster File Systems, Inc. <info@clusterfs.com>
3085 * CONFIGURATION CHANGE. This version of Lustre WILL NOT
3086 INTEROPERATE with older versions automatically. In many cases a
3087 special upgrade step is needed. Please read the
3088 user documentation before upgrading any part of a live system.
3089 * WIRE PROTOCOL CHANGE from previous 1.6 beta versions. This
3090 version will not interoperate with 1.6 betas before beta5 (1.5.95).
3091 * WARNING: Lustre configuration and startup changes are required with
3092 this release. See https://mail.clusterfs.com/wikis/lustre/MountConf
3096 Severity : enhancement
3098 Description: MountConf
3099 Details : Lustre configuration is now managed via mkfs and mount
3100 commands instead of lmc and lconf. New obd types (MGS, MGC)
3101 are added for dynamic configuration management. See
3102 https://mail.clusterfs.com/wikis/lustre/MountConf for
3105 Severity : enhancement
3107 Description: dynamic OST addition
3108 Details : OSTs can now be added to a live filesystem
3110 Severity : enhancement
3112 Description: startup order invariance
3113 Details : MDTs and OSTs can be started in any order. Clients only
3114 require the MDT to complete startup.
3116 Severity : enhancement
3118 Description: parallel, asynchronous orphan cleanup
3119 Details : orphan cleanup is now performed in separate threads for each
3120 OST, allowing parallel non-blocking operation.
3122 Severity : enhancement
3124 Description: optimized stripe assignment
3125 Details : stripe assignments are now made based on ost space available,
3126 ost previous usage, and OSS previous usage, in order to try
3127 to optimize storage space and networking resources.
3129 Severity : enhancement
3131 Description: Permanently set tunables
3132 Details : All writable /proc/fs/lustre tunables can now be permanently
3133 set on a per-server basis, at mkfs time or on a live system.
3135 Severity : enhancement
3137 Description: Lustre message v2
3138 Details : Add lustre message format v2.
3140 Severity : enhancement
3142 Description: client OST exclusion list
3143 Details : Clients can be started with a list of OSTs that should be
3144 declared "inactive" for known non-responsive OSTs.
3148 Description: SPEC SFS validation failure on NFS v2 over lustre.
3149 Details : Changes the blocksize for regular files to be 2x RPC size,
3150 and not depend on stripe size.
3152 Severity : enhancement
3154 Description: Multiple MD RPCs in flight.
3155 Details : Further unserialise some read-only MDS RPCs - learn about intents.
3156 To avoid overly-overloading MDS, introduce a limit on number of
3157 MDS RPCs in flight for a single client and add /proc controls
3158 to adjust this limit.
3160 Severity : enhancement
3162 Description: client read/write statistics
3163 Details : Add client read/write call usage stats for performance
3164 analysis of user processes.
3165 /proc/fs/lustre/llite/*/offset_stats shows non-sequential
3166 file access. extents_stats shows chunk size distribution.
3167 extents_stats_per_process show chunk size distribution per
3170 Severity : enhancement
3172 Description: mds statistics
3173 Details : Add detailed mds operations statistics in
3174 /proc/fs/lustre/mds/*/stats.
3178 Description: Failure of copying files with lustre special EAs.
3179 Details : Client side always return success for setxattr call for lustre
3180 special xattr (currently only "trusted.lov").
3185 Description: Refcount LNET uuids
3186 Details : The global LNET uuid list grew linearly with every startup;
3187 refcount repeated list entries instead of always adding to
3190 Severity : enhancement
3192 Description: Dynamic service threads
3193 Details : Within a small range, start extra service threads
3194 automatically when the request queue builds up.
3197 Frequency : mixed-endian client/server environments
3199 Description: mixed-endian crashes
3200 Details : The new msg_v2 system had some failures in mixed-endian
3203 Severity : enhancement
3205 Description: Easy OST removal
3206 Details : OSTs can be permanently deactivated with e.g. 'lctl
3207 conf_param lustre-OST0001.osc.active=0'
3209 Severity : enhancement
3211 Description: MGS proc entries
3212 Details : Added basic proc entries for the MGS showing what filesystems
3215 Severity : enhancement
3217 Description: provide MGS failover
3218 Details : Added config lock reacquisition after MGS server failover.
3220 Severity : enhancement
3222 Description: add Linux 2.4 support
3223 Details : Added support for RHEL 2.4.21 kernel for 1.6 servers and clients
3227 Description: a large application tries to do I/O to the same resource and dies
3228 in the middle of it.
3229 Details : Check the req->rq_arrival time after the call to
3230 ost_brw_lock_get(), but before we do anything about
3231 processing it & sending the BULK transfer request. This
3232 should help move old stale pending locks off the queue as
3233 quickly as obd_timeout.
3236 Frequency : when an incorrect nid is specified during startup
3238 Description: ptlrpc connect to non-existant node causes kernel crash
3239 Details : LNET can't be re-entered from an event callback, which
3240 happened when we expire a message after the export has been
3241 cleaned up. Instead, hand the zombie cleanup off to another
3244 Severity : enhancement
3246 Description: plain/inodebits lock performance improvement
3247 Details : Grouping plain/inodebits in granted list by their request modes
3248 and bits policy, thus improving the performance of search through
3252 Frequency : only if OST filesystem is corrupted
3254 Description: client incorrectly hits assertion in ptlrpc_replay_req()
3255 Details : for a short time RPCs with bulk IO are in the replay list,
3256 but replay of bulk IOs is unimplemented. If the OST filesystem
3257 is corrupted due to disk cache incoherency and then replay is
3258 started it is possible to trip an assertion. Avoid putting
3259 committed RPCs into the replay list at all to avoid this issue.
3262 Frequency : liblustre (e.g. catamount) on a large cluster with >= 8 OSTs/OSS
3264 Description: System hang on startup
3265 Details : This bug allowed the liblustre (e.g. catamount) client to
3266 return to the app before handling all startup RPCs. This
3267 could leave the node unresponsive to lustre network traffic
3268 and manifested as a server ptllnd timeout.
3270 Severity : enhancement
3272 Description: Add "/proc/sys/lustre/debug_peer_on_timeout"
3273 Details : liblustre envirable: LIBLUSTRE_DEBUG_PEER_ON_TIMEOUT
3274 boolean to control whether to print peer debug info when a
3275 client's RPC times out.
3278 Frequency : only for kernels with patches from Lustre below 1.4.3
3280 Description: Remove old rdonly API
3281 Details : Remove old rdonly API which unused from at least lustre 1.4.3
3284 Frequency : only for devices with external journals
3286 Description: Set external device read-only also
3287 Details : During a commanded failover stop, we set the disk device
3288 read-only while the server shuts down. We now also set any
3289 external journal device read-only at the same time.
3292 Frequency : when upgrading from 1.4 while trying to change parameters
3294 Description: The wrong (new) MDC name was used when setting parameters for
3295 upgraded MDT's. Also allows changing of OSC (and MDC)
3296 parameters if --writeconf is specified at tunefs upgrade time.
3299 Frequency : when setting specific ost indicies
3301 Description: QOS code breaks on skipped indicies
3302 Details : Add checks for missing OST indicies in the QOS code, so OSTs
3303 created with --index need not be sequential.
3305 Severity : enhancement
3307 Description: Add uninit_groups feature to ldiskfs2 to speed up e2fsck
3308 Details : The uninit_groups feature works in conjunction with the kernel
3309 filesystem code (ldiskfs2 only) and e2fsprogs-1.39-cfs6 to speed
3310 up the pass1 processing of e2fsck. This is a read-only feature
3311 in ldiskfs2 only, so older kernels and current ldiskfs cannot
3312 mount filesystems that have had this feature enabled.
3314 Severity : enhancement
3316 Description: Improve multi-block allocation algorithm to avoid fragmentation
3317 Details : The mballoc3 code (ldiskfs2 only) adds new mechanisms to improve
3318 allocation locality and avoid filesystem fragmentation.
3320 ------------------------------------------------------------------------------
3322 2007-04-01 Cluster File Systems, Inc. <info@clusterfs.com>
3324 * Support for kernels:
3325 2.4.21-47.0.1.EL (RHEL 3)
3326 2.6.5-7.283 (SLES 9)
3327 2.6.9-42.0.10.EL (RHEL 4)
3328 2.6.12.6 vanilla (kernel.org)
3329 2.6.16.27-0.9 (SLES 10)
3330 * Recommended e2fsprogs version: 1.39.cfs5
3332 * Note that reiserfs quotas are disabled on SLES 10 in this kernel
3336 Frequency : occasional, depends on client load and configuration
3337 Bugzilla : 12181, 12203
3338 Description: data loss for recently-modified files
3340 Details : In some cases it is possible that recently written or created
3341 files may not be written to disk in a timely manner (this should
3342 normally be within 30s unless client IO load is very high).
3343 The problem appears as zero-length files or files that are a
3344 multiple of 1MB in size after a client crash or client eviction
3345 that are missing data at the end of the file.
3347 This problem is more likely to be hit on clients where files are
3348 repeatedly created and unlinked in the same directory, clients
3349 have a large amount of RAM, have many CPUs, the filesystem has
3350 many OSTs, the clients are rebooted frequently, and/or the files
3351 are not accessed by other nodes after being written.
3353 The presence of the problem can be detected by looking at
3354 /proc/sys/fs/inode-state. If the first number (nr_inodes) is
3355 smaller than the second (nr_unused) then dirty files will not
3356 be flushed automatically to disk. "sync; sleep 10" should be
3357 run several times on the node before unmounting it to update
3358 Lustre (this is also safe to run on nodes without this problem).
3360 There is also a related kernel bug in the RHEL4 4 2.6.9 kernel
3361 that can cause this same problem, so customers using that kernel
3362 also need to update the kernel in addition to Lustre. In order
3363 to properly fix this bug, the RHEL3 2.4.21 kernel is also updated.
3365 It is normal that files written just before a client crash (less
3366 than 30s) may not yet have been flushed to disk, even for local
3370 Frequency : frequent on thin XT3 nodes
3372 Description: UUID collision on thin XT3 Linux nodes
3373 Details : UUIDs on Compute Node Linux XT3 nodes were not generated
3374 randomly, since we relied on an insufficiently-seeded PRNG.
3379 Description: OSS hangs after "All ost request buffers busy"
3380 Details : A deadlock between quota and journal operations caused OSS
3381 hangs after printing "All ost request buffers busy."
3384 Frequency : always on liblustre builds
3386 Description: Cleanup compiler warnings on liblustre
3389 Frequency : always on liblustre builds on XT3
3391 Description: LC_CONFIG_CDEBUG don't run while build liblustre on XT3.
3395 Description: Addition of EXT3_FEATURE_RO_COMPAT_DIR_NLINKS flag for
3396 > 32000 subdirectories
3397 Details : Add EXT3_FEATURE_RO_COMPAT_DIR_NLINK flag to
3398 EXT3_FEATURE_RO_COMPAT_SUPP. This flag will be set whenever
3399 subdirectory count crosses 32000. This will aid e2fsck to
3400 correctly handle more than 32000 subdirectories.
3403 Frequency : liblustre (e.g. catamount) on a large cluster with >= 8 OSTs/OSS
3405 Description: System hang on startup
3406 Details : This bug allowed the liblustre (e.g. catamount) client to
3407 return to the app before handling all startup RPCs. This
3408 could leave the node unresponsive to lustre network traffic
3409 and manifested as a server ptllnd timeout.
3411 Severity : enhancement
3413 Description: Add "/proc/sys/lustre/debug_peer_on_timeout"
3414 (liblustre envirable: LIBLUSTRE_DEBUG_PEER_ON_TIMEOUT)
3415 boolean to control whether to print peer debug info when a
3416 client's RPC times out.
3421 Description: make O_SYNC working on 2.6 kernels
3422 Details : 2.6 kernels use different method for mark pages for write,
3423 so need add a code to lustre for O_SYNC work.
3428 Description: Failure to close file and release space on NFS
3429 Details : Put inode details into lock acquired in ll_intent_file_open.
3430 Use mdc_intent_lock in ll_intent_open to properly
3431 detect all kind of errors unhandled by mdc_enqueue.
3436 Description: proc file read during shutdown sometimes raced obd removal,
3438 Details : Add lock to prevent obd access after proc file removal.
3441 Frequency : Only for files larger than 4GB on 32-bit clients.
3443 Description: improperly doing page alignment of locks
3444 Details : Modify lustre core code to use CFS_PAGE_* defines instead of
3445 PAGE_*. Make CFS_PAGE_MASK a 64-bit mask.
3450 Description: RPCs being resent when they shouldn't be
3451 Details : Some RPCs that should not be resent are being resent. This
3452 can cause inconsistencies in the RPC state machine. Do not
3453 resend such requests.
3456 Frequency : rare, only with NFS export
3458 Description: Crash on NFS re-export node
3459 Details : under very unusual load conditions an assertion is hit in
3460 ll_intent_file_open()
3463 Frequency : only if OST filesystem is corrupted
3465 Description: client incorrectly hits assertion in ptlrpc_replay_req()
3466 Details : for a short time RPCs with bulk IO are in the replay list,
3467 but replay of bulk IOs is unimplemented. If the OST filesystem
3468 is corrupted due to disk cache incoherency and then replay is
3469 started it is possible to trip an assertion. Avoid putting
3470 committed RPCs into the replay list at all to avoid this issue.
3475 Description: large O_DIRECT requests fail under memory pressure/fragmentation
3476 Details : Large single O_DIRECT read and write calls can fail to allocate
3477 a sufficiently large buffer to process the request. In case of
3478 allocation failure the allocation is retried with a smaller
3479 buffer and broken into smaller requests.
3481 Severity : enhancement
3483 Description: Add -o localflock option to simulate old noflock behaviour.
3484 Details : This will achieve local-only flock/fcntl locks coherentness.
3489 Description: versioning check is incomplete
3490 Details : Checking the version difference of client vs. server, report
3491 error if the gap is too big.
3496 Description: add support PG_writeback bit
3497 Details : add support for PG_writeback bit for Lustre, for more carefull
3498 work with page cache in 2.6 kernel. This also fix some deadlocks
3499 and remove hack for work O_SYNC with 2.6 kernel.
3501 Severity : enhancement
3503 Description: Add uninit_groups feature to ldiskfs2 to speed up e2fsck
3504 Details : The uninit_groups feature works in conjunction with the kernel
3505 filesystem code (ldiskfs2 only) and e2fsprogs-1.39-cfs6 to speed
3506 up the pass1 processing of e2fsck. This is a read-only feature
3507 in ldiskfs2 only, so older kernels and current ldiskfs cannot
3508 mount filesystems that have had this feature enabled.
3510 Severity : enhancement
3512 Description: Improve multi-block allocation algorithm to avoid fragmentation
3513 Details : The mballoc3 code (ldiskfs2 only) adds new mechanisms to improve
3514 allocation locality and avoid filesystem fragmentation.
3516 ------------------------------------------------------------------------------
3518 2007-02-09 Cluster File Systems, Inc. <info@clusterfs.com>
3520 * Support for kernels:
3521 2.6.9-42.0.3.EL (RHEL 4)
3522 2.6.5-7.276 (SLES 9)
3523 2.4.21-47.0.1.EL (RHEL 3)
3524 2.6.12.6 vanilla (kernel.org)
3525 2.6.16.21-0.8 (SLES10)
3526 * Recommended e2fsprogs version: 1.39.cfs2-0
3528 * The backwards-compatible /proc/sys/portals symlink has been removed
3529 in this release. Before upgrading, please ensure that you change
3530 any configuration scripts or /etc/sysctl.conf files that access
3531 /proc/sys/portals/* or sysctl portals.* to use the corresponding
3532 entry in /proc/sys/lnet or sysctl lnet.*. This change can be made
3533 in advance of the upgrade on any system running Lustre 1.4.6 or
3534 newer, since /proc/sys/lnet was added in that version.
3535 * Note that reiserfs quotas are disabled on SLES 10 in this kernel
3539 Frequency : only when quota is used
3541 Description: avoid scanning export list for quota master
3542 Details : Change the algorithms to avoid scanning export list in order
3543 to improve the efficiency.
3546 Frequency : MDS failover only, very rarely
3548 Description: "went back in time" messages on mds failover
3549 Details : The greatest transno may be lost when the current operation
3550 finishes with an error (transno==0) and the client's last_rcvd
3551 record is over-written. Save the greatest transno in the
3552 mds_last_transno for this case.
3555 Frequency : always for specific kernels and striping counts
3557 Description: client may get "Matching packet too big" without ACL support
3558 Details : Clients compiled without CONFIG_FS_POSIX_ACL get an error message
3559 when trying to access files in certain configurations. The
3560 clients should in fact be denied when mounting because they do
3561 not understand ACLs.
3564 Frequency : Cray XT3 with more than 4000 clients and multiple jobs
3566 Description: many clients connecting with IO in progress causes connect timeouts
3567 Details : Avoid synchronous journal commits to avoid delays caused by many
3568 clients connecting/disconnecting when bulk IO is in progress.
3569 Queue liblustre connect requests on OST_REQUEST_PORTAL instead of
3570 OST_IO_PORTAL to avoid delays behind potentially many pending
3574 Frequency : occasionally with multiple writers to a single file
3576 Description: shared writes to file may result in wrong size reported by stat()
3577 Details : Allow growing of kms when extent lock is cancelled
3580 Frequency : always with random mmap IO to multi-striped file
3582 Description: mmap write might be lost if we are writing to a 'hole' in stripe
3583 Details : Only if the hole is at the end of OST object so that kms is too
3584 small. Fix is to increase kms accordingly in ll_nopage.
