1 tbd Sun Microsystems, Inc.
4 2.6.16.54-0.2.5 (SLES 10),
5 2.6.18-53.1.21.el5 (RHEL 5),
6 2.6.22.14 vanilla (kernel.org).
7 * Client support for unpatched kernels:
8 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
9 2.6.16 - 2.6.21 vanilla (kernel.org)
10 * Recommended e2fsprogs version: 1.40.7-sun3
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 Description: sanity test 65a fails if stripecount of -1 is set
18 Details : handle -1 striping on filesystem in ll_dirstripe_verify
23 Description: ASSERTION(CheckWriteback(page,cmd)) failed
24 Details : badly clear PG_Writeback bit in ll_ap_completion can produce false
27 Severity : enhancement
29 Description: Update to RHEL5 kernel-2.6.18-53.1.21.el5.
33 Description: do not process already freed flock
34 Details : flock can possibly be freed by another thread before it reaches
35 to ldlm_flock_completion_ast.
39 Description: LBUG during stress test
40 Details : Need properly lock accesses the flock deadlock detection list.
44 Description: oops in page fault handler
45 Details : kernel page fault handler can return two special 'pages' in error case, don't
46 try dereference NOPAGE_SIGBUS and NOPAGE_OMM.
50 Description: timeout with invalidate import.
51 Details : ptlrpcd_check call obd_zombie_impexp_cull and wait request which should be
52 handled by ptlrpcd. This produce long age waiting and -ETIMEOUT
53 ptlrpc_invalidate_import and as result LASSERT.
55 Severity : enhancement
57 Description: Update to RHEL5 kernel-2.6.18-53.1.19.el5.
61 Description: enable MGS and MDT services start separately
62 Details : add a 'nomgs' option in mount.lustre to enable start a MDT with
63 a co-located MGS without starting the MGS, which is a complement
64 to 'nosvc' mount option.
68 Frequency : after recovery
69 Description: precreate to many object's after del orphan.
70 Details : del orphan st in oscc last_id == next_id and this triger growing
71 count of precreated objects. Set flag LOW to skip increase count
72 of precreated objects.
76 Frequency : rare, on clear nid stats
77 Description: ASSERTION(client_stat->nid_exp_ref_count == 0)
78 Details : when clean nid stats sometimes try destroy live entry,
79 and this produce panic in free.
83 Description: Stack overflow during MDS log replay
84 ease stack pressure by using a thread dealing llog_process.
88 Description: wait until IO finished before start new when do lock cancel.
89 Details : VM protocol want old IO finished before start new, in this case
90 need wait until PG_writeback is cleared until check dirty flag and
91 call writepages in lock cancel callback.
93 Severity : enhancement
95 Description: using special macro for print time and cleanup in includes.
99 Description: mds_mfd_close() ASSERTION(rc == 0)
100 Details : In mds_mfd_close(), we need protect inode's writecount change
101 within its orphan write semaphore to prevent possible races.
105 Description: Obsolete CURRENT_SECONDS and use cfs_time_current_sec() instead.
109 Frequency : rare, on shutdown ost
110 Description: don't hit live lock with umount ost.
111 Details : shrink_dcache_parent can be in long loop with destroy dentries,
112 use shrink_dcache_sb instead.
116 Description: don't panic with use echo client
117 Details : echo client pass NULL as client nid pointer and this produce null
122 Description: fix build on ppc32
123 Details : compile code with -m64 flag produce wrong object file for ppc32.
127 Description: add message levels for liblustreapi
131 Description: fix for occasional failure case of -ENOSPC in recovery-small tests
132 Details : Move the 'good_osts' check before the 'total_bavail' check. This
133 will result in an -EAGAIN and in the exit call path we call
134 alloc_rr() which will with increasing aggressiveness attempt to
135 aquire precreated objects on the minimum number of required OSCs.
139 Description: Use old size assignment to avoid deadlock
140 Details : This reverts the changes in bugs 2369 and bug 14138 that introduced
141 the scheduling while holding a spinlock. We do not need locking
142 for size in ll_update_inode() because size is only updated from
143 the MDS for directories or files without objects, so there is no
144 other place to do the update, and concurrent access to such inodes
145 are protected by the inode lock.
149 Description: resolve "_IOWR redefined" build error on SLES10
153 Description: dump the memory debugging after all modules are unloaded to
154 suppress false negative in conf_sanity test 39
156 Severity : enhancement
158 Description: build kernel-ib packages for OFED 1.3 in our release cycle
163 Description: fix SLES kernel versioning
164 Details : the kernel version for our SLES 10 kernel did not include a "-"
165 before the "smp" at the end. while this was not a problem in
166 general, it did mean that software trying to use the kernel
167 version to try to detect a vendor specific kernel would fail.
168 this was most evident by the OFED build scripts.
172 Description: Don't update lov_desc members until making sure they are valid
173 Details : When updating lov_desc members via proc fs, need fix their
174 validities before doing the real update.
178 Description: don't put request into delay list while invalidate in flight.
179 Details : ptlrpc_delay_request sometimes put in delay list while invalidate
180 import in flight. this produce timeout for invalidate and sometimes
181 can cause stale data.
185 Frequency : on ppc only
186 Description: not convert ost objects for directory because it's not exist.
187 Details : ll_dir_getstripe assume dirrectory has ost objects but this wrong.
191 Description: Add FMODE_EXEC file flag for SLES10 SP1 kernel.
193 Severity : enhancement
195 Description: Update to support 2.6.22.14 vanilla kernel.
199 Frequency : rare, on recovery
200 Description: read procfs can produce deadlock in some situation
201 Details : Holding lprocfs lock which send rpc can produce block for destroy
202 obd objects and this also block reconnect with -EALREADY. This isn't
203 fix all lprocfs bugs - but make it rare.
205 Severity : enhancement
207 Description: Update kernel to RHEL5 2.6.18-53.1.14.el5.
210 Frequency : frequent on X2 node
212 Description: mdc_set_open_replay_data LBUG
213 Details : Set replay data for requests that are eligible for replay.
217 Description: lustre_mgs: operation 101 on unconnected MGS
218 Details : When MGC is disconnected from MGS long enough, MGS will evict the
219 MGC, and late on MGC cannot successfully connect to MGS and a lot
220 of the error messages complaining that MGS is not connected.
224 Frequency : on network error
225 Description: panic with double free request if network error
226 Details : mdc_finish_enqueue is finish request if any network error ocuring,
227 but it's true only for synchronus enqueue, for async enqueue
228 (via ptlrpcd) this incorrect and ptlrpcd want finish request
231 Severity : enhancement
233 Description: client-side metadata stat-ahead during readdir(directory readahead)
234 Details : perform client-side metadata stat-ahead when the client detects
235 readdir and sequential stat of dir entries therein
238 Frequency : on start mds
240 Description: Implement get_info(last_id) in obdfilter.
243 Frequency : occasional
245 Description: Correctly check stale fid, not start epoch if ost not support SOM
246 Details : open with flag O_CREATE need set old fid in op_fid3 because op_fid2
247 overwrited with new generated fid, but mds can anwer with one of these
248 two fids and both is not stale. setattr incorectly start epoch and
249 assume will be called done_writeting, but without SOM done_writing
253 Frequency : rare, depends on device drivers and load
255 Description: MDS or OSS nodes crash due to stack overflow
256 Details : Code changes in 1.8.0 increased the stack usage of some functions.
257 In some cases, in conjunction with device drivers that use a lot
258 of stack the MDS (or possibly OSS) service threads could overflow
259 the stack. One change which was identified to consume additional
260 stack has been reworked to avoid the extra stack usage.
263 Frequency : occasional
265 Description: Do not fail import if osc_interpret_create gets -EAGAIN
266 Details : If osc_interpret_create got -EAGAIN it immediately exits and
267 wakeup oscc_waitq. After wakeup oscc_wait_for_objects call
268 oscc_has_objects and see OSC has no objests and call
269 oscc_internal_create to resend create request.
271 Severity : enhancement
273 Description: Update to SLES10 SP1 latest kernel-2.6.16.54-0.2.5.
275 Severity : enhancement
277 Description: Update to RHEL5 latest kernel-2.6.18-53.1.13.el5.
280 Frequency : very rare
282 Description: Fix replay if there is an un-replied request and open
283 Details : In some cases, older replay request will revert the
284 mcd->mcd_last_xid on MDS which is used to record the client's
287 Severity : enhancement
289 Description: Update to RHEL5 latest kernel-2.6.18-53.1.6.el5.
291 Severity : enhancement
293 Description: Add rhel5 support to HEAD.
295 Serverity : enhancement
297 Description: Update RHEL4 kernel to 2.6.9-67.0.4.
302 Description: Don't allow skipping OSTs if index has been specified.
303 Details : Don't allow skipping OSTs if index has been specified, make locking
304 in internal create lots better.
308 Description: LBUG in ptlrpc_check_set() bad phase ebc0de00
309 Details : access to bitfield in structure is always rounded to long
310 and this produce problem with not atomic change any bit.
314 Description: Lustre make rpms failed.
315 Details : Remove ldiskfs spec file to avoids rpmbuild be confused when
316 builds Lustre rpms from tarball.
319 Frequency : rare on shutdown ost
321 Description: If llog cancel was not send before clean_exports phase, this can
322 produce deadlock in llog code.
323 Details : If llog thread has last reference to obd and call class_import_put
324 this produce deadlock because llog_cleanup_commit_master wait when
325 last llog_commit_thread exited, but this never success because was
326 called from llog_commit_thread.
330 Description: allow userland application know is lost one of stripes.
331 Details : fill lvb_blocks with error code on ost and return it to
332 application if error flag found.
336 Description: NULL lov_tgts causing MDS oops
337 Details : more safe checks for NULL lov_tgts for avoid oops.
339 Severity : enhancement
341 Description: Update to RHEL4 latest kernel-2.6.9-67.0.1.EL.
345 Descriptoin: make lov_create() will not stuck in obd_statfs_rqset()
346 Details : If an OST is down the MDS will hang indefinitely in
347 obd_statfs_rqset() waiting for the statfs data. While for
348 MDS QOS usage of statfs, it should not stuck in waiting.
350 Severity : enhancement
352 Description: remote_acl support
353 Details : Support ACL-based permission check for remote user.
354 Support setfacl/getfacl for remote user with the utils
355 "lfs {l,r}{s,g}etfacl" which follow the same parameter format as
356 the system "{s,g}etfacl" utils.
358 Severity : enhancement
360 Description: Update to RHEL4 U6 kernel-2.6.9-67.EL.
362 Severity : enhancement
364 Description: Update to RHEL5 latest kernel-2.6.18-53.1.4.el5.
368 Description: make mgs_setparam() handle fsname containing dash
369 Details : fsname containing a dash does not work with lctl conf_param
371 Severity : enhancement
373 Description: Update to SLES10 SP1 latest kernel-2.6.16.54-0.2.3.
375 Severity : enhancement
377 Description: Update to RHEL5 Update-1 kernel 2.6.18-53.el5.
381 Frequency : rare, at shutdown
382 Description: access already free / zero obd_namespace.
383 Details : if client_disconnect_export was called without force flag set,
384 and exist connect request in flight, this can produce access to
385 NULL pointer (or already free pointer) when connect_interpret
386 store ocd flags in obd_namespace.
390 Frequency : only at startup
391 Description: not alloc memory with spinlock held.
392 Details : allocation memory with GFP_KERNEL can produce sleep deadlock,
393 if any spinlock held.
395 Severity : enhancement
397 Description: make lustre randomly fail allocating memory
398 Details : Make lustre randomly failed allocating memory for testing purpose.
400 Severity : enhancement
402 Description: lost problems with lov objid file
403 Details : Fixes some scability and access to not inited memory problems
404 in work with lov objdid file.
409 Description: lfs find does not continue on file error
410 Details : Continue other files processing when a file/dir is absent.
414 Description: Inconsistent usage of lustre_pack_reply()
415 Details : Standardize the usage of lustre_pack_reply() such that it
416 always generate a CERROR on failure.
419 Frequency : occasional
421 Description: MDS hang or stay in waiting lock
422 Details : If client receive lock with CBPENDING flag ldlm need send lock
423 cancel as separate rpc, to avoid situation when cancel request
424 can't processed due all i/o threads stay in wait lock.
428 Description: Update to RHEL5 kernel 2.6.18-8.1.15.el5.
432 Description: Update to SLES10 SP1 kernel 2.6.16.53-0.16
436 Description: Update to SLES9 kernel-2.6.5-7.287.3.
440 Description: Update to RHEL4 latest kernel.
442 Severity : enhancement
444 Description: Build SLES10 patchless client fails
445 Details : The configure was broken by run ./configure with
446 --with-linux-obj=.... argument for patchless client. When the
447 configure use --with-linux-obj, the LINUXINCLUDE= -Iinclude
448 can't search header adequately. Use absolute path such as
449 -I($LINUX)/include instead.
453 Description: interrupt oig_wait produce painc on resend.
454 Details : brw_redo_request can be used for resend requests from ptlrpcd and
455 private set, and this produce situation when rq_ptlrpcd_data not
456 copyed to new allocated request and triggered LBUG on assert
457 req->rq_ptlrpcd_data != NULL. But this member used only for wakeup
458 ptlrpcd set if request is changed and can be safety changed to use
463 Description: LASSERT_{REQ,REP}SWAB macros are buggy
464 Details : If SWAB_PARANOIA is disabled, the LASSERT_REQSWAB and
465 LASSERT_REPSWAB macros become no-ops, which is incorrect. Drop
466 these macros and replace them with their difinitions instead.
470 Description: Update kernel patches for SLES10 2.6.16.53-0.8.
471 Details : Update which_patch & target file for SLES10 latest kernel.
474 Description: Remove client patches from SLES 10 kernel.
475 Details : This causes SLES 10 clients to behave as patchless clients
476 even on a Lustre-patched (server) kernel.
478 Severity : enhancement
480 Description: self-adjustable client's lru lists
481 Details : use adaptive algorithm for managing client cached locks lru
482 lists according to current server load, other client's work
483 pattern, memory activities, etc. Both, server and client
484 side namespaces provide number of proc tunables for controlling
487 Severity : enhancement
489 Description: light-weight GSS support
490 Details : Support krb5n and krb5a mode, which keep Kerberos 5 authentication
491 and reduce performance overhead.
493 Severity : enhancement
495 Description: Linux keyring support
496 Details : Support using service of Linux keyring for Lustre GSS internal
497 context refresh/cache mechanism.
501 Description: Fix errors in lfs documentation
502 Details : Fixes man pages
506 Description: don't use GFP_* in generic Lustre code.
507 Details : Use cfs_alloc_* functions and CFS_* flags for code portability.
511 Description: obdclass is limited by single OBD_ALLOC(idarray)
512 Details : replace OBD_ALLOC/OBD_FREE with OBD_VMALLOC/OBD_VFREE
516 Description: warnings with build patchless client with vanila 2.6.19 and up
517 Details : change old ctl_table style and replace ctl_table/ctl_table_header
518 with cfs_sysctl_table_t/cfs_sysctl_table_header_t
522 Frequency : Only for SLES
523 Description: sanity_quota fail test_1
524 Details : There are multiple occurences of $TSTUSR in SLES's /etc/group
525 file, which makes TSTID[2] inunique.
529 Frequency : Only for SLES9
530 Description: Kernel patches for SLES9 2.6.5-7.286 kernel
531 Details : Update target/ChangeLog/which_patch .
535 Frequency : Only for test_10 in sanity-quota.sh
536 Description: a bug in quota test code
537 Details : checking a test flag in a wrong place
541 Frequency : Only for test_13 in sanity-quota.sh
542 Description: a bug in quota test script
543 Details : shouldn't check overfull allocated quota
547 Description: jbd statistics
548 Details : Port older jbd statistics patch for sles10
550 Severity : enhancement
552 Description: Add printing inode info into message about error in writepage.
557 Description: reply_lock_interpret crash due to race with it and lock cancel.
558 Details : Do not replay locks that are being cancelled. Do not reference
559 locks by their address during replay, just by their handle.
563 Frequency : When flocks are used.
564 Description: assertion failure in ldlm_cli_enquque_fini for non NULL lock.
565 Details : Flock locks might destroy just granted lock if it could be merged
566 with another existing flock, this is done in completion handler,
567 so teach ldlm_cli_enquque_fini that this is a valid case for
573 Description: Oops in read and write path when failing to allocate lock.
574 Details : Check if lock allocation failed and return error back.
578 Description: lstripe command fails for valid OST index
579 Details : The stripe offset is compared to lov->desc.ld_tgt_count
580 instead of lov->desc.ld_active_tgt_count.
584 Description: sanity.sh failed test 103
585 Details : RHEL mis-interpret setfacl "-X" param, so we won't test setfacl
590 Description: df doesn't work properly if diskfs blocksize != 4K
591 Details : Choose biggest blocksize of OST's as the LOV's blocksize.
595 Description: lfs find on -1 stripe looping in lsm_lmm_verify_common()
596 Details : Avoid lov_verify_lmm_common() on directory with -1 stripe count.
600 Description: obd_health_check_timeout too short
601 Details : set obd_health_check_timeout as 1.5x of obd_timeout
605 Description: llapi_file_create() does not allow some changes
606 Details : add llapi_file_open() that allows specifying the mode and
607 open flags, and also returns an open file handle.
611 Description: merge and cleanup kernel patches.
612 Details : Remove mnt_lustre_list in vfs_intent-2.6-rhel4.patch.
616 Description: Add journal checksum support.(Kernel part)
617 Details : The journal checksum feature adds two new flags i.e
618 JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT and
619 JBD2_FEATURE_COMPAT_CHECKSUM. JBD2_FEATURE_CHECKSUM flag
620 indicates that the commit block contains the checksum for
621 the blocks described by the descriptor blocks. Now commit
622 record can be sent to disk without waiting for descriptor
623 blocks to be written to disk. This behavior is controlled
624 using JBD2_FEATURE_ASYNC_COMMIT flag.
628 Description: OSS needs mutliple precreate threads
629 Details : Add ability to start more than one create thread per OSS.
633 Description: Sanity.sh test_65k failed: setstripe should have succeeded
634 Details : execute lfs setstripe on client
638 Description: mds_obd_create error creating tmp object
639 Details : When the user sets quota on root, llog will be affected and can't
640 create files and write files.
643 Frequency : Always on ia64 patchless client, and possibly others.
645 Description: Add EXPORT_SYMBOL check for node_to_cpumask symbol.
646 Details : This allows the patchless client to be loaded on architectures
651 Description: RedHat Update kernel for RHEL5
652 Details : Add the kernel config file for RHEL5.
656 Description: RedHat Update kernel for RHEL5
657 Details : Modify the kernel config file more closer RHEL5.
661 Description: Build failure against Centos5 (RHEL5)
662 Details : Define PAGE_SIZE when it isn't present.
666 Description: test_7 on sanity-quota.sh can't be used on separate machines
667 Details : add facet to handle it
671 Description: "ll_intent_file_open()) lock enqueue: err: -13" with nfs
672 Details : with NFS, the anon dentry's parent was set to itself in
673 d_alloc_anon(), so in MDS, we use rec->ur_fid1 to find the
674 corresponding dentry other than use rec->ur_name.
676 Severity : enhancement
678 Description: lfs setstripe enhancement
679 Details : Make lfs setstripe understand 'k', 'm' and 'g' for stripe size.
683 Description: enable data checksumming by default
684 Details : enable checksum by default, allow --disable-checksum
685 configure option and "-o nochecksum" mount option. Checksums
686 can also be disabled at runtime via $LPROC/osc/*/checksum_pages.
690 Description: lustre support for RHEL5
691 Details : Add support for RHEL5.
695 Description: Client eviction due to failover config
696 Details : after a connection loss, the lustre client should attempt to
697 reconnect to the last active server first before trying the
698 other potential connections.
702 Description: when mds and osts use different quota unit(32bit and 64bit),
703 quota will be released repeatly.
704 Details : void sending multiple quota reqs to mds, which will keep the
705 status between the reqs.
709 Description: osts not allocated evenly to files
710 Details : change the condition to increase offset_idx
713 Frequency : Always for filesystems larger than 2TB on 32-bit systems.
715 Description: Data corruption for OSTs that are formatted larger than 2TB
717 Details : When generating the bio request for lustre file writes the
718 sector number would overflow a temporary variable before being
719 used for the IO. The data reads correctly from Lustre (which
720 will overflow in a similar manner) but other file data or
721 filesystem metadata may be corrupted in some cases.
725 Description: Tune the kernel for good SCSI performance.
726 Details : Set the value of /sys/block/{dev}/queue/max_sectors_kb
727 to the value of /sys/block/{dev}/queue/max_hw_sectors_kb
732 Description: rewrite ext2-derived code in obdclass/uuid.c
733 Details : rewrite inherited code (uuid parsing code from ext2 utils)
734 from scratch preserving functionality.
738 Description: "lfs find -obd UUID" prints directories
739 Details : "lfs find -obd UUID" will return all directory names instead
740 of just file names. It is incorrect because the directories
741 do not reside on the OSTs.
745 Description: use i_size_read and i_size_write in 2.6 port
746 Details : replace inode->i_size access with i_size_read/write()
749 Frequency : when removing large files
751 Description: scheduling issue during removal of large Lustre files
752 Details : Don't take the BKL in fsfilt_ext3_setattr() for 2.6 kernels.
753 It causes scheduling issues when removing large files (17TB in the
757 Frequency : only with liblustre clients on XT3
759 Description: evictions taking too long
760 Details : allow llrd to evict clients directly on OSTs
763 Frequency : only on ppc
765 Description: /proc/fs/lustre/devices broken on ppc
766 Details : The patch as applied to 1.6.2 doesn't look correct for all arches.
767 We should make sure the type of 'index' is loff_t and then cast
768 explicitly as needed below. Do not assign an explicitly cast
772 Frequency : only for rhel5
774 Description: Kernel patches update for RHEL5 2.6.18-8.1.10.el5.
775 Details : Modify the target file & which_kernel.
777 Severity : enhancement
779 Description: omit set fsid for export NFS
780 Details : fix set/restore device id for avoid EMFILE error and mark lustre fs
781 as FS_REQUIRES_DEV for avoid problems with generate fsid.
785 Frequency : Always, for kernels after 2.6.16
786 Description: Fix warning idr_remove called for id=.. which is not allocated.
787 Details : Last kernels save old s_dev before kill super and not allow
788 to restore from callback - restore it before call kill_anon_super.
792 Description: buffer overruns could theoretically occur
793 Details : llapi_semantic_traverse() modifies the "path" argument by
794 appending values to the end of the origin string, and a
795 overrun may occur. Adding buffer overrun check in liblustreapi.
