1 tbd Sun Microsystems, Inc.
4 2.6.9-67.0.4.EL (RHEL 4),
5 2.6.16.54-0.2.5 (SLES 10),
6 2.6.18-53.1.14.el5 (RHEL 5).
7 * Client support for unpatched kernels:
8 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
9 2.6.16 - 2.6.21 vanilla (kernel.org)
10 * Recommended e2fsprogs version: 1.40.7-sun1
11 * Note that reiserfs quotas are disabled on SLES 10 in this kernel.
12 * RHEL 4 and RHEL 5/SLES 10 clients behaves differently on 'cd' to a
13 removed cwd "./" (refer to Bugzilla 14399).
17 Frequency : rare, on recovery
18 Description: read procfs can produce deadlock in some situation
19 Details : Holding lprocfs lock which send rpc can produce block for destroy
20 obd objects and this also block reconnect with -EALREADY. This isn't
21 fix all lprocfs bugs - but make it rare.
23 Severity : enhancement
25 Description: Update kernel to RHEL5 2.6.18-53.1.14.el5.
28 Frequency : frequent on X2 node
30 Description: mdc_set_open_replay_data LBUG
31 Details : Set replay data for requests that are eligible for replay.
35 Description: lustre_mgs: operation 101 on unconnected MGS
36 Details : When MGC is disconnected from MGS long enough, MGS will evict the
37 MGC, and late on MGC cannot successfully connect to MGS and a lot
38 of the error messages complaining that MGS is not connected.
41 Frequency : on start mds
43 Description: Implement get_info(last_id) in obdfilter.
46 Frequency : occasional
48 Description: Correctly check stale fid, not start epoch if ost not support SOM
49 Details : open with flag O_CREATE need set old fid in op_fid3 because op_fid2
50 overwrited with new generated fid, but mds can anwer with one of these
51 two fids and both is not stale. setattr incorectly start epoch and
52 assume will be called done_writeting, but without SOM done_writing
56 Frequency : rare, depends on device drivers and load
58 Description: MDS or OSS nodes crash due to stack overflow
59 Details : Code changes in 1.8.0 increased the stack usage of some functions.
60 In some cases, in conjunction with device drivers that use a lot
61 of stack the MDS (or possibly OSS) service threads could overflow
62 the stack. One change which was identified to consume additional
63 stack has been reworked to avoid the extra stack usage.
66 Frequency : occasional
68 Description: Do not fail import if osc_interpret_create gets -EAGAIN
69 Details : If osc_interpret_create got -EAGAIN it immediately exits and
70 wakeup oscc_waitq. After wakeup oscc_wait_for_objects call
71 oscc_has_objects and see OSC has no objests and call
72 oscc_internal_create to resend create request.
74 Severity : enhancement
76 Description: Update to SLES10 SP1 latest kernel-2.6.16.54-0.2.5.
78 Severity : enhancement
80 Description: Update to RHEL5 latest kernel-2.6.18-53.1.13.el5.
85 Description: Fix replay if there is an un-replied request and open
86 Details : In some cases, older replay request will revert the
87 mcd->mcd_last_xid on MDS which is used to record the client's
90 Severity : enhancement
92 Description: Update to RHEL5 latest kernel-2.6.18-53.1.6.el5.
94 Severity : enhancement
96 Description: Add rhel5 support to HEAD.
98 Serverity : enhancement
100 Description: Update RHEL4 kernel to 2.6.9-67.0.4.
105 Description: Don't allow skipping OSTs if index has been specified.
106 Details : Don't allow skipping OSTs if index has been specified, make locking
107 in internal create lots better.
111 Description: LBUG in ptlrpc_check_set() bad phase ebc0de00
112 Details : access to bitfield in structure is always rounded to long
113 and this produce problem with not atomic change any bit.
117 Description: Lustre make rpms failed.
118 Details : Remove ldiskfs spec file to avoids rpmbuild be confused when
119 builds Lustre rpms from tarball.
122 Frequency : rare on shutdown ost
124 Description: If llog cancel was not send before clean_exports phase, this can
125 produce deadlock in llog code.
126 Details : If llog thread has last reference to obd and call class_import_put
127 this produce deadlock because llog_cleanup_commit_master wait when
128 last llog_commit_thread exited, but this never success because was
129 called from llog_commit_thread.
133 Description: allow userland application know is lost one of stripes.
134 Details : fill lvb_blocks with error code on ost and return it to
135 application if error flag found.
139 Description: NULL lov_tgts causing MDS oops
140 Details : more safe checks for NULL lov_tgts for avoid oops.
142 Severity : enhancement
144 Description: Update to RHEL4 latest kernel-2.6.9-67.0.1.EL.
148 Descriptoin: make lov_create() will not stuck in obd_statfs_rqset()
149 Details : If an OST is down the MDS will hang indefinitely in
150 obd_statfs_rqset() waiting for the statfs data. While for
151 MDS QOS usage of statfs, it should not stuck in waiting.
153 Severity : enhancement
155 Description: remote_acl support
156 Details : Support ACL-based permission check for remote user.
157 Support setfacl/getfacl for remote user with the utils
158 "lfs {l,r}{s,g}etfacl" which follow the same parameter format as
159 the system "{s,g}etfacl" utils.
161 Severity : enhancement
163 Description: Update to RHEL4 U6 kernel-2.6.9-67.EL.
165 Severity : enhancement
167 Description: Update to RHEL5 latest kernel-2.6.18-53.1.4.el5.
171 Description: make mgs_setparam() handle fsname containing dash
172 Details : fsname containing a dash does not work with lctl conf_param
174 Severity : enhancement
176 Description: Update to SLES10 SP1 latest kernel-2.6.16.54-0.2.3.
178 Severity : enhancement
180 Description: Update to RHEL5 Update-1 kernel 2.6.18-53.el5.
184 Frequency : rare, at shutdown
185 Description: access already free / zero obd_namespace.
186 Details : if client_disconnect_export was called without force flag set,
187 and exist connect request in flight, this can produce access to
188 NULL pointer (or already free pointer) when connect_interpret
189 store ocd flags in obd_namespace.
193 Frequency : only at startup
194 Description: not alloc memory with spinlock held.
195 Details : allocation memory with GFP_KERNEL can produce sleep deadlock,
196 if any spinlock held.
198 Severity : enhancement
200 Description: make lustre randomly fail allocating memory
201 Details : Make lustre randomly failed allocating memory for testing purpose.
203 Severity : enhancement
205 Description: lost problems with lov objid file
206 Details : Fixes some scability and access to not inited memory problems
207 in work with lov objdid file.
212 Description: lfs find does not continue on file error
213 Details : Continue other files processing when a file/dir is absent.
217 Description: Inconsistent usage of lustre_pack_reply()
218 Details : Standardize the usage of lustre_pack_reply() such that it
219 always generate a CERROR on failure.
222 Frequency : occasional
224 Description: MDS hang or stay in waiting lock
225 Details : If client receive lock with CBPENDING flag ldlm need send lock
226 cancel as separate rpc, to avoid situation when cancel request
227 can't processed due all i/o threads stay in wait lock.
231 Description: Update to RHEL5 kernel 2.6.18-8.1.15.el5.
235 Description: Update to SLES10 SP1 kernel 2.6.16.53-0.16
239 Description: Update to SLES9 kernel-2.6.5-7.287.3.
243 Description: Update to RHEL4 latest kernel.
245 Severity : enhancement
247 Description: Build SLES10 patchless client fails
248 Details : The configure was broken by run ./configure with
249 --with-linux-obj=.... argument for patchless client. When the
250 configure use --with-linux-obj, the LINUXINCLUDE= -Iinclude
251 can't search header adequately. Use absolute path such as
252 -I($LINUX)/include instead.
256 Description: interrupt oig_wait produce painc on resend.
257 Details : brw_redo_request can be used for resend requests from ptlrpcd and
258 private set, and this produce situation when rq_ptlrpcd_data not
259 copyed to new allocated request and triggered LBUG on assert
260 req->rq_ptlrpcd_data != NULL. But this member used only for wakeup
261 ptlrpcd set if request is changed and can be safety changed to use
266 Description: LASSERT_{REQ,REP}SWAB macros are buggy
267 Details : If SWAB_PARANOIA is disabled, the LASSERT_REQSWAB and
268 LASSERT_REPSWAB macros become no-ops, which is incorrect. Drop
269 these macros and replace them with their difinitions instead.
273 Description: Update kernel patches for SLES10 2.6.16.53-0.8.
274 Details : Update which_patch & target file for SLES10 latest kernel.
277 Description: Remove client patches from SLES 10 kernel.
278 Details : This causes SLES 10 clients to behave as patchless clients
279 even on a Lustre-patched (server) kernel.
281 Severity : enhancement
283 Description: self-adjustable client's lru lists
284 Details : use adaptive algorithm for managing client cached locks lru
285 lists according to current server load, other client's work
286 pattern, memory activities, etc. Both, server and client
287 side namespaces provide number of proc tunables for controlling
290 Severity : enhancement
292 Description: light-weight GSS support
293 Details : Support krb5n and krb5a mode, which keep Kerberos 5 authentication
294 and reduce performance overhead.
296 Severity : enhancement
298 Description: Linux keyring support
299 Details : Support using service of Linux keyring for Lustre GSS internal
300 context refresh/cache mechanism.
304 Description: Fix errors in lfs documentation
305 Details : Fixes man pages
309 Description: don't use GFP_* in generic Lustre code.
310 Details : Use cfs_alloc_* functions and CFS_* flags for code portability.
314 Description: obdclass is limited by single OBD_ALLOC(idarray)
315 Details : replace OBD_ALLOC/OBD_FREE with OBD_VMALLOC/OBD_VFREE
319 Description: warnings with build patchless client with vanila 2.6.19 and up
320 Details : change old ctl_table style and replace ctl_table/ctl_table_header
321 with cfs_sysctl_table_t/cfs_sysctl_table_header_t
325 Frequency : Only for SLES
326 Description: sanity_quota fail test_1
327 Details : There are multiple occurences of $TSTUSR in SLES's /etc/group
328 file, which makes TSTID[2] inunique.
332 Frequency : Only for SLES9
333 Description: Kernel patches for SLES9 2.6.5-7.286 kernel
334 Details : Update target/ChangeLog/which_patch .
338 Frequency : Only for test_10 in sanity-quota.sh
339 Description: a bug in quota test code
340 Details : checking a test flag in a wrong place
344 Frequency : Only for test_13 in sanity-quota.sh
345 Description: a bug in quota test script
346 Details : shouldn't check overfull allocated quota
350 Description: jbd statistics
351 Details : Port older jbd statistics patch for sles10
353 Severity : enhancement
355 Description: Add printing inode info into message about error in writepage.
360 Description: reply_lock_interpret crash due to race with it and lock cancel.
361 Details : Do not replay locks that are being cancelled. Do not reference
362 locks by their address during replay, just by their handle.
366 Frequency : When flocks are used.
367 Description: assertion failure in ldlm_cli_enquque_fini for non NULL lock.
368 Details : Flock locks might destroy just granted lock if it could be merged
369 with another existing flock, this is done in completion handler,
370 so teach ldlm_cli_enquque_fini that this is a valid case for
376 Description: Oops in read and write path when failing to allocate lock.
377 Details : Check if lock allocation failed and return error back.
381 Description: lstripe command fails for valid OST index
382 Details : The stripe offset is compared to lov->desc.ld_tgt_count
383 instead of lov->desc.ld_active_tgt_count.
387 Description: sanity.sh failed test 103
388 Details : RHEL mis-interpret setfacl "-X" param, so we won't test setfacl
393 Description: df doesn't work properly if diskfs blocksize != 4K
394 Details : Choose biggest blocksize of OST's as the LOV's blocksize.
398 Description: lfs find on -1 stripe looping in lsm_lmm_verify_common()
399 Details : Avoid lov_verify_lmm_common() on directory with -1 stripe count.
403 Description: obd_health_check_timeout too short
404 Details : set obd_health_check_timeout as 1.5x of obd_timeout
408 Description: llapi_file_create() does not allow some changes
409 Details : add llapi_file_open() that allows specifying the mode and
410 open flags, and also returns an open file handle.
414 Description: merge and cleanup kernel patches.
415 Details : Remove mnt_lustre_list in vfs_intent-2.6-rhel4.patch.
419 Description: Add journal checksum support.(Kernel part)
420 Details : The journal checksum feature adds two new flags i.e
421 JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT and
422 JBD2_FEATURE_COMPAT_CHECKSUM. JBD2_FEATURE_CHECKSUM flag
423 indicates that the commit block contains the checksum for
424 the blocks described by the descriptor blocks. Now commit
425 record can be sent to disk without waiting for descriptor
426 blocks to be written to disk. This behavior is controlled
427 using JBD2_FEATURE_ASYNC_COMMIT flag.
431 Description: OSS needs mutliple precreate threads
432 Details : Add ability to start more than one create thread per OSS.
436 Description: Sanity.sh test_65k failed: setstripe should have succeeded
437 Details : execute lfs setstripe on client
441 Description: mds_obd_create error creating tmp object
442 Details : When the user sets quota on root, llog will be affected and can't
443 create files and write files.
446 Frequency : Always on ia64 patchless client, and possibly others.
448 Description: Add EXPORT_SYMBOL check for node_to_cpumask symbol.
449 Details : This allows the patchless client to be loaded on architectures
454 Description: RedHat Update kernel for RHEL5
455 Details : Add the kernel config file for RHEL5.
459 Description: RedHat Update kernel for RHEL5
460 Details : Modify the kernel config file more closer RHEL5.
464 Description: Build failure against Centos5 (RHEL5)
465 Details : Define PAGE_SIZE when it isn't present.
469 Description: test_7 on sanity-quota.sh can't be used on separate machines
470 Details : add facet to handle it
474 Description: "ll_intent_file_open()) lock enqueue: err: -13" with nfs
475 Details : with NFS, the anon dentry's parent was set to itself in
476 d_alloc_anon(), so in MDS, we use rec->ur_fid1 to find the
477 corresponding dentry other than use rec->ur_name.
479 Severity : enhancement
481 Description: lfs setstripe enhancement
482 Details : Make lfs setstripe understand 'k', 'm' and 'g' for stripe size.
486 Description: enable data checksumming by default
487 Details : enable checksum by default, allow --disable-checksum
488 configure option and "-o nochecksum" mount option. Checksums
489 can also be disabled at runtime via $LPROC/osc/*/checksum_pages.
493 Description: lustre support for RHEL5
494 Details : Add support for RHEL5.
498 Description: Client eviction due to failover config
499 Details : after a connection loss, the lustre client should attempt to
500 reconnect to the last active server first before trying the
501 other potential connections.
505 Description: when mds and osts use different quota unit(32bit and 64bit),
506 quota will be released repeatly.
507 Details : void sending multiple quota reqs to mds, which will keep the
508 status between the reqs.
512 Description: osts not allocated evenly to files
513 Details : change the condition to increase offset_idx
516 Frequency : Always for filesystems larger than 2TB on 32-bit systems.
518 Description: Data corruption for OSTs that are formatted larger than 2TB
520 Details : When generating the bio request for lustre file writes the
521 sector number would overflow a temporary variable before being
522 used for the IO. The data reads correctly from Lustre (which
523 will overflow in a similar manner) but other file data or
524 filesystem metadata may be corrupted in some cases.
528 Description: Tune the kernel for good SCSI performance.
529 Details : Set the value of /sys/block/{dev}/queue/max_sectors_kb
530 to the value of /sys/block/{dev}/queue/max_hw_sectors_kb
535 Description: rewrite ext2-derived code in obdclass/uuid.c
536 Details : rewrite inherited code (uuid parsing code from ext2 utils)
537 from scratch preserving functionality.
541 Description: "lfs find -obd UUID" prints directories
542 Details : "lfs find -obd UUID" will return all directory names instead
543 of just file names. It is incorrect because the directories
544 do not reside on the OSTs.
548 Description: use i_size_read and i_size_write in 2.6 port
549 Details : replace inode->i_size access with i_size_read/write()
552 Frequency : when removing large files
554 Description: scheduling issue during removal of large Lustre files
555 Details : Don't take the BKL in fsfilt_ext3_setattr() for 2.6 kernels.
556 It causes scheduling issues when removing large files (17TB in the
560 Frequency : only with liblustre clients on XT3
562 Description: evictions taking too long
563 Details : allow llrd to evict clients directly on OSTs
566 Frequency : only on ppc
568 Description: /proc/fs/lustre/devices broken on ppc
569 Details : The patch as applied to 1.6.2 doesn't look correct for all arches.
570 We should make sure the type of 'index' is loff_t and then cast
571 explicitly as needed below. Do not assign an explicitly cast
575 Frequency : only for rhel5
577 Description: Kernel patches update for RHEL5 2.6.18-8.1.10.el5.
578 Details : Modify the target file & which_kernel.
580 Severity : enhancement
582 Description: omit set fsid for export NFS
583 Details : fix set/restore device id for avoid EMFILE error and mark lustre fs
584 as FS_REQUIRES_DEV for avoid problems with generate fsid.
588 Frequency : Always, for kernels after 2.6.16
589 Description: Fix warning idr_remove called for id=.. which is not allocated.
590 Details : Last kernels save old s_dev before kill super and not allow
591 to restore from callback - restore it before call kill_anon_super.
595 Description: buffer overruns could theoretically occur
596 Details : llapi_semantic_traverse() modifies the "path" argument by
597 appending values to the end of the origin string, and a
598 overrun may occur. Adding buffer overrun check in liblustreapi.
602 Description: Fix error on 'ls .' at the top of the Lustre mount.
603 Details : Don't revalidate dentry if it is a root dentry.
607 Description: Kernel patches update for RHEL4 2.6.9-55.0.6.
608 Details : Modify vm-tunables-rhel4.patch.
612 Description: Kernel config for 2.6.18-vanilla.
613 Details : Modify targets/2.6-vanilla.target.in.
614 Add config file kernel-2.6.18-2.6-vanilla-i686.config.
615 Add config file kernel-2.6.18-2.6-vanilla-i686-smp.config.
616 Add config file kernel-2.6.18-2.6-vanilla-x86_64.config.
617 Add config file kernel-2.6.18-2.6-vanilla-x86_64-smp.config.
621 Description: improve handling recoverable errors
622 Details : if request processig with error which can be recoverable on server
623 request should be resend, otherwise page released from cache and
628 Description: Kernel patches update for RHEL5 2.6.18-8.1.14.el5.
629 Details : Modify target file & which_patch.
630 A flaw was found in the IA32 system call emulation provided
631 on AMD64 and Intel 64 platforms. An improperly validated 64-bit
632 value could be stored in the %RAX register, which could trigger an
633 out-of-bounds system call table access. An untrusted local user
634 could exploit this flaw to run code in the kernel
635 (ie a root privilege escalation). (CVE-2007-4573).
639 Description: O_DIRECT bypasses client statistics.
640 Details : When running with O_DIRECT I/O, neither the client rpc_stats nor
641 read_ahead_stats were updated. copied the stats section from
642 osc_send_oap_rpc() into async_internal().
646 Description: Add jbd statistics patch for RHEL5 and 2.6.18-vanilla
650 Description: change order of libsysio includes
651 Details : '#include sysio.h' should always come before '#include xtio.h'
655 Description: handle "serious error: objid * already exists" more gracefully
656 Details : If LAST_ID value on disk is smaller than the objects existing in
657 the O/0/d* directories, it indicates disk corruption and causes an
658 LBUG(). If the object is 0-length, then we should use the existing
659 object. This will help to avoid a full fsck in most cases.
661 Severity : enhancement
663 Description: adapt the lustre_config script to support the upgrade case
664 Details : Add "-u" option for lustre_config script to support upgrading 1.4
665 server targets to 1.6 in parallel.
669 Description: To avoid grant space > avaible space when the disk is almost
670 full. Without this patch you might see the error "grant XXXX >
671 available" or some LBUG about grant, when the disk is almost
673 Details : In filter_check_grant, for non_grant cache write, we should
674 check the left space by if (*left > ungranted + bytes), instead
675 of (*left > ungranted), because only we are sure the left space
676 is enough for another "bytes", then the ungrant space should be
677 increase. In client, we should update cl_avail_grant only there
678 is OBD_MD_FLGRANT in the reply.
682 Description: Update RHEL 4 kernel to fix local root privilege escalation.
683 Details : Update to the latest RHEL 4 kernel to fix the vulnerability
684 described in CVE-2007-4573. This problem could allow untrusted
685 local users to gain root access.
688 Frequency : when using O_DIRECT and quotas
690 Description: Incorrect file ownership on O_DIRECT output files
691 Details : block usage reported by 'lfs quota' does not take into account
692 files that have been written with O_DIRECT.
697 Description: touch file failed when fs is not full
698 Details : OST in recovery should not be discarded by MDS in alloc_qos(),
699 otherwise we can get ENOSP while fs is not full.
703 Description: parallel lock callbacks
704 Details : Instead of sending blocking and completion callbacks as separated
705 requests, adding them to a set and sending in parallel.
708 Frequency : only for Cray XT3
709 Bugzilla : 12829/13455
710 Description: Changing primary group doesn't change the group lustre assigns to
712 Details : When CRAY_XT3 is defined, the fsgid supplied by the client is
713 overridden with the primary group provided by the group upcall,
714 whereas the supplied fsgid can be trusted if it is in the list of
715 supplementary groups returned by the group upcall.
717 Severity : enhancement
719 Description: Allow masking D_WARNING, D_ERROR messages from console
720 Details : Console messages can now be disabled via lnet.printk.
724 Description: User code with malformed file open parameter crashes client node
725 Details : Before packing join_file req, all the related reference should be
726 checked carefully in case some malformed flags cause fake
727 join_file req on client.
731 Description: LDLM_ENQUEUE races with LDLM_CP_CALLBACK
732 Details : ldlm_completion_ast() assumes that a lock is granted when the req
733 mode is equal to the granted mode. However, it should also check
734 that LDLM_FL_CP_REQD is not set.