3587 Frequency : rare, only if OST filesystem is inconsistent with MDS filesystem
3589 Description: writes to a missing object would leak memory on the OST
3590 Details : If there is an inconsistency between the MDS and OST filesystems,
3591 such that the MDS references an object that doesn't exist, writes
3592 to that object will leak memory due to incorrect cleanup in the
3593 error handling path, eventually running out of memory on the OST.
3598 Description: Creating too long symlink causes lustre errors
3599 Details : Check symlink and name lengths before sending requests to MDS.
3602 Frequency : only if flock is enabled (not on by default)
3604 Description: posix locks not released on fd closure on 2.6.9+
3605 Details : We failed to add posix locks to list of inode locks on 2.6.9+
3606 kernels, this caused such locks not to be released on fd close and
3607 then assertions on fs unmount about still used locks.
3610 Frequency : MDS failover only, very rarely
3612 Description: clients may get ASSERTION(granted_lock != NULL)
3613 Details : When request was taking a long time, and a client was resending
3614 a getattr by name lock request. The were multiple lock requests
3615 with the same client lock handle and
3616 mds_getattr_name->fixup_handle_for_resent_request found one of the
3617 lock handles but later failed with ASSERTION(granted_lock != NULL).
3622 Description: handle->h_buffer_credits > 0, assertion failure
3623 Details : h_buffer_credits is zero after truncate, causing assertion
3624 failure. This patch extends the transaction or creates a new
3628 Frequency : NFS re-export or patchless client
3629 Bugzilla : 11179, 10796
3630 Description: Crash on NFS re-export node (__d_move)
3631 Details : We do not want to hash the dentry if we don't have a lock.
3632 But if this dentry is later used in d_move, we'd hit uninitialised
3633 list head d_hash, so we just do this to init d_hash field but
3634 leave dentry unhashed.
3637 Frequency : NFS re-export or patchless client
3639 Description: NFS exports has problem with symbolic link
3640 Details : lustre client didn't properly install dentry when re-exported
3641 to NFS or running patchless client.
3644 Frequency : NFS re-export or patchless client
3646 Description: Various nfs/patchless fixes.
3647 Details : fixes reuse disconected alias for lookup process - this fixes
3648 warning "find_exported_dentry: npd != pd",
3649 fix permission error with open files at nfs.
3650 fix apply umask when do revalidate.
3653 Frequency : occasional
3655 Description: Crash on NFS re-export node
3656 Details : calling clear_page() on the wrong pointer triggered oops in
3657 generic_mapping_read().
3660 Frequency : rarely, using O_DIRECT IO
3662 Description: unaligned directio crashes client with LASSERT
3663 Details : check for unaligned buffers before trying any requests.
3666 Frequency : rarely, using CFS RAID5 patches in non-standard kernel series
3668 Description: stale data returned from RAID cache
3669 Details : If only a small amount of IO is done to the RAID device before
3670 reading it again it is possible to get stale data from the RAID
3671 cache instead of reading it from disk.
3674 Frequency : always for sles10 kernel
3676 Description: sles10 support
3677 Details : ll_follow_link: compile fixes and using of nd_set_link
3678 under newer kernels.
3681 Frequency : depends on arch, kernel and compiler version, always on sles10
3684 Description: recursive or deep enough symlinks cause stack overflow
3685 Details : getting rid of large stack-allocated variable in
3689 Frequency : depends on hardware
3691 Description: lustre write performance loss in the SLES10 kernel
3692 Details : the performance loss is caused by using of write barriers in the
3693 ext3 code. The SLES10 kernel turns barrier support on by
3694 default. The fix is to undo that change for ldiskfs.
3696 ------------------------------------------------------------------------------
3698 2006-12-09 Cluster File Systems, Inc. <info@clusterfs.com>
3700 * Support for kernels:
3701 2.6.9-42.0.3EL (RHEL 4)
3702 2.6.5-7.276 (SLES 9)
3703 2.4.21-47.0.1.EL (RHEL 3)
3704 2.6.12.6 vanilla (kernel.org)
3708 Frequency : quota enabled and large files being deleted
3710 Description: releasing more than 4GB of quota at once hangs OST
3711 Details : If a user deletes more than 4GB of files on a single OST it
3712 will cause the OST to spin in an infinite loop. Release
3713 quota in < 4GB chunks, or use a 64-bit value for 1.4.7.1+.
3718 Description: statfs data retrieved from /proc may be stale or zero
3719 Details : When reading per-device statfs data from /proc, in the
3720 {kbytes,files}_{total,free,avail} files, it may appear
3721 as zero or be out of date.
3724 Frequency : systems with MD RAID1 external journal devices
3726 Description: lconf's call to blkid is confused by RAID1 journal devices
3727 Details : Use the "blkid -l" flag to locate the MD RAID device instead
3728 of returning all block devices that match the journal UUID.
3731 Frequency : always, for aggregate stripe size over 4GB
3733 Description: "lfs setstripe" fails assertion when setting 4GB+ stripe width
3734 Details : Using "lfs setstripe" to set stripe size * stripe count over 4GB
3735 will fail the kernel with "ASSERTION(lsm->lsm_xfersize != 0)"
3738 Frequency : always if "lfs find" used on a local file/directory
3740 Description: "lfs find" segfaults if used on a local file/directory
3741 Details : The case where a directory component was not specified wasn't
3742 handled correctly. Handle this properly.
3745 Frequency : always on ppc64
3747 Description: the write to an ext3 filesystem mounted with mballoc got stuck
3748 Details : ext3_mb_generate_buddy() uses find_next_bit() which does not
3749 perform endianness conversion.
3752 Frequency : rarely (truncate to non-zero file size after write under load)
3753 Bugzilla : 10730, 10687
3754 Description: Files padded with zeros to next 4K multiple
3755 Details : With filesystems mounted using the "extents" option (2.6 kernels)
3756 it is possible that files that are truncated to a non-zero size
3757 immediately after being written are filled with zero bytes beyond
3758 the truncated size. No file data is lost.
3760 Severity : enhancement
3762 Description: Allow recovery/failover for liblustre clients.
3763 Details : liblustre clients were unaware of failover configurations until
3766 Severity : enhancement
3768 Description: user file locks should fail when not mounting with flock option
3769 Details : Set up an error-returning stub in ll_file_operations.lock field
3770 to prevent incorrect behaviour when client is mounted without
3771 flock option. Also, set up properly f_op->flock field for
3775 Frequency : always on ia64
3777 Description: "lfs df" loops on printing out MDS statfs information
3778 Details : The obd_ioctl_data was not initialized and in some systems
3779 this caused a failure during the ioctl that did not return
3780 an error. Initialize the struct and return an error on failure.
3783 Frequency : SLES 9 only
3785 Description: Error of copying files with lustre special EAs as root
3786 Details : Client side always return success for setxattr call for lustre
3787 special xattr (currently only "trusted.lov").
3790 Frequency : rarely on clusters with both ia64+i386 clients
3792 Description: ia64+i686 clients doing shared IO on the same file may LBUG
3793 Details : In rare cases when both ia64+i686 (or other mixed-PAGE_SIZE)
3794 clients are doing concurrent writes to the same file it is
3795 possible that the ia64 clients may LASSERT because the OST
3796 extent locks are not PAGE_SIZE aligned. Ensure that grown
3797 locks are always aligned on the request boundary.
3800 Frequency : specific use, occasional
3802 Description: Overwriting in use executable truncates on-disk binary image
3803 Details : If one node attempts to overwrite an executable in use by
3804 another node, we now correctly return ETXTBSY instead of
3805 truncating the file.
3807 Severity : enhancement
3809 Description: Async OSC create to avoid the blocking unnecessarily.
3810 Details : If a OST has no remain object, system will block on the creating
3811 when need to create a new object on this OST. Now, ways use
3812 pre-created objects when available, instead of blocking on an
3813 empty osc while others are not empty. If we must block, we block
3814 for the shortest possible period of time.
3819 Description: chmod on Lustre root is propagated to other clients
3820 Details : Re-validate root's dentry in ll_lookup_it to avoid having it
3821 invalid by the follow_mount time.
3824 Frequency : liblustre clients only
3826 Description: Race in 'instant cancel' lock handling could lead to such locks
3827 never to be granted in case of SMP MDS
3828 Details : Do not destroy not yet granted but cbpending locks in
3832 Frequency : replay/resend of open
3834 Description: non null lock assetion failure in mds_intent_policy
3835 Details : Trying to replay/resend lockless open requests resulted in
3836 mds_open() returning 0 with no lock. Now it sets a flag if
3837 a lock is going to be returned.
3839 Severity : enhancement
3841 Description: Checksum enhancements
3842 Details : New checksum enhancements allow for resending RPCs that failed
3845 Severity : enhancement
3847 Description: Tunables on number of dirty pages in cacche
3848 Details : Allow to set limit on number of dirty pages cached.
3853 Description: client crash on unmount - lock still has references
3854 Details : In some error handling cases it was possible to leak a lock
3855 reference on a client while accessing a file. This was not
3856 harmful to the client during operation, but would cause the
3857 client to crash when the filesystem is unmounted.
3860 Frequency : specific case, rare
3862 Description: ETXTBSY on mds though file not in use by client
3863 Details : ETXTBSY is no longer incorrectly returned when attempting to
3864 chmod or chown a directory that the user previously tried to
3865 execute or a currently-executing binary.
3868 Frequency : extremely rare except on liblustre-based clients
3870 Description: Lustre space not freed when files are deleted
3871 Details : Clean up open-unlinked files after client eviction. Previously
3872 the unlink was skipped and the files remained as orphans.
3877 Description: OST failure "would be an LBUG" in waiting_locks_callback()
3878 Details : In some cases it was possible to send a blocking callback to
3879 a client doing a glimpse, even though that client didn't get
3880 a lock granted. When the glimpse lock is cancelled on the OST
3881 the freed lock is left on the waiting list and corrupted the list.
3884 Frequency : all core dumps
3886 Description: Broke core dumps to lustre
3887 Details : Negative dentry may be unhashed if parent does not have UPDATE
3888 lock, but some callers, e.g. do_coredump, expect dentry to be
3889 hashed after successful create, hash it in ll_create_it.
3891 ------------------------------------------------------------------------------
3893 2006-09-13 Cluster File Systems, Inc. <info@clusterfs.com>
3895 * Support for kernels:
3896 2.6.9-42.0.2.EL (RHEL 4)
3897 2.6.5-7.276 (SLES 9)
3898 2.4.21-40.EL (RHEL 3)
3899 2.6.12.6 vanilla (kernel.org)
3903 Frequency : always on RHEL 3
3905 Description: Number of open files grows over time
3906 Details : The number of open files grows over time, whether or not
3907 Lustre is started. This was due to a filp leak introduced
3908 by one of our kernel patches.
3910 ------------------------------------------------------------------------------
3912 08-20-2006 Cluster File Systems, Inc. <info@clusterfs.com>
3914 * Support for kernels:
3915 2.6.9-42.EL (RHEL 4)
3916 2.6.5-7.276 (SLES 9)
3917 2.4.21-40.EL (RHEL 3)
3918 2.6.12.6 vanilla (kernel.org)
3923 Bugzilla : 5719, 9635, 9792, 9684
3924 Description: OST (or MDS) trips assertions in (re)connection under heavy load
3925 Details : If a server is under heavy load and cannot reply to new
3926 connection requests before the client resends the (re)connect,
3927 the connection handling code can behave badly if two service
3928 threads are concurrently handing separate (re)connections from
3929 the same client. Add better locking to the connection handling
3930 code, and ensure that only a single connection will be processed
3931 for a given client UUID, even if the lock is dropped.
3933 Severity : enhancement
3935 Description: add TCP zero-copy support to kernel
3936 Details : Add support to the kernel TCP stack to allow zero-copy bulk
3937 sends if the hardware supports scatter-gather and checksumming.
3938 This allows socklnd to do client-write and server-read more
3939 efficiently and reduce CPU utilization from skbuf copying.
3942 Frequency : only if NFS exporting from client
3944 Description: NULL pointer deref in ll_iocontrol() if chattr mknod file
3945 Details : If setting attributes on a file created under NFS that had
3946 never been opened it would be possible to oops the client
3947 if the file had no objects.
3950 Frequency : always for liblustre
3952 Description: liblustre client does MDS+OSTs setattr RPC for each write
3953 Details : When doing a write from a liblustre client, the client
3954 incorrectly issued an RPC to the MDS and each OST the file was
3955 striped over in order to update the timestamps. When writing
3956 with small chunks and many clients this could overwhelm the MDS
3957 with RPCs. In all cases it would slow down the write because
3958 these RPCs are unnecessary.
3960 Severity : enhancement
3962 Description: allow number of MDS service threads to be changed at module load
3963 Details : It is now possible to change the number of MDS service threads
3964 running. Adding "options mds mds_num_threads={N}" to the MDS's
3965 /etc/modprobe.conf will set the number of threads for the next
3966 time Lustre is restarted (assuming the "mds" module is also
3967 reloaded at that time). The default number of threads will
3968 stay the same, 32 for most systems.
3973 Description: OST crash if filesystem is unformatted or corrupt
3974 Details : If an OST is started on a device that has never been formatted
3975 or if the filesystem is corrupt and cannot even mount then the
3976 error handling cleanup routines would dereference a NULL pointer.
3981 Description: NULL pointer deref in llap_from_page.
3982 Details : get_cache_page_nowait can return a page with NULL (or otherwise
3983 incorrect) mapping if the page was truncated/reclaimed while it was
3984 searched for. Check for this condition and skip such pages when
3985 doing readahead. Introduce extra check to llap_from_page() to
3986 verify page->mapping->host is non-NULL (so page is not anonymous).
3989 Frequency : Sometimes when using sys_sendfile
3991 Description: "page not covered by a lock" warnings from ll_readpage
3992 Details : sendfile called ll_readpage without right page locks present.
3993 Now we introduced ll_file_sendfile that does necessary locking
3994 around call to generic_file_sendfile() much like we do in
3998 Frequency : with certain MDS communication failures at client mount time
4000 Description: NULL pointer deref after failed client mount
4001 Details : a client connection request may delayed by the network layer
4002 and not be sent until after the PTLRPC layer has timed out the
4003 request. If the client fails the mount immediately it will try
4004 to clean up before the network times out the request. Add a
4005 reference from the request import to the obd device and delay
4006 the cleanup until the network drops the request.
4009 Frequency : occasionally during client (re)connect
4011 Description: assertion failure during client (re)connect
4012 Details : processing a client connection request may be delayed by the
4013 client or server longer than the client connect timeout. This
4014 causes the client to resend the connection request. If the
4015 original connection request is replied in this interval, the
4016 client may trip an assertion failure in ptlrpc_connect_interpret()
4017 which thought it would be the only running connect process.
4020 Frequency : only with obd_echo servers and clients that are rebooted
4022 Description: kernel BUG accessing uninitialized data structure
4023 Details : When running an obd_echo server it did not start the ping_evictor
4024 thread, and when a client was evicted an uninitialized data
4025 structure was accessed. Start the ping_evictor in the RPC
4026 service startup instead of the OBD startup.
4028 Severity : enhancement
4029 Bugzilla : 10193 (patchless)
4030 Description: Remove dependency on various unexported kernel interfaces.
4031 Details : No longer need reparent_to_init, exit_mm, exit_files,
4032 sock_getsockopt, filemap_populate, FMODE_EXEC, put_filp.
4035 Frequency : rare (only users of deprecated and unsupported LDAP config)
4037 Description: write_conf for zeroconf mount queried LDAP incorrectly for client
4038 Details : LDAP apparently contains 'lustreName' attributes instead of
4039 'name'. A simple remapping of the name is sufficient.
4042 Frequency : rare (only with non-default dump_on_timeout debug enabled)
4044 Description: waiting_locks_callback trips kernel BUG if client is evicted
4045 Details : Running with the dump_on_timeout debug flag turned on makes
4046 it possible that the waiting_locks_callback() can try to dump
4047 the Lustre kernel debug logs from an interrupt handler. Defer
4048 this log dumping to the expired_lock_main() thread.
4050 Severity : enhancement
4052 Description: Support NFS exporting on 2.6 kernels.
4053 Details : Implement non-rawops metadata methods for NFS server to use without
4054 changing NFS server code.
4057 Frequency : very rare (synthetic metadata workload only)
4059 Description: two racing renames might cause an MDS thread to deadlock
4060 Details : Running the "racer" program may cause one MDS thread to rename
4061 a file from being the source of a rename to being the target of
4062 a rename at exactly the same time that another thread is doing
4063 so, and the second thread has already enqueued these locks after
4064 doing a lookup of the target and is trying to relock them in
4065 order. Ensure that we don't try to re-lock the same resource.
4068 Frequency : only very large systems with liblustre clients
4070 Description: slow eviction of liblustre clients with the "evict_by_nid" RPC
4071 Details : Use asynchronous set_info RPCs to send the "evict_by_nid" to
4072 all OSTs in parallel. This allows the eviction of stale liblustre
4073 clients to proceed much faster than if they were done in series,
4074 and also offers similar improvements for other set_info RPCs.