799 Description: Fix error on 'ls .' at the top of the Lustre mount.
800 Details : Don't revalidate dentry if it is a root dentry.
804 Description: Kernel patches update for RHEL4 2.6.9-55.0.6.
805 Details : Modify vm-tunables-rhel4.patch.
809 Description: Kernel config for 2.6.18-vanilla.
810 Details : Modify targets/2.6-vanilla.target.in.
811 Add config file kernel-2.6.18-2.6-vanilla-i686.config.
812 Add config file kernel-2.6.18-2.6-vanilla-i686-smp.config.
813 Add config file kernel-2.6.18-2.6-vanilla-x86_64.config.
814 Add config file kernel-2.6.18-2.6-vanilla-x86_64-smp.config.
818 Description: improve handling recoverable errors
819 Details : if request processig with error which can be recoverable on server
820 request should be resend, otherwise page released from cache and
825 Description: Kernel patches update for RHEL5 2.6.18-8.1.14.el5.
826 Details : Modify target file & which_patch.
827 A flaw was found in the IA32 system call emulation provided
828 on AMD64 and Intel 64 platforms. An improperly validated 64-bit
829 value could be stored in the %RAX register, which could trigger an
830 out-of-bounds system call table access. An untrusted local user
831 could exploit this flaw to run code in the kernel
832 (ie a root privilege escalation). (CVE-2007-4573).
836 Description: O_DIRECT bypasses client statistics.
837 Details : When running with O_DIRECT I/O, neither the client rpc_stats nor
838 read_ahead_stats were updated. copied the stats section from
839 osc_send_oap_rpc() into async_internal().
843 Description: Add jbd statistics patch for RHEL5 and 2.6.18-vanilla
847 Description: change order of libsysio includes
848 Details : '#include sysio.h' should always come before '#include xtio.h'
852 Description: handle "serious error: objid * already exists" more gracefully
853 Details : If LAST_ID value on disk is smaller than the objects existing in
854 the O/0/d* directories, it indicates disk corruption and causes an
855 LBUG(). If the object is 0-length, then we should use the existing
856 object. This will help to avoid a full fsck in most cases.
858 Severity : enhancement
860 Description: adapt the lustre_config script to support the upgrade case
861 Details : Add "-u" option for lustre_config script to support upgrading 1.4
862 server targets to 1.6 in parallel.
866 Description: To avoid grant space > avaible space when the disk is almost
867 full. Without this patch you might see the error "grant XXXX >
868 available" or some LBUG about grant, when the disk is almost
870 Details : In filter_check_grant, for non_grant cache write, we should
871 check the left space by if (*left > ungranted + bytes), instead
872 of (*left > ungranted), because only we are sure the left space
873 is enough for another "bytes", then the ungrant space should be
874 increase. In client, we should update cl_avail_grant only there
875 is OBD_MD_FLGRANT in the reply.
879 Description: Update RHEL 4 kernel to fix local root privilege escalation.
880 Details : Update to the latest RHEL 4 kernel to fix the vulnerability
881 described in CVE-2007-4573. This problem could allow untrusted
882 local users to gain root access.
885 Frequency : when using O_DIRECT and quotas
887 Description: Incorrect file ownership on O_DIRECT output files
888 Details : block usage reported by 'lfs quota' does not take into account
889 files that have been written with O_DIRECT.
894 Description: touch file failed when fs is not full
895 Details : OST in recovery should not be discarded by MDS in alloc_qos(),
896 otherwise we can get ENOSP while fs is not full.
900 Description: parallel lock callbacks
901 Details : Instead of sending blocking and completion callbacks as separated
902 requests, adding them to a set and sending in parallel.
905 Frequency : only for Cray XT3
906 Bugzilla : 12829/13455
907 Description: Changing primary group doesn't change the group lustre assigns to
909 Details : When CRAY_XT3 is defined, the fsgid supplied by the client is
910 overridden with the primary group provided by the group upcall,
911 whereas the supplied fsgid can be trusted if it is in the list of
912 supplementary groups returned by the group upcall.
914 Severity : enhancement
916 Description: Allow masking D_WARNING, D_ERROR messages from console
917 Details : Console messages can now be disabled via lnet.printk.
921 Description: User code with malformed file open parameter crashes client node
922 Details : Before packing join_file req, all the related reference should be
923 checked carefully in case some malformed flags cause fake
924 join_file req on client.
928 Description: LDLM_ENQUEUE races with LDLM_CP_CALLBACK
929 Details : ldlm_completion_ast() assumes that a lock is granted when the req
930 mode is equal to the granted mode. However, it should also check
931 that LDLM_FL_CP_REQD is not set.
935 Description: Heavy nfs access might result in deadlocks
936 Details : After ELC code landed, it is now improper to enqueue any mds
937 locks under och_sem, because enqueue might want to decide to
938 cancel open locks for same inode we are holding och_sem for.
942 Description: Client eviction while running blogbench
943 Details : A lot of unlink operations with concurrent I/O can lead to a
944 deadlock causing evictions. To address the problem, the number of
945 oustanding OST_DESTROY requests is now throttled to
946 max_rpcs_in_flight per OSC and LDLM_FL_DISCARD_DATA blocking
947 callbacks are processed in priority.
951 Description: enable ACLs on MDS by default
952 Details : ACLs must be enabled on MDS by default.
955 Frequency : PPC/PPC64 only
957 Description: conflicts between asm-ppc64/types.h and lustre_types.h
958 Details : fix duplicated definitions between asm-ppc64/types.h and
959 lustre_types.h on PPC.
962 Frequency : PPC/PPC64 only
964 Description: asm-ppc/segment.h does not exist
965 Details : fix compile issue on PPC.
969 Description: better handle error messages in extents code
972 Frequency : RHEL4 only
974 Description: mkfs is very slow on IA64/RHEL4
975 Details : A performance regression has been discovered in the MPT Fusion
976 driver between versions 3.02.73rh and 3.02.99.00rh. As a
977 consequence, we have downgraded the MPT Fusion driver in the RHEL4
978 kernel from 3.02.99.00 to 3.02.73 until this problem is fixed.
980 Severity : enhancement
982 Description: SNMP support enhancement
983 Details : Adding total number of sampled request for an MDS node in snmp
986 Severity : enhancement
988 Description: Optimize ldlm waiting list processing for PR extent locks
989 Details : When processing waiting list for read extent lock and meeting read
990 lock that is same or wider to it that is not contended, skip
991 processing rest of the list and immediatelly return current
992 status of conflictness, since we are guaranteed there are no
993 conflicting locks in the rest of the list.
997 Description: Time out and refuse to reconnect
998 Details : When the failover node is the primary node, it is possible
999 to have two identical connections in imp_conn_list. We must
1000 compare not conn's pointers but NIDs, otherwise we can defeat
1001 connection throttling.
1005 Description: port llog fixes from b1_6 into HEAD
1006 Details : Port llog reference couting and some llog cleanups from b1_6
1007 (bug 10800) into HEAD, for protect from panic and access to already
1008 free llog structures.
1012 Description: Detect stride IO mode in read-ahead
1013 Details : When a client does stride read, read-ahead should detect that and
1014 read-ahead pages according to the detected stride pattern.
1018 Description: data checksumming impacts single node performance
1019 Details : add support for several checksum algorithm. Currently, only CRC32
1020 and Adler-32 are supported. The checksum type can be changed on
1021 the fly via /proc/fs/lustre/osc/*/checksum_type.
1025 Description: use adler32 for page checksums
1026 Details : when available, use the Adler-32 algorithm instead of CRC32 for
1031 Description: build for x2 fails
1032 Details : fix compile issue on Cray systems.
1036 Description: Properly match for duplicate locks
1037 Details : Due to different lock order from skiplists code, we need to
1038 traverse entire list for now
1041 Frequency : only on PPC/SLES10
1043 Description: "BITS_PER_LONG is not 32 or 64" in linux/idr.h
1044 Details : On SLES10/PPC, fs.h includes idr.h which requires BITS_PER_LONG to
1045 be defined. Add a hack in mkfs_lustre.c to work around this compile
1050 Description: LASSERT on MDS when client holding flock lock dies
1051 Details : ldlm pool logic depends on number of granted locks equal to
1052 number of released locks which is not true for flock locks, so
1053 just exclude such locks from consideration.
1057 Description: MDS deadlock with many ll_sync_lov threads and I/O stalled
1058 Details : Use fsfilt_sync() for both the whole filesystem sync and
1059 individual file sync to eliminate dangerous inode locking
1060 with I_LOCK that can lead to a deadlock.
1064 Description: performance in 1.6.3
1065 Details : Force q->max_phys_segments to MAX_PHYS_SEGMENTS on SLES10 to be
1066 sure that 1MB requests are not fragmented by the block layer.
1068 Severity : enhancement
1070 Description: organize the server-side client stats on per-nid basis
1071 Details : Change the structure of stats under obdfilter and mds to
1079 The "uuid"s file would list the uuids of _active_ exports.
1080 And the clear entry is to clear all stats and stale nids.
1082 Severity : enhancement
1084 Description: eliminate client locks in face of contention
1085 Details : file contention detection and lockless i/o implementation
1086 for contended files.
1090 Description: Reinitialize optind to 0 so that interactive lfs works in all cases
1093 Frequency : very rare, if additional xattrs are used on kernels >= 2.6.12
1095 Description: MDS may lose file striping (and hence file data) in some cases
1096 Details : If there are additional extended attributes stored on the MDS,
1097 in particular ACLs, SELinux, or user attributes (if user_xattr
1098 is specified for the client mount options) then there is a risk
1099 of attribute loss. Additionally, the Lustre file striping
1100 needs to be larger than default (e.g. striped over all OSTs),
1101 and an additional attribute must be stored initially in the
1102 inode and then increase in size enough to be moved to the
1103 external attribute block (e.g. ACL growing in size) for file
1108 Description: skiplist implementation simplification
1109 Details : skiplists are used to group compatible locks on granted list
1110 that was implemented as tracking first and last lock of each lock group
1111 the patch changes that to using doubly linked lists
1115 Description: MDS LBUG: ASSERTION(!IS_ERR(dchild))
1116 Details : Change LASSERTs to client eviction (i.e. abort client's recovery)
1117 because LASSERT on both the data supplied by a client, and the data
1118 on disk is dangerous and incorrect.
1120 Severity : enhancement
1122 Description: Slow truncate/writes to huge files at high offsets.
1123 Details : Directly associate cached pages to lock that protect those pages,
1124 this allows us to quickly find what pages to write and remove
1125 once lock callback is received.
1129 Description: more ldlm soft lockups
1130 Details : In ldlm_resource_add_lock(), call to ldlm_resource_dump()
1131 starve other threads from the resource lock for a long time in
1132 case of long waiting queue, so change the debug level from
1133 D_OTHER to the less frequently used D_INFO.
1135 Severity : enhancement
1137 Description: add -gid, -group, -uid, -user options to lfs find
1141 Description: Hung threads in invalidate_inode_pages2_range
1142 Details : The direct IO path doesn't call check_rpcs to submit a new RPC once
1143 one is completed. As a result, some RPCs are stuck in the queue
1148 Description: filter threads hungs on waiting journal commit
1149 Details : Cleanup filter group llog code, then only filter group llog will
1150 be only created in the MDS/OST syncing process.
1152 --------------------------------------------------------------------------------
1154 2007-08-10 Cluster File Systems, Inc. <info@clusterfs.com>
1156 * Support for kernels:
1157 2.6.5-7.283 (SLES 9),
1158 2.6.9-55.EL (RHEL 4),
1159 2.6.16.46-0.14 (SLES 10),
1160 2.6.18.8 vanilla (kernel.org)
1161 * Client support for unpatched kernels:
1162 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
1163 2.6.16 - 2.6.22 vanilla (kernel.org)
1164 * Due to recently discovered recovery problems, we do not recommend
1165 using patchless RHEL 4 clients with this or any earlier release.
1166 * Recommended e2fsprogs version: 1.39.cfs8
1167 * Note that reiserfs quotas are disabled on SLES 10 in this kernel.
1168 * Starting with this release, the ldiskfs backing filesystem required
1169 by Lustre is now in its own package, lustre-ldiskfs. This package
1170 should be installed. It is versioned separately from Lustre and
1171 may be released separately in future.
1175 Description: block reactivating mgc import until all deactivates complete
1176 Details : Fix race when failing back MDT/MGS to itself (testing)
1179 Frequency : at statup only
1181 Description: mds_lov_synchronize race leads to various problems
1182 Details : simultaneous MDT->OST connections at startup can cause the
1183 sync to abort, leaving the OSC in a bad state.
1185 Severity : enhancement
1187 Description: add optional extra BUILD_VERSION info
1188 Details : add a new environment variable (namely LUSTRE_VERS) which allows
1189 to override the lustre version.
1192 Frequency : 2.6.18 servers only
1194 Description: ll_kern_mount() doesn't release the module reference
1195 Details : The ldiskfs module reference count never drops down to 0
1196 because ll_kern_mount() doesn't release the module reference.
1201 Description: server LBUG when using old ost_num_threads parameter
1202 Details : Accept the old ost_num_threads parameter but warn that it
1203 is deprecated, and fix an off-by-one error that caused an LBUG.
1208 Description: Transient SCSI error results in persistent IO issue
1209 Details : iobuf->dr_error is not reinitialized to 0 between two
1213 Frequency : sometimes when underlying device returns I/O errors
1215 Description: OSTs not going read-only during write failures
1216 Details : OSTs are not remounted read-only when the journal commit threads
1217 get I/O errors because fsfilt_ext3 calls journal_start/stop()
1218 instead of the ext3 wrappers.
1222 Description: poor connect scaling with increasing client count
1223 Details : Don't run filter_grant_sanity_check for more than 100 exports
1224 to improve scaling for large numbers of clients.
1227 Frequency : SLES10 only
1229 Description: sanity-quota.sh quotacheck failed: rc = -22
1230 Details : Quotas cannot be enabled on SLES10.
1233 Frequency : liblustre clients only
1235 Description: getdirentries does not give error when run on compute nodes
1236 Details : getdirentries does not fail when the size specified as an argument
1237 is too small to contain at least one entry
1239 Severity : enhancement
1241 Description: Add LNET router traceability for debug purposes
1242 Details : If a checksum failure occurs with a router as part of the
1243 IO path, the NID of the last router that forwarded the bulk data
1244 is printed so it can be identified.
1249 Description: OST "spontaneously" evicts client; client has imp_pingable == 0
1250 Details : Due to a race condition, liblustre clients were occasionally
1251 evicted incorrectly.
1253 Severity : enhancement
1255 Description: lfs setstripe use optional parameters instead of postional
1258 Severity : enhancement
1260 Description: Nanosecond timestamp support for ldiskfs
1261 Details : The on-disk ldiskfs filesystem has added support for nanosecond
1262 resolution timestamps. There is not yet support for this at
1263 the Lustre filesystem level.
1266 Frequency : during server recovery
1268 Description: MDS failing to send precreate requests due to OSCC_FLAG_RECOVERING
1269 Details : request with rq_no_resend flag not awake l_wait_event if they get a
1273 Frequency : nfs export on patchless client
1275 Description: connectathon hang when test nfs export over patchless client
1276 Details : Disconnected dentry cannot be found with lookup, so we do not need
1277 to unhash it or make it invalid
1280 Description: fix llapi_lov_get_uuids() to allow many OSTs to be returned
1281 Details: : Change llapi_lov_get_uuids() to read the UUIDs from /proc instead
1282 of using an ioctl. This allows lfsck for > 160 OSTs to succeed.
1287 Description: open req refcounting wrong on reconnect
1288 Details : If reconnect happened between getting open reply from server and
1289 call to mdc_set_replay_data in ll_file_open, we will schedule
1290 replay for unreferenced request that we are about to free.
1291 Subsequent close will crash in variety of ways.
1292 Check that request is still eligible for replay in
1293 mdc_set_replay_data().
1298 Description: disable writes to filesystem when reading health_check file
1299 Details : the default for reading the health_check proc file has changed
1300 to NOT do a journal transaction and write to disk, because this
1301 can cause reads of the /proc file to hang and block HA state
1302 checking on a healthy but otherwise heavily loaded system. It
1303 is possible to return to the previous behaviour during configure
1304 with --enable-health-write.
1306 Severity : enhancement
1308 Description: 64-bit inode version
1309 Details: : Add a on-disk 64-bit inode version for ext3 to track changes made
1310 to the inode. This will be required for version-based recovery.
1315 Description: MDS fails to start if a duplicate client export is detected
1316 Details : in some rare cases it was possible for a client to connect to
1317 an MDS multiple times. Upon recovery the MDS would detect this
1318 and fail during startup. Handle this more gracefully.
1320 Severity : enhancement
1322 Description: Add -o localflock option to simulate old noflock
1324 Details : This will achieve local-only flock/fcntl locks
1330 Description: log_commit_thread vs filter_destroy race leads to crash
1331 Details : Take import reference before releasing llog record semaphore
1336 Description: Wrong request locking in request set processing
1337 Details : ptlrpc_check_set wrongly uses req->rq_lock for proctect add to
1338 imp_delayed_list, in this place should be used imp_lock.
1341 Frequency : when reconnection
1343 Description: Grant Leak when osc reconnect to OST
1344 Details : When osc reconnect ost, OST(filter) should check whether it
1345 should grant more space to client by comparing fed_grant and
1346 cl_avail_grant, and return the granted space to client instead
1347 of "new granted" space, because client will call osc_init_grant
1348 to update the client grant space info.
1351 Frequency : when client reconnect to OST
1353 Description: Grant Leak when osc do resend and replay bulk write
1354 Details : When osc reconnect to OST, OST(filter)should clear grant info of
1355 bulk write request, because the grant info while be sync between
1356 OSC and OST when reconnect, and we should ignore the grant info
1357 these of resend/replay write req.
1362 Description: Grant space more than avaiable left space sometimes.
1363 Details : When then OST is about to be full, if two bulk writing from
1364 different clients came to OST. Accord the avaliable space of the
1365 OST, the first req should be permitted, and the second one
1366 should be denied by ENOSPC. But if the seconde arrived before
1367 the first one is commited. The OST might wrongly permit second
1368 writing, which will cause grant space > avaiable space.
1371 Frequency : when client is evicted
1373 Description: Grant might be wrongly erased when osc is evicted by OST
1374 Details : when the import is evicted by server, it will fork another
1375 thread ptlrpc_invalidate_import_thread to invalidate the
1376 import, where the grant will be set to 0. While the original
1377 thread will update the grant it got when connecting. So if
1378 the former happened latter, the grant will be wrongly errased
1379 because of this race.
1384 Description: Checking Stale with correct fid
1385 Details : ll_revalidate_it should uses de_inode instead of op_data.fid2
1386 to check whether it is stale, because sometimes, we want the
1387 enqueue happened anyway, and op_data.fid2 will not be initialized.
1389 Severity : enhancement
1391 Description: update patchless client
1392 Details : Add support for patchless client with 2.6.20, 2.6.21 and RHEL 5
1395 Frequency : only with 2.4 kernel
1397 Description: random memory corruption
1398 Details : size of struct ll_inode_info is to big for union inode.u and this
1399 can be cause of random memory corruption.
1404 Description: Memory leak in recovery
1405 Details : Lov_mds_md was not free in an error handler in mds_create_object.
1406 It should also check obd_fail before fsfilt_start, otherwise if
1407 fsfilt_start return -EROFS,(failover mds during mds recovery).
1408 then the req will return with repmsg->transno = 0 and rc = EROFS.
1409 and we met hit the assert LASSERT(req->rq_reqmsg->transno ==
1410 req->rq_repmsg->transno) in ptlrpc_replay_interpret. Fcc should
1411 be freed no matter whether fsfilt_commit success or not.
1414 Frequency : only with huge count clients
1416 Description: Prevents from taking the superblock lock in llap_from_page for
1418 Details : using LL_ORIGIN_REMOVEPAGE origin flag instead of LL_ORIGIN_UNKNOW
1419 for llap_from_page call in ll_removepage prevents from taking the
1420 superblock lock for a soon died page.
1425 Description: Not check open intent error before release open handle
1426 Details : in some rare cases, the open intent error is not checked before
1427 release open handle, which may cause
1428 ASSERTION(open_req->rq_transno != 0), because it tries to release
1429 the failed open handle.
1434 Description: Set cat log bitmap only after create log success.
1435 Details : in some rare cases, the cat log bitmap is set too early. and it
1436 should be set only after create log success.
1440 Description: Accessing a block bevice can re-enable I/O when Lustre is
1441 tearing down a device.
1442 Details : dev_clear_rdonly(bdev) must be called in kill_bdev() instead of
1447 Description: service threads may hog cpus when there are a lot of requests
1448 Details : Insert cond_resched to give other threads a chance to use some CPU
1453 Description: the cat log was not initialized in recovery
1454 Details : When mds(mgs) do recovery, the tgt_count might be zero, so the
1455 unlink log on mds will not be initialized until mds post
1456 recovery. And also in mds post recovery, the unlink log will
1457 initialization will be done asynchronausly, so there will be race
1458 between add unlink log and unlink log initialization.
1462 Description: brw_stats were being printed incorrectly
1463 Details : brw_stats were being printed as log2 but all of them were not
1464 recorded as log2. Also remove some code duplication arising from
1465 filter_tally_{read,write}.
1469 Frequency : rare, only in recovery.
1470 Description: ASSERTION(req->rq_type != LI_POISON) failed
1471 Details : imp_lock should be held while iterating over imp_sending_list for
1472 prevent destroy request after get timeout in ptlrpc_queue_wait.
1476 Description: replay-single.sh test 52 fails
1477 Details : A lock's skiplist need to be cleanup when it being unlinked
1478 from its resource list.
1482 Description: Short directio read returns full requested size rather than
1484 Details : Direct I/O operations should return actual amount of bytes
1485 transferred rather than requested size.
1487 Severity : enhancement
1489 Description: metadata RPC reduction (e.g. for rm performance)
1490 Details : decrease the amount of synchronous RPC between clients and servers
1491 by canceling conflicing lock before the operation on the client
1492 and packing thier handles into the main operation RPC to server.
1494 Severity : enhancement
1496 Description: Async OSC create to avoid the blocking unnecessarily.
1497 Details : If a OST has no remain object, system will block on the creating
1498 when need to create a new object on this OST. Now, ways use
1499 pre-created objects when available, instead of blocking on an
1500 empty osc while others are not empty. If we must block, we block
1501 for the shortest possible period of time.
1503 Severity : enhancement
1505 Description: refine locking for avoid write wrong info into lov_objid file
1506 Details : fix possible races with add new target and write/update data in
1509 --------------------------------------------------------------------------------
1511 2007-05-03 Cluster File Systems, Inc. <info@clusterfs.com>
1516 Frequency : on some architectures
1518 Description: 1.6 client sometimes fails to mount from a 1.4 MDT
1519 Details : Uninitialized flags sometimes cause configuration commands to
1523 Frequency : patchless clients only
1525 Description: missing __iget() symbol export
1526 Details : The __iget() symbol export is missing. To avoid the need for
1527 this on patchless clients the deathrow inode reaper is turned
1528 off, and we depend on the VM to clean up old inodes. This
1529 dependency was during via the fix for bug 12181.
1531 --------------------------------------------------------------------------------
1533 2007-04-19 Cluster File Systems, Inc. <info@clusterfs.com>
1535 * CONFIGURATION CHANGE. This version of Lustre WILL NOT
1536 INTEROPERATE with older versions automatically. In many cases a
1537 special upgrade step is needed. Please read the
1538 user documentation before upgrading any part of a live system.