738 Description: Heavy nfs access might result in deadlocks
739 Details : After ELC code landed, it is now improper to enqueue any mds
740 locks under och_sem, because enqueue might want to decide to
741 cancel open locks for same inode we are holding och_sem for.
745 Description: Client eviction while running blogbench
746 Details : A lot of unlink operations with concurrent I/O can lead to a
747 deadlock causing evictions. To address the problem, the number of
748 oustanding OST_DESTROY requests is now throttled to
749 max_rpcs_in_flight per OSC and LDLM_FL_DISCARD_DATA blocking
750 callbacks are processed in priority.
754 Description: enable ACLs on MDS by default
755 Details : ACLs must be enabled on MDS by default.
758 Frequency : PPC/PPC64 only
760 Description: conflicts between asm-ppc64/types.h and lustre_types.h
761 Details : fix duplicated definitions between asm-ppc64/types.h and
762 lustre_types.h on PPC.
765 Frequency : PPC/PPC64 only
767 Description: asm-ppc/segment.h does not exist
768 Details : fix compile issue on PPC.
772 Description: better handle error messages in extents code
775 Frequency : RHEL4 only
777 Description: mkfs is very slow on IA64/RHEL4
778 Details : A performance regression has been discovered in the MPT Fusion
779 driver between versions 3.02.73rh and 3.02.99.00rh. As a
780 consequence, we have downgraded the MPT Fusion driver in the RHEL4
781 kernel from 3.02.99.00 to 3.02.73 until this problem is fixed.
783 Severity : enhancement
785 Description: SNMP support enhancement
786 Details : Adding total number of sampled request for an MDS node in snmp
789 Severity : enhancement
791 Description: Optimize ldlm waiting list processing for PR extent locks
792 Details : When processing waiting list for read extent lock and meeting read
793 lock that is same or wider to it that is not contended, skip
794 processing rest of the list and immediatelly return current
795 status of conflictness, since we are guaranteed there are no
796 conflicting locks in the rest of the list.
800 Description: Time out and refuse to reconnect
801 Details : When the failover node is the primary node, it is possible
802 to have two identical connections in imp_conn_list. We must
803 compare not conn's pointers but NIDs, otherwise we can defeat
804 connection throttling.
808 Description: port llog fixes from b1_6 into HEAD
809 Details : Port llog reference couting and some llog cleanups from b1_6
810 (bug 10800) into HEAD, for protect from panic and access to already
811 free llog structures.
815 Description: Detect stride IO mode in read-ahead
816 Details : When a client does stride read, read-ahead should detect that and
817 read-ahead pages according to the detected stride pattern.
821 Description: data checksumming impacts single node performance
822 Details : add support for several checksum algorithm. Currently, only CRC32
823 and Adler-32 are supported. The checksum type can be changed on
824 the fly via /proc/fs/lustre/osc/*/checksum_type.
828 Description: use adler32 for page checksums
829 Details : when available, use the Adler-32 algorithm instead of CRC32 for
834 Description: build for x2 fails
835 Details : fix compile issue on Cray systems.
839 Description: Properly match for duplicate locks
840 Details : Due to different lock order from skiplists code, we need to
841 traverse entire list for now
844 Frequency : only on PPC/SLES10
846 Description: "BITS_PER_LONG is not 32 or 64" in linux/idr.h
847 Details : On SLES10/PPC, fs.h includes idr.h which requires BITS_PER_LONG to
848 be defined. Add a hack in mkfs_lustre.c to work around this compile
853 Description: LASSERT on MDS when client holding flock lock dies
854 Details : ldlm pool logic depends on number of granted locks equal to
855 number of released locks which is not true for flock locks, so
856 just exclude such locks from consideration.
858 --------------------------------------------------------------------------------
860 2007-08-10 Cluster File Systems, Inc. <info@clusterfs.com>
862 * Support for kernels:
863 2.6.5-7.283 (SLES 9),
864 2.6.9-55.EL (RHEL 4),
865 2.6.16.46-0.14 (SLES 10),
866 2.6.18.8 vanilla (kernel.org)
867 * Client support for unpatched kernels:
868 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
869 2.6.16 - 2.6.22 vanilla (kernel.org)
870 * Due to recently discovered recovery problems, we do not recommend
871 using patchless RHEL 4 clients with this or any earlier release.
872 * Recommended e2fsprogs version: 1.39.cfs8
873 * Note that reiserfs quotas are disabled on SLES 10 in this kernel.
874 * Starting with this release, the ldiskfs backing filesystem required
875 by Lustre is now in its own package, lustre-ldiskfs. This package
876 should be installed. It is versioned separately from Lustre and
877 may be released separately in future.
881 Description: block reactivating mgc import until all deactivates complete
882 Details : Fix race when failing back MDT/MGS to itself (testing)
885 Frequency : at statup only
887 Description: mds_lov_synchronize race leads to various problems
888 Details : simultaneous MDT->OST connections at startup can cause the
889 sync to abort, leaving the OSC in a bad state.
891 Severity : enhancement
893 Description: add optional extra BUILD_VERSION info
894 Details : add a new environment variable (namely LUSTRE_VERS) which allows
895 to override the lustre version.
898 Frequency : 2.6.18 servers only
900 Description: ll_kern_mount() doesn't release the module reference
901 Details : The ldiskfs module reference count never drops down to 0
902 because ll_kern_mount() doesn't release the module reference.
907 Description: server LBUG when using old ost_num_threads parameter
908 Details : Accept the old ost_num_threads parameter but warn that it
909 is deprecated, and fix an off-by-one error that caused an LBUG.
914 Description: Transient SCSI error results in persistent IO issue
915 Details : iobuf->dr_error is not reinitialized to 0 between two
919 Frequency : sometimes when underlying device returns I/O errors
921 Description: OSTs not going read-only during write failures
922 Details : OSTs are not remounted read-only when the journal commit threads
923 get I/O errors because fsfilt_ext3 calls journal_start/stop()
924 instead of the ext3 wrappers.
928 Description: poor connect scaling with increasing client count
929 Details : Don't run filter_grant_sanity_check for more than 100 exports
930 to improve scaling for large numbers of clients.
933 Frequency : SLES10 only
935 Description: sanity-quota.sh quotacheck failed: rc = -22
936 Details : Quotas cannot be enabled on SLES10.
939 Frequency : liblustre clients only
941 Description: getdirentries does not give error when run on compute nodes
942 Details : getdirentries does not fail when the size specified as an argument
943 is too small to contain at least one entry
945 Severity : enhancement
947 Description: Add LNET router traceability for debug purposes
948 Details : If a checksum failure occurs with a router as part of the
949 IO path, the NID of the last router that forwarded the bulk data
950 is printed so it can be identified.
955 Description: OST "spontaneously" evicts client; client has imp_pingable == 0
956 Details : Due to a race condition, liblustre clients were occasionally
959 Severity : enhancement
961 Description: lfs setstripe use optional parameters instead of postional
964 Severity : enhancement
966 Description: Nanosecond timestamp support for ldiskfs
967 Details : The on-disk ldiskfs filesystem has added support for nanosecond
968 resolution timestamps. There is not yet support for this at
969 the Lustre filesystem level.
972 Frequency : during server recovery
974 Description: MDS failing to send precreate requests due to OSCC_FLAG_RECOVERING
975 Details : request with rq_no_resend flag not awake l_wait_event if they get a
979 Frequency : nfs export on patchless client
981 Description: connectathon hang when test nfs export over patchless client
982 Details : Disconnected dentry cannot be found with lookup, so we do not need
983 to unhash it or make it invalid
986 Description: fix llapi_lov_get_uuids() to allow many OSTs to be returned
987 Details: : Change llapi_lov_get_uuids() to read the UUIDs from /proc instead
988 of using an ioctl. This allows lfsck for > 160 OSTs to succeed.
993 Description: open req refcounting wrong on reconnect
994 Details : If reconnect happened between getting open reply from server and
995 call to mdc_set_replay_data in ll_file_open, we will schedule
996 replay for unreferenced request that we are about to free.
997 Subsequent close will crash in variety of ways.
998 Check that request is still eligible for replay in
999 mdc_set_replay_data().
1004 Description: disable writes to filesystem when reading health_check file
1005 Details : the default for reading the health_check proc file has changed
1006 to NOT do a journal transaction and write to disk, because this
1007 can cause reads of the /proc file to hang and block HA state
1008 checking on a healthy but otherwise heavily loaded system. It
1009 is possible to return to the previous behaviour during configure
1010 with --enable-health-write.
1012 Severity : enhancement
1014 Description: 64-bit inode version
1015 Details: : Add a on-disk 64-bit inode version for ext3 to track changes made
1016 to the inode. This will be required for version-based recovery.
1021 Description: MDS fails to start if a duplicate client export is detected
1022 Details : in some rare cases it was possible for a client to connect to
1023 an MDS multiple times. Upon recovery the MDS would detect this
1024 and fail during startup. Handle this more gracefully.
1026 Severity : enhancement
1028 Description: Add -o localflock option to simulate old noflock
1030 Details : This will achieve local-only flock/fcntl locks
1036 Description: log_commit_thread vs filter_destroy race leads to crash
1037 Details : Take import reference before releasing llog record semaphore
1042 Description: Wrong request locking in request set processing
1043 Details : ptlrpc_check_set wrongly uses req->rq_lock for proctect add to
1044 imp_delayed_list, in this place should be used imp_lock.
1047 Frequency : when reconnection
1049 Description: Grant Leak when osc reconnect to OST
1050 Details : When osc reconnect ost, OST(filter) should check whether it
1051 should grant more space to client by comparing fed_grant and
1052 cl_avail_grant, and return the granted space to client instead
1053 of "new granted" space, because client will call osc_init_grant
1054 to update the client grant space info.
1057 Frequency : when client reconnect to OST
1059 Description: Grant Leak when osc do resend and replay bulk write
1060 Details : When osc reconnect to OST, OST(filter)should clear grant info of
1061 bulk write request, because the grant info while be sync between
1062 OSC and OST when reconnect, and we should ignore the grant info
1063 these of resend/replay write req.
1068 Description: Grant space more than avaiable left space sometimes.
1069 Details : When then OST is about to be full, if two bulk writing from
1070 different clients came to OST. Accord the avaliable space of the
1071 OST, the first req should be permitted, and the second one
1072 should be denied by ENOSPC. But if the seconde arrived before
1073 the first one is commited. The OST might wrongly permit second
1074 writing, which will cause grant space > avaiable space.
1077 Frequency : when client is evicted
1079 Description: Grant might be wrongly erased when osc is evicted by OST
1080 Details : when the import is evicted by server, it will fork another
1081 thread ptlrpc_invalidate_import_thread to invalidate the
1082 import, where the grant will be set to 0. While the original
1083 thread will update the grant it got when connecting. So if
1084 the former happened latter, the grant will be wrongly errased
1085 because of this race.
1090 Description: Checking Stale with correct fid
1091 Details : ll_revalidate_it should uses de_inode instead of op_data.fid2
1092 to check whether it is stale, because sometimes, we want the
1093 enqueue happened anyway, and op_data.fid2 will not be initialized.
1095 Severity : enhancement
1097 Description: update patchless client
1098 Details : Add support for patchless client with 2.6.20, 2.6.21 and RHEL 5
1101 Frequency : only with 2.4 kernel
1103 Description: random memory corruption
1104 Details : size of struct ll_inode_info is to big for union inode.u and this
1105 can be cause of random memory corruption.
1110 Description: Memory leak in recovery
1111 Details : Lov_mds_md was not free in an error handler in mds_create_object.
1112 It should also check obd_fail before fsfilt_start, otherwise if
1113 fsfilt_start return -EROFS,(failover mds during mds recovery).
1114 then the req will return with repmsg->transno = 0 and rc = EROFS.
1115 and we met hit the assert LASSERT(req->rq_reqmsg->transno ==
1116 req->rq_repmsg->transno) in ptlrpc_replay_interpret. Fcc should
1117 be freed no matter whether fsfilt_commit success or not.
1120 Frequency : only with huge count clients
1122 Description: Prevents from taking the superblock lock in llap_from_page for
1124 Details : using LL_ORIGIN_REMOVEPAGE origin flag instead of LL_ORIGIN_UNKNOW
1125 for llap_from_page call in ll_removepage prevents from taking the
1126 superblock lock for a soon died page.
1131 Description: Not check open intent error before release open handle
1132 Details : in some rare cases, the open intent error is not checked before
1133 release open handle, which may cause
1134 ASSERTION(open_req->rq_transno != 0), because it tries to release
1135 the failed open handle.
1140 Description: Set cat log bitmap only after create log success.
1141 Details : in some rare cases, the cat log bitmap is set too early. and it
1142 should be set only after create log success.
1146 Description: Accessing a block bevice can re-enable I/O when Lustre is
1147 tearing down a device.
1148 Details : dev_clear_rdonly(bdev) must be called in kill_bdev() instead of
1153 Description: service threads may hog cpus when there are a lot of requests
1154 Details : Insert cond_resched to give other threads a chance to use some CPU
1159 Description: the cat log was not initialized in recovery
1160 Details : When mds(mgs) do recovery, the tgt_count might be zero, so the
1161 unlink log on mds will not be initialized until mds post
1162 recovery. And also in mds post recovery, the unlink log will
1163 initialization will be done asynchronausly, so there will be race
1164 between add unlink log and unlink log initialization.
1168 Description: brw_stats were being printed incorrectly
1169 Details : brw_stats were being printed as log2 but all of them were not
1170 recorded as log2. Also remove some code duplication arising from
1171 filter_tally_{read,write}.
1175 Frequency : rare, only in recovery.
1176 Description: ASSERTION(req->rq_type != LI_POISON) failed
1177 Details : imp_lock should be held while iterating over imp_sending_list for
1178 prevent destroy request after get timeout in ptlrpc_queue_wait.
1182 Description: replay-single.sh test 52 fails
1183 Details : A lock's skiplist need to be cleanup when it being unlinked
1184 from its resource list.
1188 Description: Short directio read returns full requested size rather than
1190 Details : Direct I/O operations should return actual amount of bytes
1191 transferred rather than requested size.
1193 Severity : enhancement
1195 Description: metadata RPC reduction (e.g. for rm performance)
1196 Details : decrease the amount of synchronous RPC between clients and servers
1197 by canceling conflicing lock before the operation on the client
1198 and packing thier handles into the main operation RPC to server.
1200 Severity : enhancement
1202 Description: Async OSC create to avoid the blocking unnecessarily.
1203 Details : If a OST has no remain object, system will block on the creating
1204 when need to create a new object on this OST. Now, ways use
1205 pre-created objects when available, instead of blocking on an
1206 empty osc while others are not empty. If we must block, we block
1207 for the shortest possible period of time.
1209 Severity : enhancement
1211 Description: refine locking for avoid write wrong info into lov_objid file
1212 Details : fix possible races with add new target and write/update data in
1215 --------------------------------------------------------------------------------
1217 2007-05-03 Cluster File Systems, Inc. <info@clusterfs.com>
1222 Frequency : on some architectures
1224 Description: 1.6 client sometimes fails to mount from a 1.4 MDT
1225 Details : Uninitialized flags sometimes cause configuration commands to
1229 Frequency : patchless clients only
1231 Description: missing __iget() symbol export
1232 Details : The __iget() symbol export is missing. To avoid the need for
1233 this on patchless clients the deathrow inode reaper is turned
1234 off, and we depend on the VM to clean up old inodes. This
1235 dependency was during via the fix for bug 12181.
1237 --------------------------------------------------------------------------------
1239 2007-04-19 Cluster File Systems, Inc. <info@clusterfs.com>
1241 * CONFIGURATION CHANGE. This version of Lustre WILL NOT
1242 INTEROPERATE with older versions automatically. In many cases a
1243 special upgrade step is needed. Please read the
1244 user documentation before upgrading any part of a live system.
1245 * WIRE PROTOCOL CHANGE from previous 1.6 beta versions. This
1246 version will not interoperate with 1.6 betas before beta5 (1.5.95).
1247 * WARNING: Lustre configuration and startup changes are required with
1248 this release. See https://mail.clusterfs.com/wikis/lustre/MountConf
1254 Severity : enhancement
1256 Description: MountConf
1257 Details : Lustre configuration is now managed via mkfs and mount
1258 commands instead of lmc and lconf. New obd types (MGS, MGC)
1259 are added for dynamic configuration management. See
1260 https://mail.clusterfs.com/wikis/lustre/MountConf for
1263 Severity : enhancement
1265 Description: dynamic OST addition
1266 Details : OSTs can now be added to a live filesystem
1268 Severity : enhancement
1270 Description: startup order invariance
1271 Details : MDTs and OSTs can be started in any order. Clients only
1272 require the MDT to complete startup.
1274 Severity : enhancement
1276 Description: parallel, asynchronous orphan cleanup
1277 Details : orphan cleanup is now performed in separate threads for each
1278 OST, allowing parallel non-blocking operation.
1280 Severity : enhancement
1282 Description: optimized stripe assignment
1283 Details : stripe assignments are now made based on ost space available,
1284 ost previous usage, and OSS previous usage, in order to try
1285 to optimize storage space and networking resources.
1287 Severity : enhancement
1289 Description: Permanently set tunables
1290 Details : All writable /proc/fs/lustre tunables can now be permanently
1291 set on a per-server basis, at mkfs time or on a live system.
1293 Severity : enhancement
1295 Description: Lustre message v2
1296 Details : Add lustre message format v2.
1298 Severity : enhancement
1300 Description: client OST exclusion list
1301 Details : Clients can be started with a list of OSTs that should be
1302 declared "inactive" for known non-responsive OSTs.
1306 Description: SPEC SFS validation failure on NFS v2 over lustre.
1307 Details : Changes the blocksize for regular files to be 2x RPC size,
1308 and not depend on stripe size.
1310 Severity : enhancement
1312 Description: Multiple MD RPCs in flight.
1313 Details : Further unserialise some read-only MDS RPCs - learn about intents.
1314 To avoid overly-overloading MDS, introduce a limit on number of
1315 MDS RPCs in flight for a single client and add /proc controls
1316 to adjust this limit.
1318 Severity : enhancement
1320 Description: client read/write statistics
1321 Details : Add client read/write call usage stats for performance
1322 analysis of user processes.
1323 /proc/fs/lustre/llite/*/offset_stats shows non-sequential
1324 file access. extents_stats shows chunk size distribution.
1325 extents_stats_per_process show chunk size distribution per
1328 Severity : enhancement
1330 Description: mds statistics
1331 Details : Add detailed mds operations statistics in
1332 /proc/fs/lustre/mds/*/stats.
1336 Description: Failure of copying files with lustre special EAs.
1337 Details : Client side always return success for setxattr call for lustre
1338 special xattr (currently only "trusted.lov").
1343 Description: Refcount LNET uuids
1344 Details : The global LNET uuid list grew linearly with every startup;
1345 refcount repeated list entries instead of always adding to
1348 Severity : enhancement
1350 Description: Dynamic service threads
1351 Details : Within a small range, start extra service threads
1352 automatically when the request queue builds up.
1355 Frequency : mixed-endian client/server environments
1357 Description: mixed-endian crashes
1358 Details : The new msg_v2 system had some failures in mixed-endian
1361 Severity : enhancement
1363 Description: Easy OST removal
1364 Details : OSTs can be permanently deactivated with e.g. 'lctl
1365 conf_param lustre-OST0001.osc.active=0'
1367 Severity : enhancement
1369 Description: MGS proc entries
1370 Details : Added basic proc entries for the MGS showing what filesystems
1373 Severity : enhancement
1375 Description: provide MGS failover
1376 Details : Added config lock reacquisition after MGS server failover.
1378 Severity : enhancement
1380 Description: add Linux 2.4 support
1381 Details : Added support for RHEL 2.4.21 kernel for 1.6 servers and clients
1385 Description: a large application tries to do I/O to the same resource and dies
1386 in the middle of it.
1387 Details : Check the req->rq_arrival time after the call to
1388 ost_brw_lock_get(), but before we do anything about
1389 processing it & sending the BULK transfer request. This
1390 should help move old stale pending locks off the queue as
1391 quickly as obd_timeout.
1394 Frequency : when an incorrect nid is specified during startup
1396 Description: ptlrpc connect to non-existant node causes kernel crash
1397 Details : LNET can't be re-entered from an event callback, which
1398 happened when we expire a message after the export has been
1399 cleaned up. Instead, hand the zombie cleanup off to another
1402 Severity : enhancement
1404 Description: plain/inodebits lock performance improvement
1405 Details : Grouping plain/inodebits in granted list by their request modes
1406 and bits policy, thus improving the performance of search through
1410 Frequency : only if OST filesystem is corrupted
1412 Description: client incorrectly hits assertion in ptlrpc_replay_req()
1413 Details : for a short time RPCs with bulk IO are in the replay list,
1414 but replay of bulk IOs is unimplemented. If the OST filesystem
1415 is corrupted due to disk cache incoherency and then replay is
1416 started it is possible to trip an assertion. Avoid putting
1417 committed RPCs into the replay list at all to avoid this issue.
1420 Frequency : liblustre (e.g. catamount) on a large cluster with >= 8 OSTs/OSS
1422 Description: System hang on startup
1423 Details : This bug allowed the liblustre (e.g. catamount) client to
1424 return to the app before handling all startup RPCs. This
1425 could leave the node unresponsive to lustre network traffic
1426 and manifested as a server ptllnd timeout.
1428 Severity : enhancement
1430 Description: Add "/proc/sys/lustre/debug_peer_on_timeout"
1431 Details : liblustre envirable: LIBLUSTRE_DEBUG_PEER_ON_TIMEOUT
1432 boolean to control whether to print peer debug info when a
1433 client's RPC times out.
1436 Frequency : only for kernels with patches from Lustre below 1.4.3
1438 Description: Remove old rdonly API
1439 Details : Remove old rdonly API which unused from at least lustre 1.4.3
1442 Frequency : only for devices with external journals
1444 Description: Set external device read-only also
1445 Details : During a commanded failover stop, we set the disk device
1446 read-only while the server shuts down. We now also set any
1447 external journal device read-only at the same time.