4079 Description: excessive CPU usage during initial read phase on client
4080 Details : During the initial read phase on a client, it would agressively
4081 retry readahead on the file, consuming too much CPU and impacting
4082 performance (since 1.4.5.8). Improve the readahead algorithm
4083 to avoid this, and also improve some other common cases (read
4084 of small files in particular, where "small" is files smaller than
4085 /proc/fs/lustre/llite/*/max_read_ahead_whole_mb, 2MB by default).
4090 Description: MDS crash when receiving packet with unknown intent.
4091 Details : Do not LBUG in unknown intent case, just return -EFAULT
4093 Severity : enhancement
4094 Bugzilla : 9293, 9385
4095 Description: MDS RPCs are serialised on client. This is unnecessary for some.
4096 Details : Do not serialize getattr (non-intent version) and statfs.
4099 Frequency : occasional, when OST network is overloaded/intermittent
4101 Description: client evicted by OST after bulk IO timeout
4102 Details : If a client sends a bulk IO request (read or write) the OST
4103 may evict the client if it is unresposive to its data GET/PUT
4104 request. This is incorrect if the network is overloaded (takes
4105 too long to transfer the RPC data) or dropped the OST GET/PUT
4106 request. There is no need to evict the client at all, since
4107 the pinger and/or lock callbacks will handle this, and the
4108 client can restart the bulk request.
4111 Frequency : Always when mmapping file with no objects
4113 Description: client crashes when mmapping file with no objects
4114 Details : Check that we actually have objects in a file before doing any
4115 operations on objects in ll_vm_open, ll_vm_close and
4121 Description: Request leak when working with deleted CWD
4122 Details : Introduce advanced request refcount tracking for requests
4123 referenced from lustre intent.
4125 Severity : Enhancement
4127 Description: Cache open file handles on client.
4128 Details : MDS now will return special lock along with openhandle, if
4129 requested and client is allowed to hold openhandle, even if unused,
4130 until such a lock is revoked. Helps NFS a lot, since NFS is opening
4131 closing files for every read/write openration.
4133 Severity : Enhancement
4135 Description: Cache open negative dentries on client when possible.
4136 Details : Guard negative dentries with UPDATE lock on parent dir, drop
4137 negative dentries on lock revocation.
4142 Description: Remounting a client read-only wasn't possible with a zconf mount
4143 Details : It wasn't possible to remount a client read-only with llmount.
4145 Severity : enhancement
4146 Description: Include MPICH 1.2.6 Lustre ADIO interface patch
4147 Details : In lustre/contrib/ or /usr/share/lustre in RPM a patch for
4148 MPICH is included to add Lustre-specific ADIO interfaces.
4149 This is based closely on the UFS ADIO layer and only differs
4150 in file creation, in order to allow the OST striping to be set.
4151 This is user-contributed code and not supported by CFS.
4156 Description: extended inode attributes (immutable, append-only) work improperly
4157 when 2.4 and 2.6 kernels are used on client/server or vice versa
4158 Details : Introduce kernel-independent values for these flags.
4160 Severity : enhancement
4163 Description: Allow fractional MB tunings for lustre in /proc/ filesystem.
4164 Details : Many of the /proc/ tunables can only be tuned at a megabyte
4165 granularity. Now, Fractional MB granularity is be supported,
4166 this is very useful for low memory system.
4168 Severity : enhancement
4170 Description: Getattr by fid
4171 Details : Getting a file attributes by its fid, obtaining UPDATE|LOOKUP
4172 locks, avoids extra getattr rpc requests to MDS, allows '/' to
4173 have locks and avoids getattr rpc requests for it on every stat.
4176 Frequency : Always, for filesystems larger than 2TB
4178 Description: ldiskfs crash at mount for filesystem larger than 2TB with mballoc
4179 Details : Kenrel kmalloc limits allocations to 128kB and this prevents
4180 filesystems larger than 2TB to be mounted with mballoc enabled.
4183 Frequency : Always, for 32-bit kernel without CONFIG_LBD and filesystem > 2TB
4185 Description: ldiskfs crash at mount for filesystem larger than 2TB with mballoc
4186 Details : If a 32-bit kernel is compiled without CONFIG_LBD enabled and a
4187 filesystems larger than 2TB is mounted then the kernel will
4188 silently corrupt the start of the filesystem. CONFIG_LBD is
4189 enabled for all CFS-supported kernels, but the possibility of
4190 this happening with a modified kernel config exists.
4192 Severity : enhancement
4194 Description: add client O_DIRECT support for 2.6 kernels
4195 Details : It is now possible to do O_DIRECT reads and writes to files
4196 in the Lustre client mountpoint on 2.6 kernel clients.
4198 Severity : enhancement
4200 Description: parallel glimpse, setattr, statfs, punch, destroy requests
4201 Details : Sends glimpse, setattr, statfs, punch, destroy requests to OSTs in
4202 parallel, not waiting for response from every OST before sending
4203 a rpc to the next OST.
4208 Description: setattr vs write race when updating file timestamps
4209 Details : Client processes that update a file timestamp into the past
4210 right after writing to the file (e.g. tar) it is possible that
4211 the updated file modification time can be reset to the current
4212 time due to a race between processing the setattr and write RPC.
4214 Severity : enhancement
4216 Description: Bring 'lfs find' closer in line with regular Linux find.
4217 Details : lfs find util supports -atime, -mtime, -ctime, -maxdepth, -print,
4218 -print0 options and obtains all the needed info through the lustre
4221 Severity : enhancement
4223 Description: support up to 1024 configured devices on one node
4224 Details : change obd_dev array from statically allocated to dynamically
4225 allocated structs as they are first used to reduce memory usage
4230 Description: Flush dirty partially truncated pages during truncate
4231 Details : Immediatelly flush partially truncated pages in filter_setattr,
4232 this way we completely avoid having any pages in page cache on OST
4233 and can retire ugly workarounds during writes to flush such pages.
4238 Description: i_sem vs transaction deadlock in mds_obd_destroy during unlink.
4239 Details : protect inode from truncation within vfs_unlink() context
4240 just take a reference before calling vfs_unlink() and release it
4241 when parent's i_sem is free.
4246 Description: last_id value checked outside lock on OST caused LASSERT failure
4247 Details : If there were multiple MDS->OST object precreate requests in
4248 flight, it was possible that the OST's last object id was checked
4249 outside a lock and incorrectly tripped an assertion. Move checks
4250 inside locks, and discard old precreate requests.
4253 Frequency : always, if extents are used on OSTs
4255 Description: index ei_leaf_hi (48-bit extension) is not zeroed in extent index
4256 Details : OSTs using the extents format would not zero the high 16 bits of
4257 the index physical block number. This is not a problem for any
4258 OST filesystems smaller than 16TB, and no kernels support ext3
4259 filesystems larger than 16TB yet. This is fixed in 1.4.7 (all
4260 new/modified files) and can be fixed for existing filesystems
4261 with e2fsprogs-1.39-cfs1.
4266 Description: import connection selection may be incorrect if timer wraps
4267 Details : Using a 32-bit jiffies timer with HZ=1000 may cause backup
4268 import connections to be ignored if the 32-bit jiffies counter
4269 wraps. Use a 64-bit jiffies counter.
4272 Frequency : very large clusters immediately after boot
4274 Description: LNET request buffers exhausted under heavy short-term load
4275 Details : If a large number of client requests are generated on a service
4276 that has previously never seen so many requests it is possible
4277 that the request buffer growth cannot keep up with the spike in
4278 demand. Instead of dropping incoming requests, they are held in
4279 the LND until the RPC service can accept more requests.
4282 Frequency : Sometimes during replay
4284 Description: Assertion failure in ll_local_open after replay.
4285 Details : If replay happened on an open request reply before we were able
4286 to set replay handler, reply will become not swabbed tripping the
4287 assertion in ll_local_open. Now we set the handler right after
4288 recognising of open request
4291 Frequency : very rare
4293 Description: kernel reports "badness in vsnprintf"
4294 Details : Reading from the "recovery_status" /proc file in small chunks
4295 may cause a negative length in lprocfs_obd_rd_recovery_status()
4296 call to vsnprintf() (which is otherwise harmless). Exit early
4297 if there is no more space in the output buffer.
4299 Severity : enhancement
4301 Description: clear OBD RPC statistics by writing to them
4302 Details : It is now possible to clear the OBD RPC statistics by writing
4303 to the "stats" file.
4308 Description: Client mtime is not the same on different clients after utimes
4309 Details : In some cases, the client was using the utimes() syscall on
4310 a file cached on another node. The clients now validate the
4311 ctime from the MDS + OSTs to determine which one is right.
4316 Description: Inability to activate failout mode
4317 Details : lconf script incorrectly assumed that in pythong string's numeric
4318 value is used in comparisons.
4321 Frequency : always with multiple stripes per file
4323 Description: Inefficient object allocation for mutli-stripe files
4324 Details : When selecting which OSTs to stripe files over, for files with
4325 a stripe count that divides evenly into the number of OSTs,
4326 the MDS is always picking the same starting OST for each file.
4327 Return the OST selection heuristic to the original design.
4332 Description: mount failures may take full timeout to return an error
4333 Details : Under some heavy load conditions it is possible that a
4334 failed mount can wait for the full obd_timeout interval,
4335 possibly several minutes, before reporting an error.
4336 Instead return an error as soon as the status is known.
4338 Frequency : quota enabled and large files being deleted
4340 Description: releasing more than 4GB of quota at once hangs OST
4341 Details : If a user deletes more than 4GB of files on a single OST it
4342 will cause the OST to spin in an infinite loop. Release
4343 quota in < 4GB chunks, or use a 64-bit value for 1.4.7.1+.
4348 Description: statfs data retrieved from /proc may be stale or zero
4349 Details : When reading per-device statfs data from /proc, in the
4350 {kbytes,files}_{total,free,avail} files, it may appear
4351 as zero or be out of date.
4354 Frequency : systems with MD RAID1 external journal devices
4356 Description: lconf's call to blkid is confused by RAID1 journal devices
4357 Details : Use the "blkid -l" flag to locate the MD RAID device instead
4358 of returning all block devices that match the journal UUID.
4361 Frequency : always, for aggregate stripe size over 4GB
4363 Description: assertion fails when trying to use 4GB stripe size
4364 Details : Use "setstripe" to set stripe size over 4GB will fail the kernel,
4365 complaining "ASSERTION(lsm->lsm_xfersize != 0)"
4368 Frequency : always on ppc64
4370 Description: the first write on an ext3 filesystem with mballoc got stuck
4371 Details : ext3_mb_generate_buddy() uses find_next_bit() which does not
4372 perform endianness conversion.
4374 ------------------------------------------------------------------------------
4376 02-14-2006 Cluster File Systems, Inc. <info@clusterfs.com>
4378 * WIRE PROTOCOL CHANGE. This version of Lustre networking WILL NOT
4379 INTEROPERATE with older versions automatically. Please read the
4380 user documentation before upgrading any part of a live system.
4381 * WARNING: Lustre networking configuration changes are required with
4382 this release. See https://bugzilla.clusterfs.com/show_bug.cgi?id=10052
4385 * Support for kernels:
4386 2.6.9-22.0.2.EL (RHEL 4)
4387 2.6.5-7.244 (SLES 9)
4388 2.6.12.6 vanilla (kernel.org)
4391 Severity : enhancement
4392 Bugzilla : 7981/8208
4393 Description: Introduced Lustre Networking (LNET)
4394 Details : LNET is new networking infrastructure for Lustre, it includes
4395 a reorganized network configuration mode (see the user
4396 documentation for full details) as well as support for routing
4397 between different network fabrics. Lustre Networking Devices
4398 (LNDs) for the supported network fabrics have also been
4399 created for this new infrastructure.
4401 Severity : enhancement
4402 Description: Introduced Access control lists
4403 Details : clients can set ACLs on files and directories in order to have
4404 more fine-grained permissions than the standard Unix UGO+RWX.
4405 The MDS must be started with the "-o acl" mount option.
4407 Severity : enhancement
4408 Description: Introduced filesystem quotas
4409 Details : Administrators may now establish per-user quotas on the
4412 Severity : enhancement
4414 Description: Configuration change for the XT3
4415 The PTLLND is now used to run Lustre over Portals on the XT3
4416 The configure option(s) --with-cray-portals are no longer used.
4417 Rather --with-portals=<path-to-portals-includes> is used to
4418 enable building on the XT3. In addition to enable XT3 specific
4419 features the option --enable-cray-xt3 must be used.
4424 Description: Running on many-way SMP OSTs can trigger oops in llcd_send()
4425 Details : A race between allocating a new llcd and re-getting the llcd_lock
4426 allowed another thread to grab newly-allocated llcd.
4428 Severity : enhancement
4430 Description: 2.6 OST async journal commit and locking fix to improve performance
4431 Details : The filter_direct_io()+filter_commitrw_write() journal commits for
4432 2.6 kernels are now async as they already were in 2.4 kernels so
4433 that they can commit concurrently with the network bulk transfer.
4434 For block-allocated files the filter allocation semaphore is held
4435 to avoid filesystem fragmentation during allocation. BKL lock
4436 removed for 2.6 xattr operations where it is no longer needed.
4441 Description: lconf incorrectly determined whether two IP networks could talk
4442 Details : In some more complicated routing and multiple-network
4443 configurations, lconf will avoid trying to make a network
4444 connection to a disjoint part of the IP space. It was doing the
4445 math incorrectly for one set of cases.
4450 Description: Fix for potential infinite loop processing records in an llog.
4451 Details : If an llog record is corrupted/zeroed, it is possible to loop
4452 forever in llog_process(). Validate the llog record length
4453 and skip the remainder of the block on error.
4456 Frequency : occasional (liblustre only)
4458 Description: liblustre could not open files whose last component is a symlink
4459 Details : sysio_path_walk() would incorrectly pass the open intent to
4460 intermediate path components.
4463 Frequency : rare (liblustre only with non-standard tuning)
4464 Bugzilla : 7201 (7350)
4465 Description: Tuning the MDC DLM LRU size to zero triggers client LASSERT
4466 Details : llu_lookup_finish_locks() tries to set lock data on a lock
4467 after it has been released, only do this for referenced locks
4469 Severity : enhancement
4471 Description: specifying an (invalid) directory default stripe_size of -1
4472 would reset the directory default striping
4473 Details : stripe_size -1 was used internally to signal directory stripe
4474 removal, now use "all default" to signal dir stripe removal
4475 as a directory striping of "all default" is not useful
4478 Frequency : common for large clusters running liblustre clients
4480 Description: doing an ls when liblustre clients are running is slow
4481 Details : sending a glimpse AST to a liblustre client waits for every AST
4482 to time out, as liblustre clients will not respond. Since they
4483 cannot cache data we refresh the OST lock LVB from disk instead.
4485 Severity : enhancement
4487 Description: doing an ls at the same time as file IO can be slow
4488 Details : enqueue and other "small" requests can be blocked behind many
4489 large IO requests. Create a new OST IO portal for non-IO
4490 requests so they can be processed faster.
4493 Frequency : rare (only HPUX clients mounting unsupported re-exported NFS vol)
4495 Description: an HPUX NFS client would get -EACCESS when ftruncate()ing a newly
4496 created file with mode 000
4497 Details : the Linux NFS server relies on an MDS_OPEN_OWNEROVERRIDE hack to
4498 allow an ftruncate() as a non-root user to a file with mode 000.
4499 Lustre now respects this flag to disable mode checks when
4500 truncating a file owned by the user
4503 Frequency : liblustre-only, when liblustre client dies unexpectedly or becomes
4506 Description: Revoking locks from clients that went dead or catatonic might take
4508 Details : New lock flags FL_CANCEL_ON_BLOCK used by liblustre makes
4509 cancellation of such locks instant on servers without waiting for
4510 any reply from clients. Clients drops these locks when cancel
4511 notification from server is received without replying.
4514 Frequency : liblustre-only, when liblustre client dies or becomes busy
4516 Description: Doing ls on Linux clients can take a long time with active
4518 Details : Liblustre client cannot handle ASTs in timely manner, so avoid
4519 granting such locks to it in the first place if possible. Locks
4520 are taken by proxy on the OST during the read or write and
4521 dropped immediately afterward. Add connect flags handling, do
4522 not grant locks to liblustre clients for glimpse ASTs.
4524 Severity : enhancement
4526 Description: Improve read-ahead algorithm to avoid excessive IO for random reads
4527 Details : Existing read-ahead algorithm is tuned for the case of streamlined
4528 sequential reads and behaves badly with applications doing random
4529 reads. Improve it by reading ahead at least read region, and
4530 avoiding excessive large RPC for small reads.
4532 Severity : enhancement
4534 Description: Creating more than 1000 files for a single job may cause a load
4535 imbalance on the OSTs if there are also a large number of OSTs.
4536 Details : qos_prep_create() uses an OST index reseed value that is an
4537 even multiple of the number of available OSTs so that if the
4538 reseed happens in the middle of the object allocation it will
4539 still utilize the OSTs as uniformly as possible.
4544 Description: OST or MDS may oops in ping_evictor_main()
4545 Details : ping_evictor_main() drops obd_dev_lock if deleting a stale export
4546 but doesn't restart at beginning of obd_exports_timed list
4549 Severity : enhancement
4551 Description: improve by-nid export eviction on the MDS and OST
4552 Details : allow multiple exports with the same NID to be evicted at one
4553 time without re-searching the exports list.