1539 * WIRE PROTOCOL CHANGE from previous 1.6 beta versions. This
1540 version will not interoperate with 1.6 betas before beta5 (1.5.95).
1541 * WARNING: Lustre configuration and startup changes are required with
1542 this release. See https://mail.clusterfs.com/wikis/lustre/MountConf
1548 Severity : enhancement
1550 Description: MountConf
1551 Details : Lustre configuration is now managed via mkfs and mount
1552 commands instead of lmc and lconf. New obd types (MGS, MGC)
1553 are added for dynamic configuration management. See
1554 https://mail.clusterfs.com/wikis/lustre/MountConf for
1557 Severity : enhancement
1559 Description: dynamic OST addition
1560 Details : OSTs can now be added to a live filesystem
1562 Severity : enhancement
1564 Description: startup order invariance
1565 Details : MDTs and OSTs can be started in any order. Clients only
1566 require the MDT to complete startup.
1568 Severity : enhancement
1570 Description: parallel, asynchronous orphan cleanup
1571 Details : orphan cleanup is now performed in separate threads for each
1572 OST, allowing parallel non-blocking operation.
1574 Severity : enhancement
1576 Description: optimized stripe assignment
1577 Details : stripe assignments are now made based on ost space available,
1578 ost previous usage, and OSS previous usage, in order to try
1579 to optimize storage space and networking resources.
1581 Severity : enhancement
1583 Description: Permanently set tunables
1584 Details : All writable /proc/fs/lustre tunables can now be permanently
1585 set on a per-server basis, at mkfs time or on a live system.
1587 Severity : enhancement
1589 Description: Lustre message v2
1590 Details : Add lustre message format v2.
1592 Severity : enhancement
1594 Description: client OST exclusion list
1595 Details : Clients can be started with a list of OSTs that should be
1596 declared "inactive" for known non-responsive OSTs.
1600 Description: SPEC SFS validation failure on NFS v2 over lustre.
1601 Details : Changes the blocksize for regular files to be 2x RPC size,
1602 and not depend on stripe size.
1604 Severity : enhancement
1606 Description: Multiple MD RPCs in flight.
1607 Details : Further unserialise some read-only MDS RPCs - learn about intents.
1608 To avoid overly-overloading MDS, introduce a limit on number of
1609 MDS RPCs in flight for a single client and add /proc controls
1610 to adjust this limit.
1612 Severity : enhancement
1614 Description: client read/write statistics
1615 Details : Add client read/write call usage stats for performance
1616 analysis of user processes.
1617 /proc/fs/lustre/llite/*/offset_stats shows non-sequential
1618 file access. extents_stats shows chunk size distribution.
1619 extents_stats_per_process show chunk size distribution per
1622 Severity : enhancement
1624 Description: mds statistics
1625 Details : Add detailed mds operations statistics in
1626 /proc/fs/lustre/mds/*/stats.
1630 Description: Failure of copying files with lustre special EAs.
1631 Details : Client side always return success for setxattr call for lustre
1632 special xattr (currently only "trusted.lov").
1637 Description: Refcount LNET uuids
1638 Details : The global LNET uuid list grew linearly with every startup;
1639 refcount repeated list entries instead of always adding to
1642 Severity : enhancement
1644 Description: Dynamic service threads
1645 Details : Within a small range, start extra service threads
1646 automatically when the request queue builds up.
1649 Frequency : mixed-endian client/server environments
1651 Description: mixed-endian crashes
1652 Details : The new msg_v2 system had some failures in mixed-endian
1655 Severity : enhancement
1657 Description: Easy OST removal
1658 Details : OSTs can be permanently deactivated with e.g. 'lctl
1659 conf_param lustre-OST0001.osc.active=0'
1661 Severity : enhancement
1663 Description: MGS proc entries
1664 Details : Added basic proc entries for the MGS showing what filesystems
1667 Severity : enhancement
1669 Description: provide MGS failover
1670 Details : Added config lock reacquisition after MGS server failover.
1672 Severity : enhancement
1674 Description: add Linux 2.4 support
1675 Details : Added support for RHEL 2.4.21 kernel for 1.6 servers and clients
1679 Description: a large application tries to do I/O to the same resource and dies
1680 in the middle of it.
1681 Details : Check the req->rq_arrival time after the call to
1682 ost_brw_lock_get(), but before we do anything about
1683 processing it & sending the BULK transfer request. This
1684 should help move old stale pending locks off the queue as
1685 quickly as obd_timeout.
1688 Frequency : when an incorrect nid is specified during startup
1690 Description: ptlrpc connect to non-existant node causes kernel crash
1691 Details : LNET can't be re-entered from an event callback, which
1692 happened when we expire a message after the export has been
1693 cleaned up. Instead, hand the zombie cleanup off to another
1696 Severity : enhancement
1698 Description: plain/inodebits lock performance improvement
1699 Details : Grouping plain/inodebits in granted list by their request modes
1700 and bits policy, thus improving the performance of search through
1704 Frequency : only if OST filesystem is corrupted
1706 Description: client incorrectly hits assertion in ptlrpc_replay_req()
1707 Details : for a short time RPCs with bulk IO are in the replay list,
1708 but replay of bulk IOs is unimplemented. If the OST filesystem
1709 is corrupted due to disk cache incoherency and then replay is
1710 started it is possible to trip an assertion. Avoid putting
1711 committed RPCs into the replay list at all to avoid this issue.
1714 Frequency : liblustre (e.g. catamount) on a large cluster with >= 8 OSTs/OSS
1716 Description: System hang on startup
1717 Details : This bug allowed the liblustre (e.g. catamount) client to
1718 return to the app before handling all startup RPCs. This
1719 could leave the node unresponsive to lustre network traffic
1720 and manifested as a server ptllnd timeout.
1722 Severity : enhancement
1724 Description: Add "/proc/sys/lustre/debug_peer_on_timeout"
1725 Details : liblustre envirable: LIBLUSTRE_DEBUG_PEER_ON_TIMEOUT
1726 boolean to control whether to print peer debug info when a
1727 client's RPC times out.
1730 Frequency : only for kernels with patches from Lustre below 1.4.3
1732 Description: Remove old rdonly API
1733 Details : Remove old rdonly API which unused from at least lustre 1.4.3
1736 Frequency : only for devices with external journals
1738 Description: Set external device read-only also
1739 Details : During a commanded failover stop, we set the disk device
1740 read-only while the server shuts down. We now also set any
1741 external journal device read-only at the same time.
1744 Frequency : when upgrading from 1.4 while trying to change parameters
1746 Description: The wrong (new) MDC name was used when setting parameters for
1747 upgraded MDT's. Also allows changing of OSC (and MDC)
1748 parameters if --writeconf is specified at tunefs upgrade time.
1751 Frequency : when setting specific ost indicies
1753 Description: QOS code breaks on skipped indicies
1754 Details : Add checks for missing OST indicies in the QOS code, so OSTs
1755 created with --index need not be sequential.
1757 Severity : enhancement
1759 Description: Add uninit_groups feature to ldiskfs2 to speed up e2fsck
1760 Details : The uninit_groups feature works in conjunction with the kernel
1761 filesystem code (ldiskfs2 only) and e2fsprogs-1.39-cfs6 to speed
1762 up the pass1 processing of e2fsck. This is a read-only feature
1763 in ldiskfs2 only, so older kernels and current ldiskfs cannot
1764 mount filesystems that have had this feature enabled.
1766 Severity : enhancement
1768 Description: Improve multi-block allocation algorithm to avoid fragmentation
1769 Details : The mballoc3 code (ldiskfs2 only) adds new mechanisms to improve
1770 allocation locality and avoid filesystem fragmentation.
1772 ------------------------------------------------------------------------------
1774 2007-04-01 Cluster File Systems, Inc. <info@clusterfs.com>
1776 * Support for kernels:
1777 2.4.21-47.0.1.EL (RHEL 3)
1778 2.6.5-7.283 (SLES 9)
1779 2.6.9-42.0.10.EL (RHEL 4)
1780 2.6.12.6 vanilla (kernel.org)
1781 2.6.16.27-0.9 (SLES 10)
1782 * Recommended e2fsprogs version: 1.39.cfs5
1784 * Note that reiserfs quotas are disabled on SLES 10 in this kernel
1788 Frequency : occasional, depends on client load and configuration
1789 Bugzilla : 12181, 12203
1790 Description: data loss for recently-modified files
1792 Details : In some cases it is possible that recently written or created
1793 files may not be written to disk in a timely manner (this should
1794 normally be within 30s unless client IO load is very high).
1795 The problem appears as zero-length files or files that are a
1796 multiple of 1MB in size after a client crash or client eviction
1797 that are missing data at the end of the file.
1799 This problem is more likely to be hit on clients where files are
1800 repeatedly created and unlinked in the same directory, clients
1801 have a large amount of RAM, have many CPUs, the filesystem has
1802 many OSTs, the clients are rebooted frequently, and/or the files
1803 are not accessed by other nodes after being written.
1805 The presence of the problem can be detected by looking at
1806 /proc/sys/fs/inode-state. If the first number (nr_inodes) is
1807 smaller than the second (nr_unused) then dirty files will not
1808 be flushed automatically to disk. "sync; sleep 10" should be
1809 run several times on the node before unmounting it to update
1810 Lustre (this is also safe to run on nodes without this problem).
1812 There is also a related kernel bug in the RHEL4 4 2.6.9 kernel
1813 that can cause this same problem, so customers using that kernel
1814 also need to update the kernel in addition to Lustre. In order
1815 to properly fix this bug, the RHEL3 2.4.21 kernel is also updated.
1817 It is normal that files written just before a client crash (less
1818 than 30s) may not yet have been flushed to disk, even for local
1822 Frequency : frequent on thin XT3 nodes
1824 Description: UUID collision on thin XT3 Linux nodes
1825 Details : UUIDs on Compute Node Linux XT3 nodes were not generated
1826 randomly, since we relied on an insufficiently-seeded PRNG.
1831 Description: OSS hangs after "All ost request buffers busy"
1832 Details : A deadlock between quota and journal operations caused OSS
1833 hangs after printing "All ost request buffers busy."
1836 Frequency : always on liblustre builds
1838 Description: Cleanup compiler warnings on liblustre
1841 Frequency : always on liblustre builds on XT3
1843 Description: LC_CONFIG_CDEBUG don't run while build liblustre on XT3.
1847 Description: Addition of EXT3_FEATURE_RO_COMPAT_DIR_NLINKS flag for
1848 > 32000 subdirectories
1849 Details : Add EXT3_FEATURE_RO_COMPAT_DIR_NLINK flag to
1850 EXT3_FEATURE_RO_COMPAT_SUPP. This flag will be set whenever
1851 subdirectory count crosses 32000. This will aid e2fsck to
1852 correctly handle more than 32000 subdirectories.
1855 Frequency : liblustre (e.g. catamount) on a large cluster with >= 8 OSTs/OSS
1857 Description: System hang on startup
1858 Details : This bug allowed the liblustre (e.g. catamount) client to
1859 return to the app before handling all startup RPCs. This
1860 could leave the node unresponsive to lustre network traffic
1861 and manifested as a server ptllnd timeout.
1863 Severity : enhancement
1865 Description: Add "/proc/sys/lustre/debug_peer_on_timeout"
1866 (liblustre envirable: LIBLUSTRE_DEBUG_PEER_ON_TIMEOUT)
1867 boolean to control whether to print peer debug info when a
1868 client's RPC times out.
1873 Description: make O_SYNC working on 2.6 kernels
1874 Details : 2.6 kernels use different method for mark pages for write,
1875 so need add a code to lustre for O_SYNC work.
1880 Description: Failure to close file and release space on NFS
1881 Details : Put inode details into lock acquired in ll_intent_file_open.
1882 Use mdc_intent_lock in ll_intent_open to properly
1883 detect all kind of errors unhandled by mdc_enqueue.
1888 Description: proc file read during shutdown sometimes raced obd removal,
1890 Details : Add lock to prevent obd access after proc file removal.
1893 Frequency : Only for files larger than 4GB on 32-bit clients.
1895 Description: improperly doing page alignment of locks
1896 Details : Modify lustre core code to use CFS_PAGE_* defines instead of
1897 PAGE_*. Make CFS_PAGE_MASK a 64-bit mask.
1902 Description: RPCs being resent when they shouldn't be
1903 Details : Some RPCs that should not be resent are being resent. This
1904 can cause inconsistencies in the RPC state machine. Do not
1905 resend such requests.
1908 Frequency : rare, only with NFS export
1910 Description: Crash on NFS re-export node
1911 Details : under very unusual load conditions an assertion is hit in
1912 ll_intent_file_open()
1915 Frequency : only if OST filesystem is corrupted
1917 Description: client incorrectly hits assertion in ptlrpc_replay_req()
1918 Details : for a short time RPCs with bulk IO are in the replay list,
1919 but replay of bulk IOs is unimplemented. If the OST filesystem
1920 is corrupted due to disk cache incoherency and then replay is
1921 started it is possible to trip an assertion. Avoid putting
1922 committed RPCs into the replay list at all to avoid this issue.
1927 Description: large O_DIRECT requests fail under memory pressure/fragmentation
1928 Details : Large single O_DIRECT read and write calls can fail to allocate
1929 a sufficiently large buffer to process the request. In case of
1930 allocation failure the allocation is retried with a smaller
1931 buffer and broken into smaller requests.
1933 Severity : enhancement
1935 Description: Add -o localflock option to simulate old noflock behaviour.
1936 Details : This will achieve local-only flock/fcntl locks coherentness.
1941 Description: versioning check is incomplete
1942 Details : Checking the version difference of client vs. server, report
1943 error if the gap is too big.
1948 Description: add support PG_writeback bit
1949 Details : add support for PG_writeback bit for Lustre, for more carefull
1950 work with page cache in 2.6 kernel. This also fix some deadlocks
1951 and remove hack for work O_SYNC with 2.6 kernel.
1953 Severity : enhancement
1955 Description: Add uninit_groups feature to ldiskfs2 to speed up e2fsck
1956 Details : The uninit_groups feature works in conjunction with the kernel
1957 filesystem code (ldiskfs2 only) and e2fsprogs-1.39-cfs6 to speed
1958 up the pass1 processing of e2fsck. This is a read-only feature
1959 in ldiskfs2 only, so older kernels and current ldiskfs cannot
1960 mount filesystems that have had this feature enabled.
1962 Severity : enhancement
1964 Description: Improve multi-block allocation algorithm to avoid fragmentation
1965 Details : The mballoc3 code (ldiskfs2 only) adds new mechanisms to improve
1966 allocation locality and avoid filesystem fragmentation.
1968 ------------------------------------------------------------------------------
1970 2007-02-09 Cluster File Systems, Inc. <info@clusterfs.com>
1972 * Support for kernels:
1973 2.6.9-42.0.3.EL (RHEL 4)
1974 2.6.5-7.276 (SLES 9)
1975 2.4.21-47.0.1.EL (RHEL 3)
1976 2.6.12.6 vanilla (kernel.org)
1977 2.6.16.21-0.8 (SLES10)
1978 * Recommended e2fsprogs version: 1.39.cfs2-0
1980 * The backwards-compatible /proc/sys/portals symlink has been removed
1981 in this release. Before upgrading, please ensure that you change
1982 any configuration scripts or /etc/sysctl.conf files that access
1983 /proc/sys/portals/* or sysctl portals.* to use the corresponding
1984 entry in /proc/sys/lnet or sysctl lnet.*. This change can be made
1985 in advance of the upgrade on any system running Lustre 1.4.6 or
1986 newer, since /proc/sys/lnet was added in that version.
1987 * Note that reiserfs quotas are disabled on SLES 10 in this kernel
1991 Frequency : only when quota is used
1993 Description: avoid scanning export list for quota master
1994 Details : Change the algorithms to avoid scanning export list in order
1995 to improve the efficiency.
1998 Frequency : MDS failover only, very rarely
2000 Description: "went back in time" messages on mds failover
2001 Details : The greatest transno may be lost when the current operation
2002 finishes with an error (transno==0) and the client's last_rcvd
2003 record is over-written. Save the greatest transno in the
2004 mds_last_transno for this case.
2007 Frequency : always for specific kernels and striping counts
2009 Description: client may get "Matching packet too big" without ACL support
2010 Details : Clients compiled without CONFIG_FS_POSIX_ACL get an error message
2011 when trying to access files in certain configurations. The
2012 clients should in fact be denied when mounting because they do
2013 not understand ACLs.
2016 Frequency : Cray XT3 with more than 4000 clients and multiple jobs
2018 Description: many clients connecting with IO in progress causes connect timeouts
2019 Details : Avoid synchronous journal commits to avoid delays caused by many
2020 clients connecting/disconnecting when bulk IO is in progress.
2021 Queue liblustre connect requests on OST_REQUEST_PORTAL instead of
2022 OST_IO_PORTAL to avoid delays behind potentially many pending
2026 Frequency : occasionally with multiple writers to a single file
2028 Description: shared writes to file may result in wrong size reported by stat()
2029 Details : Allow growing of kms when extent lock is cancelled
2032 Frequency : always with random mmap IO to multi-striped file
2034 Description: mmap write might be lost if we are writing to a 'hole' in stripe
2035 Details : Only if the hole is at the end of OST object so that kms is too
2036 small. Fix is to increase kms accordingly in ll_nopage.
2039 Frequency : rare, only if OST filesystem is inconsistent with MDS filesystem
2041 Description: writes to a missing object would leak memory on the OST
2042 Details : If there is an inconsistency between the MDS and OST filesystems,
2043 such that the MDS references an object that doesn't exist, writes
2044 to that object will leak memory due to incorrect cleanup in the
2045 error handling path, eventually running out of memory on the OST.
2050 Description: Creating too long symlink causes lustre errors
2051 Details : Check symlink and name lengths before sending requests to MDS.
2054 Frequency : only if flock is enabled (not on by default)
2056 Description: posix locks not released on fd closure on 2.6.9+
2057 Details : We failed to add posix locks to list of inode locks on 2.6.9+
2058 kernels, this caused such locks not to be released on fd close and
2059 then assertions on fs unmount about still used locks.
2062 Frequency : MDS failover only, very rarely
2064 Description: clients may get ASSERTION(granted_lock != NULL)
2065 Details : When request was taking a long time, and a client was resending
2066 a getattr by name lock request. The were multiple lock requests
2067 with the same client lock handle and
2068 mds_getattr_name->fixup_handle_for_resent_request found one of the
2069 lock handles but later failed with ASSERTION(granted_lock != NULL).
2074 Description: handle->h_buffer_credits > 0, assertion failure
2075 Details : h_buffer_credits is zero after truncate, causing assertion
2076 failure. This patch extends the transaction or creates a new
2080 Frequency : NFS re-export or patchless client
2081 Bugzilla : 11179, 10796
2082 Description: Crash on NFS re-export node (__d_move)
2083 Details : We do not want to hash the dentry if we don't have a lock.
2084 But if this dentry is later used in d_move, we'd hit uninitialised
2085 list head d_hash, so we just do this to init d_hash field but
2086 leave dentry unhashed.
2089 Frequency : NFS re-export or patchless client
2091 Description: NFS exports has problem with symbolic link
2092 Details : lustre client didn't properly install dentry when re-exported
2093 to NFS or running patchless client.
2096 Frequency : NFS re-export or patchless client
2098 Description: Various nfs/patchless fixes.
2099 Details : fixes reuse disconected alias for lookup process - this fixes
2100 warning "find_exported_dentry: npd != pd",
2101 fix permission error with open files at nfs.
2102 fix apply umask when do revalidate.
2105 Frequency : occasional
2107 Description: Crash on NFS re-export node
2108 Details : calling clear_page() on the wrong pointer triggered oops in
2109 generic_mapping_read().
2112 Frequency : rarely, using O_DIRECT IO
2114 Description: unaligned directio crashes client with LASSERT
2115 Details : check for unaligned buffers before trying any requests.
2118 Frequency : rarely, using CFS RAID5 patches in non-standard kernel series
2120 Description: stale data returned from RAID cache
2121 Details : If only a small amount of IO is done to the RAID device before
2122 reading it again it is possible to get stale data from the RAID
2123 cache instead of reading it from disk.
2126 Frequency : always for sles10 kernel
2128 Description: sles10 support
2129 Details : ll_follow_link: compile fixes and using of nd_set_link
2130 under newer kernels.
2133 Frequency : depends on arch, kernel and compiler version, always on sles10
2136 Description: recursive or deep enough symlinks cause stack overflow
2137 Details : getting rid of large stack-allocated variable in
2141 Frequency : depends on hardware
2143 Description: lustre write performance loss in the SLES10 kernel
2144 Details : the performance loss is caused by using of write barriers in the
2145 ext3 code. The SLES10 kernel turns barrier support on by
2146 default. The fix is to undo that change for ldiskfs.
2148 ------------------------------------------------------------------------------
2150 2006-12-09 Cluster File Systems, Inc. <info@clusterfs.com>
2152 * Support for kernels:
2153 2.6.9-42.0.3EL (RHEL 4)
2154 2.6.5-7.276 (SLES 9)
2155 2.4.21-47.0.1.EL (RHEL 3)
2156 2.6.12.6 vanilla (kernel.org)
2160 Frequency : quota enabled and large files being deleted
2162 Description: releasing more than 4GB of quota at once hangs OST
2163 Details : If a user deletes more than 4GB of files on a single OST it
2164 will cause the OST to spin in an infinite loop. Release
2165 quota in < 4GB chunks, or use a 64-bit value for 1.4.7.1+.
2170 Description: statfs data retrieved from /proc may be stale or zero
2171 Details : When reading per-device statfs data from /proc, in the
2172 {kbytes,files}_{total,free,avail} files, it may appear
2173 as zero or be out of date.
2176 Frequency : systems with MD RAID1 external journal devices
2178 Description: lconf's call to blkid is confused by RAID1 journal devices
2179 Details : Use the "blkid -l" flag to locate the MD RAID device instead
2180 of returning all block devices that match the journal UUID.
2183 Frequency : always, for aggregate stripe size over 4GB
2185 Description: "lfs setstripe" fails assertion when setting 4GB+ stripe width
2186 Details : Using "lfs setstripe" to set stripe size * stripe count over 4GB
2187 will fail the kernel with "ASSERTION(lsm->lsm_xfersize != 0)"
2190 Frequency : always if "lfs find" used on a local file/directory
2192 Description: "lfs find" segfaults if used on a local file/directory
2193 Details : The case where a directory component was not specified wasn't
2194 handled correctly. Handle this properly.
2197 Frequency : always on ppc64
2199 Description: the write to an ext3 filesystem mounted with mballoc got stuck
2200 Details : ext3_mb_generate_buddy() uses find_next_bit() which does not
2201 perform endianness conversion.
2204 Frequency : rarely (truncate to non-zero file size after write under load)
2205 Bugzilla : 10730, 10687
2206 Description: Files padded with zeros to next 4K multiple
2207 Details : With filesystems mounted using the "extents" option (2.6 kernels)
2208 it is possible that files that are truncated to a non-zero size
2209 immediately after being written are filled with zero bytes beyond
2210 the truncated size. No file data is lost.
