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: 1.40.4-cfs1
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.
837 --------------------------------------------------------------------------------
839 2007-08-10 Cluster File Systems, Inc. <info@clusterfs.com>
841 * Support for kernels:
842 2.6.5-7.283 (SLES 9),
843 2.6.9-55.EL (RHEL 4),
844 2.6.16.46-0.14 (SLES 10),
845 2.6.18.8 vanilla (kernel.org)
846 * Client support for unpatched kernels:
847 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
848 2.6.16 - 2.6.22 vanilla (kernel.org)
849 * Due to recently discovered recovery problems, we do not recommend
850 using patchless RHEL 4 clients with this or any earlier release.
851 * Recommended e2fsprogs version: 1.39.cfs8
852 * Note that reiserfs quotas are disabled on SLES 10 in this kernel.
853 * Starting with this release, the ldiskfs backing filesystem required
854 by Lustre is now in its own package, lustre-ldiskfs. This package
855 should be installed. It is versioned separately from Lustre and
856 may be released separately in future.
860 Description: block reactivating mgc import until all deactivates complete
861 Details : Fix race when failing back MDT/MGS to itself (testing)
864 Frequency : at statup only
866 Description: mds_lov_synchronize race leads to various problems
867 Details : simultaneous MDT->OST connections at startup can cause the
868 sync to abort, leaving the OSC in a bad state.
870 Severity : enhancement
872 Description: add optional extra BUILD_VERSION info
873 Details : add a new environment variable (namely LUSTRE_VERS) which allows
874 to override the lustre version.
877 Frequency : 2.6.18 servers only
879 Description: ll_kern_mount() doesn't release the module reference
880 Details : The ldiskfs module reference count never drops down to 0
881 because ll_kern_mount() doesn't release the module reference.
886 Description: server LBUG when using old ost_num_threads parameter
887 Details : Accept the old ost_num_threads parameter but warn that it
888 is deprecated, and fix an off-by-one error that caused an LBUG.
893 Description: Transient SCSI error results in persistent IO issue
894 Details : iobuf->dr_error is not reinitialized to 0 between two
898 Frequency : sometimes when underlying device returns I/O errors
900 Description: OSTs not going read-only during write failures
901 Details : OSTs are not remounted read-only when the journal commit threads
902 get I/O errors because fsfilt_ext3 calls journal_start/stop()
903 instead of the ext3 wrappers.
907 Description: poor connect scaling with increasing client count
908 Details : Don't run filter_grant_sanity_check for more than 100 exports
909 to improve scaling for large numbers of clients.
912 Frequency : SLES10 only
914 Description: sanity-quota.sh quotacheck failed: rc = -22
915 Details : Quotas cannot be enabled on SLES10.
918 Frequency : liblustre clients only
920 Description: getdirentries does not give error when run on compute nodes
921 Details : getdirentries does not fail when the size specified as an argument
922 is too small to contain at least one entry
924 Severity : enhancement
926 Description: Add LNET router traceability for debug purposes
927 Details : If a checksum failure occurs with a router as part of the
928 IO path, the NID of the last router that forwarded the bulk data
929 is printed so it can be identified.
934 Description: OST "spontaneously" evicts client; client has imp_pingable == 0
935 Details : Due to a race condition, liblustre clients were occasionally
938 Severity : enhancement
940 Description: lfs setstripe use optional parameters instead of postional
943 Severity : enhancement
945 Description: Nanosecond timestamp support for ldiskfs
946 Details : The on-disk ldiskfs filesystem has added support for nanosecond
947 resolution timestamps. There is not yet support for this at
948 the Lustre filesystem level.
951 Frequency : during server recovery
953 Description: MDS failing to send precreate requests due to OSCC_FLAG_RECOVERING
954 Details : request with rq_no_resend flag not awake l_wait_event if they get a
958 Frequency : nfs export on patchless client
960 Description: connectathon hang when test nfs export over patchless client
961 Details : Disconnected dentry cannot be found with lookup, so we do not need
962 to unhash it or make it invalid
965 Description: fix llapi_lov_get_uuids() to allow many OSTs to be returned
966 Details: : Change llapi_lov_get_uuids() to read the UUIDs from /proc instead
967 of using an ioctl. This allows lfsck for > 160 OSTs to succeed.
972 Description: open req refcounting wrong on reconnect
973 Details : If reconnect happened between getting open reply from server and
974 call to mdc_set_replay_data in ll_file_open, we will schedule
975 replay for unreferenced request that we are about to free.
976 Subsequent close will crash in variety of ways.
977 Check that request is still eligible for replay in
978 mdc_set_replay_data().
983 Description: disable writes to filesystem when reading health_check file
984 Details : the default for reading the health_check proc file has changed
985 to NOT do a journal transaction and write to disk, because this
986 can cause reads of the /proc file to hang and block HA state
987 checking on a healthy but otherwise heavily loaded system. It
988 is possible to return to the previous behaviour during configure
989 with --enable-health-write.
991 Severity : enhancement
993 Description: 64-bit inode version
994 Details: : Add a on-disk 64-bit inode version for ext3 to track changes made
995 to the inode. This will be required for version-based recovery.
1000 Description: MDS fails to start if a duplicate client export is detected
1001 Details : in some rare cases it was possible for a client to connect to
1002 an MDS multiple times. Upon recovery the MDS would detect this
1003 and fail during startup. Handle this more gracefully.
1005 Severity : enhancement
1007 Description: Add -o localflock option to simulate old noflock
1009 Details : This will achieve local-only flock/fcntl locks
1015 Description: log_commit_thread vs filter_destroy race leads to crash
1016 Details : Take import reference before releasing llog record semaphore
1021 Description: Wrong request locking in request set processing
1022 Details : ptlrpc_check_set wrongly uses req->rq_lock for proctect add to
1023 imp_delayed_list, in this place should be used imp_lock.
1026 Frequency : when reconnection
1028 Description: Grant Leak when osc reconnect to OST
1029 Details : When osc reconnect ost, OST(filter) should check whether it
1030 should grant more space to client by comparing fed_grant and
1031 cl_avail_grant, and return the granted space to client instead
1032 of "new granted" space, because client will call osc_init_grant
1033 to update the client grant space info.
1036 Frequency : when client reconnect to OST
1038 Description: Grant Leak when osc do resend and replay bulk write
1039 Details : When osc reconnect to OST, OST(filter)should clear grant info of
1040 bulk write request, because the grant info while be sync between
1041 OSC and OST when reconnect, and we should ignore the grant info
1042 these of resend/replay write req.
1047 Description: Grant space more than avaiable left space sometimes.
1048 Details : When then OST is about to be full, if two bulk writing from
1049 different clients came to OST. Accord the avaliable space of the
1050 OST, the first req should be permitted, and the second one
1051 should be denied by ENOSPC. But if the seconde arrived before
1052 the first one is commited. The OST might wrongly permit second
1053 writing, which will cause grant space > avaiable space.
1056 Frequency : when client is evicted
1058 Description: Grant might be wrongly erased when osc is evicted by OST
1059 Details : when the import is evicted by server, it will fork another
1060 thread ptlrpc_invalidate_import_thread to invalidate the
1061 import, where the grant will be set to 0. While the original
1062 thread will update the grant it got when connecting. So if
1063 the former happened latter, the grant will be wrongly errased
1064 because of this race.
1069 Description: Checking Stale with correct fid
1070 Details : ll_revalidate_it should uses de_inode instead of op_data.fid2
1071 to check whether it is stale, because sometimes, we want the
1072 enqueue happened anyway, and op_data.fid2 will not be initialized.
1074 Severity : enhancement
1076 Description: update patchless client
1077 Details : Add support for patchless client with 2.6.20, 2.6.21 and RHEL 5
1080 Frequency : only with 2.4 kernel
1082 Description: random memory corruption
1083 Details : size of struct ll_inode_info is to big for union inode.u and this
1084 can be cause of random memory corruption.
1089 Description: Memory leak in recovery
1090 Details : Lov_mds_md was not free in an error handler in mds_create_object.
1091 It should also check obd_fail before fsfilt_start, otherwise if
1092 fsfilt_start return -EROFS,(failover mds during mds recovery).
1093 then the req will return with repmsg->transno = 0 and rc = EROFS.
1094 and we met hit the assert LASSERT(req->rq_reqmsg->transno ==
1095 req->rq_repmsg->transno) in ptlrpc_replay_interpret. Fcc should
1096 be freed no matter whether fsfilt_commit success or not.
1099 Frequency : only with huge count clients
1101 Description: Prevents from taking the superblock lock in llap_from_page for
1103 Details : using LL_ORIGIN_REMOVEPAGE origin flag instead of LL_ORIGIN_UNKNOW
1104 for llap_from_page call in ll_removepage prevents from taking the
1105 superblock lock for a soon died page.
1110 Description: Not check open intent error before release open handle
1111 Details : in some rare cases, the open intent error is not checked before
1112 release open handle, which may cause
1113 ASSERTION(open_req->rq_transno != 0), because it tries to release
1114 the failed open handle.
1119 Description: Set cat log bitmap only after create log success.
1120 Details : in some rare cases, the cat log bitmap is set too early. and it
1121 should be set only after create log success.
1125 Description: Accessing a block bevice can re-enable I/O when Lustre is
1126 tearing down a device.
1127 Details : dev_clear_rdonly(bdev) must be called in kill_bdev() instead of
1132 Description: service threads may hog cpus when there are a lot of requests
1133 Details : Insert cond_resched to give other threads a chance to use some CPU
1138 Description: the cat log was not initialized in recovery
1139 Details : When mds(mgs) do recovery, the tgt_count might be zero, so the
1140 unlink log on mds will not be initialized until mds post
1141 recovery. And also in mds post recovery, the unlink log will
1142 initialization will be done asynchronausly, so there will be race
1143 between add unlink log and unlink log initialization.
1147 Description: brw_stats were being printed incorrectly
1148 Details : brw_stats were being printed as log2 but all of them were not
1149 recorded as log2. Also remove some code duplication arising from
1150 filter_tally_{read,write}.
1154 Frequency : rare, only in recovery.
1155 Description: ASSERTION(req->rq_type != LI_POISON) failed
1156 Details : imp_lock should be held while iterating over imp_sending_list for
1157 prevent destroy request after get timeout in ptlrpc_queue_wait.
1161 Description: replay-single.sh test 52 fails
1162 Details : A lock's skiplist need to be cleanup when it being unlinked
1163 from its resource list.
1167 Description: Short directio read returns full requested size rather than
1169 Details : Direct I/O operations should return actual amount of bytes
1170 transferred rather than requested size.
1172 Severity : enhancement
1174 Description: metadata RPC reduction (e.g. for rm performance)
1175 Details : decrease the amount of synchronous RPC between clients and servers
1176 by canceling conflicing lock before the operation on the client
1177 and packing thier handles into the main operation RPC to server.
1179 Severity : enhancement
1181 Description: Async OSC create to avoid the blocking unnecessarily.
1182 Details : If a OST has no remain object, system will block on the creating
1183 when need to create a new object on this OST. Now, ways use
1184 pre-created objects when available, instead of blocking on an
1185 empty osc while others are not empty. If we must block, we block
1186 for the shortest possible period of time.
1188 Severity : enhancement
1190 Description: refine locking for avoid write wrong info into lov_objid file
1191 Details : fix possible races with add new target and write/update data in
1194 --------------------------------------------------------------------------------
1196 2007-05-03 Cluster File Systems, Inc. <info@clusterfs.com>
1201 Frequency : on some architectures
1203 Description: 1.6 client sometimes fails to mount from a 1.4 MDT
1204 Details : Uninitialized flags sometimes cause configuration commands to
1208 Frequency : patchless clients only
1210 Description: missing __iget() symbol export
1211 Details : The __iget() symbol export is missing. To avoid the need for
1212 this on patchless clients the deathrow inode reaper is turned
1213 off, and we depend on the VM to clean up old inodes. This
1214 dependency was during via the fix for bug 12181.
1216 --------------------------------------------------------------------------------
1218 2007-04-19 Cluster File Systems, Inc. <info@clusterfs.com>
1220 * CONFIGURATION CHANGE. This version of Lustre WILL NOT
1221 INTEROPERATE with older versions automatically. In many cases a
1222 special upgrade step is needed. Please read the
1223 user documentation before upgrading any part of a live system.
1224 * WIRE PROTOCOL CHANGE from previous 1.6 beta versions. This
1225 version will not interoperate with 1.6 betas before beta5 (1.5.95).
1226 * WARNING: Lustre configuration and startup changes are required with
1227 this release. See https://mail.clusterfs.com/wikis/lustre/MountConf
1233 Severity : enhancement
1235 Description: MountConf
1236 Details : Lustre configuration is now managed via mkfs and mount
1237 commands instead of lmc and lconf. New obd types (MGS, MGC)
1238 are added for dynamic configuration management. See
1239 https://mail.clusterfs.com/wikis/lustre/MountConf for
1242 Severity : enhancement
1244 Description: dynamic OST addition
1245 Details : OSTs can now be added to a live filesystem
1247 Severity : enhancement
1249 Description: startup order invariance
1250 Details : MDTs and OSTs can be started in any order. Clients only
1251 require the MDT to complete startup.
1253 Severity : enhancement
1255 Description: parallel, asynchronous orphan cleanup
1256 Details : orphan cleanup is now performed in separate threads for each
1257 OST, allowing parallel non-blocking operation.
1259 Severity : enhancement
1261 Description: optimized stripe assignment
1262 Details : stripe assignments are now made based on ost space available,
1263 ost previous usage, and OSS previous usage, in order to try
1264 to optimize storage space and networking resources.
1266 Severity : enhancement
1268 Description: Permanently set tunables
1269 Details : All writable /proc/fs/lustre tunables can now be permanently
1270 set on a per-server basis, at mkfs time or on a live system.
1272 Severity : enhancement
1274 Description: Lustre message v2
1275 Details : Add lustre message format v2.
1277 Severity : enhancement
1279 Description: client OST exclusion list
1280 Details : Clients can be started with a list of OSTs that should be
1281 declared "inactive" for known non-responsive OSTs.
1285 Description: SPEC SFS validation failure on NFS v2 over lustre.
1286 Details : Changes the blocksize for regular files to be 2x RPC size,
1287 and not depend on stripe size.
1289 Severity : enhancement
1291 Description: Multiple MD RPCs in flight.
1292 Details : Further unserialise some read-only MDS RPCs - learn about intents.
1293 To avoid overly-overloading MDS, introduce a limit on number of
1294 MDS RPCs in flight for a single client and add /proc controls
1295 to adjust this limit.
1297 Severity : enhancement
1299 Description: client read/write statistics
1300 Details : Add client read/write call usage stats for performance
1301 analysis of user processes.
1302 /proc/fs/lustre/llite/*/offset_stats shows non-sequential
1303 file access. extents_stats shows chunk size distribution.
1304 extents_stats_per_process show chunk size distribution per
1307 Severity : enhancement
1309 Description: mds statistics
1310 Details : Add detailed mds operations statistics in
1311 /proc/fs/lustre/mds/*/stats.
1315 Description: Failure of copying files with lustre special EAs.
1316 Details : Client side always return success for setxattr call for lustre
1317 special xattr (currently only "trusted.lov").
1322 Description: Refcount LNET uuids
1323 Details : The global LNET uuid list grew linearly with every startup;
1324 refcount repeated list entries instead of always adding to
1327 Severity : enhancement
1329 Description: Dynamic service threads
1330 Details : Within a small range, start extra service threads
1331 automatically when the request queue builds up.
1334 Frequency : mixed-endian client/server environments
1336 Description: mixed-endian crashes
1337 Details : The new msg_v2 system had some failures in mixed-endian
1340 Severity : enhancement
1342 Description: Easy OST removal
1343 Details : OSTs can be permanently deactivated with e.g. 'lctl
1344 conf_param lustre-OST0001.osc.active=0'
1346 Severity : enhancement
1348 Description: MGS proc entries
1349 Details : Added basic proc entries for the MGS showing what filesystems
1352 Severity : enhancement
1354 Description: provide MGS failover
1355 Details : Added config lock reacquisition after MGS server failover.
1357 Severity : enhancement
1359 Description: add Linux 2.4 support
1360 Details : Added support for RHEL 2.4.21 kernel for 1.6 servers and clients
1364 Description: a large application tries to do I/O to the same resource and dies
1365 in the middle of it.
1366 Details : Check the req->rq_arrival time after the call to
1367 ost_brw_lock_get(), but before we do anything about
1368 processing it & sending the BULK transfer request. This
1369 should help move old stale pending locks off the queue as
1370 quickly as obd_timeout.
1373 Frequency : when an incorrect nid is specified during startup
1375 Description: ptlrpc connect to non-existant node causes kernel crash
1376 Details : LNET can't be re-entered from an event callback, which
1377 happened when we expire a message after the export has been
1378 cleaned up. Instead, hand the zombie cleanup off to another
1381 Severity : enhancement
1383 Description: plain/inodebits lock performance improvement
1384 Details : Grouping plain/inodebits in granted list by their request modes
1385 and bits policy, thus improving the performance of search through
1389 Frequency : only if OST filesystem is corrupted
1391 Description: client incorrectly hits assertion in ptlrpc_replay_req()
1392 Details : for a short time RPCs with bulk IO are in the replay list,
1393 but replay of bulk IOs is unimplemented. If the OST filesystem
1394 is corrupted due to disk cache incoherency and then replay is
1395 started it is possible to trip an assertion. Avoid putting
1396 committed RPCs into the replay list at all to avoid this issue.
1399 Frequency : liblustre (e.g. catamount) on a large cluster with >= 8 OSTs/OSS
1401 Description: System hang on startup
1402 Details : This bug allowed the liblustre (e.g. catamount) client to
1403 return to the app before handling all startup RPCs. This
1404 could leave the node unresponsive to lustre network traffic
1405 and manifested as a server ptllnd timeout.
1407 Severity : enhancement
1409 Description: Add "/proc/sys/lustre/debug_peer_on_timeout"
1410 Details : liblustre envirable: LIBLUSTRE_DEBUG_PEER_ON_TIMEOUT
1411 boolean to control whether to print peer debug info when a
1412 client's RPC times out.
1415 Frequency : only for kernels with patches from Lustre below 1.4.3
1417 Description: Remove old rdonly API
1418 Details : Remove old rdonly API which unused from at least lustre 1.4.3
1421 Frequency : only for devices with external journals
1423 Description: Set external device read-only also
1424 Details : During a commanded failover stop, we set the disk device
1425 read-only while the server shuts down. We now also set any
1426 external journal device read-only at the same time.
1429 Frequency : when upgrading from 1.4 while trying to change parameters
1431 Description: The wrong (new) MDC name was used when setting parameters for
1432 upgraded MDT's. Also allows changing of OSC (and MDC)
1433 parameters if --writeconf is specified at tunefs upgrade time.
