1 tbd Sun Microsystems, Inc.
4 2.6.16.60-0.27 (SLES 10),
5 2.6.18-92.1.10.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: 1.40.11-sun1
11 * Note that reiserfs quotas are disabled on SLES 10 in this kernel.
12 * RHEL 4 and RHEL 5/SLES 10 clients behaves differently on 'cd' to a
13 removed cwd "./" (refer to Bugzilla 14399).
14 * File join has been disabled in this release, refer to Bugzilla 16929.
16 Severity : enhancement
18 Description: Re-disable certain /proc logging
19 Details : Enable and disable client's offset_stats, extents_stats and
20 extents_stats_per_process stats logging on the fly.
23 Frequency : Only on FC kernels 2.6.22+
25 Description: oops in statahead
26 Details : Do not drop reference count for the dentry from VFS when lookup,
27 VFS will do that by itself.
29 Severity : enhancement
31 Description: Generic /proc file permissions
32 Details : Set /Proc file permissions in a more generic way to enable non-
33 root users operate on some /proc files.
37 Description: Hitting mdc_commit_close() ASSERTION
38 Details : Properly handle request reference release in
39 ll_release_openhandle().
43 Frequency : only patchless client
44 Description: add workaround for race between add/remove dentry from hash
46 Severity : enhancement
48 Description: Allow OST glimpses to return PW locks
52 Description: LBUG when llog conf file is full
53 Details : When llog bitmap is full, ENOSPC should be returned for plain
58 Description: Prevent import from entering FULL state when server in recovery
62 Description: service mount cannot take device name with ":"
63 Details : Only when device name contains ":/" will mount treat it as
69 Description: replace ptlrpcd with the statahead thread to interpret the async
70 statahead RPC callback
74 Frequency : on recovery
75 Description: I/O failures after umount during fail back
76 Details : if client reconnected to restarted server we need join to recovery
77 instead of find server handler is changed and process self eviction
78 with cancel all locks.
80 Severity : enhancement
82 Description: Update to RHEL5 kernel-2.6.18-92.1.10.el5.
84 Severity : enhancement
86 Description: Update to SLES10 SP2 kernel-2.6.16.60-0.27.
88 Severity : enhancement
90 Description: Upcall on Lustre log has been dumped
91 Details : Allow for a user mode script to be called once a Lustre log has
92 been dumped. It passes the filename of the dumped log to the
93 script, the location of the script can be specified via
94 /proc/sys/lnet/debug_log_upcall.
99 Description: avoid idr_remove called for id which is not allocated.
100 Details : Move assigment s_dev for clustered nfs to end of initialization,
101 to avoid problem with error handling.
106 Description: avoid Already found the key in hash [CONN_UNUSED_HASH] messages
107 Details : When connection is reused this not moved from CONN_UNUSED_HASH
108 into CONN_USED_HASH and this prodice warning when put connection
109 again in unused hash.
111 Severity : enhancement
113 Description: File striping can now be set to use an arbitrary pool of OSTs.
115 Severity : enhancement
117 Description: Export bytes_read/bytes_write count on OSC/OST.
121 Description: Early reply size mismatch, MGC loses connection
122 Details : Apply the MGS_CONNECT_SUPPORTED mask at reconnect time so
123 the connect flags are properly negotiated.
127 Description: Properly propagate oinfo flags from lov to osc for statfs
128 Details : restore missing copy oi_flags to lov requests.
130 Severity : enhancement
132 Description: Add man pages for llobdstat(8), llstat(8), plot-llstat(8),
133 : l_getgroups(8), lst(8), routerstat(8)
134 Details : included man pages for llobdstat(8), llstat(8),
135 : plot-llstat(8), l_getgroups(8), lst(8), routerstat(8)
137 Severity : enhancement
139 Description: configure's --enable-quota should check the
140 : kernel .config for CONFIG_QUOTA
141 Details : configure is terminated if --enable-quota is passed but
142 : no quota support is in kernel
146 Description: Remove portals compatibility
147 Details : Remove portals compatibility, not interoperable with releases
152 Description: Resolve device initialization race
153 Details : Prevent proc handler from accessing devices added to the
154 obd_devs array but yet be intialized.
156 Severity : enhancement
158 Description: Update to SLES10 SP2 kernel-2.6.16.60-0.23.
160 Severity : enhancement
162 Description: Update to RHEL5 kernel-2.6.18-92.1.6.el5.
167 Description: Using wrong pointer in osc_brw_prep_request
168 Details : Access to array[-1] can produce panic if kernel compiled with
169 CONFIG_PAGE_ALLOC enabled
173 Description: Client runs out of low memory
174 Details : Consider only lowmem when counting initial number of llap pages
178 Description: *optional* service tags registration
179 Details : if the "service tags" package is installed on a Lustre node
180 When the filesystem is mounted, a local-node service tag will
181 be created. See http://inventory.sun.com/ for more information
182 about the Service Tags asset management system.
186 Description: Kernel BUG tries to release flock
187 Details : Lustre does not destroy flock lock before last reference goes
188 away. So always drop flock locks when client is evicted and
189 perform unlock regardless of successfulness of speaking to MDS.
193 Description: add recount protection for osc callbacks, so avoid panic on shutdown
197 Description: sanity test 65a fails if stripecount of -1 is set
198 Details : handle -1 striping on filesystem in ll_dirstripe_verify
203 Description: ASSERTION(CheckWriteback(page,cmd)) failed
204 Details : badly clear PG_Writeback bit in ll_ap_completion can produce false
207 Severity : enhancement
209 Description: Update to RHEL5 kernel-2.6.18-53.1.21.el5.
213 Description: do not process already freed flock
214 Details : flock can possibly be freed by another thread before it reaches
215 to ldlm_flock_completion_ast.
219 Description: LBUG during stress test
220 Details : Need properly lock accesses the flock deadlock detection list.
224 Description: oops in page fault handler
225 Details : kernel page fault handler can return two special 'pages' in error case, don't
226 try dereference NOPAGE_SIGBUS and NOPAGE_OMM.
230 Description: timeout with invalidate import.
231 Details : ptlrpcd_check call obd_zombie_impexp_cull and wait request which should be
232 handled by ptlrpcd. This produce long age waiting and -ETIMEOUT
233 ptlrpc_invalidate_import and as result LASSERT.
235 Severity : enhancement
237 Description: Update to RHEL5 kernel-2.6.18-53.1.19.el5.
241 Description: enable MGS and MDT services start separately
242 Details : add a 'nomgs' option in mount.lustre to enable start a MDT with
243 a co-located MGS without starting the MGS, which is a complement
244 to 'nosvc' mount option.
248 Frequency : after recovery
249 Description: precreate to many object's after del orphan.
250 Details : del orphan st in oscc last_id == next_id and this triger growing
251 count of precreated objects. Set flag LOW to skip increase count
252 of precreated objects.
256 Frequency : rare, on clear nid stats
257 Description: ASSERTION(client_stat->nid_exp_ref_count == 0)
258 Details : when clean nid stats sometimes try destroy live entry,
259 and this produce panic in free.
263 Description: Stack overflow during MDS log replay
264 ease stack pressure by using a thread dealing llog_process.
268 Description: wait until IO finished before start new when do lock cancel.
269 Details : VM protocol want old IO finished before start new, in this case
270 need wait until PG_writeback is cleared until check dirty flag and
271 call writepages in lock cancel callback.
273 Severity : enhancement
275 Description: using special macro for print time and cleanup in includes.
279 Description: mds_mfd_close() ASSERTION(rc == 0)
280 Details : In mds_mfd_close(), we need protect inode's writecount change
281 within its orphan write semaphore to prevent possible races.
285 Description: Obsolete CURRENT_SECONDS and use cfs_time_current_sec() instead.
289 Frequency : rare, on shutdown ost
290 Description: don't hit live lock with umount ost.
291 Details : shrink_dcache_parent can be in long loop with destroy dentries,
292 use shrink_dcache_sb instead.
296 Description: don't panic with use echo client
297 Details : echo client pass NULL as client nid pointer and this produce null
302 Description: fix build on ppc32
303 Details : compile code with -m64 flag produce wrong object file for ppc32.
307 Description: add message levels for liblustreapi
311 Description: fix for occasional failure case of -ENOSPC in recovery-small tests
312 Details : Move the 'good_osts' check before the 'total_bavail' check. This
313 will result in an -EAGAIN and in the exit call path we call
314 alloc_rr() which will with increasing aggressiveness attempt to
315 aquire precreated objects on the minimum number of required OSCs.
319 Description: Use old size assignment to avoid deadlock
320 Details : This reverts the changes in bugs 2369 and bug 14138 that introduced
321 the scheduling while holding a spinlock. We do not need locking
322 for size in ll_update_inode() because size is only updated from
323 the MDS for directories or files without objects, so there is no
324 other place to do the update, and concurrent access to such inodes
325 are protected by the inode lock.
329 Description: resolve "_IOWR redefined" build error on SLES10
333 Description: dump the memory debugging after all modules are unloaded to
334 suppress false negative in conf_sanity test 39
336 Severity : enhancement
338 Description: build kernel-ib packages for OFED 1.3 in our release cycle
343 Description: fix SLES kernel versioning
344 Details : the kernel version for our SLES 10 kernel did not include a "-"
345 before the "smp" at the end. while this was not a problem in
346 general, it did mean that software trying to use the kernel
347 version to try to detect a vendor specific kernel would fail.
348 this was most evident by the OFED build scripts.
352 Description: Don't update lov_desc members until making sure they are valid
353 Details : When updating lov_desc members via proc fs, need fix their
354 validities before doing the real update.
358 Description: don't put request into delay list while invalidate in flight.
359 Details : ptlrpc_delay_request sometimes put in delay list while invalidate
360 import in flight. this produce timeout for invalidate and sometimes
361 can cause stale data.
365 Frequency : on ppc only
366 Description: not convert ost objects for directory because it's not exist.
367 Details : ll_dir_getstripe assume dirrectory has ost objects but this wrong.
371 Description: Add FMODE_EXEC file flag for SLES10 SP1 kernel.
373 Severity : enhancement
375 Description: Update to support 2.6.22.14 vanilla kernel.
379 Frequency : rare, on recovery
380 Description: read procfs can produce deadlock in some situation
381 Details : Holding lprocfs lock which send rpc can produce block for destroy
382 obd objects and this also block reconnect with -EALREADY. This isn't
383 fix all lprocfs bugs - but make it rare.
385 Severity : enhancement
387 Description: Update kernel to RHEL5 2.6.18-53.1.14.el5.
390 Frequency : frequent on X2 node
392 Description: mdc_set_open_replay_data LBUG
393 Details : Set replay data for requests that are eligible for replay.
397 Description: lustre_mgs: operation 101 on unconnected MGS
398 Details : When MGC is disconnected from MGS long enough, MGS will evict the
399 MGC, and late on MGC cannot successfully connect to MGS and a lot
400 of the error messages complaining that MGS is not connected.
404 Frequency : on network error
405 Description: panic with double free request if network error
406 Details : mdc_finish_enqueue is finish request if any network error ocuring,
407 but it's true only for synchronus enqueue, for async enqueue
408 (via ptlrpcd) this incorrect and ptlrpcd want finish request
411 Severity : enhancement
413 Description: client-side metadata stat-ahead during readdir(directory readahead)
414 Details : perform client-side metadata stat-ahead when the client detects
415 readdir and sequential stat of dir entries therein
418 Frequency : on start mds
420 Description: Implement get_info(last_id) in obdfilter.
423 Frequency : occasional
425 Description: Correctly check stale fid, not start epoch if ost not support SOM
426 Details : open with flag O_CREATE need set old fid in op_fid3 because op_fid2
427 overwrited with new generated fid, but mds can anwer with one of these
428 two fids and both is not stale. setattr incorectly start epoch and
429 assume will be called done_writeting, but without SOM done_writing
433 Frequency : rare, depends on device drivers and load
435 Description: MDS or OSS nodes crash due to stack overflow
436 Details : Code changes in 1.8.0 increased the stack usage of some functions.
437 In some cases, in conjunction with device drivers that use a lot
438 of stack the MDS (or possibly OSS) service threads could overflow
439 the stack. One change which was identified to consume additional
440 stack has been reworked to avoid the extra stack usage.
443 Frequency : occasional
445 Description: Do not fail import if osc_interpret_create gets -EAGAIN
446 Details : If osc_interpret_create got -EAGAIN it immediately exits and
447 wakeup oscc_waitq. After wakeup oscc_wait_for_objects call
448 oscc_has_objects and see OSC has no objests and call
449 oscc_internal_create to resend create request.
451 Severity : enhancement
453 Description: Update to SLES10 SP1 latest kernel-2.6.16.54-0.2.5.
455 Severity : enhancement
457 Description: Update to RHEL5 latest kernel-2.6.18-53.1.13.el5.
460 Frequency : very rare
462 Description: Fix replay if there is an un-replied request and open
463 Details : In some cases, older replay request will revert the
464 mcd->mcd_last_xid on MDS which is used to record the client's
467 Severity : enhancement
469 Description: Update to RHEL5 latest kernel-2.6.18-53.1.6.el5.
471 Severity : enhancement
473 Description: Add rhel5 support to HEAD.
475 Serverity : enhancement
477 Description: Update RHEL4 kernel to 2.6.9-67.0.4.
482 Description: Don't allow skipping OSTs if index has been specified.
483 Details : Don't allow skipping OSTs if index has been specified, make locking
484 in internal create lots better.
488 Description: LBUG in ptlrpc_check_set() bad phase ebc0de00
489 Details : access to bitfield in structure is always rounded to long
490 and this produce problem with not atomic change any bit.
494 Description: Lustre make rpms failed.
495 Details : Remove ldiskfs spec file to avoids rpmbuild be confused when
496 builds Lustre rpms from tarball.
499 Frequency : rare on shutdown ost
501 Description: If llog cancel was not send before clean_exports phase, this can
502 produce deadlock in llog code.
503 Details : If llog thread has last reference to obd and call class_import_put
504 this produce deadlock because llog_cleanup_commit_master wait when
505 last llog_commit_thread exited, but this never success because was
506 called from llog_commit_thread.
510 Description: allow userland application know is lost one of stripes.
511 Details : fill lvb_blocks with error code on ost and return it to
512 application if error flag found.
516 Description: NULL lov_tgts causing MDS oops
517 Details : more safe checks for NULL lov_tgts for avoid oops.
519 Severity : enhancement
521 Description: Update to RHEL4 latest kernel-2.6.9-67.0.1.EL.
525 Descriptoin: make lov_create() will not stuck in obd_statfs_rqset()
526 Details : If an OST is down the MDS will hang indefinitely in
527 obd_statfs_rqset() waiting for the statfs data. While for
528 MDS QOS usage of statfs, it should not stuck in waiting.
530 Severity : enhancement
532 Description: remote_acl support
533 Details : Support ACL-based permission check for remote user.
534 Support setfacl/getfacl for remote user with the utils
535 "lfs {l,r}{s,g}etfacl" which follow the same parameter format as
536 the system "{s,g}etfacl" utils.
538 Severity : enhancement
540 Description: Update to RHEL4 U6 kernel-2.6.9-67.EL.
542 Severity : enhancement
544 Description: Update to RHEL5 latest kernel-2.6.18-53.1.4.el5.
548 Description: make mgs_setparam() handle fsname containing dash
549 Details : fsname containing a dash does not work with lctl conf_param
551 Severity : enhancement
553 Description: Update to SLES10 SP1 latest kernel-2.6.16.54-0.2.3.
555 Severity : enhancement
557 Description: Update to RHEL5 Update-1 kernel 2.6.18-53.el5.
561 Frequency : rare, at shutdown
562 Description: access already free / zero obd_namespace.
563 Details : if client_disconnect_export was called without force flag set,
564 and exist connect request in flight, this can produce access to
565 NULL pointer (or already free pointer) when connect_interpret
566 store ocd flags in obd_namespace.
570 Frequency : only at startup
571 Description: not alloc memory with spinlock held.
572 Details : allocation memory with GFP_KERNEL can produce sleep deadlock,
573 if any spinlock held.
575 Severity : enhancement
577 Description: make lustre randomly fail allocating memory
578 Details : Make lustre randomly failed allocating memory for testing purpose.
580 Severity : enhancement
582 Description: lost problems with lov objid file
583 Details : Fixes some scability and access to not inited memory problems
584 in work with lov objdid file.
589 Description: lfs find does not continue on file error
590 Details : Continue other files processing when a file/dir is absent.
594 Description: Inconsistent usage of lustre_pack_reply()
595 Details : Standardize the usage of lustre_pack_reply() such that it
596 always generate a CERROR on failure.
599 Frequency : occasional
601 Description: MDS hang or stay in waiting lock
602 Details : If client receive lock with CBPENDING flag ldlm need send lock
603 cancel as separate rpc, to avoid situation when cancel request
604 can't processed due all i/o threads stay in wait lock.
608 Description: Update to RHEL5 kernel 2.6.18-8.1.15.el5.
612 Description: Update to SLES10 SP1 kernel 2.6.16.53-0.16
616 Description: Update to SLES9 kernel-2.6.5-7.287.3.
620 Description: Update to RHEL4 latest kernel.
622 Severity : enhancement
624 Description: Build SLES10 patchless client fails
625 Details : The configure was broken by run ./configure with
626 --with-linux-obj=.... argument for patchless client. When the
627 configure use --with-linux-obj, the LINUXINCLUDE= -Iinclude
628 can't search header adequately. Use absolute path such as
629 -I($LINUX)/include instead.
633 Description: interrupt oig_wait produce painc on resend.
634 Details : brw_redo_request can be used for resend requests from ptlrpcd and
635 private set, and this produce situation when rq_ptlrpcd_data not
636 copyed to new allocated request and triggered LBUG on assert
637 req->rq_ptlrpcd_data != NULL. But this member used only for wakeup
638 ptlrpcd set if request is changed and can be safety changed to use
643 Description: LASSERT_{REQ,REP}SWAB macros are buggy
644 Details : If SWAB_PARANOIA is disabled, the LASSERT_REQSWAB and
645 LASSERT_REPSWAB macros become no-ops, which is incorrect. Drop
646 these macros and replace them with their difinitions instead.