1450 Frequency : when upgrading from 1.4 while trying to change parameters
1452 Description: The wrong (new) MDC name was used when setting parameters for
1453 upgraded MDT's. Also allows changing of OSC (and MDC)
1454 parameters if --writeconf is specified at tunefs upgrade time.
1457 Frequency : when setting specific ost indicies
1459 Description: QOS code breaks on skipped indicies
1460 Details : Add checks for missing OST indicies in the QOS code, so OSTs
1461 created with --index need not be sequential.
1463 Severity : enhancement
1465 Description: Add uninit_groups feature to ldiskfs2 to speed up e2fsck
1466 Details : The uninit_groups feature works in conjunction with the kernel
1467 filesystem code (ldiskfs2 only) and e2fsprogs-1.39-cfs6 to speed
1468 up the pass1 processing of e2fsck. This is a read-only feature
1469 in ldiskfs2 only, so older kernels and current ldiskfs cannot
1470 mount filesystems that have had this feature enabled.
1472 Severity : enhancement
1474 Description: Improve multi-block allocation algorithm to avoid fragmentation
1475 Details : The mballoc3 code (ldiskfs2 only) adds new mechanisms to improve
1476 allocation locality and avoid filesystem fragmentation.
1478 ------------------------------------------------------------------------------
1480 2007-04-01 Cluster File Systems, Inc. <info@clusterfs.com>
1482 * Support for kernels:
1483 2.4.21-47.0.1.EL (RHEL 3)
1484 2.6.5-7.283 (SLES 9)
1485 2.6.9-42.0.10.EL (RHEL 4)
1486 2.6.12.6 vanilla (kernel.org)
1487 2.6.16.27-0.9 (SLES 10)
1488 * Recommended e2fsprogs version: 1.39.cfs5
1490 * Note that reiserfs quotas are disabled on SLES 10 in this kernel
1494 Frequency : occasional, depends on client load and configuration
1495 Bugzilla : 12181, 12203
1496 Description: data loss for recently-modified files
1498 Details : In some cases it is possible that recently written or created
1499 files may not be written to disk in a timely manner (this should
1500 normally be within 30s unless client IO load is very high).
1501 The problem appears as zero-length files or files that are a
1502 multiple of 1MB in size after a client crash or client eviction
1503 that are missing data at the end of the file.
1505 This problem is more likely to be hit on clients where files are
1506 repeatedly created and unlinked in the same directory, clients
1507 have a large amount of RAM, have many CPUs, the filesystem has
1508 many OSTs, the clients are rebooted frequently, and/or the files
1509 are not accessed by other nodes after being written.
1511 The presence of the problem can be detected by looking at
1512 /proc/sys/fs/inode-state. If the first number (nr_inodes) is
1513 smaller than the second (nr_unused) then dirty files will not
1514 be flushed automatically to disk. "sync; sleep 10" should be
1515 run several times on the node before unmounting it to update
1516 Lustre (this is also safe to run on nodes without this problem).
1518 There is also a related kernel bug in the RHEL4 4 2.6.9 kernel
1519 that can cause this same problem, so customers using that kernel
1520 also need to update the kernel in addition to Lustre. In order
1521 to properly fix this bug, the RHEL3 2.4.21 kernel is also updated.
1523 It is normal that files written just before a client crash (less
1524 than 30s) may not yet have been flushed to disk, even for local
1528 Frequency : frequent on thin XT3 nodes
1530 Description: UUID collision on thin XT3 Linux nodes
1531 Details : UUIDs on Compute Node Linux XT3 nodes were not generated
1532 randomly, since we relied on an insufficiently-seeded PRNG.
1537 Description: OSS hangs after "All ost request buffers busy"
1538 Details : A deadlock between quota and journal operations caused OSS
1539 hangs after printing "All ost request buffers busy."
1542 Frequency : always on liblustre builds
1544 Description: Cleanup compiler warnings on liblustre
1547 Frequency : always on liblustre builds on XT3
1549 Description: LC_CONFIG_CDEBUG don't run while build liblustre on XT3.
1553 Description: Addition of EXT3_FEATURE_RO_COMPAT_DIR_NLINKS flag for
1554 > 32000 subdirectories
1555 Details : Add EXT3_FEATURE_RO_COMPAT_DIR_NLINK flag to
1556 EXT3_FEATURE_RO_COMPAT_SUPP. This flag will be set whenever
1557 subdirectory count crosses 32000. This will aid e2fsck to
1558 correctly handle more than 32000 subdirectories.
1561 Frequency : liblustre (e.g. catamount) on a large cluster with >= 8 OSTs/OSS
1563 Description: System hang on startup
1564 Details : This bug allowed the liblustre (e.g. catamount) client to
1565 return to the app before handling all startup RPCs. This
1566 could leave the node unresponsive to lustre network traffic
1567 and manifested as a server ptllnd timeout.
1569 Severity : enhancement
1571 Description: Add "/proc/sys/lustre/debug_peer_on_timeout"
1572 (liblustre envirable: LIBLUSTRE_DEBUG_PEER_ON_TIMEOUT)
1573 boolean to control whether to print peer debug info when a
1574 client's RPC times out.
1579 Description: make O_SYNC working on 2.6 kernels
1580 Details : 2.6 kernels use different method for mark pages for write,
1581 so need add a code to lustre for O_SYNC work.
1586 Description: Failure to close file and release space on NFS
1587 Details : Put inode details into lock acquired in ll_intent_file_open.
1588 Use mdc_intent_lock in ll_intent_open to properly
1589 detect all kind of errors unhandled by mdc_enqueue.
1594 Description: proc file read during shutdown sometimes raced obd removal,
1596 Details : Add lock to prevent obd access after proc file removal.
1599 Frequency : Only for files larger than 4GB on 32-bit clients.
1601 Description: improperly doing page alignment of locks
1602 Details : Modify lustre core code to use CFS_PAGE_* defines instead of
1603 PAGE_*. Make CFS_PAGE_MASK a 64-bit mask.
1608 Description: RPCs being resent when they shouldn't be
1609 Details : Some RPCs that should not be resent are being resent. This
1610 can cause inconsistencies in the RPC state machine. Do not
1611 resend such requests.
1614 Frequency : rare, only with NFS export
1616 Description: Crash on NFS re-export node
1617 Details : under very unusual load conditions an assertion is hit in
1618 ll_intent_file_open()
1621 Frequency : only if OST filesystem is corrupted
1623 Description: client incorrectly hits assertion in ptlrpc_replay_req()
1624 Details : for a short time RPCs with bulk IO are in the replay list,
1625 but replay of bulk IOs is unimplemented. If the OST filesystem
1626 is corrupted due to disk cache incoherency and then replay is
1627 started it is possible to trip an assertion. Avoid putting
1628 committed RPCs into the replay list at all to avoid this issue.
1633 Description: large O_DIRECT requests fail under memory pressure/fragmentation
1634 Details : Large single O_DIRECT read and write calls can fail to allocate
1635 a sufficiently large buffer to process the request. In case of
1636 allocation failure the allocation is retried with a smaller
1637 buffer and broken into smaller requests.
1639 Severity : enhancement
1641 Description: Add -o localflock option to simulate old noflock behaviour.
1642 Details : This will achieve local-only flock/fcntl locks coherentness.
1647 Description: versioning check is incomplete
1648 Details : Checking the version difference of client vs. server, report
1649 error if the gap is too big.
1654 Description: add support PG_writeback bit
1655 Details : add support for PG_writeback bit for Lustre, for more carefull
1656 work with page cache in 2.6 kernel. This also fix some deadlocks
1657 and remove hack for work O_SYNC with 2.6 kernel.
1659 Severity : enhancement
1661 Description: Add uninit_groups feature to ldiskfs2 to speed up e2fsck
1662 Details : The uninit_groups feature works in conjunction with the kernel
1663 filesystem code (ldiskfs2 only) and e2fsprogs-1.39-cfs6 to speed
1664 up the pass1 processing of e2fsck. This is a read-only feature
1665 in ldiskfs2 only, so older kernels and current ldiskfs cannot
1666 mount filesystems that have had this feature enabled.
1668 Severity : enhancement
1670 Description: Improve multi-block allocation algorithm to avoid fragmentation
1671 Details : The mballoc3 code (ldiskfs2 only) adds new mechanisms to improve
1672 allocation locality and avoid filesystem fragmentation.
1674 ------------------------------------------------------------------------------
1676 2007-02-09 Cluster File Systems, Inc. <info@clusterfs.com>
1678 * Support for kernels:
1679 2.6.9-42.0.3.EL (RHEL 4)
1680 2.6.5-7.276 (SLES 9)
1681 2.4.21-47.0.1.EL (RHEL 3)
1682 2.6.12.6 vanilla (kernel.org)
1683 2.6.16.21-0.8 (SLES10)
1684 * Recommended e2fsprogs version: 1.39.cfs2-0
1686 * The backwards-compatible /proc/sys/portals symlink has been removed
1687 in this release. Before upgrading, please ensure that you change
1688 any configuration scripts or /etc/sysctl.conf files that access
1689 /proc/sys/portals/* or sysctl portals.* to use the corresponding
1690 entry in /proc/sys/lnet or sysctl lnet.*. This change can be made
1691 in advance of the upgrade on any system running Lustre 1.4.6 or
1692 newer, since /proc/sys/lnet was added in that version.
1693 * Note that reiserfs quotas are disabled on SLES 10 in this kernel
1697 Frequency : only when quota is used
1699 Description: avoid scanning export list for quota master
1700 Details : Change the algorithms to avoid scanning export list in order
1701 to improve the efficiency.
1704 Frequency : MDS failover only, very rarely
1706 Description: "went back in time" messages on mds failover
1707 Details : The greatest transno may be lost when the current operation
1708 finishes with an error (transno==0) and the client's last_rcvd
1709 record is over-written. Save the greatest transno in the
1710 mds_last_transno for this case.
1713 Frequency : always for specific kernels and striping counts
1715 Description: client may get "Matching packet too big" without ACL support
1716 Details : Clients compiled without CONFIG_FS_POSIX_ACL get an error message
1717 when trying to access files in certain configurations. The
1718 clients should in fact be denied when mounting because they do
1719 not understand ACLs.
1722 Frequency : Cray XT3 with more than 4000 clients and multiple jobs
1724 Description: many clients connecting with IO in progress causes connect timeouts
1725 Details : Avoid synchronous journal commits to avoid delays caused by many
1726 clients connecting/disconnecting when bulk IO is in progress.
1727 Queue liblustre connect requests on OST_REQUEST_PORTAL instead of
1728 OST_IO_PORTAL to avoid delays behind potentially many pending
1732 Frequency : occasionally with multiple writers to a single file
1734 Description: shared writes to file may result in wrong size reported by stat()
1735 Details : Allow growing of kms when extent lock is cancelled
1738 Frequency : always with random mmap IO to multi-striped file
1740 Description: mmap write might be lost if we are writing to a 'hole' in stripe
1741 Details : Only if the hole is at the end of OST object so that kms is too
1742 small. Fix is to increase kms accordingly in ll_nopage.
1745 Frequency : rare, only if OST filesystem is inconsistent with MDS filesystem
1747 Description: writes to a missing object would leak memory on the OST
1748 Details : If there is an inconsistency between the MDS and OST filesystems,
1749 such that the MDS references an object that doesn't exist, writes
1750 to that object will leak memory due to incorrect cleanup in the
1751 error handling path, eventually running out of memory on the OST.
1756 Description: Creating too long symlink causes lustre errors
1757 Details : Check symlink and name lengths before sending requests to MDS.
1760 Frequency : only if flock is enabled (not on by default)
1762 Description: posix locks not released on fd closure on 2.6.9+
1763 Details : We failed to add posix locks to list of inode locks on 2.6.9+
1764 kernels, this caused such locks not to be released on fd close and
1765 then assertions on fs unmount about still used locks.
1768 Frequency : MDS failover only, very rarely
1770 Description: clients may get ASSERTION(granted_lock != NULL)
1771 Details : When request was taking a long time, and a client was resending
1772 a getattr by name lock request. The were multiple lock requests
1773 with the same client lock handle and
1774 mds_getattr_name->fixup_handle_for_resent_request found one of the
1775 lock handles but later failed with ASSERTION(granted_lock != NULL).
1780 Description: handle->h_buffer_credits > 0, assertion failure
1781 Details : h_buffer_credits is zero after truncate, causing assertion
1782 failure. This patch extends the transaction or creates a new
1786 Frequency : NFS re-export or patchless client
1787 Bugzilla : 11179, 10796
1788 Description: Crash on NFS re-export node (__d_move)
1789 Details : We do not want to hash the dentry if we don't have a lock.
1790 But if this dentry is later used in d_move, we'd hit uninitialised
1791 list head d_hash, so we just do this to init d_hash field but
1792 leave dentry unhashed.
1795 Frequency : NFS re-export or patchless client
1797 Description: NFS exports has problem with symbolic link
1798 Details : lustre client didn't properly install dentry when re-exported
1799 to NFS or running patchless client.
1802 Frequency : NFS re-export or patchless client
1804 Description: Various nfs/patchless fixes.
1805 Details : fixes reuse disconected alias for lookup process - this fixes
1806 warning "find_exported_dentry: npd != pd",
1807 fix permission error with open files at nfs.
1808 fix apply umask when do revalidate.
1811 Frequency : occasional
1813 Description: Crash on NFS re-export node
1814 Details : calling clear_page() on the wrong pointer triggered oops in
1815 generic_mapping_read().
1818 Frequency : rarely, using O_DIRECT IO
1820 Description: unaligned directio crashes client with LASSERT
1821 Details : check for unaligned buffers before trying any requests.
1824 Frequency : rarely, using CFS RAID5 patches in non-standard kernel series
1826 Description: stale data returned from RAID cache
1827 Details : If only a small amount of IO is done to the RAID device before
1828 reading it again it is possible to get stale data from the RAID
1829 cache instead of reading it from disk.
1832 Frequency : always for sles10 kernel
1834 Description: sles10 support
1835 Details : ll_follow_link: compile fixes and using of nd_set_link
1836 under newer kernels.
1839 Frequency : depends on arch, kernel and compiler version, always on sles10
1842 Description: recursive or deep enough symlinks cause stack overflow
1843 Details : getting rid of large stack-allocated variable in
1847 Frequency : depends on hardware
1849 Description: lustre write performance loss in the SLES10 kernel
1850 Details : the performance loss is caused by using of write barriers in the
1851 ext3 code. The SLES10 kernel turns barrier support on by
1852 default. The fix is to undo that change for ldiskfs.
1854 ------------------------------------------------------------------------------
1856 2006-12-09 Cluster File Systems, Inc. <info@clusterfs.com>
1858 * Support for kernels:
1859 2.6.9-42.0.3EL (RHEL 4)
1860 2.6.5-7.276 (SLES 9)
1861 2.4.21-47.0.1.EL (RHEL 3)
1862 2.6.12.6 vanilla (kernel.org)
1866 Frequency : quota enabled and large files being deleted
1868 Description: releasing more than 4GB of quota at once hangs OST
1869 Details : If a user deletes more than 4GB of files on a single OST it
1870 will cause the OST to spin in an infinite loop. Release
1871 quota in < 4GB chunks, or use a 64-bit value for 1.4.7.1+.
1876 Description: statfs data retrieved from /proc may be stale or zero
1877 Details : When reading per-device statfs data from /proc, in the
1878 {kbytes,files}_{total,free,avail} files, it may appear
1879 as zero or be out of date.
1882 Frequency : systems with MD RAID1 external journal devices
1884 Description: lconf's call to blkid is confused by RAID1 journal devices
1885 Details : Use the "blkid -l" flag to locate the MD RAID device instead
1886 of returning all block devices that match the journal UUID.
1889 Frequency : always, for aggregate stripe size over 4GB
1891 Description: "lfs setstripe" fails assertion when setting 4GB+ stripe width
1892 Details : Using "lfs setstripe" to set stripe size * stripe count over 4GB
1893 will fail the kernel with "ASSERTION(lsm->lsm_xfersize != 0)"
1896 Frequency : always if "lfs find" used on a local file/directory
1898 Description: "lfs find" segfaults if used on a local file/directory
1899 Details : The case where a directory component was not specified wasn't
1900 handled correctly. Handle this properly.
1903 Frequency : always on ppc64
1905 Description: the write to an ext3 filesystem mounted with mballoc got stuck
1906 Details : ext3_mb_generate_buddy() uses find_next_bit() which does not
1907 perform endianness conversion.
1910 Frequency : rarely (truncate to non-zero file size after write under load)
1911 Bugzilla : 10730, 10687
1912 Description: Files padded with zeros to next 4K multiple
1913 Details : With filesystems mounted using the "extents" option (2.6 kernels)
1914 it is possible that files that are truncated to a non-zero size
1915 immediately after being written are filled with zero bytes beyond
1916 the truncated size. No file data is lost.
1918 Severity : enhancement
1920 Description: Allow recovery/failover for liblustre clients.
1921 Details : liblustre clients were unaware of failover configurations until
1924 Severity : enhancement
1926 Description: user file locks should fail when not mounting with flock option
1927 Details : Set up an error-returning stub in ll_file_operations.lock field
1928 to prevent incorrect behaviour when client is mounted without
1929 flock option. Also, set up properly f_op->flock field for
1933 Frequency : always on ia64
1935 Description: "lfs df" loops on printing out MDS statfs information
1936 Details : The obd_ioctl_data was not initialized and in some systems
1937 this caused a failure during the ioctl that did not return
1938 an error. Initialize the struct and return an error on failure.
1941 Frequency : SLES 9 only
1943 Description: Error of copying files with lustre special EAs as root
1944 Details : Client side always return success for setxattr call for lustre
1945 special xattr (currently only "trusted.lov").
1948 Frequency : rarely on clusters with both ia64+i386 clients
1950 Description: ia64+i686 clients doing shared IO on the same file may LBUG
1951 Details : In rare cases when both ia64+i686 (or other mixed-PAGE_SIZE)
1952 clients are doing concurrent writes to the same file it is
1953 possible that the ia64 clients may LASSERT because the OST
1954 extent locks are not PAGE_SIZE aligned. Ensure that grown
1955 locks are always aligned on the request boundary.
1958 Frequency : specific use, occasional
1960 Description: Overwriting in use executable truncates on-disk binary image
1961 Details : If one node attempts to overwrite an executable in use by
1962 another node, we now correctly return ETXTBSY instead of
1963 truncating the file.
1965 Severity : enhancement
1967 Description: Async OSC create to avoid the blocking unnecessarily.
1968 Details : If a OST has no remain object, system will block on the creating
1969 when need to create a new object on this OST. Now, ways use
1970 pre-created objects when available, instead of blocking on an
1971 empty osc while others are not empty. If we must block, we block
1972 for the shortest possible period of time.
1977 Description: chmod on Lustre root is propagated to other clients
1978 Details : Re-validate root's dentry in ll_lookup_it to avoid having it
1979 invalid by the follow_mount time.
1982 Frequency : liblustre clients only
1984 Description: Race in 'instant cancel' lock handling could lead to such locks
1985 never to be granted in case of SMP MDS
1986 Details : Do not destroy not yet granted but cbpending locks in
1990 Frequency : replay/resend of open
1992 Description: non null lock assetion failure in mds_intent_policy
1993 Details : Trying to replay/resend lockless open requests resulted in
1994 mds_open() returning 0 with no lock. Now it sets a flag if
1995 a lock is going to be returned.
1997 Severity : enhancement
1999 Description: Checksum enhancements
2000 Details : New checksum enhancements allow for resending RPCs that failed
2003 Severity : enhancement
2005 Description: Tunables on number of dirty pages in cacche
2006 Details : Allow to set limit on number of dirty pages cached.
2011 Description: client crash on unmount - lock still has references
2012 Details : In some error handling cases it was possible to leak a lock
2013 reference on a client while accessing a file. This was not
2014 harmful to the client during operation, but would cause the
2015 client to crash when the filesystem is unmounted.
2018 Frequency : specific case, rare
2020 Description: ETXTBSY on mds though file not in use by client
2021 Details : ETXTBSY is no longer incorrectly returned when attempting to
2022 chmod or chown a directory that the user previously tried to
2023 execute or a currently-executing binary.
2026 Frequency : extremely rare except on liblustre-based clients
2028 Description: Lustre space not freed when files are deleted
2029 Details : Clean up open-unlinked files after client eviction. Previously
2030 the unlink was skipped and the files remained as orphans.
2035 Description: OST failure "would be an LBUG" in waiting_locks_callback()
2036 Details : In some cases it was possible to send a blocking callback to
2037 a client doing a glimpse, even though that client didn't get
2038 a lock granted. When the glimpse lock is cancelled on the OST
2039 the freed lock is left on the waiting list and corrupted the list.
2042 Frequency : all core dumps
2044 Description: Broke core dumps to lustre
2045 Details : Negative dentry may be unhashed if parent does not have UPDATE
2046 lock, but some callers, e.g. do_coredump, expect dentry to be
2047 hashed after successful create, hash it in ll_create_it.
2049 ------------------------------------------------------------------------------
2051 2006-09-13 Cluster File Systems, Inc. <info@clusterfs.com>
2053 * Support for kernels:
2054 2.6.9-42.0.2.EL (RHEL 4)
2055 2.6.5-7.276 (SLES 9)
2056 2.4.21-40.EL (RHEL 3)
2057 2.6.12.6 vanilla (kernel.org)
2061 Frequency : always on RHEL 3
2063 Description: Number of open files grows over time
2064 Details : The number of open files grows over time, whether or not
2065 Lustre is started. This was due to a filp leak introduced
2066 by one of our kernel patches.