1436 Frequency : when setting specific ost indicies
1438 Description: QOS code breaks on skipped indicies
1439 Details : Add checks for missing OST indicies in the QOS code, so OSTs
1440 created with --index need not be sequential.
1442 Severity : enhancement
1444 Description: Add uninit_groups feature to ldiskfs2 to speed up e2fsck
1445 Details : The uninit_groups feature works in conjunction with the kernel
1446 filesystem code (ldiskfs2 only) and e2fsprogs-1.39-cfs6 to speed
1447 up the pass1 processing of e2fsck. This is a read-only feature
1448 in ldiskfs2 only, so older kernels and current ldiskfs cannot
1449 mount filesystems that have had this feature enabled.
1451 Severity : enhancement
1453 Description: Improve multi-block allocation algorithm to avoid fragmentation
1454 Details : The mballoc3 code (ldiskfs2 only) adds new mechanisms to improve
1455 allocation locality and avoid filesystem fragmentation.
1457 ------------------------------------------------------------------------------
1459 2007-04-01 Cluster File Systems, Inc. <info@clusterfs.com>
1461 * Support for kernels:
1462 2.4.21-47.0.1.EL (RHEL 3)
1463 2.6.5-7.283 (SLES 9)
1464 2.6.9-42.0.10.EL (RHEL 4)
1465 2.6.12.6 vanilla (kernel.org)
1466 2.6.16.27-0.9 (SLES 10)
1467 * Recommended e2fsprogs version: 1.39.cfs5
1469 * Note that reiserfs quotas are disabled on SLES 10 in this kernel
1473 Frequency : occasional, depends on client load and configuration
1474 Bugzilla : 12181, 12203
1475 Description: data loss for recently-modified files
1477 Details : In some cases it is possible that recently written or created
1478 files may not be written to disk in a timely manner (this should
1479 normally be within 30s unless client IO load is very high).
1480 The problem appears as zero-length files or files that are a
1481 multiple of 1MB in size after a client crash or client eviction
1482 that are missing data at the end of the file.
1484 This problem is more likely to be hit on clients where files are
1485 repeatedly created and unlinked in the same directory, clients
1486 have a large amount of RAM, have many CPUs, the filesystem has
1487 many OSTs, the clients are rebooted frequently, and/or the files
1488 are not accessed by other nodes after being written.
1490 The presence of the problem can be detected by looking at
1491 /proc/sys/fs/inode-state. If the first number (nr_inodes) is
1492 smaller than the second (nr_unused) then dirty files will not
1493 be flushed automatically to disk. "sync; sleep 10" should be
1494 run several times on the node before unmounting it to update
1495 Lustre (this is also safe to run on nodes without this problem).
1497 There is also a related kernel bug in the RHEL4 4 2.6.9 kernel
1498 that can cause this same problem, so customers using that kernel
1499 also need to update the kernel in addition to Lustre. In order
1500 to properly fix this bug, the RHEL3 2.4.21 kernel is also updated.
1502 It is normal that files written just before a client crash (less
1503 than 30s) may not yet have been flushed to disk, even for local
1507 Frequency : frequent on thin XT3 nodes
1509 Description: UUID collision on thin XT3 Linux nodes
1510 Details : UUIDs on Compute Node Linux XT3 nodes were not generated
1511 randomly, since we relied on an insufficiently-seeded PRNG.
1516 Description: OSS hangs after "All ost request buffers busy"
1517 Details : A deadlock between quota and journal operations caused OSS
1518 hangs after printing "All ost request buffers busy."
1521 Frequency : always on liblustre builds
1523 Description: Cleanup compiler warnings on liblustre
1526 Frequency : always on liblustre builds on XT3
1528 Description: LC_CONFIG_CDEBUG don't run while build liblustre on XT3.
1532 Description: Addition of EXT3_FEATURE_RO_COMPAT_DIR_NLINKS flag for
1533 > 32000 subdirectories
1534 Details : Add EXT3_FEATURE_RO_COMPAT_DIR_NLINK flag to
1535 EXT3_FEATURE_RO_COMPAT_SUPP. This flag will be set whenever
1536 subdirectory count crosses 32000. This will aid e2fsck to
1537 correctly handle more than 32000 subdirectories.
1540 Frequency : liblustre (e.g. catamount) on a large cluster with >= 8 OSTs/OSS
1542 Description: System hang on startup
1543 Details : This bug allowed the liblustre (e.g. catamount) client to
1544 return to the app before handling all startup RPCs. This
1545 could leave the node unresponsive to lustre network traffic
1546 and manifested as a server ptllnd timeout.
1548 Severity : enhancement
1550 Description: Add "/proc/sys/lustre/debug_peer_on_timeout"
1551 (liblustre envirable: LIBLUSTRE_DEBUG_PEER_ON_TIMEOUT)
1552 boolean to control whether to print peer debug info when a
1553 client's RPC times out.
1558 Description: make O_SYNC working on 2.6 kernels
1559 Details : 2.6 kernels use different method for mark pages for write,
1560 so need add a code to lustre for O_SYNC work.
1565 Description: Failure to close file and release space on NFS
1566 Details : Put inode details into lock acquired in ll_intent_file_open.
1567 Use mdc_intent_lock in ll_intent_open to properly
1568 detect all kind of errors unhandled by mdc_enqueue.
1573 Description: proc file read during shutdown sometimes raced obd removal,
1575 Details : Add lock to prevent obd access after proc file removal.
1578 Frequency : Only for files larger than 4GB on 32-bit clients.
1580 Description: improperly doing page alignment of locks
1581 Details : Modify lustre core code to use CFS_PAGE_* defines instead of
1582 PAGE_*. Make CFS_PAGE_MASK a 64-bit mask.
1587 Description: RPCs being resent when they shouldn't be
1588 Details : Some RPCs that should not be resent are being resent. This
1589 can cause inconsistencies in the RPC state machine. Do not
1590 resend such requests.
1593 Frequency : rare, only with NFS export
1595 Description: Crash on NFS re-export node
1596 Details : under very unusual load conditions an assertion is hit in
1597 ll_intent_file_open()
1600 Frequency : only if OST filesystem is corrupted
1602 Description: client incorrectly hits assertion in ptlrpc_replay_req()
1603 Details : for a short time RPCs with bulk IO are in the replay list,
1604 but replay of bulk IOs is unimplemented. If the OST filesystem
1605 is corrupted due to disk cache incoherency and then replay is
1606 started it is possible to trip an assertion. Avoid putting
1607 committed RPCs into the replay list at all to avoid this issue.
1612 Description: large O_DIRECT requests fail under memory pressure/fragmentation
1613 Details : Large single O_DIRECT read and write calls can fail to allocate
1614 a sufficiently large buffer to process the request. In case of
1615 allocation failure the allocation is retried with a smaller
1616 buffer and broken into smaller requests.
1618 Severity : enhancement
1620 Description: Add -o localflock option to simulate old noflock behaviour.
1621 Details : This will achieve local-only flock/fcntl locks coherentness.
1626 Description: versioning check is incomplete
1627 Details : Checking the version difference of client vs. server, report
1628 error if the gap is too big.
1633 Description: add support PG_writeback bit
1634 Details : add support for PG_writeback bit for Lustre, for more carefull
1635 work with page cache in 2.6 kernel. This also fix some deadlocks
1636 and remove hack for work O_SYNC with 2.6 kernel.
1638 Severity : enhancement
1640 Description: Add uninit_groups feature to ldiskfs2 to speed up e2fsck
1641 Details : The uninit_groups feature works in conjunction with the kernel
1642 filesystem code (ldiskfs2 only) and e2fsprogs-1.39-cfs6 to speed
1643 up the pass1 processing of e2fsck. This is a read-only feature
1644 in ldiskfs2 only, so older kernels and current ldiskfs cannot
1645 mount filesystems that have had this feature enabled.
1647 Severity : enhancement
1649 Description: Improve multi-block allocation algorithm to avoid fragmentation
1650 Details : The mballoc3 code (ldiskfs2 only) adds new mechanisms to improve
1651 allocation locality and avoid filesystem fragmentation.
1653 ------------------------------------------------------------------------------
1655 2007-02-09 Cluster File Systems, Inc. <info@clusterfs.com>
1657 * Support for kernels:
1658 2.6.9-42.0.3.EL (RHEL 4)
1659 2.6.5-7.276 (SLES 9)
1660 2.4.21-47.0.1.EL (RHEL 3)
1661 2.6.12.6 vanilla (kernel.org)
1662 2.6.16.21-0.8 (SLES10)
1663 * Recommended e2fsprogs version: 1.39.cfs2-0
1665 * The backwards-compatible /proc/sys/portals symlink has been removed
1666 in this release. Before upgrading, please ensure that you change
1667 any configuration scripts or /etc/sysctl.conf files that access
1668 /proc/sys/portals/* or sysctl portals.* to use the corresponding
1669 entry in /proc/sys/lnet or sysctl lnet.*. This change can be made
1670 in advance of the upgrade on any system running Lustre 1.4.6 or
1671 newer, since /proc/sys/lnet was added in that version.
1672 * Note that reiserfs quotas are disabled on SLES 10 in this kernel
1676 Frequency : only when quota is used
1678 Description: avoid scanning export list for quota master
1679 Details : Change the algorithms to avoid scanning export list in order
1680 to improve the efficiency.
1683 Frequency : MDS failover only, very rarely
1685 Description: "went back in time" messages on mds failover
1686 Details : The greatest transno may be lost when the current operation
1687 finishes with an error (transno==0) and the client's last_rcvd
1688 record is over-written. Save the greatest transno in the
1689 mds_last_transno for this case.
1692 Frequency : always for specific kernels and striping counts
1694 Description: client may get "Matching packet too big" without ACL support
1695 Details : Clients compiled without CONFIG_FS_POSIX_ACL get an error message
1696 when trying to access files in certain configurations. The
1697 clients should in fact be denied when mounting because they do
1698 not understand ACLs.
1701 Frequency : Cray XT3 with more than 4000 clients and multiple jobs
1703 Description: many clients connecting with IO in progress causes connect timeouts
1704 Details : Avoid synchronous journal commits to avoid delays caused by many
1705 clients connecting/disconnecting when bulk IO is in progress.
1706 Queue liblustre connect requests on OST_REQUEST_PORTAL instead of
1707 OST_IO_PORTAL to avoid delays behind potentially many pending
1711 Frequency : occasionally with multiple writers to a single file
1713 Description: shared writes to file may result in wrong size reported by stat()
1714 Details : Allow growing of kms when extent lock is cancelled
1717 Frequency : always with random mmap IO to multi-striped file
1719 Description: mmap write might be lost if we are writing to a 'hole' in stripe
1720 Details : Only if the hole is at the end of OST object so that kms is too
1721 small. Fix is to increase kms accordingly in ll_nopage.
1724 Frequency : rare, only if OST filesystem is inconsistent with MDS filesystem
1726 Description: writes to a missing object would leak memory on the OST
1727 Details : If there is an inconsistency between the MDS and OST filesystems,
1728 such that the MDS references an object that doesn't exist, writes
1729 to that object will leak memory due to incorrect cleanup in the
1730 error handling path, eventually running out of memory on the OST.
1735 Description: Creating too long symlink causes lustre errors
1736 Details : Check symlink and name lengths before sending requests to MDS.
1739 Frequency : only if flock is enabled (not on by default)
1741 Description: posix locks not released on fd closure on 2.6.9+
1742 Details : We failed to add posix locks to list of inode locks on 2.6.9+
1743 kernels, this caused such locks not to be released on fd close and
1744 then assertions on fs unmount about still used locks.
1747 Frequency : MDS failover only, very rarely
1749 Description: clients may get ASSERTION(granted_lock != NULL)
1750 Details : When request was taking a long time, and a client was resending
1751 a getattr by name lock request. The were multiple lock requests
1752 with the same client lock handle and
1753 mds_getattr_name->fixup_handle_for_resent_request found one of the
1754 lock handles but later failed with ASSERTION(granted_lock != NULL).
1759 Description: handle->h_buffer_credits > 0, assertion failure
1760 Details : h_buffer_credits is zero after truncate, causing assertion
1761 failure. This patch extends the transaction or creates a new
1765 Frequency : NFS re-export or patchless client
1766 Bugzilla : 11179, 10796
1767 Description: Crash on NFS re-export node (__d_move)
1768 Details : We do not want to hash the dentry if we don't have a lock.
1769 But if this dentry is later used in d_move, we'd hit uninitialised
1770 list head d_hash, so we just do this to init d_hash field but
1771 leave dentry unhashed.
1774 Frequency : NFS re-export or patchless client
1776 Description: NFS exports has problem with symbolic link
1777 Details : lustre client didn't properly install dentry when re-exported
1778 to NFS or running patchless client.
1781 Frequency : NFS re-export or patchless client
1783 Description: Various nfs/patchless fixes.
1784 Details : fixes reuse disconected alias for lookup process - this fixes
1785 warning "find_exported_dentry: npd != pd",
1786 fix permission error with open files at nfs.
1787 fix apply umask when do revalidate.
1790 Frequency : occasional
1792 Description: Crash on NFS re-export node
1793 Details : calling clear_page() on the wrong pointer triggered oops in
1794 generic_mapping_read().
1797 Frequency : rarely, using O_DIRECT IO
1799 Description: unaligned directio crashes client with LASSERT
1800 Details : check for unaligned buffers before trying any requests.
1803 Frequency : rarely, using CFS RAID5 patches in non-standard kernel series
1805 Description: stale data returned from RAID cache
1806 Details : If only a small amount of IO is done to the RAID device before
1807 reading it again it is possible to get stale data from the RAID
1808 cache instead of reading it from disk.
1811 Frequency : always for sles10 kernel
1813 Description: sles10 support
1814 Details : ll_follow_link: compile fixes and using of nd_set_link
1815 under newer kernels.
1818 Frequency : depends on arch, kernel and compiler version, always on sles10
1821 Description: recursive or deep enough symlinks cause stack overflow
1822 Details : getting rid of large stack-allocated variable in
1826 Frequency : depends on hardware
1828 Description: lustre write performance loss in the SLES10 kernel
1829 Details : the performance loss is caused by using of write barriers in the
1830 ext3 code. The SLES10 kernel turns barrier support on by
1831 default. The fix is to undo that change for ldiskfs.
1833 ------------------------------------------------------------------------------
1835 2006-12-09 Cluster File Systems, Inc. <info@clusterfs.com>
1837 * Support for kernels:
1838 2.6.9-42.0.3EL (RHEL 4)
1839 2.6.5-7.276 (SLES 9)
1840 2.4.21-47.0.1.EL (RHEL 3)
1841 2.6.12.6 vanilla (kernel.org)
1845 Frequency : quota enabled and large files being deleted
1847 Description: releasing more than 4GB of quota at once hangs OST
1848 Details : If a user deletes more than 4GB of files on a single OST it
1849 will cause the OST to spin in an infinite loop. Release
1850 quota in < 4GB chunks, or use a 64-bit value for 1.4.7.1+.
1855 Description: statfs data retrieved from /proc may be stale or zero
1856 Details : When reading per-device statfs data from /proc, in the
1857 {kbytes,files}_{total,free,avail} files, it may appear
1858 as zero or be out of date.
1861 Frequency : systems with MD RAID1 external journal devices
1863 Description: lconf's call to blkid is confused by RAID1 journal devices
1864 Details : Use the "blkid -l" flag to locate the MD RAID device instead
1865 of returning all block devices that match the journal UUID.
1868 Frequency : always, for aggregate stripe size over 4GB
1870 Description: "lfs setstripe" fails assertion when setting 4GB+ stripe width
1871 Details : Using "lfs setstripe" to set stripe size * stripe count over 4GB
1872 will fail the kernel with "ASSERTION(lsm->lsm_xfersize != 0)"
1875 Frequency : always if "lfs find" used on a local file/directory
1877 Description: "lfs find" segfaults if used on a local file/directory
1878 Details : The case where a directory component was not specified wasn't
1879 handled correctly. Handle this properly.
1882 Frequency : always on ppc64
1884 Description: the write to an ext3 filesystem mounted with mballoc got stuck
1885 Details : ext3_mb_generate_buddy() uses find_next_bit() which does not
1886 perform endianness conversion.
1889 Frequency : rarely (truncate to non-zero file size after write under load)
1890 Bugzilla : 10730, 10687
1891 Description: Files padded with zeros to next 4K multiple
1892 Details : With filesystems mounted using the "extents" option (2.6 kernels)
1893 it is possible that files that are truncated to a non-zero size
1894 immediately after being written are filled with zero bytes beyond
1895 the truncated size. No file data is lost.
1897 Severity : enhancement
1899 Description: Allow recovery/failover for liblustre clients.
1900 Details : liblustre clients were unaware of failover configurations until
1903 Severity : enhancement
1905 Description: user file locks should fail when not mounting with flock option
1906 Details : Set up an error-returning stub in ll_file_operations.lock field
1907 to prevent incorrect behaviour when client is mounted without
1908 flock option. Also, set up properly f_op->flock field for
1912 Frequency : always on ia64
1914 Description: "lfs df" loops on printing out MDS statfs information
1915 Details : The obd_ioctl_data was not initialized and in some systems
1916 this caused a failure during the ioctl that did not return
1917 an error. Initialize the struct and return an error on failure.
1920 Frequency : SLES 9 only
1922 Description: Error of copying files with lustre special EAs as root
1923 Details : Client side always return success for setxattr call for lustre
1924 special xattr (currently only "trusted.lov").
1927 Frequency : rarely on clusters with both ia64+i386 clients
1929 Description: ia64+i686 clients doing shared IO on the same file may LBUG
1930 Details : In rare cases when both ia64+i686 (or other mixed-PAGE_SIZE)
1931 clients are doing concurrent writes to the same file it is
1932 possible that the ia64 clients may LASSERT because the OST
1933 extent locks are not PAGE_SIZE aligned. Ensure that grown
1934 locks are always aligned on the request boundary.
1937 Frequency : specific use, occasional
1939 Description: Overwriting in use executable truncates on-disk binary image
1940 Details : If one node attempts to overwrite an executable in use by
1941 another node, we now correctly return ETXTBSY instead of
1942 truncating the file.
1944 Severity : enhancement
1946 Description: Async OSC create to avoid the blocking unnecessarily.
1947 Details : If a OST has no remain object, system will block on the creating
1948 when need to create a new object on this OST. Now, ways use
1949 pre-created objects when available, instead of blocking on an
1950 empty osc while others are not empty. If we must block, we block
1951 for the shortest possible period of time.
1956 Description: chmod on Lustre root is propagated to other clients
1957 Details : Re-validate root's dentry in ll_lookup_it to avoid having it
1958 invalid by the follow_mount time.