4556 Frequency : rare, only with supplementary groups enabled on SMP 2.6 kernels
4558 Description: MDS may oops in groups_free()
4559 Details : in rare race conditions a newly allocated group_info struct is
4560 freed again, and this can be NULL. The 2.4 compatibility code
4561 for groups_free() checked for a NULL pointer, but 2.6 did not.
4564 Frequency : common for liblustre clients doing little filesystem IO
4565 Bugzilla : 9352, 7313
4566 Description: server may evict liblustre clients accessing contended locks
4567 Details : if a client is granted a lock or receives a completion AST
4568 with a blocking AST already set it would not reply to the AST
4569 for LDLM_FL_CANCEL_ON_BLOCK locks. It now replies to such ASTs.
4572 Frequency : lfs setstripe, only systems with more than 160 OSTs
4574 Description: unable to set striping with a starting offset beyond OST 160
4575 Details : llapi_create_file() incorrectly limited the starting stripe
4576 index to the maximum single-file stripe count.
4579 Frequency : LDAP users only
4581 Description: lconf did not handle in-kernel recovery with LDAP properly
4582 Details : lconf/LustreDB get_refs() is searching the wrong namespace
4584 Severity : enhancement
4586 Description: bind OST threads to NUMA nodes to improve performance
4587 Details : all OST threads are uniformly bound to CPUs on a single NUMA
4588 node and do their allocations there to localize memory access
4590 Severity : enhancement
4592 Description: llmount can determine client NID directly from Myrinet (GM)
4593 Details : the client NID code from gmnalnid was moved directly into
4594 llmount, removing the need to use this or specifying the
4595 client NID explicitly when mounting GM clients with zeroconf
4598 Frequency : if client is started with down MDS
4600 Description: if client is started with down MDS mount hangs in ptlrpc_queue_wait
4601 Details : Having an LWI_INTR() wait event (interruptible, but no timeout)
4602 will wait indefinitely in ptlrpc_queue_wait->l_wait_event() after
4603 ptlrpc_import_delayed_req() because we didn't check if the
4604 request was interrupted, and we also didn't break out of the
4605 event loop if there was no timeout
4610 Description: data loss during non-page-aligned writes to a single file from
4611 both multiple nodes and multiple threads on one node at same time
4612 Details : updates to KMS and lsm weren't protected by common lock. Resulting
4613 inconsistency led to false short-reads, that were cached and later
4614 used by ->prepare_write() to fill in partially written page,
4615 leading to data loss.
4618 Frequency : always, if lconf --abort_recovery used
4620 Description: lconf --abort_recovery fails with 'Operation not supported'
4621 Details : lconf was attempting to abort recovery on the MDT device and not
4624 Severity : enhancement
4626 Description: remove cleanup logs
4627 Details : replace lconf-generated cleanup logs with lustre internal
4628 cleanup routines. Eliminates the need for client-cleanup and
4631 Severity : enhancement
4633 Description: add support for EAs (user and system) on lustre filesystems
4634 Details : it is now possible to store extended attributes in the Lustre
4635 client filesystem, and with the user_xattr mount option it
4636 is possible to allow users to store EAs on their files also
4638 Severity : enhancement
4640 Description: Add possibility (config option) to show minimal available OST free
4642 Details : When compiled with --enable-mindf configure option, statfs(2)
4643 (and so, df) will return least minimal free space available from
4644 all OSTs as amount of free space on FS, instead of summary of
4645 free spaces of all OSTs.
4647 Severity : enhancement
4649 Description: do not expand extent locks acquired on OST-side
4650 Details : Modify ldlm_extent_policy() to not expand local locks, acquired
4651 by server: they are not cached anyway.
4654 Frequency : when mmap is used/binaries executed from Lustre
4656 Description: Unmmap pages before throwing them away from read cache.
4657 Details : llap_shrink cache now attempts to unmap pages before discarding
4658 them (if unmapping failed - do not discard). SLES9 kernel has
4659 extra checks that trigger if this unmapping is not done first.
4664 Description: lconf didn't resolve symlinks before checking to see whether a
4665 given mountpoint was already in use
4668 Frequency : when migrating failover services
4669 Bugzilla : 6395, 9514
4670 Description: When migrating a subset of services from a node (e.g. failback
4671 from a failover service node) the remaining services would
4672 time out and evict clients.
4673 Details : lconf --force (implied by --failover) sets the global obd_timeout
4674 to 5 seconds in order to quickly disconnect, but this caused
4675 other RPCs to time out too quickly. Do not change the global
4676 obd_timeout for force cleanup, only set it for DISCONNECT RPCs.
4678 Severity : enhancement
4679 Frequency : if MDS is started with down OST
4680 Bugzilla : 9439,5706
4681 Description: Allow startup/shutdown of an MDS without depending on the
4682 availability of the OSTs.
4683 Details : Asynchronously call mds_lov_synchronize during MDS startup.
4684 Add appropriate locking and lov-osc refcounts for safe
4685 cleaning. Add osc abort_inflight calls in case the
4686 synchronize never started.
4689 Frequency : occasional (Cray XT3 only)
4691 Description: root not authorized to access files in CRAY_PORTALS environment
4692 Details : The client process capabilities were not honoured on the MDS in
4693 a CRAY_PORTALS/CRAY_XT3 environment. If the file had previously
4694 been accessed by an authorized user then root was able to access
4695 the file on the local client also. The root user capabilities
4696 are now allowed on the MDS, as this environment has secure UID.
4699 Frequency : occasional
4701 Description: ldiskfs "too long searching" message happens too often
4702 Details : A debugging message (otherwise harmless) prints too often on
4703 the OST console. This has been reduced to only happen when
4704 there are fragmentation problems on the filesystem.
4709 Description: Division by zero in statfs when all OSCs are inactive
4710 Details : lov_get_stripecnt() returns zero due to incorrect order of checks,
4711 lov_statfs divides by value returned by lov_get_stripecnt().
4715 Bugzilla : 9489, 3273
4716 Description: First write from each client to each OST was only 4kB in size,
4717 to initialize client writeback cache, which caused sub-optimal
4718 RPCs and poor layout on disk for the first writen file.
4719 Details : Clients now request an initial cache grant at (re)connect time
4720 and so that they can start streaming writes to the cache right
4721 away and always do full-sized RPCs if there is enough data.
4722 If the OST is rebooted the client also re-establishes its grant
4723 so that client cached writes will be honoured under the grant.
4728 Description: Slow ls (and stat(2) syscall) on files residing on IO-loaded OSTs
4729 Details : Now I/O RPCs go to different portal number and (presumably) fast
4730 lock requests (and glimses) and other RPCs get their own service
4731 threads pool that should be able to service those RPCs
4734 Severity : enhancement
4736 Description: Ability to exchange lustre version between client and servers and
4737 issue warnings at client side if client is too old. Also for
4738 liblustre clients there is ability to refuse connection of too old
4740 Details : New 'version' field is added to connect data structure that is
4741 filled with version info. That info is later checked by server and
4745 Frequency : rare, liblustre only.
4746 Bugzilla : 9296, 9581
4747 Description: Two simultaneous writes from liblustre at offset within same page
4748 might proceed at the same time overwriting eachother with stale
4750 Details : I/O lock withing llu_file_prwv was released too early, before data
4751 actually was hitting the wire. Extended lock-holding time until
4752 server acknowledges receiving data.
4755 Frequency : extremely rare. Never observed in practice.
4757 Description: avoid generating lustre_handle cookie of 0.
4758 Details : class_handle_hash() generates handle cookies by incrementing
4759 global counter, and can hit 0 occasionaly (this is unlikely, but
4760 not impossible, because initial value of cookie counter is
4761 selected randonly). Value of 0 is used as a sentinel meaning
4762 "unassigned handle" --- avoid it. Also coalesce two critical
4763 sections in this function into one.
4765 Severity : enhancement
4767 Description: allow liblustre clients to delegate truncate locking to OST
4768 Details : To avoid overhead of locking, liblustre client instructs OST to
4769 take extent lock in ost_punch() on client's behalf. New connection
4770 flag is added to handle backward compatibility.
4772 Severity : enhancement
4773 Bugzilla : 4928, 7341, 9758
4774 Description: allow number of OST service threads to be specified
4775 Details : a module parameter allows the number of OST service threads
4776 to be specified via "options ost ost_num_threads={N}" in the
4777 OSS's /etc/modules.conf or /etc/modprobe.conf.
4781 Bugzilla : 6146, 9635, 9895
4782 Description: servers crash with bad pointer in target_handle_connect()
4783 Details : In rare cases when a client is reconnecting it was possible that
4784 the connection request was the last reference for that export.
4785 We would temporarily drop the export reference and get a new
4786 one, but this may have been the last reference and the export
4787 was just destroyed. Get new reference before dropping old one.
4789 Severity : enhancement
4790 Frequency : if client is started with failover MDS
4792 Description: Allow multiple MDS hostnames in the mount command
4793 Details : Try to read the configuration from all specified MDS
4794 hostnames during a client mount in case the "primary"
4797 Severity : enhancement
4799 Description: Stop sending data to evicted clients as soon as possible.
4800 Details : Check if the client we are about to send or are sending data to
4801 was evicted already. (Check is done every second of waiting,
4802 for which l_wait_event interface was extended to allow checking
4803 of exit condition at specified intervals).
4806 Frequency : rare, normally only when NFS exporting is done from client
4808 Description: 'bad disk LOV MAGIC: 0x00000000' error when chown'ing files
4810 Details : Make mds_get_md() recognise empty md case and set lmm size to 0.
4813 Frequency : always, if srand() is called before liblustre initialization
4815 Description: Liblustre uses system PRNG disturbing its usage by user application
4816 Details : Introduce internal to lustre fast and high-quality PRNG for
4817 lustre usage and make liblustre and some other places in generic
4818 lustre code to use it.
4820 Severity : enhancement
4821 Bugzilla : 9477, 9557, 9870
4822 Description: Verify that the MDS configuration logs are updated when xml is
4823 Details : Check if the .xml configuration logs are newer than the config
4824 logs stored on the MDS and report an error if this is the case.
4825 Request --write-conf, or allow starting with --old_conf.
4827 Severity : enhancement
4829 Description: Handle symlinks in the path when checking if Lustre is mounted.
4830 Details : Resolve intermediate symlinks when checking if a client has
4831 mounted a filesystem to avoid duplicate client mounts.
4836 Description: lconf can hit an error exception but still return success.
4837 Details : The lconf command catches the Command error exception at the top
4838 level script context and will exit with the associated exit
4839 status, but doesn't ensure that this exit status is non-zero.
4844 Description: failure of ptlrpc thread startup can cause oops
4845 Details : Starting a ptlrpc service thread can fail if there are a large
4846 number of threads or the server memory is very fragmented.
4847 Handle this without oopsing.
4850 Frequency : always, only if liblustre and non-default acceptor port was used
4852 Description: liblustre cannot connect to servers with non-default acceptor port
4853 Details : tcpnal_set_default_params() was not called and was therefore
4854 ignoring the environment varaible TCPNAL_PORT, as well as other
4855 TCPNAL_ environment variables
4860 Description: two objects could be created on the same OST for a single file
4861 Details : If an OST is down, in some cases it was possible to create two
4862 objects on a single OST for a single file. No problems other
4863 than potential performance impact and spurious error messages.
4867 Bugzilla : 5681, 9562
4868 Description: Client may oops in ll_unhash_aliases
4869 Details : Client dcache may become inconsistent in race condition.
4870 In some cases "getcwd" can fail if the current directory is
4876 Description: Inode refcounting problems in NFS export code
4877 Details : link_raw functions used to call d_instantiate without obtaining
4878 extra inode reference first.
4882 Bugzilla : 9942, 9903
4883 Description: Referencing freed requests leading to crash, memleaks with NFS.
4884 Details : We used to require that call to ll_revalidate_it was always
4885 followed by ll_lookup_it. Also with revalidate_special() it is
4886 possible to call ll_revalidate_it() twice for the same dentry
4887 even if first occurence returned success. This fix changes semantic
4888 between DISP_ENQ_COMPLETE disposition flag to mean there is extra
4889 reference on a request referred from the intent.
4890 ll_intent_release() then releases such a request.
4893 Frequency : rare, normally benchmark loads only
4895 Description: unlinked inodes were kept in memory on the client
4896 Details : If a client is repeatedly creating and unlinking files it
4897 can accumulate a lot of stale inodes in the inode slab cache.
4898 If there is no other client load running this can cause the
4899 client node to run out of memory. Instead flush old inodes
4900 from client cache that have the same inode number as a new inode.
4903 Frequency : SLES9 2.6.5 kernel and long filenames only
4904 Bugzilla : 9969, 10379
4905 Description: utime reports stale NFS file handle
4906 Details : SLES9 uses out-of-dentry names in some cases, which confused
4907 the lustre dentry revalidation. Change it to always use the
4911 Frequency : rare, unless heavy write-truncate concurrency is continuous
4912 Bugzilla : 4180, 6984, 7171, 9963, 9331
4913 Description: OST becomes very slow and/or deadlocked during object unlink
4914 Details : filter_destroy() was holding onto the parent directory lock
4915 while truncating+unlinking objects. For very large objects this
4916 may block other threads for a long time and slow overall OST
4917 responsiveness. It may also be possible to get a lock ordering
4918 deadlock in this case, or run out of journal credits because of
4919 the combined truncate+unlink. Solution is to do object truncate
4920 first in one transaction without parent lock, and then do the
4921 final unlink in a new transaction with the parent lock. This
4922 reduces the lock hold time dramatically.
4925 Frequency : rare, 2.4 kernels only
4927 Description: MDS or OST cleanup may trip kernel BUG when dropping kernel lock
4928 Details : mds_cleanup() and filter_cleanup() need to drop the kernel lock
4929 before unmounting their filesystem in order to avoid deadlock.
4930 The kernel_locked() function in 2.4 kernels only checks whether
4931 the kernel lock is held, not whether it is this process that is
4932 holding it as 2.6 kernels do.
4937 Description: MDS or OST may oops/LBUG if a client is connecting multiple times
4938 Details : The client ptlrpc code may be trying to reconnect to a down
4939 server before a previous connection attempt has timed out.
4940 Increase the reconnect interval to be longer than the connection
4941 timeout interval to avoid sending duplicate connections to
4945 Frequency : echo_client brw_test command
4947 Description: fix echo_client to work with OST preallocated code
4948 Details : OST preallocation code (5137) didn't take echo_client IO path
4949 into account: echo_client calls filter methods outside of any
4950 OST thread and, hence, there is no per-thread preallocated
4951 pages and buffers to use. Solution: hijack pga pages for IO. As
4952 a byproduct, this avoids unnecessary data copying.
4956 Bugzilla : 3555, 5962, 6025, 6155, 6296, 9574
4957 Description: Client can oops in mdc_commit_close() after open replay
4958 Details : It was possible for the MDS to return an open request with no
4959 transaction number in mds_finish_transno() if the client was
4960 evicted, but without actually returning an error. Clients
4961 would later try to replay that open and may trip an assertion
4962 Simplify the client close codepath, and always return an error
4963 from the MDS in case the open is not successful.
4966 Frequency : rare, 2.6 OSTs only
4968 Description: OST may deadlock under high load on fragmented files
4969 Details : If there was a heavy load and highly-fragmented OST filesystems
4970 it was possible to have all the OST threads deadlock waiting on
4971 allocation of biovecs, because the biovecs were not released
4972 until the entire RPC IO was completed. Instead, release biovecs
4973 as soon as they are complete to ensure forward IO progress.
4975 Severity : enhancement
4977 Description: Support for specifying external journal device at mount
4978 Details : If an OST or MDS device is formatted with an external journal
4979 device, this device major/minor is stored in the ext3 superblock
4980 and may not be valid for failover. Allow detecting and
4981 specifying the external journal at mount time.
4986 Description: Mounting an MDS with pending unlinked files may cause oops
4987 Details : target_finish_recovery() calls mds_postrecov() which returned
4988 the number of orphans unlinked. mds_lov_connect->mds_postsetup()
4989 considers this an error and immediately begins cleaning up the
4990 lov, just after starting the mds_lov process
4992 Severity : enhancement
4994 Description: Implement 'lfs df' to report actual free space on per-OST basis
4995 Details : Add sub-command 'df' on 'lfs' to report the disk space usage of
4996 MDS/OSDs. Usage: lfs df [-i][-h]. Command Options: '-i' to report
4997 usage of objects; '-h' to report in human readable format.
4999 ------------------------------------------------------------------------------
5001 08-26-2005 Cluster File Systems, Inc. <info@clusterfs.com>
5008 Description: Mounting an ldiskfs file system with mballoc may crash OST node.
5009 Details : ldiskfs mballoc code may reference an uninitialized buddy struct
5010 at startup during orphan unlinking. Instead, skip buddy update
5011 before setup, as it will be regenerated after recovery is complete.
5016 Description: If an OST is inactive, its locks might reference stale inodes.
5017 Details : lov_change_cbdata() must iterate over all namespaces, even if
5018 they are inactive to clear inode references from the lock.
5020 Severity : enhancement
5021 Frequency : occasional, if non-standard max_dirty_mb used
5023 Description: Client will block write RPCs if not enough grant
5024 Details : If a client has max_dirty_mb smaller than max_rpcs_in_flight,
5025 then the client will block writes while waiting for another RPC
5026 to complete instead of consuming its dirty limit. With change
5027 we get improved performance when max_dirty_mb is small.