2068 ------------------------------------------------------------------------------
2070 08-20-2006 Cluster File Systems, Inc. <info@clusterfs.com>
2072 * Support for kernels:
2073 2.6.9-42.EL (RHEL 4)
2074 2.6.5-7.276 (SLES 9)
2075 2.4.21-40.EL (RHEL 3)
2076 2.6.12.6 vanilla (kernel.org)
2081 Bugzilla : 5719, 9635, 9792, 9684
2082 Description: OST (or MDS) trips assertions in (re)connection under heavy load
2083 Details : If a server is under heavy load and cannot reply to new
2084 connection requests before the client resends the (re)connect,
2085 the connection handling code can behave badly if two service
2086 threads are concurrently handing separate (re)connections from
2087 the same client. Add better locking to the connection handling
2088 code, and ensure that only a single connection will be processed
2089 for a given client UUID, even if the lock is dropped.
2091 Severity : enhancement
2093 Description: add TCP zero-copy support to kernel
2094 Details : Add support to the kernel TCP stack to allow zero-copy bulk
2095 sends if the hardware supports scatter-gather and checksumming.
2096 This allows socklnd to do client-write and server-read more
2097 efficiently and reduce CPU utilization from skbuf copying.
2100 Frequency : only if NFS exporting from client
2102 Description: NULL pointer deref in ll_iocontrol() if chattr mknod file
2103 Details : If setting attributes on a file created under NFS that had
2104 never been opened it would be possible to oops the client
2105 if the file had no objects.
2108 Frequency : always for liblustre
2110 Description: liblustre client does MDS+OSTs setattr RPC for each write
2111 Details : When doing a write from a liblustre client, the client
2112 incorrectly issued an RPC to the MDS and each OST the file was
2113 striped over in order to update the timestamps. When writing
2114 with small chunks and many clients this could overwhelm the MDS
2115 with RPCs. In all cases it would slow down the write because
2116 these RPCs are unnecessary.
2118 Severity : enhancement
2120 Description: allow number of MDS service threads to be changed at module load
2121 Details : It is now possible to change the number of MDS service threads
2122 running. Adding "options mds mds_num_threads={N}" to the MDS's
2123 /etc/modprobe.conf will set the number of threads for the next
2124 time Lustre is restarted (assuming the "mds" module is also
2125 reloaded at that time). The default number of threads will
2126 stay the same, 32 for most systems.
2131 Description: OST crash if filesystem is unformatted or corrupt
2132 Details : If an OST is started on a device that has never been formatted
2133 or if the filesystem is corrupt and cannot even mount then the
2134 error handling cleanup routines would dereference a NULL pointer.
2139 Description: NULL pointer deref in llap_from_page.
2140 Details : get_cache_page_nowait can return a page with NULL (or otherwise
2141 incorrect) mapping if the page was truncated/reclaimed while it was
2142 searched for. Check for this condition and skip such pages when
2143 doing readahead. Introduce extra check to llap_from_page() to
2144 verify page->mapping->host is non-NULL (so page is not anonymous).
2147 Frequency : Sometimes when using sys_sendfile
2149 Description: "page not covered by a lock" warnings from ll_readpage
2150 Details : sendfile called ll_readpage without right page locks present.
2151 Now we introduced ll_file_sendfile that does necessary locking
2152 around call to generic_file_sendfile() much like we do in
2156 Frequency : with certain MDS communication failures at client mount time
2158 Description: NULL pointer deref after failed client mount
2159 Details : a client connection request may delayed by the network layer
2160 and not be sent until after the PTLRPC layer has timed out the
2161 request. If the client fails the mount immediately it will try
2162 to clean up before the network times out the request. Add a
2163 reference from the request import to the obd device and delay
2164 the cleanup until the network drops the request.
2167 Frequency : occasionally during client (re)connect
2169 Description: assertion failure during client (re)connect
2170 Details : processing a client connection request may be delayed by the
2171 client or server longer than the client connect timeout. This
2172 causes the client to resend the connection request. If the
2173 original connection request is replied in this interval, the
2174 client may trip an assertion failure in ptlrpc_connect_interpret()
2175 which thought it would be the only running connect process.
2178 Frequency : only with obd_echo servers and clients that are rebooted
2180 Description: kernel BUG accessing uninitialized data structure
2181 Details : When running an obd_echo server it did not start the ping_evictor
2182 thread, and when a client was evicted an uninitialized data
2183 structure was accessed. Start the ping_evictor in the RPC
2184 service startup instead of the OBD startup.
2186 Severity : enhancement
2187 Bugzilla : 10193 (patchless)
2188 Description: Remove dependency on various unexported kernel interfaces.
2189 Details : No longer need reparent_to_init, exit_mm, exit_files,
2190 sock_getsockopt, filemap_populate, FMODE_EXEC, put_filp.
2193 Frequency : rare (only users of deprecated and unsupported LDAP config)
2195 Description: write_conf for zeroconf mount queried LDAP incorrectly for client
2196 Details : LDAP apparently contains 'lustreName' attributes instead of
2197 'name'. A simple remapping of the name is sufficient.
2200 Frequency : rare (only with non-default dump_on_timeout debug enabled)
2202 Description: waiting_locks_callback trips kernel BUG if client is evicted
2203 Details : Running with the dump_on_timeout debug flag turned on makes
2204 it possible that the waiting_locks_callback() can try to dump
2205 the Lustre kernel debug logs from an interrupt handler. Defer
2206 this log dumping to the expired_lock_main() thread.
2208 Severity : enhancement
2210 Description: Support NFS exporting on 2.6 kernels.
2211 Details : Implement non-rawops metadata methods for NFS server to use without
2212 changing NFS server code.
2215 Frequency : very rare (synthetic metadata workload only)
2217 Description: two racing renames might cause an MDS thread to deadlock
2218 Details : Running the "racer" program may cause one MDS thread to rename
2219 a file from being the source of a rename to being the target of
2220 a rename at exactly the same time that another thread is doing
2221 so, and the second thread has already enqueued these locks after
2222 doing a lookup of the target and is trying to relock them in
2223 order. Ensure that we don't try to re-lock the same resource.
2226 Frequency : only very large systems with liblustre clients
2228 Description: slow eviction of liblustre clients with the "evict_by_nid" RPC
2229 Details : Use asynchronous set_info RPCs to send the "evict_by_nid" to
2230 all OSTs in parallel. This allows the eviction of stale liblustre
2231 clients to proceed much faster than if they were done in series,
2232 and also offers similar improvements for other set_info RPCs.
2237 Description: excessive CPU usage during initial read phase on client
2238 Details : During the initial read phase on a client, it would agressively
2239 retry readahead on the file, consuming too much CPU and impacting
2240 performance (since 1.4.5.8). Improve the readahead algorithm
2241 to avoid this, and also improve some other common cases (read
2242 of small files in particular, where "small" is files smaller than
2243 /proc/fs/lustre/llite/*/max_read_ahead_whole_mb, 2MB by default).
2248 Description: MDS crash when receiving packet with unknown intent.
2249 Details : Do not LBUG in unknown intent case, just return -EFAULT
2251 Severity : enhancement
2252 Bugzilla : 9293, 9385
2253 Description: MDS RPCs are serialised on client. This is unnecessary for some.
2254 Details : Do not serialize getattr (non-intent version) and statfs.
2257 Frequency : occasional, when OST network is overloaded/intermittent
2259 Description: client evicted by OST after bulk IO timeout
2260 Details : If a client sends a bulk IO request (read or write) the OST
2261 may evict the client if it is unresposive to its data GET/PUT
2262 request. This is incorrect if the network is overloaded (takes
2263 too long to transfer the RPC data) or dropped the OST GET/PUT
2264 request. There is no need to evict the client at all, since
2265 the pinger and/or lock callbacks will handle this, and the
2266 client can restart the bulk request.
2269 Frequency : Always when mmapping file with no objects
2271 Description: client crashes when mmapping file with no objects
2272 Details : Check that we actually have objects in a file before doing any
2273 operations on objects in ll_vm_open, ll_vm_close and
2279 Description: Request leak when working with deleted CWD
2280 Details : Introduce advanced request refcount tracking for requests
2281 referenced from lustre intent.
2283 Severity : Enhancement
2285 Description: Cache open file handles on client.
2286 Details : MDS now will return special lock along with openhandle, if
2287 requested and client is allowed to hold openhandle, even if unused,
2288 until such a lock is revoked. Helps NFS a lot, since NFS is opening
2289 closing files for every read/write openration.
2291 Severity : Enhancement
2293 Description: Cache open negative dentries on client when possible.
2294 Details : Guard negative dentries with UPDATE lock on parent dir, drop
2295 negative dentries on lock revocation.
2300 Description: Remounting a client read-only wasn't possible with a zconf mount
2301 Details : It wasn't possible to remount a client read-only with llmount.
2303 Severity : enhancement
2304 Description: Include MPICH 1.2.6 Lustre ADIO interface patch
2305 Details : In lustre/contrib/ or /usr/share/lustre in RPM a patch for
2306 MPICH is included to add Lustre-specific ADIO interfaces.
2307 This is based closely on the UFS ADIO layer and only differs
2308 in file creation, in order to allow the OST striping to be set.
2309 This is user-contributed code and not supported by CFS.
2314 Description: extended inode attributes (immutable, append-only) work improperly
2315 when 2.4 and 2.6 kernels are used on client/server or vice versa
2316 Details : Introduce kernel-independent values for these flags.
2318 Severity : enhancement
2321 Description: Allow fractional MB tunings for lustre in /proc/ filesystem.
2322 Details : Many of the /proc/ tunables can only be tuned at a megabyte
2323 granularity. Now, Fractional MB granularity is be supported,
2324 this is very useful for low memory system.
2326 Severity : enhancement
2328 Description: Getattr by fid
2329 Details : Getting a file attributes by its fid, obtaining UPDATE|LOOKUP
2330 locks, avoids extra getattr rpc requests to MDS, allows '/' to
2331 have locks and avoids getattr rpc requests for it on every stat.
2334 Frequency : Always, for filesystems larger than 2TB
2336 Description: ldiskfs crash at mount for filesystem larger than 2TB with mballoc
2337 Details : Kenrel kmalloc limits allocations to 128kB and this prevents
2338 filesystems larger than 2TB to be mounted with mballoc enabled.
2341 Frequency : Always, for 32-bit kernel without CONFIG_LBD and filesystem > 2TB
2343 Description: ldiskfs crash at mount for filesystem larger than 2TB with mballoc
2344 Details : If a 32-bit kernel is compiled without CONFIG_LBD enabled and a
2345 filesystems larger than 2TB is mounted then the kernel will
2346 silently corrupt the start of the filesystem. CONFIG_LBD is
2347 enabled for all CFS-supported kernels, but the possibility of
2348 this happening with a modified kernel config exists.
2350 Severity : enhancement
2352 Description: add client O_DIRECT support for 2.6 kernels
2353 Details : It is now possible to do O_DIRECT reads and writes to files
2354 in the Lustre client mountpoint on 2.6 kernel clients.
2356 Severity : enhancement
2358 Description: parallel glimpse, setattr, statfs, punch, destroy requests
2359 Details : Sends glimpse, setattr, statfs, punch, destroy requests to OSTs in
2360 parallel, not waiting for response from every OST before sending
2361 a rpc to the next OST.
2366 Description: setattr vs write race when updating file timestamps
2367 Details : Client processes that update a file timestamp into the past
2368 right after writing to the file (e.g. tar) it is possible that
2369 the updated file modification time can be reset to the current
2370 time due to a race between processing the setattr and write RPC.
2372 Severity : enhancement
2374 Description: Bring 'lfs find' closer in line with regular Linux find.
2375 Details : lfs find util supports -atime, -mtime, -ctime, -maxdepth, -print,
2376 -print0 options and obtains all the needed info through the lustre
2379 Severity : enhancement
2381 Description: support up to 1024 configured devices on one node
2382 Details : change obd_dev array from statically allocated to dynamically
2383 allocated structs as they are first used to reduce memory usage
2388 Description: Flush dirty partially truncated pages during truncate
2389 Details : Immediatelly flush partially truncated pages in filter_setattr,
2390 this way we completely avoid having any pages in page cache on OST
2391 and can retire ugly workarounds during writes to flush such pages.
2396 Description: i_sem vs transaction deadlock in mds_obd_destroy during unlink.
2397 Details : protect inode from truncation within vfs_unlink() context
2398 just take a reference before calling vfs_unlink() and release it
2399 when parent's i_sem is free.
2404 Description: last_id value checked outside lock on OST caused LASSERT failure
2405 Details : If there were multiple MDS->OST object precreate requests in
2406 flight, it was possible that the OST's last object id was checked
2407 outside a lock and incorrectly tripped an assertion. Move checks
2408 inside locks, and discard old precreate requests.
2411 Frequency : always, if extents are used on OSTs
2413 Description: index ei_leaf_hi (48-bit extension) is not zeroed in extent index
2414 Details : OSTs using the extents format would not zero the high 16 bits of
2415 the index physical block number. This is not a problem for any
2416 OST filesystems smaller than 16TB, and no kernels support ext3
2417 filesystems larger than 16TB yet. This is fixed in 1.4.7 (all
2418 new/modified files) and can be fixed for existing filesystems
2419 with e2fsprogs-1.39-cfs1.
2424 Description: import connection selection may be incorrect if timer wraps
2425 Details : Using a 32-bit jiffies timer with HZ=1000 may cause backup
2426 import connections to be ignored if the 32-bit jiffies counter
2427 wraps. Use a 64-bit jiffies counter.
2430 Frequency : very large clusters immediately after boot
2432 Description: LNET request buffers exhausted under heavy short-term load
2433 Details : If a large number of client requests are generated on a service
2434 that has previously never seen so many requests it is possible
2435 that the request buffer growth cannot keep up with the spike in
2436 demand. Instead of dropping incoming requests, they are held in
2437 the LND until the RPC service can accept more requests.
2440 Frequency : Sometimes during replay
2442 Description: Assertion failure in ll_local_open after replay.
2443 Details : If replay happened on an open request reply before we were able
2444 to set replay handler, reply will become not swabbed tripping the
2445 assertion in ll_local_open. Now we set the handler right after
2446 recognising of open request
2449 Frequency : very rare
2451 Description: kernel reports "badness in vsnprintf"
2452 Details : Reading from the "recovery_status" /proc file in small chunks
2453 may cause a negative length in lprocfs_obd_rd_recovery_status()
2454 call to vsnprintf() (which is otherwise harmless). Exit early
2455 if there is no more space in the output buffer.
2457 Severity : enhancement
2459 Description: clear OBD RPC statistics by writing to them
2460 Details : It is now possible to clear the OBD RPC statistics by writing
2461 to the "stats" file.
2466 Description: Client mtime is not the same on different clients after utimes
2467 Details : In some cases, the client was using the utimes() syscall on
2468 a file cached on another node. The clients now validate the
2469 ctime from the MDS + OSTs to determine which one is right.
2474 Description: Inability to activate failout mode
2475 Details : lconf script incorrectly assumed that in pythong string's numeric
2476 value is used in comparisons.
2479 Frequency : always with multiple stripes per file
2481 Description: Inefficient object allocation for mutli-stripe files
2482 Details : When selecting which OSTs to stripe files over, for files with
2483 a stripe count that divides evenly into the number of OSTs,
2484 the MDS is always picking the same starting OST for each file.
2485 Return the OST selection heuristic to the original design.
2490 Description: mount failures may take full timeout to return an error
2491 Details : Under some heavy load conditions it is possible that a
2492 failed mount can wait for the full obd_timeout interval,
2493 possibly several minutes, before reporting an error.
2494 Instead return an error as soon as the status is known.
2496 Frequency : quota enabled and large files being deleted
2498 Description: releasing more than 4GB of quota at once hangs OST
2499 Details : If a user deletes more than 4GB of files on a single OST it
2500 will cause the OST to spin in an infinite loop. Release
2501 quota in < 4GB chunks, or use a 64-bit value for 1.4.7.1+.
2506 Description: statfs data retrieved from /proc may be stale or zero
2507 Details : When reading per-device statfs data from /proc, in the
2508 {kbytes,files}_{total,free,avail} files, it may appear
2509 as zero or be out of date.
2512 Frequency : systems with MD RAID1 external journal devices
2514 Description: lconf's call to blkid is confused by RAID1 journal devices
2515 Details : Use the "blkid -l" flag to locate the MD RAID device instead
2516 of returning all block devices that match the journal UUID.
2519 Frequency : always, for aggregate stripe size over 4GB
2521 Description: assertion fails when trying to use 4GB stripe size
2522 Details : Use "setstripe" to set stripe size over 4GB will fail the kernel,
2523 complaining "ASSERTION(lsm->lsm_xfersize != 0)"
2526 Frequency : always on ppc64
2528 Description: the first write on an ext3 filesystem with mballoc got stuck
2529 Details : ext3_mb_generate_buddy() uses find_next_bit() which does not
2530 perform endianness conversion.
2532 ------------------------------------------------------------------------------
2534 02-14-2006 Cluster File Systems, Inc. <info@clusterfs.com>
2536 * WIRE PROTOCOL CHANGE. This version of Lustre networking WILL NOT
2537 INTEROPERATE with older versions automatically. Please read the
2538 user documentation before upgrading any part of a live system.
2539 * WARNING: Lustre networking configuration changes are required with
2540 this release. See https://bugzilla.clusterfs.com/show_bug.cgi?id=10052
2543 * Support for kernels:
2544 2.6.9-22.0.2.EL (RHEL 4)
2545 2.6.5-7.244 (SLES 9)
2546 2.6.12.6 vanilla (kernel.org)
2549 Severity : enhancement
2550 Bugzilla : 7981/8208
2551 Description: Introduced Lustre Networking (LNET)
2552 Details : LNET is new networking infrastructure for Lustre, it includes
2553 a reorganized network configuration mode (see the user
2554 documentation for full details) as well as support for routing
2555 between different network fabrics. Lustre Networking Devices
2556 (LNDs) for the supported network fabrics have also been
2557 created for this new infrastructure.
2559 Severity : enhancement
2560 Description: Introduced Access control lists
2561 Details : clients can set ACLs on files and directories in order to have
2562 more fine-grained permissions than the standard Unix UGO+RWX.
2563 The MDS must be started with the "-o acl" mount option.
2565 Severity : enhancement
2566 Description: Introduced filesystem quotas
2567 Details : Administrators may now establish per-user quotas on the
2570 Severity : enhancement
2572 Description: Configuration change for the XT3
2573 The PTLLND is now used to run Lustre over Portals on the XT3
2574 The configure option(s) --with-cray-portals are no longer used.
2575 Rather --with-portals=<path-to-portals-includes> is used to
2576 enable building on the XT3. In addition to enable XT3 specific
2577 features the option --enable-cray-xt3 must be used.
2582 Description: Running on many-way SMP OSTs can trigger oops in llcd_send()
2583 Details : A race between allocating a new llcd and re-getting the llcd_lock
2584 allowed another thread to grab newly-allocated llcd.
2586 Severity : enhancement
2588 Description: 2.6 OST async journal commit and locking fix to improve performance
2589 Details : The filter_direct_io()+filter_commitrw_write() journal commits for
2590 2.6 kernels are now async as they already were in 2.4 kernels so
2591 that they can commit concurrently with the network bulk transfer.
2592 For block-allocated files the filter allocation semaphore is held
2593 to avoid filesystem fragmentation during allocation. BKL lock
2594 removed for 2.6 xattr operations where it is no longer needed.
2599 Description: lconf incorrectly determined whether two IP networks could talk
2600 Details : In some more complicated routing and multiple-network
2601 configurations, lconf will avoid trying to make a network
2602 connection to a disjoint part of the IP space. It was doing the
2603 math incorrectly for one set of cases.
2608 Description: Fix for potential infinite loop processing records in an llog.
2609 Details : If an llog record is corrupted/zeroed, it is possible to loop
2610 forever in llog_process(). Validate the llog record length
2611 and skip the remainder of the block on error.
2614 Frequency : occasional (liblustre only)
2616 Description: liblustre could not open files whose last component is a symlink
2617 Details : sysio_path_walk() would incorrectly pass the open intent to
2618 intermediate path components.
2621 Frequency : rare (liblustre only with non-standard tuning)
2622 Bugzilla : 7201 (7350)
2623 Description: Tuning the MDC DLM LRU size to zero triggers client LASSERT
2624 Details : llu_lookup_finish_locks() tries to set lock data on a lock
2625 after it has been released, only do this for referenced locks
2627 Severity : enhancement
2629 Description: specifying an (invalid) directory default stripe_size of -1
2630 would reset the directory default striping
2631 Details : stripe_size -1 was used internally to signal directory stripe
2632 removal, now use "all default" to signal dir stripe removal
2633 as a directory striping of "all default" is not useful
2636 Frequency : common for large clusters running liblustre clients
2638 Description: doing an ls when liblustre clients are running is slow
2639 Details : sending a glimpse AST to a liblustre client waits for every AST
2640 to time out, as liblustre clients will not respond. Since they
2641 cannot cache data we refresh the OST lock LVB from disk instead.
2643 Severity : enhancement
2645 Description: doing an ls at the same time as file IO can be slow
2646 Details : enqueue and other "small" requests can be blocked behind many
2647 large IO requests. Create a new OST IO portal for non-IO
2648 requests so they can be processed faster.
2651 Frequency : rare (only HPUX clients mounting unsupported re-exported NFS vol)
2653 Description: an HPUX NFS client would get -EACCESS when ftruncate()ing a newly
2654 created file with mode 000
2655 Details : the Linux NFS server relies on an MDS_OPEN_OWNEROVERRIDE hack to
2656 allow an ftruncate() as a non-root user to a file with mode 000.
2657 Lustre now respects this flag to disable mode checks when
2658 truncating a file owned by the user
2661 Frequency : liblustre-only, when liblustre client dies unexpectedly or becomes
2664 Description: Revoking locks from clients that went dead or catatonic might take
2666 Details : New lock flags FL_CANCEL_ON_BLOCK used by liblustre makes
2667 cancellation of such locks instant on servers without waiting for
2668 any reply from clients. Clients drops these locks when cancel
2669 notification from server is received without replying.