650 Description: Update kernel patches for SLES10 2.6.16.53-0.8.
651 Details : Update which_patch & target file for SLES10 latest kernel.
654 Description: Remove client patches from SLES 10 kernel.
655 Details : This causes SLES 10 clients to behave as patchless clients
656 even on a Lustre-patched (server) kernel.
658 Severity : enhancement
660 Description: self-adjustable client's lru lists
661 Details : use adaptive algorithm for managing client cached locks lru
662 lists according to current server load, other client's work
663 pattern, memory activities, etc. Both, server and client
664 side namespaces provide number of proc tunables for controlling
667 Severity : enhancement
669 Description: light-weight GSS support
670 Details : Support krb5n and krb5a mode, which keep Kerberos 5 authentication
671 and reduce performance overhead.
673 Severity : enhancement
675 Description: Linux keyring support
676 Details : Support using service of Linux keyring for Lustre GSS internal
677 context refresh/cache mechanism.
681 Description: Fix errors in lfs documentation
682 Details : Fixes man pages
686 Description: don't use GFP_* in generic Lustre code.
687 Details : Use cfs_alloc_* functions and CFS_* flags for code portability.
691 Description: obdclass is limited by single OBD_ALLOC(idarray)
692 Details : replace OBD_ALLOC/OBD_FREE with OBD_VMALLOC/OBD_VFREE
696 Description: warnings with build patchless client with vanila 2.6.19 and up
697 Details : change old ctl_table style and replace ctl_table/ctl_table_header
698 with cfs_sysctl_table_t/cfs_sysctl_table_header_t
702 Frequency : Only for SLES
703 Description: sanity_quota fail test_1
704 Details : There are multiple occurences of $TSTUSR in SLES's /etc/group
705 file, which makes TSTID[2] inunique.
709 Frequency : Only for SLES9
710 Description: Kernel patches for SLES9 2.6.5-7.286 kernel
711 Details : Update target/ChangeLog/which_patch .
715 Frequency : Only for test_10 in sanity-quota.sh
716 Description: a bug in quota test code
717 Details : checking a test flag in a wrong place
721 Frequency : Only for test_13 in sanity-quota.sh
722 Description: a bug in quota test script
723 Details : shouldn't check overfull allocated quota
727 Description: jbd statistics
728 Details : Port older jbd statistics patch for sles10
730 Severity : enhancement
732 Description: Add printing inode info into message about error in writepage.
737 Description: reply_lock_interpret crash due to race with it and lock cancel.
738 Details : Do not replay locks that are being cancelled. Do not reference
739 locks by their address during replay, just by their handle.
743 Frequency : When flocks are used.
744 Description: assertion failure in ldlm_cli_enquque_fini for non NULL lock.
745 Details : Flock locks might destroy just granted lock if it could be merged
746 with another existing flock, this is done in completion handler,
747 so teach ldlm_cli_enquque_fini that this is a valid case for
753 Description: Oops in read and write path when failing to allocate lock.
754 Details : Check if lock allocation failed and return error back.
758 Description: lstripe command fails for valid OST index
759 Details : The stripe offset is compared to lov->desc.ld_tgt_count
760 instead of lov->desc.ld_active_tgt_count.
764 Description: sanity.sh failed test 103
765 Details : RHEL mis-interpret setfacl "-X" param, so we won't test setfacl
770 Description: df doesn't work properly if diskfs blocksize != 4K
771 Details : Choose biggest blocksize of OST's as the LOV's blocksize.
775 Description: lfs find on -1 stripe looping in lsm_lmm_verify_common()
776 Details : Avoid lov_verify_lmm_common() on directory with -1 stripe count.
778 Severity : enhancement
780 Description: Adaptive timeouts
781 Details : RPC timeouts adapt to changing server load and network
782 conditions to reduce resend attempts and improve recovery time.
786 Description: llapi_file_create() does not allow some changes
787 Details : add llapi_file_open() that allows specifying the mode and
788 open flags, and also returns an open file handle.
792 Description: merge and cleanup kernel patches.
793 Details : Remove mnt_lustre_list in vfs_intent-2.6-rhel4.patch.
797 Description: Add journal checksum support.(Kernel part)
798 Details : The journal checksum feature adds two new flags i.e
799 JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT and
800 JBD2_FEATURE_COMPAT_CHECKSUM. JBD2_FEATURE_CHECKSUM flag
801 indicates that the commit block contains the checksum for
802 the blocks described by the descriptor blocks. Now commit
803 record can be sent to disk without waiting for descriptor
804 blocks to be written to disk. This behavior is controlled
805 using JBD2_FEATURE_ASYNC_COMMIT flag.
809 Description: OSS needs mutliple precreate threads
810 Details : Add ability to start more than one create thread per OSS.
814 Description: Sanity.sh test_65k failed: setstripe should have succeeded
815 Details : execute lfs setstripe on client
819 Description: mds_obd_create error creating tmp object
820 Details : When the user sets quota on root, llog will be affected and can't
821 create files and write files.
824 Frequency : Always on ia64 patchless client, and possibly others.
826 Description: Add EXPORT_SYMBOL check for node_to_cpumask symbol.
827 Details : This allows the patchless client to be loaded on architectures
832 Description: RedHat Update kernel for RHEL5
833 Details : Add the kernel config file for RHEL5.
837 Description: RedHat Update kernel for RHEL5
838 Details : Modify the kernel config file more closer RHEL5.
842 Description: Build failure against Centos5 (RHEL5)
843 Details : Define PAGE_SIZE when it isn't present.
847 Description: test_7 on sanity-quota.sh can't be used on separate machines
848 Details : add facet to handle it
852 Description: "ll_intent_file_open()) lock enqueue: err: -13" with nfs
853 Details : with NFS, the anon dentry's parent was set to itself in
854 d_alloc_anon(), so in MDS, we use rec->ur_fid1 to find the
855 corresponding dentry other than use rec->ur_name.
857 Severity : enhancement
859 Description: lfs setstripe enhancement
860 Details : Make lfs setstripe understand 'k', 'm' and 'g' for stripe size.
864 Description: enable data checksumming by default
865 Details : enable checksum by default, allow --disable-checksum
866 configure option and "-o nochecksum" mount option. Checksums
867 can also be disabled at runtime via $LPROC/osc/*/checksum_pages.
871 Description: lustre support for RHEL5
872 Details : Add support for RHEL5.
876 Description: Client eviction due to failover config
877 Details : after a connection loss, the lustre client should attempt to
878 reconnect to the last active server first before trying the
879 other potential connections.
883 Description: when mds and osts use different quota unit(32bit and 64bit),
884 quota will be released repeatly.
885 Details : void sending multiple quota reqs to mds, which will keep the
886 status between the reqs.
890 Description: osts not allocated evenly to files
891 Details : change the condition to increase offset_idx
894 Frequency : Always for filesystems larger than 2TB on 32-bit systems.
896 Description: Data corruption for OSTs that are formatted larger than 2TB
898 Details : When generating the bio request for lustre file writes the
899 sector number would overflow a temporary variable before being
900 used for the IO. The data reads correctly from Lustre (which
901 will overflow in a similar manner) but other file data or
902 filesystem metadata may be corrupted in some cases.
906 Description: Tune the kernel for good SCSI performance.
907 Details : Set the value of /sys/block/{dev}/queue/max_sectors_kb
908 to the value of /sys/block/{dev}/queue/max_hw_sectors_kb
913 Description: rewrite ext2-derived code in obdclass/uuid.c
914 Details : rewrite inherited code (uuid parsing code from ext2 utils)
915 from scratch preserving functionality.
919 Description: "lfs find -obd UUID" prints directories
920 Details : "lfs find -obd UUID" will return all directory names instead
921 of just file names. It is incorrect because the directories
922 do not reside on the OSTs.
926 Description: use i_size_read and i_size_write in 2.6 port
927 Details : replace inode->i_size access with i_size_read/write()
930 Frequency : when removing large files
932 Description: scheduling issue during removal of large Lustre files
933 Details : Don't take the BKL in fsfilt_ext3_setattr() for 2.6 kernels.
934 It causes scheduling issues when removing large files (17TB in the
938 Frequency : only with liblustre clients on XT3
940 Description: evictions taking too long
941 Details : allow llrd to evict clients directly on OSTs
944 Frequency : only on ppc
946 Description: /proc/fs/lustre/devices broken on ppc
947 Details : The patch as applied to 1.6.2 doesn't look correct for all arches.
948 We should make sure the type of 'index' is loff_t and then cast
949 explicitly as needed below. Do not assign an explicitly cast
953 Frequency : only for rhel5
955 Description: Kernel patches update for RHEL5 2.6.18-8.1.10.el5.
956 Details : Modify the target file & which_kernel.
958 Severity : enhancement
960 Description: omit set fsid for export NFS
961 Details : fix set/restore device id for avoid EMFILE error and mark lustre fs
962 as FS_REQUIRES_DEV for avoid problems with generate fsid.
966 Frequency : Always, for kernels after 2.6.16
967 Description: Fix warning idr_remove called for id=.. which is not allocated.
968 Details : Last kernels save old s_dev before kill super and not allow
969 to restore from callback - restore it before call kill_anon_super.
973 Description: buffer overruns could theoretically occur
974 Details : llapi_semantic_traverse() modifies the "path" argument by
975 appending values to the end of the origin string, and a
976 overrun may occur. Adding buffer overrun check in liblustreapi.
980 Description: Fix error on 'ls .' at the top of the Lustre mount.
981 Details : Don't revalidate dentry if it is a root dentry.
985 Description: Kernel patches update for RHEL4 2.6.9-55.0.6.
986 Details : Modify vm-tunables-rhel4.patch.
990 Description: Kernel config for 2.6.18-vanilla.
991 Details : Modify targets/2.6-vanilla.target.in.
992 Add config file kernel-2.6.18-2.6-vanilla-i686.config.
993 Add config file kernel-2.6.18-2.6-vanilla-i686-smp.config.
994 Add config file kernel-2.6.18-2.6-vanilla-x86_64.config.
995 Add config file kernel-2.6.18-2.6-vanilla-x86_64-smp.config.
999 Description: improve handling recoverable errors
1000 Details : if request processig with error which can be recoverable on server
1001 request should be resend, otherwise page released from cache and
1006 Description: Kernel patches update for RHEL5 2.6.18-8.1.14.el5.
1007 Details : Modify target file & which_patch.
1008 A flaw was found in the IA32 system call emulation provided
1009 on AMD64 and Intel 64 platforms. An improperly validated 64-bit
1010 value could be stored in the %RAX register, which could trigger an
1011 out-of-bounds system call table access. An untrusted local user
1012 could exploit this flaw to run code in the kernel
1013 (ie a root privilege escalation). (CVE-2007-4573).
1017 Description: O_DIRECT bypasses client statistics.
1018 Details : When running with O_DIRECT I/O, neither the client rpc_stats nor
1019 read_ahead_stats were updated. copied the stats section from
1020 osc_send_oap_rpc() into async_internal().
1024 Description: Add jbd statistics patch for RHEL5 and 2.6.18-vanilla
1028 Description: change order of libsysio includes
1029 Details : '#include sysio.h' should always come before '#include xtio.h'
1033 Description: handle "serious error: objid * already exists" more gracefully
1034 Details : If LAST_ID value on disk is smaller than the objects existing in
1035 the O/0/d* directories, it indicates disk corruption and causes an
1036 LBUG(). If the object is 0-length, then we should use the existing
1037 object. This will help to avoid a full fsck in most cases.
1039 Severity : enhancement
1041 Description: adapt the lustre_config script to support the upgrade case
1042 Details : Add "-u" option for lustre_config script to support upgrading 1.4
1043 server targets to 1.6 in parallel.
1047 Description: To avoid grant space > avaible space when the disk is almost
1048 full. Without this patch you might see the error "grant XXXX >
1049 available" or some LBUG about grant, when the disk is almost
1051 Details : In filter_check_grant, for non_grant cache write, we should
1052 check the left space by if (*left > ungranted + bytes), instead
1053 of (*left > ungranted), because only we are sure the left space
1054 is enough for another "bytes", then the ungrant space should be
1055 increase. In client, we should update cl_avail_grant only there
1056 is OBD_MD_FLGRANT in the reply.
1060 Description: Update RHEL 4 kernel to fix local root privilege escalation.
1061 Details : Update to the latest RHEL 4 kernel to fix the vulnerability
1062 described in CVE-2007-4573. This problem could allow untrusted
1063 local users to gain root access.
1066 Frequency : when using O_DIRECT and quotas
1068 Description: Incorrect file ownership on O_DIRECT output files
1069 Details : block usage reported by 'lfs quota' does not take into account
1070 files that have been written with O_DIRECT.
1075 Description: touch file failed when fs is not full
1076 Details : OST in recovery should not be discarded by MDS in alloc_qos(),
1077 otherwise we can get ENOSP while fs is not full.
1081 Description: parallel lock callbacks
1082 Details : Instead of sending blocking and completion callbacks as separated
1083 requests, adding them to a set and sending in parallel.
1086 Frequency : only for Cray XT3
1087 Bugzilla : 12829/13455
1088 Description: Changing primary group doesn't change the group lustre assigns to
1090 Details : When CRAY_XT3 is defined, the fsgid supplied by the client is
1091 overridden with the primary group provided by the group upcall,
1092 whereas the supplied fsgid can be trusted if it is in the list of
1093 supplementary groups returned by the group upcall.
1095 Severity : enhancement
1097 Description: Allow masking D_WARNING, D_ERROR messages from console
1098 Details : Console messages can now be disabled via lnet.printk.
1102 Description: User code with malformed file open parameter crashes client node
1103 Details : Before packing join_file req, all the related reference should be
1104 checked carefully in case some malformed flags cause fake
1105 join_file req on client.
1109 Description: LDLM_ENQUEUE races with LDLM_CP_CALLBACK
1110 Details : ldlm_completion_ast() assumes that a lock is granted when the req
1111 mode is equal to the granted mode. However, it should also check
1112 that LDLM_FL_CP_REQD is not set.
1116 Description: Heavy nfs access might result in deadlocks
1117 Details : After ELC code landed, it is now improper to enqueue any mds
1118 locks under och_sem, because enqueue might want to decide to
1119 cancel open locks for same inode we are holding och_sem for.
1123 Description: Client eviction while running blogbench
1124 Details : A lot of unlink operations with concurrent I/O can lead to a
1125 deadlock causing evictions. To address the problem, the number of
1126 oustanding OST_DESTROY requests is now throttled to
1127 max_rpcs_in_flight per OSC and LDLM_FL_DISCARD_DATA blocking
1128 callbacks are processed in priority.
1132 Description: enable ACLs on MDS by default
1133 Details : ACLs must be enabled on MDS by default.
1136 Frequency : PPC/PPC64 only
1138 Description: conflicts between asm-ppc64/types.h and lustre_types.h
1139 Details : fix duplicated definitions between asm-ppc64/types.h and
1140 lustre_types.h on PPC.
1143 Frequency : PPC/PPC64 only
1145 Description: asm-ppc/segment.h does not exist
1146 Details : fix compile issue on PPC.
1150 Description: better handle error messages in extents code
1153 Frequency : RHEL4 only
1155 Description: mkfs is very slow on IA64/RHEL4
1156 Details : A performance regression has been discovered in the MPT Fusion
1157 driver between versions 3.02.73rh and 3.02.99.00rh. As a
1158 consequence, we have downgraded the MPT Fusion driver in the RHEL4
1159 kernel from 3.02.99.00 to 3.02.73 until this problem is fixed.
1161 Severity : enhancement
1163 Description: SNMP support enhancement
1164 Details : Adding total number of sampled request for an MDS node in snmp
1167 Severity : enhancement
1169 Description: Optimize ldlm waiting list processing for PR extent locks
1170 Details : When processing waiting list for read extent lock and meeting read
1171 lock that is same or wider to it that is not contended, skip
1172 processing rest of the list and immediatelly return current
1173 status of conflictness, since we are guaranteed there are no
1174 conflicting locks in the rest of the list.
1178 Description: Time out and refuse to reconnect
1179 Details : When the failover node is the primary node, it is possible
1180 to have two identical connections in imp_conn_list. We must
1181 compare not conn's pointers but NIDs, otherwise we can defeat
1182 connection throttling.
1186 Description: port llog fixes from b1_6 into HEAD
1187 Details : Port llog reference couting and some llog cleanups from b1_6
1188 (bug 10800) into HEAD, for protect from panic and access to already
1189 free llog structures.
1193 Description: Detect stride IO mode in read-ahead
1194 Details : When a client does stride read, read-ahead should detect that and
1195 read-ahead pages according to the detected stride pattern.
1199 Description: data checksumming impacts single node performance
1200 Details : add support for several checksum algorithm. Currently, only CRC32
1201 and Adler-32 are supported. The checksum type can be changed on
1202 the fly via /proc/fs/lustre/osc/*/checksum_type.
1206 Description: use adler32 for page checksums
1207 Details : when available, use the Adler-32 algorithm instead of CRC32 for
1212 Description: build for x2 fails
1213 Details : fix compile issue on Cray systems.
1217 Description: Properly match for duplicate locks
1218 Details : Due to different lock order from skiplists code, we need to
1219 traverse entire list for now
1222 Frequency : only on PPC/SLES10
1224 Description: "BITS_PER_LONG is not 32 or 64" in linux/idr.h
1225 Details : On SLES10/PPC, fs.h includes idr.h which requires BITS_PER_LONG to
1226 be defined. Add a hack in mkfs_lustre.c to work around this compile
1231 Description: LASSERT on MDS when client holding flock lock dies
1232 Details : ldlm pool logic depends on number of granted locks equal to
1233 number of released locks which is not true for flock locks, so
1234 just exclude such locks from consideration.
1238 Description: MDS deadlock with many ll_sync_lov threads and I/O stalled
1239 Details : Use fsfilt_sync() for both the whole filesystem sync and
1240 individual file sync to eliminate dangerous inode locking
1241 with I_LOCK that can lead to a deadlock.
1245 Description: performance in 1.6.3
1246 Details : Force q->max_phys_segments to MAX_PHYS_SEGMENTS on SLES10 to be
1247 sure that 1MB requests are not fragmented by the block layer.