1961 Frequency : liblustre clients only
1963 Description: Race in 'instant cancel' lock handling could lead to such locks
1964 never to be granted in case of SMP MDS
1965 Details : Do not destroy not yet granted but cbpending locks in
1969 Frequency : replay/resend of open
1971 Description: non null lock assetion failure in mds_intent_policy
1972 Details : Trying to replay/resend lockless open requests resulted in
1973 mds_open() returning 0 with no lock. Now it sets a flag if
1974 a lock is going to be returned.
1976 Severity : enhancement
1978 Description: Checksum enhancements
1979 Details : New checksum enhancements allow for resending RPCs that failed
1982 Severity : enhancement
1984 Description: Tunables on number of dirty pages in cacche
1985 Details : Allow to set limit on number of dirty pages cached.
1990 Description: client crash on unmount - lock still has references
1991 Details : In some error handling cases it was possible to leak a lock
1992 reference on a client while accessing a file. This was not
1993 harmful to the client during operation, but would cause the
1994 client to crash when the filesystem is unmounted.
1997 Frequency : specific case, rare
1999 Description: ETXTBSY on mds though file not in use by client
2000 Details : ETXTBSY is no longer incorrectly returned when attempting to
2001 chmod or chown a directory that the user previously tried to
2002 execute or a currently-executing binary.
2005 Frequency : extremely rare except on liblustre-based clients
2007 Description: Lustre space not freed when files are deleted
2008 Details : Clean up open-unlinked files after client eviction. Previously
2009 the unlink was skipped and the files remained as orphans.
2014 Description: OST failure "would be an LBUG" in waiting_locks_callback()
2015 Details : In some cases it was possible to send a blocking callback to
2016 a client doing a glimpse, even though that client didn't get
2017 a lock granted. When the glimpse lock is cancelled on the OST
2018 the freed lock is left on the waiting list and corrupted the list.
2021 Frequency : all core dumps
2023 Description: Broke core dumps to lustre
2024 Details : Negative dentry may be unhashed if parent does not have UPDATE
2025 lock, but some callers, e.g. do_coredump, expect dentry to be
2026 hashed after successful create, hash it in ll_create_it.
2028 ------------------------------------------------------------------------------
2030 2006-09-13 Cluster File Systems, Inc. <info@clusterfs.com>
2032 * Support for kernels:
2033 2.6.9-42.0.2.EL (RHEL 4)
2034 2.6.5-7.276 (SLES 9)
2035 2.4.21-40.EL (RHEL 3)
2036 2.6.12.6 vanilla (kernel.org)
2040 Frequency : always on RHEL 3
2042 Description: Number of open files grows over time
2043 Details : The number of open files grows over time, whether or not
2044 Lustre is started. This was due to a filp leak introduced
2045 by one of our kernel patches.
2047 ------------------------------------------------------------------------------
2049 08-20-2006 Cluster File Systems, Inc. <info@clusterfs.com>
2051 * Support for kernels:
2052 2.6.9-42.EL (RHEL 4)
2053 2.6.5-7.276 (SLES 9)
2054 2.4.21-40.EL (RHEL 3)
2055 2.6.12.6 vanilla (kernel.org)
2060 Bugzilla : 5719, 9635, 9792, 9684
2061 Description: OST (or MDS) trips assertions in (re)connection under heavy load
2062 Details : If a server is under heavy load and cannot reply to new
2063 connection requests before the client resends the (re)connect,
2064 the connection handling code can behave badly if two service
2065 threads are concurrently handing separate (re)connections from
2066 the same client. Add better locking to the connection handling
2067 code, and ensure that only a single connection will be processed
2068 for a given client UUID, even if the lock is dropped.
2070 Severity : enhancement
2072 Description: add TCP zero-copy support to kernel
2073 Details : Add support to the kernel TCP stack to allow zero-copy bulk
2074 sends if the hardware supports scatter-gather and checksumming.
2075 This allows socklnd to do client-write and server-read more
2076 efficiently and reduce CPU utilization from skbuf copying.
2079 Frequency : only if NFS exporting from client
2081 Description: NULL pointer deref in ll_iocontrol() if chattr mknod file
2082 Details : If setting attributes on a file created under NFS that had
2083 never been opened it would be possible to oops the client
2084 if the file had no objects.
2087 Frequency : always for liblustre
2089 Description: liblustre client does MDS+OSTs setattr RPC for each write
2090 Details : When doing a write from a liblustre client, the client
2091 incorrectly issued an RPC to the MDS and each OST the file was
2092 striped over in order to update the timestamps. When writing
2093 with small chunks and many clients this could overwhelm the MDS
2094 with RPCs. In all cases it would slow down the write because
2095 these RPCs are unnecessary.
2097 Severity : enhancement
2099 Description: allow number of MDS service threads to be changed at module load
2100 Details : It is now possible to change the number of MDS service threads
2101 running. Adding "options mds mds_num_threads={N}" to the MDS's
2102 /etc/modprobe.conf will set the number of threads for the next
2103 time Lustre is restarted (assuming the "mds" module is also
2104 reloaded at that time). The default number of threads will
2105 stay the same, 32 for most systems.
2110 Description: OST crash if filesystem is unformatted or corrupt
2111 Details : If an OST is started on a device that has never been formatted
2112 or if the filesystem is corrupt and cannot even mount then the
2113 error handling cleanup routines would dereference a NULL pointer.
2118 Description: NULL pointer deref in llap_from_page.
2119 Details : get_cache_page_nowait can return a page with NULL (or otherwise
2120 incorrect) mapping if the page was truncated/reclaimed while it was
2121 searched for. Check for this condition and skip such pages when
2122 doing readahead. Introduce extra check to llap_from_page() to
2123 verify page->mapping->host is non-NULL (so page is not anonymous).
2126 Frequency : Sometimes when using sys_sendfile
2128 Description: "page not covered by a lock" warnings from ll_readpage
2129 Details : sendfile called ll_readpage without right page locks present.
2130 Now we introduced ll_file_sendfile that does necessary locking
2131 around call to generic_file_sendfile() much like we do in
2135 Frequency : with certain MDS communication failures at client mount time
2137 Description: NULL pointer deref after failed client mount
2138 Details : a client connection request may delayed by the network layer
2139 and not be sent until after the PTLRPC layer has timed out the
2140 request. If the client fails the mount immediately it will try
2141 to clean up before the network times out the request. Add a
2142 reference from the request import to the obd device and delay
2143 the cleanup until the network drops the request.
2146 Frequency : occasionally during client (re)connect
2148 Description: assertion failure during client (re)connect
2149 Details : processing a client connection request may be delayed by the
2150 client or server longer than the client connect timeout. This
2151 causes the client to resend the connection request. If the
2152 original connection request is replied in this interval, the
2153 client may trip an assertion failure in ptlrpc_connect_interpret()
2154 which thought it would be the only running connect process.
2157 Frequency : only with obd_echo servers and clients that are rebooted
2159 Description: kernel BUG accessing uninitialized data structure
2160 Details : When running an obd_echo server it did not start the ping_evictor
2161 thread, and when a client was evicted an uninitialized data
2162 structure was accessed. Start the ping_evictor in the RPC
2163 service startup instead of the OBD startup.
2165 Severity : enhancement
2166 Bugzilla : 10193 (patchless)
2167 Description: Remove dependency on various unexported kernel interfaces.
2168 Details : No longer need reparent_to_init, exit_mm, exit_files,
2169 sock_getsockopt, filemap_populate, FMODE_EXEC, put_filp.
2172 Frequency : rare (only users of deprecated and unsupported LDAP config)
2174 Description: write_conf for zeroconf mount queried LDAP incorrectly for client
2175 Details : LDAP apparently contains 'lustreName' attributes instead of
2176 'name'. A simple remapping of the name is sufficient.
2179 Frequency : rare (only with non-default dump_on_timeout debug enabled)
2181 Description: waiting_locks_callback trips kernel BUG if client is evicted
2182 Details : Running with the dump_on_timeout debug flag turned on makes
2183 it possible that the waiting_locks_callback() can try to dump
2184 the Lustre kernel debug logs from an interrupt handler. Defer
2185 this log dumping to the expired_lock_main() thread.
2187 Severity : enhancement
2189 Description: Support NFS exporting on 2.6 kernels.
2190 Details : Implement non-rawops metadata methods for NFS server to use without
2191 changing NFS server code.
2194 Frequency : very rare (synthetic metadata workload only)
2196 Description: two racing renames might cause an MDS thread to deadlock
2197 Details : Running the "racer" program may cause one MDS thread to rename
2198 a file from being the source of a rename to being the target of
2199 a rename at exactly the same time that another thread is doing
2200 so, and the second thread has already enqueued these locks after
2201 doing a lookup of the target and is trying to relock them in
2202 order. Ensure that we don't try to re-lock the same resource.
2205 Frequency : only very large systems with liblustre clients
2207 Description: slow eviction of liblustre clients with the "evict_by_nid" RPC
2208 Details : Use asynchronous set_info RPCs to send the "evict_by_nid" to
2209 all OSTs in parallel. This allows the eviction of stale liblustre
2210 clients to proceed much faster than if they were done in series,
2211 and also offers similar improvements for other set_info RPCs.
2216 Description: excessive CPU usage during initial read phase on client
2217 Details : During the initial read phase on a client, it would agressively
2218 retry readahead on the file, consuming too much CPU and impacting
2219 performance (since 1.4.5.8). Improve the readahead algorithm
2220 to avoid this, and also improve some other common cases (read
2221 of small files in particular, where "small" is files smaller than
2222 /proc/fs/lustre/llite/*/max_read_ahead_whole_mb, 2MB by default).
2227 Description: MDS crash when receiving packet with unknown intent.
2228 Details : Do not LBUG in unknown intent case, just return -EFAULT
2230 Severity : enhancement
2231 Bugzilla : 9293, 9385
2232 Description: MDS RPCs are serialised on client. This is unnecessary for some.
2233 Details : Do not serialize getattr (non-intent version) and statfs.
2236 Frequency : occasional, when OST network is overloaded/intermittent
2238 Description: client evicted by OST after bulk IO timeout
2239 Details : If a client sends a bulk IO request (read or write) the OST
2240 may evict the client if it is unresposive to its data GET/PUT
2241 request. This is incorrect if the network is overloaded (takes
2242 too long to transfer the RPC data) or dropped the OST GET/PUT
2243 request. There is no need to evict the client at all, since
2244 the pinger and/or lock callbacks will handle this, and the
2245 client can restart the bulk request.
2248 Frequency : Always when mmapping file with no objects
2250 Description: client crashes when mmapping file with no objects
2251 Details : Check that we actually have objects in a file before doing any
2252 operations on objects in ll_vm_open, ll_vm_close and
2258 Description: Request leak when working with deleted CWD
2259 Details : Introduce advanced request refcount tracking for requests
2260 referenced from lustre intent.
2262 Severity : Enhancement
2264 Description: Cache open file handles on client.
2265 Details : MDS now will return special lock along with openhandle, if
2266 requested and client is allowed to hold openhandle, even if unused,
2267 until such a lock is revoked. Helps NFS a lot, since NFS is opening
2268 closing files for every read/write openration.
2270 Severity : Enhancement
2272 Description: Cache open negative dentries on client when possible.
2273 Details : Guard negative dentries with UPDATE lock on parent dir, drop
2274 negative dentries on lock revocation.
2279 Description: Remounting a client read-only wasn't possible with a zconf mount
2280 Details : It wasn't possible to remount a client read-only with llmount.
2282 Severity : enhancement
2283 Description: Include MPICH 1.2.6 Lustre ADIO interface patch
2284 Details : In lustre/contrib/ or /usr/share/lustre in RPM a patch for
2285 MPICH is included to add Lustre-specific ADIO interfaces.
2286 This is based closely on the UFS ADIO layer and only differs
2287 in file creation, in order to allow the OST striping to be set.
2288 This is user-contributed code and not supported by CFS.
2293 Description: extended inode attributes (immutable, append-only) work improperly
2294 when 2.4 and 2.6 kernels are used on client/server or vice versa
2295 Details : Introduce kernel-independent values for these flags.
2297 Severity : enhancement
2300 Description: Allow fractional MB tunings for lustre in /proc/ filesystem.
2301 Details : Many of the /proc/ tunables can only be tuned at a megabyte
2302 granularity. Now, Fractional MB granularity is be supported,
2303 this is very useful for low memory system.
2305 Severity : enhancement
2307 Description: Getattr by fid
2308 Details : Getting a file attributes by its fid, obtaining UPDATE|LOOKUP
2309 locks, avoids extra getattr rpc requests to MDS, allows '/' to
2310 have locks and avoids getattr rpc requests for it on every stat.
2313 Frequency : Always, for filesystems larger than 2TB
2315 Description: ldiskfs crash at mount for filesystem larger than 2TB with mballoc
2316 Details : Kenrel kmalloc limits allocations to 128kB and this prevents
2317 filesystems larger than 2TB to be mounted with mballoc enabled.
2320 Frequency : Always, for 32-bit kernel without CONFIG_LBD and filesystem > 2TB
2322 Description: ldiskfs crash at mount for filesystem larger than 2TB with mballoc
2323 Details : If a 32-bit kernel is compiled without CONFIG_LBD enabled and a
2324 filesystems larger than 2TB is mounted then the kernel will
2325 silently corrupt the start of the filesystem. CONFIG_LBD is
2326 enabled for all CFS-supported kernels, but the possibility of
2327 this happening with a modified kernel config exists.
2329 Severity : enhancement
2331 Description: add client O_DIRECT support for 2.6 kernels
2332 Details : It is now possible to do O_DIRECT reads and writes to files
2333 in the Lustre client mountpoint on 2.6 kernel clients.
2335 Severity : enhancement
2337 Description: parallel glimpse, setattr, statfs, punch, destroy requests
2338 Details : Sends glimpse, setattr, statfs, punch, destroy requests to OSTs in
2339 parallel, not waiting for response from every OST before sending
2340 a rpc to the next OST.
2345 Description: setattr vs write race when updating file timestamps
2346 Details : Client processes that update a file timestamp into the past
2347 right after writing to the file (e.g. tar) it is possible that
2348 the updated file modification time can be reset to the current
2349 time due to a race between processing the setattr and write RPC.
2351 Severity : enhancement
2353 Description: Bring 'lfs find' closer in line with regular Linux find.
2354 Details : lfs find util supports -atime, -mtime, -ctime, -maxdepth, -print,
2355 -print0 options and obtains all the needed info through the lustre
2358 Severity : enhancement
2360 Description: support up to 1024 configured devices on one node
2361 Details : change obd_dev array from statically allocated to dynamically
2362 allocated structs as they are first used to reduce memory usage
2367 Description: Flush dirty partially truncated pages during truncate
2368 Details : Immediatelly flush partially truncated pages in filter_setattr,
2369 this way we completely avoid having any pages in page cache on OST
2370 and can retire ugly workarounds during writes to flush such pages.
2375 Description: i_sem vs transaction deadlock in mds_obd_destroy during unlink.
2376 Details : protect inode from truncation within vfs_unlink() context
2377 just take a reference before calling vfs_unlink() and release it
2378 when parent's i_sem is free.
2383 Description: last_id value checked outside lock on OST caused LASSERT failure
2384 Details : If there were multiple MDS->OST object precreate requests in
2385 flight, it was possible that the OST's last object id was checked
2386 outside a lock and incorrectly tripped an assertion. Move checks
2387 inside locks, and discard old precreate requests.
2390 Frequency : always, if extents are used on OSTs
2392 Description: index ei_leaf_hi (48-bit extension) is not zeroed in extent index
2393 Details : OSTs using the extents format would not zero the high 16 bits of
2394 the index physical block number. This is not a problem for any
2395 OST filesystems smaller than 16TB, and no kernels support ext3
2396 filesystems larger than 16TB yet. This is fixed in 1.4.7 (all
2397 new/modified files) and can be fixed for existing filesystems
2398 with e2fsprogs-1.39-cfs1.
2403 Description: import connection selection may be incorrect if timer wraps
2404 Details : Using a 32-bit jiffies timer with HZ=1000 may cause backup
2405 import connections to be ignored if the 32-bit jiffies counter
2406 wraps. Use a 64-bit jiffies counter.
2409 Frequency : very large clusters immediately after boot
2411 Description: LNET request buffers exhausted under heavy short-term load
2412 Details : If a large number of client requests are generated on a service
2413 that has previously never seen so many requests it is possible
2414 that the request buffer growth cannot keep up with the spike in
2415 demand. Instead of dropping incoming requests, they are held in
2416 the LND until the RPC service can accept more requests.
2419 Frequency : Sometimes during replay
2421 Description: Assertion failure in ll_local_open after replay.
2422 Details : If replay happened on an open request reply before we were able
2423 to set replay handler, reply will become not swabbed tripping the
2424 assertion in ll_local_open. Now we set the handler right after
2425 recognising of open request
2428 Frequency : very rare
2430 Description: kernel reports "badness in vsnprintf"
2431 Details : Reading from the "recovery_status" /proc file in small chunks
2432 may cause a negative length in lprocfs_obd_rd_recovery_status()
2433 call to vsnprintf() (which is otherwise harmless). Exit early
2434 if there is no more space in the output buffer.
2436 Severity : enhancement
2438 Description: clear OBD RPC statistics by writing to them
2439 Details : It is now possible to clear the OBD RPC statistics by writing
2440 to the "stats" file.
2445 Description: Client mtime is not the same on different clients after utimes
2446 Details : In some cases, the client was using the utimes() syscall on
2447 a file cached on another node. The clients now validate the
2448 ctime from the MDS + OSTs to determine which one is right.
2453 Description: Inability to activate failout mode
2454 Details : lconf script incorrectly assumed that in pythong string's numeric
2455 value is used in comparisons.
2458 Frequency : always with multiple stripes per file
2460 Description: Inefficient object allocation for mutli-stripe files
2461 Details : When selecting which OSTs to stripe files over, for files with
2462 a stripe count that divides evenly into the number of OSTs,
2463 the MDS is always picking the same starting OST for each file.
2464 Return the OST selection heuristic to the original design.
2469 Description: mount failures may take full timeout to return an error
2470 Details : Under some heavy load conditions it is possible that a
2471 failed mount can wait for the full obd_timeout interval,
2472 possibly several minutes, before reporting an error.
2473 Instead return an error as soon as the status is known.
2475 Frequency : quota enabled and large files being deleted
2477 Description: releasing more than 4GB of quota at once hangs OST
2478 Details : If a user deletes more than 4GB of files on a single OST it
2479 will cause the OST to spin in an infinite loop. Release
2480 quota in < 4GB chunks, or use a 64-bit value for 1.4.7.1+.
2485 Description: statfs data retrieved from /proc may be stale or zero
2486 Details : When reading per-device statfs data from /proc, in the
2487 {kbytes,files}_{total,free,avail} files, it may appear
2488 as zero or be out of date.
2491 Frequency : systems with MD RAID1 external journal devices
2493 Description: lconf's call to blkid is confused by RAID1 journal devices
2494 Details : Use the "blkid -l" flag to locate the MD RAID device instead
2495 of returning all block devices that match the journal UUID.