5029 Severity : enhancement
5030 Bugzilla : 3389, 6253
5031 Description: Add support for supplementary groups on the MDS.
5032 Details : The MDS has an upcall /proc/fs/lustre/mds/{mds}/group_upcall
5033 (set to /usr/sbin/l_getgroups if enabled) which will do MDS-side
5034 lookups for user supplementary groups into a cache.
5038 Description: O_CREAT|O_EXCL open flags in liblustre always return -EEXIST
5039 Details : Make libsysio to not enforce O_EXCL by clearing the flag,
5040 for liblustre O_EXCL is enforced by MDS.
5044 Description: readdir never returns NULL in liblustre.
5045 Details : Corrected llu_iop_getdirentries logic, to return offset of next
5046 dentry in struct dirent.
5050 Frequency : liblustre only, depends on application IO pattern
5051 Description: liblustre clients evicted if not contacting servers
5052 Details : Don't put liblustre clients into the ping_evictor list, so
5053 they will not be evicted by the pinger ever.
5055 Severity : enhancement
5057 Description: Add ability to evict clients by NID from MDS.
5058 Details : By echoing "nid:$NID" string into
5059 /proc/fs/lustre/mds/.../evict_client client with nid that equals to
5060 $NID would be instantly evicted from this MDS and from all active
5061 OSTs connected to it.
5065 Description: Do not query file size twice, somewhat slowing stat(2) calls.
5066 Details : lookup_it_finish() used to query file size from OSTs that was not
5071 Description: service threads change working directory to that of init
5072 Details : Starting lustre service threads may pin the working directory
5073 of the parent thread, making that filesystem busy. Threads
5074 now change to the working directory of init to avoid this.
5078 Frequency : during shutdown only
5079 Description: shutdown with a failed MDS or OST can cause unmount to hang
5080 Details : Don't resend DISCONNECT messages in ptlrpc_disconnect_import()
5085 Frequency : 2.6 only
5086 Description: chmod/chown may include an extra supplementary group
5087 Details : ll{,u}_mdc_pack_op_data() does not properly initialize the
5088 supplementary group and if none is specified this is used.
5091 Bugzilla : 5479 (6816)
5093 Description: Racing open + rm can assert client in mdc_set_open_replay_data()
5094 Details : If lookup is in progress on a file that is unlinked we might try
5095 to revalidate the inode and fail in revalidate after lookup is
5096 complete and ll_file_open() enqueues the open again but
5097 it_open_error() was not checking DISP_OPEN_OPEN errors correctly.
5100 Frequency : always, if lconf --abort_recovery used
5102 Description: lconf --abort_recovery fails with 'Operation not supported'
5103 Details : lconf was attempting to abort recovery on the MDT device and not
5106 ------------------------------------------------------------------------------
5108 2005-08-08 Cluster File Systems, Inc. <info@clusterfs.com>
5113 Frequency : rare (only unsupported configurations with a node running as an
5115 Bugzilla : 6514, 5137
5116 Description: Mounting a Lustre file system on a node running as an OST could
5118 Details : OSTs now preallocates memory needed to write out data at
5119 startup, instead of when needed, to avoid having to
5120 allocate memory in possibly low memory situations.
5121 Specifically, if the file system is mounted on on OST,
5122 memory pressure could force it to try to write out data,
5123 which it needed to allocate memory to do. Due to the low
5124 memory, it would be unable to do so and the node would
5125 become unresponsive.
5127 Severity : enhancement
5129 Description: Addition of lconf --service command line option
5130 Details : lconf now accepts a '--service <arg>' option, which is
5131 shorthand for 'lconf --group <arg> --select <arg>=<hostname>'
5133 Severity : enhancement
5135 Description: Failover mode is now the default for OSTs.
5136 Details : By default, OSTs will now run in failover mode. To return to
5137 the old behaviour, add '--failout' to the lmc line for OSTs.
5139 Severity : enhancement
5141 Description: Health checks are now provided for MDS and OSTs
5142 Details : Additional detailed health check information on MSD and OSTs
5143 is now provided through the procfs health_check value.
5146 Frequency : occasional, depends on IO load
5148 Description: Disk fragmentation on the OSTs could eventually cause slowdowns
5149 after numerous create/delete cycles
5150 Details : The ext3 inode allocation policy would not allocate new inodes
5151 very well on the OSTs because there are no new directories
5152 being created. Instead we look for groups with free space if
5153 the parent directories are nearly full.
5158 Description: Network or server problems during mount may cause partially
5159 mounted clients instead of returning an error.
5160 Details : The config llog parsing code may overwrite the error return
5161 code during mount error handling, returning success instead
5167 Description: MDS can fail to allocate large reply buffers
5168 Details : After long uptimes the MDS can fail to allocate large reply
5169 buffers (e.g. zconf client mount config records) due to memory
5170 fragmentation or consumption by the buffer cache. Preallocate
5171 some large reply buffers so that these replies can be sent even
5172 under memory pressure.
5176 Frequency : rare (liblustre)
5177 Description: fsx running with liblustre complained that using truncate() to
5178 extend the file doesn't work. This patch corrects that issue.
5179 Details : This is the liblustre equivalent of the fix for bug 6196. Fixes
5180 ATTR_SIZE and lsm use in llu_setattr_raw.
5184 Frequency : rare, only 2.6 kernels
5185 Description: Unusual file access patterns on the MDS may result in inode
5186 data being lost in very rare circumstances.
5187 Details : Bad interaction between the ea-in-inode patch and the "no-read"
5188 code in the 2.6 kernel caused the inode and/or EA data not to
5189 be read from disk, causing single-file corruption.
5193 Frequency : rare, only 2.6 filesystems using extents
5194 Description: Heavy concurrent write and delete load may cause data corruption.
5195 Details : It was possible under high-load situations to have an extent
5196 metadata block in the block device cache from a just-unlinked
5197 file overwrite a newly-allocated data block. We now unmap any
5198 metadata buffers that alias just-allocated data blocks.
5202 Frequency : filesystems with default stripe_count larger than 77
5203 Description: lconf+mke2fs fail when formatting filesystem with > 77 stripes
5204 Details : lconf specifies an inode size of 4096 bytes when the default
5205 stripe_count is larger than 77. This conflicts with the default
5206 inode density of 1 per 4096 bytes. Allocate smaller inodes in
5207 this case to avoid pinning too much memory for large EAs.
5209 ------------------------------------------------------------------------------
5211 2005-07-07 Cluster File Systems, Inc. <info@clusterfs.com>
5216 Frequency : rare (extremely heavy IO load with hundreds of clients)
5218 Description: Client is evicted, gets IO error writing to file
5219 Details : lock ordering changes for bug 5492 reintroduced bug 3267 and
5220 caused clients to be evicted for AST timeouts. The fixes in
5221 bug 5192 mean we no longer need to have such short AST timeouts
5222 so ldlm_timeout has been increased.
5225 Frequency : occasional during --force or --failover shutdown under load
5226 Bugzilla : 5949, 4834
5227 Description: Server oops/LBUG if stopped with --force or --failover under load
5228 Details : a collection of import/export refcount and cleanup ordering
5229 issues fixed for safer force cleanup
5232 Frequency : only filesystems larger than 120 OSTs
5233 Bugzilla : 5990, 6223
5234 Description: lfs getstripe would oops on a very large filesystem
5235 Details : lov_getconfig used kfree on vmalloc'd memory
5238 Frequency : only filesystems exporting via NFS to Solaris 10 clients
5239 Bugzilla : 6242, 6243
5240 Description: reading from files that had been truncated to a non-zero size
5241 but never opened returned no data
5242 Details : ll_file_read() reads zeros from no-object files to EOF
5247 Description: A bug in MDS/OSS recovery could cause the OSS to fail an assertion
5248 Details : There's little harm in aborting MDS/OSS recovery and letting it
5249 try again, so I removed the LASSERT and return an error instead.
5251 Severity : enhancement
5253 Description: New debugging infrastructure for tracking down data corruption
5254 Details : The I/O checksum code was replaced to: (a) control it at runtime,
5255 (b) cover more of the client-side code path, and (c) try to narrow
5256 down where problems occurred
5260 Bugzilla : 3819, 4364, 4397, 6313
5261 Description: Racing close and eviction MDS could cause assertion in mds_close
5262 Details : It was possible to get multiple mfd references during close and
5263 client eviction, leading to one thread referencing a freed mfd.
5265 Severity: : enhancement
5266 Bugzilla : 3262, 6359
5267 Description: Attempts to reconnect to servers are now more aggressive.
5268 Details : This builds on the enhanced upcall-less recovery that was added
5269 in 1.4.2. When trying to reconnect to servers, clients will
5270 now try each server in the failover group every 10 seconds. By
5271 default, clients would previously try one server every 25 seconds.
5276 Description: After recovery, certain operations trigger a failed
5277 assertion on a client.
5278 Details : Failing over an mds, using lconf -d --failover, while a
5279 client was doing a readdir() call would cause the client to
5280 LBUG after recovery completed and the readdir() was resent.
5282 Severity : enhancement
5284 Description: Default groups are now added by lconf
5285 Details : You can now run lconf --group <servicename> without having to
5286 manually add groups with lmc.
5289 Frequency : occasional
5291 Description: Nodes with an elan id of 0 trigger a failed assertion
5294 Frequency : always when accessing e.g. tty/console device nodes
5296 Description: tty and some other devices nodes cannot be used on lustre
5297 Details : file's private_data field is used by device data and lustre
5298 values in there got lost. New field was added to struct file to
5299 store fs-specific private data.
5302 Frequency : when exporting Lustre via NFS
5304 Description: NFSD failed occasionally when looking up a path component
5305 Details : NFSD is looking up ".." which was broken in ext3 directories
5306 that had grown large enough to become hashed.
5309 Frequency : Clusters with multiple interfaces not on the same subnet
5311 Description: Nodes will repeatedly try to reconnect to an interface which it
5312 cannot reach and report an error to the log.
5313 Details : Extra peer list entries will be created by lconf with some peers
5314 unreachable. lconf now validates the peer before adding it.
5317 Frequency : Only if a default stripe is set on the filesystem root.
5319 Description: Setting a default stripe on the filesystem root prevented the
5320 filesystem from being remounted.
5321 Details : The client was sending extra request flags in the root getattr
5322 request and did not allocate a reply buffer for the dir EA.
5325 Frequency : occasional, higher if lots of files are accessed by one client
5326 Bugzilla : 6159, 6097
5327 Description: Client trips assertion regarding lsm mismatch/magic
5328 Details : While revalidating inodes the VFS looks up inodes with ifind()
5329 and in rare cases can find an inode that is being freed.
5330 The ll_test_inode() code will free the lsm during ifind()
5331 when it finds an existing inode and then the VFS later attaches
5332 this free lsm to a new inode.
5336 Bugzilla : 6422, 7030
5337 Description: MDS deadlock between mkdir and client eviction
5338 Details : Creating a new file via mkdir or mknod (starting a transaction
5339 and getting the ns lock) can deadlock with client eviction
5340 (gets ns lock and trying to finish a synchronous transaction).
5343 Frequency : occasional
5344 Description: While starting a server, the fsfilt_ext3 module could not be
5346 Details : CFS's improved ext3 filesystem is named ldiskfs for 2.6
5347 kernels. Previously, lconf would still use the ext3 name
5348 when trying to load modules. Now, it will correctly use
5349 ext3 on 2.4 and ldiskfs on 2.6.
5351 Severity : enhancement
5352 Description: The default stripe count has been changed to 1
5353 Details : The interpretation of the default stripe count (0, to lfs
5354 or lmc) has been changed to mean striping across a single
5355 OST, rather than all available. For general usage we have
5356 found a stripe count of 1 or 2 works best.
5358 Severity : enhancement
5359 Description: Add support for compiling against Cray portals.
5360 Details : Conditional compiling for some areas that are different
5364 Frequency : occasional
5365 Bugzilla : 6409, 6834
5366 Description: Creating files with an explicit stripe count may lead to
5367 a failed assertion on the MDS
5368 Details : If some OSTs are full or unavailable, creating files may
5369 trigger a failed assertion on the MDS. Now, Lustre will
5370 try to use other servers or return an error to the
5374 Frequency : occasional
5376 Description: Multiple concurrent overlapping read+write on multiple SMP nodes
5377 caused lock timeout during readahead (since 1.4.2).
5378 Details : Processes doing readahead might match a lock that hasn't been
5379 granted yet if there are overlapping and conflicting lock
5380 requests. The readahead process waits on ungranted lock
5381 (original lock is CBPENDING), while OST waits for that process
5382 to cancel CBPENDING read lock and eventually evicts client.
5384 Severity : enhancement
5386 Description: Initial enabling of flock support for clients
5387 Details : Implements fcntl advisory locking and file status functions.
5388 This feature is provided as an optional mount flag (default
5389 off), and is NOT CURRENTLY SUPPORTED. Not all types of record
5390 locking are implemented yet, and those that are are not guaranteed
5391 to be completely correct in production environments.
5392 mount -t lustre -o [flock|noflock] ...
5395 Frequency : occasional
5397 Description: OSTs running 2.4 kernels but with extents enabled might trip an
5398 assertion in the ext3 JBD (journaling) layer.
5399 Details : The b_committed_data struct is protected by the big kernel lock
5400 in 2.4 kernels, serializing journal_commit_transaction() and
5401 ext3_get_block_handle->ext3_new_block->find_next_usable_block()
5402 access to this struct. In 2.6 kernels there is finer grained
5403 locking to improve SMP performance of the JBD layer.
5407 Description: Changes the "SCSI I/O Stats" kernel patch to default to "enabled"
5412 Description: merge and cleanup kernel patches.
5415 -----------------------------------------------------------------------------
5417 2005-05-05 Cluster File Systems, Inc. <info@clusterfs.com>
5419 NOTE: Lustre 1.4.2 uses an incompatible network protocol than previous
5420 versions of Lustre. Please update all servers and clients to
5421 version 1.4.2 or later at the same time. You must also run
5422 "lconf --write-conf {config}.xml" on the MDS while it is stopped
5423 to update the configuration logs.
5425 - fix for HPUX NFS client breakage when NFS exporting Lustre (5781)
5426 - mdc_enqueue does not need max_mds_easize request buffer on send (5707)
5427 - swab llog records of type '0' so we get proper header size/idx (5861)
5428 - send llog cancel req to DLM cancel portal instead of cb portal (5515)
5429 - fix rename of one directory over another leaking an inode (5953)
5430 - avoid SetPageDirty on 2.6 (5981)
5431 - don't re-add just-being-destroyed locks to the waiting list (5653)
5432 - when creating new directories, inherit the parent's custom
5433 striping settings if present parent (3048)
5434 - flush buffers from cache before direct IO in 2.6 obdfilter (4982)
5435 - don't hold i_size_sem in ll_nopage() and ll_ap_refresh_count (6077)
5436 - don't hold client locks on temporary worklist from l_lru (5666)
5437 - handle IO errors in 2.6 obdfilter bio completion routine (6046)
5438 - automatically evict dead clients (5921)
5439 - Update file size properly in create+truncate+fstat case (6196)
5440 - Do not unhash mountpoint dentries, do not allow removal of
5442 - Avoid lock ordering deadlock issue with write/truncate (6203,5654)
5443 - reserve enough journal credits in fsfilt_start_log for setattr (4554)
5444 - ldlm_enqueue freed-export error path would always LBUG (6149,6184)
5445 - don't reference lr_lvb_data until after we hold lr_lvb_sem (6170)
5446 - don't overwrite last_rcvd if there is a *_client_add() error (6086)
5447 - Correctly handle reads of files with no objects (6243)
5448 - lctl recover will also mark a device active if deactivate used (5933)
5450 - by default create 1 inode per 4kB space on MDS, per 16kB on OSTs
5451 - allow --write-conf on an MDS with different nettype than client (5619)
5452 - don't write config llogs to MDS for mounts not from that MDS (5617)
5453 - lconf should create multiple TCP connections from a client (5201)
5454 - init scripts are now turned off by default; run chkconfig --on
5455 lustre and chkconfig --on lustrefs to use them
5456 - upcalls are no longer needed for clients to recover to failover
5458 - add --abort-recovery option to lconf to abort recovery on device
5460 - add support for an arbitrary number of OSTs (3026)
5461 - Quota support protocol changes.