2672 Frequency : liblustre-only, when liblustre client dies or becomes busy
2674 Description: Doing ls on Linux clients can take a long time with active
2676 Details : Liblustre client cannot handle ASTs in timely manner, so avoid
2677 granting such locks to it in the first place if possible. Locks
2678 are taken by proxy on the OST during the read or write and
2679 dropped immediately afterward. Add connect flags handling, do
2680 not grant locks to liblustre clients for glimpse ASTs.
2682 Severity : enhancement
2684 Description: Improve read-ahead algorithm to avoid excessive IO for random reads
2685 Details : Existing read-ahead algorithm is tuned for the case of streamlined
2686 sequential reads and behaves badly with applications doing random
2687 reads. Improve it by reading ahead at least read region, and
2688 avoiding excessive large RPC for small reads.
2690 Severity : enhancement
2692 Description: Creating more than 1000 files for a single job may cause a load
2693 imbalance on the OSTs if there are also a large number of OSTs.
2694 Details : qos_prep_create() uses an OST index reseed value that is an
2695 even multiple of the number of available OSTs so that if the
2696 reseed happens in the middle of the object allocation it will
2697 still utilize the OSTs as uniformly as possible.
2702 Description: OST or MDS may oops in ping_evictor_main()
2703 Details : ping_evictor_main() drops obd_dev_lock if deleting a stale export
2704 but doesn't restart at beginning of obd_exports_timed list
2707 Severity : enhancement
2709 Description: improve by-nid export eviction on the MDS and OST
2710 Details : allow multiple exports with the same NID to be evicted at one
2711 time without re-searching the exports list.
2714 Frequency : rare, only with supplementary groups enabled on SMP 2.6 kernels
2716 Description: MDS may oops in groups_free()
2717 Details : in rare race conditions a newly allocated group_info struct is
2718 freed again, and this can be NULL. The 2.4 compatibility code
2719 for groups_free() checked for a NULL pointer, but 2.6 did not.
2722 Frequency : common for liblustre clients doing little filesystem IO
2723 Bugzilla : 9352, 7313
2724 Description: server may evict liblustre clients accessing contended locks
2725 Details : if a client is granted a lock or receives a completion AST
2726 with a blocking AST already set it would not reply to the AST
2727 for LDLM_FL_CANCEL_ON_BLOCK locks. It now replies to such ASTs.
2730 Frequency : lfs setstripe, only systems with more than 160 OSTs
2732 Description: unable to set striping with a starting offset beyond OST 160
2733 Details : llapi_create_file() incorrectly limited the starting stripe
2734 index to the maximum single-file stripe count.
2737 Frequency : LDAP users only
2739 Description: lconf did not handle in-kernel recovery with LDAP properly
2740 Details : lconf/LustreDB get_refs() is searching the wrong namespace
2742 Severity : enhancement
2744 Description: bind OST threads to NUMA nodes to improve performance
2745 Details : all OST threads are uniformly bound to CPUs on a single NUMA
2746 node and do their allocations there to localize memory access
2748 Severity : enhancement
2750 Description: llmount can determine client NID directly from Myrinet (GM)
2751 Details : the client NID code from gmnalnid was moved directly into
2752 llmount, removing the need to use this or specifying the
2753 client NID explicitly when mounting GM clients with zeroconf
2756 Frequency : if client is started with down MDS
2758 Description: if client is started with down MDS mount hangs in ptlrpc_queue_wait
2759 Details : Having an LWI_INTR() wait event (interruptible, but no timeout)
2760 will wait indefinitely in ptlrpc_queue_wait->l_wait_event() after
2761 ptlrpc_import_delayed_req() because we didn't check if the
2762 request was interrupted, and we also didn't break out of the
2763 event loop if there was no timeout
2768 Description: data loss during non-page-aligned writes to a single file from
2769 both multiple nodes and multiple threads on one node at same time
2770 Details : updates to KMS and lsm weren't protected by common lock. Resulting
2771 inconsistency led to false short-reads, that were cached and later
2772 used by ->prepare_write() to fill in partially written page,
2773 leading to data loss.
2776 Frequency : always, if lconf --abort_recovery used
2778 Description: lconf --abort_recovery fails with 'Operation not supported'
2779 Details : lconf was attempting to abort recovery on the MDT device and not
2782 Severity : enhancement
2784 Description: remove cleanup logs
2785 Details : replace lconf-generated cleanup logs with lustre internal
2786 cleanup routines. Eliminates the need for client-cleanup and
2789 Severity : enhancement
2791 Description: add support for EAs (user and system) on lustre filesystems
2792 Details : it is now possible to store extended attributes in the Lustre
2793 client filesystem, and with the user_xattr mount option it
2794 is possible to allow users to store EAs on their files also
2796 Severity : enhancement
2798 Description: Add possibility (config option) to show minimal available OST free
2800 Details : When compiled with --enable-mindf configure option, statfs(2)
2801 (and so, df) will return least minimal free space available from
2802 all OSTs as amount of free space on FS, instead of summary of
2803 free spaces of all OSTs.
2805 Severity : enhancement
2807 Description: do not expand extent locks acquired on OST-side
2808 Details : Modify ldlm_extent_policy() to not expand local locks, acquired
2809 by server: they are not cached anyway.
2812 Frequency : when mmap is used/binaries executed from Lustre
2814 Description: Unmmap pages before throwing them away from read cache.
2815 Details : llap_shrink cache now attempts to unmap pages before discarding
2816 them (if unmapping failed - do not discard). SLES9 kernel has
2817 extra checks that trigger if this unmapping is not done first.
2822 Description: lconf didn't resolve symlinks before checking to see whether a
2823 given mountpoint was already in use
2826 Frequency : when migrating failover services
2827 Bugzilla : 6395, 9514
2828 Description: When migrating a subset of services from a node (e.g. failback
2829 from a failover service node) the remaining services would
2830 time out and evict clients.
2831 Details : lconf --force (implied by --failover) sets the global obd_timeout
2832 to 5 seconds in order to quickly disconnect, but this caused
2833 other RPCs to time out too quickly. Do not change the global
2834 obd_timeout for force cleanup, only set it for DISCONNECT RPCs.
2836 Severity : enhancement
2837 Frequency : if MDS is started with down OST
2838 Bugzilla : 9439,5706
2839 Description: Allow startup/shutdown of an MDS without depending on the
2840 availability of the OSTs.
2841 Details : Asynchronously call mds_lov_synchronize during MDS startup.
2842 Add appropriate locking and lov-osc refcounts for safe
2843 cleaning. Add osc abort_inflight calls in case the
2844 synchronize never started.
2847 Frequency : occasional (Cray XT3 only)
2849 Description: root not authorized to access files in CRAY_PORTALS environment
2850 Details : The client process capabilities were not honoured on the MDS in
2851 a CRAY_PORTALS/CRAY_XT3 environment. If the file had previously
2852 been accessed by an authorized user then root was able to access
2853 the file on the local client also. The root user capabilities
2854 are now allowed on the MDS, as this environment has secure UID.
2857 Frequency : occasional
2859 Description: ldiskfs "too long searching" message happens too often
2860 Details : A debugging message (otherwise harmless) prints too often on
2861 the OST console. This has been reduced to only happen when
2862 there are fragmentation problems on the filesystem.
2867 Description: Division by zero in statfs when all OSCs are inactive
2868 Details : lov_get_stripecnt() returns zero due to incorrect order of checks,
2869 lov_statfs divides by value returned by lov_get_stripecnt().
2873 Bugzilla : 9489, 3273
2874 Description: First write from each client to each OST was only 4kB in size,
2875 to initialize client writeback cache, which caused sub-optimal
2876 RPCs and poor layout on disk for the first writen file.
2877 Details : Clients now request an initial cache grant at (re)connect time
2878 and so that they can start streaming writes to the cache right
2879 away and always do full-sized RPCs if there is enough data.
2880 If the OST is rebooted the client also re-establishes its grant
2881 so that client cached writes will be honoured under the grant.
2886 Description: Slow ls (and stat(2) syscall) on files residing on IO-loaded OSTs
2887 Details : Now I/O RPCs go to different portal number and (presumably) fast
2888 lock requests (and glimses) and other RPCs get their own service
2889 threads pool that should be able to service those RPCs
2892 Severity : enhancement
2894 Description: Ability to exchange lustre version between client and servers and
2895 issue warnings at client side if client is too old. Also for
2896 liblustre clients there is ability to refuse connection of too old
2898 Details : New 'version' field is added to connect data structure that is
2899 filled with version info. That info is later checked by server and
2903 Frequency : rare, liblustre only.
2904 Bugzilla : 9296, 9581
2905 Description: Two simultaneous writes from liblustre at offset within same page
2906 might proceed at the same time overwriting eachother with stale
2908 Details : I/O lock withing llu_file_prwv was released too early, before data
2909 actually was hitting the wire. Extended lock-holding time until
2910 server acknowledges receiving data.
2913 Frequency : extremely rare. Never observed in practice.
2915 Description: avoid generating lustre_handle cookie of 0.
2916 Details : class_handle_hash() generates handle cookies by incrementing
2917 global counter, and can hit 0 occasionaly (this is unlikely, but
2918 not impossible, because initial value of cookie counter is
2919 selected randonly). Value of 0 is used as a sentinel meaning
2920 "unassigned handle" --- avoid it. Also coalesce two critical
2921 sections in this function into one.
2923 Severity : enhancement
2925 Description: allow liblustre clients to delegate truncate locking to OST
2926 Details : To avoid overhead of locking, liblustre client instructs OST to
2927 take extent lock in ost_punch() on client's behalf. New connection
2928 flag is added to handle backward compatibility.
2930 Severity : enhancement
2931 Bugzilla : 4928, 7341, 9758
2932 Description: allow number of OST service threads to be specified
2933 Details : a module parameter allows the number of OST service threads
2934 to be specified via "options ost ost_num_threads={N}" in the
2935 OSS's /etc/modules.conf or /etc/modprobe.conf.
2939 Bugzilla : 6146, 9635, 9895
2940 Description: servers crash with bad pointer in target_handle_connect()
2941 Details : In rare cases when a client is reconnecting it was possible that
2942 the connection request was the last reference for that export.
2943 We would temporarily drop the export reference and get a new
2944 one, but this may have been the last reference and the export
2945 was just destroyed. Get new reference before dropping old one.
2947 Severity : enhancement
2948 Frequency : if client is started with failover MDS
2950 Description: Allow multiple MDS hostnames in the mount command
2951 Details : Try to read the configuration from all specified MDS
2952 hostnames during a client mount in case the "primary"
2955 Severity : enhancement
2957 Description: Stop sending data to evicted clients as soon as possible.
2958 Details : Check if the client we are about to send or are sending data to
2959 was evicted already. (Check is done every second of waiting,
2960 for which l_wait_event interface was extended to allow checking
2961 of exit condition at specified intervals).
2964 Frequency : rare, normally only when NFS exporting is done from client
2966 Description: 'bad disk LOV MAGIC: 0x00000000' error when chown'ing files
2968 Details : Make mds_get_md() recognise empty md case and set lmm size to 0.
2971 Frequency : always, if srand() is called before liblustre initialization
2973 Description: Liblustre uses system PRNG disturbing its usage by user application
2974 Details : Introduce internal to lustre fast and high-quality PRNG for
2975 lustre usage and make liblustre and some other places in generic
2976 lustre code to use it.
2978 Severity : enhancement
2979 Bugzilla : 9477, 9557, 9870
2980 Description: Verify that the MDS configuration logs are updated when xml is
2981 Details : Check if the .xml configuration logs are newer than the config
2982 logs stored on the MDS and report an error if this is the case.
2983 Request --write-conf, or allow starting with --old_conf.
2985 Severity : enhancement
2987 Description: Handle symlinks in the path when checking if Lustre is mounted.
2988 Details : Resolve intermediate symlinks when checking if a client has
2989 mounted a filesystem to avoid duplicate client mounts.
2994 Description: lconf can hit an error exception but still return success.
2995 Details : The lconf command catches the Command error exception at the top
2996 level script context and will exit with the associated exit
2997 status, but doesn't ensure that this exit status is non-zero.
3002 Description: failure of ptlrpc thread startup can cause oops
3003 Details : Starting a ptlrpc service thread can fail if there are a large
3004 number of threads or the server memory is very fragmented.
3005 Handle this without oopsing.
3008 Frequency : always, only if liblustre and non-default acceptor port was used
3010 Description: liblustre cannot connect to servers with non-default acceptor port
3011 Details : tcpnal_set_default_params() was not called and was therefore
3012 ignoring the environment varaible TCPNAL_PORT, as well as other
3013 TCPNAL_ environment variables
3018 Description: two objects could be created on the same OST for a single file
3019 Details : If an OST is down, in some cases it was possible to create two
3020 objects on a single OST for a single file. No problems other
3021 than potential performance impact and spurious error messages.
3025 Bugzilla : 5681, 9562
3026 Description: Client may oops in ll_unhash_aliases
3027 Details : Client dcache may become inconsistent in race condition.
3028 In some cases "getcwd" can fail if the current directory is
3034 Description: Inode refcounting problems in NFS export code
3035 Details : link_raw functions used to call d_instantiate without obtaining
3036 extra inode reference first.
3040 Bugzilla : 9942, 9903
3041 Description: Referencing freed requests leading to crash, memleaks with NFS.
3042 Details : We used to require that call to ll_revalidate_it was always
3043 followed by ll_lookup_it. Also with revalidate_special() it is
3044 possible to call ll_revalidate_it() twice for the same dentry
3045 even if first occurence returned success. This fix changes semantic
3046 between DISP_ENQ_COMPLETE disposition flag to mean there is extra
3047 reference on a request referred from the intent.
3048 ll_intent_release() then releases such a request.
3051 Frequency : rare, normally benchmark loads only
3053 Description: unlinked inodes were kept in memory on the client
3054 Details : If a client is repeatedly creating and unlinking files it
3055 can accumulate a lot of stale inodes in the inode slab cache.
3056 If there is no other client load running this can cause the
3057 client node to run out of memory. Instead flush old inodes
3058 from client cache that have the same inode number as a new inode.
3061 Frequency : SLES9 2.6.5 kernel and long filenames only
3062 Bugzilla : 9969, 10379
3063 Description: utime reports stale NFS file handle
3064 Details : SLES9 uses out-of-dentry names in some cases, which confused
3065 the lustre dentry revalidation. Change it to always use the
3069 Frequency : rare, unless heavy write-truncate concurrency is continuous
3070 Bugzilla : 4180, 6984, 7171, 9963, 9331
3071 Description: OST becomes very slow and/or deadlocked during object unlink
3072 Details : filter_destroy() was holding onto the parent directory lock
3073 while truncating+unlinking objects. For very large objects this
3074 may block other threads for a long time and slow overall OST
3075 responsiveness. It may also be possible to get a lock ordering
3076 deadlock in this case, or run out of journal credits because of
3077 the combined truncate+unlink. Solution is to do object truncate
3078 first in one transaction without parent lock, and then do the
3079 final unlink in a new transaction with the parent lock. This
3080 reduces the lock hold time dramatically.
3083 Frequency : rare, 2.4 kernels only
3085 Description: MDS or OST cleanup may trip kernel BUG when dropping kernel lock
3086 Details : mds_cleanup() and filter_cleanup() need to drop the kernel lock
3087 before unmounting their filesystem in order to avoid deadlock.
3088 The kernel_locked() function in 2.4 kernels only checks whether
3089 the kernel lock is held, not whether it is this process that is
3090 holding it as 2.6 kernels do.
3095 Description: MDS or OST may oops/LBUG if a client is connecting multiple times
3096 Details : The client ptlrpc code may be trying to reconnect to a down
3097 server before a previous connection attempt has timed out.
3098 Increase the reconnect interval to be longer than the connection
3099 timeout interval to avoid sending duplicate connections to
3103 Frequency : echo_client brw_test command
3105 Description: fix echo_client to work with OST preallocated code
3106 Details : OST preallocation code (5137) didn't take echo_client IO path
3107 into account: echo_client calls filter methods outside of any
3108 OST thread and, hence, there is no per-thread preallocated
3109 pages and buffers to use. Solution: hijack pga pages for IO. As
3110 a byproduct, this avoids unnecessary data copying.
3114 Bugzilla : 3555, 5962, 6025, 6155, 6296, 9574
3115 Description: Client can oops in mdc_commit_close() after open replay
3116 Details : It was possible for the MDS to return an open request with no
3117 transaction number in mds_finish_transno() if the client was
3118 evicted, but without actually returning an error. Clients
3119 would later try to replay that open and may trip an assertion
3120 Simplify the client close codepath, and always return an error
3121 from the MDS in case the open is not successful.
3124 Frequency : rare, 2.6 OSTs only
3126 Description: OST may deadlock under high load on fragmented files
3127 Details : If there was a heavy load and highly-fragmented OST filesystems
3128 it was possible to have all the OST threads deadlock waiting on
3129 allocation of biovecs, because the biovecs were not released
3130 until the entire RPC IO was completed. Instead, release biovecs
3131 as soon as they are complete to ensure forward IO progress.
3133 Severity : enhancement
3135 Description: Support for specifying external journal device at mount
3136 Details : If an OST or MDS device is formatted with an external journal
3137 device, this device major/minor is stored in the ext3 superblock
3138 and may not be valid for failover. Allow detecting and
3139 specifying the external journal at mount time.
3144 Description: Mounting an MDS with pending unlinked files may cause oops
3145 Details : target_finish_recovery() calls mds_postrecov() which returned
3146 the number of orphans unlinked. mds_lov_connect->mds_postsetup()
3147 considers this an error and immediately begins cleaning up the
3148 lov, just after starting the mds_lov process
3150 Severity : enhancement
3152 Description: Implement 'lfs df' to report actual free space on per-OST basis
3153 Details : Add sub-command 'df' on 'lfs' to report the disk space usage of
3154 MDS/OSDs. Usage: lfs df [-i][-h]. Command Options: '-i' to report
3155 usage of objects; '-h' to report in human readable format.
3157 ------------------------------------------------------------------------------
3159 08-26-2005 Cluster File Systems, Inc. <info@clusterfs.com>
3166 Description: Mounting an ldiskfs file system with mballoc may crash OST node.
3167 Details : ldiskfs mballoc code may reference an uninitialized buddy struct
3168 at startup during orphan unlinking. Instead, skip buddy update
3169 before setup, as it will be regenerated after recovery is complete.
3174 Description: If an OST is inactive, its locks might reference stale inodes.
3175 Details : lov_change_cbdata() must iterate over all namespaces, even if
3176 they are inactive to clear inode references from the lock.
3178 Severity : enhancement
3179 Frequency : occasional, if non-standard max_dirty_mb used
3181 Description: Client will block write RPCs if not enough grant
3182 Details : If a client has max_dirty_mb smaller than max_rpcs_in_flight,
3183 then the client will block writes while waiting for another RPC
3184 to complete instead of consuming its dirty limit. With change
3185 we get improved performance when max_dirty_mb is small.
3187 Severity : enhancement
3188 Bugzilla : 3389, 6253
3189 Description: Add support for supplementary groups on the MDS.
3190 Details : The MDS has an upcall /proc/fs/lustre/mds/{mds}/group_upcall
3191 (set to /usr/sbin/l_getgroups if enabled) which will do MDS-side
3192 lookups for user supplementary groups into a cache.
3196 Description: O_CREAT|O_EXCL open flags in liblustre always return -EEXIST
3197 Details : Make libsysio to not enforce O_EXCL by clearing the flag,
3198 for liblustre O_EXCL is enforced by MDS.
3202 Description: readdir never returns NULL in liblustre.
3203 Details : Corrected llu_iop_getdirentries logic, to return offset of next
3204 dentry in struct dirent.
3208 Frequency : liblustre only, depends on application IO pattern
3209 Description: liblustre clients evicted if not contacting servers
3210 Details : Don't put liblustre clients into the ping_evictor list, so
3211 they will not be evicted by the pinger ever.
3213 Severity : enhancement
3215 Description: Add ability to evict clients by NID from MDS.
3216 Details : By echoing "nid:$NID" string into
3217 /proc/fs/lustre/mds/.../evict_client client with nid that equals to
3218 $NID would be instantly evicted from this MDS and from all active
3219 OSTs connected to it.
3223 Description: Do not query file size twice, somewhat slowing stat(2) calls.
3224 Details : lookup_it_finish() used to query file size from OSTs that was not
3229 Description: service threads change working directory to that of init
3230 Details : Starting lustre service threads may pin the working directory
3231 of the parent thread, making that filesystem busy. Threads
3232 now change to the working directory of init to avoid this.
3236 Frequency : during shutdown only
3237 Description: shutdown with a failed MDS or OST can cause unmount to hang
3238 Details : Don't resend DISCONNECT messages in ptlrpc_disconnect_import()
3243 Frequency : 2.6 only
3244 Description: chmod/chown may include an extra supplementary group
3245 Details : ll{,u}_mdc_pack_op_data() does not properly initialize the
3246 supplementary group and if none is specified this is used.
3249 Bugzilla : 5479 (6816)
3251 Description: Racing open + rm can assert client in mdc_set_open_replay_data()
3252 Details : If lookup is in progress on a file that is unlinked we might try
3253 to revalidate the inode and fail in revalidate after lookup is
3254 complete and ll_file_open() enqueues the open again but
3255 it_open_error() was not checking DISP_OPEN_OPEN errors correctly.
3258 Frequency : always, if lconf --abort_recovery used
3260 Description: lconf --abort_recovery fails with 'Operation not supported'
3261 Details : lconf was attempting to abort recovery on the MDT device and not
3264 ------------------------------------------------------------------------------
3266 2005-08-08 Cluster File Systems, Inc. <info@clusterfs.com>
3271 Frequency : rare (only unsupported configurations with a node running as an
3273 Bugzilla : 6514, 5137
3274 Description: Mounting a Lustre file system on a node running as an OST could
3276 Details : OSTs now preallocates memory needed to write out data at
3277 startup, instead of when needed, to avoid having to
3278 allocate memory in possibly low memory situations.
3279 Specifically, if the file system is mounted on on OST,
3280 memory pressure could force it to try to write out data,
3281 which it needed to allocate memory to do. Due to the low
3282 memory, it would be unable to do so and the node would
3283 become unresponsive.