1249 Severity : enhancement
1251 Description: organize the server-side client stats on per-nid basis
1252 Details : Change the structure of stats under obdfilter and mds to
1260 The "uuid"s file would list the uuids of _active_ exports.
1261 And the clear entry is to clear all stats and stale nids.
1263 Severity : enhancement
1265 Description: eliminate client locks in face of contention
1266 Details : file contention detection and lockless i/o implementation
1267 for contended files.
1271 Description: Reinitialize optind to 0 so that interactive lfs works in all cases
1274 Frequency : very rare, if additional xattrs are used on kernels >= 2.6.12
1276 Description: MDS may lose file striping (and hence file data) in some cases
1277 Details : If there are additional extended attributes stored on the MDS,
1278 in particular ACLs, SELinux, or user attributes (if user_xattr
1279 is specified for the client mount options) then there is a risk
1280 of attribute loss. Additionally, the Lustre file striping
1281 needs to be larger than default (e.g. striped over all OSTs),
1282 and an additional attribute must be stored initially in the
1283 inode and then increase in size enough to be moved to the
1284 external attribute block (e.g. ACL growing in size) for file
1289 Description: skiplist implementation simplification
1290 Details : skiplists are used to group compatible locks on granted list
1291 that was implemented as tracking first and last lock of each
1292 lock group the patch changes that to using doubly linked lists
1296 Description: MDS LBUG: ASSERTION(!IS_ERR(dchild))
1297 Details : Change LASSERTs to client eviction (i.e. abort client's recovery)
1298 because LASSERT on both the data supplied by a client, and the
1299 data on disk is dangerous and incorrect.
1301 Severity : enhancement
1303 Description: Slow truncate/writes to huge files at high offsets.
1304 Details : Directly associate cached pages to lock that protect those pages,
1305 this allows us to quickly find what pages to write and remove
1306 once lock callback is received.
1310 Description: more ldlm soft lockups
1311 Details : In ldlm_resource_add_lock(), call to ldlm_resource_dump()
1312 starve other threads from the resource lock for a long time in
1313 case of long waiting queue, so change the debug level from
1314 D_OTHER to the less frequently used D_INFO.
1316 Severity : enhancement
1318 Description: add -gid, -group, -uid, -user options to lfs find
1322 Description: Hung threads in invalidate_inode_pages2_range
1323 Details : The direct IO path doesn't call check_rpcs to submit a new RPC once
1324 one is completed. As a result, some RPCs are stuck in the queue
1329 Description: filter threads hungs on waiting journal commit
1330 Details : Cleanup filter group llog code, then only filter group llog will
1331 be only created in the MDS/OST syncing process.
1335 Description: Procfs and llog threads access destoryed import sometimes.
1336 Details : Sync the import destoryed process with procfs and llog threads by
1337 the import refcount and semaphore.
1339 Severity : enhancement
1341 Description: openlock cache of b1_6 port to HEAD
1346 Description: kernel BUG at ldiskfs2_ext_new_extent_cb
1347 Details : If insertion of an extent fails, then discard the inode
1348 preallocation and free data blocks else it can lead to duplicate
1353 Description: don't always update ctime in ext3_xattr_set_handle()
1354 Details : Current xattr code updates the inode ctime in ext3_xattr_set_handle.
1355 In some cases the ctime should not be updated, for example for
1356 2.0->1.8 compatibility it is necessary to delete an xattr and it
1357 should not update the ctime.
1361 Bugzilla : 15713/16362
1362 Description: Assertion in iopen_connect_dentry in 1.6.3
1363 Details : looking up an inode via iopen with the wrong generation number can
1364 populate the dcache with a disconneced dentry while the inode
1365 number is in the process of being reallocated. This causes an
1366 assertion failure in iopen since the inode's dentry list contains
1367 both a connected and disconnected dentry.
1371 Description: assertion failure in ldlm_handle2lock()
1372 Details : fix a race between class_handle_unhash() and class_handle2object()
1373 introduced in lustre 1.6.5 by bug 13622.
1378 Description: Kernel BUG: sd_iostats_bump: unexpected disk index
1379 Details : remove the limit of 256 scsi disks in the sd_iostat patch
1384 Description: oops in sd_iostats_seq_show()
1385 Details : unloading/reloading the scsi low level driver triggers a kernel
1386 bug when trying to access the sd iostat file.
1391 Description: Kernel panics during QLogic driver reload
1392 Details : REQ_BLOCK_PC requests are not handled properly in the sd iostat
1393 patch, causing memory corruption.
1398 Description: journal_dev option does not work in b1_6
1399 Details : pass mount option during pre-mount.
1401 Severity : enhancement
1403 Description: Add a FIEMAP(FIle Extent MAP) ioctl
1404 Details : FIEMAP ioctl will allow an application to efficiently fetch the
1405 extent information of a file. It can be used to map logical blocks
1406 in a file to physical blocks in the block device.
1410 Description: LDLM soft lockups - improvement
1411 Details : It is be possible to send the lock handle along with each read
1412 or write request because the client is already doing a lock match
1413 itself so there isn't any reason the OST should have to re-do that
1417 Frequency : only on Cray X2
1419 Description: X2 build failures
1420 Details : fix build failures on Cray X2.
1424 Description: xid & resent requests
1425 Details : Initialize RPC XID from clock at startup (randomly if clock is
1428 Severity : enhancement
1430 Description: Add lustre_start utility to start or stop multiple Lustre servers
1435 Description: Lustre GPF in {:ptlrpc:ptlrpc_server_free_request+373}
1436 Details : In case of memory pressure, list_del() can be called twice on
1437 req->rq_history_list, causing a kernel oops.
1441 Description: (ptllnd_peer.c:557:kptllnd_peer_check_sends()) ASSERTION(!in_interrupt()) failed
1442 Details : fix stack overflow in the distributed lock manager by defering export
1443 eviction after a failed ast to the elt thread instead of handling
1444 it in the dlm interpret routine.
1448 Description: Convert some comments to new format.
1449 Details : Update documenting comments to match doxygen conventions.
1453 Description: Grammar fixes.
1454 Details : A couple of trivial spelling fixes.
1458 Description: OSD_COUNTERS-mandatory
1459 Details : Make previously optional ->oti_{w,r}_locks sanity checks mandatory
1460 to simplify the code.
1464 Description: simplify cmm_device freeing logic.
1465 Details : Call cmm_device_free() in the failure path of cmm_device_alloc().
1469 Description: Add lockdep support to dt_object_operations locking interface.
1470 Details : Augment ->do_{read,write}_lock() prototypes with a `role' parameter
1471 indicating lock ordering. Update mdd code to use new locking
1476 Description: Introduce failloc constants for lockless IO tests.
1477 Details : Add two new failloc constants to test lockless IO. Only one of
1478 them in implemented---another is checked in yet to be landed
1483 Description: Add lockdep support for inode mutex.
1484 Details : Introduce and use new LOCK_INODE_MUTEX_PARENT() macro to be used
1485 in the situations where Lustre has to lock more than one inode
1490 Description: Add optional invariants checking support.
1491 Details : Add new LINVRNT() macro, optional on new --enable-invariants
1492 configure switch. This macro is to be used for consistency and
1493 sanity checks that are too expensive to be left in `production'
1498 Description: Zap lock->l_granted_mode with explicit LCK_MINMODE.
1499 Details : Use LCK_MINMODE rather than 0 to reset lock->l_granted_mode to
1504 Description: Add lockdep support for ldlm_lock and ldlm_resource.
1505 Details : Use spin_lock_nested() in (the only) situation where more than
1506 one ldlm_lock is locked simultaneously. Also, fix possible
1507 dead-lock in ldlm_lock_change_resource() by enforcing particular
1512 Description: Use struct ldlm_callback_suite in ldlm_lock_create().
1513 Details : Instead of specifying each ldlm_lock call-back through separate
1514 parameter, wrap them into struct ldlm_callback_suite.
1518 Description: Kill join_lru obd method and its callers.
1519 Details : CLIO uses lock weighting policy to keep locks over mmapped regions
1520 in memory---a requirement implemented through ->o_join_lru() obd
1521 method in HEAD. Remove this method and its users.
1525 Description: Add asynchronous ldlm ENQUEUE completion handler.
1526 Details : CLIO posts ENQUEUE requests asynchronously through ptlrpcd---a
1527 case that stock ldlm_completion_ast() cannot handle as it waits
1528 until lock is granted. Introduce new ldlm_completion_ast_async()
1529 for this. Also comment ldlm_completion_ast().
1533 Description: ldlm_error <-> errno conversion.
1534 Details : Add functions to map (rather arbitrary) between LDLM error codes
1535 and standard errno values. CLIO needs this to prevent LDLM specific
1536 constants from escaping ldlm and osc.
1540 Description: Kill unused ldlm_handle2lock_ns() function.
1541 Details : Kill unused ldlm_handle2lock_ns() function.
1545 Description: Add lu_ref support to ldlm_lock
1546 Details : lu_ref support for ldlm_lock and ldlm_resource. See lu_ref patch.
1547 lu_ref fields ->l_reference and ->lr_reference are added to ldlm_lock
1548 and ldlm_resource. LDLM interface has to be changed, because code that
1549 releases a reference on a lock, has to "know" what reference this is.
1550 In the most frequent case
1552 lock = ldlm_handle2lock(handle);
1554 LDLM_LOCK_PUT(lock);
1556 no changes are required. When any other reference (received _not_ from
1557 ldlm_handle2lock()) is released, LDLM_LOCK_RELEASE() has to be called
1558 instead of LDLM_LOCK_PUT().
1560 Arguably, changes are pervasive, and interface requires some discipline
1561 for proper use. On the other hand, it was very instrumental in finding
1562 a few leaked lock references.
1566 Description: Add ldlm_lock_addref_try().
1567 Details : Introduce ldlm_lock_addref_try() function (used by CLIO) that
1568 attempts to addref a lock that might be being canceled
1573 Description: Add ldlm_weigh_callback().
1574 Details : Add new ->l_weigh_ast() call-back to ldlm_lock. It is called
1575 by ldlm_cancel_shrink_policy() to estimate lock "value", instead of
1576 hard-coded `number of pages' logic.
1578 --------------------------------------------------------------------------------
1580 2007-08-10 Cluster File Systems, Inc. <info@clusterfs.com>
1582 * Support for kernels:
1583 2.6.5-7.283 (SLES 9),
1584 2.6.9-55.EL (RHEL 4),
1585 2.6.16.46-0.14 (SLES 10),
1586 2.6.18.8 vanilla (kernel.org)
1587 * Client support for unpatched kernels:
1588 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
1589 2.6.16 - 2.6.22 vanilla (kernel.org)
1590 * Due to recently discovered recovery problems, we do not recommend
1591 using patchless RHEL 4 clients with this or any earlier release.
1592 * Recommended e2fsprogs version: 1.39.cfs8
1593 * Note that reiserfs quotas are disabled on SLES 10 in this kernel.
1594 * Starting with this release, the ldiskfs backing filesystem required
1595 by Lustre is now in its own package, lustre-ldiskfs. This package
1596 should be installed. It is versioned separately from Lustre and
1597 may be released separately in future.
1601 Description: block reactivating mgc import until all deactivates complete
1602 Details : Fix race when failing back MDT/MGS to itself (testing)
1605 Frequency : at statup only
1607 Description: mds_lov_synchronize race leads to various problems
1608 Details : simultaneous MDT->OST connections at startup can cause the
1609 sync to abort, leaving the OSC in a bad state.
1611 Severity : enhancement
1613 Description: add optional extra BUILD_VERSION info
1614 Details : add a new environment variable (namely LUSTRE_VERS) which allows
1615 to override the lustre version.
1618 Frequency : 2.6.18 servers only
1620 Description: ll_kern_mount() doesn't release the module reference
1621 Details : The ldiskfs module reference count never drops down to 0
1622 because ll_kern_mount() doesn't release the module reference.
1627 Description: server LBUG when using old ost_num_threads parameter
1628 Details : Accept the old ost_num_threads parameter but warn that it
1629 is deprecated, and fix an off-by-one error that caused an LBUG.
1634 Description: Transient SCSI error results in persistent IO issue
1635 Details : iobuf->dr_error is not reinitialized to 0 between two
1639 Frequency : sometimes when underlying device returns I/O errors
1641 Description: OSTs not going read-only during write failures
1642 Details : OSTs are not remounted read-only when the journal commit threads
1643 get I/O errors because fsfilt_ext3 calls journal_start/stop()
1644 instead of the ext3 wrappers.
1648 Description: poor connect scaling with increasing client count
1649 Details : Don't run filter_grant_sanity_check for more than 100 exports
1650 to improve scaling for large numbers of clients.
1653 Frequency : SLES10 only
1655 Description: sanity-quota.sh quotacheck failed: rc = -22
1656 Details : Quotas cannot be enabled on SLES10.
1659 Frequency : liblustre clients only
1661 Description: getdirentries does not give error when run on compute nodes
1662 Details : getdirentries does not fail when the size specified as an argument
1663 is too small to contain at least one entry
1665 Severity : enhancement
1667 Description: Add LNET router traceability for debug purposes
1668 Details : If a checksum failure occurs with a router as part of the
1669 IO path, the NID of the last router that forwarded the bulk data
1670 is printed so it can be identified.
1675 Description: OST "spontaneously" evicts client; client has imp_pingable == 0
1676 Details : Due to a race condition, liblustre clients were occasionally
1677 evicted incorrectly.
1679 Severity : enhancement
1681 Description: lfs setstripe use optional parameters instead of postional
1684 Severity : enhancement
1686 Description: Nanosecond timestamp support for ldiskfs
1687 Details : The on-disk ldiskfs filesystem has added support for nanosecond
1688 resolution timestamps. There is not yet support for this at
1689 the Lustre filesystem level.
1692 Frequency : during server recovery
1694 Description: MDS failing to send precreate requests due to OSCC_FLAG_RECOVERING
1695 Details : request with rq_no_resend flag not awake l_wait_event if they get a
1699 Frequency : nfs export on patchless client
1701 Description: connectathon hang when test nfs export over patchless client
1702 Details : Disconnected dentry cannot be found with lookup, so we do not need
1703 to unhash it or make it invalid
1706 Description: fix llapi_lov_get_uuids() to allow many OSTs to be returned
1707 Details: : Change llapi_lov_get_uuids() to read the UUIDs from /proc instead
1708 of using an ioctl. This allows lfsck for > 160 OSTs to succeed.
1713 Description: open req refcounting wrong on reconnect
1714 Details : If reconnect happened between getting open reply from server and
1715 call to mdc_set_replay_data in ll_file_open, we will schedule
1716 replay for unreferenced request that we are about to free.
1717 Subsequent close will crash in variety of ways.
1718 Check that request is still eligible for replay in
1719 mdc_set_replay_data().
1724 Description: disable writes to filesystem when reading health_check file
1725 Details : the default for reading the health_check proc file has changed
1726 to NOT do a journal transaction and write to disk, because this
1727 can cause reads of the /proc file to hang and block HA state
1728 checking on a healthy but otherwise heavily loaded system. It
1729 is possible to return to the previous behaviour during configure
1730 with --enable-health-write.
1732 Severity : enhancement
1734 Description: 64-bit inode version
1735 Details: : Add a on-disk 64-bit inode version for ext3 to track changes made
1736 to the inode. This will be required for version-based recovery.
1741 Description: MDS fails to start if a duplicate client export is detected
1742 Details : in some rare cases it was possible for a client to connect to
1743 an MDS multiple times. Upon recovery the MDS would detect this
1744 and fail during startup. Handle this more gracefully.
1746 Severity : enhancement
1748 Description: Add -o localflock option to simulate old noflock
1750 Details : This will achieve local-only flock/fcntl locks
1756 Description: log_commit_thread vs filter_destroy race leads to crash
1757 Details : Take import reference before releasing llog record semaphore
1762 Description: Wrong request locking in request set processing
1763 Details : ptlrpc_check_set wrongly uses req->rq_lock for proctect add to
1764 imp_delayed_list, in this place should be used imp_lock.
1767 Frequency : when reconnection
1769 Description: Grant Leak when osc reconnect to OST
1770 Details : When osc reconnect ost, OST(filter) should check whether it
1771 should grant more space to client by comparing fed_grant and
1772 cl_avail_grant, and return the granted space to client instead
1773 of "new granted" space, because client will call osc_init_grant
1774 to update the client grant space info.
1777 Frequency : when client reconnect to OST
1779 Description: Grant Leak when osc do resend and replay bulk write
1780 Details : When osc reconnect to OST, OST(filter)should clear grant info of
1781 bulk write request, because the grant info while be sync between
1782 OSC and OST when reconnect, and we should ignore the grant info
1783 these of resend/replay write req.
1788 Description: Grant space more than avaiable left space sometimes.
1789 Details : When then OST is about to be full, if two bulk writing from
1790 different clients came to OST. Accord the avaliable space of the
1791 OST, the first req should be permitted, and the second one
1792 should be denied by ENOSPC. But if the seconde arrived before
1793 the first one is commited. The OST might wrongly permit second
1794 writing, which will cause grant space > avaiable space.
1797 Frequency : when client is evicted
1799 Description: Grant might be wrongly erased when osc is evicted by OST
1800 Details : when the import is evicted by server, it will fork another
1801 thread ptlrpc_invalidate_import_thread to invalidate the
1802 import, where the grant will be set to 0. While the original
1803 thread will update the grant it got when connecting. So if
1804 the former happened latter, the grant will be wrongly errased
1805 because of this race.
1810 Description: Checking Stale with correct fid
1811 Details : ll_revalidate_it should uses de_inode instead of op_data.fid2
1812 to check whether it is stale, because sometimes, we want the
1813 enqueue happened anyway, and op_data.fid2 will not be initialized.
1815 Severity : enhancement
1817 Description: update patchless client
1818 Details : Add support for patchless client with 2.6.20, 2.6.21 and RHEL 5
1821 Frequency : only with 2.4 kernel
1823 Description: random memory corruption
1824 Details : size of struct ll_inode_info is to big for union inode.u and this
1825 can be cause of random memory corruption.