5462 - forward compatibility changes to wire structs (6007)
5463 - rmmod NALs that might be loaded because of /etc/modules.conf (6133)
5464 - support for mountfsoptions and clientoptions to the Lustre LDAP (5873)
5465 - improved "lustre status" script
5466 - initialize blocksize for non-regular files (6062)
5467 - added --disable-server and --disable-client configure options (5782)
5468 - introduce a lookup cache for lconf to avoid repeated DB scans (6204)
5469 - Vanilla 2.4.29 support
5470 - increase maximum number of obd devices to 520 (6242)
5471 - remove the tcp-zero-copy patch from the suse-2.4 series (5902)
5472 - Quadrics Elan drivers are now included for the RHEL 3 2.4.21 and
5473 SLES 9 2.6.5 kernels
5474 - limit stripes per file to 160 (the maximum EA size) (6093)
5476 2005-03-22 Cluster File Systems, Inc. <info@clusterfs.com>
5479 - don't LASSERT in ll_release on NULL lld with NFS export (4655, 5760)
5480 - hold NS lock when calling handle_ast_error->del_waiting_lock (5746)
5481 - fix setattr mtime regression from lovcleanup merge (4829, 5669)
5482 - workaround for 2.6 crash in ll_unhash_aliases (5687, 5210)
5483 - small ext3 extents cleanups and fixes (5733)
5484 - improved mballoc code, several small races and bugs fixed (5733, 5638)
5485 - kernel version 43 - fix remove_suid bugs in both 2.4 and 2.6 (5695)
5486 - avoid needless client->OST connect, fix handle mismatch (5317)
5487 - fix DLM error path that led to out-of-sync client, long delays (5779)
5488 - support common vfs-enforced mount options (nodev,nosuid,noexec) (5637)
5489 - fix several locking issues related to i_size (5492,5624,5654,5672)
5490 - don't move pending lock onto export if it is already evicted (5683)
5491 - fix kernel oops when creating .foo in unlinked directory (5548)
5492 - fix deadlock in obdfilter statistics vs. object create (5811)
5493 - use time_{before,after} to avoid timer jiffies wrap (5882)
5494 - shutdown --force/--failover stability (3607,3651,4797,5203,4834)
5495 - Do not leak request if server was not able to process it (5154)
5496 - If mds_open unable to find parent dir, make that negative lookup(5154)
5497 - don't create new directories with extent-mapping (5909, 5936)
5499 - fix lustre/lustrefs init scripts for SuSE (patch from Scali, 5702)
5500 - don't hold the pinger_sem in ptlrpc_pinger_sending_on_import
5501 - change obd_increase_kms to obd_adjust_kms (up or down) (5654)
5502 - lconf, lmc search both /usr/lib and /usr/lib64 for Python libs (5800)
5503 - support for RHEL4 kernel on i686 (5773)
5504 - provide error messages when incompatible logs are encountered (5898)
5506 2005-02-18 Cluster File Systems, Inc. <info@clusterfs.com>
5507 * version 1.4.0.10 (1.4.1 release candidate 1)
5509 - don't keep a lock reference when lock is not granted (4238)
5510 - unsafe list practices (rarely) led to infinite eviction loop (4908)
5511 - add per-fs limit of Lustre pages in page cache, avoid OOM (4699)
5512 - drop import inflight refcount on signal_completed_replay error (5255)
5513 - unlock page after async write error during send (3677)
5514 - handle missing objects in filter_preprw_read properly (5265)
5515 - no transno return for symlink open, don't save no-trasno open (3440)
5516 - don't try to complete elan receive that already failed (4012)
5517 - free RPC server reply state on error (5406)
5518 - clean up thread from ptlrpc_start_thread() on error (5160)
5519 - readahead could read extra page into cache that wasn't ejected (5388)
5520 - prevent races in class_attach/setup/cleanup/detach (5260)
5521 - don't dereference de->d_inode after l_dput of de (5458)
5522 - use "int" for stripe value returned from lock_to_stripe (5544)
5523 - mballoc allocation and error-checking fixes in 2.6 (5504)
5524 - block device patches to fix I/O request sizes in 2.6 (5482)
5525 - look up hostnames for IB nals (5602)
5526 - 2.6 changed lock ordering of 2 semaphores, caused deadlock (5654)
5527 - don't start multiple acceptors for the same port (5277)
5528 - fix incorrect LASSERT in mds_getattr_name (5635)
5529 - export a proc file for general "ping" checking (5628)
5530 - fix "lfs check" to not block when the MDS is down (5628)
5532 - service request history (4965)
5533 - put {ll,lov,osc}_async_page structs in a single slab (4699)
5534 - create an "evict_client" /proc entry on OSTs, like the MDS has
5535 - fix mount usage message, return errors per mount(8) (5168)
5536 - change grep [] to grep "[]" in tests so they work in more UMLs
5537 - fix ppc64/x86_64 spec to use %{_libdir} instead of /usr/lib (5389)
5538 - remove ancient LOV_MAGIC_V0 EA support (5047)
5539 - add "disk I/Os in flight" and "I/O req time" stats in obdfilter
5540 - align r/w RPCs to PTLRPC_MAX_BRW_SIZE boundary for performance (3451)
5541 - allow readahead allocations to fail when low on memory (5383)
5542 - mmap locking landed again, after considerable improvement (2828)
5543 - add get_hostaddr() to lustreDB.py for LDAP support (5459)
5545 2004-11-23 Cluster File Systems, Inc. <info@clusterfs.com>
5548 - send OST transaction number in read/write reply to free req (4966)
5549 - don't ASSERT in ptl_send_rpc() if we run out of memory (5119)
5550 - lock /proc/sys/portals/routes internal state, avoiding oops (4827)
5551 - the watchdog thread now runs as interruptible (5246)
5552 - flock/lockf fixes (but it's still disabled, pending 5135)
5553 - don't use EXT3 constants in llite code (5094)
5554 - memory shortage at startup could cause assertion (5176)
5556 - reorganization of lov code
5557 - single portals codebase
5559 - add extents/mballoc support (5025)
5560 - direct I/O reads in the obdfilter (4048)
5561 - kernel patches from LNXI for 2.6 (bluesmoke, perfctr, mtd, kexec)
5563 tbd Cluster File Systems, Inc. <info@clusterfs.com>
5566 - send OST transaction number in read/write reply to free req (4966)
5567 - don't ASSERT in ptl_send_rpc() if we run out of memory (5119)
5568 - lock /proc/sys/portals/routes internal state, avoiding oops (4827)
5569 - the watchdog thread now runs as interruptible (5246)
5570 - handle missing objects in filter_preprw_read properly (5265)
5571 - unsafe list practices (rarely) led to infinite eviction loop (4908)
5572 - drop import inflight refcount on signal_completed_replay error (5255)
5573 - unlock page after async write error during send (3677)
5574 - return original error code on reconstructed replies (3761)
5575 - no transno return for symlink open, don't save no-trasno open (3440)
5577 - add pid to ldlm debugging output (4922)
5578 - bump the watchdog timeouts -- we can't handle 30sec yet
5579 - extra debugging for orphan dentry/inode bug (5259)
5581 2004-11-16 Cluster File Systems, Inc. <info@clusterfs.com>
5584 - fix TCP_NODELAY bug, which caused extreme perf regression (5134)
5585 - allocate qswnal tx descriptors singly to avoid fragmentation (4504)
5586 - don't LBUG on obdo_alloc() failure, use OBD_SLAB_ALLOC() (4800)
5587 - fix NULL dereference in /proc/sys/portals/routes (4827)
5588 - allow failed mdc_close() operations to be interrupted (4561)
5589 - stop precreate on OST before MDS would time out on it (4778)
5590 - don't send partial-page writes before EOF from client (4410)
5591 - discard client grant for sub-page writes on large-page clients (4520)
5592 - don't free dentries not owned by NFS code, check generation (4806)
5593 - fix lsm leak if mds_create_objects() fails (4801)
5594 - limit debug_daemon file size, always print CERROR messages (4789)
5595 - use transno after validating reply (3892)
5596 - process timed out requests if import state changes (3754)
5597 - update mtime on OST during writes, return in glimpse (4829)
5598 - add mkfsoptions to LDAP (4679)
5599 - use ->max_readahead method instead of zapping global ra (5039)
5600 - don't interrupt __l_wait_event() during strace
5602 - add software watchdogs to catch hung threads quickly (4941)
5603 - make lustrefs init script start after nfs is mounted
5604 - fix CWARN/ERROR duplication (4930)
5605 - return async write errors to application if possible (2248)
5606 - add /proc/sys/portal/memused (bytes allocated by PORTALS_ALLOC)
5607 - print NAL number in %x format (4645)
5608 - update barely-supported suse-2.4.21-171 series (4842)
5609 - support for sles 9 %post scripts
5610 - support for building 2.6 kernel-source packages
5611 - support for sles km_* packages
5613 2004-10-07 Cluster File Systems, Inc. <info@clusterfs.com>
5616 - ignore -ENOENT errors in osc_destroy (3639)
5617 - notify osc create thread that OSC is being cleaned up (4600)
5618 - add nettype argument for llmount in #5d in conf-sanity.sh (3936)
5619 - reconstruct ost_handle() like mds_handle() (4657)
5620 - create a new thread to do import eviction to avoid deadlock (3969)
5621 - let lconf resolve symlinked-to devices (4629)
5622 - don't unlink "objects" from directory with default EA (4554)
5623 - hold socknal file ref over connect in case target is down (4394)
5624 - allow more than 32000 subdirectories in a single directory (3244)
5625 - fix blocks count for O_DIRECT writes (3751)
5626 - OST returns ENOSPC from object create when no space left (4539)
5627 - don't send truncate RPC if file size isn't changing (4410)
5628 - limit OSC precreate to 1/2 of value OST considers bogus (4778)
5629 - bind to privileged port in socknal and tcpnal (3689)
5631 - rate limit CERROR/CWARN console message to avoid overload (4519)
5632 - GETFILEINFO dir ioctl returns LOV EA + MDS stat in 1 call (3327)
5633 - basic mmap support (3918)
5634 - kernel patch series update from b1_4 (4711)
5636 2004-09-16 Cluster File Systems, Inc. <info@clusterfs.com>
5639 - avoid crash during MDS cleanup with OST shut down (2775)
5640 - fix loi_list_lock/oig_lock inversion on interrupted IO (4136)
5641 - don't use bad inodes on the MDS (3744)
5642 - dynamic object preallocation to improve recovery speed (4236)
5643 - don't hold spinlock over lock dumping or change debug flags (4401)
5644 - don't zero obd_dev when it is force cleaned (3651)
5645 - print grants to console if they go negative (4431)
5646 - "lctl deactivate" will stop automatic recovery attempts (3406)
5647 - look for existing locks in ldlm_handle_enqueue() (3764)
5648 - don't resolve lock handle twice in recovery avoiding race (4401)
5649 - revalidate should check working dir is a directory (4134)
5651 - don't always mark "slow" obdfilter messages as errors (4418)
5653 2004-08-24 Cluster File Systems, Inc. <info@clusterfs.com>
5656 - don't close LustreDB during write_conf until it is done (3860)
5657 - fix typo in lconf for_each_profile (3821)
5658 - allow dumping logs from multiple threads at one time (3820)
5659 - don't allow multiple threads in OSC recovery (3812)
5660 - fix debug_size parameters (3864)
5661 - fix mds_postrecov to initialize import for llog ctxt (3121)
5662 - replace config semaphore with spinlock (3306)
5663 - be sure to send a reply for a CANCEL rpc with bad export (3863)
5664 - don't allow enqueue to complete on a destroyed export (3822)
5665 - down write_lock before checking llog header bitmap (3825)
5666 - recover from lock replay timeout (3764)
5667 - up llog sem before sending rpc (3652)
5668 - reduce ns lock hold times when setting kms (3267)
5669 - change a dlm LBUG to LASSERTF, to maybe learn something (4228)
5670 - fix NULL deref and obd_dev leak on setup error (3312)
5671 - replace some LBUG about llog ops with error handling (3841)
5672 - don't match INVALID dentries from d_lookup and spin (3784)
5673 - hold dcache_lock while marking dentries INVALID and hashing (4255)
5674 - fix invalid assertion in ptlrpc_set_wait (3880)
5676 - add libwrap support for the TCP acceptor (3996)
5677 - add /proc/sys/portals/routes for non-root route listing (3994)
5678 - allow setting MDS UUID in .xml (2580)
5679 - print the stack of a process that LBUGs (4228)
5681 2004-07-14 Cluster File Systems, Inc. <info@clusterfs.com>
5684 - don't cleanup request in ll_file_open() on failed MDS open (3430)
5685 - make sure to unset replay flag from failed open requests (3440)
5686 - if default stripe count is 0, use OST count for inode size (3636)
5687 - update parent mtime/ctime on client for create/unlink (2611)
5688 - drop dentry ref in ext3_add_link from open_connect_dentry (3266)
5689 - free recovery state on server during a forced cleanup (3571)
5690 - unregister_reply for resent reqs (3063)
5691 - loop back devices mounting and status check on 2.6 (3563)
5692 - fix resource-creation race that can provoke i_size == 0 (3513)
5693 - don't try to use bad inodes returned from MDS/OST fs lookup (3688)
5694 - more debugging for page-accounting assertion (3746)
5695 - return -ENOENT instead of asserting if ost getattr+unlink race (3558)
5696 - avoid deadlock after precreation failure (3758)
5697 - fix race and lock order deadlock in orphan handling (3450, 3750)
5698 - add validity checks when grabbing inodes from l_ast_data (3599)
5700 - add /proc/.../recovery_status to obdfilter (3428)
5701 - lightweight CDEBUG infrastructure, debug daemon (3668)
5702 - change default OSC RPC parameters to be better on small clusters
5703 - turn off OST read cache for files smaller than 32MB
5704 - install man pages and include them in rpms (3100)
5705 - add new init script for (un)mounting lustre filesystems (2593)
5706 - run chkconfig in %post for init scripts (3701)
5707 - drop scimac NAL (unmaintained)
5709 2004-06-17 Cluster File Systems, Inc. <info@clusterfs.com>
5712 - clean kiobufs before and after use (3485)
5713 - strip trailing '/'s before comparing paths with /proc/mounts (3486)
5714 - remove assertions to work around "in-flight rpcs" recovery bug (3063)
5715 - change init script to fail more clearly if not run as root (1528)
5716 - allow clients to reconnect during replay (1742)
5717 - fix ns_lock/i_sem lock ordering deadlock for kms update (3477)
5718 - don't do DNS lookups on NIDs too small for IP addresses (3442)
5719 - re-awaken ptlrpcd if new requests arrive during check_set (3554)
5720 - fix cond_resched (3554)
5721 - only evict unfinished clients after recovery (3515)
5722 - allow bulk resend, prevent data loss (3570)
5723 - dynamic ptlrpc request buffer allocation (2102)
5724 - don't allow unlinking open directory if it isn't empty (2904)
5725 - set MDS/OST threads to umask 0 to not clobber client modes (3359)
5726 - remove extraneous obd dereference causing LASSERT failure (3334)
5727 - don't use get_cycles() when creating temp. files on the mds (3156)
5728 - hold i_sem when setting i_size in ll_extent_lock() (3564)
5729 - handle EEXIST for set-stripe, set proper directory name (3336)
5731 - servers can dump a log evicting a client - lustre.dump_on_timeout=1
5732 - fix ksocknal_fmb_callback() error messages (2918)
5734 2004-05-27 Cluster File Systems, Inc. <info@clusterfs.com>
5737 - don't copy lvb into (possibly NULL) reply on error (2983)
5738 - don't deref dentry after dput, don't free lvb on error (2922)
5739 - use the kms to determine writeback rpc length (2947)
5740 - increment oti_logcookies when osc is inactive (2948)
5741 - update client's i_blocks count via lvb messages (2543)
5742 - handle intent open/close of special files properly (1557)
5743 - mount MDS with errors=remount-ro, like obdfilter (2009)
5744 - initialize lock handle to avoid ASSERT on error cleanup (3057)
5745 - don't use cancelling-locks' kms values (2947)
5746 - use highest lock extent for kms, not last one (2925)
5747 - don't dereference ERR_PTR() dentry in error handling path (3107)
5748 - fix thread race in portals_debug_dumplog() (3122)
5749 - create lprocfs device entries at setup instead of at attach (1519)
5750 - common AST error handler, don't evict client on completion race (3145)
5751 - zero nameidata in detach_mnt in 2.6 (3118)
5752 - verify d_inode after revalidate_special is valid in 2.6 (3116)
5753 - use lustre_put_super() to handle zconf unmounts in 2.6 (3064)
5754 - initialize RPC timeout timer earlier for 2.6 (3219)
5755 - don't dereference NULL reply buffer if mdc_close was never sent (2410)
5756 - print nal/nid for unknown nid (3258)
5757 - additional checks for oscc recovery before doing precreate (3284)
5758 - fix ll_extent_lock() error return code for 64-bit systems (3043)
5759 - don't crash in mdc_close for bad permissions on open (3285)
5760 - zero i_rdev for non-device files (3147)
5761 - clear page->private before handing to FS, better assertion (3119)
5762 - tune the read pipeline (3236)
5763 - fix incorrect decref of invalidated dentry (2350)
5764 - provide read-ahead stats and refine rpc in flight stats (3328)
5765 - don't hold journal transaction open across create RPC (3313)
5766 - update atime on MDS at close time (3265)
5767 - close LDAP connection when recovering to avoid server load (3315)
5768 - update iopen-2.6 patch with fixes from 2399,2517,2904 (3301)
5769 - don't leak open file on MDS after open resend (3325)
5770 - serialize filter_precreate and filter_destroy_precreated (3329)
5771 - loop device shouldn't call sync_dev() for nul device (3092)
5772 - clear page cache after eviction (2766)
5773 - resynchronize MDS->OST in background (2824)
5774 - refuse to mount the same filesystem twice on same mountpoint (3394)
5775 - allow llmount to create routes for mounting behind routers (3320)
5776 - push lock cancellation to blocking thread for glimpse ASTs (3409)
5777 - don't call osc_set_data_with_check() for TEST_LOCK matches (3159)
5778 - fix rare problem with rename on htree directories (3417)
5780 - allow default OST striping configuration per directory (1414)
5781 - fix compilation for qswnal for 2.6 kernels (3125)
5782 - increase maximum number of MDS request buffers for large systems
5783 - change liblustreapi to be useful for external progs like lfsck (3098)
5784 - increase local configuration timeout for slow disks (3353)
5785 - allow configuring ldlm AST timeout - lustre.ldlm_timeout=<seconds>
5787 2004-03-22 Cluster File Systems, Inc. <info@clusterfs.com>
5790 - fixes for glimpse AST timeouts / incorrectly 0-sized files (2818)
5791 - don't overwrite extent policy data in reply if lock was blocked (2901)
5792 - drop filter export grants atomically with removal from device (2663)
5793 - del obd_self_export from work_list in class_disconnect_exports (2908)
5794 - don't LBUG if MDS recovery times out during orphan cleanup (2530)
5795 - swab reply message in mdc_close, other PPC fixes (2464)
5796 - fix destroying of named logs (2325)
5797 - overwrite old logs when running lconf --write_conf (2264)
5798 - bump LLOG_CHUNKSIZE to 8k to allow for larger clusters (2306)
5799 - fix race in target_handle_connect (2898)
5800 - mds_reint_create() should take same inode create lock (2926)