3285 Severity : enhancement
3287 Description: Addition of lconf --service command line option
3288 Details : lconf now accepts a '--service <arg>' option, which is
3289 shorthand for 'lconf --group <arg> --select <arg>=<hostname>'
3291 Severity : enhancement
3293 Description: Failover mode is now the default for OSTs.
3294 Details : By default, OSTs will now run in failover mode. To return to
3295 the old behaviour, add '--failout' to the lmc line for OSTs.
3297 Severity : enhancement
3299 Description: Health checks are now provided for MDS and OSTs
3300 Details : Additional detailed health check information on MSD and OSTs
3301 is now provided through the procfs health_check value.
3304 Frequency : occasional, depends on IO load
3306 Description: Disk fragmentation on the OSTs could eventually cause slowdowns
3307 after numerous create/delete cycles
3308 Details : The ext3 inode allocation policy would not allocate new inodes
3309 very well on the OSTs because there are no new directories
3310 being created. Instead we look for groups with free space if
3311 the parent directories are nearly full.
3316 Description: Network or server problems during mount may cause partially
3317 mounted clients instead of returning an error.
3318 Details : The config llog parsing code may overwrite the error return
3319 code during mount error handling, returning success instead
3325 Description: MDS can fail to allocate large reply buffers
3326 Details : After long uptimes the MDS can fail to allocate large reply
3327 buffers (e.g. zconf client mount config records) due to memory
3328 fragmentation or consumption by the buffer cache. Preallocate
3329 some large reply buffers so that these replies can be sent even
3330 under memory pressure.
3334 Frequency : rare (liblustre)
3335 Description: fsx running with liblustre complained that using truncate() to
3336 extend the file doesn't work. This patch corrects that issue.
3337 Details : This is the liblustre equivalent of the fix for bug 6196. Fixes
3338 ATTR_SIZE and lsm use in llu_setattr_raw.
3342 Frequency : rare, only 2.6 kernels
3343 Description: Unusual file access patterns on the MDS may result in inode
3344 data being lost in very rare circumstances.
3345 Details : Bad interaction between the ea-in-inode patch and the "no-read"
3346 code in the 2.6 kernel caused the inode and/or EA data not to
3347 be read from disk, causing single-file corruption.
3351 Frequency : rare, only 2.6 filesystems using extents
3352 Description: Heavy concurrent write and delete load may cause data corruption.
3353 Details : It was possible under high-load situations to have an extent
3354 metadata block in the block device cache from a just-unlinked
3355 file overwrite a newly-allocated data block. We now unmap any
3356 metadata buffers that alias just-allocated data blocks.
3360 Frequency : filesystems with default stripe_count larger than 77
3361 Description: lconf+mke2fs fail when formatting filesystem with > 77 stripes
3362 Details : lconf specifies an inode size of 4096 bytes when the default
3363 stripe_count is larger than 77. This conflicts with the default
3364 inode density of 1 per 4096 bytes. Allocate smaller inodes in
3365 this case to avoid pinning too much memory for large EAs.
3367 ------------------------------------------------------------------------------
3369 2005-07-07 Cluster File Systems, Inc. <info@clusterfs.com>
3374 Frequency : rare (extremely heavy IO load with hundreds of clients)
3376 Description: Client is evicted, gets IO error writing to file
3377 Details : lock ordering changes for bug 5492 reintroduced bug 3267 and
3378 caused clients to be evicted for AST timeouts. The fixes in
3379 bug 5192 mean we no longer need to have such short AST timeouts
3380 so ldlm_timeout has been increased.
3383 Frequency : occasional during --force or --failover shutdown under load
3384 Bugzilla : 5949, 4834
3385 Description: Server oops/LBUG if stopped with --force or --failover under load
3386 Details : a collection of import/export refcount and cleanup ordering
3387 issues fixed for safer force cleanup
3390 Frequency : only filesystems larger than 120 OSTs
3391 Bugzilla : 5990, 6223
3392 Description: lfs getstripe would oops on a very large filesystem
3393 Details : lov_getconfig used kfree on vmalloc'd memory
3396 Frequency : only filesystems exporting via NFS to Solaris 10 clients
3397 Bugzilla : 6242, 6243
3398 Description: reading from files that had been truncated to a non-zero size
3399 but never opened returned no data
3400 Details : ll_file_read() reads zeros from no-object files to EOF
3405 Description: A bug in MDS/OSS recovery could cause the OSS to fail an assertion
3406 Details : There's little harm in aborting MDS/OSS recovery and letting it
3407 try again, so I removed the LASSERT and return an error instead.
3409 Severity : enhancement
3411 Description: New debugging infrastructure for tracking down data corruption
3412 Details : The I/O checksum code was replaced to: (a) control it at runtime,
3413 (b) cover more of the client-side code path, and (c) try to narrow
3414 down where problems occurred
3418 Bugzilla : 3819, 4364, 4397, 6313
3419 Description: Racing close and eviction MDS could cause assertion in mds_close
3420 Details : It was possible to get multiple mfd references during close and
3421 client eviction, leading to one thread referencing a freed mfd.
3423 Severity: : enhancement
3424 Bugzilla : 3262, 6359
3425 Description: Attempts to reconnect to servers are now more aggressive.
3426 Details : This builds on the enhanced upcall-less recovery that was added
3427 in 1.4.2. When trying to reconnect to servers, clients will
3428 now try each server in the failover group every 10 seconds. By
3429 default, clients would previously try one server every 25 seconds.
3434 Description: After recovery, certain operations trigger a failed
3435 assertion on a client.
3436 Details : Failing over an mds, using lconf -d --failover, while a
3437 client was doing a readdir() call would cause the client to
3438 LBUG after recovery completed and the readdir() was resent.
3440 Severity : enhancement
3442 Description: Default groups are now added by lconf
3443 Details : You can now run lconf --group <servicename> without having to
3444 manually add groups with lmc.
3447 Frequency : occasional
3449 Description: Nodes with an elan id of 0 trigger a failed assertion
3452 Frequency : always when accessing e.g. tty/console device nodes
3454 Description: tty and some other devices nodes cannot be used on lustre
3455 Details : file's private_data field is used by device data and lustre
3456 values in there got lost. New field was added to struct file to
3457 store fs-specific private data.
3460 Frequency : when exporting Lustre via NFS
3462 Description: NFSD failed occasionally when looking up a path component
3463 Details : NFSD is looking up ".." which was broken in ext3 directories
3464 that had grown large enough to become hashed.
3467 Frequency : Clusters with multiple interfaces not on the same subnet
3469 Description: Nodes will repeatedly try to reconnect to an interface which it
3470 cannot reach and report an error to the log.
3471 Details : Extra peer list entries will be created by lconf with some peers
3472 unreachable. lconf now validates the peer before adding it.
3475 Frequency : Only if a default stripe is set on the filesystem root.
3477 Description: Setting a default stripe on the filesystem root prevented the
3478 filesystem from being remounted.
3479 Details : The client was sending extra request flags in the root getattr
3480 request and did not allocate a reply buffer for the dir EA.
3483 Frequency : occasional, higher if lots of files are accessed by one client
3484 Bugzilla : 6159, 6097
3485 Description: Client trips assertion regarding lsm mismatch/magic
3486 Details : While revalidating inodes the VFS looks up inodes with ifind()
3487 and in rare cases can find an inode that is being freed.
3488 The ll_test_inode() code will free the lsm during ifind()
3489 when it finds an existing inode and then the VFS later attaches
3490 this free lsm to a new inode.
3494 Bugzilla : 6422, 7030
3495 Description: MDS deadlock between mkdir and client eviction
3496 Details : Creating a new file via mkdir or mknod (starting a transaction
3497 and getting the ns lock) can deadlock with client eviction
3498 (gets ns lock and trying to finish a synchronous transaction).
3501 Frequency : occasional
3502 Description: While starting a server, the fsfilt_ext3 module could not be
3504 Details : CFS's improved ext3 filesystem is named ldiskfs for 2.6
3505 kernels. Previously, lconf would still use the ext3 name
3506 when trying to load modules. Now, it will correctly use
3507 ext3 on 2.4 and ldiskfs on 2.6.
3509 Severity : enhancement
3510 Description: The default stripe count has been changed to 1
3511 Details : The interpretation of the default stripe count (0, to lfs
3512 or lmc) has been changed to mean striping across a single
3513 OST, rather than all available. For general usage we have
3514 found a stripe count of 1 or 2 works best.
3516 Severity : enhancement
3517 Description: Add support for compiling against Cray portals.
3518 Details : Conditional compiling for some areas that are different
3522 Frequency : occasional
3523 Bugzilla : 6409, 6834
3524 Description: Creating files with an explicit stripe count may lead to
3525 a failed assertion on the MDS
3526 Details : If some OSTs are full or unavailable, creating files may
3527 trigger a failed assertion on the MDS. Now, Lustre will
3528 try to use other servers or return an error to the
3532 Frequency : occasional
3534 Description: Multiple concurrent overlapping read+write on multiple SMP nodes
3535 caused lock timeout during readahead (since 1.4.2).
3536 Details : Processes doing readahead might match a lock that hasn't been
3537 granted yet if there are overlapping and conflicting lock
3538 requests. The readahead process waits on ungranted lock
3539 (original lock is CBPENDING), while OST waits for that process
3540 to cancel CBPENDING read lock and eventually evicts client.
3542 Severity : enhancement
3544 Description: Initial enabling of flock support for clients
3545 Details : Implements fcntl advisory locking and file status functions.
3546 This feature is provided as an optional mount flag (default
3547 off), and is NOT CURRENTLY SUPPORTED. Not all types of record
3548 locking are implemented yet, and those that are are not guaranteed
3549 to be completely correct in production environments.
3550 mount -t lustre -o [flock|noflock] ...
3553 Frequency : occasional
3555 Description: OSTs running 2.4 kernels but with extents enabled might trip an
3556 assertion in the ext3 JBD (journaling) layer.
3557 Details : The b_committed_data struct is protected by the big kernel lock
3558 in 2.4 kernels, serializing journal_commit_transaction() and
3559 ext3_get_block_handle->ext3_new_block->find_next_usable_block()
3560 access to this struct. In 2.6 kernels there is finer grained
3561 locking to improve SMP performance of the JBD layer.
3565 Description: Changes the "SCSI I/O Stats" kernel patch to default to "enabled"
3570 Description: merge and cleanup kernel patches.
3573 -----------------------------------------------------------------------------
3575 2005-05-05 Cluster File Systems, Inc. <info@clusterfs.com>
3577 NOTE: Lustre 1.4.2 uses an incompatible network protocol than previous
3578 versions of Lustre. Please update all servers and clients to
3579 version 1.4.2 or later at the same time. You must also run
3580 "lconf --write-conf {config}.xml" on the MDS while it is stopped
3581 to update the configuration logs.
3583 - fix for HPUX NFS client breakage when NFS exporting Lustre (5781)
3584 - mdc_enqueue does not need max_mds_easize request buffer on send (5707)
3585 - swab llog records of type '0' so we get proper header size/idx (5861)
3586 - send llog cancel req to DLM cancel portal instead of cb portal (5515)
3587 - fix rename of one directory over another leaking an inode (5953)
3588 - avoid SetPageDirty on 2.6 (5981)
3589 - don't re-add just-being-destroyed locks to the waiting list (5653)
3590 - when creating new directories, inherit the parent's custom
3591 striping settings if present parent (3048)
3592 - flush buffers from cache before direct IO in 2.6 obdfilter (4982)
3593 - don't hold i_size_sem in ll_nopage() and ll_ap_refresh_count (6077)
3594 - don't hold client locks on temporary worklist from l_lru (5666)
3595 - handle IO errors in 2.6 obdfilter bio completion routine (6046)
3596 - automatically evict dead clients (5921)
3597 - Update file size properly in create+truncate+fstat case (6196)
3598 - Do not unhash mountpoint dentries, do not allow removal of
3600 - Avoid lock ordering deadlock issue with write/truncate (6203,5654)
3601 - reserve enough journal credits in fsfilt_start_log for setattr (4554)
3602 - ldlm_enqueue freed-export error path would always LBUG (6149,6184)
3603 - don't reference lr_lvb_data until after we hold lr_lvb_sem (6170)
3604 - don't overwrite last_rcvd if there is a *_client_add() error (6086)
3605 - Correctly handle reads of files with no objects (6243)
3606 - lctl recover will also mark a device active if deactivate used (5933)
3608 - by default create 1 inode per 4kB space on MDS, per 16kB on OSTs
3609 - allow --write-conf on an MDS with different nettype than client (5619)
3610 - don't write config llogs to MDS for mounts not from that MDS (5617)
3611 - lconf should create multiple TCP connections from a client (5201)
3612 - init scripts are now turned off by default; run chkconfig --on
3613 lustre and chkconfig --on lustrefs to use them
3614 - upcalls are no longer needed for clients to recover to failover
3616 - add --abort-recovery option to lconf to abort recovery on device
3618 - add support for an arbitrary number of OSTs (3026)
3619 - Quota support protocol changes.
3620 - forward compatibility changes to wire structs (6007)
3621 - rmmod NALs that might be loaded because of /etc/modules.conf (6133)
3622 - support for mountfsoptions and clientoptions to the Lustre LDAP (5873)
3623 - improved "lustre status" script
3624 - initialize blocksize for non-regular files (6062)
3625 - added --disable-server and --disable-client configure options (5782)
3626 - introduce a lookup cache for lconf to avoid repeated DB scans (6204)
3627 - Vanilla 2.4.29 support
3628 - increase maximum number of obd devices to 520 (6242)
3629 - remove the tcp-zero-copy patch from the suse-2.4 series (5902)
3630 - Quadrics Elan drivers are now included for the RHEL 3 2.4.21 and
3631 SLES 9 2.6.5 kernels
3632 - limit stripes per file to 160 (the maximum EA size) (6093)
3634 2005-03-22 Cluster File Systems, Inc. <info@clusterfs.com>
3637 - don't LASSERT in ll_release on NULL lld with NFS export (4655, 5760)
3638 - hold NS lock when calling handle_ast_error->del_waiting_lock (5746)
3639 - fix setattr mtime regression from lovcleanup merge (4829, 5669)
3640 - workaround for 2.6 crash in ll_unhash_aliases (5687, 5210)
3641 - small ext3 extents cleanups and fixes (5733)
3642 - improved mballoc code, several small races and bugs fixed (5733, 5638)
3643 - kernel version 43 - fix remove_suid bugs in both 2.4 and 2.6 (5695)
3644 - avoid needless client->OST connect, fix handle mismatch (5317)
3645 - fix DLM error path that led to out-of-sync client, long delays (5779)
3646 - support common vfs-enforced mount options (nodev,nosuid,noexec) (5637)
3647 - fix several locking issues related to i_size (5492,5624,5654,5672)
3648 - don't move pending lock onto export if it is already evicted (5683)
3649 - fix kernel oops when creating .foo in unlinked directory (5548)
3650 - fix deadlock in obdfilter statistics vs. object create (5811)
3651 - use time_{before,after} to avoid timer jiffies wrap (5882)
3652 - shutdown --force/--failover stability (3607,3651,4797,5203,4834)
3653 - Do not leak request if server was not able to process it (5154)
3654 - If mds_open unable to find parent dir, make that negative lookup(5154)
3655 - don't create new directories with extent-mapping (5909, 5936)
3657 - fix lustre/lustrefs init scripts for SuSE (patch from Scali, 5702)
3658 - don't hold the pinger_sem in ptlrpc_pinger_sending_on_import
3659 - change obd_increase_kms to obd_adjust_kms (up or down) (5654)
3660 - lconf, lmc search both /usr/lib and /usr/lib64 for Python libs (5800)
3661 - support for RHEL4 kernel on i686 (5773)
3662 - provide error messages when incompatible logs are encountered (5898)
3664 2005-02-18 Cluster File Systems, Inc. <info@clusterfs.com>
3665 * version 1.4.0.10 (1.4.1 release candidate 1)
3667 - don't keep a lock reference when lock is not granted (4238)
3668 - unsafe list practices (rarely) led to infinite eviction loop (4908)
3669 - add per-fs limit of Lustre pages in page cache, avoid OOM (4699)
3670 - drop import inflight refcount on signal_completed_replay error (5255)
3671 - unlock page after async write error during send (3677)
3672 - handle missing objects in filter_preprw_read properly (5265)
3673 - no transno return for symlink open, don't save no-trasno open (3440)
3674 - don't try to complete elan receive that already failed (4012)
3675 - free RPC server reply state on error (5406)
3676 - clean up thread from ptlrpc_start_thread() on error (5160)
3677 - readahead could read extra page into cache that wasn't ejected (5388)
3678 - prevent races in class_attach/setup/cleanup/detach (5260)
3679 - don't dereference de->d_inode after l_dput of de (5458)
3680 - use "int" for stripe value returned from lock_to_stripe (5544)
3681 - mballoc allocation and error-checking fixes in 2.6 (5504)
3682 - block device patches to fix I/O request sizes in 2.6 (5482)
3683 - look up hostnames for IB nals (5602)
3684 - 2.6 changed lock ordering of 2 semaphores, caused deadlock (5654)
3685 - don't start multiple acceptors for the same port (5277)
3686 - fix incorrect LASSERT in mds_getattr_name (5635)
3687 - export a proc file for general "ping" checking (5628)
3688 - fix "lfs check" to not block when the MDS is down (5628)
3690 - service request history (4965)
3691 - put {ll,lov,osc}_async_page structs in a single slab (4699)
3692 - create an "evict_client" /proc entry on OSTs, like the MDS has
3693 - fix mount usage message, return errors per mount(8) (5168)
3694 - change grep [] to grep "[]" in tests so they work in more UMLs
3695 - fix ppc64/x86_64 spec to use %{_libdir} instead of /usr/lib (5389)
3696 - remove ancient LOV_MAGIC_V0 EA support (5047)
3697 - add "disk I/Os in flight" and "I/O req time" stats in obdfilter
3698 - align r/w RPCs to PTLRPC_MAX_BRW_SIZE boundary for performance (3451)
3699 - allow readahead allocations to fail when low on memory (5383)
3700 - mmap locking landed again, after considerable improvement (2828)
3701 - add get_hostaddr() to lustreDB.py for LDAP support (5459)
3703 2004-11-23 Cluster File Systems, Inc. <info@clusterfs.com>
3706 - send OST transaction number in read/write reply to free req (4966)
3707 - don't ASSERT in ptl_send_rpc() if we run out of memory (5119)
3708 - lock /proc/sys/portals/routes internal state, avoiding oops (4827)
3709 - the watchdog thread now runs as interruptible (5246)
3710 - flock/lockf fixes (but it's still disabled, pending 5135)
3711 - don't use EXT3 constants in llite code (5094)
3712 - memory shortage at startup could cause assertion (5176)
3714 - reorganization of lov code
3715 - single portals codebase
3717 - add extents/mballoc support (5025)
3718 - direct I/O reads in the obdfilter (4048)
3719 - kernel patches from LNXI for 2.6 (bluesmoke, perfctr, mtd, kexec)
3721 tbd Cluster File Systems, Inc. <info@clusterfs.com>
3724 - send OST transaction number in read/write reply to free req (4966)
3725 - don't ASSERT in ptl_send_rpc() if we run out of memory (5119)
3726 - lock /proc/sys/portals/routes internal state, avoiding oops (4827)
3727 - the watchdog thread now runs as interruptible (5246)
3728 - handle missing objects in filter_preprw_read properly (5265)
3729 - unsafe list practices (rarely) led to infinite eviction loop (4908)
3730 - drop import inflight refcount on signal_completed_replay error (5255)
3731 - unlock page after async write error during send (3677)
3732 - return original error code on reconstructed replies (3761)
3733 - no transno return for symlink open, don't save no-trasno open (3440)
3735 - add pid to ldlm debugging output (4922)
3736 - bump the watchdog timeouts -- we can't handle 30sec yet
3737 - extra debugging for orphan dentry/inode bug (5259)
3739 2004-11-16 Cluster File Systems, Inc. <info@clusterfs.com>
3742 - fix TCP_NODELAY bug, which caused extreme perf regression (5134)
3743 - allocate qswnal tx descriptors singly to avoid fragmentation (4504)
3744 - don't LBUG on obdo_alloc() failure, use OBD_SLAB_ALLOC() (4800)
3745 - fix NULL dereference in /proc/sys/portals/routes (4827)
3746 - allow failed mdc_close() operations to be interrupted (4561)
3747 - stop precreate on OST before MDS would time out on it (4778)
3748 - don't send partial-page writes before EOF from client (4410)
3749 - discard client grant for sub-page writes on large-page clients (4520)
3750 - don't free dentries not owned by NFS code, check generation (4806)
3751 - fix lsm leak if mds_create_objects() fails (4801)
3752 - limit debug_daemon file size, always print CERROR messages (4789)
3753 - use transno after validating reply (3892)
3754 - process timed out requests if import state changes (3754)
3755 - update mtime on OST during writes, return in glimpse (4829)
3756 - add mkfsoptions to LDAP (4679)
3757 - use ->max_readahead method instead of zapping global ra (5039)