1830 Description: Memory leak in recovery
1831 Details : Lov_mds_md was not free in an error handler in mds_create_object.
1832 It should also check obd_fail before fsfilt_start, otherwise if
1833 fsfilt_start return -EROFS,(failover mds during mds recovery).
1834 then the req will return with repmsg->transno = 0 and rc = EROFS.
1835 and we met hit the assert LASSERT(req->rq_reqmsg->transno ==
1836 req->rq_repmsg->transno) in ptlrpc_replay_interpret. Fcc should
1837 be freed no matter whether fsfilt_commit success or not.
1840 Frequency : only with huge count clients
1842 Description: Prevents from taking the superblock lock in llap_from_page for
1844 Details : using LL_ORIGIN_REMOVEPAGE origin flag instead of LL_ORIGIN_UNKNOW
1845 for llap_from_page call in ll_removepage prevents from taking the
1846 superblock lock for a soon died page.
1851 Description: Not check open intent error before release open handle
1852 Details : in some rare cases, the open intent error is not checked before
1853 release open handle, which may cause
1854 ASSERTION(open_req->rq_transno != 0), because it tries to release
1855 the failed open handle.
1860 Description: Set cat log bitmap only after create log success.
1861 Details : in some rare cases, the cat log bitmap is set too early. and it
1862 should be set only after create log success.
1866 Description: Accessing a block bevice can re-enable I/O when Lustre is
1867 tearing down a device.
1868 Details : dev_clear_rdonly(bdev) must be called in kill_bdev() instead of
1873 Description: service threads may hog cpus when there are a lot of requests
1874 Details : Insert cond_resched to give other threads a chance to use some CPU
1879 Description: the cat log was not initialized in recovery
1880 Details : When mds(mgs) do recovery, the tgt_count might be zero, so the
1881 unlink log on mds will not be initialized until mds post
1882 recovery. And also in mds post recovery, the unlink log will
1883 initialization will be done asynchronausly, so there will be race
1884 between add unlink log and unlink log initialization.
1888 Description: brw_stats were being printed incorrectly
1889 Details : brw_stats were being printed as log2 but all of them were not
1890 recorded as log2. Also remove some code duplication arising from
1891 filter_tally_{read,write}.
1895 Frequency : rare, only in recovery.
1896 Description: ASSERTION(req->rq_type != LI_POISON) failed
1897 Details : imp_lock should be held while iterating over imp_sending_list for
1898 prevent destroy request after get timeout in ptlrpc_queue_wait.
1902 Description: replay-single.sh test 52 fails
1903 Details : A lock's skiplist need to be cleanup when it being unlinked
1904 from its resource list.
1908 Description: Short directio read returns full requested size rather than
1910 Details : Direct I/O operations should return actual amount of bytes
1911 transferred rather than requested size.
1913 Severity : enhancement
1915 Description: metadata RPC reduction (e.g. for rm performance)
1916 Details : decrease the amount of synchronous RPC between clients and servers
1917 by canceling conflicing lock before the operation on the client
1918 and packing thier handles into the main operation RPC to server.
1920 Severity : enhancement
1922 Description: Async OSC create to avoid the blocking unnecessarily.
1923 Details : If a OST has no remain object, system will block on the creating
1924 when need to create a new object on this OST. Now, ways use
1925 pre-created objects when available, instead of blocking on an
1926 empty osc while others are not empty. If we must block, we block
1927 for the shortest possible period of time.
1931 Description: improve handling recoverable errors
1932 Details : if request processig with error which can be recoverable on server
1933 request should be resend, otherwise page released from cache and
1936 Severity : enhancement
1938 Description: refine locking for avoid write wrong info into lov_objid file
1939 Details : fix possible races with add new target and write/update data in
1942 --------------------------------------------------------------------------------
1944 2007-05-03 Cluster File Systems, Inc. <info@clusterfs.com>
1949 Frequency : on some architectures
1951 Description: 1.6 client sometimes fails to mount from a 1.4 MDT
1952 Details : Uninitialized flags sometimes cause configuration commands to
1956 Frequency : patchless clients only
1958 Description: missing __iget() symbol export
1959 Details : The __iget() symbol export is missing. To avoid the need for
1960 this on patchless clients the deathrow inode reaper is turned
1961 off, and we depend on the VM to clean up old inodes. This
1962 dependency was during via the fix for bug 12181.
1964 --------------------------------------------------------------------------------
1966 2007-04-19 Cluster File Systems, Inc. <info@clusterfs.com>
1968 * CONFIGURATION CHANGE. This version of Lustre WILL NOT
1969 INTEROPERATE with older versions automatically. In many cases a
1970 special upgrade step is needed. Please read the
1971 user documentation before upgrading any part of a live system.
1972 * WIRE PROTOCOL CHANGE from previous 1.6 beta versions. This
1973 version will not interoperate with 1.6 betas before beta5 (1.5.95).
1974 * WARNING: Lustre configuration and startup changes are required with
1975 this release. See https://mail.clusterfs.com/wikis/lustre/MountConf
1979 Severity : enhancement
1981 Description: MountConf
1982 Details : Lustre configuration is now managed via mkfs and mount
1983 commands instead of lmc and lconf. New obd types (MGS, MGC)
1984 are added for dynamic configuration management. See
1985 https://mail.clusterfs.com/wikis/lustre/MountConf for
1988 Severity : enhancement
1990 Description: dynamic OST addition
1991 Details : OSTs can now be added to a live filesystem
1993 Severity : enhancement
1995 Description: startup order invariance
1996 Details : MDTs and OSTs can be started in any order. Clients only
1997 require the MDT to complete startup.
1999 Severity : enhancement
2001 Description: parallel, asynchronous orphan cleanup
2002 Details : orphan cleanup is now performed in separate threads for each
2003 OST, allowing parallel non-blocking operation.
2005 Severity : enhancement
2007 Description: optimized stripe assignment
2008 Details : stripe assignments are now made based on ost space available,
2009 ost previous usage, and OSS previous usage, in order to try
2010 to optimize storage space and networking resources.
2012 Severity : enhancement
2014 Description: Permanently set tunables
2015 Details : All writable /proc/fs/lustre tunables can now be permanently
2016 set on a per-server basis, at mkfs time or on a live system.
2018 Severity : enhancement
2020 Description: Lustre message v2
2021 Details : Add lustre message format v2.
2023 Severity : enhancement
2025 Description: client OST exclusion list
2026 Details : Clients can be started with a list of OSTs that should be
2027 declared "inactive" for known non-responsive OSTs.
2031 Description: SPEC SFS validation failure on NFS v2 over lustre.
2032 Details : Changes the blocksize for regular files to be 2x RPC size,
2033 and not depend on stripe size.
2035 Severity : enhancement
2037 Description: Multiple MD RPCs in flight.
2038 Details : Further unserialise some read-only MDS RPCs - learn about intents.
2039 To avoid overly-overloading MDS, introduce a limit on number of
2040 MDS RPCs in flight for a single client and add /proc controls
2041 to adjust this limit.
2043 Severity : enhancement
2045 Description: client read/write statistics
2046 Details : Add client read/write call usage stats for performance
2047 analysis of user processes.
2048 /proc/fs/lustre/llite/*/offset_stats shows non-sequential
2049 file access. extents_stats shows chunk size distribution.
2050 extents_stats_per_process show chunk size distribution per
2053 Severity : enhancement
2055 Description: mds statistics
2056 Details : Add detailed mds operations statistics in
2057 /proc/fs/lustre/mds/*/stats.
2061 Description: Failure of copying files with lustre special EAs.
2062 Details : Client side always return success for setxattr call for lustre
2063 special xattr (currently only "trusted.lov").
2068 Description: Refcount LNET uuids
2069 Details : The global LNET uuid list grew linearly with every startup;
2070 refcount repeated list entries instead of always adding to
2073 Severity : enhancement
2075 Description: Dynamic service threads
2076 Details : Within a small range, start extra service threads
2077 automatically when the request queue builds up.
2080 Frequency : mixed-endian client/server environments
2082 Description: mixed-endian crashes
2083 Details : The new msg_v2 system had some failures in mixed-endian
2086 Severity : enhancement
2088 Description: Easy OST removal
2089 Details : OSTs can be permanently deactivated with e.g. 'lctl
2090 conf_param lustre-OST0001.osc.active=0'
2092 Severity : enhancement
2094 Description: MGS proc entries
2095 Details : Added basic proc entries for the MGS showing what filesystems
2098 Severity : enhancement
2100 Description: provide MGS failover
2101 Details : Added config lock reacquisition after MGS server failover.
2103 Severity : enhancement
2105 Description: add Linux 2.4 support
2106 Details : Added support for RHEL 2.4.21 kernel for 1.6 servers and clients
2110 Description: a large application tries to do I/O to the same resource and dies
2111 in the middle of it.
2112 Details : Check the req->rq_arrival time after the call to
2113 ost_brw_lock_get(), but before we do anything about
2114 processing it & sending the BULK transfer request. This
2115 should help move old stale pending locks off the queue as
2116 quickly as obd_timeout.
2119 Frequency : when an incorrect nid is specified during startup
2121 Description: ptlrpc connect to non-existant node causes kernel crash
2122 Details : LNET can't be re-entered from an event callback, which
2123 happened when we expire a message after the export has been
2124 cleaned up. Instead, hand the zombie cleanup off to another
2127 Severity : enhancement
2129 Description: plain/inodebits lock performance improvement
2130 Details : Grouping plain/inodebits in granted list by their request modes
2131 and bits policy, thus improving the performance of search through
2135 Frequency : only if OST filesystem is corrupted
2137 Description: client incorrectly hits assertion in ptlrpc_replay_req()
2138 Details : for a short time RPCs with bulk IO are in the replay list,
2139 but replay of bulk IOs is unimplemented. If the OST filesystem
2140 is corrupted due to disk cache incoherency and then replay is
2141 started it is possible to trip an assertion. Avoid putting
2142 committed RPCs into the replay list at all to avoid this issue.
2145 Frequency : liblustre (e.g. catamount) on a large cluster with >= 8 OSTs/OSS
2147 Description: System hang on startup
2148 Details : This bug allowed the liblustre (e.g. catamount) client to
2149 return to the app before handling all startup RPCs. This
2150 could leave the node unresponsive to lustre network traffic
2151 and manifested as a server ptllnd timeout.
2153 Severity : enhancement
2155 Description: Add "/proc/sys/lustre/debug_peer_on_timeout"
2156 Details : liblustre envirable: LIBLUSTRE_DEBUG_PEER_ON_TIMEOUT
2157 boolean to control whether to print peer debug info when a
2158 client's RPC times out.
2161 Frequency : only for kernels with patches from Lustre below 1.4.3
2163 Description: Remove old rdonly API
2164 Details : Remove old rdonly API which unused from at least lustre 1.4.3
2167 Frequency : only for devices with external journals
2169 Description: Set external device read-only also
2170 Details : During a commanded failover stop, we set the disk device
2171 read-only while the server shuts down. We now also set any
2172 external journal device read-only at the same time.
2175 Frequency : when upgrading from 1.4 while trying to change parameters
2177 Description: The wrong (new) MDC name was used when setting parameters for
2178 upgraded MDT's. Also allows changing of OSC (and MDC)
2179 parameters if --writeconf is specified at tunefs upgrade time.
2182 Frequency : when setting specific ost indicies
2184 Description: QOS code breaks on skipped indicies
2185 Details : Add checks for missing OST indicies in the QOS code, so OSTs
2186 created with --index need not be sequential.
2188 Severity : enhancement
2190 Description: Add uninit_groups feature to ldiskfs2 to speed up e2fsck
2191 Details : The uninit_groups feature works in conjunction with the kernel
2192 filesystem code (ldiskfs2 only) and e2fsprogs-1.39-cfs6 to speed
2193 up the pass1 processing of e2fsck. This is a read-only feature
2194 in ldiskfs2 only, so older kernels and current ldiskfs cannot
2195 mount filesystems that have had this feature enabled.
2197 Severity : enhancement
2199 Description: Improve multi-block allocation algorithm to avoid fragmentation
2200 Details : The mballoc3 code (ldiskfs2 only) adds new mechanisms to improve
2201 allocation locality and avoid filesystem fragmentation.
2203 ------------------------------------------------------------------------------
2205 2007-04-01 Cluster File Systems, Inc. <info@clusterfs.com>
2207 * Support for kernels:
2208 2.4.21-47.0.1.EL (RHEL 3)
2209 2.6.5-7.283 (SLES 9)
2210 2.6.9-42.0.10.EL (RHEL 4)
2211 2.6.12.6 vanilla (kernel.org)
2212 2.6.16.27-0.9 (SLES 10)
2213 * Recommended e2fsprogs version: 1.39.cfs5
2215 * Note that reiserfs quotas are disabled on SLES 10 in this kernel
2219 Frequency : occasional, depends on client load and configuration
2220 Bugzilla : 12181, 12203
2221 Description: data loss for recently-modified files
2223 Details : In some cases it is possible that recently written or created
2224 files may not be written to disk in a timely manner (this should
2225 normally be within 30s unless client IO load is very high).
2226 The problem appears as zero-length files or files that are a
2227 multiple of 1MB in size after a client crash or client eviction
2228 that are missing data at the end of the file.
2230 This problem is more likely to be hit on clients where files are
2231 repeatedly created and unlinked in the same directory, clients
2232 have a large amount of RAM, have many CPUs, the filesystem has
2233 many OSTs, the clients are rebooted frequently, and/or the files
2234 are not accessed by other nodes after being written.
2236 The presence of the problem can be detected by looking at
2237 /proc/sys/fs/inode-state. If the first number (nr_inodes) is
2238 smaller than the second (nr_unused) then dirty files will not
2239 be flushed automatically to disk. "sync; sleep 10" should be
2240 run several times on the node before unmounting it to update
2241 Lustre (this is also safe to run on nodes without this problem).
2243 There is also a related kernel bug in the RHEL4 4 2.6.9 kernel
2244 that can cause this same problem, so customers using that kernel
2245 also need to update the kernel in addition to Lustre. In order
2246 to properly fix this bug, the RHEL3 2.4.21 kernel is also updated.
2248 It is normal that files written just before a client crash (less
2249 than 30s) may not yet have been flushed to disk, even for local
2253 Frequency : frequent on thin XT3 nodes
2255 Description: UUID collision on thin XT3 Linux nodes
2256 Details : UUIDs on Compute Node Linux XT3 nodes were not generated
2257 randomly, since we relied on an insufficiently-seeded PRNG.
2262 Description: OSS hangs after "All ost request buffers busy"
2263 Details : A deadlock between quota and journal operations caused OSS
2264 hangs after printing "All ost request buffers busy."
2267 Frequency : always on liblustre builds
2269 Description: Cleanup compiler warnings on liblustre
2272 Frequency : always on liblustre builds on XT3
2274 Description: LC_CONFIG_CDEBUG don't run while build liblustre on XT3.
2278 Description: Addition of EXT3_FEATURE_RO_COMPAT_DIR_NLINKS flag for
2279 > 32000 subdirectories
2280 Details : Add EXT3_FEATURE_RO_COMPAT_DIR_NLINK flag to
2281 EXT3_FEATURE_RO_COMPAT_SUPP. This flag will be set whenever
2282 subdirectory count crosses 32000. This will aid e2fsck to
2283 correctly handle more than 32000 subdirectories.
2286 Frequency : liblustre (e.g. catamount) on a large cluster with >= 8 OSTs/OSS
2288 Description: System hang on startup
2289 Details : This bug allowed the liblustre (e.g. catamount) client to
2290 return to the app before handling all startup RPCs. This
2291 could leave the node unresponsive to lustre network traffic
2292 and manifested as a server ptllnd timeout.
2294 Severity : enhancement
2296 Description: Add "/proc/sys/lustre/debug_peer_on_timeout"
2297 (liblustre envirable: LIBLUSTRE_DEBUG_PEER_ON_TIMEOUT)
2298 boolean to control whether to print peer debug info when a
2299 client's RPC times out.
2304 Description: make O_SYNC working on 2.6 kernels
2305 Details : 2.6 kernels use different method for mark pages for write,
2306 so need add a code to lustre for O_SYNC work.
2311 Description: Failure to close file and release space on NFS
2312 Details : Put inode details into lock acquired in ll_intent_file_open.
2313 Use mdc_intent_lock in ll_intent_open to properly
2314 detect all kind of errors unhandled by mdc_enqueue.
2319 Description: proc file read during shutdown sometimes raced obd removal,
2321 Details : Add lock to prevent obd access after proc file removal.
2324 Frequency : Only for files larger than 4GB on 32-bit clients.
2326 Description: improperly doing page alignment of locks
2327 Details : Modify lustre core code to use CFS_PAGE_* defines instead of
2328 PAGE_*. Make CFS_PAGE_MASK a 64-bit mask.
2333 Description: RPCs being resent when they shouldn't be
2334 Details : Some RPCs that should not be resent are being resent. This
2335 can cause inconsistencies in the RPC state machine. Do not
2336 resend such requests.
2339 Frequency : rare, only with NFS export
2341 Description: Crash on NFS re-export node
2342 Details : under very unusual load conditions an assertion is hit in
2343 ll_intent_file_open()
2346 Frequency : only if OST filesystem is corrupted
2348 Description: client incorrectly hits assertion in ptlrpc_replay_req()
2349 Details : for a short time RPCs with bulk IO are in the replay list,
2350 but replay of bulk IOs is unimplemented. If the OST filesystem
2351 is corrupted due to disk cache incoherency and then replay is
2352 started it is possible to trip an assertion. Avoid putting
2353 committed RPCs into the replay list at all to avoid this issue.
2358 Description: large O_DIRECT requests fail under memory pressure/fragmentation
2359 Details : Large single O_DIRECT read and write calls can fail to allocate
2360 a sufficiently large buffer to process the request. In case of
2361 allocation failure the allocation is retried with a smaller
2362 buffer and broken into smaller requests.
2364 Severity : enhancement
2366 Description: Add -o localflock option to simulate old noflock behaviour.
2367 Details : This will achieve local-only flock/fcntl locks coherentness.
2372 Description: versioning check is incomplete
2373 Details : Checking the version difference of client vs. server, report
2374 error if the gap is too big.