2212 Severity : enhancement
2214 Description: Allow recovery/failover for liblustre clients.
2215 Details : liblustre clients were unaware of failover configurations until
2218 Severity : enhancement
2220 Description: user file locks should fail when not mounting with flock option
2221 Details : Set up an error-returning stub in ll_file_operations.lock field
2222 to prevent incorrect behaviour when client is mounted without
2223 flock option. Also, set up properly f_op->flock field for
2227 Frequency : always on ia64
2229 Description: "lfs df" loops on printing out MDS statfs information
2230 Details : The obd_ioctl_data was not initialized and in some systems
2231 this caused a failure during the ioctl that did not return
2232 an error. Initialize the struct and return an error on failure.
2235 Frequency : SLES 9 only
2237 Description: Error of copying files with lustre special EAs as root
2238 Details : Client side always return success for setxattr call for lustre
2239 special xattr (currently only "trusted.lov").
2242 Frequency : rarely on clusters with both ia64+i386 clients
2244 Description: ia64+i686 clients doing shared IO on the same file may LBUG
2245 Details : In rare cases when both ia64+i686 (or other mixed-PAGE_SIZE)
2246 clients are doing concurrent writes to the same file it is
2247 possible that the ia64 clients may LASSERT because the OST
2248 extent locks are not PAGE_SIZE aligned. Ensure that grown
2249 locks are always aligned on the request boundary.
2252 Frequency : specific use, occasional
2254 Description: Overwriting in use executable truncates on-disk binary image
2255 Details : If one node attempts to overwrite an executable in use by
2256 another node, we now correctly return ETXTBSY instead of
2257 truncating the file.
2259 Severity : enhancement
2261 Description: Async OSC create to avoid the blocking unnecessarily.
2262 Details : If a OST has no remain object, system will block on the creating
2263 when need to create a new object on this OST. Now, ways use
2264 pre-created objects when available, instead of blocking on an
2265 empty osc while others are not empty. If we must block, we block
2266 for the shortest possible period of time.
2271 Description: chmod on Lustre root is propagated to other clients
2272 Details : Re-validate root's dentry in ll_lookup_it to avoid having it
2273 invalid by the follow_mount time.
2276 Frequency : liblustre clients only
2278 Description: Race in 'instant cancel' lock handling could lead to such locks
2279 never to be granted in case of SMP MDS
2280 Details : Do not destroy not yet granted but cbpending locks in
2284 Frequency : replay/resend of open
2286 Description: non null lock assetion failure in mds_intent_policy
2287 Details : Trying to replay/resend lockless open requests resulted in
2288 mds_open() returning 0 with no lock. Now it sets a flag if
2289 a lock is going to be returned.
2291 Severity : enhancement
2293 Description: Checksum enhancements
2294 Details : New checksum enhancements allow for resending RPCs that failed
2297 Severity : enhancement
2299 Description: Tunables on number of dirty pages in cacche
2300 Details : Allow to set limit on number of dirty pages cached.
2305 Description: client crash on unmount - lock still has references
2306 Details : In some error handling cases it was possible to leak a lock
2307 reference on a client while accessing a file. This was not
2308 harmful to the client during operation, but would cause the
2309 client to crash when the filesystem is unmounted.
2312 Frequency : specific case, rare
2314 Description: ETXTBSY on mds though file not in use by client
2315 Details : ETXTBSY is no longer incorrectly returned when attempting to
2316 chmod or chown a directory that the user previously tried to
2317 execute or a currently-executing binary.
2320 Frequency : extremely rare except on liblustre-based clients
2322 Description: Lustre space not freed when files are deleted
2323 Details : Clean up open-unlinked files after client eviction. Previously
2324 the unlink was skipped and the files remained as orphans.
2329 Description: OST failure "would be an LBUG" in waiting_locks_callback()
2330 Details : In some cases it was possible to send a blocking callback to
2331 a client doing a glimpse, even though that client didn't get
2332 a lock granted. When the glimpse lock is cancelled on the OST
2333 the freed lock is left on the waiting list and corrupted the list.
2336 Frequency : all core dumps
2338 Description: Broke core dumps to lustre
2339 Details : Negative dentry may be unhashed if parent does not have UPDATE
2340 lock, but some callers, e.g. do_coredump, expect dentry to be
2341 hashed after successful create, hash it in ll_create_it.
2343 ------------------------------------------------------------------------------
2345 2006-09-13 Cluster File Systems, Inc. <info@clusterfs.com>
2347 * Support for kernels:
2348 2.6.9-42.0.2.EL (RHEL 4)
2349 2.6.5-7.276 (SLES 9)
2350 2.4.21-40.EL (RHEL 3)
2351 2.6.12.6 vanilla (kernel.org)
2355 Frequency : always on RHEL 3
2357 Description: Number of open files grows over time
2358 Details : The number of open files grows over time, whether or not
2359 Lustre is started. This was due to a filp leak introduced
2360 by one of our kernel patches.
2362 ------------------------------------------------------------------------------
2364 08-20-2006 Cluster File Systems, Inc. <info@clusterfs.com>
2366 * Support for kernels:
2367 2.6.9-42.EL (RHEL 4)
2368 2.6.5-7.276 (SLES 9)
2369 2.4.21-40.EL (RHEL 3)
2370 2.6.12.6 vanilla (kernel.org)
2375 Bugzilla : 5719, 9635, 9792, 9684
2376 Description: OST (or MDS) trips assertions in (re)connection under heavy load
2377 Details : If a server is under heavy load and cannot reply to new
2378 connection requests before the client resends the (re)connect,
2379 the connection handling code can behave badly if two service
2380 threads are concurrently handing separate (re)connections from
2381 the same client. Add better locking to the connection handling
2382 code, and ensure that only a single connection will be processed
2383 for a given client UUID, even if the lock is dropped.
2385 Severity : enhancement
2387 Description: add TCP zero-copy support to kernel
2388 Details : Add support to the kernel TCP stack to allow zero-copy bulk
2389 sends if the hardware supports scatter-gather and checksumming.
2390 This allows socklnd to do client-write and server-read more
2391 efficiently and reduce CPU utilization from skbuf copying.
2394 Frequency : only if NFS exporting from client
2396 Description: NULL pointer deref in ll_iocontrol() if chattr mknod file
2397 Details : If setting attributes on a file created under NFS that had
2398 never been opened it would be possible to oops the client
2399 if the file had no objects.
2402 Frequency : always for liblustre
2404 Description: liblustre client does MDS+OSTs setattr RPC for each write
2405 Details : When doing a write from a liblustre client, the client
2406 incorrectly issued an RPC to the MDS and each OST the file was
2407 striped over in order to update the timestamps. When writing
2408 with small chunks and many clients this could overwhelm the MDS
2409 with RPCs. In all cases it would slow down the write because
2410 these RPCs are unnecessary.
2412 Severity : enhancement
2414 Description: allow number of MDS service threads to be changed at module load
2415 Details : It is now possible to change the number of MDS service threads
2416 running. Adding "options mds mds_num_threads={N}" to the MDS's
2417 /etc/modprobe.conf will set the number of threads for the next
2418 time Lustre is restarted (assuming the "mds" module is also
2419 reloaded at that time). The default number of threads will
2420 stay the same, 32 for most systems.
2425 Description: OST crash if filesystem is unformatted or corrupt
2426 Details : If an OST is started on a device that has never been formatted
2427 or if the filesystem is corrupt and cannot even mount then the
2428 error handling cleanup routines would dereference a NULL pointer.
2433 Description: NULL pointer deref in llap_from_page.
2434 Details : get_cache_page_nowait can return a page with NULL (or otherwise
2435 incorrect) mapping if the page was truncated/reclaimed while it was
2436 searched for. Check for this condition and skip such pages when
2437 doing readahead. Introduce extra check to llap_from_page() to
2438 verify page->mapping->host is non-NULL (so page is not anonymous).
2441 Frequency : Sometimes when using sys_sendfile
2443 Description: "page not covered by a lock" warnings from ll_readpage
2444 Details : sendfile called ll_readpage without right page locks present.
2445 Now we introduced ll_file_sendfile that does necessary locking
2446 around call to generic_file_sendfile() much like we do in
2450 Frequency : with certain MDS communication failures at client mount time
2452 Description: NULL pointer deref after failed client mount
2453 Details : a client connection request may delayed by the network layer
2454 and not be sent until after the PTLRPC layer has timed out the
2455 request. If the client fails the mount immediately it will try
2456 to clean up before the network times out the request. Add a
2457 reference from the request import to the obd device and delay
2458 the cleanup until the network drops the request.
2461 Frequency : occasionally during client (re)connect
2463 Description: assertion failure during client (re)connect
2464 Details : processing a client connection request may be delayed by the
2465 client or server longer than the client connect timeout. This
2466 causes the client to resend the connection request. If the
2467 original connection request is replied in this interval, the
2468 client may trip an assertion failure in ptlrpc_connect_interpret()
2469 which thought it would be the only running connect process.
2472 Frequency : only with obd_echo servers and clients that are rebooted
2474 Description: kernel BUG accessing uninitialized data structure
2475 Details : When running an obd_echo server it did not start the ping_evictor
2476 thread, and when a client was evicted an uninitialized data
2477 structure was accessed. Start the ping_evictor in the RPC
2478 service startup instead of the OBD startup.
2480 Severity : enhancement
2481 Bugzilla : 10193 (patchless)
2482 Description: Remove dependency on various unexported kernel interfaces.
2483 Details : No longer need reparent_to_init, exit_mm, exit_files,
2484 sock_getsockopt, filemap_populate, FMODE_EXEC, put_filp.
2487 Frequency : rare (only users of deprecated and unsupported LDAP config)
2489 Description: write_conf for zeroconf mount queried LDAP incorrectly for client
2490 Details : LDAP apparently contains 'lustreName' attributes instead of
2491 'name'. A simple remapping of the name is sufficient.
2494 Frequency : rare (only with non-default dump_on_timeout debug enabled)
2496 Description: waiting_locks_callback trips kernel BUG if client is evicted
2497 Details : Running with the dump_on_timeout debug flag turned on makes
2498 it possible that the waiting_locks_callback() can try to dump
2499 the Lustre kernel debug logs from an interrupt handler. Defer
2500 this log dumping to the expired_lock_main() thread.
2502 Severity : enhancement
2504 Description: Support NFS exporting on 2.6 kernels.
2505 Details : Implement non-rawops metadata methods for NFS server to use without
2506 changing NFS server code.
2509 Frequency : very rare (synthetic metadata workload only)
2511 Description: two racing renames might cause an MDS thread to deadlock
2512 Details : Running the "racer" program may cause one MDS thread to rename
2513 a file from being the source of a rename to being the target of
2514 a rename at exactly the same time that another thread is doing
2515 so, and the second thread has already enqueued these locks after
2516 doing a lookup of the target and is trying to relock them in
2517 order. Ensure that we don't try to re-lock the same resource.
2520 Frequency : only very large systems with liblustre clients
2522 Description: slow eviction of liblustre clients with the "evict_by_nid" RPC
2523 Details : Use asynchronous set_info RPCs to send the "evict_by_nid" to
2524 all OSTs in parallel. This allows the eviction of stale liblustre
2525 clients to proceed much faster than if they were done in series,
2526 and also offers similar improvements for other set_info RPCs.
2531 Description: excessive CPU usage during initial read phase on client
2532 Details : During the initial read phase on a client, it would agressively
2533 retry readahead on the file, consuming too much CPU and impacting
2534 performance (since 1.4.5.8). Improve the readahead algorithm
2535 to avoid this, and also improve some other common cases (read
2536 of small files in particular, where "small" is files smaller than
2537 /proc/fs/lustre/llite/*/max_read_ahead_whole_mb, 2MB by default).
2542 Description: MDS crash when receiving packet with unknown intent.
2543 Details : Do not LBUG in unknown intent case, just return -EFAULT
2545 Severity : enhancement
2546 Bugzilla : 9293, 9385
2547 Description: MDS RPCs are serialised on client. This is unnecessary for some.
2548 Details : Do not serialize getattr (non-intent version) and statfs.
2551 Frequency : occasional, when OST network is overloaded/intermittent
2553 Description: client evicted by OST after bulk IO timeout
2554 Details : If a client sends a bulk IO request (read or write) the OST
2555 may evict the client if it is unresposive to its data GET/PUT
2556 request. This is incorrect if the network is overloaded (takes
2557 too long to transfer the RPC data) or dropped the OST GET/PUT
2558 request. There is no need to evict the client at all, since
2559 the pinger and/or lock callbacks will handle this, and the
2560 client can restart the bulk request.
2563 Frequency : Always when mmapping file with no objects
2565 Description: client crashes when mmapping file with no objects
2566 Details : Check that we actually have objects in a file before doing any
2567 operations on objects in ll_vm_open, ll_vm_close and
2573 Description: Request leak when working with deleted CWD
2574 Details : Introduce advanced request refcount tracking for requests
2575 referenced from lustre intent.
2577 Severity : Enhancement
2579 Description: Cache open file handles on client.
2580 Details : MDS now will return special lock along with openhandle, if
2581 requested and client is allowed to hold openhandle, even if unused,
2582 until such a lock is revoked. Helps NFS a lot, since NFS is opening
2583 closing files for every read/write openration.
2585 Severity : Enhancement
2587 Description: Cache open negative dentries on client when possible.
2588 Details : Guard negative dentries with UPDATE lock on parent dir, drop
2589 negative dentries on lock revocation.
2594 Description: Remounting a client read-only wasn't possible with a zconf mount
2595 Details : It wasn't possible to remount a client read-only with llmount.
2597 Severity : enhancement
2598 Description: Include MPICH 1.2.6 Lustre ADIO interface patch
2599 Details : In lustre/contrib/ or /usr/share/lustre in RPM a patch for
2600 MPICH is included to add Lustre-specific ADIO interfaces.
2601 This is based closely on the UFS ADIO layer and only differs
2602 in file creation, in order to allow the OST striping to be set.
2603 This is user-contributed code and not supported by CFS.
2608 Description: extended inode attributes (immutable, append-only) work improperly
2609 when 2.4 and 2.6 kernels are used on client/server or vice versa
2610 Details : Introduce kernel-independent values for these flags.
2612 Severity : enhancement
2615 Description: Allow fractional MB tunings for lustre in /proc/ filesystem.
2616 Details : Many of the /proc/ tunables can only be tuned at a megabyte
2617 granularity. Now, Fractional MB granularity is be supported,
2618 this is very useful for low memory system.
2620 Severity : enhancement
2622 Description: Getattr by fid
2623 Details : Getting a file attributes by its fid, obtaining UPDATE|LOOKUP
2624 locks, avoids extra getattr rpc requests to MDS, allows '/' to
2625 have locks and avoids getattr rpc requests for it on every stat.
2628 Frequency : Always, for filesystems larger than 2TB
2630 Description: ldiskfs crash at mount for filesystem larger than 2TB with mballoc
2631 Details : Kenrel kmalloc limits allocations to 128kB and this prevents
2632 filesystems larger than 2TB to be mounted with mballoc enabled.
2635 Frequency : Always, for 32-bit kernel without CONFIG_LBD and filesystem > 2TB
2637 Description: ldiskfs crash at mount for filesystem larger than 2TB with mballoc
2638 Details : If a 32-bit kernel is compiled without CONFIG_LBD enabled and a
2639 filesystems larger than 2TB is mounted then the kernel will
2640 silently corrupt the start of the filesystem. CONFIG_LBD is
2641 enabled for all CFS-supported kernels, but the possibility of
2642 this happening with a modified kernel config exists.
2644 Severity : enhancement
2646 Description: add client O_DIRECT support for 2.6 kernels
2647 Details : It is now possible to do O_DIRECT reads and writes to files
2648 in the Lustre client mountpoint on 2.6 kernel clients.
2650 Severity : enhancement
2652 Description: parallel glimpse, setattr, statfs, punch, destroy requests
2653 Details : Sends glimpse, setattr, statfs, punch, destroy requests to OSTs in
2654 parallel, not waiting for response from every OST before sending
2655 a rpc to the next OST.
2660 Description: setattr vs write race when updating file timestamps
2661 Details : Client processes that update a file timestamp into the past
2662 right after writing to the file (e.g. tar) it is possible that
2663 the updated file modification time can be reset to the current
2664 time due to a race between processing the setattr and write RPC.
2666 Severity : enhancement
2668 Description: Bring 'lfs find' closer in line with regular Linux find.
2669 Details : lfs find util supports -atime, -mtime, -ctime, -maxdepth, -print,
2670 -print0 options and obtains all the needed info through the lustre
2673 Severity : enhancement
2675 Description: support up to 1024 configured devices on one node
2676 Details : change obd_dev array from statically allocated to dynamically
2677 allocated structs as they are first used to reduce memory usage
2682 Description: Flush dirty partially truncated pages during truncate
2683 Details : Immediatelly flush partially truncated pages in filter_setattr,
2684 this way we completely avoid having any pages in page cache on OST
2685 and can retire ugly workarounds during writes to flush such pages.
2690 Description: i_sem vs transaction deadlock in mds_obd_destroy during unlink.
2691 Details : protect inode from truncation within vfs_unlink() context
2692 just take a reference before calling vfs_unlink() and release it
2693 when parent's i_sem is free.
2698 Description: last_id value checked outside lock on OST caused LASSERT failure
2699 Details : If there were multiple MDS->OST object precreate requests in
2700 flight, it was possible that the OST's last object id was checked
2701 outside a lock and incorrectly tripped an assertion. Move checks
2702 inside locks, and discard old precreate requests.
2705 Frequency : always, if extents are used on OSTs
2707 Description: index ei_leaf_hi (48-bit extension) is not zeroed in extent index
2708 Details : OSTs using the extents format would not zero the high 16 bits of
2709 the index physical block number. This is not a problem for any
2710 OST filesystems smaller than 16TB, and no kernels support ext3
2711 filesystems larger than 16TB yet. This is fixed in 1.4.7 (all
2712 new/modified files) and can be fixed for existing filesystems
2713 with e2fsprogs-1.39-cfs1.
2718 Description: import connection selection may be incorrect if timer wraps
2719 Details : Using a 32-bit jiffies timer with HZ=1000 may cause backup
2720 import connections to be ignored if the 32-bit jiffies counter
2721 wraps. Use a 64-bit jiffies counter.
2724 Frequency : very large clusters immediately after boot
2726 Description: LNET request buffers exhausted under heavy short-term load
2727 Details : If a large number of client requests are generated on a service
2728 that has previously never seen so many requests it is possible
2729 that the request buffer growth cannot keep up with the spike in
2730 demand. Instead of dropping incoming requests, they are held in
2731 the LND until the RPC service can accept more requests.
2734 Frequency : Sometimes during replay
2736 Description: Assertion failure in ll_local_open after replay.
2737 Details : If replay happened on an open request reply before we were able
2738 to set replay handler, reply will become not swabbed tripping the
2739 assertion in ll_local_open. Now we set the handler right after
2740 recognising of open request
2743 Frequency : very rare
2745 Description: kernel reports "badness in vsnprintf"
2746 Details : Reading from the "recovery_status" /proc file in small chunks
2747 may cause a negative length in lprocfs_obd_rd_recovery_status()
2748 call to vsnprintf() (which is otherwise harmless). Exit early
2749 if there is no more space in the output buffer.
2751 Severity : enhancement
2753 Description: clear OBD RPC statistics by writing to them
2754 Details : It is now possible to clear the OBD RPC statistics by writing
2755 to the "stats" file.
2760 Description: Client mtime is not the same on different clients after utimes
2761 Details : In some cases, the client was using the utimes() syscall on
2762 a file cached on another node. The clients now validate the
2763 ctime from the MDS + OSTs to determine which one is right.
2768 Description: Inability to activate failout mode
2769 Details : lconf script incorrectly assumed that in pythong string's numeric
2770 value is used in comparisons.
2773 Frequency : always with multiple stripes per file
2775 Description: Inefficient object allocation for mutli-stripe files
2776 Details : When selecting which OSTs to stripe files over, for files with
2777 a stripe count that divides evenly into the number of OSTs,
2778 the MDS is always picking the same starting OST for each file.
2779 Return the OST selection heuristic to the original design.
2784 Description: mount failures may take full timeout to return an error
2785 Details : Under some heavy load conditions it is possible that a
2786 failed mount can wait for the full obd_timeout interval,
2787 possibly several minutes, before reporting an error.
2788 Instead return an error as soon as the status is known.
2790 Frequency : quota enabled and large files being deleted
2792 Description: releasing more than 4GB of quota at once hangs OST
2793 Details : If a user deletes more than 4GB of files on a single OST it
2794 will cause the OST to spin in an infinite loop. Release
2795 quota in < 4GB chunks, or use a 64-bit value for 1.4.7.1+.
2800 Description: statfs data retrieved from /proc may be stale or zero
2801 Details : When reading per-device statfs data from /proc, in the
2802 {kbytes,files}_{total,free,avail} files, it may appear
2803 as zero or be out of date.
2806 Frequency : systems with MD RAID1 external journal devices
2808 Description: lconf's call to blkid is confused by RAID1 journal devices
2809 Details : Use the "blkid -l" flag to locate the MD RAID device instead
2810 of returning all block devices that match the journal UUID.
2813 Frequency : always, for aggregate stripe size over 4GB
2815 Description: assertion fails when trying to use 4GB stripe size
2816 Details : Use "setstripe" to set stripe size over 4GB will fail the kernel,
2817 complaining "ASSERTION(lsm->lsm_xfersize != 0)"
2820 Frequency : always on ppc64
2822 Description: the first write on an ext3 filesystem with mballoc got stuck
2823 Details : ext3_mb_generate_buddy() uses find_next_bit() which does not
2824 perform endianness conversion.
2826 ------------------------------------------------------------------------------
2828 02-14-2006 Cluster File Systems, Inc. <info@clusterfs.com>
2830 * WIRE PROTOCOL CHANGE. This version of Lustre networking WILL NOT
2831 INTEROPERATE with older versions automatically. Please read the
2832 user documentation before upgrading any part of a live system.
2833 * WARNING: Lustre networking configuration changes are required with
2834 this release. See https://bugzilla.clusterfs.com/show_bug.cgi?id=10052
2837 * Support for kernels:
2838 2.6.9-22.0.2.EL (RHEL 4)
2839 2.6.5-7.244 (SLES 9)
2840 2.6.12.6 vanilla (kernel.org)
2843 Severity : enhancement
2844 Bugzilla : 7981/8208
2845 Description: Introduced Lustre Networking (LNET)
2846 Details : LNET is new networking infrastructure for Lustre, it includes
2847 a reorganized network configuration mode (see the user
2848 documentation for full details) as well as support for routing
2849 between different network fabrics. Lustre Networking Devices
2850 (LNDs) for the supported network fabrics have also been
2851 created for this new infrastructure.