5801 - correct journal credits calculated for CANCEL_UNLINK_LOG (2931)
5802 - don't close files for self_export to avoid uninitialized obd (2936)
5803 - allow MDS with the same name as client node (2939)
5804 - hold dentry reference for closed log files for unlink (2325)
5805 - reserve space for all logs during transactions (2059)
5806 - don't evict page beyond end of stripe extent (2925)
5807 - don't oops on a deleted current working directory (2399)
5808 - handle hard links to targets without a parent properly (2517)
5809 - don't dereference NULL lock when racing during eviction (2867)
5810 - don't grow lock extents when lots of conflicting locks (2919)
5812 2004-03-04 Cluster File Systems, Inc. <info@clusterfs.com>
5815 - account for cache space usage on clients to avoid data loss (974)
5816 - lfsck support in lustre kernel code (2349)
5817 - reduce journal credits needed for BRW writes (2370)
5818 - orphan handling to avoid losing space on client/server crashes
5819 - ptlrpcd can be blocked, stopping ALL progress (2477)
5820 - use lock value blocks to assist in proper KMS, faster stat (1021)
5821 - takes i_sem instead of DLM locks internally on obdfilter (2720)
5822 - recovery for initial connections (2355)
5823 - fixes for mds_cleanup_orphans (1934)
5824 - abort_recovery crashes MDS in b_eq (mds_unlink_orphan) (2584)
5825 - block all file creations until orphan recovery completes (1901)
5826 - client remove rq_connection from request struct (2423)
5827 - conf-sanity test_5, proper cleanup in umount log not availale (2640)
5828 - recovery timer race (2670)
5829 - mdc_close recovey bug (2532)
5830 - ptlrpc cleanup bug (2710)
5831 - mds timeout on local locks (2588)
5832 - namespace lock held during RPCs (2431)
5833 - handle interrupted sync write properly (2503)
5834 - don't try to handle a message that hasn't been replied to (2699)
5835 - client assert failure during cleanup after abort recovery (2701)
5836 - leak mdc device after failed mount (2712)
5837 - ptlrpc_check_set allows timedout requests to complete (2714)
5838 - wait for inflight reqs when ptlrpcd finishes (2710)
5839 - make sure unregistered services are removed from the srv_list
5840 - reset bulk XID's when resending them (caught by 1138 test)
5841 - unregister_bulk after timeout
5842 - fix lconf error (2694)
5843 - handle write after unfinished setstripe, stripe-only getstripe (2388)
5844 - readahead locks pages, leaves pending causing memory pressure (2673)
5845 - increase OST request buffers to 4096 on large machines (2729)
5846 - fix up permission of existing directories in simple_mkdir (2661)
5847 - init deleted item, add assertions ptlrpc_abort_inflight() (2725)
5848 - don't assign transno to errored transactions (2742)
5849 - don't delete objects on OST if given a bogus objid from MDS (2751)
5850 - handle large client PAGE_SIZE readdir on small PAGE_SIZE MDS (2777)
5851 - if rq_no_resend, then timeout request after recovery (2432)
5852 - fix MDS llog_logid record size, 64-bit array alignment (2733)
5853 - don't call usermode_helper from ptlrpcd, DEFAULT upcall (2773)
5854 - put magic in mount.lustre data, check for bad/NULL mount data (2529)
5855 - MDS recovery shouldn't delete objects that it has given out (2730)
5856 - if enqueue arrives after completion, don't clobber LVB (2819)
5857 - don't unlock pages twice when trigger_group_io returns error (2814)
5858 - don't deref NULL rq_repmsg if ldlm_handle_enqueue failed (2822)
5859 - don't write pages to disk if there was an error (1450)
5860 - don't ping imports that have recovery disabled (2676)
5861 - take buffered bytes into account when balancing socknal conn (2817)
5862 - hold a DLM lock over readdir always, use truncate_inode_pages (2706)
5863 - reconnect unlink llog connection after MDS reconnects to OST (2816)
5864 - remove little-endian swabbing of llog records (1987)
5865 - set/limit i_blksize to LL_MAX_BLKSIZE on client (2884)
5866 - retry reposting request buffers if they fail (1191)
5867 - grow extent at grant time to avoid granting a revoked lock (2809)
5868 - lock revoke doesn't evict page if covered by a second lock (2765)
5869 - disable VM readahead to avoid reading outside lock extents (2805)
5871 - return LL_SUPER_MAGIC from statfs for the filesystem type (1972)
5872 - updated kernel patches for hp-2.4.20 kernel (2681)
5874 2004-02-07 Cluster File Systems, Inc. <info@clusterfs.com>
5877 - fix truncated write corruption (2366)
5878 - fix for failed assertion in iopen_connect_dentry (1792,2517)
5880 - don't flag the ptlrpcd thread with PF_MEMALLOC (2636)
5881 - ensure len(uuid) < 37 in lmc (1171)
5882 - fix ia64 OOPS in llog_test (2255)
5883 - zero end of page at obdfilter for partial page writes (2648)
5884 - don't leave stale dentries around after renames (bug 2428)
5885 - fix timeouts when evicting a client with a single lock held (2642)
5886 - set deadline for the initial HELLO message to drain (2634)
5887 - print out dotted-quad IP addresses in the socknal (2302)
5889 - additional debugging for MDS client eviction problem (2443)
5890 - fix mkfsoptions support for osts (2603, 2604)
5892 2004-01-27 Cluster File Systems, Inc. <info@clusterfs.com>
5895 - add series for the vanilla 2.6.0 kernel
5896 - add series for the vanilla 2.4.24 kernel
5897 - add series for a cray x86/64 UL kernel drop
5898 - fix xattr patches for the vanilla 2.4.19 series
5900 - generate true UUIDs in lmc (1171)
5901 - have portals stack dumping break in UML (2466)
5902 - avoid bad dchild deref; avoid inum lock w/o creation (2362)
5903 - allocate with _NOFS in ldlm to avoid deadlock (1933)
5904 - wake callback waiting threads on client eviction (2460)
5905 - Add --ptldebug and --subsystem to lmc (1719)
5906 - update assertion to allow safe interrupt allocation
5907 - set rq_no_resend for cancel requests (2432)
5908 - recalculate ptlrpcd timeout after resend (2494)
5909 - call vfs_rmdir when removing pending directories (2368)
5910 - fix renaming a file to itself (2429)
5911 - lmc creates a default one-stripe lov (2454)
5912 - expand procfs space to handle large clusters (2326)
5913 - increase UML stack to avoid overflow
5914 - update lconf's list of debug and subsystem masks
5915 - fix lfs find --obd (2510)
5916 - /proc tunable for disabling filter read caching (2591)
5917 - stop rpm packages from altering slapd.conf (2301)
5918 - disable nagle in the socknal under 0conf (2578)
5919 - choose mds inode size based on stripe count (2572)
5920 - fix kernel-source rpm problems (2516)
5922 - add --disable-doc to avoid pdf generation (2421)
5923 - update documentation, tests, type-os, comments
5924 - avoid format warnings on ia64
5925 - remove the TOE NAL
5926 - tiny code cleanups by removing unused fields
5928 2004-01-07 Cluster File Systems, Inc. <info@clusterfs.com>
5931 - fix obvious semaphore misuse in as-yet-unused setattr path (2348)
5932 - remove the most blatant lies from BUILDING file (2371)
5933 - change default debug level to reasonable production setting
5934 - reduce client side cache size to reduce cache flush time
5935 - reduce max RPCs in flight to avoid unnecessary file fragmentation
5936 - make TCP zerocopy and pinger support enabled by default (2476)
5937 - sync writes completed after process exits caused crashes (2319)
5938 - maintain correct mount count on the MDS (2356)
5939 - backout 1557, because 2316 wasn't really fixed
5940 - better file I/O statistics gathering in /proc
5941 - don't take unnecessary, deadlock-inducing bug in readpage (2383)
5942 - another kernel patch to fix zero-copy TCP function export
5943 - don't take duplicate lock when processing re-sent getattr (2420)
5944 - lctl uses obd_self_export instead of creating new conn (2353)
5945 - MDS/OST recovery case which requires object creation asserted (2425)
5946 - move lfs from /usr/sbin to /usr/bin in packages
5947 - fix race between mds_client_add and mds_client_free (2417)
5948 - use kmalloc instead of slabs in portals (2430)
5949 - don't create duplicate records when a failover MDS is present (2442)
5950 - remove unnecessary mount age check (2332)
5951 - don't remove directory inodes from locks prematurely (2451)
5952 - don't break if MDS service name is the same as hostname (2103)
5953 - fix races in client write RPC generation when cache full (2482)
5955 2003-12-13 Cluster File Systems, Inc. <info@clusterfs.com>
5958 - remove now-unused request->rq_obd (278)
5959 - if an allocation fails, print out how much memory we've used (1933)
5960 - use PORTAL_SLAB_ALLOC for structures, to get GFP_MEMALLOC (1933)
5961 - add the "configurable stack size" patch to most series files (1256)
5962 - ability to write large log records, for 100+ OST configs (2306)
5963 - fix NULL deref when filter_prep fails (2314)
5964 - fix operator precedence error in filter_sync
5965 - dynamic allocation of socknal TX descriptors (2315)
5966 - fix a missed case in the GFP_MEMALLOC patch, can cause deadlock (2310)
5967 - fix gcc 2.96 compilation problem in xattr kernel patch (2294)
5968 - ensure that CWARN messages in Portals always get to the syslog
5969 - __init/__exit are not for prototype decls (ldlm_init/exit)
5970 - x86-64 compile warning fixes
5971 - fix gateway LMC keyword conflict (2318)
5972 - fix MDS lock inversions in getattr/reint paths (1844)
5973 - fix a rare lock re-ordering bug, which caused deadlock (2322)
5974 - fix i_sem/journal inversion in fsfilt_ext3_write_record (2306)
5975 - DLM race condition prevented some lock evictions (2328)
5976 - ENOMEM detection and retry on socknal sends (2230)
5977 - use GFP_NOFS throughout Lustre, to combat ENOMEM (2230)
5978 - move osc_rpcd into ptlrpc, for use in MDC and others (2329)
5979 - protect MDS inode fsdata with stronger locking; fixes assertion (2313)
5980 - better error messages when a client is rejected during recovery (1505)
5981 - avoid cancelling locks which were never granted, after failure (2330)
5982 - fix i_sem/journal inversion in mds_client_add (2333)
5983 - fix truncate/getattr lock cycle deadlock (2334)
5984 - use rpcd to send close; allows resend after timeout, avoid leak (1897)
5985 - fix two rare exit paths which could leak an l_lock() ref (2321)
5986 - fencepost error in MDS/OST orphan recovery (2226)
5987 - make log record alignment 8 bytes (1988)
5988 - lstripe now fails when requested offset > ost_count (2237)
5989 - ensure that all kernel series have a complete list.h (1607)
5990 - fix crashes in special-file operations (2316)
5991 - lctl create/brw OID mismatch, caused by obsolete filter loop (2339)
5993 - allow configurable automake binary, for testing new versions
5994 - small update to the lfs documentation
5996 2003-12-03 Cluster File Systems, Inc. <info@clusterfs.com>
5998 * fix negative export reference count in fsfilt_sync (2312)
6000 2003-12-01 Cluster File Systems, Inc. <info@clusterfs.com>
6001 * release candidate 0.9.1
6003 - orphans are moved into the PENDING directory for possible recovery
6004 - replayed opens now open by fid for orphan/rename safety (1042)
6005 - last close of an orphan inode generates a transno (683)
6006 - chdir() and mount() now pin the directory entry (1020)
6007 - avoid CERROR in normal ll_setattr_raw() error case (1500)
6008 - discard very old requests without processing them (1502)
6009 - remove some common, well-understood CERRORs (1505)
6010 - require O_DIRECT I/O to be page-sized to workaround IA64 crash (1609)
6011 - clear "grant" flags in OST replies until OST grant code lands (1644)
6012 - fix read performance by not clobbering i_blksize on client (1598)
6013 - fix __ldlm_handle2lock oops by not dereferencing lock after PUT (1625)
6014 - make LRU size a /proc tunable, clears locks when reduced (707)
6015 - fix some lprocfs rot that prevented ptlbd from loading (1732)
6016 - server locks take references on exports now (1558)
6017 - build fixes for 2.4.20-rh trees (1663)
6018 - return an error from lov_create if all OSCs are inactive (1751)
6019 - fix import levels when a reconnect happens without a timeout (1597)
6020 - exit early from mds_open if we get a lookup error (1749)
6021 - partial page read at EOF wouldn't wait for disk before sending (1642)
6022 - avoid NULL deref in obdfilter when reading page past EOF (1592)
6023 - avoid LASSERT in ll_intent_lock if server failed very early (1090)
6024 - fix LBUG in ll_it_open_error with rc = -2 (1861)
6025 - write/truncate lock inversion (1639)
6026 - Don't auto-load obdclass, portals modules during cleanup (1495)
6027 - fix timestamps from jumping to "now" (1763)
6028 - extra journal assertions (1648)
6029 - add an extra multiunlink test (1771)
6030 - fix read_record/write_record API (1776)
6031 - fix leak of offset_extent, possible incorrect i_size later (1772)
6032 - fix lasserts in mis-matched transnos during open-unlink testing (1541)
6033 - Debugging for the kqswnal_get_idle_tx problems (1820)
6034 - Allow recovery to be attempted multiple times (1536)
6035 - Write out MDS last_rcvd file after it is first created (1600)
6036 - Fix tx_descriptor leak in failed transmit situations (1827)
6037 - ext3 journaling fixes for assertion failure after IO error (1871)
6038 - class_export_put() on freed export after completion AST error (1896)
6039 - Fix revalidate looping in VFS (1322)
6040 - Don't access a freed export during MDS_REINT timeout (1521)
6041 - Add open-unlink recovery support on the MDS (1673,1764)
6042 - Return an error if no MDS data was read from last_rcvd (1946)
6043 - Fix for lookup "." or ".." crash on error (1932,1931,1935)
6044 - Don't setup a disk device that doesn't match exported UUID (317)
6045 - Reduce bulk RPC timeout to avoid cascading client/OST failures (1845)
6046 - avoid committing NULL handle in force close
6047 - local.sh is now a one-stripe LOV configuration
6048 - POSIX utime.4 -EPERM on FIFO not owned by user (56)
6049 - fix ext3 htree duplicate directory entry corruption (1516)
6050 - POSIX creat.13, fstat.1, open.18, stat.3 new file atime/mtime (2020)
6051 - update to new LOV EA format (2097)
6052 - interoperability for different PAGE_SIZE/wordsize (686,1821,1343,2042)
6054 2003-06-15 Phil Schwan <phil@clusterfs.com>
6057 - imports and exports cleanup too early, need refcounts (349, 879, 1045)
6058 - per-import/export recovery handling (958, 931, 959)
6059 - multiple last-rcvd slots, for serving multiple FSes (949)
6060 - connections are again shared between multiple imp/exports (963, 964)
6061 - "umount -f" would hang if any requests needed to be sent (393, 978)
6062 - avoid pinning large req buffer by copying for queued messages (989)
6063 - add "uuid" to "lctl device" command to help upcalls (991)
6064 - "open" RPCs with transnos would confuse recovery counters (1037)
6065 - do proper endian conversion of all wire messages (288, 340, 891)
6066 - remove OST bulk get LBUGs, fix ost_brw_write cleanup (1126)
6067 - call waiting locks callback from LDLM recovery thread (1127, 1151)
6068 - fix ptlrpc_connection leak in target_handle_connect (1174)
6069 - fix import refcounting bug in OST and MDS cleanup (1134)
6070 - if an invalid-at-open-time OSC returned before close(), LBUG (1150)
6071 - fix very unlikely obd_types race condition (501)
6072 - remove osc_open hack for echo_client (1187)
6073 - we leaked exports/dlmimps for forcibly disconnected clients (1143)
6074 - a failure in read_inode2 leads to deadlock (1139)
6075 - cancel ack-locks as soon as transaction is committed (1072)
6076 - fix major leaks and crashes in the bulk I/O path (937, 1057)
6077 - make sure to commitrw after any preprw to avoid deadlock (1162)
6078 - failing to execute a file in a lustre FS would lock inode (1203)
6079 - small DEBUG_REQ fix to avoid dereferencing a NULL (1227)
6080 - don't ASSERT while cleaning up an incompletely-setup obd (1248)
6081 - obd_uuid2tgt would walk off the end of the list (1255)
6082 - on IA64 the osc would give portals incorrect bulk size (1258)
6083 - fix debug daemon ioctl interface; allows daemon on ia64 (1274)
6084 - fix lock inversion caused by new llite matching code (1282)
6085 - limit the number of dirty pages on a client to 10MB (1286)
6086 - timed out locks were not being corrected cancelled (1289)
6087 - fix O_DIRECT above 4GB on IA-32 (1292)
6088 * major user-visible changes
6089 - fail out/fail over policy now controlled by the upcall (993)
6091 - add OBD_PING to check server availability and failure (954)
6092 - lustre messages are now sent in sending host order (288, 340, 891)
6093 - add eadatalen to MDS getattr reply (340)
6094 - OST read replies may contain second buffer, with per-page status (593)
6096 2003-03-11 Phil Schwan <phil@clusterfs.com>
6099 - LDLM_DEBUG macro fix, for gcc 3.2 (850)
6100 - failed open()s could cause deadlock; fixed (867, 869)
6101 - stop cancelling OST locks when files are closed (481)
6102 - overlapping XID spaces caused network corruption (851, 853)
6103 - fix unsafe fsfilt counter arithmetic; change to atomic_t
6104 - setattr_raw added, to do single-RPC, server-side setattrs
6105 - lmc/lconf syntax change for OST UUIDs
6106 - fix crashy race condition between ptlrpc_free_req and osc_close
6107 - don't use request in mdc_enqueue if we hit a timeout (889)
6108 - don't set the inode i_size for regular files from the MDS (896)
6109 - handle out of order completion AST (842)
6110 - don't LBUG if a lock request times out after receiving AST (913)
6111 - avoid d_rehash race in ll_find_alias by rehashing inside dcache_lock
6112 - if a bad lock AST arrives, send an error instead of dropping entirely
6113 - return 0 from revalidate2 if ll_intent_lock returns -EINTR (912)
6114 - fix leak in bulk IO when only partially completed (899, 900, 926)
6115 - fix O_DIRECT for ia64 (55)
6116 - (almost) eliminate Lustre-kernel-thread effects on load average (722)
6117 - C-z after timeout could hang a process forever; fixed (977)
6119 - client-side I/O cache (678, 924, 929, 941, 970)
6121 - READPAGE and SETATTRs which don't take server-side locks get
6124 2003-02-11 Phil Schwan <phil@clusterfs.com>
6127 - Fix ldlm_lock_match on the MDS to avoid matching remote locks (592)
6128 - Fix fsfilt_extN_readpage() to read a full page of directory
6129 entries, or fake the remainder if PAGE_SIZE != blocksize (500)
6130 - Avoid extra mdc_getattr() in ll_intent_lock when possible (534, 604)
6131 - Fix imbalanced LOV object allocation and out-of-bound access (469)
6132 - Most intent operations were removed, in favour of a new RPC mode
6133 that does a single RPC to the server and bypasses most of the VFS
6134 - All LDLM resource ID arrays were removed in favour of ldlm_res_id
6135 - Aggressively cancel local locks on DLM servers
6136 - mds_reint_unlink sends EA to the client if it's the last nlink.