2379 Description: add support PG_writeback bit
2380 Details : add support for PG_writeback bit for Lustre, for more carefull
2381 work with page cache in 2.6 kernel. This also fix some deadlocks
2382 and remove hack for work O_SYNC with 2.6 kernel.
2384 Severity : enhancement
2386 Description: Add uninit_groups feature to ldiskfs2 to speed up e2fsck
2387 Details : The uninit_groups feature works in conjunction with the kernel
2388 filesystem code (ldiskfs2 only) and e2fsprogs-1.39-cfs6 to speed
2389 up the pass1 processing of e2fsck. This is a read-only feature
2390 in ldiskfs2 only, so older kernels and current ldiskfs cannot
2391 mount filesystems that have had this feature enabled.
2393 Severity : enhancement
2395 Description: Improve multi-block allocation algorithm to avoid fragmentation
2396 Details : The mballoc3 code (ldiskfs2 only) adds new mechanisms to improve
2397 allocation locality and avoid filesystem fragmentation.
2399 ------------------------------------------------------------------------------
2401 2007-02-09 Cluster File Systems, Inc. <info@clusterfs.com>
2403 * Support for kernels:
2404 2.6.9-42.0.3.EL (RHEL 4)
2405 2.6.5-7.276 (SLES 9)
2406 2.4.21-47.0.1.EL (RHEL 3)
2407 2.6.12.6 vanilla (kernel.org)
2408 2.6.16.21-0.8 (SLES10)
2409 * Recommended e2fsprogs version: 1.39.cfs2-0
2411 * The backwards-compatible /proc/sys/portals symlink has been removed
2412 in this release. Before upgrading, please ensure that you change
2413 any configuration scripts or /etc/sysctl.conf files that access
2414 /proc/sys/portals/* or sysctl portals.* to use the corresponding
2415 entry in /proc/sys/lnet or sysctl lnet.*. This change can be made
2416 in advance of the upgrade on any system running Lustre 1.4.6 or
2417 newer, since /proc/sys/lnet was added in that version.
2418 * Note that reiserfs quotas are disabled on SLES 10 in this kernel
2422 Frequency : only when quota is used
2424 Description: avoid scanning export list for quota master
2425 Details : Change the algorithms to avoid scanning export list in order
2426 to improve the efficiency.
2429 Frequency : MDS failover only, very rarely
2431 Description: "went back in time" messages on mds failover
2432 Details : The greatest transno may be lost when the current operation
2433 finishes with an error (transno==0) and the client's last_rcvd
2434 record is over-written. Save the greatest transno in the
2435 mds_last_transno for this case.
2438 Frequency : always for specific kernels and striping counts
2440 Description: client may get "Matching packet too big" without ACL support
2441 Details : Clients compiled without CONFIG_FS_POSIX_ACL get an error message
2442 when trying to access files in certain configurations. The
2443 clients should in fact be denied when mounting because they do
2444 not understand ACLs.
2447 Frequency : Cray XT3 with more than 4000 clients and multiple jobs
2449 Description: many clients connecting with IO in progress causes connect timeouts
2450 Details : Avoid synchronous journal commits to avoid delays caused by many
2451 clients connecting/disconnecting when bulk IO is in progress.
2452 Queue liblustre connect requests on OST_REQUEST_PORTAL instead of
2453 OST_IO_PORTAL to avoid delays behind potentially many pending
2457 Frequency : occasionally with multiple writers to a single file
2459 Description: shared writes to file may result in wrong size reported by stat()
2460 Details : Allow growing of kms when extent lock is cancelled
2463 Frequency : always with random mmap IO to multi-striped file
2465 Description: mmap write might be lost if we are writing to a 'hole' in stripe
2466 Details : Only if the hole is at the end of OST object so that kms is too
2467 small. Fix is to increase kms accordingly in ll_nopage.
2470 Frequency : rare, only if OST filesystem is inconsistent with MDS filesystem
2472 Description: writes to a missing object would leak memory on the OST
2473 Details : If there is an inconsistency between the MDS and OST filesystems,
2474 such that the MDS references an object that doesn't exist, writes
2475 to that object will leak memory due to incorrect cleanup in the
2476 error handling path, eventually running out of memory on the OST.
2481 Description: Creating too long symlink causes lustre errors
2482 Details : Check symlink and name lengths before sending requests to MDS.
2485 Frequency : only if flock is enabled (not on by default)
2487 Description: posix locks not released on fd closure on 2.6.9+
2488 Details : We failed to add posix locks to list of inode locks on 2.6.9+
2489 kernels, this caused such locks not to be released on fd close and
2490 then assertions on fs unmount about still used locks.
2493 Frequency : MDS failover only, very rarely
2495 Description: clients may get ASSERTION(granted_lock != NULL)
2496 Details : When request was taking a long time, and a client was resending
2497 a getattr by name lock request. The were multiple lock requests
2498 with the same client lock handle and
2499 mds_getattr_name->fixup_handle_for_resent_request found one of the
2500 lock handles but later failed with ASSERTION(granted_lock != NULL).
2505 Description: handle->h_buffer_credits > 0, assertion failure
2506 Details : h_buffer_credits is zero after truncate, causing assertion
2507 failure. This patch extends the transaction or creates a new
2511 Frequency : NFS re-export or patchless client
2512 Bugzilla : 11179, 10796
2513 Description: Crash on NFS re-export node (__d_move)
2514 Details : We do not want to hash the dentry if we don't have a lock.
2515 But if this dentry is later used in d_move, we'd hit uninitialised
2516 list head d_hash, so we just do this to init d_hash field but
2517 leave dentry unhashed.
2520 Frequency : NFS re-export or patchless client
2522 Description: NFS exports has problem with symbolic link
2523 Details : lustre client didn't properly install dentry when re-exported
2524 to NFS or running patchless client.
2527 Frequency : NFS re-export or patchless client
2529 Description: Various nfs/patchless fixes.
2530 Details : fixes reuse disconected alias for lookup process - this fixes
2531 warning "find_exported_dentry: npd != pd",
2532 fix permission error with open files at nfs.
2533 fix apply umask when do revalidate.
2536 Frequency : occasional
2538 Description: Crash on NFS re-export node
2539 Details : calling clear_page() on the wrong pointer triggered oops in
2540 generic_mapping_read().
2543 Frequency : rarely, using O_DIRECT IO
2545 Description: unaligned directio crashes client with LASSERT
2546 Details : check for unaligned buffers before trying any requests.
2549 Frequency : rarely, using CFS RAID5 patches in non-standard kernel series
2551 Description: stale data returned from RAID cache
2552 Details : If only a small amount of IO is done to the RAID device before
2553 reading it again it is possible to get stale data from the RAID
2554 cache instead of reading it from disk.
2557 Frequency : always for sles10 kernel
2559 Description: sles10 support
2560 Details : ll_follow_link: compile fixes and using of nd_set_link
2561 under newer kernels.
2564 Frequency : depends on arch, kernel and compiler version, always on sles10
2567 Description: recursive or deep enough symlinks cause stack overflow
2568 Details : getting rid of large stack-allocated variable in
2572 Frequency : depends on hardware
2574 Description: lustre write performance loss in the SLES10 kernel
2575 Details : the performance loss is caused by using of write barriers in the
2576 ext3 code. The SLES10 kernel turns barrier support on by
2577 default. The fix is to undo that change for ldiskfs.
2579 ------------------------------------------------------------------------------
2581 2006-12-09 Cluster File Systems, Inc. <info@clusterfs.com>
2583 * Support for kernels:
2584 2.6.9-42.0.3EL (RHEL 4)
2585 2.6.5-7.276 (SLES 9)
2586 2.4.21-47.0.1.EL (RHEL 3)
2587 2.6.12.6 vanilla (kernel.org)
2591 Frequency : quota enabled and large files being deleted
2593 Description: releasing more than 4GB of quota at once hangs OST
2594 Details : If a user deletes more than 4GB of files on a single OST it
2595 will cause the OST to spin in an infinite loop. Release
2596 quota in < 4GB chunks, or use a 64-bit value for 1.4.7.1+.
2601 Description: statfs data retrieved from /proc may be stale or zero
2602 Details : When reading per-device statfs data from /proc, in the
2603 {kbytes,files}_{total,free,avail} files, it may appear
2604 as zero or be out of date.
2607 Frequency : systems with MD RAID1 external journal devices
2609 Description: lconf's call to blkid is confused by RAID1 journal devices
2610 Details : Use the "blkid -l" flag to locate the MD RAID device instead
2611 of returning all block devices that match the journal UUID.
2614 Frequency : always, for aggregate stripe size over 4GB
2616 Description: "lfs setstripe" fails assertion when setting 4GB+ stripe width
2617 Details : Using "lfs setstripe" to set stripe size * stripe count over 4GB
2618 will fail the kernel with "ASSERTION(lsm->lsm_xfersize != 0)"
2621 Frequency : always if "lfs find" used on a local file/directory
2623 Description: "lfs find" segfaults if used on a local file/directory
2624 Details : The case where a directory component was not specified wasn't
2625 handled correctly. Handle this properly.
2628 Frequency : always on ppc64
2630 Description: the write to an ext3 filesystem mounted with mballoc got stuck
2631 Details : ext3_mb_generate_buddy() uses find_next_bit() which does not
2632 perform endianness conversion.
2635 Frequency : rarely (truncate to non-zero file size after write under load)
2636 Bugzilla : 10730, 10687
2637 Description: Files padded with zeros to next 4K multiple
2638 Details : With filesystems mounted using the "extents" option (2.6 kernels)
2639 it is possible that files that are truncated to a non-zero size
2640 immediately after being written are filled with zero bytes beyond
2641 the truncated size. No file data is lost.
2643 Severity : enhancement
2645 Description: Allow recovery/failover for liblustre clients.
2646 Details : liblustre clients were unaware of failover configurations until
2649 Severity : enhancement
2651 Description: user file locks should fail when not mounting with flock option
2652 Details : Set up an error-returning stub in ll_file_operations.lock field
2653 to prevent incorrect behaviour when client is mounted without
2654 flock option. Also, set up properly f_op->flock field for
2658 Frequency : always on ia64
2660 Description: "lfs df" loops on printing out MDS statfs information
2661 Details : The obd_ioctl_data was not initialized and in some systems
2662 this caused a failure during the ioctl that did not return
2663 an error. Initialize the struct and return an error on failure.
2666 Frequency : SLES 9 only
2668 Description: Error of copying files with lustre special EAs as root
2669 Details : Client side always return success for setxattr call for lustre
2670 special xattr (currently only "trusted.lov").
2673 Frequency : rarely on clusters with both ia64+i386 clients
2675 Description: ia64+i686 clients doing shared IO on the same file may LBUG
2676 Details : In rare cases when both ia64+i686 (or other mixed-PAGE_SIZE)
2677 clients are doing concurrent writes to the same file it is
2678 possible that the ia64 clients may LASSERT because the OST
2679 extent locks are not PAGE_SIZE aligned. Ensure that grown
2680 locks are always aligned on the request boundary.
2683 Frequency : specific use, occasional
2685 Description: Overwriting in use executable truncates on-disk binary image
2686 Details : If one node attempts to overwrite an executable in use by
2687 another node, we now correctly return ETXTBSY instead of
2688 truncating the file.
2690 Severity : enhancement
2692 Description: Async OSC create to avoid the blocking unnecessarily.
2693 Details : If a OST has no remain object, system will block on the creating
2694 when need to create a new object on this OST. Now, ways use
2695 pre-created objects when available, instead of blocking on an
2696 empty osc while others are not empty. If we must block, we block
2697 for the shortest possible period of time.
2702 Description: chmod on Lustre root is propagated to other clients
2703 Details : Re-validate root's dentry in ll_lookup_it to avoid having it
2704 invalid by the follow_mount time.
2707 Frequency : liblustre clients only
2709 Description: Race in 'instant cancel' lock handling could lead to such locks
2710 never to be granted in case of SMP MDS
2711 Details : Do not destroy not yet granted but cbpending locks in
2715 Frequency : replay/resend of open
2717 Description: non null lock assetion failure in mds_intent_policy
2718 Details : Trying to replay/resend lockless open requests resulted in
2719 mds_open() returning 0 with no lock. Now it sets a flag if
2720 a lock is going to be returned.
2722 Severity : enhancement
2724 Description: Checksum enhancements
2725 Details : New checksum enhancements allow for resending RPCs that failed
2728 Severity : enhancement
2730 Description: Tunables on number of dirty pages in cacche
2731 Details : Allow to set limit on number of dirty pages cached.
2736 Description: client crash on unmount - lock still has references
2737 Details : In some error handling cases it was possible to leak a lock
2738 reference on a client while accessing a file. This was not
2739 harmful to the client during operation, but would cause the
2740 client to crash when the filesystem is unmounted.
2743 Frequency : specific case, rare
2745 Description: ETXTBSY on mds though file not in use by client
2746 Details : ETXTBSY is no longer incorrectly returned when attempting to
2747 chmod or chown a directory that the user previously tried to
2748 execute or a currently-executing binary.
2751 Frequency : extremely rare except on liblustre-based clients
2753 Description: Lustre space not freed when files are deleted
2754 Details : Clean up open-unlinked files after client eviction. Previously
2755 the unlink was skipped and the files remained as orphans.
2760 Description: OST failure "would be an LBUG" in waiting_locks_callback()
2761 Details : In some cases it was possible to send a blocking callback to
2762 a client doing a glimpse, even though that client didn't get
2763 a lock granted. When the glimpse lock is cancelled on the OST
2764 the freed lock is left on the waiting list and corrupted the list.
2767 Frequency : all core dumps
2769 Description: Broke core dumps to lustre
2770 Details : Negative dentry may be unhashed if parent does not have UPDATE
2771 lock, but some callers, e.g. do_coredump, expect dentry to be
2772 hashed after successful create, hash it in ll_create_it.
2774 ------------------------------------------------------------------------------
2776 2006-09-13 Cluster File Systems, Inc. <info@clusterfs.com>
2778 * Support for kernels:
2779 2.6.9-42.0.2.EL (RHEL 4)
2780 2.6.5-7.276 (SLES 9)
2781 2.4.21-40.EL (RHEL 3)
2782 2.6.12.6 vanilla (kernel.org)
2786 Frequency : always on RHEL 3
2788 Description: Number of open files grows over time
2789 Details : The number of open files grows over time, whether or not
2790 Lustre is started. This was due to a filp leak introduced
2791 by one of our kernel patches.
2793 ------------------------------------------------------------------------------
2795 08-20-2006 Cluster File Systems, Inc. <info@clusterfs.com>
2797 * Support for kernels:
2798 2.6.9-42.EL (RHEL 4)
2799 2.6.5-7.276 (SLES 9)
2800 2.4.21-40.EL (RHEL 3)
2801 2.6.12.6 vanilla (kernel.org)
2806 Bugzilla : 5719, 9635, 9792, 9684
2807 Description: OST (or MDS) trips assertions in (re)connection under heavy load
2808 Details : If a server is under heavy load and cannot reply to new
2809 connection requests before the client resends the (re)connect,
2810 the connection handling code can behave badly if two service
2811 threads are concurrently handing separate (re)connections from
2812 the same client. Add better locking to the connection handling
2813 code, and ensure that only a single connection will be processed
2814 for a given client UUID, even if the lock is dropped.
2816 Severity : enhancement
2818 Description: add TCP zero-copy support to kernel
2819 Details : Add support to the kernel TCP stack to allow zero-copy bulk
2820 sends if the hardware supports scatter-gather and checksumming.
2821 This allows socklnd to do client-write and server-read more
2822 efficiently and reduce CPU utilization from skbuf copying.
2825 Frequency : only if NFS exporting from client
2827 Description: NULL pointer deref in ll_iocontrol() if chattr mknod file
2828 Details : If setting attributes on a file created under NFS that had
2829 never been opened it would be possible to oops the client
2830 if the file had no objects.
2833 Frequency : always for liblustre
2835 Description: liblustre client does MDS+OSTs setattr RPC for each write
2836 Details : When doing a write from a liblustre client, the client
2837 incorrectly issued an RPC to the MDS and each OST the file was
2838 striped over in order to update the timestamps. When writing
2839 with small chunks and many clients this could overwhelm the MDS
2840 with RPCs. In all cases it would slow down the write because
2841 these RPCs are unnecessary.
2843 Severity : enhancement
2845 Description: allow number of MDS service threads to be changed at module load
2846 Details : It is now possible to change the number of MDS service threads
2847 running. Adding "options mds mds_num_threads={N}" to the MDS's
2848 /etc/modprobe.conf will set the number of threads for the next
2849 time Lustre is restarted (assuming the "mds" module is also
2850 reloaded at that time). The default number of threads will
2851 stay the same, 32 for most systems.
2856 Description: OST crash if filesystem is unformatted or corrupt
2857 Details : If an OST is started on a device that has never been formatted
2858 or if the filesystem is corrupt and cannot even mount then the
2859 error handling cleanup routines would dereference a NULL pointer.
2864 Description: NULL pointer deref in llap_from_page.
2865 Details : get_cache_page_nowait can return a page with NULL (or otherwise
2866 incorrect) mapping if the page was truncated/reclaimed while it was
2867 searched for. Check for this condition and skip such pages when
2868 doing readahead. Introduce extra check to llap_from_page() to
2869 verify page->mapping->host is non-NULL (so page is not anonymous).
2872 Frequency : Sometimes when using sys_sendfile
2874 Description: "page not covered by a lock" warnings from ll_readpage
2875 Details : sendfile called ll_readpage without right page locks present.
2876 Now we introduced ll_file_sendfile that does necessary locking
2877 around call to generic_file_sendfile() much like we do in
2881 Frequency : with certain MDS communication failures at client mount time
2883 Description: NULL pointer deref after failed client mount
2884 Details : a client connection request may delayed by the network layer
2885 and not be sent until after the PTLRPC layer has timed out the
2886 request. If the client fails the mount immediately it will try
2887 to clean up before the network times out the request. Add a
2888 reference from the request import to the obd device and delay
2889 the cleanup until the network drops the request.