2853 Severity : enhancement
2854 Description: Introduced Access control lists
2855 Details : clients can set ACLs on files and directories in order to have
2856 more fine-grained permissions than the standard Unix UGO+RWX.
2857 The MDS must be started with the "-o acl" mount option.
2859 Severity : enhancement
2860 Description: Introduced filesystem quotas
2861 Details : Administrators may now establish per-user quotas on the
2864 Severity : enhancement
2866 Description: Configuration change for the XT3
2867 The PTLLND is now used to run Lustre over Portals on the XT3
2868 The configure option(s) --with-cray-portals are no longer used.
2869 Rather --with-portals=<path-to-portals-includes> is used to
2870 enable building on the XT3. In addition to enable XT3 specific
2871 features the option --enable-cray-xt3 must be used.
2876 Description: Running on many-way SMP OSTs can trigger oops in llcd_send()
2877 Details : A race between allocating a new llcd and re-getting the llcd_lock
2878 allowed another thread to grab newly-allocated llcd.
2880 Severity : enhancement
2882 Description: 2.6 OST async journal commit and locking fix to improve performance
2883 Details : The filter_direct_io()+filter_commitrw_write() journal commits for
2884 2.6 kernels are now async as they already were in 2.4 kernels so
2885 that they can commit concurrently with the network bulk transfer.
2886 For block-allocated files the filter allocation semaphore is held
2887 to avoid filesystem fragmentation during allocation. BKL lock
2888 removed for 2.6 xattr operations where it is no longer needed.
2893 Description: lconf incorrectly determined whether two IP networks could talk
2894 Details : In some more complicated routing and multiple-network
2895 configurations, lconf will avoid trying to make a network
2896 connection to a disjoint part of the IP space. It was doing the
2897 math incorrectly for one set of cases.
2902 Description: Fix for potential infinite loop processing records in an llog.
2903 Details : If an llog record is corrupted/zeroed, it is possible to loop
2904 forever in llog_process(). Validate the llog record length
2905 and skip the remainder of the block on error.
2908 Frequency : occasional (liblustre only)
2910 Description: liblustre could not open files whose last component is a symlink
2911 Details : sysio_path_walk() would incorrectly pass the open intent to
2912 intermediate path components.
2915 Frequency : rare (liblustre only with non-standard tuning)
2916 Bugzilla : 7201 (7350)
2917 Description: Tuning the MDC DLM LRU size to zero triggers client LASSERT
2918 Details : llu_lookup_finish_locks() tries to set lock data on a lock
2919 after it has been released, only do this for referenced locks
2921 Severity : enhancement
2923 Description: specifying an (invalid) directory default stripe_size of -1
2924 would reset the directory default striping
2925 Details : stripe_size -1 was used internally to signal directory stripe
2926 removal, now use "all default" to signal dir stripe removal
2927 as a directory striping of "all default" is not useful
2930 Frequency : common for large clusters running liblustre clients
2932 Description: doing an ls when liblustre clients are running is slow
2933 Details : sending a glimpse AST to a liblustre client waits for every AST
2934 to time out, as liblustre clients will not respond. Since they
2935 cannot cache data we refresh the OST lock LVB from disk instead.
2937 Severity : enhancement
2939 Description: doing an ls at the same time as file IO can be slow
2940 Details : enqueue and other "small" requests can be blocked behind many
2941 large IO requests. Create a new OST IO portal for non-IO
2942 requests so they can be processed faster.
2945 Frequency : rare (only HPUX clients mounting unsupported re-exported NFS vol)
2947 Description: an HPUX NFS client would get -EACCESS when ftruncate()ing a newly
2948 created file with mode 000
2949 Details : the Linux NFS server relies on an MDS_OPEN_OWNEROVERRIDE hack to
2950 allow an ftruncate() as a non-root user to a file with mode 000.
2951 Lustre now respects this flag to disable mode checks when
2952 truncating a file owned by the user
2955 Frequency : liblustre-only, when liblustre client dies unexpectedly or becomes
2958 Description: Revoking locks from clients that went dead or catatonic might take
2960 Details : New lock flags FL_CANCEL_ON_BLOCK used by liblustre makes
2961 cancellation of such locks instant on servers without waiting for
2962 any reply from clients. Clients drops these locks when cancel
2963 notification from server is received without replying.
2966 Frequency : liblustre-only, when liblustre client dies or becomes busy
2968 Description: Doing ls on Linux clients can take a long time with active
2970 Details : Liblustre client cannot handle ASTs in timely manner, so avoid
2971 granting such locks to it in the first place if possible. Locks
2972 are taken by proxy on the OST during the read or write and
2973 dropped immediately afterward. Add connect flags handling, do
2974 not grant locks to liblustre clients for glimpse ASTs.
2976 Severity : enhancement
2978 Description: Improve read-ahead algorithm to avoid excessive IO for random reads
2979 Details : Existing read-ahead algorithm is tuned for the case of streamlined
2980 sequential reads and behaves badly with applications doing random
2981 reads. Improve it by reading ahead at least read region, and
2982 avoiding excessive large RPC for small reads.
2984 Severity : enhancement
2986 Description: Creating more than 1000 files for a single job may cause a load
2987 imbalance on the OSTs if there are also a large number of OSTs.
2988 Details : qos_prep_create() uses an OST index reseed value that is an
2989 even multiple of the number of available OSTs so that if the
2990 reseed happens in the middle of the object allocation it will
2991 still utilize the OSTs as uniformly as possible.
2996 Description: OST or MDS may oops in ping_evictor_main()
2997 Details : ping_evictor_main() drops obd_dev_lock if deleting a stale export
2998 but doesn't restart at beginning of obd_exports_timed list
3001 Severity : enhancement
3003 Description: improve by-nid export eviction on the MDS and OST
3004 Details : allow multiple exports with the same NID to be evicted at one
3005 time without re-searching the exports list.
3008 Frequency : rare, only with supplementary groups enabled on SMP 2.6 kernels
3010 Description: MDS may oops in groups_free()
3011 Details : in rare race conditions a newly allocated group_info struct is
3012 freed again, and this can be NULL. The 2.4 compatibility code
3013 for groups_free() checked for a NULL pointer, but 2.6 did not.
3016 Frequency : common for liblustre clients doing little filesystem IO
3017 Bugzilla : 9352, 7313
3018 Description: server may evict liblustre clients accessing contended locks
3019 Details : if a client is granted a lock or receives a completion AST
3020 with a blocking AST already set it would not reply to the AST
3021 for LDLM_FL_CANCEL_ON_BLOCK locks. It now replies to such ASTs.
3024 Frequency : lfs setstripe, only systems with more than 160 OSTs
3026 Description: unable to set striping with a starting offset beyond OST 160
3027 Details : llapi_create_file() incorrectly limited the starting stripe
3028 index to the maximum single-file stripe count.
3031 Frequency : LDAP users only
3033 Description: lconf did not handle in-kernel recovery with LDAP properly
3034 Details : lconf/LustreDB get_refs() is searching the wrong namespace
3036 Severity : enhancement
3038 Description: bind OST threads to NUMA nodes to improve performance
3039 Details : all OST threads are uniformly bound to CPUs on a single NUMA
3040 node and do their allocations there to localize memory access
3042 Severity : enhancement
3044 Description: llmount can determine client NID directly from Myrinet (GM)
3045 Details : the client NID code from gmnalnid was moved directly into
3046 llmount, removing the need to use this or specifying the
3047 client NID explicitly when mounting GM clients with zeroconf
3050 Frequency : if client is started with down MDS
3052 Description: if client is started with down MDS mount hangs in ptlrpc_queue_wait
3053 Details : Having an LWI_INTR() wait event (interruptible, but no timeout)
3054 will wait indefinitely in ptlrpc_queue_wait->l_wait_event() after
3055 ptlrpc_import_delayed_req() because we didn't check if the
3056 request was interrupted, and we also didn't break out of the
3057 event loop if there was no timeout
3062 Description: data loss during non-page-aligned writes to a single file from
3063 both multiple nodes and multiple threads on one node at same time
3064 Details : updates to KMS and lsm weren't protected by common lock. Resulting
3065 inconsistency led to false short-reads, that were cached and later
3066 used by ->prepare_write() to fill in partially written page,
3067 leading to data loss.
3070 Frequency : always, if lconf --abort_recovery used
3072 Description: lconf --abort_recovery fails with 'Operation not supported'
3073 Details : lconf was attempting to abort recovery on the MDT device and not
3076 Severity : enhancement
3078 Description: remove cleanup logs
3079 Details : replace lconf-generated cleanup logs with lustre internal
3080 cleanup routines. Eliminates the need for client-cleanup and
3083 Severity : enhancement
3085 Description: add support for EAs (user and system) on lustre filesystems
3086 Details : it is now possible to store extended attributes in the Lustre
3087 client filesystem, and with the user_xattr mount option it
3088 is possible to allow users to store EAs on their files also
3090 Severity : enhancement
3092 Description: Add possibility (config option) to show minimal available OST free
3094 Details : When compiled with --enable-mindf configure option, statfs(2)
3095 (and so, df) will return least minimal free space available from
3096 all OSTs as amount of free space on FS, instead of summary of
3097 free spaces of all OSTs.
3099 Severity : enhancement
3101 Description: do not expand extent locks acquired on OST-side
3102 Details : Modify ldlm_extent_policy() to not expand local locks, acquired
3103 by server: they are not cached anyway.
3106 Frequency : when mmap is used/binaries executed from Lustre
3108 Description: Unmmap pages before throwing them away from read cache.
3109 Details : llap_shrink cache now attempts to unmap pages before discarding
3110 them (if unmapping failed - do not discard). SLES9 kernel has
3111 extra checks that trigger if this unmapping is not done first.
3116 Description: lconf didn't resolve symlinks before checking to see whether a
3117 given mountpoint was already in use
3120 Frequency : when migrating failover services
3121 Bugzilla : 6395, 9514
3122 Description: When migrating a subset of services from a node (e.g. failback
3123 from a failover service node) the remaining services would
3124 time out and evict clients.
3125 Details : lconf --force (implied by --failover) sets the global obd_timeout
3126 to 5 seconds in order to quickly disconnect, but this caused
3127 other RPCs to time out too quickly. Do not change the global
3128 obd_timeout for force cleanup, only set it for DISCONNECT RPCs.
3130 Severity : enhancement
3131 Frequency : if MDS is started with down OST
3132 Bugzilla : 9439,5706
3133 Description: Allow startup/shutdown of an MDS without depending on the
3134 availability of the OSTs.
3135 Details : Asynchronously call mds_lov_synchronize during MDS startup.
3136 Add appropriate locking and lov-osc refcounts for safe
3137 cleaning. Add osc abort_inflight calls in case the
3138 synchronize never started.
3141 Frequency : occasional (Cray XT3 only)
3143 Description: root not authorized to access files in CRAY_PORTALS environment
3144 Details : The client process capabilities were not honoured on the MDS in
3145 a CRAY_PORTALS/CRAY_XT3 environment. If the file had previously
3146 been accessed by an authorized user then root was able to access
3147 the file on the local client also. The root user capabilities
3148 are now allowed on the MDS, as this environment has secure UID.
3151 Frequency : occasional
3153 Description: ldiskfs "too long searching" message happens too often
3154 Details : A debugging message (otherwise harmless) prints too often on
3155 the OST console. This has been reduced to only happen when
3156 there are fragmentation problems on the filesystem.
3161 Description: Division by zero in statfs when all OSCs are inactive
3162 Details : lov_get_stripecnt() returns zero due to incorrect order of checks,
3163 lov_statfs divides by value returned by lov_get_stripecnt().
3167 Bugzilla : 9489, 3273
3168 Description: First write from each client to each OST was only 4kB in size,
3169 to initialize client writeback cache, which caused sub-optimal
3170 RPCs and poor layout on disk for the first writen file.
3171 Details : Clients now request an initial cache grant at (re)connect time
3172 and so that they can start streaming writes to the cache right
3173 away and always do full-sized RPCs if there is enough data.
3174 If the OST is rebooted the client also re-establishes its grant
3175 so that client cached writes will be honoured under the grant.
3180 Description: Slow ls (and stat(2) syscall) on files residing on IO-loaded OSTs
3181 Details : Now I/O RPCs go to different portal number and (presumably) fast
3182 lock requests (and glimses) and other RPCs get their own service
3183 threads pool that should be able to service those RPCs
3186 Severity : enhancement
3188 Description: Ability to exchange lustre version between client and servers and
3189 issue warnings at client side if client is too old. Also for
3190 liblustre clients there is ability to refuse connection of too old
3192 Details : New 'version' field is added to connect data structure that is
3193 filled with version info. That info is later checked by server and
3197 Frequency : rare, liblustre only.
3198 Bugzilla : 9296, 9581
3199 Description: Two simultaneous writes from liblustre at offset within same page
3200 might proceed at the same time overwriting eachother with stale
3202 Details : I/O lock withing llu_file_prwv was released too early, before data
3203 actually was hitting the wire. Extended lock-holding time until
3204 server acknowledges receiving data.
3207 Frequency : extremely rare. Never observed in practice.
3209 Description: avoid generating lustre_handle cookie of 0.
3210 Details : class_handle_hash() generates handle cookies by incrementing
3211 global counter, and can hit 0 occasionaly (this is unlikely, but
3212 not impossible, because initial value of cookie counter is
3213 selected randonly). Value of 0 is used as a sentinel meaning
3214 "unassigned handle" --- avoid it. Also coalesce two critical
3215 sections in this function into one.
3217 Severity : enhancement
3219 Description: allow liblustre clients to delegate truncate locking to OST
3220 Details : To avoid overhead of locking, liblustre client instructs OST to
3221 take extent lock in ost_punch() on client's behalf. New connection
3222 flag is added to handle backward compatibility.
3224 Severity : enhancement
3225 Bugzilla : 4928, 7341, 9758
3226 Description: allow number of OST service threads to be specified
3227 Details : a module parameter allows the number of OST service threads
3228 to be specified via "options ost ost_num_threads={N}" in the
3229 OSS's /etc/modules.conf or /etc/modprobe.conf.
3233 Bugzilla : 6146, 9635, 9895
3234 Description: servers crash with bad pointer in target_handle_connect()
3235 Details : In rare cases when a client is reconnecting it was possible that
3236 the connection request was the last reference for that export.
3237 We would temporarily drop the export reference and get a new
3238 one, but this may have been the last reference and the export
3239 was just destroyed. Get new reference before dropping old one.
3241 Severity : enhancement
3242 Frequency : if client is started with failover MDS
3244 Description: Allow multiple MDS hostnames in the mount command
3245 Details : Try to read the configuration from all specified MDS
3246 hostnames during a client mount in case the "primary"
3249 Severity : enhancement
3251 Description: Stop sending data to evicted clients as soon as possible.
3252 Details : Check if the client we are about to send or are sending data to
3253 was evicted already. (Check is done every second of waiting,
3254 for which l_wait_event interface was extended to allow checking
3255 of exit condition at specified intervals).
3258 Frequency : rare, normally only when NFS exporting is done from client
3260 Description: 'bad disk LOV MAGIC: 0x00000000' error when chown'ing files
3262 Details : Make mds_get_md() recognise empty md case and set lmm size to 0.
3265 Frequency : always, if srand() is called before liblustre initialization
3267 Description: Liblustre uses system PRNG disturbing its usage by user application
3268 Details : Introduce internal to lustre fast and high-quality PRNG for
3269 lustre usage and make liblustre and some other places in generic
3270 lustre code to use it.
3272 Severity : enhancement
3273 Bugzilla : 9477, 9557, 9870
3274 Description: Verify that the MDS configuration logs are updated when xml is
3275 Details : Check if the .xml configuration logs are newer than the config
3276 logs stored on the MDS and report an error if this is the case.
3277 Request --write-conf, or allow starting with --old_conf.
3279 Severity : enhancement
3281 Description: Handle symlinks in the path when checking if Lustre is mounted.
3282 Details : Resolve intermediate symlinks when checking if a client has
3283 mounted a filesystem to avoid duplicate client mounts.
3288 Description: lconf can hit an error exception but still return success.
3289 Details : The lconf command catches the Command error exception at the top
3290 level script context and will exit with the associated exit
3291 status, but doesn't ensure that this exit status is non-zero.
3296 Description: failure of ptlrpc thread startup can cause oops
3297 Details : Starting a ptlrpc service thread can fail if there are a large
3298 number of threads or the server memory is very fragmented.
3299 Handle this without oopsing.
3302 Frequency : always, only if liblustre and non-default acceptor port was used
3304 Description: liblustre cannot connect to servers with non-default acceptor port
3305 Details : tcpnal_set_default_params() was not called and was therefore
3306 ignoring the environment varaible TCPNAL_PORT, as well as other
3307 TCPNAL_ environment variables
3312 Description: two objects could be created on the same OST for a single file
3313 Details : If an OST is down, in some cases it was possible to create two
3314 objects on a single OST for a single file. No problems other
3315 than potential performance impact and spurious error messages.
3319 Bugzilla : 5681, 9562
3320 Description: Client may oops in ll_unhash_aliases
3321 Details : Client dcache may become inconsistent in race condition.
3322 In some cases "getcwd" can fail if the current directory is
3328 Description: Inode refcounting problems in NFS export code
3329 Details : link_raw functions used to call d_instantiate without obtaining
3330 extra inode reference first.
3334 Bugzilla : 9942, 9903
3335 Description: Referencing freed requests leading to crash, memleaks with NFS.
3336 Details : We used to require that call to ll_revalidate_it was always
3337 followed by ll_lookup_it. Also with revalidate_special() it is
3338 possible to call ll_revalidate_it() twice for the same dentry
3339 even if first occurence returned success. This fix changes semantic
3340 between DISP_ENQ_COMPLETE disposition flag to mean there is extra
3341 reference on a request referred from the intent.
3342 ll_intent_release() then releases such a request.
3345 Frequency : rare, normally benchmark loads only
3347 Description: unlinked inodes were kept in memory on the client
3348 Details : If a client is repeatedly creating and unlinking files it
3349 can accumulate a lot of stale inodes in the inode slab cache.
3350 If there is no other client load running this can cause the
3351 client node to run out of memory. Instead flush old inodes
3352 from client cache that have the same inode number as a new inode.
3355 Frequency : SLES9 2.6.5 kernel and long filenames only
3356 Bugzilla : 9969, 10379
3357 Description: utime reports stale NFS file handle
3358 Details : SLES9 uses out-of-dentry names in some cases, which confused
3359 the lustre dentry revalidation. Change it to always use the
3363 Frequency : rare, unless heavy write-truncate concurrency is continuous
3364 Bugzilla : 4180, 6984, 7171, 9963, 9331
3365 Description: OST becomes very slow and/or deadlocked during object unlink
3366 Details : filter_destroy() was holding onto the parent directory lock
3367 while truncating+unlinking objects. For very large objects this
3368 may block other threads for a long time and slow overall OST
3369 responsiveness. It may also be possible to get a lock ordering
3370 deadlock in this case, or run out of journal credits because of
3371 the combined truncate+unlink. Solution is to do object truncate
3372 first in one transaction without parent lock, and then do the
3373 final unlink in a new transaction with the parent lock. This
3374 reduces the lock hold time dramatically.
3377 Frequency : rare, 2.4 kernels only
3379 Description: MDS or OST cleanup may trip kernel BUG when dropping kernel lock
3380 Details : mds_cleanup() and filter_cleanup() need to drop the kernel lock
3381 before unmounting their filesystem in order to avoid deadlock.
3382 The kernel_locked() function in 2.4 kernels only checks whether
3383 the kernel lock is held, not whether it is this process that is
3384 holding it as 2.6 kernels do.
3389 Description: MDS or OST may oops/LBUG if a client is connecting multiple times
3390 Details : The client ptlrpc code may be trying to reconnect to a down
3391 server before a previous connection attempt has timed out.
3392 Increase the reconnect interval to be longer than the connection
3393 timeout interval to avoid sending duplicate connections to
3397 Frequency : echo_client brw_test command
3399 Description: fix echo_client to work with OST preallocated code
3400 Details : OST preallocation code (5137) didn't take echo_client IO path
3401 into account: echo_client calls filter methods outside of any
3402 OST thread and, hence, there is no per-thread preallocated
3403 pages and buffers to use. Solution: hijack pga pages for IO. As
3404 a byproduct, this avoids unnecessary data copying.
3408 Bugzilla : 3555, 5962, 6025, 6155, 6296, 9574
3409 Description: Client can oops in mdc_commit_close() after open replay
3410 Details : It was possible for the MDS to return an open request with no
3411 transaction number in mds_finish_transno() if the client was
3412 evicted, but without actually returning an error. Clients
3413 would later try to replay that open and may trip an assertion
3414 Simplify the client close codepath, and always return an error
3415 from the MDS in case the open is not successful.
3418 Frequency : rare, 2.6 OSTs only
3420 Description: OST may deadlock under high load on fragmented files
3421 Details : If there was a heavy load and highly-fragmented OST filesystems
3422 it was possible to have all the OST threads deadlock waiting on
3423 allocation of biovecs, because the biovecs were not released
3424 until the entire RPC IO was completed. Instead, release biovecs
3425 as soon as they are complete to ensure forward IO progress.
3427 Severity : enhancement
3429 Description: Support for specifying external journal device at mount
3430 Details : If an OST or MDS device is formatted with an external journal
3431 device, this device major/minor is stored in the ext3 superblock
3432 and may not be valid for failover. Allow detecting and
3433 specifying the external journal at mount time.
3438 Description: Mounting an MDS with pending unlinked files may cause oops
3439 Details : target_finish_recovery() calls mds_postrecov() which returned
3440 the number of orphans unlinked. mds_lov_connect->mds_postsetup()
3441 considers this an error and immediately begins cleaning up the
3442 lov, just after starting the mds_lov process
3444 Severity : enhancement
3446 Description: Implement 'lfs df' to report actual free space on per-OST basis
3447 Details : Add sub-command 'df' on 'lfs' to report the disk space usage of
3448 MDS/OSDs. Usage: lfs df [-i][-h]. Command Options: '-i' to report
3449 usage of objects; '-h' to report in human readable format.
3451 ------------------------------------------------------------------------------
3453 08-26-2005 Cluster File Systems, Inc. <info@clusterfs.com>
3460 Description: Mounting an ldiskfs file system with mballoc may crash OST node.
3461 Details : ldiskfs mballoc code may reference an uninitialized buddy struct
3462 at startup during orphan unlinking. Instead, skip buddy update
3463 before setup, as it will be regenerated after recovery is complete.
3468 Description: If an OST is inactive, its locks might reference stale inodes.
3469 Details : lov_change_cbdata() must iterate over all namespaces, even if
3470 they are inactive to clear inode references from the lock.
3472 Severity : enhancement
3473 Frequency : occasional, if non-standard max_dirty_mb used
3475 Description: Client will block write RPCs if not enough grant
3476 Details : If a client has max_dirty_mb smaller than max_rpcs_in_flight,
3477 then the client will block writes while waiting for another RPC
3478 to complete instead of consuming its dirty limit. With change
3479 we get improved performance when max_dirty_mb is small.