2498 Frequency : always, for aggregate stripe size over 4GB
2500 Description: assertion fails when trying to use 4GB stripe size
2501 Details : Use "setstripe" to set stripe size over 4GB will fail the kernel,
2502 complaining "ASSERTION(lsm->lsm_xfersize != 0)"
2505 Frequency : always on ppc64
2507 Description: the first write on an ext3 filesystem with mballoc got stuck
2508 Details : ext3_mb_generate_buddy() uses find_next_bit() which does not
2509 perform endianness conversion.
2511 ------------------------------------------------------------------------------
2513 02-14-2006 Cluster File Systems, Inc. <info@clusterfs.com>
2515 * WIRE PROTOCOL CHANGE. This version of Lustre networking WILL NOT
2516 INTEROPERATE with older versions automatically. Please read the
2517 user documentation before upgrading any part of a live system.
2518 * WARNING: Lustre networking configuration changes are required with
2519 this release. See https://bugzilla.clusterfs.com/show_bug.cgi?id=10052
2522 * Support for kernels:
2523 2.6.9-22.0.2.EL (RHEL 4)
2524 2.6.5-7.244 (SLES 9)
2525 2.6.12.6 vanilla (kernel.org)
2528 Severity : enhancement
2529 Bugzilla : 7981/8208
2530 Description: Introduced Lustre Networking (LNET)
2531 Details : LNET is new networking infrastructure for Lustre, it includes
2532 a reorganized network configuration mode (see the user
2533 documentation for full details) as well as support for routing
2534 between different network fabrics. Lustre Networking Devices
2535 (LNDs) for the supported network fabrics have also been
2536 created for this new infrastructure.
2538 Severity : enhancement
2539 Description: Introduced Access control lists
2540 Details : clients can set ACLs on files and directories in order to have
2541 more fine-grained permissions than the standard Unix UGO+RWX.
2542 The MDS must be started with the "-o acl" mount option.
2544 Severity : enhancement
2545 Description: Introduced filesystem quotas
2546 Details : Administrators may now establish per-user quotas on the
2549 Severity : enhancement
2551 Description: Configuration change for the XT3
2552 The PTLLND is now used to run Lustre over Portals on the XT3
2553 The configure option(s) --with-cray-portals are no longer used.
2554 Rather --with-portals=<path-to-portals-includes> is used to
2555 enable building on the XT3. In addition to enable XT3 specific
2556 features the option --enable-cray-xt3 must be used.
2561 Description: Running on many-way SMP OSTs can trigger oops in llcd_send()
2562 Details : A race between allocating a new llcd and re-getting the llcd_lock
2563 allowed another thread to grab newly-allocated llcd.
2565 Severity : enhancement
2567 Description: 2.6 OST async journal commit and locking fix to improve performance
2568 Details : The filter_direct_io()+filter_commitrw_write() journal commits for
2569 2.6 kernels are now async as they already were in 2.4 kernels so
2570 that they can commit concurrently with the network bulk transfer.
2571 For block-allocated files the filter allocation semaphore is held
2572 to avoid filesystem fragmentation during allocation. BKL lock
2573 removed for 2.6 xattr operations where it is no longer needed.
2578 Description: lconf incorrectly determined whether two IP networks could talk
2579 Details : In some more complicated routing and multiple-network
2580 configurations, lconf will avoid trying to make a network
2581 connection to a disjoint part of the IP space. It was doing the
2582 math incorrectly for one set of cases.
2587 Description: Fix for potential infinite loop processing records in an llog.
2588 Details : If an llog record is corrupted/zeroed, it is possible to loop
2589 forever in llog_process(). Validate the llog record length
2590 and skip the remainder of the block on error.
2593 Frequency : occasional (liblustre only)
2595 Description: liblustre could not open files whose last component is a symlink
2596 Details : sysio_path_walk() would incorrectly pass the open intent to
2597 intermediate path components.
2600 Frequency : rare (liblustre only with non-standard tuning)
2601 Bugzilla : 7201 (7350)
2602 Description: Tuning the MDC DLM LRU size to zero triggers client LASSERT
2603 Details : llu_lookup_finish_locks() tries to set lock data on a lock
2604 after it has been released, only do this for referenced locks
2606 Severity : enhancement
2608 Description: specifying an (invalid) directory default stripe_size of -1
2609 would reset the directory default striping
2610 Details : stripe_size -1 was used internally to signal directory stripe
2611 removal, now use "all default" to signal dir stripe removal
2612 as a directory striping of "all default" is not useful
2615 Frequency : common for large clusters running liblustre clients
2617 Description: doing an ls when liblustre clients are running is slow
2618 Details : sending a glimpse AST to a liblustre client waits for every AST
2619 to time out, as liblustre clients will not respond. Since they
2620 cannot cache data we refresh the OST lock LVB from disk instead.
2622 Severity : enhancement
2624 Description: doing an ls at the same time as file IO can be slow
2625 Details : enqueue and other "small" requests can be blocked behind many
2626 large IO requests. Create a new OST IO portal for non-IO
2627 requests so they can be processed faster.
2630 Frequency : rare (only HPUX clients mounting unsupported re-exported NFS vol)
2632 Description: an HPUX NFS client would get -EACCESS when ftruncate()ing a newly
2633 created file with mode 000
2634 Details : the Linux NFS server relies on an MDS_OPEN_OWNEROVERRIDE hack to
2635 allow an ftruncate() as a non-root user to a file with mode 000.
2636 Lustre now respects this flag to disable mode checks when
2637 truncating a file owned by the user
2640 Frequency : liblustre-only, when liblustre client dies unexpectedly or becomes
2643 Description: Revoking locks from clients that went dead or catatonic might take
2645 Details : New lock flags FL_CANCEL_ON_BLOCK used by liblustre makes
2646 cancellation of such locks instant on servers without waiting for
2647 any reply from clients. Clients drops these locks when cancel
2648 notification from server is received without replying.
2651 Frequency : liblustre-only, when liblustre client dies or becomes busy
2653 Description: Doing ls on Linux clients can take a long time with active
2655 Details : Liblustre client cannot handle ASTs in timely manner, so avoid
2656 granting such locks to it in the first place if possible. Locks
2657 are taken by proxy on the OST during the read or write and
2658 dropped immediately afterward. Add connect flags handling, do
2659 not grant locks to liblustre clients for glimpse ASTs.
2661 Severity : enhancement
2663 Description: Improve read-ahead algorithm to avoid excessive IO for random reads
2664 Details : Existing read-ahead algorithm is tuned for the case of streamlined
2665 sequential reads and behaves badly with applications doing random
2666 reads. Improve it by reading ahead at least read region, and
2667 avoiding excessive large RPC for small reads.
2669 Severity : enhancement
2671 Description: Creating more than 1000 files for a single job may cause a load
2672 imbalance on the OSTs if there are also a large number of OSTs.
2673 Details : qos_prep_create() uses an OST index reseed value that is an
2674 even multiple of the number of available OSTs so that if the
2675 reseed happens in the middle of the object allocation it will
2676 still utilize the OSTs as uniformly as possible.
2681 Description: OST or MDS may oops in ping_evictor_main()
2682 Details : ping_evictor_main() drops obd_dev_lock if deleting a stale export
2683 but doesn't restart at beginning of obd_exports_timed list
2686 Severity : enhancement
2688 Description: improve by-nid export eviction on the MDS and OST
2689 Details : allow multiple exports with the same NID to be evicted at one
2690 time without re-searching the exports list.
2693 Frequency : rare, only with supplementary groups enabled on SMP 2.6 kernels
2695 Description: MDS may oops in groups_free()
2696 Details : in rare race conditions a newly allocated group_info struct is
2697 freed again, and this can be NULL. The 2.4 compatibility code
2698 for groups_free() checked for a NULL pointer, but 2.6 did not.
2701 Frequency : common for liblustre clients doing little filesystem IO
2702 Bugzilla : 9352, 7313
2703 Description: server may evict liblustre clients accessing contended locks
2704 Details : if a client is granted a lock or receives a completion AST
2705 with a blocking AST already set it would not reply to the AST
2706 for LDLM_FL_CANCEL_ON_BLOCK locks. It now replies to such ASTs.
2709 Frequency : lfs setstripe, only systems with more than 160 OSTs
2711 Description: unable to set striping with a starting offset beyond OST 160
2712 Details : llapi_create_file() incorrectly limited the starting stripe
2713 index to the maximum single-file stripe count.
2716 Frequency : LDAP users only
2718 Description: lconf did not handle in-kernel recovery with LDAP properly
2719 Details : lconf/LustreDB get_refs() is searching the wrong namespace
2721 Severity : enhancement
2723 Description: bind OST threads to NUMA nodes to improve performance
2724 Details : all OST threads are uniformly bound to CPUs on a single NUMA
2725 node and do their allocations there to localize memory access
2727 Severity : enhancement
2729 Description: llmount can determine client NID directly from Myrinet (GM)
2730 Details : the client NID code from gmnalnid was moved directly into
2731 llmount, removing the need to use this or specifying the
2732 client NID explicitly when mounting GM clients with zeroconf
2735 Frequency : if client is started with down MDS
2737 Description: if client is started with down MDS mount hangs in ptlrpc_queue_wait
2738 Details : Having an LWI_INTR() wait event (interruptible, but no timeout)
2739 will wait indefinitely in ptlrpc_queue_wait->l_wait_event() after
2740 ptlrpc_import_delayed_req() because we didn't check if the
2741 request was interrupted, and we also didn't break out of the
2742 event loop if there was no timeout
2747 Description: data loss during non-page-aligned writes to a single file from
2748 both multiple nodes and multiple threads on one node at same time
2749 Details : updates to KMS and lsm weren't protected by common lock. Resulting
2750 inconsistency led to false short-reads, that were cached and later
2751 used by ->prepare_write() to fill in partially written page,
2752 leading to data loss.
2755 Frequency : always, if lconf --abort_recovery used
2757 Description: lconf --abort_recovery fails with 'Operation not supported'
2758 Details : lconf was attempting to abort recovery on the MDT device and not
2761 Severity : enhancement
2763 Description: remove cleanup logs
2764 Details : replace lconf-generated cleanup logs with lustre internal
2765 cleanup routines. Eliminates the need for client-cleanup and
2768 Severity : enhancement
2770 Description: add support for EAs (user and system) on lustre filesystems
2771 Details : it is now possible to store extended attributes in the Lustre
2772 client filesystem, and with the user_xattr mount option it
2773 is possible to allow users to store EAs on their files also
2775 Severity : enhancement
2777 Description: Add possibility (config option) to show minimal available OST free
2779 Details : When compiled with --enable-mindf configure option, statfs(2)
2780 (and so, df) will return least minimal free space available from
2781 all OSTs as amount of free space on FS, instead of summary of
2782 free spaces of all OSTs.
2784 Severity : enhancement
2786 Description: do not expand extent locks acquired on OST-side
2787 Details : Modify ldlm_extent_policy() to not expand local locks, acquired
2788 by server: they are not cached anyway.
2791 Frequency : when mmap is used/binaries executed from Lustre
2793 Description: Unmmap pages before throwing them away from read cache.
2794 Details : llap_shrink cache now attempts to unmap pages before discarding
2795 them (if unmapping failed - do not discard). SLES9 kernel has
2796 extra checks that trigger if this unmapping is not done first.
2801 Description: lconf didn't resolve symlinks before checking to see whether a
2802 given mountpoint was already in use
2805 Frequency : when migrating failover services
2806 Bugzilla : 6395, 9514
2807 Description: When migrating a subset of services from a node (e.g. failback
2808 from a failover service node) the remaining services would
2809 time out and evict clients.
2810 Details : lconf --force (implied by --failover) sets the global obd_timeout
2811 to 5 seconds in order to quickly disconnect, but this caused
2812 other RPCs to time out too quickly. Do not change the global
2813 obd_timeout for force cleanup, only set it for DISCONNECT RPCs.
2815 Severity : enhancement
2816 Frequency : if MDS is started with down OST
2817 Bugzilla : 9439,5706
2818 Description: Allow startup/shutdown of an MDS without depending on the
2819 availability of the OSTs.
2820 Details : Asynchronously call mds_lov_synchronize during MDS startup.
2821 Add appropriate locking and lov-osc refcounts for safe
2822 cleaning. Add osc abort_inflight calls in case the
2823 synchronize never started.
2826 Frequency : occasional (Cray XT3 only)
2828 Description: root not authorized to access files in CRAY_PORTALS environment
2829 Details : The client process capabilities were not honoured on the MDS in
2830 a CRAY_PORTALS/CRAY_XT3 environment. If the file had previously
2831 been accessed by an authorized user then root was able to access
2832 the file on the local client also. The root user capabilities
2833 are now allowed on the MDS, as this environment has secure UID.
2836 Frequency : occasional
2838 Description: ldiskfs "too long searching" message happens too often
2839 Details : A debugging message (otherwise harmless) prints too often on
2840 the OST console. This has been reduced to only happen when
2841 there are fragmentation problems on the filesystem.
2846 Description: Division by zero in statfs when all OSCs are inactive
2847 Details : lov_get_stripecnt() returns zero due to incorrect order of checks,
2848 lov_statfs divides by value returned by lov_get_stripecnt().
2852 Bugzilla : 9489, 3273
2853 Description: First write from each client to each OST was only 4kB in size,
2854 to initialize client writeback cache, which caused sub-optimal
2855 RPCs and poor layout on disk for the first writen file.
2856 Details : Clients now request an initial cache grant at (re)connect time
2857 and so that they can start streaming writes to the cache right
2858 away and always do full-sized RPCs if there is enough data.
2859 If the OST is rebooted the client also re-establishes its grant
2860 so that client cached writes will be honoured under the grant.
2865 Description: Slow ls (and stat(2) syscall) on files residing on IO-loaded OSTs
2866 Details : Now I/O RPCs go to different portal number and (presumably) fast
2867 lock requests (and glimses) and other RPCs get their own service
2868 threads pool that should be able to service those RPCs
2871 Severity : enhancement
2873 Description: Ability to exchange lustre version between client and servers and
2874 issue warnings at client side if client is too old. Also for
2875 liblustre clients there is ability to refuse connection of too old
2877 Details : New 'version' field is added to connect data structure that is
2878 filled with version info. That info is later checked by server and
2882 Frequency : rare, liblustre only.
2883 Bugzilla : 9296, 9581
2884 Description: Two simultaneous writes from liblustre at offset within same page
2885 might proceed at the same time overwriting eachother with stale
2887 Details : I/O lock withing llu_file_prwv was released too early, before data
2888 actually was hitting the wire. Extended lock-holding time until
2889 server acknowledges receiving data.
2892 Frequency : extremely rare. Never observed in practice.
2894 Description: avoid generating lustre_handle cookie of 0.
2895 Details : class_handle_hash() generates handle cookies by incrementing
2896 global counter, and can hit 0 occasionaly (this is unlikely, but
2897 not impossible, because initial value of cookie counter is
2898 selected randonly). Value of 0 is used as a sentinel meaning
2899 "unassigned handle" --- avoid it. Also coalesce two critical
2900 sections in this function into one.
2902 Severity : enhancement
2904 Description: allow liblustre clients to delegate truncate locking to OST
2905 Details : To avoid overhead of locking, liblustre client instructs OST to
2906 take extent lock in ost_punch() on client's behalf. New connection
2907 flag is added to handle backward compatibility.
2909 Severity : enhancement
2910 Bugzilla : 4928, 7341, 9758
2911 Description: allow number of OST service threads to be specified
2912 Details : a module parameter allows the number of OST service threads
2913 to be specified via "options ost ost_num_threads={N}" in the
2914 OSS's /etc/modules.conf or /etc/modprobe.conf.
2918 Bugzilla : 6146, 9635, 9895
2919 Description: servers crash with bad pointer in target_handle_connect()
2920 Details : In rare cases when a client is reconnecting it was possible that
2921 the connection request was the last reference for that export.
2922 We would temporarily drop the export reference and get a new
2923 one, but this may have been the last reference and the export
2924 was just destroyed. Get new reference before dropping old one.
2926 Severity : enhancement
2927 Frequency : if client is started with failover MDS
2929 Description: Allow multiple MDS hostnames in the mount command
2930 Details : Try to read the configuration from all specified MDS
2931 hostnames during a client mount in case the "primary"
2934 Severity : enhancement
2936 Description: Stop sending data to evicted clients as soon as possible.
2937 Details : Check if the client we are about to send or are sending data to
2938 was evicted already. (Check is done every second of waiting,
2939 for which l_wait_event interface was extended to allow checking
2940 of exit condition at specified intervals).
2943 Frequency : rare, normally only when NFS exporting is done from client
2945 Description: 'bad disk LOV MAGIC: 0x00000000' error when chown'ing files
2947 Details : Make mds_get_md() recognise empty md case and set lmm size to 0.
2950 Frequency : always, if srand() is called before liblustre initialization
2952 Description: Liblustre uses system PRNG disturbing its usage by user application
2953 Details : Introduce internal to lustre fast and high-quality PRNG for
2954 lustre usage and make liblustre and some other places in generic
2955 lustre code to use it.
2957 Severity : enhancement
2958 Bugzilla : 9477, 9557, 9870
2959 Description: Verify that the MDS configuration logs are updated when xml is
2960 Details : Check if the .xml configuration logs are newer than the config
2961 logs stored on the MDS and report an error if this is the case.
2962 Request --write-conf, or allow starting with --old_conf.
2964 Severity : enhancement
2966 Description: Handle symlinks in the path when checking if Lustre is mounted.
2967 Details : Resolve intermediate symlinks when checking if a client has
2968 mounted a filesystem to avoid duplicate client mounts.
2973 Description: lconf can hit an error exception but still return success.
2974 Details : The lconf command catches the Command error exception at the top
2975 level script context and will exit with the associated exit
2976 status, but doesn't ensure that this exit status is non-zero.
2981 Description: failure of ptlrpc thread startup can cause oops
2982 Details : Starting a ptlrpc service thread can fail if there are a large
2983 number of threads or the server memory is very fragmented.
2984 Handle this without oopsing.
2987 Frequency : always, only if liblustre and non-default acceptor port was used
2989 Description: liblustre cannot connect to servers with non-default acceptor port
2990 Details : tcpnal_set_default_params() was not called and was therefore
2991 ignoring the environment varaible TCPNAL_PORT, as well as other
2992 TCPNAL_ environment variables
2997 Description: two objects could be created on the same OST for a single file
2998 Details : If an OST is down, in some cases it was possible to create two
2999 objects on a single OST for a single file. No problems other
3000 than potential performance impact and spurious error messages.
3004 Bugzilla : 5681, 9562
3005 Description: Client may oops in ll_unhash_aliases
3006 Details : Client dcache may become inconsistent in race condition.
3007 In some cases "getcwd" can fail if the current directory is
3013 Description: Inode refcounting problems in NFS export code
3014 Details : link_raw functions used to call d_instantiate without obtaining
3015 extra inode reference first.