2892 Frequency : occasionally during client (re)connect
2894 Description: assertion failure during client (re)connect
2895 Details : processing a client connection request may be delayed by the
2896 client or server longer than the client connect timeout. This
2897 causes the client to resend the connection request. If the
2898 original connection request is replied in this interval, the
2899 client may trip an assertion failure in ptlrpc_connect_interpret()
2900 which thought it would be the only running connect process.
2903 Frequency : only with obd_echo servers and clients that are rebooted
2905 Description: kernel BUG accessing uninitialized data structure
2906 Details : When running an obd_echo server it did not start the ping_evictor
2907 thread, and when a client was evicted an uninitialized data
2908 structure was accessed. Start the ping_evictor in the RPC
2909 service startup instead of the OBD startup.
2911 Severity : enhancement
2912 Bugzilla : 10193 (patchless)
2913 Description: Remove dependency on various unexported kernel interfaces.
2914 Details : No longer need reparent_to_init, exit_mm, exit_files,
2915 sock_getsockopt, filemap_populate, FMODE_EXEC, put_filp.
2918 Frequency : rare (only users of deprecated and unsupported LDAP config)
2920 Description: write_conf for zeroconf mount queried LDAP incorrectly for client
2921 Details : LDAP apparently contains 'lustreName' attributes instead of
2922 'name'. A simple remapping of the name is sufficient.
2925 Frequency : rare (only with non-default dump_on_timeout debug enabled)
2927 Description: waiting_locks_callback trips kernel BUG if client is evicted
2928 Details : Running with the dump_on_timeout debug flag turned on makes
2929 it possible that the waiting_locks_callback() can try to dump
2930 the Lustre kernel debug logs from an interrupt handler. Defer
2931 this log dumping to the expired_lock_main() thread.
2933 Severity : enhancement
2935 Description: Support NFS exporting on 2.6 kernels.
2936 Details : Implement non-rawops metadata methods for NFS server to use without
2937 changing NFS server code.
2940 Frequency : very rare (synthetic metadata workload only)
2942 Description: two racing renames might cause an MDS thread to deadlock
2943 Details : Running the "racer" program may cause one MDS thread to rename
2944 a file from being the source of a rename to being the target of
2945 a rename at exactly the same time that another thread is doing
2946 so, and the second thread has already enqueued these locks after
2947 doing a lookup of the target and is trying to relock them in
2948 order. Ensure that we don't try to re-lock the same resource.
2951 Frequency : only very large systems with liblustre clients
2953 Description: slow eviction of liblustre clients with the "evict_by_nid" RPC
2954 Details : Use asynchronous set_info RPCs to send the "evict_by_nid" to
2955 all OSTs in parallel. This allows the eviction of stale liblustre
2956 clients to proceed much faster than if they were done in series,
2957 and also offers similar improvements for other set_info RPCs.
2962 Description: excessive CPU usage during initial read phase on client
2963 Details : During the initial read phase on a client, it would agressively
2964 retry readahead on the file, consuming too much CPU and impacting
2965 performance (since 1.4.5.8). Improve the readahead algorithm
2966 to avoid this, and also improve some other common cases (read
2967 of small files in particular, where "small" is files smaller than
2968 /proc/fs/lustre/llite/*/max_read_ahead_whole_mb, 2MB by default).
2973 Description: MDS crash when receiving packet with unknown intent.
2974 Details : Do not LBUG in unknown intent case, just return -EFAULT
2976 Severity : enhancement
2977 Bugzilla : 9293, 9385
2978 Description: MDS RPCs are serialised on client. This is unnecessary for some.
2979 Details : Do not serialize getattr (non-intent version) and statfs.
2982 Frequency : occasional, when OST network is overloaded/intermittent
2984 Description: client evicted by OST after bulk IO timeout
2985 Details : If a client sends a bulk IO request (read or write) the OST
2986 may evict the client if it is unresposive to its data GET/PUT
2987 request. This is incorrect if the network is overloaded (takes
2988 too long to transfer the RPC data) or dropped the OST GET/PUT
2989 request. There is no need to evict the client at all, since
2990 the pinger and/or lock callbacks will handle this, and the
2991 client can restart the bulk request.
2994 Frequency : Always when mmapping file with no objects
2996 Description: client crashes when mmapping file with no objects
2997 Details : Check that we actually have objects in a file before doing any
2998 operations on objects in ll_vm_open, ll_vm_close and
3004 Description: Request leak when working with deleted CWD
3005 Details : Introduce advanced request refcount tracking for requests
3006 referenced from lustre intent.
3008 Severity : Enhancement
3010 Description: Cache open file handles on client.
3011 Details : MDS now will return special lock along with openhandle, if
3012 requested and client is allowed to hold openhandle, even if unused,
3013 until such a lock is revoked. Helps NFS a lot, since NFS is opening
3014 closing files for every read/write openration.
3016 Severity : Enhancement
3018 Description: Cache open negative dentries on client when possible.
3019 Details : Guard negative dentries with UPDATE lock on parent dir, drop
3020 negative dentries on lock revocation.
3025 Description: Remounting a client read-only wasn't possible with a zconf mount
3026 Details : It wasn't possible to remount a client read-only with llmount.
3028 Severity : enhancement
3029 Description: Include MPICH 1.2.6 Lustre ADIO interface patch
3030 Details : In lustre/contrib/ or /usr/share/lustre in RPM a patch for
3031 MPICH is included to add Lustre-specific ADIO interfaces.
3032 This is based closely on the UFS ADIO layer and only differs
3033 in file creation, in order to allow the OST striping to be set.
3034 This is user-contributed code and not supported by CFS.
3039 Description: extended inode attributes (immutable, append-only) work improperly
3040 when 2.4 and 2.6 kernels are used on client/server or vice versa
3041 Details : Introduce kernel-independent values for these flags.
3043 Severity : enhancement
3046 Description: Allow fractional MB tunings for lustre in /proc/ filesystem.
3047 Details : Many of the /proc/ tunables can only be tuned at a megabyte
3048 granularity. Now, Fractional MB granularity is be supported,
3049 this is very useful for low memory system.
3051 Severity : enhancement
3053 Description: Getattr by fid
3054 Details : Getting a file attributes by its fid, obtaining UPDATE|LOOKUP
3055 locks, avoids extra getattr rpc requests to MDS, allows '/' to
3056 have locks and avoids getattr rpc requests for it on every stat.
3059 Frequency : Always, for filesystems larger than 2TB
3061 Description: ldiskfs crash at mount for filesystem larger than 2TB with mballoc
3062 Details : Kenrel kmalloc limits allocations to 128kB and this prevents
3063 filesystems larger than 2TB to be mounted with mballoc enabled.
3066 Frequency : Always, for 32-bit kernel without CONFIG_LBD and filesystem > 2TB
3068 Description: ldiskfs crash at mount for filesystem larger than 2TB with mballoc
3069 Details : If a 32-bit kernel is compiled without CONFIG_LBD enabled and a
3070 filesystems larger than 2TB is mounted then the kernel will
3071 silently corrupt the start of the filesystem. CONFIG_LBD is
3072 enabled for all CFS-supported kernels, but the possibility of
3073 this happening with a modified kernel config exists.
3075 Severity : enhancement
3077 Description: add client O_DIRECT support for 2.6 kernels
3078 Details : It is now possible to do O_DIRECT reads and writes to files
3079 in the Lustre client mountpoint on 2.6 kernel clients.
3081 Severity : enhancement
3083 Description: parallel glimpse, setattr, statfs, punch, destroy requests
3084 Details : Sends glimpse, setattr, statfs, punch, destroy requests to OSTs in
3085 parallel, not waiting for response from every OST before sending
3086 a rpc to the next OST.
3091 Description: setattr vs write race when updating file timestamps
3092 Details : Client processes that update a file timestamp into the past
3093 right after writing to the file (e.g. tar) it is possible that
3094 the updated file modification time can be reset to the current
3095 time due to a race between processing the setattr and write RPC.
3097 Severity : enhancement
3099 Description: Bring 'lfs find' closer in line with regular Linux find.
3100 Details : lfs find util supports -atime, -mtime, -ctime, -maxdepth, -print,
3101 -print0 options and obtains all the needed info through the lustre
3104 Severity : enhancement
3106 Description: support up to 1024 configured devices on one node
3107 Details : change obd_dev array from statically allocated to dynamically
3108 allocated structs as they are first used to reduce memory usage
3113 Description: Flush dirty partially truncated pages during truncate
3114 Details : Immediatelly flush partially truncated pages in filter_setattr,
3115 this way we completely avoid having any pages in page cache on OST
3116 and can retire ugly workarounds during writes to flush such pages.
3121 Description: i_sem vs transaction deadlock in mds_obd_destroy during unlink.
3122 Details : protect inode from truncation within vfs_unlink() context
3123 just take a reference before calling vfs_unlink() and release it
3124 when parent's i_sem is free.
3129 Description: last_id value checked outside lock on OST caused LASSERT failure
3130 Details : If there were multiple MDS->OST object precreate requests in
3131 flight, it was possible that the OST's last object id was checked
3132 outside a lock and incorrectly tripped an assertion. Move checks
3133 inside locks, and discard old precreate requests.
3136 Frequency : always, if extents are used on OSTs
3138 Description: index ei_leaf_hi (48-bit extension) is not zeroed in extent index
3139 Details : OSTs using the extents format would not zero the high 16 bits of
3140 the index physical block number. This is not a problem for any
3141 OST filesystems smaller than 16TB, and no kernels support ext3
3142 filesystems larger than 16TB yet. This is fixed in 1.4.7 (all
3143 new/modified files) and can be fixed for existing filesystems
3144 with e2fsprogs-1.39-cfs1.
3149 Description: import connection selection may be incorrect if timer wraps
3150 Details : Using a 32-bit jiffies timer with HZ=1000 may cause backup
3151 import connections to be ignored if the 32-bit jiffies counter
3152 wraps. Use a 64-bit jiffies counter.
3155 Frequency : very large clusters immediately after boot
3157 Description: LNET request buffers exhausted under heavy short-term load
3158 Details : If a large number of client requests are generated on a service
3159 that has previously never seen so many requests it is possible
3160 that the request buffer growth cannot keep up with the spike in
3161 demand. Instead of dropping incoming requests, they are held in
3162 the LND until the RPC service can accept more requests.
3165 Frequency : Sometimes during replay
3167 Description: Assertion failure in ll_local_open after replay.
3168 Details : If replay happened on an open request reply before we were able
3169 to set replay handler, reply will become not swabbed tripping the
3170 assertion in ll_local_open. Now we set the handler right after
3171 recognising of open request
3174 Frequency : very rare
3176 Description: kernel reports "badness in vsnprintf"
3177 Details : Reading from the "recovery_status" /proc file in small chunks
3178 may cause a negative length in lprocfs_obd_rd_recovery_status()
3179 call to vsnprintf() (which is otherwise harmless). Exit early
3180 if there is no more space in the output buffer.
3182 Severity : enhancement
3184 Description: clear OBD RPC statistics by writing to them
3185 Details : It is now possible to clear the OBD RPC statistics by writing
3186 to the "stats" file.
3191 Description: Client mtime is not the same on different clients after utimes
3192 Details : In some cases, the client was using the utimes() syscall on
3193 a file cached on another node. The clients now validate the
3194 ctime from the MDS + OSTs to determine which one is right.
3199 Description: Inability to activate failout mode
3200 Details : lconf script incorrectly assumed that in pythong string's numeric
3201 value is used in comparisons.
3204 Frequency : always with multiple stripes per file
3206 Description: Inefficient object allocation for mutli-stripe files
3207 Details : When selecting which OSTs to stripe files over, for files with
3208 a stripe count that divides evenly into the number of OSTs,
3209 the MDS is always picking the same starting OST for each file.
3210 Return the OST selection heuristic to the original design.
3215 Description: mount failures may take full timeout to return an error
3216 Details : Under some heavy load conditions it is possible that a
3217 failed mount can wait for the full obd_timeout interval,
3218 possibly several minutes, before reporting an error.
3219 Instead return an error as soon as the status is known.
3221 Frequency : quota enabled and large files being deleted
3223 Description: releasing more than 4GB of quota at once hangs OST
3224 Details : If a user deletes more than 4GB of files on a single OST it
3225 will cause the OST to spin in an infinite loop. Release
3226 quota in < 4GB chunks, or use a 64-bit value for 1.4.7.1+.
3231 Description: statfs data retrieved from /proc may be stale or zero
3232 Details : When reading per-device statfs data from /proc, in the
3233 {kbytes,files}_{total,free,avail} files, it may appear
3234 as zero or be out of date.
3237 Frequency : systems with MD RAID1 external journal devices
3239 Description: lconf's call to blkid is confused by RAID1 journal devices
3240 Details : Use the "blkid -l" flag to locate the MD RAID device instead
3241 of returning all block devices that match the journal UUID.
3244 Frequency : always, for aggregate stripe size over 4GB
3246 Description: assertion fails when trying to use 4GB stripe size
3247 Details : Use "setstripe" to set stripe size over 4GB will fail the kernel,
3248 complaining "ASSERTION(lsm->lsm_xfersize != 0)"
3251 Frequency : always on ppc64
3253 Description: the first write on an ext3 filesystem with mballoc got stuck
3254 Details : ext3_mb_generate_buddy() uses find_next_bit() which does not
3255 perform endianness conversion.
3257 ------------------------------------------------------------------------------
3259 02-14-2006 Cluster File Systems, Inc. <info@clusterfs.com>
3261 * WIRE PROTOCOL CHANGE. This version of Lustre networking WILL NOT
3262 INTEROPERATE with older versions automatically. Please read the
3263 user documentation before upgrading any part of a live system.
3264 * WARNING: Lustre networking configuration changes are required with
3265 this release. See https://bugzilla.clusterfs.com/show_bug.cgi?id=10052
3268 * Support for kernels:
3269 2.6.9-22.0.2.EL (RHEL 4)
3270 2.6.5-7.244 (SLES 9)
3271 2.6.12.6 vanilla (kernel.org)
3274 Severity : enhancement
3275 Bugzilla : 7981/8208
3276 Description: Introduced Lustre Networking (LNET)
3277 Details : LNET is new networking infrastructure for Lustre, it includes
3278 a reorganized network configuration mode (see the user
3279 documentation for full details) as well as support for routing
3280 between different network fabrics. Lustre Networking Devices
3281 (LNDs) for the supported network fabrics have also been
3282 created for this new infrastructure.
3284 Severity : enhancement
3285 Description: Introduced Access control lists
3286 Details : clients can set ACLs on files and directories in order to have
3287 more fine-grained permissions than the standard Unix UGO+RWX.
3288 The MDS must be started with the "-o acl" mount option.
3290 Severity : enhancement
3291 Description: Introduced filesystem quotas
3292 Details : Administrators may now establish per-user quotas on the
3295 Severity : enhancement
3297 Description: Configuration change for the XT3
3298 The PTLLND is now used to run Lustre over Portals on the XT3
3299 The configure option(s) --with-cray-portals are no longer used.
3300 Rather --with-portals=<path-to-portals-includes> is used to
3301 enable building on the XT3. In addition to enable XT3 specific
3302 features the option --enable-cray-xt3 must be used.
3307 Description: Running on many-way SMP OSTs can trigger oops in llcd_send()
3308 Details : A race between allocating a new llcd and re-getting the llcd_lock
3309 allowed another thread to grab newly-allocated llcd.
3311 Severity : enhancement
3313 Description: 2.6 OST async journal commit and locking fix to improve performance
3314 Details : The filter_direct_io()+filter_commitrw_write() journal commits for
3315 2.6 kernels are now async as they already were in 2.4 kernels so
3316 that they can commit concurrently with the network bulk transfer.
3317 For block-allocated files the filter allocation semaphore is held
3318 to avoid filesystem fragmentation during allocation. BKL lock
3319 removed for 2.6 xattr operations where it is no longer needed.
3324 Description: lconf incorrectly determined whether two IP networks could talk
3325 Details : In some more complicated routing and multiple-network
3326 configurations, lconf will avoid trying to make a network
3327 connection to a disjoint part of the IP space. It was doing the
3328 math incorrectly for one set of cases.
3333 Description: Fix for potential infinite loop processing records in an llog.
3334 Details : If an llog record is corrupted/zeroed, it is possible to loop
3335 forever in llog_process(). Validate the llog record length
3336 and skip the remainder of the block on error.
3339 Frequency : occasional (liblustre only)
3341 Description: liblustre could not open files whose last component is a symlink
3342 Details : sysio_path_walk() would incorrectly pass the open intent to
3343 intermediate path components.
3346 Frequency : rare (liblustre only with non-standard tuning)
3347 Bugzilla : 7201 (7350)
3348 Description: Tuning the MDC DLM LRU size to zero triggers client LASSERT
3349 Details : llu_lookup_finish_locks() tries to set lock data on a lock
3350 after it has been released, only do this for referenced locks
3352 Severity : enhancement
3354 Description: specifying an (invalid) directory default stripe_size of -1
3355 would reset the directory default striping
3356 Details : stripe_size -1 was used internally to signal directory stripe
3357 removal, now use "all default" to signal dir stripe removal
3358 as a directory striping of "all default" is not useful
3361 Frequency : common for large clusters running liblustre clients
3363 Description: doing an ls when liblustre clients are running is slow
3364 Details : sending a glimpse AST to a liblustre client waits for every AST
3365 to time out, as liblustre clients will not respond. Since they
3366 cannot cache data we refresh the OST lock LVB from disk instead.
3368 Severity : enhancement
3370 Description: doing an ls at the same time as file IO can be slow
3371 Details : enqueue and other "small" requests can be blocked behind many
3372 large IO requests. Create a new OST IO portal for non-IO
3373 requests so they can be processed faster.
3376 Frequency : rare (only HPUX clients mounting unsupported re-exported NFS vol)
3378 Description: an HPUX NFS client would get -EACCESS when ftruncate()ing a newly
3379 created file with mode 000
3380 Details : the Linux NFS server relies on an MDS_OPEN_OWNEROVERRIDE hack to
3381 allow an ftruncate() as a non-root user to a file with mode 000.