3481 Severity : enhancement
3482 Bugzilla : 3389, 6253
3483 Description: Add support for supplementary groups on the MDS.
3484 Details : The MDS has an upcall /proc/fs/lustre/mds/{mds}/group_upcall
3485 (set to /usr/sbin/l_getgroups if enabled) which will do MDS-side
3486 lookups for user supplementary groups into a cache.
3490 Description: O_CREAT|O_EXCL open flags in liblustre always return -EEXIST
3491 Details : Make libsysio to not enforce O_EXCL by clearing the flag,
3492 for liblustre O_EXCL is enforced by MDS.
3496 Description: readdir never returns NULL in liblustre.
3497 Details : Corrected llu_iop_getdirentries logic, to return offset of next
3498 dentry in struct dirent.
3502 Frequency : liblustre only, depends on application IO pattern
3503 Description: liblustre clients evicted if not contacting servers
3504 Details : Don't put liblustre clients into the ping_evictor list, so
3505 they will not be evicted by the pinger ever.
3507 Severity : enhancement
3509 Description: Add ability to evict clients by NID from MDS.
3510 Details : By echoing "nid:$NID" string into
3511 /proc/fs/lustre/mds/.../evict_client client with nid that equals to
3512 $NID would be instantly evicted from this MDS and from all active
3513 OSTs connected to it.
3517 Description: Do not query file size twice, somewhat slowing stat(2) calls.
3518 Details : lookup_it_finish() used to query file size from OSTs that was not
3523 Description: service threads change working directory to that of init
3524 Details : Starting lustre service threads may pin the working directory
3525 of the parent thread, making that filesystem busy. Threads
3526 now change to the working directory of init to avoid this.
3530 Frequency : during shutdown only
3531 Description: shutdown with a failed MDS or OST can cause unmount to hang
3532 Details : Don't resend DISCONNECT messages in ptlrpc_disconnect_import()
3537 Frequency : 2.6 only
3538 Description: chmod/chown may include an extra supplementary group
3539 Details : ll{,u}_mdc_pack_op_data() does not properly initialize the
3540 supplementary group and if none is specified this is used.
3543 Bugzilla : 5479 (6816)
3545 Description: Racing open + rm can assert client in mdc_set_open_replay_data()
3546 Details : If lookup is in progress on a file that is unlinked we might try
3547 to revalidate the inode and fail in revalidate after lookup is
3548 complete and ll_file_open() enqueues the open again but
3549 it_open_error() was not checking DISP_OPEN_OPEN errors correctly.
3552 Frequency : always, if lconf --abort_recovery used
3554 Description: lconf --abort_recovery fails with 'Operation not supported'
3555 Details : lconf was attempting to abort recovery on the MDT device and not
3558 ------------------------------------------------------------------------------
3560 2005-08-08 Cluster File Systems, Inc. <info@clusterfs.com>
3565 Frequency : rare (only unsupported configurations with a node running as an
3567 Bugzilla : 6514, 5137
3568 Description: Mounting a Lustre file system on a node running as an OST could
3570 Details : OSTs now preallocates memory needed to write out data at
3571 startup, instead of when needed, to avoid having to
3572 allocate memory in possibly low memory situations.
3573 Specifically, if the file system is mounted on on OST,
3574 memory pressure could force it to try to write out data,
3575 which it needed to allocate memory to do. Due to the low
3576 memory, it would be unable to do so and the node would
3577 become unresponsive.
3579 Severity : enhancement
3581 Description: Addition of lconf --service command line option
3582 Details : lconf now accepts a '--service <arg>' option, which is
3583 shorthand for 'lconf --group <arg> --select <arg>=<hostname>'
3585 Severity : enhancement
3587 Description: Failover mode is now the default for OSTs.
3588 Details : By default, OSTs will now run in failover mode. To return to
3589 the old behaviour, add '--failout' to the lmc line for OSTs.
3591 Severity : enhancement
3593 Description: Health checks are now provided for MDS and OSTs
3594 Details : Additional detailed health check information on MSD and OSTs
3595 is now provided through the procfs health_check value.
3598 Frequency : occasional, depends on IO load
3600 Description: Disk fragmentation on the OSTs could eventually cause slowdowns
3601 after numerous create/delete cycles
3602 Details : The ext3 inode allocation policy would not allocate new inodes
3603 very well on the OSTs because there are no new directories
3604 being created. Instead we look for groups with free space if
3605 the parent directories are nearly full.
3610 Description: Network or server problems during mount may cause partially
3611 mounted clients instead of returning an error.
3612 Details : The config llog parsing code may overwrite the error return
3613 code during mount error handling, returning success instead
3619 Description: MDS can fail to allocate large reply buffers
3620 Details : After long uptimes the MDS can fail to allocate large reply
3621 buffers (e.g. zconf client mount config records) due to memory
3622 fragmentation or consumption by the buffer cache. Preallocate
3623 some large reply buffers so that these replies can be sent even
3624 under memory pressure.
3628 Frequency : rare (liblustre)
3629 Description: fsx running with liblustre complained that using truncate() to
3630 extend the file doesn't work. This patch corrects that issue.
3631 Details : This is the liblustre equivalent of the fix for bug 6196. Fixes
3632 ATTR_SIZE and lsm use in llu_setattr_raw.
3636 Frequency : rare, only 2.6 kernels
3637 Description: Unusual file access patterns on the MDS may result in inode
3638 data being lost in very rare circumstances.
3639 Details : Bad interaction between the ea-in-inode patch and the "no-read"
3640 code in the 2.6 kernel caused the inode and/or EA data not to
3641 be read from disk, causing single-file corruption.
3645 Frequency : rare, only 2.6 filesystems using extents
3646 Description: Heavy concurrent write and delete load may cause data corruption.
3647 Details : It was possible under high-load situations to have an extent
3648 metadata block in the block device cache from a just-unlinked
3649 file overwrite a newly-allocated data block. We now unmap any
3650 metadata buffers that alias just-allocated data blocks.
3654 Frequency : filesystems with default stripe_count larger than 77
3655 Description: lconf+mke2fs fail when formatting filesystem with > 77 stripes
3656 Details : lconf specifies an inode size of 4096 bytes when the default
3657 stripe_count is larger than 77. This conflicts with the default
3658 inode density of 1 per 4096 bytes. Allocate smaller inodes in
3659 this case to avoid pinning too much memory for large EAs.
3661 ------------------------------------------------------------------------------
3663 2005-07-07 Cluster File Systems, Inc. <info@clusterfs.com>
3668 Frequency : rare (extremely heavy IO load with hundreds of clients)
3670 Description: Client is evicted, gets IO error writing to file
3671 Details : lock ordering changes for bug 5492 reintroduced bug 3267 and
3672 caused clients to be evicted for AST timeouts. The fixes in
3673 bug 5192 mean we no longer need to have such short AST timeouts
3674 so ldlm_timeout has been increased.
3677 Frequency : occasional during --force or --failover shutdown under load
3678 Bugzilla : 5949, 4834
3679 Description: Server oops/LBUG if stopped with --force or --failover under load
3680 Details : a collection of import/export refcount and cleanup ordering
3681 issues fixed for safer force cleanup
3684 Frequency : only filesystems larger than 120 OSTs
3685 Bugzilla : 5990, 6223
3686 Description: lfs getstripe would oops on a very large filesystem
3687 Details : lov_getconfig used kfree on vmalloc'd memory
3690 Frequency : only filesystems exporting via NFS to Solaris 10 clients
3691 Bugzilla : 6242, 6243
3692 Description: reading from files that had been truncated to a non-zero size
3693 but never opened returned no data
3694 Details : ll_file_read() reads zeros from no-object files to EOF
3699 Description: A bug in MDS/OSS recovery could cause the OSS to fail an assertion
3700 Details : There's little harm in aborting MDS/OSS recovery and letting it
3701 try again, so I removed the LASSERT and return an error instead.
3703 Severity : enhancement
3705 Description: New debugging infrastructure for tracking down data corruption
3706 Details : The I/O checksum code was replaced to: (a) control it at runtime,
3707 (b) cover more of the client-side code path, and (c) try to narrow
3708 down where problems occurred
3712 Bugzilla : 3819, 4364, 4397, 6313
3713 Description: Racing close and eviction MDS could cause assertion in mds_close
3714 Details : It was possible to get multiple mfd references during close and
3715 client eviction, leading to one thread referencing a freed mfd.
3717 Severity: : enhancement
3718 Bugzilla : 3262, 6359
3719 Description: Attempts to reconnect to servers are now more aggressive.
3720 Details : This builds on the enhanced upcall-less recovery that was added
3721 in 1.4.2. When trying to reconnect to servers, clients will
3722 now try each server in the failover group every 10 seconds. By
3723 default, clients would previously try one server every 25 seconds.
3728 Description: After recovery, certain operations trigger a failed
3729 assertion on a client.
3730 Details : Failing over an mds, using lconf -d --failover, while a
3731 client was doing a readdir() call would cause the client to
3732 LBUG after recovery completed and the readdir() was resent.
3734 Severity : enhancement
3736 Description: Default groups are now added by lconf
3737 Details : You can now run lconf --group <servicename> without having to
3738 manually add groups with lmc.
3741 Frequency : occasional
3743 Description: Nodes with an elan id of 0 trigger a failed assertion
3746 Frequency : always when accessing e.g. tty/console device nodes
3748 Description: tty and some other devices nodes cannot be used on lustre
3749 Details : file's private_data field is used by device data and lustre
3750 values in there got lost. New field was added to struct file to
3751 store fs-specific private data.
3754 Frequency : when exporting Lustre via NFS
3756 Description: NFSD failed occasionally when looking up a path component
3757 Details : NFSD is looking up ".." which was broken in ext3 directories
3758 that had grown large enough to become hashed.
3761 Frequency : Clusters with multiple interfaces not on the same subnet
3763 Description: Nodes will repeatedly try to reconnect to an interface which it
3764 cannot reach and report an error to the log.
3765 Details : Extra peer list entries will be created by lconf with some peers
3766 unreachable. lconf now validates the peer before adding it.
3769 Frequency : Only if a default stripe is set on the filesystem root.
3771 Description: Setting a default stripe on the filesystem root prevented the
3772 filesystem from being remounted.
3773 Details : The client was sending extra request flags in the root getattr
3774 request and did not allocate a reply buffer for the dir EA.
3777 Frequency : occasional, higher if lots of files are accessed by one client
3778 Bugzilla : 6159, 6097
3779 Description: Client trips assertion regarding lsm mismatch/magic
3780 Details : While revalidating inodes the VFS looks up inodes with ifind()
3781 and in rare cases can find an inode that is being freed.
3782 The ll_test_inode() code will free the lsm during ifind()
3783 when it finds an existing inode and then the VFS later attaches
3784 this free lsm to a new inode.
3788 Bugzilla : 6422, 7030
3789 Description: MDS deadlock between mkdir and client eviction
3790 Details : Creating a new file via mkdir or mknod (starting a transaction
3791 and getting the ns lock) can deadlock with client eviction
3792 (gets ns lock and trying to finish a synchronous transaction).
3795 Frequency : occasional
3796 Description: While starting a server, the fsfilt_ext3 module could not be
3798 Details : CFS's improved ext3 filesystem is named ldiskfs for 2.6
3799 kernels. Previously, lconf would still use the ext3 name
3800 when trying to load modules. Now, it will correctly use
3801 ext3 on 2.4 and ldiskfs on 2.6.
3803 Severity : enhancement
3804 Description: The default stripe count has been changed to 1
3805 Details : The interpretation of the default stripe count (0, to lfs
3806 or lmc) has been changed to mean striping across a single
3807 OST, rather than all available. For general usage we have
3808 found a stripe count of 1 or 2 works best.
3810 Severity : enhancement
3811 Description: Add support for compiling against Cray portals.
3812 Details : Conditional compiling for some areas that are different
3816 Frequency : occasional
3817 Bugzilla : 6409, 6834
3818 Description: Creating files with an explicit stripe count may lead to
3819 a failed assertion on the MDS
3820 Details : If some OSTs are full or unavailable, creating files may
3821 trigger a failed assertion on the MDS. Now, Lustre will
3822 try to use other servers or return an error to the
3826 Frequency : occasional
3828 Description: Multiple concurrent overlapping read+write on multiple SMP nodes
3829 caused lock timeout during readahead (since 1.4.2).
3830 Details : Processes doing readahead might match a lock that hasn't been
3831 granted yet if there are overlapping and conflicting lock
3832 requests. The readahead process waits on ungranted lock
3833 (original lock is CBPENDING), while OST waits for that process
3834 to cancel CBPENDING read lock and eventually evicts client.
3836 Severity : enhancement
3838 Description: Initial enabling of flock support for clients
3839 Details : Implements fcntl advisory locking and file status functions.
3840 This feature is provided as an optional mount flag (default
3841 off), and is NOT CURRENTLY SUPPORTED. Not all types of record
3842 locking are implemented yet, and those that are are not guaranteed
3843 to be completely correct in production environments.
3844 mount -t lustre -o [flock|noflock] ...
3847 Frequency : occasional
3849 Description: OSTs running 2.4 kernels but with extents enabled might trip an
3850 assertion in the ext3 JBD (journaling) layer.
3851 Details : The b_committed_data struct is protected by the big kernel lock
3852 in 2.4 kernels, serializing journal_commit_transaction() and
3853 ext3_get_block_handle->ext3_new_block->find_next_usable_block()
3854 access to this struct. In 2.6 kernels there is finer grained
3855 locking to improve SMP performance of the JBD layer.
3859 Description: Changes the "SCSI I/O Stats" kernel patch to default to "enabled"
3864 Description: merge and cleanup kernel patches.
3867 -----------------------------------------------------------------------------
3869 2005-05-05 Cluster File Systems, Inc. <info@clusterfs.com>
3871 NOTE: Lustre 1.4.2 uses an incompatible network protocol than previous
3872 versions of Lustre. Please update all servers and clients to
3873 version 1.4.2 or later at the same time. You must also run
3874 "lconf --write-conf {config}.xml" on the MDS while it is stopped
3875 to update the configuration logs.
3877 - fix for HPUX NFS client breakage when NFS exporting Lustre (5781)
3878 - mdc_enqueue does not need max_mds_easize request buffer on send (5707)
3879 - swab llog records of type '0' so we get proper header size/idx (5861)
3880 - send llog cancel req to DLM cancel portal instead of cb portal (5515)
3881 - fix rename of one directory over another leaking an inode (5953)
3882 - avoid SetPageDirty on 2.6 (5981)
3883 - don't re-add just-being-destroyed locks to the waiting list (5653)
3884 - when creating new directories, inherit the parent's custom
3885 striping settings if present parent (3048)
3886 - flush buffers from cache before direct IO in 2.6 obdfilter (4982)
3887 - don't hold i_size_sem in ll_nopage() and ll_ap_refresh_count (6077)
3888 - don't hold client locks on temporary worklist from l_lru (5666)
3889 - handle IO errors in 2.6 obdfilter bio completion routine (6046)
3890 - automatically evict dead clients (5921)
3891 - Update file size properly in create+truncate+fstat case (6196)
3892 - Do not unhash mountpoint dentries, do not allow removal of
3894 - Avoid lock ordering deadlock issue with write/truncate (6203,5654)
3895 - reserve enough journal credits in fsfilt_start_log for setattr (4554)
3896 - ldlm_enqueue freed-export error path would always LBUG (6149,6184)
3897 - don't reference lr_lvb_data until after we hold lr_lvb_sem (6170)
3898 - don't overwrite last_rcvd if there is a *_client_add() error (6086)
3899 - Correctly handle reads of files with no objects (6243)
3900 - lctl recover will also mark a device active if deactivate used (5933)
3902 - by default create 1 inode per 4kB space on MDS, per 16kB on OSTs
3903 - allow --write-conf on an MDS with different nettype than client (5619)
3904 - don't write config llogs to MDS for mounts not from that MDS (5617)
3905 - lconf should create multiple TCP connections from a client (5201)
3906 - init scripts are now turned off by default; run chkconfig --on
3907 lustre and chkconfig --on lustrefs to use them
3908 - upcalls are no longer needed for clients to recover to failover
3910 - add --abort-recovery option to lconf to abort recovery on device
3912 - add support for an arbitrary number of OSTs (3026)
3913 - Quota support protocol changes.