3019 Bugzilla : 9942, 9903
3020 Description: Referencing freed requests leading to crash, memleaks with NFS.
3021 Details : We used to require that call to ll_revalidate_it was always
3022 followed by ll_lookup_it. Also with revalidate_special() it is
3023 possible to call ll_revalidate_it() twice for the same dentry
3024 even if first occurence returned success. This fix changes semantic
3025 between DISP_ENQ_COMPLETE disposition flag to mean there is extra
3026 reference on a request referred from the intent.
3027 ll_intent_release() then releases such a request.
3030 Frequency : rare, normally benchmark loads only
3032 Description: unlinked inodes were kept in memory on the client
3033 Details : If a client is repeatedly creating and unlinking files it
3034 can accumulate a lot of stale inodes in the inode slab cache.
3035 If there is no other client load running this can cause the
3036 client node to run out of memory. Instead flush old inodes
3037 from client cache that have the same inode number as a new inode.
3040 Frequency : SLES9 2.6.5 kernel and long filenames only
3041 Bugzilla : 9969, 10379
3042 Description: utime reports stale NFS file handle
3043 Details : SLES9 uses out-of-dentry names in some cases, which confused
3044 the lustre dentry revalidation. Change it to always use the
3048 Frequency : rare, unless heavy write-truncate concurrency is continuous
3049 Bugzilla : 4180, 6984, 7171, 9963, 9331
3050 Description: OST becomes very slow and/or deadlocked during object unlink
3051 Details : filter_destroy() was holding onto the parent directory lock
3052 while truncating+unlinking objects. For very large objects this
3053 may block other threads for a long time and slow overall OST
3054 responsiveness. It may also be possible to get a lock ordering
3055 deadlock in this case, or run out of journal credits because of
3056 the combined truncate+unlink. Solution is to do object truncate
3057 first in one transaction without parent lock, and then do the
3058 final unlink in a new transaction with the parent lock. This
3059 reduces the lock hold time dramatically.
3062 Frequency : rare, 2.4 kernels only
3064 Description: MDS or OST cleanup may trip kernel BUG when dropping kernel lock
3065 Details : mds_cleanup() and filter_cleanup() need to drop the kernel lock
3066 before unmounting their filesystem in order to avoid deadlock.
3067 The kernel_locked() function in 2.4 kernels only checks whether
3068 the kernel lock is held, not whether it is this process that is
3069 holding it as 2.6 kernels do.
3074 Description: MDS or OST may oops/LBUG if a client is connecting multiple times
3075 Details : The client ptlrpc code may be trying to reconnect to a down
3076 server before a previous connection attempt has timed out.
3077 Increase the reconnect interval to be longer than the connection
3078 timeout interval to avoid sending duplicate connections to
3082 Frequency : echo_client brw_test command
3084 Description: fix echo_client to work with OST preallocated code
3085 Details : OST preallocation code (5137) didn't take echo_client IO path
3086 into account: echo_client calls filter methods outside of any
3087 OST thread and, hence, there is no per-thread preallocated
3088 pages and buffers to use. Solution: hijack pga pages for IO. As
3089 a byproduct, this avoids unnecessary data copying.
3093 Bugzilla : 3555, 5962, 6025, 6155, 6296, 9574
3094 Description: Client can oops in mdc_commit_close() after open replay
3095 Details : It was possible for the MDS to return an open request with no
3096 transaction number in mds_finish_transno() if the client was
3097 evicted, but without actually returning an error. Clients
3098 would later try to replay that open and may trip an assertion
3099 Simplify the client close codepath, and always return an error
3100 from the MDS in case the open is not successful.
3103 Frequency : rare, 2.6 OSTs only
3105 Description: OST may deadlock under high load on fragmented files
3106 Details : If there was a heavy load and highly-fragmented OST filesystems
3107 it was possible to have all the OST threads deadlock waiting on
3108 allocation of biovecs, because the biovecs were not released
3109 until the entire RPC IO was completed. Instead, release biovecs
3110 as soon as they are complete to ensure forward IO progress.
3112 Severity : enhancement
3114 Description: Support for specifying external journal device at mount
3115 Details : If an OST or MDS device is formatted with an external journal
3116 device, this device major/minor is stored in the ext3 superblock
3117 and may not be valid for failover. Allow detecting and
3118 specifying the external journal at mount time.
3123 Description: Mounting an MDS with pending unlinked files may cause oops
3124 Details : target_finish_recovery() calls mds_postrecov() which returned
3125 the number of orphans unlinked. mds_lov_connect->mds_postsetup()
3126 considers this an error and immediately begins cleaning up the
3127 lov, just after starting the mds_lov process
3129 Severity : enhancement
3131 Description: Implement 'lfs df' to report actual free space on per-OST basis
3132 Details : Add sub-command 'df' on 'lfs' to report the disk space usage of
3133 MDS/OSDs. Usage: lfs df [-i][-h]. Command Options: '-i' to report
3134 usage of objects; '-h' to report in human readable format.
3136 ------------------------------------------------------------------------------
3138 08-26-2005 Cluster File Systems, Inc. <info@clusterfs.com>
3145 Description: Mounting an ldiskfs file system with mballoc may crash OST node.
3146 Details : ldiskfs mballoc code may reference an uninitialized buddy struct
3147 at startup during orphan unlinking. Instead, skip buddy update
3148 before setup, as it will be regenerated after recovery is complete.
3153 Description: If an OST is inactive, its locks might reference stale inodes.
3154 Details : lov_change_cbdata() must iterate over all namespaces, even if
3155 they are inactive to clear inode references from the lock.
3157 Severity : enhancement
3158 Frequency : occasional, if non-standard max_dirty_mb used
3160 Description: Client will block write RPCs if not enough grant
3161 Details : If a client has max_dirty_mb smaller than max_rpcs_in_flight,
3162 then the client will block writes while waiting for another RPC
3163 to complete instead of consuming its dirty limit. With change
3164 we get improved performance when max_dirty_mb is small.
3166 Severity : enhancement
3167 Bugzilla : 3389, 6253
3168 Description: Add support for supplementary groups on the MDS.
3169 Details : The MDS has an upcall /proc/fs/lustre/mds/{mds}/group_upcall
3170 (set to /usr/sbin/l_getgroups if enabled) which will do MDS-side
3171 lookups for user supplementary groups into a cache.
3175 Description: O_CREAT|O_EXCL open flags in liblustre always return -EEXIST
3176 Details : Make libsysio to not enforce O_EXCL by clearing the flag,
3177 for liblustre O_EXCL is enforced by MDS.
3181 Description: readdir never returns NULL in liblustre.
3182 Details : Corrected llu_iop_getdirentries logic, to return offset of next
3183 dentry in struct dirent.
3187 Frequency : liblustre only, depends on application IO pattern
3188 Description: liblustre clients evicted if not contacting servers
3189 Details : Don't put liblustre clients into the ping_evictor list, so
3190 they will not be evicted by the pinger ever.
3192 Severity : enhancement
3194 Description: Add ability to evict clients by NID from MDS.
3195 Details : By echoing "nid:$NID" string into
3196 /proc/fs/lustre/mds/.../evict_client client with nid that equals to
3197 $NID would be instantly evicted from this MDS and from all active
3198 OSTs connected to it.
3202 Description: Do not query file size twice, somewhat slowing stat(2) calls.
3203 Details : lookup_it_finish() used to query file size from OSTs that was not
3208 Description: service threads change working directory to that of init
3209 Details : Starting lustre service threads may pin the working directory
3210 of the parent thread, making that filesystem busy. Threads
3211 now change to the working directory of init to avoid this.
3215 Frequency : during shutdown only
3216 Description: shutdown with a failed MDS or OST can cause unmount to hang
3217 Details : Don't resend DISCONNECT messages in ptlrpc_disconnect_import()
3222 Frequency : 2.6 only
3223 Description: chmod/chown may include an extra supplementary group
3224 Details : ll{,u}_mdc_pack_op_data() does not properly initialize the
3225 supplementary group and if none is specified this is used.
3228 Bugzilla : 5479 (6816)
3230 Description: Racing open + rm can assert client in mdc_set_open_replay_data()
3231 Details : If lookup is in progress on a file that is unlinked we might try
3232 to revalidate the inode and fail in revalidate after lookup is
3233 complete and ll_file_open() enqueues the open again but
3234 it_open_error() was not checking DISP_OPEN_OPEN errors correctly.
3237 Frequency : always, if lconf --abort_recovery used
3239 Description: lconf --abort_recovery fails with 'Operation not supported'
3240 Details : lconf was attempting to abort recovery on the MDT device and not
3243 ------------------------------------------------------------------------------
3245 2005-08-08 Cluster File Systems, Inc. <info@clusterfs.com>
3250 Frequency : rare (only unsupported configurations with a node running as an
3252 Bugzilla : 6514, 5137
3253 Description: Mounting a Lustre file system on a node running as an OST could
3255 Details : OSTs now preallocates memory needed to write out data at
3256 startup, instead of when needed, to avoid having to
3257 allocate memory in possibly low memory situations.
3258 Specifically, if the file system is mounted on on OST,
3259 memory pressure could force it to try to write out data,
3260 which it needed to allocate memory to do. Due to the low
3261 memory, it would be unable to do so and the node would
3262 become unresponsive.
3264 Severity : enhancement
3266 Description: Addition of lconf --service command line option
3267 Details : lconf now accepts a '--service <arg>' option, which is
3268 shorthand for 'lconf --group <arg> --select <arg>=<hostname>'
3270 Severity : enhancement
3272 Description: Failover mode is now the default for OSTs.
3273 Details : By default, OSTs will now run in failover mode. To return to
3274 the old behaviour, add '--failout' to the lmc line for OSTs.
3276 Severity : enhancement
3278 Description: Health checks are now provided for MDS and OSTs
3279 Details : Additional detailed health check information on MSD and OSTs
3280 is now provided through the procfs health_check value.
3283 Frequency : occasional, depends on IO load
3285 Description: Disk fragmentation on the OSTs could eventually cause slowdowns
3286 after numerous create/delete cycles
3287 Details : The ext3 inode allocation policy would not allocate new inodes
3288 very well on the OSTs because there are no new directories
3289 being created. Instead we look for groups with free space if
3290 the parent directories are nearly full.
3295 Description: Network or server problems during mount may cause partially
3296 mounted clients instead of returning an error.
3297 Details : The config llog parsing code may overwrite the error return
3298 code during mount error handling, returning success instead
3304 Description: MDS can fail to allocate large reply buffers
3305 Details : After long uptimes the MDS can fail to allocate large reply
3306 buffers (e.g. zconf client mount config records) due to memory
3307 fragmentation or consumption by the buffer cache. Preallocate
3308 some large reply buffers so that these replies can be sent even
3309 under memory pressure.
3313 Frequency : rare (liblustre)
3314 Description: fsx running with liblustre complained that using truncate() to
3315 extend the file doesn't work. This patch corrects that issue.
3316 Details : This is the liblustre equivalent of the fix for bug 6196. Fixes
3317 ATTR_SIZE and lsm use in llu_setattr_raw.
3321 Frequency : rare, only 2.6 kernels
3322 Description: Unusual file access patterns on the MDS may result in inode
3323 data being lost in very rare circumstances.
3324 Details : Bad interaction between the ea-in-inode patch and the "no-read"
3325 code in the 2.6 kernel caused the inode and/or EA data not to
3326 be read from disk, causing single-file corruption.
3330 Frequency : rare, only 2.6 filesystems using extents
3331 Description: Heavy concurrent write and delete load may cause data corruption.
3332 Details : It was possible under high-load situations to have an extent
3333 metadata block in the block device cache from a just-unlinked
3334 file overwrite a newly-allocated data block. We now unmap any
3335 metadata buffers that alias just-allocated data blocks.
3339 Frequency : filesystems with default stripe_count larger than 77
3340 Description: lconf+mke2fs fail when formatting filesystem with > 77 stripes
3341 Details : lconf specifies an inode size of 4096 bytes when the default
3342 stripe_count is larger than 77. This conflicts with the default
3343 inode density of 1 per 4096 bytes. Allocate smaller inodes in
3344 this case to avoid pinning too much memory for large EAs.
3346 ------------------------------------------------------------------------------
3348 2005-07-07 Cluster File Systems, Inc. <info@clusterfs.com>
3353 Frequency : rare (extremely heavy IO load with hundreds of clients)
3355 Description: Client is evicted, gets IO error writing to file
3356 Details : lock ordering changes for bug 5492 reintroduced bug 3267 and
3357 caused clients to be evicted for AST timeouts. The fixes in
3358 bug 5192 mean we no longer need to have such short AST timeouts
3359 so ldlm_timeout has been increased.
3362 Frequency : occasional during --force or --failover shutdown under load
3363 Bugzilla : 5949, 4834
3364 Description: Server oops/LBUG if stopped with --force or --failover under load
3365 Details : a collection of import/export refcount and cleanup ordering
3366 issues fixed for safer force cleanup
3369 Frequency : only filesystems larger than 120 OSTs
3370 Bugzilla : 5990, 6223
3371 Description: lfs getstripe would oops on a very large filesystem
3372 Details : lov_getconfig used kfree on vmalloc'd memory
3375 Frequency : only filesystems exporting via NFS to Solaris 10 clients
3376 Bugzilla : 6242, 6243
3377 Description: reading from files that had been truncated to a non-zero size
3378 but never opened returned no data
3379 Details : ll_file_read() reads zeros from no-object files to EOF
3384 Description: A bug in MDS/OSS recovery could cause the OSS to fail an assertion
3385 Details : There's little harm in aborting MDS/OSS recovery and letting it
3386 try again, so I removed the LASSERT and return an error instead.
3388 Severity : enhancement
3390 Description: New debugging infrastructure for tracking down data corruption
3391 Details : The I/O checksum code was replaced to: (a) control it at runtime,
3392 (b) cover more of the client-side code path, and (c) try to narrow
3393 down where problems occurred
3397 Bugzilla : 3819, 4364, 4397, 6313
3398 Description: Racing close and eviction MDS could cause assertion in mds_close
3399 Details : It was possible to get multiple mfd references during close and
3400 client eviction, leading to one thread referencing a freed mfd.
3402 Severity: : enhancement
3403 Bugzilla : 3262, 6359
3404 Description: Attempts to reconnect to servers are now more aggressive.
3405 Details : This builds on the enhanced upcall-less recovery that was added
3406 in 1.4.2. When trying to reconnect to servers, clients will
3407 now try each server in the failover group every 10 seconds. By
3408 default, clients would previously try one server every 25 seconds.
3413 Description: After recovery, certain operations trigger a failed
3414 assertion on a client.
3415 Details : Failing over an mds, using lconf -d --failover, while a
3416 client was doing a readdir() call would cause the client to
3417 LBUG after recovery completed and the readdir() was resent.
3419 Severity : enhancement
3421 Description: Default groups are now added by lconf
3422 Details : You can now run lconf --group <servicename> without having to
3423 manually add groups with lmc.
3426 Frequency : occasional
3428 Description: Nodes with an elan id of 0 trigger a failed assertion
3431 Frequency : always when accessing e.g. tty/console device nodes
3433 Description: tty and some other devices nodes cannot be used on lustre
3434 Details : file's private_data field is used by device data and lustre
3435 values in there got lost. New field was added to struct file to
3436 store fs-specific private data.
3439 Frequency : when exporting Lustre via NFS
3441 Description: NFSD failed occasionally when looking up a path component
3442 Details : NFSD is looking up ".." which was broken in ext3 directories
3443 that had grown large enough to become hashed.
3446 Frequency : Clusters with multiple interfaces not on the same subnet
3448 Description: Nodes will repeatedly try to reconnect to an interface which it
3449 cannot reach and report an error to the log.
3450 Details : Extra peer list entries will be created by lconf with some peers
3451 unreachable. lconf now validates the peer before adding it.
3454 Frequency : Only if a default stripe is set on the filesystem root.
3456 Description: Setting a default stripe on the filesystem root prevented the
3457 filesystem from being remounted.
3458 Details : The client was sending extra request flags in the root getattr
3459 request and did not allocate a reply buffer for the dir EA.
3462 Frequency : occasional, higher if lots of files are accessed by one client
3463 Bugzilla : 6159, 6097
3464 Description: Client trips assertion regarding lsm mismatch/magic
3465 Details : While revalidating inodes the VFS looks up inodes with ifind()
3466 and in rare cases can find an inode that is being freed.
3467 The ll_test_inode() code will free the lsm during ifind()
3468 when it finds an existing inode and then the VFS later attaches
3469 this free lsm to a new inode.
3473 Bugzilla : 6422, 7030
3474 Description: MDS deadlock between mkdir and client eviction
3475 Details : Creating a new file via mkdir or mknod (starting a transaction
3476 and getting the ns lock) can deadlock with client eviction
3477 (gets ns lock and trying to finish a synchronous transaction).
3480 Frequency : occasional
3481 Description: While starting a server, the fsfilt_ext3 module could not be
3483 Details : CFS's improved ext3 filesystem is named ldiskfs for 2.6
3484 kernels. Previously, lconf would still use the ext3 name
3485 when trying to load modules. Now, it will correctly use
3486 ext3 on 2.4 and ldiskfs on 2.6.
3488 Severity : enhancement
3489 Description: The default stripe count has been changed to 1
3490 Details : The interpretation of the default stripe count (0, to lfs
3491 or lmc) has been changed to mean striping across a single
3492 OST, rather than all available. For general usage we have
3493 found a stripe count of 1 or 2 works best.
3495 Severity : enhancement
3496 Description: Add support for compiling against Cray portals.
3497 Details : Conditional compiling for some areas that are different
3501 Frequency : occasional
3502 Bugzilla : 6409, 6834
3503 Description: Creating files with an explicit stripe count may lead to
3504 a failed assertion on the MDS
3505 Details : If some OSTs are full or unavailable, creating files may
3506 trigger a failed assertion on the MDS. Now, Lustre will
3507 try to use other servers or return an error to the
3511 Frequency : occasional
3513 Description: Multiple concurrent overlapping read+write on multiple SMP nodes
3514 caused lock timeout during readahead (since 1.4.2).
3515 Details : Processes doing readahead might match a lock that hasn't been
3516 granted yet if there are overlapping and conflicting lock
3517 requests. The readahead process waits on ungranted lock
3518 (original lock is CBPENDING), while OST waits for that process
3519 to cancel CBPENDING read lock and eventually evicts client.
3521 Severity : enhancement
3523 Description: Initial enabling of flock support for clients
3524 Details : Implements fcntl advisory locking and file status functions.
3525 This feature is provided as an optional mount flag (default
3526 off), and is NOT CURRENTLY SUPPORTED. Not all types of record
3527 locking are implemented yet, and those that are are not guaranteed
3528 to be completely correct in production environments.