3382 Lustre now respects this flag to disable mode checks when
3383 truncating a file owned by the user
3386 Frequency : liblustre-only, when liblustre client dies unexpectedly or becomes
3389 Description: Revoking locks from clients that went dead or catatonic might take
3391 Details : New lock flags FL_CANCEL_ON_BLOCK used by liblustre makes
3392 cancellation of such locks instant on servers without waiting for
3393 any reply from clients. Clients drops these locks when cancel
3394 notification from server is received without replying.
3397 Frequency : liblustre-only, when liblustre client dies or becomes busy
3399 Description: Doing ls on Linux clients can take a long time with active
3401 Details : Liblustre client cannot handle ASTs in timely manner, so avoid
3402 granting such locks to it in the first place if possible. Locks
3403 are taken by proxy on the OST during the read or write and
3404 dropped immediately afterward. Add connect flags handling, do
3405 not grant locks to liblustre clients for glimpse ASTs.
3407 Severity : enhancement
3409 Description: Improve read-ahead algorithm to avoid excessive IO for random reads
3410 Details : Existing read-ahead algorithm is tuned for the case of streamlined
3411 sequential reads and behaves badly with applications doing random
3412 reads. Improve it by reading ahead at least read region, and
3413 avoiding excessive large RPC for small reads.
3415 Severity : enhancement
3417 Description: Creating more than 1000 files for a single job may cause a load
3418 imbalance on the OSTs if there are also a large number of OSTs.
3419 Details : qos_prep_create() uses an OST index reseed value that is an
3420 even multiple of the number of available OSTs so that if the
3421 reseed happens in the middle of the object allocation it will
3422 still utilize the OSTs as uniformly as possible.
3427 Description: OST or MDS may oops in ping_evictor_main()
3428 Details : ping_evictor_main() drops obd_dev_lock if deleting a stale export
3429 but doesn't restart at beginning of obd_exports_timed list
3432 Severity : enhancement
3434 Description: improve by-nid export eviction on the MDS and OST
3435 Details : allow multiple exports with the same NID to be evicted at one
3436 time without re-searching the exports list.
3439 Frequency : rare, only with supplementary groups enabled on SMP 2.6 kernels
3441 Description: MDS may oops in groups_free()
3442 Details : in rare race conditions a newly allocated group_info struct is
3443 freed again, and this can be NULL. The 2.4 compatibility code
3444 for groups_free() checked for a NULL pointer, but 2.6 did not.
3447 Frequency : common for liblustre clients doing little filesystem IO
3448 Bugzilla : 9352, 7313
3449 Description: server may evict liblustre clients accessing contended locks
3450 Details : if a client is granted a lock or receives a completion AST
3451 with a blocking AST already set it would not reply to the AST
3452 for LDLM_FL_CANCEL_ON_BLOCK locks. It now replies to such ASTs.
3455 Frequency : lfs setstripe, only systems with more than 160 OSTs
3457 Description: unable to set striping with a starting offset beyond OST 160
3458 Details : llapi_create_file() incorrectly limited the starting stripe
3459 index to the maximum single-file stripe count.
3462 Frequency : LDAP users only
3464 Description: lconf did not handle in-kernel recovery with LDAP properly
3465 Details : lconf/LustreDB get_refs() is searching the wrong namespace
3467 Severity : enhancement
3469 Description: bind OST threads to NUMA nodes to improve performance
3470 Details : all OST threads are uniformly bound to CPUs on a single NUMA
3471 node and do their allocations there to localize memory access
3473 Severity : enhancement
3475 Description: llmount can determine client NID directly from Myrinet (GM)
3476 Details : the client NID code from gmnalnid was moved directly into
3477 llmount, removing the need to use this or specifying the
3478 client NID explicitly when mounting GM clients with zeroconf
3481 Frequency : if client is started with down MDS
3483 Description: if client is started with down MDS mount hangs in ptlrpc_queue_wait
3484 Details : Having an LWI_INTR() wait event (interruptible, but no timeout)
3485 will wait indefinitely in ptlrpc_queue_wait->l_wait_event() after
3486 ptlrpc_import_delayed_req() because we didn't check if the
3487 request was interrupted, and we also didn't break out of the
3488 event loop if there was no timeout
3493 Description: data loss during non-page-aligned writes to a single file from
3494 both multiple nodes and multiple threads on one node at same time
3495 Details : updates to KMS and lsm weren't protected by common lock. Resulting
3496 inconsistency led to false short-reads, that were cached and later
3497 used by ->prepare_write() to fill in partially written page,
3498 leading to data loss.
3501 Frequency : always, if lconf --abort_recovery used
3503 Description: lconf --abort_recovery fails with 'Operation not supported'
3504 Details : lconf was attempting to abort recovery on the MDT device and not
3507 Severity : enhancement
3509 Description: remove cleanup logs
3510 Details : replace lconf-generated cleanup logs with lustre internal
3511 cleanup routines. Eliminates the need for client-cleanup and
3514 Severity : enhancement
3516 Description: add support for EAs (user and system) on lustre filesystems
3517 Details : it is now possible to store extended attributes in the Lustre
3518 client filesystem, and with the user_xattr mount option it
3519 is possible to allow users to store EAs on their files also
3521 Severity : enhancement
3523 Description: Add possibility (config option) to show minimal available OST free
3525 Details : When compiled with --enable-mindf configure option, statfs(2)
3526 (and so, df) will return least minimal free space available from
3527 all OSTs as amount of free space on FS, instead of summary of
3528 free spaces of all OSTs.
3530 Severity : enhancement
3532 Description: do not expand extent locks acquired on OST-side
3533 Details : Modify ldlm_extent_policy() to not expand local locks, acquired
3534 by server: they are not cached anyway.
3537 Frequency : when mmap is used/binaries executed from Lustre
3539 Description: Unmmap pages before throwing them away from read cache.
3540 Details : llap_shrink cache now attempts to unmap pages before discarding
3541 them (if unmapping failed - do not discard). SLES9 kernel has
3542 extra checks that trigger if this unmapping is not done first.
3547 Description: lconf didn't resolve symlinks before checking to see whether a
3548 given mountpoint was already in use
3551 Frequency : when migrating failover services
3552 Bugzilla : 6395, 9514
3553 Description: When migrating a subset of services from a node (e.g. failback
3554 from a failover service node) the remaining services would
3555 time out and evict clients.
3556 Details : lconf --force (implied by --failover) sets the global obd_timeout
3557 to 5 seconds in order to quickly disconnect, but this caused
3558 other RPCs to time out too quickly. Do not change the global
3559 obd_timeout for force cleanup, only set it for DISCONNECT RPCs.
3561 Severity : enhancement
3562 Frequency : if MDS is started with down OST
3563 Bugzilla : 9439,5706
3564 Description: Allow startup/shutdown of an MDS without depending on the
3565 availability of the OSTs.
3566 Details : Asynchronously call mds_lov_synchronize during MDS startup.
3567 Add appropriate locking and lov-osc refcounts for safe
3568 cleaning. Add osc abort_inflight calls in case the
3569 synchronize never started.
3572 Frequency : occasional (Cray XT3 only)
3574 Description: root not authorized to access files in CRAY_PORTALS environment
3575 Details : The client process capabilities were not honoured on the MDS in
3576 a CRAY_PORTALS/CRAY_XT3 environment. If the file had previously
3577 been accessed by an authorized user then root was able to access
3578 the file on the local client also. The root user capabilities
3579 are now allowed on the MDS, as this environment has secure UID.
3582 Frequency : occasional
3584 Description: ldiskfs "too long searching" message happens too often
3585 Details : A debugging message (otherwise harmless) prints too often on
3586 the OST console. This has been reduced to only happen when
3587 there are fragmentation problems on the filesystem.
3592 Description: Division by zero in statfs when all OSCs are inactive
3593 Details : lov_get_stripecnt() returns zero due to incorrect order of checks,
3594 lov_statfs divides by value returned by lov_get_stripecnt().
3598 Bugzilla : 9489, 3273
3599 Description: First write from each client to each OST was only 4kB in size,
3600 to initialize client writeback cache, which caused sub-optimal
3601 RPCs and poor layout on disk for the first writen file.
3602 Details : Clients now request an initial cache grant at (re)connect time
3603 and so that they can start streaming writes to the cache right
3604 away and always do full-sized RPCs if there is enough data.
3605 If the OST is rebooted the client also re-establishes its grant
3606 so that client cached writes will be honoured under the grant.
3611 Description: Slow ls (and stat(2) syscall) on files residing on IO-loaded OSTs
3612 Details : Now I/O RPCs go to different portal number and (presumably) fast
3613 lock requests (and glimses) and other RPCs get their own service
3614 threads pool that should be able to service those RPCs
3617 Severity : enhancement
3619 Description: Ability to exchange lustre version between client and servers and
3620 issue warnings at client side if client is too old. Also for
3621 liblustre clients there is ability to refuse connection of too old
3623 Details : New 'version' field is added to connect data structure that is
3624 filled with version info. That info is later checked by server and
3628 Frequency : rare, liblustre only.
3629 Bugzilla : 9296, 9581
3630 Description: Two simultaneous writes from liblustre at offset within same page
3631 might proceed at the same time overwriting eachother with stale
3633 Details : I/O lock withing llu_file_prwv was released too early, before data
3634 actually was hitting the wire. Extended lock-holding time until
3635 server acknowledges receiving data.
3638 Frequency : extremely rare. Never observed in practice.
3640 Description: avoid generating lustre_handle cookie of 0.
3641 Details : class_handle_hash() generates handle cookies by incrementing
3642 global counter, and can hit 0 occasionaly (this is unlikely, but
3643 not impossible, because initial value of cookie counter is
3644 selected randonly). Value of 0 is used as a sentinel meaning
3645 "unassigned handle" --- avoid it. Also coalesce two critical
3646 sections in this function into one.
3648 Severity : enhancement
3650 Description: allow liblustre clients to delegate truncate locking to OST
3651 Details : To avoid overhead of locking, liblustre client instructs OST to
3652 take extent lock in ost_punch() on client's behalf. New connection
3653 flag is added to handle backward compatibility.
3655 Severity : enhancement
3656 Bugzilla : 4928, 7341, 9758
3657 Description: allow number of OST service threads to be specified
3658 Details : a module parameter allows the number of OST service threads
3659 to be specified via "options ost ost_num_threads={N}" in the
3660 OSS's /etc/modules.conf or /etc/modprobe.conf.
3664 Bugzilla : 6146, 9635, 9895
3665 Description: servers crash with bad pointer in target_handle_connect()
3666 Details : In rare cases when a client is reconnecting it was possible that
3667 the connection request was the last reference for that export.
3668 We would temporarily drop the export reference and get a new
3669 one, but this may have been the last reference and the export
3670 was just destroyed. Get new reference before dropping old one.
3672 Severity : enhancement
3673 Frequency : if client is started with failover MDS
3675 Description: Allow multiple MDS hostnames in the mount command
3676 Details : Try to read the configuration from all specified MDS
3677 hostnames during a client mount in case the "primary"
3680 Severity : enhancement
3682 Description: Stop sending data to evicted clients as soon as possible.
3683 Details : Check if the client we are about to send or are sending data to
3684 was evicted already. (Check is done every second of waiting,
3685 for which l_wait_event interface was extended to allow checking
3686 of exit condition at specified intervals).
3689 Frequency : rare, normally only when NFS exporting is done from client
3691 Description: 'bad disk LOV MAGIC: 0x00000000' error when chown'ing files
3693 Details : Make mds_get_md() recognise empty md case and set lmm size to 0.
3696 Frequency : always, if srand() is called before liblustre initialization
3698 Description: Liblustre uses system PRNG disturbing its usage by user application
3699 Details : Introduce internal to lustre fast and high-quality PRNG for
3700 lustre usage and make liblustre and some other places in generic
3701 lustre code to use it.
3703 Severity : enhancement
3704 Bugzilla : 9477, 9557, 9870
3705 Description: Verify that the MDS configuration logs are updated when xml is
3706 Details : Check if the .xml configuration logs are newer than the config
3707 logs stored on the MDS and report an error if this is the case.
3708 Request --write-conf, or allow starting with --old_conf.
3710 Severity : enhancement
3712 Description: Handle symlinks in the path when checking if Lustre is mounted.
3713 Details : Resolve intermediate symlinks when checking if a client has
3714 mounted a filesystem to avoid duplicate client mounts.
3719 Description: lconf can hit an error exception but still return success.
3720 Details : The lconf command catches the Command error exception at the top
3721 level script context and will exit with the associated exit
3722 status, but doesn't ensure that this exit status is non-zero.
3727 Description: failure of ptlrpc thread startup can cause oops
3728 Details : Starting a ptlrpc service thread can fail if there are a large
3729 number of threads or the server memory is very fragmented.
3730 Handle this without oopsing.
3733 Frequency : always, only if liblustre and non-default acceptor port was used
3735 Description: liblustre cannot connect to servers with non-default acceptor port
3736 Details : tcpnal_set_default_params() was not called and was therefore
3737 ignoring the environment varaible TCPNAL_PORT, as well as other
3738 TCPNAL_ environment variables
3743 Description: two objects could be created on the same OST for a single file
3744 Details : If an OST is down, in some cases it was possible to create two
3745 objects on a single OST for a single file. No problems other
3746 than potential performance impact and spurious error messages.
3750 Bugzilla : 5681, 9562
3751 Description: Client may oops in ll_unhash_aliases
3752 Details : Client dcache may become inconsistent in race condition.
3753 In some cases "getcwd" can fail if the current directory is
3759 Description: Inode refcounting problems in NFS export code
3760 Details : link_raw functions used to call d_instantiate without obtaining
3761 extra inode reference first.
3765 Bugzilla : 9942, 9903
3766 Description: Referencing freed requests leading to crash, memleaks with NFS.
3767 Details : We used to require that call to ll_revalidate_it was always
3768 followed by ll_lookup_it. Also with revalidate_special() it is
3769 possible to call ll_revalidate_it() twice for the same dentry
3770 even if first occurence returned success. This fix changes semantic
3771 between DISP_ENQ_COMPLETE disposition flag to mean there is extra
3772 reference on a request referred from the intent.
3773 ll_intent_release() then releases such a request.
3776 Frequency : rare, normally benchmark loads only
3778 Description: unlinked inodes were kept in memory on the client
3779 Details : If a client is repeatedly creating and unlinking files it
3780 can accumulate a lot of stale inodes in the inode slab cache.
3781 If there is no other client load running this can cause the
3782 client node to run out of memory. Instead flush old inodes
3783 from client cache that have the same inode number as a new inode.
3786 Frequency : SLES9 2.6.5 kernel and long filenames only
3787 Bugzilla : 9969, 10379
3788 Description: utime reports stale NFS file handle
3789 Details : SLES9 uses out-of-dentry names in some cases, which confused
3790 the lustre dentry revalidation. Change it to always use the
3794 Frequency : rare, unless heavy write-truncate concurrency is continuous
3795 Bugzilla : 4180, 6984, 7171, 9963, 9331
3796 Description: OST becomes very slow and/or deadlocked during object unlink
3797 Details : filter_destroy() was holding onto the parent directory lock
3798 while truncating+unlinking objects. For very large objects this
3799 may block other threads for a long time and slow overall OST
3800 responsiveness. It may also be possible to get a lock ordering
3801 deadlock in this case, or run out of journal credits because of
3802 the combined truncate+unlink. Solution is to do object truncate
3803 first in one transaction without parent lock, and then do the
3804 final unlink in a new transaction with the parent lock. This
3805 reduces the lock hold time dramatically.
3808 Frequency : rare, 2.4 kernels only
3810 Description: MDS or OST cleanup may trip kernel BUG when dropping kernel lock
3811 Details : mds_cleanup() and filter_cleanup() need to drop the kernel lock
3812 before unmounting their filesystem in order to avoid deadlock.
3813 The kernel_locked() function in 2.4 kernels only checks whether
3814 the kernel lock is held, not whether it is this process that is
3815 holding it as 2.6 kernels do.
3820 Description: MDS or OST may oops/LBUG if a client is connecting multiple times
3821 Details : The client ptlrpc code may be trying to reconnect to a down
3822 server before a previous connection attempt has timed out.
3823 Increase the reconnect interval to be longer than the connection
3824 timeout interval to avoid sending duplicate connections to
3828 Frequency : echo_client brw_test command
3830 Description: fix echo_client to work with OST preallocated code
3831 Details : OST preallocation code (5137) didn't take echo_client IO path
3832 into account: echo_client calls filter methods outside of any
3833 OST thread and, hence, there is no per-thread preallocated
3834 pages and buffers to use. Solution: hijack pga pages for IO. As
3835 a byproduct, this avoids unnecessary data copying.
3839 Bugzilla : 3555, 5962, 6025, 6155, 6296, 9574
3840 Description: Client can oops in mdc_commit_close() after open replay
3841 Details : It was possible for the MDS to return an open request with no
3842 transaction number in mds_finish_transno() if the client was
3843 evicted, but without actually returning an error. Clients
3844 would later try to replay that open and may trip an assertion
3845 Simplify the client close codepath, and always return an error
3846 from the MDS in case the open is not successful.
3849 Frequency : rare, 2.6 OSTs only
3851 Description: OST may deadlock under high load on fragmented files
3852 Details : If there was a heavy load and highly-fragmented OST filesystems
3853 it was possible to have all the OST threads deadlock waiting on
3854 allocation of biovecs, because the biovecs were not released
3855 until the entire RPC IO was completed. Instead, release biovecs
3856 as soon as they are complete to ensure forward IO progress.
3858 Severity : enhancement
3860 Description: Support for specifying external journal device at mount
3861 Details : If an OST or MDS device is formatted with an external journal
3862 device, this device major/minor is stored in the ext3 superblock
3863 and may not be valid for failover. Allow detecting and
3864 specifying the external journal at mount time.
3869 Description: Mounting an MDS with pending unlinked files may cause oops
3870 Details : target_finish_recovery() calls mds_postrecov() which returned
3871 the number of orphans unlinked. mds_lov_connect->mds_postsetup()
3872 considers this an error and immediately begins cleaning up the
3873 lov, just after starting the mds_lov process
3875 Severity : enhancement
3877 Description: Implement 'lfs df' to report actual free space on per-OST basis
3878 Details : Add sub-command 'df' on 'lfs' to report the disk space usage of
3879 MDS/OSDs. Usage: lfs df [-i][-h]. Command Options: '-i' to report
3880 usage of objects; '-h' to report in human readable format.