6137 client uses that EA to unlink OST objects.
6138 - mds_reint_{rename,unlink,link} were rewritten to take ordered locks
6139 - recursive symlinks were fixed (439)
6140 - fixed NULL deref in DEBUG_REQ
6141 - filter_update_lastobjid no longer calls sync, which annoyed extN
6142 - fixed multi-client small-writes to a single file problem (445)
6143 - fixed mtime updates during file writes (607)
6144 - fixed vector writes on obdfilter causing problems when ENOSPC (670)
6145 - fixed bug in obd_brw_read/write() (under guise of testing 367)
6146 - fixed Linux OST size reporting problem (444, 656)
6147 - OST now updates object mtime with writes or setattr (607, 619)
6148 - client verifies file size before zeroing page past EOF (445)
6149 - OST now writes last allocated objid to disk with allocation (108)
6150 - LOV on echo now works (409)
6152 - mds_reint_unlink sends a new buffer, with the EA included. this
6153 buffer is only valid if body->valid & OBD_MD_FLEASIZE, which is only
6154 set if a regular file was being unlinked, and it was the last link
6155 - use PtlGet from the target for bulk writes (315)
6156 - OST now updates object mtime with writes or setattr (607, 619)
6157 - LDLM now has a grant-time callback to revalidate locked items, if
6159 - Many MDS operations were reorganized to combat race conditions
6161 - Merge b_intel branch (updated lprocfs code) - now at /proc/fs/lustre
6162 - configure check to avoid gcc version 2.96 20000731-2.96-98) (606)
6164 2003-01-06 Andreas Dilger <adilger@clusterfs.com>
6167 - Fully reactivate OST imports after reconnection (512, others)
6168 - Make sure client sees our -ENOTCONN from mds_handle (513 - partial)
6169 - More graceful error handling for truncating on dead OST (515)
6170 - Don't error out unless we're actually accessing dead stripes (474)
6171 - Fix garbage sizes when stripes are missing (410)
6172 - LRU counters were broken, causing constant lock purge (433, 432)
6173 - garbage on read from stripes with failed OSTs (441)
6174 - mark OSCs as active before reconnecting during recovery (438)
6175 - lov_enqueue and lov_cancel need to handle inactive OSTs (403)
6176 - lfind did not preserve OST order in output (443)
6177 - symlinks cause hung clients, incorrect data (439)
6178 - stop dereferencing request after dropping refcount (457)
6179 - don't LASSERT(spin_is_locked) on non-SMP (455)
6180 - fixes for many rename() bugs
6181 - fstat didn't correctly synchronize attributes (399)
6182 - server must handle lock cancellation during blocking AST prep (487)
6183 - bulk descriptors were free()d too soon (511)
6184 - fix paths in lconf, which would load incorrect modules (451, 507)
6185 - fix confusing lconf 'host not found' error message (386)
6186 - fix lock order deadlock on OST (O/R i_sem before journal ops, 478)
6187 - fix race condition in mdc_blocking_ast() for inode access (526)
6188 - fix lov_unpackmd() unpacking wrong number of stripes (537)
6189 - fix lov_set_osc_active() marking wrong OSC inactive (440)
6190 - fix bad lstripe lov_unpackmd() assertion (fix layering too) (527)
6191 - fix multiple writes of stripe MD to MDS (358, maybe 519)
6192 - fix lstripe in several ways (kernel side) (527)
6193 - fix request leak in ldlm_cli_enqueue (262)
6194 - incorrect OSC was marked inactive after OST failure
6195 - call mds_fs_cleanup before unmounting filesystem (524)
6196 - fix races between taking ns_lock and ldlm_lock_change_resource
6197 - fix races updating LOV export open file list
6198 - fix lov_enqueue error path, avoid decref-ing bad lock handle
6199 - fix recovery NULL deref in ldlm_cli_cancel_unused
6200 - fix some DLM races by using new hash table for lock handles (419)
6201 - permit the client to specify desired inodes, at replay
6202 - duplicate requests when we queue them for replay reintegration
6203 - fix last_rcvd offset calculation
6204 - sync after each recovered transaction, so we always make progress
6205 - never, not always, ERESTART requests without transnos
6206 - store the lov_desc in the MDS, so we don't depend on getlovinfo to
6208 - skip replay if the MDS says that the client is already connected
6209 - don't check for a recovery-enabled export to match lctl's UUID
6210 - don't INC_USE_COUNT for phantom exports
6211 - don't crash when cleaning up phantom exports (567)
6212 - don't double-finish or set replay data for errored mdc_open requests
6213 - abort requests when they time out, so we don't get old replies
6214 - send/receive replies for AST messages again
6215 - if the client says that it doesn't have the lock, cancel it on the
6217 - if we timeout during I/O, don't try to cancel an in-use lock; instead
6218 mark it as destroyed, it will all work out when decref is called
6219 - fix module use counts (22, 581)
6221 - ASTs now expect a reply (server cancels lock on error reply)
6223 2002-12-02 Andreas Dilger <adilger@clusterfs.com>
6226 - fix many simultaneous client startup (392)
6227 - fix dentry->d_it clobbering
6228 - credentials weren't being shipped for readdir/getattr operations
6229 - remove invalid assertions triggered during some concurrent MD
6231 - proper Lustre versions added (336, 389)
6232 - fix memory leak for create error case (398)
6233 - fix LOV locking bug that would get cli/srv out of sync
6234 - fix echo client over LOV (409)
6235 - fix dbench 2, extN refcount problem (170, 258, 356, 418)
6236 - fix double-O_EXCL intent crash (424)
6237 - avoid sending multiple lock CANCELs (352)
6239 - MDS can do multi-client recovery (modulo bugs in new code)
6241 - many updates, edits, cleanups
6243 2002-11-18 Phil Schwan <phil@clusterfs.com>
6246 - fix null d_it dereference (346)
6247 - fix full OST/dbench hang (333)
6248 - fix permission problem with file removal (286)
6249 - fix removal of OSCs from LOV when they fail
6250 - fix NULL deref during bulk timeout (214)
6251 - fix problems related to multiple filesystems on one MDS (241)
6252 - fixed serious subtle metadata locking bugs
6253 - free locks on clients when inodes are removed due to memory
6255 - fix inode pointer in lock data (285)
6256 - partial support for multiple MDS on a single host (241)
6257 - data locks weren't cancelled at clear_inode time (290, 311)
6258 - intent locks could lead to unbounded lock growth (205)
6259 - added a maximum lock count, an LRU list, and a flusher
6260 - fix multiple rename (365)
6261 - properly abstracted the echo client
6262 - OSC locked 1 byte too many; fixed
6263 - rewrote brw callback code:
6264 - fixed recovery bugs related to LOVs (306)
6265 - fixed too-many-pages-in-one-write crash (191)
6266 - fixed (again) crash in sync_io_timeout (214)
6267 - probably fixed callback-related race (385)
6269 - Add capability to MDS protocol
6270 - LDLM cancellations and callbacks on different portals
6272 2002-10-28 Andreas Dilger <adilger@clusterfs.com>
6275 - limit client IOV size to PTL_MD_MAX_IOV (611336, 191)
6276 - defer open object destruction to close time (601981, 138)
6277 - open/close OST file handle in obdo (OBD_MD_FLHANDLE) (601981, 138)
6278 - move LDLM_ENQUEUE/CONVERT back to MDS portal (625069)
6279 - abstract ll_lookup2, fix ll_revalidate2 to use abstraction (256)
6280 - don't call obd_setattr in ll_file_release for destroyed objects
6281 * protocol change to lustre_msg: move |version| and add |flags|
6282 * protocol change to osc_punch: "start" in "o_size", "end" in "o_blocks"
6283 * lock replay: for LDLM_FL_REPLAY trust client to do right thing
6284 * added replay of create, unlink, link and rename operations during
6285 MDS failover; recovery should be much more robust now
6286 * remove failed OSCs from LOVs (only lov_create uses this so far)
6287 * the lustre-HOWTO was brought (more) up to date (582544)
6289 2002-10-23 Phil Schwan <phil@clusterfs.com>
6292 - in-use dentries weren't being reused properly (617851)
6293 - prevent multiple LDLM setup (599178)
6294 - fix LOV size calculations for truncate (617853)
6295 - fix client handling of MDS intent errors (POSIX)
6296 - fix permission bug in lovstripe.c test (624321)
6297 - fix MDS thread deadlock - move LDLM handler to DLM portal (625069)
6298 - truncate past end of file could corrupt data
6299 - proper cleanup after timeouts, crashes, etc (592524, 550815)
6300 - a race in recovery could return ETIMEDOUT to apps (623947)
6301 - building outside the source directory was fixed
6302 * the lustre-HOWTO was brought (more) up to date (582544)
6303 * major progress was made on recovery functionality
6305 2002-10-10 Phil Schwan <phil@clusterfs.com>
6308 - recovery deadlock fix
6309 - rm -rf causes LBUG fix (617817)
6310 - file open by multiple tasks fix (618962)
6311 - directory permissions bugs (602707 and 620007)
6312 - journal_stop fixed with locking (611313)
6313 - O_APPEND failures resolved (618273, perhaps 614459)
6314 - lconf PATH fix (619770)
6315 - IA64 build fix (621450)
6316 - RPC buffer sizes scale with amount of memory
6318 2002-10-01 Phil Schwan <phil@clusterfs.com>
6321 - locks would be cancelled without throwing away data pages,
6322 resulting in inconsistent data (605627)
6323 - inode attributes were not always being refreshed (605627, 612449)
6324 - lconf now continues to cleanup after lctl reports an error
6325 - MDS now enforces user permissions (602707)
6326 - lprocfs cleanup fixed, but not yet enabled (614157)
6327 - fixed infinite server hang, should a client not respond to an AST
6328 - avoid going into recovery if user calls readlink() with a buffer
6329 that's too small (613941)
6330 - AST RPCs no longer require replies (614867) -- this may be changed
6331 - don't crash server if client sends an IOV that's too big (611336)
6332 - fixed lock conversion deadlock (611892)
6333 - fixed the following of symlinks (614622)
6334 * recovery: the server can remove locks from a client that dies, other
6335 clients can make progress
6336 * more extN patch fixes
6337 * compile-time configurable ptlrpc buffer allocations
6339 - collaborative read cache document
6340 - Lustre Lite Performance CDR document-in-progress
6342 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
6345 - fix typo in patch-2.4.18
6347 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
6350 - clear ptlrpc request each time in handle_incoming_request()
6351 - unlink of files now destroys the object on the OST
6353 2002-09-19 Peter Braam <braam@clusterfs.com>
6355 * add hard link support
6356 * change obdfile creation method
6357 * kernel patch changed
6359 2002-09-19 Peter Braam <braam@clusterfs.com>
6362 - stack overflow bug in extN fixed
6364 2002-09-18 Andreas Dilger <adilger@clusterfs.com>
6366 * documentation updates
6367 - add man pages for config tools
6368 - update tests/README to describe testing with new config tools
6369 - finish metadata API descriptions
6370 * bug fixes and cleanups
6371 - statfs workaround for 16TB limit
6372 - LOV stripe allocation improved, can stripe on subset of OSTs
6373 - LOV file size/IO offset was wrong for files > 4GB in size
6374 - object EA data was being dropped, caused files to be unreadable
6375 - memory overflow with non-LOV OST caused memory corruption
6376 - fixed regression tests to work with new config tools, obdfilter
6377 - fixed bug when directory size became larger than 1 block
6378 - fixed bug (for single client case) when PWD was deleted
6379 - invalidate local directory pages when doing intent-based ops
6380 - avoid LDLM oops when lock callback contained bad data
6382 2002-09-09 Andreas Dilger <adilger@clusterfs.com>
6384 * documentation updates
6385 * bug fixes and cleanups
6386 - configuration tools
6389 - 64-bit compile warnings
6390 - 64-bit internal statfs data
6392 * test_brw on persistent OST devices
6394 * lprocfs (disabled)
6396 2002-09-04 Andreas Dilger <adilger@clusterfs.com>
6398 * documentation updates
6399 * bug fixes and cleanups
6400 * configuration tools
6402 2002-08-30 Peter J. Braam <braam@clusterfs.com>
6405 * many small fixes to 0_5_4
6406 * io/network handling
6407 * thinkos in MDS operations
6409 2002-08-24 Peter J. Braam <braam@clusterfs.com>
6412 * crucial basic fixes to 0.5.3
6413 * IOR, Iozone work over Elan
6416 2002-08-07 Phil Schwan <phil@clusterfs.com>
6417 * version 0_5_3, our first alpha
6418 * we use the new Portals iovs
6419 * documentation updates
6420 * bug fixes and cleanups
6421 * small changes in the DLM wire protocol
6423 2002-07-25 Peter J. Braam <braam@clusterfs.com>
6424 * version 0_5_1 with some initial stability,
6425 * locking on MD and file I/O.
6426 * documentation updates
6427 * several bug fixes since 0.5.0
6428 * small changes in wire protocol
6430 2002-07-18 Phil Schwan <phil@clusterfs.com>
6432 * delivered as Lustre Light Alpha
6433 * fixed a crash after handling invalid MDS requests
6434 * fixed directory pages for architectures with non-4k pages sizes
6436 2002-07-11 Andreas Dilger <adilger@clusterfs.com>
6437 * release version v0_4_4
6438 * Moves TCP acceptor to be on port 2432 (unused Coda port) instead
6440 * Fixes a number of interruption problems with OST operations.
6441 * Update documentation for portals header changes
6442 * Move all wire protocol structs/defines to lustre_idl.h
6443 * Fixes symlink length bug.
6444 * Add tcpdump to repository.
6446 2002-07-05 Andreas Dilger <adilger@clusterfs.com>
6447 * release version v0_4_3
6448 * Fixes statfs for inodes on extN.
6449 * Fixes bug in runtests which would delete /etc/hosts.
6450 * Use 64-bit object IDs wherever possible (not into VFS though)
6451 Remove ost_get_info, which is unused by lustre, and out of date.
6453 2002-07-03 Peter Braam <braam@clusterfs.com>
6454 * release version v0_4_2 Fixes a lookup error (type not passed)
6455 * move forward to head of Portals
6456 * move forward to latest Lustre kernel
6458 2002-06-25 Peter Braam <braam@clusterfs.com>
6459 * release version v0_4_1. Hopefully stable on single node use.