3529 mount -t lustre -o [flock|noflock] ...
3532 Frequency : occasional
3534 Description: OSTs running 2.4 kernels but with extents enabled might trip an
3535 assertion in the ext3 JBD (journaling) layer.
3536 Details : The b_committed_data struct is protected by the big kernel lock
3537 in 2.4 kernels, serializing journal_commit_transaction() and
3538 ext3_get_block_handle->ext3_new_block->find_next_usable_block()
3539 access to this struct. In 2.6 kernels there is finer grained
3540 locking to improve SMP performance of the JBD layer.
3544 Description: Changes the "SCSI I/O Stats" kernel patch to default to "enabled"
3549 Description: merge and cleanup kernel patches.
3552 -----------------------------------------------------------------------------
3554 2005-05-05 Cluster File Systems, Inc. <info@clusterfs.com>
3556 NOTE: Lustre 1.4.2 uses an incompatible network protocol than previous
3557 versions of Lustre. Please update all servers and clients to
3558 version 1.4.2 or later at the same time. You must also run
3559 "lconf --write-conf {config}.xml" on the MDS while it is stopped
3560 to update the configuration logs.
3562 - fix for HPUX NFS client breakage when NFS exporting Lustre (5781)
3563 - mdc_enqueue does not need max_mds_easize request buffer on send (5707)
3564 - swab llog records of type '0' so we get proper header size/idx (5861)
3565 - send llog cancel req to DLM cancel portal instead of cb portal (5515)
3566 - fix rename of one directory over another leaking an inode (5953)
3567 - avoid SetPageDirty on 2.6 (5981)
3568 - don't re-add just-being-destroyed locks to the waiting list (5653)
3569 - when creating new directories, inherit the parent's custom
3570 striping settings if present parent (3048)
3571 - flush buffers from cache before direct IO in 2.6 obdfilter (4982)
3572 - don't hold i_size_sem in ll_nopage() and ll_ap_refresh_count (6077)
3573 - don't hold client locks on temporary worklist from l_lru (5666)
3574 - handle IO errors in 2.6 obdfilter bio completion routine (6046)
3575 - automatically evict dead clients (5921)
3576 - Update file size properly in create+truncate+fstat case (6196)
3577 - Do not unhash mountpoint dentries, do not allow removal of
3579 - Avoid lock ordering deadlock issue with write/truncate (6203,5654)
3580 - reserve enough journal credits in fsfilt_start_log for setattr (4554)
3581 - ldlm_enqueue freed-export error path would always LBUG (6149,6184)
3582 - don't reference lr_lvb_data until after we hold lr_lvb_sem (6170)
3583 - don't overwrite last_rcvd if there is a *_client_add() error (6086)
3584 - Correctly handle reads of files with no objects (6243)
3585 - lctl recover will also mark a device active if deactivate used (5933)
3587 - by default create 1 inode per 4kB space on MDS, per 16kB on OSTs
3588 - allow --write-conf on an MDS with different nettype than client (5619)
3589 - don't write config llogs to MDS for mounts not from that MDS (5617)
3590 - lconf should create multiple TCP connections from a client (5201)
3591 - init scripts are now turned off by default; run chkconfig --on
3592 lustre and chkconfig --on lustrefs to use them
3593 - upcalls are no longer needed for clients to recover to failover
3595 - add --abort-recovery option to lconf to abort recovery on device
3597 - add support for an arbitrary number of OSTs (3026)
3598 - Quota support protocol changes.
3599 - forward compatibility changes to wire structs (6007)
3600 - rmmod NALs that might be loaded because of /etc/modules.conf (6133)
3601 - support for mountfsoptions and clientoptions to the Lustre LDAP (5873)
3602 - improved "lustre status" script
3603 - initialize blocksize for non-regular files (6062)
3604 - added --disable-server and --disable-client configure options (5782)
3605 - introduce a lookup cache for lconf to avoid repeated DB scans (6204)
3606 - Vanilla 2.4.29 support
3607 - increase maximum number of obd devices to 520 (6242)
3608 - remove the tcp-zero-copy patch from the suse-2.4 series (5902)
3609 - Quadrics Elan drivers are now included for the RHEL 3 2.4.21 and
3610 SLES 9 2.6.5 kernels
3611 - limit stripes per file to 160 (the maximum EA size) (6093)
3613 2005-03-22 Cluster File Systems, Inc. <info@clusterfs.com>
3616 - don't LASSERT in ll_release on NULL lld with NFS export (4655, 5760)
3617 - hold NS lock when calling handle_ast_error->del_waiting_lock (5746)
3618 - fix setattr mtime regression from lovcleanup merge (4829, 5669)
3619 - workaround for 2.6 crash in ll_unhash_aliases (5687, 5210)
3620 - small ext3 extents cleanups and fixes (5733)
3621 - improved mballoc code, several small races and bugs fixed (5733, 5638)
3622 - kernel version 43 - fix remove_suid bugs in both 2.4 and 2.6 (5695)
3623 - avoid needless client->OST connect, fix handle mismatch (5317)
3624 - fix DLM error path that led to out-of-sync client, long delays (5779)
3625 - support common vfs-enforced mount options (nodev,nosuid,noexec) (5637)
3626 - fix several locking issues related to i_size (5492,5624,5654,5672)
3627 - don't move pending lock onto export if it is already evicted (5683)
3628 - fix kernel oops when creating .foo in unlinked directory (5548)
3629 - fix deadlock in obdfilter statistics vs. object create (5811)
3630 - use time_{before,after} to avoid timer jiffies wrap (5882)
3631 - shutdown --force/--failover stability (3607,3651,4797,5203,4834)
3632 - Do not leak request if server was not able to process it (5154)
3633 - If mds_open unable to find parent dir, make that negative lookup(5154)
3634 - don't create new directories with extent-mapping (5909, 5936)
3636 - fix lustre/lustrefs init scripts for SuSE (patch from Scali, 5702)
3637 - don't hold the pinger_sem in ptlrpc_pinger_sending_on_import
3638 - change obd_increase_kms to obd_adjust_kms (up or down) (5654)
3639 - lconf, lmc search both /usr/lib and /usr/lib64 for Python libs (5800)
3640 - support for RHEL4 kernel on i686 (5773)
3641 - provide error messages when incompatible logs are encountered (5898)
3643 2005-02-18 Cluster File Systems, Inc. <info@clusterfs.com>
3644 * version 1.4.0.10 (1.4.1 release candidate 1)
3646 - don't keep a lock reference when lock is not granted (4238)
3647 - unsafe list practices (rarely) led to infinite eviction loop (4908)
3648 - add per-fs limit of Lustre pages in page cache, avoid OOM (4699)
3649 - drop import inflight refcount on signal_completed_replay error (5255)
3650 - unlock page after async write error during send (3677)
3651 - handle missing objects in filter_preprw_read properly (5265)
3652 - no transno return for symlink open, don't save no-trasno open (3440)
3653 - don't try to complete elan receive that already failed (4012)
3654 - free RPC server reply state on error (5406)
3655 - clean up thread from ptlrpc_start_thread() on error (5160)
3656 - readahead could read extra page into cache that wasn't ejected (5388)
3657 - prevent races in class_attach/setup/cleanup/detach (5260)
3658 - don't dereference de->d_inode after l_dput of de (5458)
3659 - use "int" for stripe value returned from lock_to_stripe (5544)
3660 - mballoc allocation and error-checking fixes in 2.6 (5504)
3661 - block device patches to fix I/O request sizes in 2.6 (5482)
3662 - look up hostnames for IB nals (5602)
3663 - 2.6 changed lock ordering of 2 semaphores, caused deadlock (5654)
3664 - don't start multiple acceptors for the same port (5277)
3665 - fix incorrect LASSERT in mds_getattr_name (5635)
3666 - export a proc file for general "ping" checking (5628)
3667 - fix "lfs check" to not block when the MDS is down (5628)
3669 - service request history (4965)
3670 - put {ll,lov,osc}_async_page structs in a single slab (4699)
3671 - create an "evict_client" /proc entry on OSTs, like the MDS has
3672 - fix mount usage message, return errors per mount(8) (5168)
3673 - change grep [] to grep "[]" in tests so they work in more UMLs
3674 - fix ppc64/x86_64 spec to use %{_libdir} instead of /usr/lib (5389)
3675 - remove ancient LOV_MAGIC_V0 EA support (5047)
3676 - add "disk I/Os in flight" and "I/O req time" stats in obdfilter
3677 - align r/w RPCs to PTLRPC_MAX_BRW_SIZE boundary for performance (3451)
3678 - allow readahead allocations to fail when low on memory (5383)
3679 - mmap locking landed again, after considerable improvement (2828)
3680 - add get_hostaddr() to lustreDB.py for LDAP support (5459)
3682 2004-11-23 Cluster File Systems, Inc. <info@clusterfs.com>
3685 - send OST transaction number in read/write reply to free req (4966)
3686 - don't ASSERT in ptl_send_rpc() if we run out of memory (5119)
3687 - lock /proc/sys/portals/routes internal state, avoiding oops (4827)
3688 - the watchdog thread now runs as interruptible (5246)
3689 - flock/lockf fixes (but it's still disabled, pending 5135)
3690 - don't use EXT3 constants in llite code (5094)
3691 - memory shortage at startup could cause assertion (5176)
3693 - reorganization of lov code
3694 - single portals codebase
3696 - add extents/mballoc support (5025)
3697 - direct I/O reads in the obdfilter (4048)
3698 - kernel patches from LNXI for 2.6 (bluesmoke, perfctr, mtd, kexec)
3700 tbd Cluster File Systems, Inc. <info@clusterfs.com>
3703 - send OST transaction number in read/write reply to free req (4966)
3704 - don't ASSERT in ptl_send_rpc() if we run out of memory (5119)
3705 - lock /proc/sys/portals/routes internal state, avoiding oops (4827)
3706 - the watchdog thread now runs as interruptible (5246)
3707 - handle missing objects in filter_preprw_read properly (5265)
3708 - unsafe list practices (rarely) led to infinite eviction loop (4908)
3709 - drop import inflight refcount on signal_completed_replay error (5255)
3710 - unlock page after async write error during send (3677)
3711 - return original error code on reconstructed replies (3761)
3712 - no transno return for symlink open, don't save no-trasno open (3440)
3714 - add pid to ldlm debugging output (4922)
3715 - bump the watchdog timeouts -- we can't handle 30sec yet
3716 - extra debugging for orphan dentry/inode bug (5259)
3718 2004-11-16 Cluster File Systems, Inc. <info@clusterfs.com>
3721 - fix TCP_NODELAY bug, which caused extreme perf regression (5134)
3722 - allocate qswnal tx descriptors singly to avoid fragmentation (4504)
3723 - don't LBUG on obdo_alloc() failure, use OBD_SLAB_ALLOC() (4800)
3724 - fix NULL dereference in /proc/sys/portals/routes (4827)
3725 - allow failed mdc_close() operations to be interrupted (4561)
3726 - stop precreate on OST before MDS would time out on it (4778)
3727 - don't send partial-page writes before EOF from client (4410)
3728 - discard client grant for sub-page writes on large-page clients (4520)
3729 - don't free dentries not owned by NFS code, check generation (4806)
3730 - fix lsm leak if mds_create_objects() fails (4801)
3731 - limit debug_daemon file size, always print CERROR messages (4789)
3732 - use transno after validating reply (3892)
3733 - process timed out requests if import state changes (3754)
3734 - update mtime on OST during writes, return in glimpse (4829)
3735 - add mkfsoptions to LDAP (4679)
3736 - use ->max_readahead method instead of zapping global ra (5039)
3737 - don't interrupt __l_wait_event() during strace
3739 - add software watchdogs to catch hung threads quickly (4941)
3740 - make lustrefs init script start after nfs is mounted
3741 - fix CWARN/ERROR duplication (4930)
3742 - return async write errors to application if possible (2248)
3743 - add /proc/sys/portal/memused (bytes allocated by PORTALS_ALLOC)
3744 - print NAL number in %x format (4645)
3745 - update barely-supported suse-2.4.21-171 series (4842)
3746 - support for sles 9 %post scripts
3747 - support for building 2.6 kernel-source packages
3748 - support for sles km_* packages
3750 2004-10-07 Cluster File Systems, Inc. <info@clusterfs.com>
3753 - ignore -ENOENT errors in osc_destroy (3639)
3754 - notify osc create thread that OSC is being cleaned up (4600)
3755 - add nettype argument for llmount in #5d in conf-sanity.sh (3936)
3756 - reconstruct ost_handle() like mds_handle() (4657)
3757 - create a new thread to do import eviction to avoid deadlock (3969)
3758 - let lconf resolve symlinked-to devices (4629)
3759 - don't unlink "objects" from directory with default EA (4554)
3760 - hold socknal file ref over connect in case target is down (4394)
3761 - allow more than 32000 subdirectories in a single directory (3244)
3762 - fix blocks count for O_DIRECT writes (3751)
3763 - OST returns ENOSPC from object create when no space left (4539)
3764 - don't send truncate RPC if file size isn't changing (4410)
3765 - limit OSC precreate to 1/2 of value OST considers bogus (4778)
3766 - bind to privileged port in socknal and tcpnal (3689)
3768 - rate limit CERROR/CWARN console message to avoid overload (4519)
3769 - GETFILEINFO dir ioctl returns LOV EA + MDS stat in 1 call (3327)
3770 - basic mmap support (3918)
3771 - kernel patch series update from b1_4 (4711)
3773 2004-09-16 Cluster File Systems, Inc. <info@clusterfs.com>
3776 - avoid crash during MDS cleanup with OST shut down (2775)
3777 - fix loi_list_lock/oig_lock inversion on interrupted IO (4136)
3778 - don't use bad inodes on the MDS (3744)
3779 - dynamic object preallocation to improve recovery speed (4236)
3780 - don't hold spinlock over lock dumping or change debug flags (4401)
3781 - don't zero obd_dev when it is force cleaned (3651)
3782 - print grants to console if they go negative (4431)
3783 - "lctl deactivate" will stop automatic recovery attempts (3406)
3784 - look for existing locks in ldlm_handle_enqueue() (3764)
3785 - don't resolve lock handle twice in recovery avoiding race (4401)
3786 - revalidate should check working dir is a directory (4134)
3788 - don't always mark "slow" obdfilter messages as errors (4418)
3790 2004-08-24 Cluster File Systems, Inc. <info@clusterfs.com>
3793 - don't close LustreDB during write_conf until it is done (3860)
3794 - fix typo in lconf for_each_profile (3821)
3795 - allow dumping logs from multiple threads at one time (3820)
3796 - don't allow multiple threads in OSC recovery (3812)
3797 - fix debug_size parameters (3864)
3798 - fix mds_postrecov to initialize import for llog ctxt (3121)
3799 - replace config semaphore with spinlock (3306)
3800 - be sure to send a reply for a CANCEL rpc with bad export (3863)
3801 - don't allow enqueue to complete on a destroyed export (3822)
3802 - down write_lock before checking llog header bitmap (3825)
3803 - recover from lock replay timeout (3764)
3804 - up llog sem before sending rpc (3652)
3805 - reduce ns lock hold times when setting kms (3267)
3806 - change a dlm LBUG to LASSERTF, to maybe learn something (4228)
3807 - fix NULL deref and obd_dev leak on setup error (3312)
3808 - replace some LBUG about llog ops with error handling (3841)
3809 - don't match INVALID dentries from d_lookup and spin (3784)
3810 - hold dcache_lock while marking dentries INVALID and hashing (4255)
3811 - fix invalid assertion in ptlrpc_set_wait (3880)
3813 - add libwrap support for the TCP acceptor (3996)
3814 - add /proc/sys/portals/routes for non-root route listing (3994)
3815 - allow setting MDS UUID in .xml (2580)
3816 - print the stack of a process that LBUGs (4228)
3818 2004-07-14 Cluster File Systems, Inc. <info@clusterfs.com>
3821 - don't cleanup request in ll_file_open() on failed MDS open (3430)
3822 - make sure to unset replay flag from failed open requests (3440)
3823 - if default stripe count is 0, use OST count for inode size (3636)
3824 - update parent mtime/ctime on client for create/unlink (2611)
3825 - drop dentry ref in ext3_add_link from open_connect_dentry (3266)
3826 - free recovery state on server during a forced cleanup (3571)
3827 - unregister_reply for resent reqs (3063)
3828 - loop back devices mounting and status check on 2.6 (3563)
3829 - fix resource-creation race that can provoke i_size == 0 (3513)
3830 - don't try to use bad inodes returned from MDS/OST fs lookup (3688)
3831 - more debugging for page-accounting assertion (3746)
3832 - return -ENOENT instead of asserting if ost getattr+unlink race (3558)
3833 - avoid deadlock after precreation failure (3758)
3834 - fix race and lock order deadlock in orphan handling (3450, 3750)
3835 - add validity checks when grabbing inodes from l_ast_data (3599)
3837 - add /proc/.../recovery_status to obdfilter (3428)
3838 - lightweight CDEBUG infrastructure, debug daemon (3668)
3839 - change default OSC RPC parameters to be better on small clusters
3840 - turn off OST read cache for files smaller than 32MB
3841 - install man pages and include them in rpms (3100)
3842 - add new init script for (un)mounting lustre filesystems (2593)
3843 - run chkconfig in %post for init scripts (3701)
3844 - drop scimac NAL (unmaintained)
3846 2004-06-17 Cluster File Systems, Inc. <info@clusterfs.com>
3849 - clean kiobufs before and after use (3485)
3850 - strip trailing '/'s before comparing paths with /proc/mounts (3486)
3851 - remove assertions to work around "in-flight rpcs" recovery bug (3063)
3852 - change init script to fail more clearly if not run as root (1528)
3853 - allow clients to reconnect during replay (1742)
3854 - fix ns_lock/i_sem lock ordering deadlock for kms update (3477)
3855 - don't do DNS lookups on NIDs too small for IP addresses (3442)
3856 - re-awaken ptlrpcd if new requests arrive during check_set (3554)
3857 - fix cond_resched (3554)
3858 - only evict unfinished clients after recovery (3515)
3859 - allow bulk resend, prevent data loss (3570)
3860 - dynamic ptlrpc request buffer allocation (2102)
3861 - don't allow unlinking open directory if it isn't empty (2904)
3862 - set MDS/OST threads to umask 0 to not clobber client modes (3359)
3863 - remove extraneous obd dereference causing LASSERT failure (3334)
3864 - don't use get_cycles() when creating temp. files on the mds (3156)
3865 - hold i_sem when setting i_size in ll_extent_lock() (3564)
3866 - handle EEXIST for set-stripe, set proper directory name (3336)
3868 - servers can dump a log evicting a client - lustre.dump_on_timeout=1
3869 - fix ksocknal_fmb_callback() error messages (2918)
3871 2004-05-27 Cluster File Systems, Inc. <info@clusterfs.com>
3874 - don't copy lvb into (possibly NULL) reply on error (2983)
3875 - don't deref dentry after dput, don't free lvb on error (2922)
3876 - use the kms to determine writeback rpc length (2947)
3877 - increment oti_logcookies when osc is inactive (2948)
3878 - update client's i_blocks count via lvb messages (2543)
3879 - handle intent open/close of special files properly (1557)
3880 - mount MDS with errors=remount-ro, like obdfilter (2009)
3881 - initialize lock handle to avoid ASSERT on error cleanup (3057)
3882 - don't use cancelling-locks' kms values (2947)
3883 - use highest lock extent for kms, not last one (2925)
3884 - don't dereference ERR_PTR() dentry in error handling path (3107)
3885 - fix thread race in portals_debug_dumplog() (3122)
3886 - create lprocfs device entries at setup instead of at attach (1519)
3887 - common AST error handler, don't evict client on completion race (3145)
3888 - zero nameidata in detach_mnt in 2.6 (3118)
3889 - verify d_inode after revalidate_special is valid in 2.6 (3116)
3890 - use lustre_put_super() to handle zconf unmounts in 2.6 (3064)
3891 - initialize RPC timeout timer earlier for 2.6 (3219)
3892 - don't dereference NULL reply buffer if mdc_close was never sent (2410)
3893 - print nal/nid for unknown nid (3258)
3894 - additional checks for oscc recovery before doing precreate (3284)
3895 - fix ll_extent_lock() error return code for 64-bit systems (3043)
3896 - don't crash in mdc_close for bad permissions on open (3285)
3897 - zero i_rdev for non-device files (3147)
3898 - clear page->private before handing to FS, better assertion (3119)
3899 - tune the read pipeline (3236)
3900 - fix incorrect decref of invalidated dentry (2350)
3901 - provide read-ahead stats and refine rpc in flight stats (3328)
3902 - don't hold journal transaction open across create RPC (3313)
3903 - update atime on MDS at close time (3265)
3904 - close LDAP connection when recovering to avoid server load (3315)
3905 - update iopen-2.6 patch with fixes from 2399,2517,2904 (3301)
3906 - don't leak open file on MDS after open resend (3325)
3907 - serialize filter_precreate and filter_destroy_precreated (3329)
3908 - loop device shouldn't call sync_dev() for nul device (3092)