3758 - don't interrupt __l_wait_event() during strace
3760 - add software watchdogs to catch hung threads quickly (4941)
3761 - make lustrefs init script start after nfs is mounted
3762 - fix CWARN/ERROR duplication (4930)
3763 - return async write errors to application if possible (2248)
3764 - add /proc/sys/portal/memused (bytes allocated by PORTALS_ALLOC)
3765 - print NAL number in %x format (4645)
3766 - update barely-supported suse-2.4.21-171 series (4842)
3767 - support for sles 9 %post scripts
3768 - support for building 2.6 kernel-source packages
3769 - support for sles km_* packages
3771 2004-10-07 Cluster File Systems, Inc. <info@clusterfs.com>
3774 - ignore -ENOENT errors in osc_destroy (3639)
3775 - notify osc create thread that OSC is being cleaned up (4600)
3776 - add nettype argument for llmount in #5d in conf-sanity.sh (3936)
3777 - reconstruct ost_handle() like mds_handle() (4657)
3778 - create a new thread to do import eviction to avoid deadlock (3969)
3779 - let lconf resolve symlinked-to devices (4629)
3780 - don't unlink "objects" from directory with default EA (4554)
3781 - hold socknal file ref over connect in case target is down (4394)
3782 - allow more than 32000 subdirectories in a single directory (3244)
3783 - fix blocks count for O_DIRECT writes (3751)
3784 - OST returns ENOSPC from object create when no space left (4539)
3785 - don't send truncate RPC if file size isn't changing (4410)
3786 - limit OSC precreate to 1/2 of value OST considers bogus (4778)
3787 - bind to privileged port in socknal and tcpnal (3689)
3789 - rate limit CERROR/CWARN console message to avoid overload (4519)
3790 - GETFILEINFO dir ioctl returns LOV EA + MDS stat in 1 call (3327)
3791 - basic mmap support (3918)
3792 - kernel patch series update from b1_4 (4711)
3794 2004-09-16 Cluster File Systems, Inc. <info@clusterfs.com>
3797 - avoid crash during MDS cleanup with OST shut down (2775)
3798 - fix loi_list_lock/oig_lock inversion on interrupted IO (4136)
3799 - don't use bad inodes on the MDS (3744)
3800 - dynamic object preallocation to improve recovery speed (4236)
3801 - don't hold spinlock over lock dumping or change debug flags (4401)
3802 - don't zero obd_dev when it is force cleaned (3651)
3803 - print grants to console if they go negative (4431)
3804 - "lctl deactivate" will stop automatic recovery attempts (3406)
3805 - look for existing locks in ldlm_handle_enqueue() (3764)
3806 - don't resolve lock handle twice in recovery avoiding race (4401)
3807 - revalidate should check working dir is a directory (4134)
3809 - don't always mark "slow" obdfilter messages as errors (4418)
3811 2004-08-24 Cluster File Systems, Inc. <info@clusterfs.com>
3814 - don't close LustreDB during write_conf until it is done (3860)
3815 - fix typo in lconf for_each_profile (3821)
3816 - allow dumping logs from multiple threads at one time (3820)
3817 - don't allow multiple threads in OSC recovery (3812)
3818 - fix debug_size parameters (3864)
3819 - fix mds_postrecov to initialize import for llog ctxt (3121)
3820 - replace config semaphore with spinlock (3306)
3821 - be sure to send a reply for a CANCEL rpc with bad export (3863)
3822 - don't allow enqueue to complete on a destroyed export (3822)
3823 - down write_lock before checking llog header bitmap (3825)
3824 - recover from lock replay timeout (3764)
3825 - up llog sem before sending rpc (3652)
3826 - reduce ns lock hold times when setting kms (3267)
3827 - change a dlm LBUG to LASSERTF, to maybe learn something (4228)
3828 - fix NULL deref and obd_dev leak on setup error (3312)
3829 - replace some LBUG about llog ops with error handling (3841)
3830 - don't match INVALID dentries from d_lookup and spin (3784)
3831 - hold dcache_lock while marking dentries INVALID and hashing (4255)
3832 - fix invalid assertion in ptlrpc_set_wait (3880)
3834 - add libwrap support for the TCP acceptor (3996)
3835 - add /proc/sys/portals/routes for non-root route listing (3994)
3836 - allow setting MDS UUID in .xml (2580)
3837 - print the stack of a process that LBUGs (4228)
3839 2004-07-14 Cluster File Systems, Inc. <info@clusterfs.com>
3842 - don't cleanup request in ll_file_open() on failed MDS open (3430)
3843 - make sure to unset replay flag from failed open requests (3440)
3844 - if default stripe count is 0, use OST count for inode size (3636)
3845 - update parent mtime/ctime on client for create/unlink (2611)
3846 - drop dentry ref in ext3_add_link from open_connect_dentry (3266)
3847 - free recovery state on server during a forced cleanup (3571)
3848 - unregister_reply for resent reqs (3063)
3849 - loop back devices mounting and status check on 2.6 (3563)
3850 - fix resource-creation race that can provoke i_size == 0 (3513)
3851 - don't try to use bad inodes returned from MDS/OST fs lookup (3688)
3852 - more debugging for page-accounting assertion (3746)
3853 - return -ENOENT instead of asserting if ost getattr+unlink race (3558)
3854 - avoid deadlock after precreation failure (3758)
3855 - fix race and lock order deadlock in orphan handling (3450, 3750)
3856 - add validity checks when grabbing inodes from l_ast_data (3599)
3858 - add /proc/.../recovery_status to obdfilter (3428)
3859 - lightweight CDEBUG infrastructure, debug daemon (3668)
3860 - change default OSC RPC parameters to be better on small clusters
3861 - turn off OST read cache for files smaller than 32MB
3862 - install man pages and include them in rpms (3100)
3863 - add new init script for (un)mounting lustre filesystems (2593)
3864 - run chkconfig in %post for init scripts (3701)
3865 - drop scimac NAL (unmaintained)
3867 2004-06-17 Cluster File Systems, Inc. <info@clusterfs.com>
3870 - clean kiobufs before and after use (3485)
3871 - strip trailing '/'s before comparing paths with /proc/mounts (3486)
3872 - remove assertions to work around "in-flight rpcs" recovery bug (3063)
3873 - change init script to fail more clearly if not run as root (1528)
3874 - allow clients to reconnect during replay (1742)
3875 - fix ns_lock/i_sem lock ordering deadlock for kms update (3477)
3876 - don't do DNS lookups on NIDs too small for IP addresses (3442)
3877 - re-awaken ptlrpcd if new requests arrive during check_set (3554)
3878 - fix cond_resched (3554)
3879 - only evict unfinished clients after recovery (3515)
3880 - allow bulk resend, prevent data loss (3570)
3881 - dynamic ptlrpc request buffer allocation (2102)
3882 - don't allow unlinking open directory if it isn't empty (2904)
3883 - set MDS/OST threads to umask 0 to not clobber client modes (3359)
3884 - remove extraneous obd dereference causing LASSERT failure (3334)
3885 - don't use get_cycles() when creating temp. files on the mds (3156)
3886 - hold i_sem when setting i_size in ll_extent_lock() (3564)
3887 - handle EEXIST for set-stripe, set proper directory name (3336)
3889 - servers can dump a log evicting a client - lustre.dump_on_timeout=1
3890 - fix ksocknal_fmb_callback() error messages (2918)
3892 2004-05-27 Cluster File Systems, Inc. <info@clusterfs.com>
3895 - don't copy lvb into (possibly NULL) reply on error (2983)
3896 - don't deref dentry after dput, don't free lvb on error (2922)
3897 - use the kms to determine writeback rpc length (2947)
3898 - increment oti_logcookies when osc is inactive (2948)
3899 - update client's i_blocks count via lvb messages (2543)
3900 - handle intent open/close of special files properly (1557)
3901 - mount MDS with errors=remount-ro, like obdfilter (2009)
3902 - initialize lock handle to avoid ASSERT on error cleanup (3057)
3903 - don't use cancelling-locks' kms values (2947)
3904 - use highest lock extent for kms, not last one (2925)
3905 - don't dereference ERR_PTR() dentry in error handling path (3107)
3906 - fix thread race in portals_debug_dumplog() (3122)
3907 - create lprocfs device entries at setup instead of at attach (1519)
3908 - common AST error handler, don't evict client on completion race (3145)
3909 - zero nameidata in detach_mnt in 2.6 (3118)
3910 - verify d_inode after revalidate_special is valid in 2.6 (3116)
3911 - use lustre_put_super() to handle zconf unmounts in 2.6 (3064)
3912 - initialize RPC timeout timer earlier for 2.6 (3219)
3913 - don't dereference NULL reply buffer if mdc_close was never sent (2410)
3914 - print nal/nid for unknown nid (3258)
3915 - additional checks for oscc recovery before doing precreate (3284)
3916 - fix ll_extent_lock() error return code for 64-bit systems (3043)
3917 - don't crash in mdc_close for bad permissions on open (3285)
3918 - zero i_rdev for non-device files (3147)
3919 - clear page->private before handing to FS, better assertion (3119)
3920 - tune the read pipeline (3236)
3921 - fix incorrect decref of invalidated dentry (2350)
3922 - provide read-ahead stats and refine rpc in flight stats (3328)
3923 - don't hold journal transaction open across create RPC (3313)
3924 - update atime on MDS at close time (3265)
3925 - close LDAP connection when recovering to avoid server load (3315)
3926 - update iopen-2.6 patch with fixes from 2399,2517,2904 (3301)
3927 - don't leak open file on MDS after open resend (3325)
3928 - serialize filter_precreate and filter_destroy_precreated (3329)
3929 - loop device shouldn't call sync_dev() for nul device (3092)
3930 - clear page cache after eviction (2766)
3931 - resynchronize MDS->OST in background (2824)
3932 - refuse to mount the same filesystem twice on same mountpoint (3394)
3933 - allow llmount to create routes for mounting behind routers (3320)
3934 - push lock cancellation to blocking thread for glimpse ASTs (3409)
3935 - don't call osc_set_data_with_check() for TEST_LOCK matches (3159)
3936 - fix rare problem with rename on htree directories (3417)
3938 - allow default OST striping configuration per directory (1414)
3939 - fix compilation for qswnal for 2.6 kernels (3125)
3940 - increase maximum number of MDS request buffers for large systems
3941 - change liblustreapi to be useful for external progs like lfsck (3098)
3942 - increase local configuration timeout for slow disks (3353)
3943 - allow configuring ldlm AST timeout - lustre.ldlm_timeout=<seconds>
3945 2004-03-22 Cluster File Systems, Inc. <info@clusterfs.com>
3948 - fixes for glimpse AST timeouts / incorrectly 0-sized files (2818)
3949 - don't overwrite extent policy data in reply if lock was blocked (2901)
3950 - drop filter export grants atomically with removal from device (2663)
3951 - del obd_self_export from work_list in class_disconnect_exports (2908)
3952 - don't LBUG if MDS recovery times out during orphan cleanup (2530)
3953 - swab reply message in mdc_close, other PPC fixes (2464)
3954 - fix destroying of named logs (2325)
3955 - overwrite old logs when running lconf --write_conf (2264)
3956 - bump LLOG_CHUNKSIZE to 8k to allow for larger clusters (2306)
3957 - fix race in target_handle_connect (2898)
3958 - mds_reint_create() should take same inode create lock (2926)
3959 - correct journal credits calculated for CANCEL_UNLINK_LOG (2931)
3960 - don't close files for self_export to avoid uninitialized obd (2936)
3961 - allow MDS with the same name as client node (2939)
3962 - hold dentry reference for closed log files for unlink (2325)
3963 - reserve space for all logs during transactions (2059)
3964 - don't evict page beyond end of stripe extent (2925)
3965 - don't oops on a deleted current working directory (2399)
3966 - handle hard links to targets without a parent properly (2517)
3967 - don't dereference NULL lock when racing during eviction (2867)
3968 - don't grow lock extents when lots of conflicting locks (2919)
3970 2004-03-04 Cluster File Systems, Inc. <info@clusterfs.com>
3973 - account for cache space usage on clients to avoid data loss (974)
3974 - lfsck support in lustre kernel code (2349)
3975 - reduce journal credits needed for BRW writes (2370)
3976 - orphan handling to avoid losing space on client/server crashes
3977 - ptlrpcd can be blocked, stopping ALL progress (2477)
3978 - use lock value blocks to assist in proper KMS, faster stat (1021)
3979 - takes i_sem instead of DLM locks internally on obdfilter (2720)
3980 - recovery for initial connections (2355)
3981 - fixes for mds_cleanup_orphans (1934)
3982 - abort_recovery crashes MDS in b_eq (mds_unlink_orphan) (2584)
3983 - block all file creations until orphan recovery completes (1901)
3984 - client remove rq_connection from request struct (2423)
3985 - conf-sanity test_5, proper cleanup in umount log not availale (2640)
3986 - recovery timer race (2670)
3987 - mdc_close recovey bug (2532)
3988 - ptlrpc cleanup bug (2710)
3989 - mds timeout on local locks (2588)
3990 - namespace lock held during RPCs (2431)
3991 - handle interrupted sync write properly (2503)
3992 - don't try to handle a message that hasn't been replied to (2699)
3993 - client assert failure during cleanup after abort recovery (2701)
3994 - leak mdc device after failed mount (2712)
3995 - ptlrpc_check_set allows timedout requests to complete (2714)
3996 - wait for inflight reqs when ptlrpcd finishes (2710)
3997 - make sure unregistered services are removed from the srv_list
3998 - reset bulk XID's when resending them (caught by 1138 test)
3999 - unregister_bulk after timeout
4000 - fix lconf error (2694)
4001 - handle write after unfinished setstripe, stripe-only getstripe (2388)
4002 - readahead locks pages, leaves pending causing memory pressure (2673)
4003 - increase OST request buffers to 4096 on large machines (2729)
4004 - fix up permission of existing directories in simple_mkdir (2661)
4005 - init deleted item, add assertions ptlrpc_abort_inflight() (2725)
4006 - don't assign transno to errored transactions (2742)
4007 - don't delete objects on OST if given a bogus objid from MDS (2751)
4008 - handle large client PAGE_SIZE readdir on small PAGE_SIZE MDS (2777)
4009 - if rq_no_resend, then timeout request after recovery (2432)
4010 - fix MDS llog_logid record size, 64-bit array alignment (2733)
4011 - don't call usermode_helper from ptlrpcd, DEFAULT upcall (2773)
4012 - put magic in mount.lustre data, check for bad/NULL mount data (2529)
4013 - MDS recovery shouldn't delete objects that it has given out (2730)
4014 - if enqueue arrives after completion, don't clobber LVB (2819)
4015 - don't unlock pages twice when trigger_group_io returns error (2814)
4016 - don't deref NULL rq_repmsg if ldlm_handle_enqueue failed (2822)
4017 - don't write pages to disk if there was an error (1450)
4018 - don't ping imports that have recovery disabled (2676)
4019 - take buffered bytes into account when balancing socknal conn (2817)
4020 - hold a DLM lock over readdir always, use truncate_inode_pages (2706)
4021 - reconnect unlink llog connection after MDS reconnects to OST (2816)
4022 - remove little-endian swabbing of llog records (1987)
4023 - set/limit i_blksize to LL_MAX_BLKSIZE on client (2884)
4024 - retry reposting request buffers if they fail (1191)
4025 - grow extent at grant time to avoid granting a revoked lock (2809)
4026 - lock revoke doesn't evict page if covered by a second lock (2765)
4027 - disable VM readahead to avoid reading outside lock extents (2805)
4029 - return LL_SUPER_MAGIC from statfs for the filesystem type (1972)
4030 - updated kernel patches for hp-2.4.20 kernel (2681)
4032 2004-02-07 Cluster File Systems, Inc. <info@clusterfs.com>
4035 - fix truncated write corruption (2366)
4036 - fix for failed assertion in iopen_connect_dentry (1792,2517)
4038 - don't flag the ptlrpcd thread with PF_MEMALLOC (2636)
4039 - ensure len(uuid) < 37 in lmc (1171)
4040 - fix ia64 OOPS in llog_test (2255)
4041 - zero end of page at obdfilter for partial page writes (2648)
4042 - don't leave stale dentries around after renames (bug 2428)
4043 - fix timeouts when evicting a client with a single lock held (2642)
4044 - set deadline for the initial HELLO message to drain (2634)
4045 - print out dotted-quad IP addresses in the socknal (2302)
4047 - additional debugging for MDS client eviction problem (2443)
4048 - fix mkfsoptions support for osts (2603, 2604)
4050 2004-01-27 Cluster File Systems, Inc. <info@clusterfs.com>
4053 - add series for the vanilla 2.6.0 kernel
4054 - add series for the vanilla 2.4.24 kernel
4055 - add series for a cray x86/64 UL kernel drop
4056 - fix xattr patches for the vanilla 2.4.19 series
4058 - generate true UUIDs in lmc (1171)
4059 - have portals stack dumping break in UML (2466)
4060 - avoid bad dchild deref; avoid inum lock w/o creation (2362)
4061 - allocate with _NOFS in ldlm to avoid deadlock (1933)
4062 - wake callback waiting threads on client eviction (2460)
4063 - Add --ptldebug and --subsystem to lmc (1719)
4064 - update assertion to allow safe interrupt allocation
4065 - set rq_no_resend for cancel requests (2432)
4066 - recalculate ptlrpcd timeout after resend (2494)
4067 - call vfs_rmdir when removing pending directories (2368)
4068 - fix renaming a file to itself (2429)
4069 - lmc creates a default one-stripe lov (2454)
4070 - expand procfs space to handle large clusters (2326)
4071 - increase UML stack to avoid overflow
4072 - update lconf's list of debug and subsystem masks
4073 - fix lfs find --obd (2510)
4074 - /proc tunable for disabling filter read caching (2591)
4075 - stop rpm packages from altering slapd.conf (2301)
4076 - disable nagle in the socknal under 0conf (2578)
4077 - choose mds inode size based on stripe count (2572)
4078 - fix kernel-source rpm problems (2516)
4080 - add --disable-doc to avoid pdf generation (2421)
4081 - update documentation, tests, type-os, comments
4082 - avoid format warnings on ia64
4083 - remove the TOE NAL
4084 - tiny code cleanups by removing unused fields
4086 2004-01-07 Cluster File Systems, Inc. <info@clusterfs.com>
4089 - fix obvious semaphore misuse in as-yet-unused setattr path (2348)
4090 - remove the most blatant lies from BUILDING file (2371)
4091 - change default debug level to reasonable production setting
4092 - reduce client side cache size to reduce cache flush time
4093 - reduce max RPCs in flight to avoid unnecessary file fragmentation
4094 - make TCP zerocopy and pinger support enabled by default (2476)
4095 - sync writes completed after process exits caused crashes (2319)
4096 - maintain correct mount count on the MDS (2356)
4097 - backout 1557, because 2316 wasn't really fixed
4098 - better file I/O statistics gathering in /proc
4099 - don't take unnecessary, deadlock-inducing bug in readpage (2383)
4100 - another kernel patch to fix zero-copy TCP function export
4101 - don't take duplicate lock when processing re-sent getattr (2420)
4102 - lctl uses obd_self_export instead of creating new conn (2353)
4103 - MDS/OST recovery case which requires object creation asserted (2425)
4104 - move lfs from /usr/sbin to /usr/bin in packages
4105 - fix race between mds_client_add and mds_client_free (2417)
4106 - use kmalloc instead of slabs in portals (2430)
4107 - don't create duplicate records when a failover MDS is present (2442)
4108 - remove unnecessary mount age check (2332)
4109 - don't remove directory inodes from locks prematurely (2451)
4110 - don't break if MDS service name is the same as hostname (2103)
4111 - fix races in client write RPC generation when cache full (2482)
4113 2003-12-13 Cluster File Systems, Inc. <info@clusterfs.com>
4116 - remove now-unused request->rq_obd (278)
4117 - if an allocation fails, print out how much memory we've used (1933)
4118 - use PORTAL_SLAB_ALLOC for structures, to get GFP_MEMALLOC (1933)
4119 - add the "configurable stack size" patch to most series files (1256)
4120 - ability to write large log records, for 100+ OST configs (2306)
4121 - fix NULL deref when filter_prep fails (2314)
4122 - fix operator precedence error in filter_sync
4123 - dynamic allocation of socknal TX descriptors (2315)
4124 - fix a missed case in the GFP_MEMALLOC patch, can cause deadlock (2310)
4125 - fix gcc 2.96 compilation problem in xattr kernel patch (2294)
4126 - ensure that CWARN messages in Portals always get to the syslog
4127 - __init/__exit are not for prototype decls (ldlm_init/exit)
4128 - x86-64 compile warning fixes
4129 - fix gateway LMC keyword conflict (2318)
4130 - fix MDS lock inversions in getattr/reint paths (1844)
4131 - fix a rare lock re-ordering bug, which caused deadlock (2322)
4132 - fix i_sem/journal inversion in fsfilt_ext3_write_record (2306)
4133 - DLM race condition prevented some lock evictions (2328)
4134 - ENOMEM detection and retry on socknal sends (2230)
4135 - use GFP_NOFS throughout Lustre, to combat ENOMEM (2230)
4136 - move osc_rpcd into ptlrpc, for use in MDC and others (2329)
4137 - protect MDS inode fsdata with stronger locking; fixes assertion (2313)
4138 - better error messages when a client is rejected during recovery (1505)
4139 - avoid cancelling locks which were never granted, after failure (2330)
4140 - fix i_sem/journal inversion in mds_client_add (2333)
4141 - fix truncate/getattr lock cycle deadlock (2334)
4142 - use rpcd to send close; allows resend after timeout, avoid leak (1897)
4143 - fix two rare exit paths which could leak an l_lock() ref (2321)
4144 - fencepost error in MDS/OST orphan recovery (2226)
4145 - make log record alignment 8 bytes (1988)
4146 - lstripe now fails when requested offset > ost_count (2237)
4147 - ensure that all kernel series have a complete list.h (1607)
4148 - fix crashes in special-file operations (2316)
4149 - lctl create/brw OID mismatch, caused by obsolete filter loop (2339)
4151 - allow configurable automake binary, for testing new versions
4152 - small update to the lfs documentation
4154 2003-12-03 Cluster File Systems, Inc. <info@clusterfs.com>
4156 * fix negative export reference count in fsfilt_sync (2312)
4158 2003-12-01 Cluster File Systems, Inc. <info@clusterfs.com>