3914 - forward compatibility changes to wire structs (6007)
3915 - rmmod NALs that might be loaded because of /etc/modules.conf (6133)
3916 - support for mountfsoptions and clientoptions to the Lustre LDAP (5873)
3917 - improved "lustre status" script
3918 - initialize blocksize for non-regular files (6062)
3919 - added --disable-server and --disable-client configure options (5782)
3920 - introduce a lookup cache for lconf to avoid repeated DB scans (6204)
3921 - Vanilla 2.4.29 support
3922 - increase maximum number of obd devices to 520 (6242)
3923 - remove the tcp-zero-copy patch from the suse-2.4 series (5902)
3924 - Quadrics Elan drivers are now included for the RHEL 3 2.4.21 and
3925 SLES 9 2.6.5 kernels
3926 - limit stripes per file to 160 (the maximum EA size) (6093)
3928 2005-03-22 Cluster File Systems, Inc. <info@clusterfs.com>
3931 - don't LASSERT in ll_release on NULL lld with NFS export (4655, 5760)
3932 - hold NS lock when calling handle_ast_error->del_waiting_lock (5746)
3933 - fix setattr mtime regression from lovcleanup merge (4829, 5669)
3934 - workaround for 2.6 crash in ll_unhash_aliases (5687, 5210)
3935 - small ext3 extents cleanups and fixes (5733)
3936 - improved mballoc code, several small races and bugs fixed (5733, 5638)
3937 - kernel version 43 - fix remove_suid bugs in both 2.4 and 2.6 (5695)
3938 - avoid needless client->OST connect, fix handle mismatch (5317)
3939 - fix DLM error path that led to out-of-sync client, long delays (5779)
3940 - support common vfs-enforced mount options (nodev,nosuid,noexec) (5637)
3941 - fix several locking issues related to i_size (5492,5624,5654,5672)
3942 - don't move pending lock onto export if it is already evicted (5683)
3943 - fix kernel oops when creating .foo in unlinked directory (5548)
3944 - fix deadlock in obdfilter statistics vs. object create (5811)
3945 - use time_{before,after} to avoid timer jiffies wrap (5882)
3946 - shutdown --force/--failover stability (3607,3651,4797,5203,4834)
3947 - Do not leak request if server was not able to process it (5154)
3948 - If mds_open unable to find parent dir, make that negative lookup(5154)
3949 - don't create new directories with extent-mapping (5909, 5936)
3951 - fix lustre/lustrefs init scripts for SuSE (patch from Scali, 5702)
3952 - don't hold the pinger_sem in ptlrpc_pinger_sending_on_import
3953 - change obd_increase_kms to obd_adjust_kms (up or down) (5654)
3954 - lconf, lmc search both /usr/lib and /usr/lib64 for Python libs (5800)
3955 - support for RHEL4 kernel on i686 (5773)
3956 - provide error messages when incompatible logs are encountered (5898)
3958 2005-02-18 Cluster File Systems, Inc. <info@clusterfs.com>
3959 * version 1.4.0.10 (1.4.1 release candidate 1)
3961 - don't keep a lock reference when lock is not granted (4238)
3962 - unsafe list practices (rarely) led to infinite eviction loop (4908)
3963 - add per-fs limit of Lustre pages in page cache, avoid OOM (4699)
3964 - drop import inflight refcount on signal_completed_replay error (5255)
3965 - unlock page after async write error during send (3677)
3966 - handle missing objects in filter_preprw_read properly (5265)
3967 - no transno return for symlink open, don't save no-trasno open (3440)
3968 - don't try to complete elan receive that already failed (4012)
3969 - free RPC server reply state on error (5406)
3970 - clean up thread from ptlrpc_start_thread() on error (5160)
3971 - readahead could read extra page into cache that wasn't ejected (5388)
3972 - prevent races in class_attach/setup/cleanup/detach (5260)
3973 - don't dereference de->d_inode after l_dput of de (5458)
3974 - use "int" for stripe value returned from lock_to_stripe (5544)
3975 - mballoc allocation and error-checking fixes in 2.6 (5504)
3976 - block device patches to fix I/O request sizes in 2.6 (5482)
3977 - look up hostnames for IB nals (5602)
3978 - 2.6 changed lock ordering of 2 semaphores, caused deadlock (5654)
3979 - don't start multiple acceptors for the same port (5277)
3980 - fix incorrect LASSERT in mds_getattr_name (5635)
3981 - export a proc file for general "ping" checking (5628)
3982 - fix "lfs check" to not block when the MDS is down (5628)
3984 - service request history (4965)
3985 - put {ll,lov,osc}_async_page structs in a single slab (4699)
3986 - create an "evict_client" /proc entry on OSTs, like the MDS has
3987 - fix mount usage message, return errors per mount(8) (5168)
3988 - change grep [] to grep "[]" in tests so they work in more UMLs
3989 - fix ppc64/x86_64 spec to use %{_libdir} instead of /usr/lib (5389)
3990 - remove ancient LOV_MAGIC_V0 EA support (5047)
3991 - add "disk I/Os in flight" and "I/O req time" stats in obdfilter
3992 - align r/w RPCs to PTLRPC_MAX_BRW_SIZE boundary for performance (3451)
3993 - allow readahead allocations to fail when low on memory (5383)
3994 - mmap locking landed again, after considerable improvement (2828)
3995 - add get_hostaddr() to lustreDB.py for LDAP support (5459)
3997 2004-11-23 Cluster File Systems, Inc. <info@clusterfs.com>
4000 - send OST transaction number in read/write reply to free req (4966)
4001 - don't ASSERT in ptl_send_rpc() if we run out of memory (5119)
4002 - lock /proc/sys/portals/routes internal state, avoiding oops (4827)
4003 - the watchdog thread now runs as interruptible (5246)
4004 - flock/lockf fixes (but it's still disabled, pending 5135)
4005 - don't use EXT3 constants in llite code (5094)
4006 - memory shortage at startup could cause assertion (5176)
4008 - reorganization of lov code
4009 - single portals codebase
4011 - add extents/mballoc support (5025)
4012 - direct I/O reads in the obdfilter (4048)
4013 - kernel patches from LNXI for 2.6 (bluesmoke, perfctr, mtd, kexec)
4015 tbd Cluster File Systems, Inc. <info@clusterfs.com>
4018 - send OST transaction number in read/write reply to free req (4966)
4019 - don't ASSERT in ptl_send_rpc() if we run out of memory (5119)
4020 - lock /proc/sys/portals/routes internal state, avoiding oops (4827)
4021 - the watchdog thread now runs as interruptible (5246)
4022 - handle missing objects in filter_preprw_read properly (5265)
4023 - unsafe list practices (rarely) led to infinite eviction loop (4908)
4024 - drop import inflight refcount on signal_completed_replay error (5255)
4025 - unlock page after async write error during send (3677)
4026 - return original error code on reconstructed replies (3761)
4027 - no transno return for symlink open, don't save no-trasno open (3440)
4029 - add pid to ldlm debugging output (4922)
4030 - bump the watchdog timeouts -- we can't handle 30sec yet
4031 - extra debugging for orphan dentry/inode bug (5259)
4033 2004-11-16 Cluster File Systems, Inc. <info@clusterfs.com>
4036 - fix TCP_NODELAY bug, which caused extreme perf regression (5134)
4037 - allocate qswnal tx descriptors singly to avoid fragmentation (4504)
4038 - don't LBUG on obdo_alloc() failure, use OBD_SLAB_ALLOC() (4800)
4039 - fix NULL dereference in /proc/sys/portals/routes (4827)
4040 - allow failed mdc_close() operations to be interrupted (4561)
4041 - stop precreate on OST before MDS would time out on it (4778)
4042 - don't send partial-page writes before EOF from client (4410)
4043 - discard client grant for sub-page writes on large-page clients (4520)
4044 - don't free dentries not owned by NFS code, check generation (4806)
4045 - fix lsm leak if mds_create_objects() fails (4801)
4046 - limit debug_daemon file size, always print CERROR messages (4789)
4047 - use transno after validating reply (3892)
4048 - process timed out requests if import state changes (3754)
4049 - update mtime on OST during writes, return in glimpse (4829)
4050 - add mkfsoptions to LDAP (4679)
4051 - use ->max_readahead method instead of zapping global ra (5039)
4052 - don't interrupt __l_wait_event() during strace
4054 - add software watchdogs to catch hung threads quickly (4941)
4055 - make lustrefs init script start after nfs is mounted
4056 - fix CWARN/ERROR duplication (4930)
4057 - return async write errors to application if possible (2248)
4058 - add /proc/sys/portal/memused (bytes allocated by PORTALS_ALLOC)
4059 - print NAL number in %x format (4645)
4060 - update barely-supported suse-2.4.21-171 series (4842)
4061 - support for sles 9 %post scripts
4062 - support for building 2.6 kernel-source packages
4063 - support for sles km_* packages
4065 2004-10-07 Cluster File Systems, Inc. <info@clusterfs.com>
4068 - ignore -ENOENT errors in osc_destroy (3639)
4069 - notify osc create thread that OSC is being cleaned up (4600)
4070 - add nettype argument for llmount in #5d in conf-sanity.sh (3936)
4071 - reconstruct ost_handle() like mds_handle() (4657)
4072 - create a new thread to do import eviction to avoid deadlock (3969)
4073 - let lconf resolve symlinked-to devices (4629)
4074 - don't unlink "objects" from directory with default EA (4554)
4075 - hold socknal file ref over connect in case target is down (4394)
4076 - allow more than 32000 subdirectories in a single directory (3244)
4077 - fix blocks count for O_DIRECT writes (3751)
4078 - OST returns ENOSPC from object create when no space left (4539)
4079 - don't send truncate RPC if file size isn't changing (4410)
4080 - limit OSC precreate to 1/2 of value OST considers bogus (4778)
4081 - bind to privileged port in socknal and tcpnal (3689)
4083 - rate limit CERROR/CWARN console message to avoid overload (4519)
4084 - GETFILEINFO dir ioctl returns LOV EA + MDS stat in 1 call (3327)
4085 - basic mmap support (3918)
4086 - kernel patch series update from b1_4 (4711)
4088 2004-09-16 Cluster File Systems, Inc. <info@clusterfs.com>
4091 - avoid crash during MDS cleanup with OST shut down (2775)
4092 - fix loi_list_lock/oig_lock inversion on interrupted IO (4136)
4093 - don't use bad inodes on the MDS (3744)
4094 - dynamic object preallocation to improve recovery speed (4236)
4095 - don't hold spinlock over lock dumping or change debug flags (4401)
4096 - don't zero obd_dev when it is force cleaned (3651)
4097 - print grants to console if they go negative (4431)
4098 - "lctl deactivate" will stop automatic recovery attempts (3406)
4099 - look for existing locks in ldlm_handle_enqueue() (3764)
4100 - don't resolve lock handle twice in recovery avoiding race (4401)
4101 - revalidate should check working dir is a directory (4134)
4103 - don't always mark "slow" obdfilter messages as errors (4418)
4105 2004-08-24 Cluster File Systems, Inc. <info@clusterfs.com>
4108 - don't close LustreDB during write_conf until it is done (3860)
4109 - fix typo in lconf for_each_profile (3821)
4110 - allow dumping logs from multiple threads at one time (3820)
4111 - don't allow multiple threads in OSC recovery (3812)
4112 - fix debug_size parameters (3864)
4113 - fix mds_postrecov to initialize import for llog ctxt (3121)
4114 - replace config semaphore with spinlock (3306)
4115 - be sure to send a reply for a CANCEL rpc with bad export (3863)
4116 - don't allow enqueue to complete on a destroyed export (3822)
4117 - down write_lock before checking llog header bitmap (3825)
4118 - recover from lock replay timeout (3764)
4119 - up llog sem before sending rpc (3652)
4120 - reduce ns lock hold times when setting kms (3267)
4121 - change a dlm LBUG to LASSERTF, to maybe learn something (4228)
4122 - fix NULL deref and obd_dev leak on setup error (3312)
4123 - replace some LBUG about llog ops with error handling (3841)
4124 - don't match INVALID dentries from d_lookup and spin (3784)
4125 - hold dcache_lock while marking dentries INVALID and hashing (4255)
4126 - fix invalid assertion in ptlrpc_set_wait (3880)
4128 - add libwrap support for the TCP acceptor (3996)
4129 - add /proc/sys/portals/routes for non-root route listing (3994)
4130 - allow setting MDS UUID in .xml (2580)
4131 - print the stack of a process that LBUGs (4228)
4133 2004-07-14 Cluster File Systems, Inc. <info@clusterfs.com>
4136 - don't cleanup request in ll_file_open() on failed MDS open (3430)
4137 - make sure to unset replay flag from failed open requests (3440)
4138 - if default stripe count is 0, use OST count for inode size (3636)
4139 - update parent mtime/ctime on client for create/unlink (2611)
4140 - drop dentry ref in ext3_add_link from open_connect_dentry (3266)
4141 - free recovery state on server during a forced cleanup (3571)
4142 - unregister_reply for resent reqs (3063)
4143 - loop back devices mounting and status check on 2.6 (3563)
4144 - fix resource-creation race that can provoke i_size == 0 (3513)
4145 - don't try to use bad inodes returned from MDS/OST fs lookup (3688)
4146 - more debugging for page-accounting assertion (3746)
4147 - return -ENOENT instead of asserting if ost getattr+unlink race (3558)
4148 - avoid deadlock after precreation failure (3758)
4149 - fix race and lock order deadlock in orphan handling (3450, 3750)
4150 - add validity checks when grabbing inodes from l_ast_data (3599)
4152 - add /proc/.../recovery_status to obdfilter (3428)
4153 - lightweight CDEBUG infrastructure, debug daemon (3668)
4154 - change default OSC RPC parameters to be better on small clusters
4155 - turn off OST read cache for files smaller than 32MB
4156 - install man pages and include them in rpms (3100)
4157 - add new init script for (un)mounting lustre filesystems (2593)
4158 - run chkconfig in %post for init scripts (3701)
4159 - drop scimac NAL (unmaintained)
4161 2004-06-17 Cluster File Systems, Inc. <info@clusterfs.com>
4164 - clean kiobufs before and after use (3485)
4165 - strip trailing '/'s before comparing paths with /proc/mounts (3486)
4166 - remove assertions to work around "in-flight rpcs" recovery bug (3063)
4167 - change init script to fail more clearly if not run as root (1528)
4168 - allow clients to reconnect during replay (1742)
4169 - fix ns_lock/i_sem lock ordering deadlock for kms update (3477)
4170 - don't do DNS lookups on NIDs too small for IP addresses (3442)
4171 - re-awaken ptlrpcd if new requests arrive during check_set (3554)
4172 - fix cond_resched (3554)
4173 - only evict unfinished clients after recovery (3515)
4174 - allow bulk resend, prevent data loss (3570)
4175 - dynamic ptlrpc request buffer allocation (2102)
4176 - don't allow unlinking open directory if it isn't empty (2904)
4177 - set MDS/OST threads to umask 0 to not clobber client modes (3359)
4178 - remove extraneous obd dereference causing LASSERT failure (3334)
4179 - don't use get_cycles() when creating temp. files on the mds (3156)
4180 - hold i_sem when setting i_size in ll_extent_lock() (3564)
4181 - handle EEXIST for set-stripe, set proper directory name (3336)
4183 - servers can dump a log evicting a client - lustre.dump_on_timeout=1
4184 - fix ksocknal_fmb_callback() error messages (2918)
4186 2004-05-27 Cluster File Systems, Inc. <info@clusterfs.com>
4189 - don't copy lvb into (possibly NULL) reply on error (2983)
4190 - don't deref dentry after dput, don't free lvb on error (2922)
4191 - use the kms to determine writeback rpc length (2947)
4192 - increment oti_logcookies when osc is inactive (2948)
4193 - update client's i_blocks count via lvb messages (2543)
4194 - handle intent open/close of special files properly (1557)
4195 - mount MDS with errors=remount-ro, like obdfilter (2009)
4196 - initialize lock handle to avoid ASSERT on error cleanup (3057)
4197 - don't use cancelling-locks' kms values (2947)
4198 - use highest lock extent for kms, not last one (2925)
4199 - don't dereference ERR_PTR() dentry in error handling path (3107)
4200 - fix thread race in portals_debug_dumplog() (3122)
4201 - create lprocfs device entries at setup instead of at attach (1519)
4202 - common AST error handler, don't evict client on completion race (3145)
4203 - zero nameidata in detach_mnt in 2.6 (3118)
4204 - verify d_inode after revalidate_special is valid in 2.6 (3116)
4205 - use lustre_put_super() to handle zconf unmounts in 2.6 (3064)
4206 - initialize RPC timeout timer earlier for 2.6 (3219)
4207 - don't dereference NULL reply buffer if mdc_close was never sent (2410)
4208 - print nal/nid for unknown nid (3258)
4209 - additional checks for oscc recovery before doing precreate (3284)
4210 - fix ll_extent_lock() error return code for 64-bit systems (3043)
4211 - don't crash in mdc_close for bad permissions on open (3285)
4212 - zero i_rdev for non-device files (3147)
4213 - clear page->private before handing to FS, better assertion (3119)
4214 - tune the read pipeline (3236)
4215 - fix incorrect decref of invalidated dentry (2350)
4216 - provide read-ahead stats and refine rpc in flight stats (3328)
4217 - don't hold journal transaction open across create RPC (3313)
4218 - update atime on MDS at close time (3265)
4219 - close LDAP connection when recovering to avoid server load (3315)
4220 - update iopen-2.6 patch with fixes from 2399,2517,2904 (3301)
4221 - don't leak open file on MDS after open resend (3325)
4222 - serialize filter_precreate and filter_destroy_precreated (3329)
4223 - loop device shouldn't call sync_dev() for nul device (3092)
4224 - clear page cache after eviction (2766)
4225 - resynchronize MDS->OST in background (2824)
4226 - refuse to mount the same filesystem twice on same mountpoint (3394)
4227 - allow llmount to create routes for mounting behind routers (3320)
4228 - push lock cancellation to blocking thread for glimpse ASTs (3409)
4229 - don't call osc_set_data_with_check() for TEST_LOCK matches (3159)
4230 - fix rare problem with rename on htree directories (3417)
4232 - allow default OST striping configuration per directory (1414)
4233 - fix compilation for qswnal for 2.6 kernels (3125)
4234 - increase maximum number of MDS request buffers for large systems
4235 - change liblustreapi to be useful for external progs like lfsck (3098)
4236 - increase local configuration timeout for slow disks (3353)
4237 - allow configuring ldlm AST timeout - lustre.ldlm_timeout=<seconds>
4239 2004-03-22 Cluster File Systems, Inc. <info@clusterfs.com>
4242 - fixes for glimpse AST timeouts / incorrectly 0-sized files (2818)
4243 - don't overwrite extent policy data in reply if lock was blocked (2901)
4244 - drop filter export grants atomically with removal from device (2663)
4245 - del obd_self_export from work_list in class_disconnect_exports (2908)
4246 - don't LBUG if MDS recovery times out during orphan cleanup (2530)
4247 - swab reply message in mdc_close, other PPC fixes (2464)
4248 - fix destroying of named logs (2325)
4249 - overwrite old logs when running lconf --write_conf (2264)
4250 - bump LLOG_CHUNKSIZE to 8k to allow for larger clusters (2306)
4251 - fix race in target_handle_connect (2898)
4252 - mds_reint_create() should take same inode create lock (2926)
4253 - correct journal credits calculated for CANCEL_UNLINK_LOG (2931)
4254 - don't close files for self_export to avoid uninitialized obd (2936)
4255 - allow MDS with the same name as client node (2939)
4256 - hold dentry reference for closed log files for unlink (2325)
4257 - reserve space for all logs during transactions (2059)
4258 - don't evict page beyond end of stripe extent (2925)
4259 - don't oops on a deleted current working directory (2399)
4260 - handle hard links to targets without a parent properly (2517)
4261 - don't dereference NULL lock when racing during eviction (2867)
4262 - don't grow lock extents when lots of conflicting locks (2919)
4264 2004-03-04 Cluster File Systems, Inc. <info@clusterfs.com>
4267 - account for cache space usage on clients to avoid data loss (974)
4268 - lfsck support in lustre kernel code (2349)
4269 - reduce journal credits needed for BRW writes (2370)
4270 - orphan handling to avoid losing space on client/server crashes
4271 - ptlrpcd can be blocked, stopping ALL progress (2477)
4272 - use lock value blocks to assist in proper KMS, faster stat (1021)
4273 - takes i_sem instead of DLM locks internally on obdfilter (2720)
4274 - recovery for initial connections (2355)
4275 - fixes for mds_cleanup_orphans (1934)
4276 - abort_recovery crashes MDS in b_eq (mds_unlink_orphan) (2584)
4277 - block all file creations until orphan recovery completes (1901)
4278 - client remove rq_connection from request struct (2423)
4279 - conf-sanity test_5, proper cleanup in umount log not availale (2640)
4280 - recovery timer race (2670)
4281 - mdc_close recovey bug (2532)
4282 - ptlrpc cleanup bug (2710)
4283 - mds timeout on local locks (2588)
4284 - namespace lock held during RPCs (2431)
4285 - handle interrupted sync write properly (2503)
4286 - don't try to handle a message that hasn't been replied to (2699)
4287 - client assert failure during cleanup after abort recovery (2701)
4288 - leak mdc device after failed mount (2712)
4289 - ptlrpc_check_set allows timedout requests to complete (2714)
4290 - wait for inflight reqs when ptlrpcd finishes (2710)
4291 - make sure unregistered services are removed from the srv_list
4292 - reset bulk XID's when resending them (caught by 1138 test)
4293 - unregister_bulk after timeout
4294 - fix lconf error (2694)
4295 - handle write after unfinished setstripe, stripe-only getstripe (2388)
4296 - readahead locks pages, leaves pending causing memory pressure (2673)
4297 - increase OST request buffers to 4096 on large machines (2729)
4298 - fix up permission of existing directories in simple_mkdir (2661)
4299 - init deleted item, add assertions ptlrpc_abort_inflight() (2725)
4300 - don't assign transno to errored transactions (2742)
4301 - don't delete objects on OST if given a bogus objid from MDS (2751)
4302 - handle large client PAGE_SIZE readdir on small PAGE_SIZE MDS (2777)
4303 - if rq_no_resend, then timeout request after recovery (2432)
4304 - fix MDS llog_logid record size, 64-bit array alignment (2733)
4305 - don't call usermode_helper from ptlrpcd, DEFAULT upcall (2773)
4306 - put magic in mount.lustre data, check for bad/NULL mount data (2529)
4307 - MDS recovery shouldn't delete objects that it has given out (2730)
4308 - if enqueue arrives after completion, don't clobber LVB (2819)
4309 - don't unlock pages twice when trigger_group_io returns error (2814)
4310 - don't deref NULL rq_repmsg if ldlm_handle_enqueue failed (2822)
4311 - don't write pages to disk if there was an error (1450)
4312 - don't ping imports that have recovery disabled (2676)
4313 - take buffered bytes into account when balancing socknal conn (2817)
4314 - hold a DLM lock over readdir always, use truncate_inode_pages (2706)
4315 - reconnect unlink llog connection after MDS reconnects to OST (2816)
4316 - remove little-endian swabbing of llog records (1987)
4317 - set/limit i_blksize to LL_MAX_BLKSIZE on client (2884)
4318 - retry reposting request buffers if they fail (1191)
4319 - grow extent at grant time to avoid granting a revoked lock (2809)
4320 - lock revoke doesn't evict page if covered by a second lock (2765)
4321 - disable VM readahead to avoid reading outside lock extents (2805)
4323 - return LL_SUPER_MAGIC from statfs for the filesystem type (1972)
4324 - updated kernel patches for hp-2.4.20 kernel (2681)
4326 2004-02-07 Cluster File Systems, Inc. <info@clusterfs.com>
4329 - fix truncated write corruption (2366)
4330 - fix for failed assertion in iopen_connect_dentry (1792,2517)
4332 - don't flag the ptlrpcd thread with PF_MEMALLOC (2636)
4333 - ensure len(uuid) < 37 in lmc (1171)
4334 - fix ia64 OOPS in llog_test (2255)
4335 - zero end of page at obdfilter for partial page writes (2648)
4336 - don't leave stale dentries around after renames (bug 2428)
4337 - fix timeouts when evicting a client with a single lock held (2642)
4338 - set deadline for the initial HELLO message to drain (2634)
4339 - print out dotted-quad IP addresses in the socknal (2302)
4341 - additional debugging for MDS client eviction problem (2443)
4342 - fix mkfsoptions support for osts (2603, 2604)
4344 2004-01-27 Cluster File Systems, Inc. <info@clusterfs.com>
4347 - add series for the vanilla 2.6.0 kernel
4348 - add series for the vanilla 2.4.24 kernel
4349 - add series for a cray x86/64 UL kernel drop
4350 - fix xattr patches for the vanilla 2.4.19 series
4352 - generate true UUIDs in lmc (1171)
4353 - have portals stack dumping break in UML (2466)
4354 - avoid bad dchild deref; avoid inum lock w/o creation (2362)
4355 - allocate with _NOFS in ldlm to avoid deadlock (1933)
4356 - wake callback waiting threads on client eviction (2460)
4357 - Add --ptldebug and --subsystem to lmc (1719)
4358 - update assertion to allow safe interrupt allocation
4359 - set rq_no_resend for cancel requests (2432)
4360 - recalculate ptlrpcd timeout after resend (2494)
4361 - call vfs_rmdir when removing pending directories (2368)
4362 - fix renaming a file to itself (2429)
4363 - lmc creates a default one-stripe lov (2454)
4364 - expand procfs space to handle large clusters (2326)
4365 - increase UML stack to avoid overflow
4366 - update lconf's list of debug and subsystem masks
4367 - fix lfs find --obd (2510)
4368 - /proc tunable for disabling filter read caching (2591)
4369 - stop rpm packages from altering slapd.conf (2301)
4370 - disable nagle in the socknal under 0conf (2578)
4371 - choose mds inode size based on stripe count (2572)
4372 - fix kernel-source rpm problems (2516)
4374 - add --disable-doc to avoid pdf generation (2421)
4375 - update documentation, tests, type-os, comments
4376 - avoid format warnings on ia64
4377 - remove the TOE NAL
4378 - tiny code cleanups by removing unused fields
4380 2004-01-07 Cluster File Systems, Inc. <info@clusterfs.com>
4383 - fix obvious semaphore misuse in as-yet-unused setattr path (2348)
4384 - remove the most blatant lies from BUILDING file (2371)