3909 - clear page cache after eviction (2766)
3910 - resynchronize MDS->OST in background (2824)
3911 - refuse to mount the same filesystem twice on same mountpoint (3394)
3912 - allow llmount to create routes for mounting behind routers (3320)
3913 - push lock cancellation to blocking thread for glimpse ASTs (3409)
3914 - don't call osc_set_data_with_check() for TEST_LOCK matches (3159)
3915 - fix rare problem with rename on htree directories (3417)
3917 - allow default OST striping configuration per directory (1414)
3918 - fix compilation for qswnal for 2.6 kernels (3125)
3919 - increase maximum number of MDS request buffers for large systems
3920 - change liblustreapi to be useful for external progs like lfsck (3098)
3921 - increase local configuration timeout for slow disks (3353)
3922 - allow configuring ldlm AST timeout - lustre.ldlm_timeout=<seconds>
3924 2004-03-22 Cluster File Systems, Inc. <info@clusterfs.com>
3927 - fixes for glimpse AST timeouts / incorrectly 0-sized files (2818)
3928 - don't overwrite extent policy data in reply if lock was blocked (2901)
3929 - drop filter export grants atomically with removal from device (2663)
3930 - del obd_self_export from work_list in class_disconnect_exports (2908)
3931 - don't LBUG if MDS recovery times out during orphan cleanup (2530)
3932 - swab reply message in mdc_close, other PPC fixes (2464)
3933 - fix destroying of named logs (2325)
3934 - overwrite old logs when running lconf --write_conf (2264)
3935 - bump LLOG_CHUNKSIZE to 8k to allow for larger clusters (2306)
3936 - fix race in target_handle_connect (2898)
3937 - mds_reint_create() should take same inode create lock (2926)
3938 - correct journal credits calculated for CANCEL_UNLINK_LOG (2931)
3939 - don't close files for self_export to avoid uninitialized obd (2936)
3940 - allow MDS with the same name as client node (2939)
3941 - hold dentry reference for closed log files for unlink (2325)
3942 - reserve space for all logs during transactions (2059)
3943 - don't evict page beyond end of stripe extent (2925)
3944 - don't oops on a deleted current working directory (2399)
3945 - handle hard links to targets without a parent properly (2517)
3946 - don't dereference NULL lock when racing during eviction (2867)
3947 - don't grow lock extents when lots of conflicting locks (2919)
3949 2004-03-04 Cluster File Systems, Inc. <info@clusterfs.com>
3952 - account for cache space usage on clients to avoid data loss (974)
3953 - lfsck support in lustre kernel code (2349)
3954 - reduce journal credits needed for BRW writes (2370)
3955 - orphan handling to avoid losing space on client/server crashes
3956 - ptlrpcd can be blocked, stopping ALL progress (2477)
3957 - use lock value blocks to assist in proper KMS, faster stat (1021)
3958 - takes i_sem instead of DLM locks internally on obdfilter (2720)
3959 - recovery for initial connections (2355)
3960 - fixes for mds_cleanup_orphans (1934)
3961 - abort_recovery crashes MDS in b_eq (mds_unlink_orphan) (2584)
3962 - block all file creations until orphan recovery completes (1901)
3963 - client remove rq_connection from request struct (2423)
3964 - conf-sanity test_5, proper cleanup in umount log not availale (2640)
3965 - recovery timer race (2670)
3966 - mdc_close recovey bug (2532)
3967 - ptlrpc cleanup bug (2710)
3968 - mds timeout on local locks (2588)
3969 - namespace lock held during RPCs (2431)
3970 - handle interrupted sync write properly (2503)
3971 - don't try to handle a message that hasn't been replied to (2699)
3972 - client assert failure during cleanup after abort recovery (2701)
3973 - leak mdc device after failed mount (2712)
3974 - ptlrpc_check_set allows timedout requests to complete (2714)
3975 - wait for inflight reqs when ptlrpcd finishes (2710)
3976 - make sure unregistered services are removed from the srv_list
3977 - reset bulk XID's when resending them (caught by 1138 test)
3978 - unregister_bulk after timeout
3979 - fix lconf error (2694)
3980 - handle write after unfinished setstripe, stripe-only getstripe (2388)
3981 - readahead locks pages, leaves pending causing memory pressure (2673)
3982 - increase OST request buffers to 4096 on large machines (2729)
3983 - fix up permission of existing directories in simple_mkdir (2661)
3984 - init deleted item, add assertions ptlrpc_abort_inflight() (2725)
3985 - don't assign transno to errored transactions (2742)
3986 - don't delete objects on OST if given a bogus objid from MDS (2751)
3987 - handle large client PAGE_SIZE readdir on small PAGE_SIZE MDS (2777)
3988 - if rq_no_resend, then timeout request after recovery (2432)
3989 - fix MDS llog_logid record size, 64-bit array alignment (2733)
3990 - don't call usermode_helper from ptlrpcd, DEFAULT upcall (2773)
3991 - put magic in mount.lustre data, check for bad/NULL mount data (2529)
3992 - MDS recovery shouldn't delete objects that it has given out (2730)
3993 - if enqueue arrives after completion, don't clobber LVB (2819)
3994 - don't unlock pages twice when trigger_group_io returns error (2814)
3995 - don't deref NULL rq_repmsg if ldlm_handle_enqueue failed (2822)
3996 - don't write pages to disk if there was an error (1450)
3997 - don't ping imports that have recovery disabled (2676)
3998 - take buffered bytes into account when balancing socknal conn (2817)
3999 - hold a DLM lock over readdir always, use truncate_inode_pages (2706)
4000 - reconnect unlink llog connection after MDS reconnects to OST (2816)
4001 - remove little-endian swabbing of llog records (1987)
4002 - set/limit i_blksize to LL_MAX_BLKSIZE on client (2884)
4003 - retry reposting request buffers if they fail (1191)
4004 - grow extent at grant time to avoid granting a revoked lock (2809)
4005 - lock revoke doesn't evict page if covered by a second lock (2765)
4006 - disable VM readahead to avoid reading outside lock extents (2805)
4008 - return LL_SUPER_MAGIC from statfs for the filesystem type (1972)
4009 - updated kernel patches for hp-2.4.20 kernel (2681)
4011 2004-02-07 Cluster File Systems, Inc. <info@clusterfs.com>
4014 - fix truncated write corruption (2366)
4015 - fix for failed assertion in iopen_connect_dentry (1792,2517)
4017 - don't flag the ptlrpcd thread with PF_MEMALLOC (2636)
4018 - ensure len(uuid) < 37 in lmc (1171)
4019 - fix ia64 OOPS in llog_test (2255)
4020 - zero end of page at obdfilter for partial page writes (2648)
4021 - don't leave stale dentries around after renames (bug 2428)
4022 - fix timeouts when evicting a client with a single lock held (2642)
4023 - set deadline for the initial HELLO message to drain (2634)
4024 - print out dotted-quad IP addresses in the socknal (2302)
4026 - additional debugging for MDS client eviction problem (2443)
4027 - fix mkfsoptions support for osts (2603, 2604)
4029 2004-01-27 Cluster File Systems, Inc. <info@clusterfs.com>
4032 - add series for the vanilla 2.6.0 kernel
4033 - add series for the vanilla 2.4.24 kernel
4034 - add series for a cray x86/64 UL kernel drop
4035 - fix xattr patches for the vanilla 2.4.19 series
4037 - generate true UUIDs in lmc (1171)
4038 - have portals stack dumping break in UML (2466)
4039 - avoid bad dchild deref; avoid inum lock w/o creation (2362)
4040 - allocate with _NOFS in ldlm to avoid deadlock (1933)
4041 - wake callback waiting threads on client eviction (2460)
4042 - Add --ptldebug and --subsystem to lmc (1719)
4043 - update assertion to allow safe interrupt allocation
4044 - set rq_no_resend for cancel requests (2432)
4045 - recalculate ptlrpcd timeout after resend (2494)
4046 - call vfs_rmdir when removing pending directories (2368)
4047 - fix renaming a file to itself (2429)
4048 - lmc creates a default one-stripe lov (2454)
4049 - expand procfs space to handle large clusters (2326)
4050 - increase UML stack to avoid overflow
4051 - update lconf's list of debug and subsystem masks
4052 - fix lfs find --obd (2510)
4053 - /proc tunable for disabling filter read caching (2591)
4054 - stop rpm packages from altering slapd.conf (2301)
4055 - disable nagle in the socknal under 0conf (2578)
4056 - choose mds inode size based on stripe count (2572)
4057 - fix kernel-source rpm problems (2516)
4059 - add --disable-doc to avoid pdf generation (2421)
4060 - update documentation, tests, type-os, comments
4061 - avoid format warnings on ia64
4062 - remove the TOE NAL
4063 - tiny code cleanups by removing unused fields
4065 2004-01-07 Cluster File Systems, Inc. <info@clusterfs.com>
4068 - fix obvious semaphore misuse in as-yet-unused setattr path (2348)
4069 - remove the most blatant lies from BUILDING file (2371)
4070 - change default debug level to reasonable production setting
4071 - reduce client side cache size to reduce cache flush time
4072 - reduce max RPCs in flight to avoid unnecessary file fragmentation
4073 - make TCP zerocopy and pinger support enabled by default (2476)
4074 - sync writes completed after process exits caused crashes (2319)
4075 - maintain correct mount count on the MDS (2356)
4076 - backout 1557, because 2316 wasn't really fixed
4077 - better file I/O statistics gathering in /proc
4078 - don't take unnecessary, deadlock-inducing bug in readpage (2383)
4079 - another kernel patch to fix zero-copy TCP function export
4080 - don't take duplicate lock when processing re-sent getattr (2420)
4081 - lctl uses obd_self_export instead of creating new conn (2353)
4082 - MDS/OST recovery case which requires object creation asserted (2425)
4083 - move lfs from /usr/sbin to /usr/bin in packages
4084 - fix race between mds_client_add and mds_client_free (2417)
4085 - use kmalloc instead of slabs in portals (2430)
4086 - don't create duplicate records when a failover MDS is present (2442)
4087 - remove unnecessary mount age check (2332)
4088 - don't remove directory inodes from locks prematurely (2451)
4089 - don't break if MDS service name is the same as hostname (2103)
4090 - fix races in client write RPC generation when cache full (2482)
4092 2003-12-13 Cluster File Systems, Inc. <info@clusterfs.com>
4095 - remove now-unused request->rq_obd (278)
4096 - if an allocation fails, print out how much memory we've used (1933)
4097 - use PORTAL_SLAB_ALLOC for structures, to get GFP_MEMALLOC (1933)
4098 - add the "configurable stack size" patch to most series files (1256)
4099 - ability to write large log records, for 100+ OST configs (2306)
4100 - fix NULL deref when filter_prep fails (2314)
4101 - fix operator precedence error in filter_sync
4102 - dynamic allocation of socknal TX descriptors (2315)
4103 - fix a missed case in the GFP_MEMALLOC patch, can cause deadlock (2310)
4104 - fix gcc 2.96 compilation problem in xattr kernel patch (2294)
4105 - ensure that CWARN messages in Portals always get to the syslog
4106 - __init/__exit are not for prototype decls (ldlm_init/exit)
4107 - x86-64 compile warning fixes
4108 - fix gateway LMC keyword conflict (2318)
4109 - fix MDS lock inversions in getattr/reint paths (1844)
4110 - fix a rare lock re-ordering bug, which caused deadlock (2322)
4111 - fix i_sem/journal inversion in fsfilt_ext3_write_record (2306)
4112 - DLM race condition prevented some lock evictions (2328)
4113 - ENOMEM detection and retry on socknal sends (2230)
4114 - use GFP_NOFS throughout Lustre, to combat ENOMEM (2230)
4115 - move osc_rpcd into ptlrpc, for use in MDC and others (2329)
4116 - protect MDS inode fsdata with stronger locking; fixes assertion (2313)
4117 - better error messages when a client is rejected during recovery (1505)
4118 - avoid cancelling locks which were never granted, after failure (2330)
4119 - fix i_sem/journal inversion in mds_client_add (2333)
4120 - fix truncate/getattr lock cycle deadlock (2334)
4121 - use rpcd to send close; allows resend after timeout, avoid leak (1897)
4122 - fix two rare exit paths which could leak an l_lock() ref (2321)
4123 - fencepost error in MDS/OST orphan recovery (2226)
4124 - make log record alignment 8 bytes (1988)
4125 - lstripe now fails when requested offset > ost_count (2237)
4126 - ensure that all kernel series have a complete list.h (1607)
4127 - fix crashes in special-file operations (2316)
4128 - lctl create/brw OID mismatch, caused by obsolete filter loop (2339)
4130 - allow configurable automake binary, for testing new versions
4131 - small update to the lfs documentation
4133 2003-12-03 Cluster File Systems, Inc. <info@clusterfs.com>
4135 * fix negative export reference count in fsfilt_sync (2312)
4137 2003-12-01 Cluster File Systems, Inc. <info@clusterfs.com>
4138 * release candidate 0.9.1
4140 - orphans are moved into the PENDING directory for possible recovery
4141 - replayed opens now open by fid for orphan/rename safety (1042)
4142 - last close of an orphan inode generates a transno (683)
4143 - chdir() and mount() now pin the directory entry (1020)
4144 - avoid CERROR in normal ll_setattr_raw() error case (1500)
4145 - discard very old requests without processing them (1502)
4146 - remove some common, well-understood CERRORs (1505)
4147 - require O_DIRECT I/O to be page-sized to workaround IA64 crash (1609)
4148 - clear "grant" flags in OST replies until OST grant code lands (1644)
4149 - fix read performance by not clobbering i_blksize on client (1598)
4150 - fix __ldlm_handle2lock oops by not dereferencing lock after PUT (1625)
4151 - make LRU size a /proc tunable, clears locks when reduced (707)
4152 - fix some lprocfs rot that prevented ptlbd from loading (1732)
4153 - server locks take references on exports now (1558)
4154 - build fixes for 2.4.20-rh trees (1663)
4155 - return an error from lov_create if all OSCs are inactive (1751)
4156 - fix import levels when a reconnect happens without a timeout (1597)
4157 - exit early from mds_open if we get a lookup error (1749)
4158 - partial page read at EOF wouldn't wait for disk before sending (1642)
4159 - avoid NULL deref in obdfilter when reading page past EOF (1592)
4160 - avoid LASSERT in ll_intent_lock if server failed very early (1090)
4161 - fix LBUG in ll_it_open_error with rc = -2 (1861)
4162 - write/truncate lock inversion (1639)
4163 - Don't auto-load obdclass, portals modules during cleanup (1495)
4164 - fix timestamps from jumping to "now" (1763)
4165 - extra journal assertions (1648)
4166 - add an extra multiunlink test (1771)
4167 - fix read_record/write_record API (1776)
4168 - fix leak of offset_extent, possible incorrect i_size later (1772)
4169 - fix lasserts in mis-matched transnos during open-unlink testing (1541)
4170 - Debugging for the kqswnal_get_idle_tx problems (1820)
4171 - Allow recovery to be attempted multiple times (1536)
4172 - Write out MDS last_rcvd file after it is first created (1600)
4173 - Fix tx_descriptor leak in failed transmit situations (1827)
4174 - ext3 journaling fixes for assertion failure after IO error (1871)
4175 - class_export_put() on freed export after completion AST error (1896)
4176 - Fix revalidate looping in VFS (1322)
4177 - Don't access a freed export during MDS_REINT timeout (1521)
4178 - Add open-unlink recovery support on the MDS (1673,1764)
4179 - Return an error if no MDS data was read from last_rcvd (1946)
4180 - Fix for lookup "." or ".." crash on error (1932,1931,1935)
4181 - Don't setup a disk device that doesn't match exported UUID (317)
4182 - Reduce bulk RPC timeout to avoid cascading client/OST failures (1845)
4183 - avoid committing NULL handle in force close
4184 - local.sh is now a one-stripe LOV configuration
4185 - POSIX utime.4 -EPERM on FIFO not owned by user (56)
4186 - fix ext3 htree duplicate directory entry corruption (1516)
4187 - POSIX creat.13, fstat.1, open.18, stat.3 new file atime/mtime (2020)
4188 - update to new LOV EA format (2097)
4189 - interoperability for different PAGE_SIZE/wordsize (686,1821,1343,2042)
4191 2003-06-15 Phil Schwan <phil@clusterfs.com>
4194 - imports and exports cleanup too early, need refcounts (349, 879, 1045)
4195 - per-import/export recovery handling (958, 931, 959)
4196 - multiple last-rcvd slots, for serving multiple FSes (949)
4197 - connections are again shared between multiple imp/exports (963, 964)
4198 - "umount -f" would hang if any requests needed to be sent (393, 978)
4199 - avoid pinning large req buffer by copying for queued messages (989)
4200 - add "uuid" to "lctl device" command to help upcalls (991)
4201 - "open" RPCs with transnos would confuse recovery counters (1037)
4202 - do proper endian conversion of all wire messages (288, 340, 891)
4203 - remove OST bulk get LBUGs, fix ost_brw_write cleanup (1126)
4204 - call waiting locks callback from LDLM recovery thread (1127, 1151)
4205 - fix ptlrpc_connection leak in target_handle_connect (1174)
4206 - fix import refcounting bug in OST and MDS cleanup (1134)
4207 - if an invalid-at-open-time OSC returned before close(), LBUG (1150)
4208 - fix very unlikely obd_types race condition (501)
4209 - remove osc_open hack for echo_client (1187)
4210 - we leaked exports/dlmimps for forcibly disconnected clients (1143)
4211 - a failure in read_inode2 leads to deadlock (1139)
4212 - cancel ack-locks as soon as transaction is committed (1072)
4213 - fix major leaks and crashes in the bulk I/O path (937, 1057)
4214 - make sure to commitrw after any preprw to avoid deadlock (1162)
4215 - failing to execute a file in a lustre FS would lock inode (1203)
4216 - small DEBUG_REQ fix to avoid dereferencing a NULL (1227)
4217 - don't ASSERT while cleaning up an incompletely-setup obd (1248)
4218 - obd_uuid2tgt would walk off the end of the list (1255)
4219 - on IA64 the osc would give portals incorrect bulk size (1258)
4220 - fix debug daemon ioctl interface; allows daemon on ia64 (1274)
4221 - fix lock inversion caused by new llite matching code (1282)
4222 - limit the number of dirty pages on a client to 10MB (1286)
4223 - timed out locks were not being corrected cancelled (1289)
4224 - fix O_DIRECT above 4GB on IA-32 (1292)
4225 * major user-visible changes
4226 - fail out/fail over policy now controlled by the upcall (993)
4228 - add OBD_PING to check server availability and failure (954)
4229 - lustre messages are now sent in sending host order (288, 340, 891)
4230 - add eadatalen to MDS getattr reply (340)
4231 - OST read replies may contain second buffer, with per-page status (593)
4233 2003-03-11 Phil Schwan <phil@clusterfs.com>
4236 - LDLM_DEBUG macro fix, for gcc 3.2 (850)
4237 - failed open()s could cause deadlock; fixed (867, 869)
4238 - stop cancelling OST locks when files are closed (481)
4239 - overlapping XID spaces caused network corruption (851, 853)
4240 - fix unsafe fsfilt counter arithmetic; change to atomic_t
4241 - setattr_raw added, to do single-RPC, server-side setattrs
4242 - lmc/lconf syntax change for OST UUIDs
4243 - fix crashy race condition between ptlrpc_free_req and osc_close
4244 - don't use request in mdc_enqueue if we hit a timeout (889)
4245 - don't set the inode i_size for regular files from the MDS (896)
4246 - handle out of order completion AST (842)
4247 - don't LBUG if a lock request times out after receiving AST (913)
4248 - avoid d_rehash race in ll_find_alias by rehashing inside dcache_lock
4249 - if a bad lock AST arrives, send an error instead of dropping entirely
4250 - return 0 from revalidate2 if ll_intent_lock returns -EINTR (912)
4251 - fix leak in bulk IO when only partially completed (899, 900, 926)
4252 - fix O_DIRECT for ia64 (55)
4253 - (almost) eliminate Lustre-kernel-thread effects on load average (722)
4254 - C-z after timeout could hang a process forever; fixed (977)
4256 - client-side I/O cache (678, 924, 929, 941, 970)
4258 - READPAGE and SETATTRs which don't take server-side locks get
4261 2003-02-11 Phil Schwan <phil@clusterfs.com>
4264 - Fix ldlm_lock_match on the MDS to avoid matching remote locks (592)
4265 - Fix fsfilt_extN_readpage() to read a full page of directory
4266 entries, or fake the remainder if PAGE_SIZE != blocksize (500)
4267 - Avoid extra mdc_getattr() in ll_intent_lock when possible (534, 604)
4268 - Fix imbalanced LOV object allocation and out-of-bound access (469)
4269 - Most intent operations were removed, in favour of a new RPC mode
4270 that does a single RPC to the server and bypasses most of the VFS
4271 - All LDLM resource ID arrays were removed in favour of ldlm_res_id
4272 - Aggressively cancel local locks on DLM servers
4273 - mds_reint_unlink sends EA to the client if it's the last nlink.