4159 * release candidate 0.9.1
4161 - orphans are moved into the PENDING directory for possible recovery
4162 - replayed opens now open by fid for orphan/rename safety (1042)
4163 - last close of an orphan inode generates a transno (683)
4164 - chdir() and mount() now pin the directory entry (1020)
4165 - avoid CERROR in normal ll_setattr_raw() error case (1500)
4166 - discard very old requests without processing them (1502)
4167 - remove some common, well-understood CERRORs (1505)
4168 - require O_DIRECT I/O to be page-sized to workaround IA64 crash (1609)
4169 - clear "grant" flags in OST replies until OST grant code lands (1644)
4170 - fix read performance by not clobbering i_blksize on client (1598)
4171 - fix __ldlm_handle2lock oops by not dereferencing lock after PUT (1625)
4172 - make LRU size a /proc tunable, clears locks when reduced (707)
4173 - fix some lprocfs rot that prevented ptlbd from loading (1732)
4174 - server locks take references on exports now (1558)
4175 - build fixes for 2.4.20-rh trees (1663)
4176 - return an error from lov_create if all OSCs are inactive (1751)
4177 - fix import levels when a reconnect happens without a timeout (1597)
4178 - exit early from mds_open if we get a lookup error (1749)
4179 - partial page read at EOF wouldn't wait for disk before sending (1642)
4180 - avoid NULL deref in obdfilter when reading page past EOF (1592)
4181 - avoid LASSERT in ll_intent_lock if server failed very early (1090)
4182 - fix LBUG in ll_it_open_error with rc = -2 (1861)
4183 - write/truncate lock inversion (1639)
4184 - Don't auto-load obdclass, portals modules during cleanup (1495)
4185 - fix timestamps from jumping to "now" (1763)
4186 - extra journal assertions (1648)
4187 - add an extra multiunlink test (1771)
4188 - fix read_record/write_record API (1776)
4189 - fix leak of offset_extent, possible incorrect i_size later (1772)
4190 - fix lasserts in mis-matched transnos during open-unlink testing (1541)
4191 - Debugging for the kqswnal_get_idle_tx problems (1820)
4192 - Allow recovery to be attempted multiple times (1536)
4193 - Write out MDS last_rcvd file after it is first created (1600)
4194 - Fix tx_descriptor leak in failed transmit situations (1827)
4195 - ext3 journaling fixes for assertion failure after IO error (1871)
4196 - class_export_put() on freed export after completion AST error (1896)
4197 - Fix revalidate looping in VFS (1322)
4198 - Don't access a freed export during MDS_REINT timeout (1521)
4199 - Add open-unlink recovery support on the MDS (1673,1764)
4200 - Return an error if no MDS data was read from last_rcvd (1946)
4201 - Fix for lookup "." or ".." crash on error (1932,1931,1935)
4202 - Don't setup a disk device that doesn't match exported UUID (317)
4203 - Reduce bulk RPC timeout to avoid cascading client/OST failures (1845)
4204 - avoid committing NULL handle in force close
4205 - local.sh is now a one-stripe LOV configuration
4206 - POSIX utime.4 -EPERM on FIFO not owned by user (56)
4207 - fix ext3 htree duplicate directory entry corruption (1516)
4208 - POSIX creat.13, fstat.1, open.18, stat.3 new file atime/mtime (2020)
4209 - update to new LOV EA format (2097)
4210 - interoperability for different PAGE_SIZE/wordsize (686,1821,1343,2042)
4212 2003-06-15 Phil Schwan <phil@clusterfs.com>
4215 - imports and exports cleanup too early, need refcounts (349, 879, 1045)
4216 - per-import/export recovery handling (958, 931, 959)
4217 - multiple last-rcvd slots, for serving multiple FSes (949)
4218 - connections are again shared between multiple imp/exports (963, 964)
4219 - "umount -f" would hang if any requests needed to be sent (393, 978)
4220 - avoid pinning large req buffer by copying for queued messages (989)
4221 - add "uuid" to "lctl device" command to help upcalls (991)
4222 - "open" RPCs with transnos would confuse recovery counters (1037)
4223 - do proper endian conversion of all wire messages (288, 340, 891)
4224 - remove OST bulk get LBUGs, fix ost_brw_write cleanup (1126)
4225 - call waiting locks callback from LDLM recovery thread (1127, 1151)
4226 - fix ptlrpc_connection leak in target_handle_connect (1174)
4227 - fix import refcounting bug in OST and MDS cleanup (1134)
4228 - if an invalid-at-open-time OSC returned before close(), LBUG (1150)
4229 - fix very unlikely obd_types race condition (501)
4230 - remove osc_open hack for echo_client (1187)
4231 - we leaked exports/dlmimps for forcibly disconnected clients (1143)
4232 - a failure in read_inode2 leads to deadlock (1139)
4233 - cancel ack-locks as soon as transaction is committed (1072)
4234 - fix major leaks and crashes in the bulk I/O path (937, 1057)
4235 - make sure to commitrw after any preprw to avoid deadlock (1162)
4236 - failing to execute a file in a lustre FS would lock inode (1203)
4237 - small DEBUG_REQ fix to avoid dereferencing a NULL (1227)
4238 - don't ASSERT while cleaning up an incompletely-setup obd (1248)
4239 - obd_uuid2tgt would walk off the end of the list (1255)
4240 - on IA64 the osc would give portals incorrect bulk size (1258)
4241 - fix debug daemon ioctl interface; allows daemon on ia64 (1274)
4242 - fix lock inversion caused by new llite matching code (1282)
4243 - limit the number of dirty pages on a client to 10MB (1286)
4244 - timed out locks were not being corrected cancelled (1289)
4245 - fix O_DIRECT above 4GB on IA-32 (1292)
4246 * major user-visible changes
4247 - fail out/fail over policy now controlled by the upcall (993)
4249 - add OBD_PING to check server availability and failure (954)
4250 - lustre messages are now sent in sending host order (288, 340, 891)
4251 - add eadatalen to MDS getattr reply (340)
4252 - OST read replies may contain second buffer, with per-page status (593)
4254 2003-03-11 Phil Schwan <phil@clusterfs.com>
4257 - LDLM_DEBUG macro fix, for gcc 3.2 (850)
4258 - failed open()s could cause deadlock; fixed (867, 869)
4259 - stop cancelling OST locks when files are closed (481)
4260 - overlapping XID spaces caused network corruption (851, 853)
4261 - fix unsafe fsfilt counter arithmetic; change to atomic_t
4262 - setattr_raw added, to do single-RPC, server-side setattrs
4263 - lmc/lconf syntax change for OST UUIDs
4264 - fix crashy race condition between ptlrpc_free_req and osc_close
4265 - don't use request in mdc_enqueue if we hit a timeout (889)
4266 - don't set the inode i_size for regular files from the MDS (896)
4267 - handle out of order completion AST (842)
4268 - don't LBUG if a lock request times out after receiving AST (913)
4269 - avoid d_rehash race in ll_find_alias by rehashing inside dcache_lock
4270 - if a bad lock AST arrives, send an error instead of dropping entirely
4271 - return 0 from revalidate2 if ll_intent_lock returns -EINTR (912)
4272 - fix leak in bulk IO when only partially completed (899, 900, 926)
4273 - fix O_DIRECT for ia64 (55)
4274 - (almost) eliminate Lustre-kernel-thread effects on load average (722)
4275 - C-z after timeout could hang a process forever; fixed (977)
4277 - client-side I/O cache (678, 924, 929, 941, 970)
4279 - READPAGE and SETATTRs which don't take server-side locks get
4282 2003-02-11 Phil Schwan <phil@clusterfs.com>
4285 - Fix ldlm_lock_match on the MDS to avoid matching remote locks (592)
4286 - Fix fsfilt_extN_readpage() to read a full page of directory
4287 entries, or fake the remainder if PAGE_SIZE != blocksize (500)
4288 - Avoid extra mdc_getattr() in ll_intent_lock when possible (534, 604)
4289 - Fix imbalanced LOV object allocation and out-of-bound access (469)
4290 - Most intent operations were removed, in favour of a new RPC mode
4291 that does a single RPC to the server and bypasses most of the VFS
4292 - All LDLM resource ID arrays were removed in favour of ldlm_res_id
4293 - Aggressively cancel local locks on DLM servers
4294 - mds_reint_unlink sends EA to the client if it's the last nlink.
4295 client uses that EA to unlink OST objects.
4296 - mds_reint_{rename,unlink,link} were rewritten to take ordered locks
4297 - recursive symlinks were fixed (439)
4298 - fixed NULL deref in DEBUG_REQ
4299 - filter_update_lastobjid no longer calls sync, which annoyed extN
4300 - fixed multi-client small-writes to a single file problem (445)
4301 - fixed mtime updates during file writes (607)
4302 - fixed vector writes on obdfilter causing problems when ENOSPC (670)
4303 - fixed bug in obd_brw_read/write() (under guise of testing 367)
4304 - fixed Linux OST size reporting problem (444, 656)
4305 - OST now updates object mtime with writes or setattr (607, 619)
4306 - client verifies file size before zeroing page past EOF (445)
4307 - OST now writes last allocated objid to disk with allocation (108)
4308 - LOV on echo now works (409)
4310 - mds_reint_unlink sends a new buffer, with the EA included. this
4311 buffer is only valid if body->valid & OBD_MD_FLEASIZE, which is only
4312 set if a regular file was being unlinked, and it was the last link
4313 - use PtlGet from the target for bulk writes (315)
4314 - OST now updates object mtime with writes or setattr (607, 619)
4315 - LDLM now has a grant-time callback to revalidate locked items, if
4317 - Many MDS operations were reorganized to combat race conditions
4319 - Merge b_intel branch (updated lprocfs code) - now at /proc/fs/lustre
4320 - configure check to avoid gcc version 2.96 20000731-2.96-98) (606)
4322 2003-01-06 Andreas Dilger <adilger@clusterfs.com>
4325 - Fully reactivate OST imports after reconnection (512, others)
4326 - Make sure client sees our -ENOTCONN from mds_handle (513 - partial)
4327 - More graceful error handling for truncating on dead OST (515)
4328 - Don't error out unless we're actually accessing dead stripes (474)
4329 - Fix garbage sizes when stripes are missing (410)
4330 - LRU counters were broken, causing constant lock purge (433, 432)
4331 - garbage on read from stripes with failed OSTs (441)
4332 - mark OSCs as active before reconnecting during recovery (438)
4333 - lov_enqueue and lov_cancel need to handle inactive OSTs (403)
4334 - lfind did not preserve OST order in output (443)
4335 - symlinks cause hung clients, incorrect data (439)
4336 - stop dereferencing request after dropping refcount (457)
4337 - don't LASSERT(spin_is_locked) on non-SMP (455)
4338 - fixes for many rename() bugs
4339 - fstat didn't correctly synchronize attributes (399)
4340 - server must handle lock cancellation during blocking AST prep (487)
4341 - bulk descriptors were free()d too soon (511)
4342 - fix paths in lconf, which would load incorrect modules (451, 507)
4343 - fix confusing lconf 'host not found' error message (386)
4344 - fix lock order deadlock on OST (O/R i_sem before journal ops, 478)
4345 - fix race condition in mdc_blocking_ast() for inode access (526)
4346 - fix lov_unpackmd() unpacking wrong number of stripes (537)
4347 - fix lov_set_osc_active() marking wrong OSC inactive (440)
4348 - fix bad lstripe lov_unpackmd() assertion (fix layering too) (527)
4349 - fix multiple writes of stripe MD to MDS (358, maybe 519)
4350 - fix lstripe in several ways (kernel side) (527)
4351 - fix request leak in ldlm_cli_enqueue (262)
4352 - incorrect OSC was marked inactive after OST failure
4353 - call mds_fs_cleanup before unmounting filesystem (524)
4354 - fix races between taking ns_lock and ldlm_lock_change_resource
4355 - fix races updating LOV export open file list
4356 - fix lov_enqueue error path, avoid decref-ing bad lock handle
4357 - fix recovery NULL deref in ldlm_cli_cancel_unused
4358 - fix some DLM races by using new hash table for lock handles (419)
4359 - permit the client to specify desired inodes, at replay
4360 - duplicate requests when we queue them for replay reintegration
4361 - fix last_rcvd offset calculation
4362 - sync after each recovered transaction, so we always make progress
4363 - never, not always, ERESTART requests without transnos
4364 - store the lov_desc in the MDS, so we don't depend on getlovinfo to
4366 - skip replay if the MDS says that the client is already connected
4367 - don't check for a recovery-enabled export to match lctl's UUID
4368 - don't INC_USE_COUNT for phantom exports
4369 - don't crash when cleaning up phantom exports (567)
4370 - don't double-finish or set replay data for errored mdc_open requests
4371 - abort requests when they time out, so we don't get old replies
4372 - send/receive replies for AST messages again
4373 - if the client says that it doesn't have the lock, cancel it on the
4375 - if we timeout during I/O, don't try to cancel an in-use lock; instead
4376 mark it as destroyed, it will all work out when decref is called
4377 - fix module use counts (22, 581)
4379 - ASTs now expect a reply (server cancels lock on error reply)
4381 2002-12-02 Andreas Dilger <adilger@clusterfs.com>
4384 - fix many simultaneous client startup (392)
4385 - fix dentry->d_it clobbering
4386 - credentials weren't being shipped for readdir/getattr operations
4387 - remove invalid assertions triggered during some concurrent MD
4389 - proper Lustre versions added (336, 389)
4390 - fix memory leak for create error case (398)
4391 - fix LOV locking bug that would get cli/srv out of sync
4392 - fix echo client over LOV (409)
4393 - fix dbench 2, extN refcount problem (170, 258, 356, 418)
4394 - fix double-O_EXCL intent crash (424)
4395 - avoid sending multiple lock CANCELs (352)
4397 - MDS can do multi-client recovery (modulo bugs in new code)
4399 - many updates, edits, cleanups
4401 2002-11-18 Phil Schwan <phil@clusterfs.com>
4404 - fix null d_it dereference (346)
4405 - fix full OST/dbench hang (333)
4406 - fix permission problem with file removal (286)
4407 - fix removal of OSCs from LOV when they fail
4408 - fix NULL deref during bulk timeout (214)
4409 - fix problems related to multiple filesystems on one MDS (241)
4410 - fixed serious subtle metadata locking bugs
4411 - free locks on clients when inodes are removed due to memory
4413 - fix inode pointer in lock data (285)
4414 - partial support for multiple MDS on a single host (241)
4415 - data locks weren't cancelled at clear_inode time (290, 311)
4416 - intent locks could lead to unbounded lock growth (205)
4417 - added a maximum lock count, an LRU list, and a flusher
4418 - fix multiple rename (365)
4419 - properly abstracted the echo client
4420 - OSC locked 1 byte too many; fixed
4421 - rewrote brw callback code:
4422 - fixed recovery bugs related to LOVs (306)
4423 - fixed too-many-pages-in-one-write crash (191)
4424 - fixed (again) crash in sync_io_timeout (214)
4425 - probably fixed callback-related race (385)
4427 - Add capability to MDS protocol
4428 - LDLM cancellations and callbacks on different portals
4430 2002-10-28 Andreas Dilger <adilger@clusterfs.com>
4433 - limit client IOV size to PTL_MD_MAX_IOV (611336, 191)
4434 - defer open object destruction to close time (601981, 138)
4435 - open/close OST file handle in obdo (OBD_MD_FLHANDLE) (601981, 138)
4436 - move LDLM_ENQUEUE/CONVERT back to MDS portal (625069)
4437 - abstract ll_lookup2, fix ll_revalidate2 to use abstraction (256)
4438 - don't call obd_setattr in ll_file_release for destroyed objects
4439 * protocol change to lustre_msg: move |version| and add |flags|
4440 * protocol change to osc_punch: "start" in "o_size", "end" in "o_blocks"
4441 * lock replay: for LDLM_FL_REPLAY trust client to do right thing
4442 * added replay of create, unlink, link and rename operations during
4443 MDS failover; recovery should be much more robust now
4444 * remove failed OSCs from LOVs (only lov_create uses this so far)
4445 * the lustre-HOWTO was brought (more) up to date (582544)
4447 2002-10-23 Phil Schwan <phil@clusterfs.com>
4450 - in-use dentries weren't being reused properly (617851)
4451 - prevent multiple LDLM setup (599178)
4452 - fix LOV size calculations for truncate (617853)
4453 - fix client handling of MDS intent errors (POSIX)
4454 - fix permission bug in lovstripe.c test (624321)
4455 - fix MDS thread deadlock - move LDLM handler to DLM portal (625069)
4456 - truncate past end of file could corrupt data
4457 - proper cleanup after timeouts, crashes, etc (592524, 550815)
4458 - a race in recovery could return ETIMEDOUT to apps (623947)
4459 - building outside the source directory was fixed
4460 * the lustre-HOWTO was brought (more) up to date (582544)
4461 * major progress was made on recovery functionality
4463 2002-10-10 Phil Schwan <phil@clusterfs.com>
4466 - recovery deadlock fix
4467 - rm -rf causes LBUG fix (617817)
4468 - file open by multiple tasks fix (618962)
4469 - directory permissions bugs (602707 and 620007)
4470 - journal_stop fixed with locking (611313)
4471 - O_APPEND failures resolved (618273, perhaps 614459)
4472 - lconf PATH fix (619770)
4473 - IA64 build fix (621450)
4474 - RPC buffer sizes scale with amount of memory
4476 2002-10-01 Phil Schwan <phil@clusterfs.com>
4479 - locks would be cancelled without throwing away data pages,
4480 resulting in inconsistent data (605627)
4481 - inode attributes were not always being refreshed (605627, 612449)
4482 - lconf now continues to cleanup after lctl reports an error
4483 - MDS now enforces user permissions (602707)
4484 - lprocfs cleanup fixed, but not yet enabled (614157)
4485 - fixed infinite server hang, should a client not respond to an AST
4486 - avoid going into recovery if user calls readlink() with a buffer
4487 that's too small (613941)
4488 - AST RPCs no longer require replies (614867) -- this may be changed
4489 - don't crash server if client sends an IOV that's too big (611336)
4490 - fixed lock conversion deadlock (611892)
4491 - fixed the following of symlinks (614622)
4492 * recovery: the server can remove locks from a client that dies, other
4493 clients can make progress
4494 * more extN patch fixes
4495 * compile-time configurable ptlrpc buffer allocations
4497 - collaborative read cache document
4498 - Lustre Lite Performance CDR document-in-progress
4500 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
4503 - fix typo in patch-2.4.18
4505 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
4508 - clear ptlrpc request each time in handle_incoming_request()
4509 - unlink of files now destroys the object on the OST
4511 2002-09-19 Peter Braam <braam@clusterfs.com>
4513 * add hard link support
4514 * change obdfile creation method
4515 * kernel patch changed
4517 2002-09-19 Peter Braam <braam@clusterfs.com>
4520 - stack overflow bug in extN fixed
4522 2002-09-18 Andreas Dilger <adilger@clusterfs.com>
4524 * documentation updates
4525 - add man pages for config tools
4526 - update tests/README to describe testing with new config tools
4527 - finish metadata API descriptions
4528 * bug fixes and cleanups
4529 - statfs workaround for 16TB limit
4530 - LOV stripe allocation improved, can stripe on subset of OSTs
4531 - LOV file size/IO offset was wrong for files > 4GB in size
4532 - object EA data was being dropped, caused files to be unreadable
4533 - memory overflow with non-LOV OST caused memory corruption
4534 - fixed regression tests to work with new config tools, obdfilter
4535 - fixed bug when directory size became larger than 1 block
4536 - fixed bug (for single client case) when PWD was deleted
4537 - invalidate local directory pages when doing intent-based ops
4538 - avoid LDLM oops when lock callback contained bad data
4540 2002-09-09 Andreas Dilger <adilger@clusterfs.com>
4542 * documentation updates
4543 * bug fixes and cleanups
4544 - configuration tools
4547 - 64-bit compile warnings
4548 - 64-bit internal statfs data
4550 * test_brw on persistent OST devices
4552 * lprocfs (disabled)
4554 2002-09-04 Andreas Dilger <adilger@clusterfs.com>
4556 * documentation updates
4557 * bug fixes and cleanups
4558 * configuration tools
4560 2002-08-30 Peter J. Braam <braam@clusterfs.com>
4563 * many small fixes to 0_5_4
4564 * io/network handling
4565 * thinkos in MDS operations
4567 2002-08-24 Peter J. Braam <braam@clusterfs.com>
4570 * crucial basic fixes to 0.5.3
4571 * IOR, Iozone work over Elan
4574 2002-08-07 Phil Schwan <phil@clusterfs.com>
4575 * version 0_5_3, our first alpha
4576 * we use the new Portals iovs
4577 * documentation updates
4578 * bug fixes and cleanups
4579 * small changes in the DLM wire protocol
4581 2002-07-25 Peter J. Braam <braam@clusterfs.com>
4582 * version 0_5_1 with some initial stability,
4583 * locking on MD and file I/O.
4584 * documentation updates
4585 * several bug fixes since 0.5.0
4586 * small changes in wire protocol
4588 2002-07-18 Phil Schwan <phil@clusterfs.com>
4590 * delivered as Lustre Light Alpha
4591 * fixed a crash after handling invalid MDS requests
4592 * fixed directory pages for architectures with non-4k pages sizes
4594 2002-07-11 Andreas Dilger <adilger@clusterfs.com>
4595 * release version v0_4_4
4596 * Moves TCP acceptor to be on port 2432 (unused Coda port) instead
4598 * Fixes a number of interruption problems with OST operations.
4599 * Update documentation for portals header changes
4600 * Move all wire protocol structs/defines to lustre_idl.h
4601 * Fixes symlink length bug.
4602 * Add tcpdump to repository.
4604 2002-07-05 Andreas Dilger <adilger@clusterfs.com>
4605 * release version v0_4_3
4606 * Fixes statfs for inodes on extN.
4607 * Fixes bug in runtests which would delete /etc/hosts.
4608 * Use 64-bit object IDs wherever possible (not into VFS though)
4609 Remove ost_get_info, which is unused by lustre, and out of date.
4611 2002-07-03 Peter Braam <braam@clusterfs.com>
4612 * release version v0_4_2 Fixes a lookup error (type not passed)
4613 * move forward to head of Portals
4614 * move forward to latest Lustre kernel
4616 2002-06-25 Peter Braam <braam@clusterfs.com>
4617 * release version v0_4_1. Hopefully stable on single node use.