4385 - change default debug level to reasonable production setting
4386 - reduce client side cache size to reduce cache flush time
4387 - reduce max RPCs in flight to avoid unnecessary file fragmentation
4388 - make TCP zerocopy and pinger support enabled by default (2476)
4389 - sync writes completed after process exits caused crashes (2319)
4390 - maintain correct mount count on the MDS (2356)
4391 - backout 1557, because 2316 wasn't really fixed
4392 - better file I/O statistics gathering in /proc
4393 - don't take unnecessary, deadlock-inducing bug in readpage (2383)
4394 - another kernel patch to fix zero-copy TCP function export
4395 - don't take duplicate lock when processing re-sent getattr (2420)
4396 - lctl uses obd_self_export instead of creating new conn (2353)
4397 - MDS/OST recovery case which requires object creation asserted (2425)
4398 - move lfs from /usr/sbin to /usr/bin in packages
4399 - fix race between mds_client_add and mds_client_free (2417)
4400 - use kmalloc instead of slabs in portals (2430)
4401 - don't create duplicate records when a failover MDS is present (2442)
4402 - remove unnecessary mount age check (2332)
4403 - don't remove directory inodes from locks prematurely (2451)
4404 - don't break if MDS service name is the same as hostname (2103)
4405 - fix races in client write RPC generation when cache full (2482)
4407 2003-12-13 Cluster File Systems, Inc. <info@clusterfs.com>
4410 - remove now-unused request->rq_obd (278)
4411 - if an allocation fails, print out how much memory we've used (1933)
4412 - use PORTAL_SLAB_ALLOC for structures, to get GFP_MEMALLOC (1933)
4413 - add the "configurable stack size" patch to most series files (1256)
4414 - ability to write large log records, for 100+ OST configs (2306)
4415 - fix NULL deref when filter_prep fails (2314)
4416 - fix operator precedence error in filter_sync
4417 - dynamic allocation of socknal TX descriptors (2315)
4418 - fix a missed case in the GFP_MEMALLOC patch, can cause deadlock (2310)
4419 - fix gcc 2.96 compilation problem in xattr kernel patch (2294)
4420 - ensure that CWARN messages in Portals always get to the syslog
4421 - __init/__exit are not for prototype decls (ldlm_init/exit)
4422 - x86-64 compile warning fixes
4423 - fix gateway LMC keyword conflict (2318)
4424 - fix MDS lock inversions in getattr/reint paths (1844)
4425 - fix a rare lock re-ordering bug, which caused deadlock (2322)
4426 - fix i_sem/journal inversion in fsfilt_ext3_write_record (2306)
4427 - DLM race condition prevented some lock evictions (2328)
4428 - ENOMEM detection and retry on socknal sends (2230)
4429 - use GFP_NOFS throughout Lustre, to combat ENOMEM (2230)
4430 - move osc_rpcd into ptlrpc, for use in MDC and others (2329)
4431 - protect MDS inode fsdata with stronger locking; fixes assertion (2313)
4432 - better error messages when a client is rejected during recovery (1505)
4433 - avoid cancelling locks which were never granted, after failure (2330)
4434 - fix i_sem/journal inversion in mds_client_add (2333)
4435 - fix truncate/getattr lock cycle deadlock (2334)
4436 - use rpcd to send close; allows resend after timeout, avoid leak (1897)
4437 - fix two rare exit paths which could leak an l_lock() ref (2321)
4438 - fencepost error in MDS/OST orphan recovery (2226)
4439 - make log record alignment 8 bytes (1988)
4440 - lstripe now fails when requested offset > ost_count (2237)
4441 - ensure that all kernel series have a complete list.h (1607)
4442 - fix crashes in special-file operations (2316)
4443 - lctl create/brw OID mismatch, caused by obsolete filter loop (2339)
4445 - allow configurable automake binary, for testing new versions
4446 - small update to the lfs documentation
4448 2003-12-03 Cluster File Systems, Inc. <info@clusterfs.com>
4450 * fix negative export reference count in fsfilt_sync (2312)
4452 2003-12-01 Cluster File Systems, Inc. <info@clusterfs.com>
4453 * release candidate 0.9.1
4455 - orphans are moved into the PENDING directory for possible recovery
4456 - replayed opens now open by fid for orphan/rename safety (1042)
4457 - last close of an orphan inode generates a transno (683)
4458 - chdir() and mount() now pin the directory entry (1020)
4459 - avoid CERROR in normal ll_setattr_raw() error case (1500)
4460 - discard very old requests without processing them (1502)
4461 - remove some common, well-understood CERRORs (1505)
4462 - require O_DIRECT I/O to be page-sized to workaround IA64 crash (1609)
4463 - clear "grant" flags in OST replies until OST grant code lands (1644)
4464 - fix read performance by not clobbering i_blksize on client (1598)
4465 - fix __ldlm_handle2lock oops by not dereferencing lock after PUT (1625)
4466 - make LRU size a /proc tunable, clears locks when reduced (707)
4467 - fix some lprocfs rot that prevented ptlbd from loading (1732)
4468 - server locks take references on exports now (1558)
4469 - build fixes for 2.4.20-rh trees (1663)
4470 - return an error from lov_create if all OSCs are inactive (1751)
4471 - fix import levels when a reconnect happens without a timeout (1597)
4472 - exit early from mds_open if we get a lookup error (1749)
4473 - partial page read at EOF wouldn't wait for disk before sending (1642)
4474 - avoid NULL deref in obdfilter when reading page past EOF (1592)
4475 - avoid LASSERT in ll_intent_lock if server failed very early (1090)
4476 - fix LBUG in ll_it_open_error with rc = -2 (1861)
4477 - write/truncate lock inversion (1639)
4478 - Don't auto-load obdclass, portals modules during cleanup (1495)
4479 - fix timestamps from jumping to "now" (1763)
4480 - extra journal assertions (1648)
4481 - add an extra multiunlink test (1771)
4482 - fix read_record/write_record API (1776)
4483 - fix leak of offset_extent, possible incorrect i_size later (1772)
4484 - fix lasserts in mis-matched transnos during open-unlink testing (1541)
4485 - Debugging for the kqswnal_get_idle_tx problems (1820)
4486 - Allow recovery to be attempted multiple times (1536)
4487 - Write out MDS last_rcvd file after it is first created (1600)
4488 - Fix tx_descriptor leak in failed transmit situations (1827)
4489 - ext3 journaling fixes for assertion failure after IO error (1871)
4490 - class_export_put() on freed export after completion AST error (1896)
4491 - Fix revalidate looping in VFS (1322)
4492 - Don't access a freed export during MDS_REINT timeout (1521)
4493 - Add open-unlink recovery support on the MDS (1673,1764)
4494 - Return an error if no MDS data was read from last_rcvd (1946)
4495 - Fix for lookup "." or ".." crash on error (1932,1931,1935)
4496 - Don't setup a disk device that doesn't match exported UUID (317)
4497 - Reduce bulk RPC timeout to avoid cascading client/OST failures (1845)
4498 - avoid committing NULL handle in force close
4499 - local.sh is now a one-stripe LOV configuration
4500 - POSIX utime.4 -EPERM on FIFO not owned by user (56)
4501 - fix ext3 htree duplicate directory entry corruption (1516)
4502 - POSIX creat.13, fstat.1, open.18, stat.3 new file atime/mtime (2020)
4503 - update to new LOV EA format (2097)
4504 - interoperability for different PAGE_SIZE/wordsize (686,1821,1343,2042)
4506 2003-06-15 Phil Schwan <phil@clusterfs.com>
4509 - imports and exports cleanup too early, need refcounts (349, 879, 1045)
4510 - per-import/export recovery handling (958, 931, 959)
4511 - multiple last-rcvd slots, for serving multiple FSes (949)
4512 - connections are again shared between multiple imp/exports (963, 964)
4513 - "umount -f" would hang if any requests needed to be sent (393, 978)
4514 - avoid pinning large req buffer by copying for queued messages (989)
4515 - add "uuid" to "lctl device" command to help upcalls (991)
4516 - "open" RPCs with transnos would confuse recovery counters (1037)
4517 - do proper endian conversion of all wire messages (288, 340, 891)
4518 - remove OST bulk get LBUGs, fix ost_brw_write cleanup (1126)
4519 - call waiting locks callback from LDLM recovery thread (1127, 1151)
4520 - fix ptlrpc_connection leak in target_handle_connect (1174)
4521 - fix import refcounting bug in OST and MDS cleanup (1134)
4522 - if an invalid-at-open-time OSC returned before close(), LBUG (1150)
4523 - fix very unlikely obd_types race condition (501)
4524 - remove osc_open hack for echo_client (1187)
4525 - we leaked exports/dlmimps for forcibly disconnected clients (1143)
4526 - a failure in read_inode2 leads to deadlock (1139)
4527 - cancel ack-locks as soon as transaction is committed (1072)
4528 - fix major leaks and crashes in the bulk I/O path (937, 1057)
4529 - make sure to commitrw after any preprw to avoid deadlock (1162)
4530 - failing to execute a file in a lustre FS would lock inode (1203)
4531 - small DEBUG_REQ fix to avoid dereferencing a NULL (1227)
4532 - don't ASSERT while cleaning up an incompletely-setup obd (1248)
4533 - obd_uuid2tgt would walk off the end of the list (1255)
4534 - on IA64 the osc would give portals incorrect bulk size (1258)
4535 - fix debug daemon ioctl interface; allows daemon on ia64 (1274)
4536 - fix lock inversion caused by new llite matching code (1282)
4537 - limit the number of dirty pages on a client to 10MB (1286)
4538 - timed out locks were not being corrected cancelled (1289)
4539 - fix O_DIRECT above 4GB on IA-32 (1292)
4540 * major user-visible changes
4541 - fail out/fail over policy now controlled by the upcall (993)
4543 - add OBD_PING to check server availability and failure (954)
4544 - lustre messages are now sent in sending host order (288, 340, 891)
4545 - add eadatalen to MDS getattr reply (340)
4546 - OST read replies may contain second buffer, with per-page status (593)
4548 2003-03-11 Phil Schwan <phil@clusterfs.com>
4551 - LDLM_DEBUG macro fix, for gcc 3.2 (850)
4552 - failed open()s could cause deadlock; fixed (867, 869)
4553 - stop cancelling OST locks when files are closed (481)
4554 - overlapping XID spaces caused network corruption (851, 853)
4555 - fix unsafe fsfilt counter arithmetic; change to atomic_t
4556 - setattr_raw added, to do single-RPC, server-side setattrs
4557 - lmc/lconf syntax change for OST UUIDs
4558 - fix crashy race condition between ptlrpc_free_req and osc_close
4559 - don't use request in mdc_enqueue if we hit a timeout (889)
4560 - don't set the inode i_size for regular files from the MDS (896)
4561 - handle out of order completion AST (842)
4562 - don't LBUG if a lock request times out after receiving AST (913)
4563 - avoid d_rehash race in ll_find_alias by rehashing inside dcache_lock
4564 - if a bad lock AST arrives, send an error instead of dropping entirely
4565 - return 0 from revalidate2 if ll_intent_lock returns -EINTR (912)
4566 - fix leak in bulk IO when only partially completed (899, 900, 926)
4567 - fix O_DIRECT for ia64 (55)
4568 - (almost) eliminate Lustre-kernel-thread effects on load average (722)
4569 - C-z after timeout could hang a process forever; fixed (977)
4571 - client-side I/O cache (678, 924, 929, 941, 970)
4573 - READPAGE and SETATTRs which don't take server-side locks get
4576 2003-02-11 Phil Schwan <phil@clusterfs.com>
4579 - Fix ldlm_lock_match on the MDS to avoid matching remote locks (592)
4580 - Fix fsfilt_extN_readpage() to read a full page of directory
4581 entries, or fake the remainder if PAGE_SIZE != blocksize (500)
4582 - Avoid extra mdc_getattr() in ll_intent_lock when possible (534, 604)
4583 - Fix imbalanced LOV object allocation and out-of-bound access (469)
4584 - Most intent operations were removed, in favour of a new RPC mode
4585 that does a single RPC to the server and bypasses most of the VFS
4586 - All LDLM resource ID arrays were removed in favour of ldlm_res_id
4587 - Aggressively cancel local locks on DLM servers
4588 - mds_reint_unlink sends EA to the client if it's the last nlink.
4589 client uses that EA to unlink OST objects.
4590 - mds_reint_{rename,unlink,link} were rewritten to take ordered locks
4591 - recursive symlinks were fixed (439)
4592 - fixed NULL deref in DEBUG_REQ
4593 - filter_update_lastobjid no longer calls sync, which annoyed extN
4594 - fixed multi-client small-writes to a single file problem (445)
4595 - fixed mtime updates during file writes (607)
4596 - fixed vector writes on obdfilter causing problems when ENOSPC (670)
4597 - fixed bug in obd_brw_read/write() (under guise of testing 367)
4598 - fixed Linux OST size reporting problem (444, 656)
4599 - OST now updates object mtime with writes or setattr (607, 619)
4600 - client verifies file size before zeroing page past EOF (445)
4601 - OST now writes last allocated objid to disk with allocation (108)
4602 - LOV on echo now works (409)
4604 - mds_reint_unlink sends a new buffer, with the EA included. this
4605 buffer is only valid if body->valid & OBD_MD_FLEASIZE, which is only
4606 set if a regular file was being unlinked, and it was the last link
4607 - use PtlGet from the target for bulk writes (315)
4608 - OST now updates object mtime with writes or setattr (607, 619)
4609 - LDLM now has a grant-time callback to revalidate locked items, if
4611 - Many MDS operations were reorganized to combat race conditions
4613 - Merge b_intel branch (updated lprocfs code) - now at /proc/fs/lustre
4614 - configure check to avoid gcc version 2.96 20000731-2.96-98) (606)
4616 2003-01-06 Andreas Dilger <adilger@clusterfs.com>
4619 - Fully reactivate OST imports after reconnection (512, others)
4620 - Make sure client sees our -ENOTCONN from mds_handle (513 - partial)
4621 - More graceful error handling for truncating on dead OST (515)
4622 - Don't error out unless we're actually accessing dead stripes (474)
4623 - Fix garbage sizes when stripes are missing (410)
4624 - LRU counters were broken, causing constant lock purge (433, 432)
4625 - garbage on read from stripes with failed OSTs (441)
4626 - mark OSCs as active before reconnecting during recovery (438)
4627 - lov_enqueue and lov_cancel need to handle inactive OSTs (403)
4628 - lfind did not preserve OST order in output (443)
4629 - symlinks cause hung clients, incorrect data (439)
4630 - stop dereferencing request after dropping refcount (457)
4631 - don't LASSERT(spin_is_locked) on non-SMP (455)
4632 - fixes for many rename() bugs
4633 - fstat didn't correctly synchronize attributes (399)
4634 - server must handle lock cancellation during blocking AST prep (487)
4635 - bulk descriptors were free()d too soon (511)
4636 - fix paths in lconf, which would load incorrect modules (451, 507)
4637 - fix confusing lconf 'host not found' error message (386)
4638 - fix lock order deadlock on OST (O/R i_sem before journal ops, 478)
4639 - fix race condition in mdc_blocking_ast() for inode access (526)
4640 - fix lov_unpackmd() unpacking wrong number of stripes (537)
4641 - fix lov_set_osc_active() marking wrong OSC inactive (440)
4642 - fix bad lstripe lov_unpackmd() assertion (fix layering too) (527)
4643 - fix multiple writes of stripe MD to MDS (358, maybe 519)
4644 - fix lstripe in several ways (kernel side) (527)
4645 - fix request leak in ldlm_cli_enqueue (262)
4646 - incorrect OSC was marked inactive after OST failure
4647 - call mds_fs_cleanup before unmounting filesystem (524)
4648 - fix races between taking ns_lock and ldlm_lock_change_resource
4649 - fix races updating LOV export open file list
4650 - fix lov_enqueue error path, avoid decref-ing bad lock handle
4651 - fix recovery NULL deref in ldlm_cli_cancel_unused
4652 - fix some DLM races by using new hash table for lock handles (419)
4653 - permit the client to specify desired inodes, at replay
4654 - duplicate requests when we queue them for replay reintegration
4655 - fix last_rcvd offset calculation
4656 - sync after each recovered transaction, so we always make progress
4657 - never, not always, ERESTART requests without transnos
4658 - store the lov_desc in the MDS, so we don't depend on getlovinfo to
4660 - skip replay if the MDS says that the client is already connected
4661 - don't check for a recovery-enabled export to match lctl's UUID
4662 - don't INC_USE_COUNT for phantom exports
4663 - don't crash when cleaning up phantom exports (567)
4664 - don't double-finish or set replay data for errored mdc_open requests
4665 - abort requests when they time out, so we don't get old replies
4666 - send/receive replies for AST messages again
4667 - if the client says that it doesn't have the lock, cancel it on the
4669 - if we timeout during I/O, don't try to cancel an in-use lock; instead
4670 mark it as destroyed, it will all work out when decref is called
4671 - fix module use counts (22, 581)
4673 - ASTs now expect a reply (server cancels lock on error reply)
4675 2002-12-02 Andreas Dilger <adilger@clusterfs.com>
4678 - fix many simultaneous client startup (392)
4679 - fix dentry->d_it clobbering
4680 - credentials weren't being shipped for readdir/getattr operations
4681 - remove invalid assertions triggered during some concurrent MD
4683 - proper Lustre versions added (336, 389)
4684 - fix memory leak for create error case (398)
4685 - fix LOV locking bug that would get cli/srv out of sync
4686 - fix echo client over LOV (409)
4687 - fix dbench 2, extN refcount problem (170, 258, 356, 418)
4688 - fix double-O_EXCL intent crash (424)
4689 - avoid sending multiple lock CANCELs (352)
4691 - MDS can do multi-client recovery (modulo bugs in new code)
4693 - many updates, edits, cleanups
4695 2002-11-18 Phil Schwan <phil@clusterfs.com>
4698 - fix null d_it dereference (346)
4699 - fix full OST/dbench hang (333)
4700 - fix permission problem with file removal (286)
4701 - fix removal of OSCs from LOV when they fail
4702 - fix NULL deref during bulk timeout (214)
4703 - fix problems related to multiple filesystems on one MDS (241)
4704 - fixed serious subtle metadata locking bugs
4705 - free locks on clients when inodes are removed due to memory
4707 - fix inode pointer in lock data (285)
4708 - partial support for multiple MDS on a single host (241)
4709 - data locks weren't cancelled at clear_inode time (290, 311)
4710 - intent locks could lead to unbounded lock growth (205)
4711 - added a maximum lock count, an LRU list, and a flusher
4712 - fix multiple rename (365)
4713 - properly abstracted the echo client
4714 - OSC locked 1 byte too many; fixed
4715 - rewrote brw callback code:
4716 - fixed recovery bugs related to LOVs (306)
4717 - fixed too-many-pages-in-one-write crash (191)
4718 - fixed (again) crash in sync_io_timeout (214)
4719 - probably fixed callback-related race (385)
4721 - Add capability to MDS protocol
4722 - LDLM cancellations and callbacks on different portals
4724 2002-10-28 Andreas Dilger <adilger@clusterfs.com>
4727 - limit client IOV size to PTL_MD_MAX_IOV (611336, 191)
4728 - defer open object destruction to close time (601981, 138)
4729 - open/close OST file handle in obdo (OBD_MD_FLHANDLE) (601981, 138)
4730 - move LDLM_ENQUEUE/CONVERT back to MDS portal (625069)
4731 - abstract ll_lookup2, fix ll_revalidate2 to use abstraction (256)
4732 - don't call obd_setattr in ll_file_release for destroyed objects
4733 * protocol change to lustre_msg: move |version| and add |flags|
4734 * protocol change to osc_punch: "start" in "o_size", "end" in "o_blocks"
4735 * lock replay: for LDLM_FL_REPLAY trust client to do right thing
4736 * added replay of create, unlink, link and rename operations during
4737 MDS failover; recovery should be much more robust now
4738 * remove failed OSCs from LOVs (only lov_create uses this so far)
4739 * the lustre-HOWTO was brought (more) up to date (582544)
4741 2002-10-23 Phil Schwan <phil@clusterfs.com>
4744 - in-use dentries weren't being reused properly (617851)
4745 - prevent multiple LDLM setup (599178)
4746 - fix LOV size calculations for truncate (617853)
4747 - fix client handling of MDS intent errors (POSIX)
4748 - fix permission bug in lovstripe.c test (624321)
4749 - fix MDS thread deadlock - move LDLM handler to DLM portal (625069)
4750 - truncate past end of file could corrupt data
4751 - proper cleanup after timeouts, crashes, etc (592524, 550815)
4752 - a race in recovery could return ETIMEDOUT to apps (623947)
4753 - building outside the source directory was fixed
4754 * the lustre-HOWTO was brought (more) up to date (582544)
4755 * major progress was made on recovery functionality
4757 2002-10-10 Phil Schwan <phil@clusterfs.com>
4760 - recovery deadlock fix
4761 - rm -rf causes LBUG fix (617817)
4762 - file open by multiple tasks fix (618962)
4763 - directory permissions bugs (602707 and 620007)
4764 - journal_stop fixed with locking (611313)
4765 - O_APPEND failures resolved (618273, perhaps 614459)
4766 - lconf PATH fix (619770)
4767 - IA64 build fix (621450)
4768 - RPC buffer sizes scale with amount of memory
4770 2002-10-01 Phil Schwan <phil@clusterfs.com>
4773 - locks would be cancelled without throwing away data pages,
4774 resulting in inconsistent data (605627)
4775 - inode attributes were not always being refreshed (605627, 612449)
4776 - lconf now continues to cleanup after lctl reports an error
4777 - MDS now enforces user permissions (602707)
4778 - lprocfs cleanup fixed, but not yet enabled (614157)
4779 - fixed infinite server hang, should a client not respond to an AST
4780 - avoid going into recovery if user calls readlink() with a buffer
4781 that's too small (613941)
4782 - AST RPCs no longer require replies (614867) -- this may be changed
4783 - don't crash server if client sends an IOV that's too big (611336)
4784 - fixed lock conversion deadlock (611892)
4785 - fixed the following of symlinks (614622)
4786 * recovery: the server can remove locks from a client that dies, other
4787 clients can make progress
4788 * more extN patch fixes
4789 * compile-time configurable ptlrpc buffer allocations
4791 - collaborative read cache document
4792 - Lustre Lite Performance CDR document-in-progress
4794 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
4797 - fix typo in patch-2.4.18
4799 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
4802 - clear ptlrpc request each time in handle_incoming_request()
4803 - unlink of files now destroys the object on the OST
4805 2002-09-19 Peter Braam <braam@clusterfs.com>
4807 * add hard link support
4808 * change obdfile creation method
4809 * kernel patch changed
4811 2002-09-19 Peter Braam <braam@clusterfs.com>
4814 - stack overflow bug in extN fixed
4816 2002-09-18 Andreas Dilger <adilger@clusterfs.com>
4818 * documentation updates
4819 - add man pages for config tools
4820 - update tests/README to describe testing with new config tools
4821 - finish metadata API descriptions
4822 * bug fixes and cleanups
4823 - statfs workaround for 16TB limit
4824 - LOV stripe allocation improved, can stripe on subset of OSTs
4825 - LOV file size/IO offset was wrong for files > 4GB in size
4826 - object EA data was being dropped, caused files to be unreadable
4827 - memory overflow with non-LOV OST caused memory corruption
4828 - fixed regression tests to work with new config tools, obdfilter
4829 - fixed bug when directory size became larger than 1 block
4830 - fixed bug (for single client case) when PWD was deleted
4831 - invalidate local directory pages when doing intent-based ops
4832 - avoid LDLM oops when lock callback contained bad data
4834 2002-09-09 Andreas Dilger <adilger@clusterfs.com>
4836 * documentation updates
4837 * bug fixes and cleanups
4838 - configuration tools
4841 - 64-bit compile warnings
4842 - 64-bit internal statfs data
4844 * test_brw on persistent OST devices
4846 * lprocfs (disabled)
4848 2002-09-04 Andreas Dilger <adilger@clusterfs.com>
4850 * documentation updates
4851 * bug fixes and cleanups
4852 * configuration tools
4854 2002-08-30 Peter J. Braam <braam@clusterfs.com>
4857 * many small fixes to 0_5_4
4858 * io/network handling
4859 * thinkos in MDS operations
4861 2002-08-24 Peter J. Braam <braam@clusterfs.com>
4864 * crucial basic fixes to 0.5.3
4865 * IOR, Iozone work over Elan
4868 2002-08-07 Phil Schwan <phil@clusterfs.com>
4869 * version 0_5_3, our first alpha
4870 * we use the new Portals iovs
4871 * documentation updates
4872 * bug fixes and cleanups
4873 * small changes in the DLM wire protocol
4875 2002-07-25 Peter J. Braam <braam@clusterfs.com>
4876 * version 0_5_1 with some initial stability,
4877 * locking on MD and file I/O.
4878 * documentation updates
4879 * several bug fixes since 0.5.0
4880 * small changes in wire protocol
4882 2002-07-18 Phil Schwan <phil@clusterfs.com>
4884 * delivered as Lustre Light Alpha
4885 * fixed a crash after handling invalid MDS requests
4886 * fixed directory pages for architectures with non-4k pages sizes
4888 2002-07-11 Andreas Dilger <adilger@clusterfs.com>
4889 * release version v0_4_4
4890 * Moves TCP acceptor to be on port 2432 (unused Coda port) instead
4892 * Fixes a number of interruption problems with OST operations.
4893 * Update documentation for portals header changes
4894 * Move all wire protocol structs/defines to lustre_idl.h
4895 * Fixes symlink length bug.
4896 * Add tcpdump to repository.
4898 2002-07-05 Andreas Dilger <adilger@clusterfs.com>
4899 * release version v0_4_3
4900 * Fixes statfs for inodes on extN.
4901 * Fixes bug in runtests which would delete /etc/hosts.
4902 * Use 64-bit object IDs wherever possible (not into VFS though)
4903 Remove ost_get_info, which is unused by lustre, and out of date.
4905 2002-07-03 Peter Braam <braam@clusterfs.com>
4906 * release version v0_4_2 Fixes a lookup error (type not passed)
4907 * move forward to head of Portals
4908 * move forward to latest Lustre kernel
4910 2002-06-25 Peter Braam <braam@clusterfs.com>
4911 * release version v0_4_1. Hopefully stable on single node use.