4274 client uses that EA to unlink OST objects.
4275 - mds_reint_{rename,unlink,link} were rewritten to take ordered locks
4276 - recursive symlinks were fixed (439)
4277 - fixed NULL deref in DEBUG_REQ
4278 - filter_update_lastobjid no longer calls sync, which annoyed extN
4279 - fixed multi-client small-writes to a single file problem (445)
4280 - fixed mtime updates during file writes (607)
4281 - fixed vector writes on obdfilter causing problems when ENOSPC (670)
4282 - fixed bug in obd_brw_read/write() (under guise of testing 367)
4283 - fixed Linux OST size reporting problem (444, 656)
4284 - OST now updates object mtime with writes or setattr (607, 619)
4285 - client verifies file size before zeroing page past EOF (445)
4286 - OST now writes last allocated objid to disk with allocation (108)
4287 - LOV on echo now works (409)
4289 - mds_reint_unlink sends a new buffer, with the EA included. this
4290 buffer is only valid if body->valid & OBD_MD_FLEASIZE, which is only
4291 set if a regular file was being unlinked, and it was the last link
4292 - use PtlGet from the target for bulk writes (315)
4293 - OST now updates object mtime with writes or setattr (607, 619)
4294 - LDLM now has a grant-time callback to revalidate locked items, if
4296 - Many MDS operations were reorganized to combat race conditions
4298 - Merge b_intel branch (updated lprocfs code) - now at /proc/fs/lustre
4299 - configure check to avoid gcc version 2.96 20000731-2.96-98) (606)
4301 2003-01-06 Andreas Dilger <adilger@clusterfs.com>
4304 - Fully reactivate OST imports after reconnection (512, others)
4305 - Make sure client sees our -ENOTCONN from mds_handle (513 - partial)
4306 - More graceful error handling for truncating on dead OST (515)
4307 - Don't error out unless we're actually accessing dead stripes (474)
4308 - Fix garbage sizes when stripes are missing (410)
4309 - LRU counters were broken, causing constant lock purge (433, 432)
4310 - garbage on read from stripes with failed OSTs (441)
4311 - mark OSCs as active before reconnecting during recovery (438)
4312 - lov_enqueue and lov_cancel need to handle inactive OSTs (403)
4313 - lfind did not preserve OST order in output (443)
4314 - symlinks cause hung clients, incorrect data (439)
4315 - stop dereferencing request after dropping refcount (457)
4316 - don't LASSERT(spin_is_locked) on non-SMP (455)
4317 - fixes for many rename() bugs
4318 - fstat didn't correctly synchronize attributes (399)
4319 - server must handle lock cancellation during blocking AST prep (487)
4320 - bulk descriptors were free()d too soon (511)
4321 - fix paths in lconf, which would load incorrect modules (451, 507)
4322 - fix confusing lconf 'host not found' error message (386)
4323 - fix lock order deadlock on OST (O/R i_sem before journal ops, 478)
4324 - fix race condition in mdc_blocking_ast() for inode access (526)
4325 - fix lov_unpackmd() unpacking wrong number of stripes (537)
4326 - fix lov_set_osc_active() marking wrong OSC inactive (440)
4327 - fix bad lstripe lov_unpackmd() assertion (fix layering too) (527)
4328 - fix multiple writes of stripe MD to MDS (358, maybe 519)
4329 - fix lstripe in several ways (kernel side) (527)
4330 - fix request leak in ldlm_cli_enqueue (262)
4331 - incorrect OSC was marked inactive after OST failure
4332 - call mds_fs_cleanup before unmounting filesystem (524)
4333 - fix races between taking ns_lock and ldlm_lock_change_resource
4334 - fix races updating LOV export open file list
4335 - fix lov_enqueue error path, avoid decref-ing bad lock handle
4336 - fix recovery NULL deref in ldlm_cli_cancel_unused
4337 - fix some DLM races by using new hash table for lock handles (419)
4338 - permit the client to specify desired inodes, at replay
4339 - duplicate requests when we queue them for replay reintegration
4340 - fix last_rcvd offset calculation
4341 - sync after each recovered transaction, so we always make progress
4342 - never, not always, ERESTART requests without transnos
4343 - store the lov_desc in the MDS, so we don't depend on getlovinfo to
4345 - skip replay if the MDS says that the client is already connected
4346 - don't check for a recovery-enabled export to match lctl's UUID
4347 - don't INC_USE_COUNT for phantom exports
4348 - don't crash when cleaning up phantom exports (567)
4349 - don't double-finish or set replay data for errored mdc_open requests
4350 - abort requests when they time out, so we don't get old replies
4351 - send/receive replies for AST messages again
4352 - if the client says that it doesn't have the lock, cancel it on the
4354 - if we timeout during I/O, don't try to cancel an in-use lock; instead
4355 mark it as destroyed, it will all work out when decref is called
4356 - fix module use counts (22, 581)
4358 - ASTs now expect a reply (server cancels lock on error reply)
4360 2002-12-02 Andreas Dilger <adilger@clusterfs.com>
4363 - fix many simultaneous client startup (392)
4364 - fix dentry->d_it clobbering
4365 - credentials weren't being shipped for readdir/getattr operations
4366 - remove invalid assertions triggered during some concurrent MD
4368 - proper Lustre versions added (336, 389)
4369 - fix memory leak for create error case (398)
4370 - fix LOV locking bug that would get cli/srv out of sync
4371 - fix echo client over LOV (409)
4372 - fix dbench 2, extN refcount problem (170, 258, 356, 418)
4373 - fix double-O_EXCL intent crash (424)
4374 - avoid sending multiple lock CANCELs (352)
4376 - MDS can do multi-client recovery (modulo bugs in new code)
4378 - many updates, edits, cleanups
4380 2002-11-18 Phil Schwan <phil@clusterfs.com>
4383 - fix null d_it dereference (346)
4384 - fix full OST/dbench hang (333)
4385 - fix permission problem with file removal (286)
4386 - fix removal of OSCs from LOV when they fail
4387 - fix NULL deref during bulk timeout (214)
4388 - fix problems related to multiple filesystems on one MDS (241)
4389 - fixed serious subtle metadata locking bugs
4390 - free locks on clients when inodes are removed due to memory
4392 - fix inode pointer in lock data (285)
4393 - partial support for multiple MDS on a single host (241)
4394 - data locks weren't cancelled at clear_inode time (290, 311)
4395 - intent locks could lead to unbounded lock growth (205)
4396 - added a maximum lock count, an LRU list, and a flusher
4397 - fix multiple rename (365)
4398 - properly abstracted the echo client
4399 - OSC locked 1 byte too many; fixed
4400 - rewrote brw callback code:
4401 - fixed recovery bugs related to LOVs (306)
4402 - fixed too-many-pages-in-one-write crash (191)
4403 - fixed (again) crash in sync_io_timeout (214)
4404 - probably fixed callback-related race (385)
4406 - Add capability to MDS protocol
4407 - LDLM cancellations and callbacks on different portals
4409 2002-10-28 Andreas Dilger <adilger@clusterfs.com>
4412 - limit client IOV size to PTL_MD_MAX_IOV (611336, 191)
4413 - defer open object destruction to close time (601981, 138)
4414 - open/close OST file handle in obdo (OBD_MD_FLHANDLE) (601981, 138)
4415 - move LDLM_ENQUEUE/CONVERT back to MDS portal (625069)
4416 - abstract ll_lookup2, fix ll_revalidate2 to use abstraction (256)
4417 - don't call obd_setattr in ll_file_release for destroyed objects
4418 * protocol change to lustre_msg: move |version| and add |flags|
4419 * protocol change to osc_punch: "start" in "o_size", "end" in "o_blocks"
4420 * lock replay: for LDLM_FL_REPLAY trust client to do right thing
4421 * added replay of create, unlink, link and rename operations during
4422 MDS failover; recovery should be much more robust now
4423 * remove failed OSCs from LOVs (only lov_create uses this so far)
4424 * the lustre-HOWTO was brought (more) up to date (582544)
4426 2002-10-23 Phil Schwan <phil@clusterfs.com>
4429 - in-use dentries weren't being reused properly (617851)
4430 - prevent multiple LDLM setup (599178)
4431 - fix LOV size calculations for truncate (617853)
4432 - fix client handling of MDS intent errors (POSIX)
4433 - fix permission bug in lovstripe.c test (624321)
4434 - fix MDS thread deadlock - move LDLM handler to DLM portal (625069)
4435 - truncate past end of file could corrupt data
4436 - proper cleanup after timeouts, crashes, etc (592524, 550815)
4437 - a race in recovery could return ETIMEDOUT to apps (623947)
4438 - building outside the source directory was fixed
4439 * the lustre-HOWTO was brought (more) up to date (582544)
4440 * major progress was made on recovery functionality
4442 2002-10-10 Phil Schwan <phil@clusterfs.com>
4445 - recovery deadlock fix
4446 - rm -rf causes LBUG fix (617817)
4447 - file open by multiple tasks fix (618962)
4448 - directory permissions bugs (602707 and 620007)
4449 - journal_stop fixed with locking (611313)
4450 - O_APPEND failures resolved (618273, perhaps 614459)
4451 - lconf PATH fix (619770)
4452 - IA64 build fix (621450)
4453 - RPC buffer sizes scale with amount of memory
4455 2002-10-01 Phil Schwan <phil@clusterfs.com>
4458 - locks would be cancelled without throwing away data pages,
4459 resulting in inconsistent data (605627)
4460 - inode attributes were not always being refreshed (605627, 612449)
4461 - lconf now continues to cleanup after lctl reports an error
4462 - MDS now enforces user permissions (602707)
4463 - lprocfs cleanup fixed, but not yet enabled (614157)
4464 - fixed infinite server hang, should a client not respond to an AST
4465 - avoid going into recovery if user calls readlink() with a buffer
4466 that's too small (613941)
4467 - AST RPCs no longer require replies (614867) -- this may be changed
4468 - don't crash server if client sends an IOV that's too big (611336)
4469 - fixed lock conversion deadlock (611892)
4470 - fixed the following of symlinks (614622)
4471 * recovery: the server can remove locks from a client that dies, other
4472 clients can make progress
4473 * more extN patch fixes
4474 * compile-time configurable ptlrpc buffer allocations
4476 - collaborative read cache document
4477 - Lustre Lite Performance CDR document-in-progress
4479 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
4482 - fix typo in patch-2.4.18
4484 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
4487 - clear ptlrpc request each time in handle_incoming_request()
4488 - unlink of files now destroys the object on the OST
4490 2002-09-19 Peter Braam <braam@clusterfs.com>
4492 * add hard link support
4493 * change obdfile creation method
4494 * kernel patch changed
4496 2002-09-19 Peter Braam <braam@clusterfs.com>
4499 - stack overflow bug in extN fixed
4501 2002-09-18 Andreas Dilger <adilger@clusterfs.com>
4503 * documentation updates
4504 - add man pages for config tools
4505 - update tests/README to describe testing with new config tools
4506 - finish metadata API descriptions
4507 * bug fixes and cleanups
4508 - statfs workaround for 16TB limit
4509 - LOV stripe allocation improved, can stripe on subset of OSTs
4510 - LOV file size/IO offset was wrong for files > 4GB in size
4511 - object EA data was being dropped, caused files to be unreadable
4512 - memory overflow with non-LOV OST caused memory corruption
4513 - fixed regression tests to work with new config tools, obdfilter
4514 - fixed bug when directory size became larger than 1 block
4515 - fixed bug (for single client case) when PWD was deleted
4516 - invalidate local directory pages when doing intent-based ops
4517 - avoid LDLM oops when lock callback contained bad data
4519 2002-09-09 Andreas Dilger <adilger@clusterfs.com>
4521 * documentation updates
4522 * bug fixes and cleanups
4523 - configuration tools
4526 - 64-bit compile warnings
4527 - 64-bit internal statfs data
4529 * test_brw on persistent OST devices
4531 * lprocfs (disabled)
4533 2002-09-04 Andreas Dilger <adilger@clusterfs.com>
4535 * documentation updates
4536 * bug fixes and cleanups
4537 * configuration tools
4539 2002-08-30 Peter J. Braam <braam@clusterfs.com>
4542 * many small fixes to 0_5_4
4543 * io/network handling
4544 * thinkos in MDS operations
4546 2002-08-24 Peter J. Braam <braam@clusterfs.com>
4549 * crucial basic fixes to 0.5.3
4550 * IOR, Iozone work over Elan
4553 2002-08-07 Phil Schwan <phil@clusterfs.com>
4554 * version 0_5_3, our first alpha
4555 * we use the new Portals iovs
4556 * documentation updates
4557 * bug fixes and cleanups
4558 * small changes in the DLM wire protocol
4560 2002-07-25 Peter J. Braam <braam@clusterfs.com>
4561 * version 0_5_1 with some initial stability,
4562 * locking on MD and file I/O.
4563 * documentation updates
4564 * several bug fixes since 0.5.0
4565 * small changes in wire protocol
4567 2002-07-18 Phil Schwan <phil@clusterfs.com>
4569 * delivered as Lustre Light Alpha
4570 * fixed a crash after handling invalid MDS requests
4571 * fixed directory pages for architectures with non-4k pages sizes
4573 2002-07-11 Andreas Dilger <adilger@clusterfs.com>
4574 * release version v0_4_4
4575 * Moves TCP acceptor to be on port 2432 (unused Coda port) instead
4577 * Fixes a number of interruption problems with OST operations.
4578 * Update documentation for portals header changes
4579 * Move all wire protocol structs/defines to lustre_idl.h
4580 * Fixes symlink length bug.
4581 * Add tcpdump to repository.
4583 2002-07-05 Andreas Dilger <adilger@clusterfs.com>
4584 * release version v0_4_3
4585 * Fixes statfs for inodes on extN.
4586 * Fixes bug in runtests which would delete /etc/hosts.
4587 * Use 64-bit object IDs wherever possible (not into VFS though)
4588 Remove ost_get_info, which is unused by lustre, and out of date.
4590 2002-07-03 Peter Braam <braam@clusterfs.com>
4591 * release version v0_4_2 Fixes a lookup error (type not passed)
4592 * move forward to head of Portals
4593 * move forward to latest Lustre kernel
4595 2002-06-25 Peter Braam <braam@clusterfs.com>
4596 * release version v0_4_1. Hopefully stable on single node use.