3882 ------------------------------------------------------------------------------
3884 08-26-2005 Cluster File Systems, Inc. <info@clusterfs.com>
3891 Description: Mounting an ldiskfs file system with mballoc may crash OST node.
3892 Details : ldiskfs mballoc code may reference an uninitialized buddy struct
3893 at startup during orphan unlinking. Instead, skip buddy update
3894 before setup, as it will be regenerated after recovery is complete.
3899 Description: If an OST is inactive, its locks might reference stale inodes.
3900 Details : lov_change_cbdata() must iterate over all namespaces, even if
3901 they are inactive to clear inode references from the lock.
3903 Severity : enhancement
3904 Frequency : occasional, if non-standard max_dirty_mb used
3906 Description: Client will block write RPCs if not enough grant
3907 Details : If a client has max_dirty_mb smaller than max_rpcs_in_flight,
3908 then the client will block writes while waiting for another RPC
3909 to complete instead of consuming its dirty limit. With change
3910 we get improved performance when max_dirty_mb is small.
3912 Severity : enhancement
3913 Bugzilla : 3389, 6253
3914 Description: Add support for supplementary groups on the MDS.
3915 Details : The MDS has an upcall /proc/fs/lustre/mds/{mds}/group_upcall
3916 (set to /usr/sbin/l_getgroups if enabled) which will do MDS-side
3917 lookups for user supplementary groups into a cache.
3921 Description: O_CREAT|O_EXCL open flags in liblustre always return -EEXIST
3922 Details : Make libsysio to not enforce O_EXCL by clearing the flag,
3923 for liblustre O_EXCL is enforced by MDS.
3927 Description: readdir never returns NULL in liblustre.
3928 Details : Corrected llu_iop_getdirentries logic, to return offset of next
3929 dentry in struct dirent.
3933 Frequency : liblustre only, depends on application IO pattern
3934 Description: liblustre clients evicted if not contacting servers
3935 Details : Don't put liblustre clients into the ping_evictor list, so
3936 they will not be evicted by the pinger ever.
3938 Severity : enhancement
3940 Description: Add ability to evict clients by NID from MDS.
3941 Details : By echoing "nid:$NID" string into
3942 /proc/fs/lustre/mds/.../evict_client client with nid that equals to
3943 $NID would be instantly evicted from this MDS and from all active
3944 OSTs connected to it.
3948 Description: Do not query file size twice, somewhat slowing stat(2) calls.
3949 Details : lookup_it_finish() used to query file size from OSTs that was not
3954 Description: service threads change working directory to that of init
3955 Details : Starting lustre service threads may pin the working directory
3956 of the parent thread, making that filesystem busy. Threads
3957 now change to the working directory of init to avoid this.
3961 Frequency : during shutdown only
3962 Description: shutdown with a failed MDS or OST can cause unmount to hang
3963 Details : Don't resend DISCONNECT messages in ptlrpc_disconnect_import()
3968 Frequency : 2.6 only
3969 Description: chmod/chown may include an extra supplementary group
3970 Details : ll{,u}_mdc_pack_op_data() does not properly initialize the
3971 supplementary group and if none is specified this is used.
3974 Bugzilla : 5479 (6816)
3976 Description: Racing open + rm can assert client in mdc_set_open_replay_data()
3977 Details : If lookup is in progress on a file that is unlinked we might try
3978 to revalidate the inode and fail in revalidate after lookup is
3979 complete and ll_file_open() enqueues the open again but
3980 it_open_error() was not checking DISP_OPEN_OPEN errors correctly.
3983 Frequency : always, if lconf --abort_recovery used
3985 Description: lconf --abort_recovery fails with 'Operation not supported'
3986 Details : lconf was attempting to abort recovery on the MDT device and not
3989 ------------------------------------------------------------------------------
3991 2005-08-08 Cluster File Systems, Inc. <info@clusterfs.com>
3996 Frequency : rare (only unsupported configurations with a node running as an
3998 Bugzilla : 6514, 5137
3999 Description: Mounting a Lustre file system on a node running as an OST could
4001 Details : OSTs now preallocates memory needed to write out data at
4002 startup, instead of when needed, to avoid having to
4003 allocate memory in possibly low memory situations.
4004 Specifically, if the file system is mounted on on OST,
4005 memory pressure could force it to try to write out data,
4006 which it needed to allocate memory to do. Due to the low
4007 memory, it would be unable to do so and the node would
4008 become unresponsive.
4010 Severity : enhancement
4012 Description: Addition of lconf --service command line option
4013 Details : lconf now accepts a '--service <arg>' option, which is
4014 shorthand for 'lconf --group <arg> --select <arg>=<hostname>'
4016 Severity : enhancement
4018 Description: Failover mode is now the default for OSTs.
4019 Details : By default, OSTs will now run in failover mode. To return to
4020 the old behaviour, add '--failout' to the lmc line for OSTs.
4022 Severity : enhancement
4024 Description: Health checks are now provided for MDS and OSTs
4025 Details : Additional detailed health check information on MSD and OSTs
4026 is now provided through the procfs health_check value.
4029 Frequency : occasional, depends on IO load
4031 Description: Disk fragmentation on the OSTs could eventually cause slowdowns
4032 after numerous create/delete cycles
4033 Details : The ext3 inode allocation policy would not allocate new inodes
4034 very well on the OSTs because there are no new directories
4035 being created. Instead we look for groups with free space if
4036 the parent directories are nearly full.
4041 Description: Network or server problems during mount may cause partially
4042 mounted clients instead of returning an error.
4043 Details : The config llog parsing code may overwrite the error return
4044 code during mount error handling, returning success instead
4050 Description: MDS can fail to allocate large reply buffers
4051 Details : After long uptimes the MDS can fail to allocate large reply
4052 buffers (e.g. zconf client mount config records) due to memory
4053 fragmentation or consumption by the buffer cache. Preallocate
4054 some large reply buffers so that these replies can be sent even
4055 under memory pressure.
4059 Frequency : rare (liblustre)
4060 Description: fsx running with liblustre complained that using truncate() to
4061 extend the file doesn't work. This patch corrects that issue.
4062 Details : This is the liblustre equivalent of the fix for bug 6196. Fixes
4063 ATTR_SIZE and lsm use in llu_setattr_raw.
4067 Frequency : rare, only 2.6 kernels
4068 Description: Unusual file access patterns on the MDS may result in inode
4069 data being lost in very rare circumstances.
4070 Details : Bad interaction between the ea-in-inode patch and the "no-read"
4071 code in the 2.6 kernel caused the inode and/or EA data not to
4072 be read from disk, causing single-file corruption.
4076 Frequency : rare, only 2.6 filesystems using extents
4077 Description: Heavy concurrent write and delete load may cause data corruption.
4078 Details : It was possible under high-load situations to have an extent
4079 metadata block in the block device cache from a just-unlinked
4080 file overwrite a newly-allocated data block. We now unmap any
4081 metadata buffers that alias just-allocated data blocks.
4085 Frequency : filesystems with default stripe_count larger than 77
4086 Description: lconf+mke2fs fail when formatting filesystem with > 77 stripes
4087 Details : lconf specifies an inode size of 4096 bytes when the default
4088 stripe_count is larger than 77. This conflicts with the default
4089 inode density of 1 per 4096 bytes. Allocate smaller inodes in
4090 this case to avoid pinning too much memory for large EAs.
4092 ------------------------------------------------------------------------------
4094 2005-07-07 Cluster File Systems, Inc. <info@clusterfs.com>
4099 Frequency : rare (extremely heavy IO load with hundreds of clients)
4101 Description: Client is evicted, gets IO error writing to file
4102 Details : lock ordering changes for bug 5492 reintroduced bug 3267 and
4103 caused clients to be evicted for AST timeouts. The fixes in
4104 bug 5192 mean we no longer need to have such short AST timeouts
4105 so ldlm_timeout has been increased.
4108 Frequency : occasional during --force or --failover shutdown under load
4109 Bugzilla : 5949, 4834
4110 Description: Server oops/LBUG if stopped with --force or --failover under load
4111 Details : a collection of import/export refcount and cleanup ordering
4112 issues fixed for safer force cleanup
4115 Frequency : only filesystems larger than 120 OSTs
4116 Bugzilla : 5990, 6223
4117 Description: lfs getstripe would oops on a very large filesystem
4118 Details : lov_getconfig used kfree on vmalloc'd memory
4121 Frequency : only filesystems exporting via NFS to Solaris 10 clients
4122 Bugzilla : 6242, 6243
4123 Description: reading from files that had been truncated to a non-zero size
4124 but never opened returned no data
4125 Details : ll_file_read() reads zeros from no-object files to EOF
4130 Description: A bug in MDS/OSS recovery could cause the OSS to fail an assertion
4131 Details : There's little harm in aborting MDS/OSS recovery and letting it
4132 try again, so I removed the LASSERT and return an error instead.
4134 Severity : enhancement
4136 Description: New debugging infrastructure for tracking down data corruption
4137 Details : The I/O checksum code was replaced to: (a) control it at runtime,
4138 (b) cover more of the client-side code path, and (c) try to narrow
4139 down where problems occurred
4143 Bugzilla : 3819, 4364, 4397, 6313
4144 Description: Racing close and eviction MDS could cause assertion in mds_close
4145 Details : It was possible to get multiple mfd references during close and
4146 client eviction, leading to one thread referencing a freed mfd.
4148 Severity: : enhancement
4149 Bugzilla : 3262, 6359
4150 Description: Attempts to reconnect to servers are now more aggressive.
4151 Details : This builds on the enhanced upcall-less recovery that was added
4152 in 1.4.2. When trying to reconnect to servers, clients will
4153 now try each server in the failover group every 10 seconds. By
4154 default, clients would previously try one server every 25 seconds.
4159 Description: After recovery, certain operations trigger a failed
4160 assertion on a client.
4161 Details : Failing over an mds, using lconf -d --failover, while a
4162 client was doing a readdir() call would cause the client to
4163 LBUG after recovery completed and the readdir() was resent.
4165 Severity : enhancement
4167 Description: Default groups are now added by lconf
4168 Details : You can now run lconf --group <servicename> without having to
4169 manually add groups with lmc.
4172 Frequency : occasional
4174 Description: Nodes with an elan id of 0 trigger a failed assertion
4177 Frequency : always when accessing e.g. tty/console device nodes
4179 Description: tty and some other devices nodes cannot be used on lustre
4180 Details : file's private_data field is used by device data and lustre
4181 values in there got lost. New field was added to struct file to
4182 store fs-specific private data.
4185 Frequency : when exporting Lustre via NFS
4187 Description: NFSD failed occasionally when looking up a path component
4188 Details : NFSD is looking up ".." which was broken in ext3 directories
4189 that had grown large enough to become hashed.
4192 Frequency : Clusters with multiple interfaces not on the same subnet
4194 Description: Nodes will repeatedly try to reconnect to an interface which it
4195 cannot reach and report an error to the log.
4196 Details : Extra peer list entries will be created by lconf with some peers
4197 unreachable. lconf now validates the peer before adding it.
4200 Frequency : Only if a default stripe is set on the filesystem root.
4202 Description: Setting a default stripe on the filesystem root prevented the
4203 filesystem from being remounted.
4204 Details : The client was sending extra request flags in the root getattr
4205 request and did not allocate a reply buffer for the dir EA.
4208 Frequency : occasional, higher if lots of files are accessed by one client
4209 Bugzilla : 6159, 6097
4210 Description: Client trips assertion regarding lsm mismatch/magic
4211 Details : While revalidating inodes the VFS looks up inodes with ifind()
4212 and in rare cases can find an inode that is being freed.
4213 The ll_test_inode() code will free the lsm during ifind()
4214 when it finds an existing inode and then the VFS later attaches
4215 this free lsm to a new inode.
4219 Bugzilla : 6422, 7030
4220 Description: MDS deadlock between mkdir and client eviction
4221 Details : Creating a new file via mkdir or mknod (starting a transaction
4222 and getting the ns lock) can deadlock with client eviction
4223 (gets ns lock and trying to finish a synchronous transaction).
4226 Frequency : occasional
4227 Description: While starting a server, the fsfilt_ext3 module could not be
4229 Details : CFS's improved ext3 filesystem is named ldiskfs for 2.6
4230 kernels. Previously, lconf would still use the ext3 name
4231 when trying to load modules. Now, it will correctly use
4232 ext3 on 2.4 and ldiskfs on 2.6.
4234 Severity : enhancement
4235 Description: The default stripe count has been changed to 1
4236 Details : The interpretation of the default stripe count (0, to lfs
4237 or lmc) has been changed to mean striping across a single
4238 OST, rather than all available. For general usage we have
4239 found a stripe count of 1 or 2 works best.
4241 Severity : enhancement
4242 Description: Add support for compiling against Cray portals.
4243 Details : Conditional compiling for some areas that are different
4247 Frequency : occasional
4248 Bugzilla : 6409, 6834
4249 Description: Creating files with an explicit stripe count may lead to
4250 a failed assertion on the MDS
4251 Details : If some OSTs are full or unavailable, creating files may
4252 trigger a failed assertion on the MDS. Now, Lustre will
4253 try to use other servers or return an error to the
4257 Frequency : occasional
4259 Description: Multiple concurrent overlapping read+write on multiple SMP nodes
4260 caused lock timeout during readahead (since 1.4.2).
4261 Details : Processes doing readahead might match a lock that hasn't been
4262 granted yet if there are overlapping and conflicting lock
4263 requests. The readahead process waits on ungranted lock
4264 (original lock is CBPENDING), while OST waits for that process
4265 to cancel CBPENDING read lock and eventually evicts client.
4267 Severity : enhancement
4269 Description: Initial enabling of flock support for clients
4270 Details : Implements fcntl advisory locking and file status functions.
4271 This feature is provided as an optional mount flag (default
4272 off), and is NOT CURRENTLY SUPPORTED. Not all types of record
4273 locking are implemented yet, and those that are are not guaranteed
4274 to be completely correct in production environments.
4275 mount -t lustre -o [flock|noflock] ...
4278 Frequency : occasional
4280 Description: OSTs running 2.4 kernels but with extents enabled might trip an
4281 assertion in the ext3 JBD (journaling) layer.
4282 Details : The b_committed_data struct is protected by the big kernel lock
4283 in 2.4 kernels, serializing journal_commit_transaction() and
4284 ext3_get_block_handle->ext3_new_block->find_next_usable_block()
4285 access to this struct. In 2.6 kernels there is finer grained
4286 locking to improve SMP performance of the JBD layer.
4290 Description: Changes the "SCSI I/O Stats" kernel patch to default to "enabled"
4295 Description: merge and cleanup kernel patches.
4298 -----------------------------------------------------------------------------
4300 2005-05-05 Cluster File Systems, Inc. <info@clusterfs.com>
4302 NOTE: Lustre 1.4.2 uses an incompatible network protocol than previous
4303 versions of Lustre. Please update all servers and clients to
4304 version 1.4.2 or later at the same time. You must also run
4305 "lconf --write-conf {config}.xml" on the MDS while it is stopped
4306 to update the configuration logs.
4308 - fix for HPUX NFS client breakage when NFS exporting Lustre (5781)
4309 - mdc_enqueue does not need max_mds_easize request buffer on send (5707)
4310 - swab llog records of type '0' so we get proper header size/idx (5861)
4311 - send llog cancel req to DLM cancel portal instead of cb portal (5515)
4312 - fix rename of one directory over another leaking an inode (5953)
4313 - avoid SetPageDirty on 2.6 (5981)
4314 - don't re-add just-being-destroyed locks to the waiting list (5653)
4315 - when creating new directories, inherit the parent's custom
4316 striping settings if present parent (3048)
4317 - flush buffers from cache before direct IO in 2.6 obdfilter (4982)
4318 - don't hold i_size_sem in ll_nopage() and ll_ap_refresh_count (6077)
4319 - don't hold client locks on temporary worklist from l_lru (5666)
4320 - handle IO errors in 2.6 obdfilter bio completion routine (6046)
4321 - automatically evict dead clients (5921)
4322 - Update file size properly in create+truncate+fstat case (6196)
4323 - Do not unhash mountpoint dentries, do not allow removal of
4325 - Avoid lock ordering deadlock issue with write/truncate (6203,5654)
4326 - reserve enough journal credits in fsfilt_start_log for setattr (4554)
4327 - ldlm_enqueue freed-export error path would always LBUG (6149,6184)
4328 - don't reference lr_lvb_data until after we hold lr_lvb_sem (6170)
4329 - don't overwrite last_rcvd if there is a *_client_add() error (6086)
4330 - Correctly handle reads of files with no objects (6243)
4331 - lctl recover will also mark a device active if deactivate used (5933)
4333 - by default create 1 inode per 4kB space on MDS, per 16kB on OSTs
4334 - allow --write-conf on an MDS with different nettype than client (5619)
4335 - don't write config llogs to MDS for mounts not from that MDS (5617)
4336 - lconf should create multiple TCP connections from a client (5201)
4337 - init scripts are now turned off by default; run chkconfig --on
4338 lustre and chkconfig --on lustrefs to use them
4339 - upcalls are no longer needed for clients to recover to failover
4341 - add --abort-recovery option to lconf to abort recovery on device
4343 - add support for an arbitrary number of OSTs (3026)
4344 - Quota support protocol changes.
4345 - forward compatibility changes to wire structs (6007)
4346 - rmmod NALs that might be loaded because of /etc/modules.conf (6133)
4347 - support for mountfsoptions and clientoptions to the Lustre LDAP (5873)
4348 - improved "lustre status" script
4349 - initialize blocksize for non-regular files (6062)
4350 - added --disable-server and --disable-client configure options (5782)
4351 - introduce a lookup cache for lconf to avoid repeated DB scans (6204)
4352 - Vanilla 2.4.29 support
4353 - increase maximum number of obd devices to 520 (6242)
4354 - remove the tcp-zero-copy patch from the suse-2.4 series (5902)
4355 - Quadrics Elan drivers are now included for the RHEL 3 2.4.21 and
4356 SLES 9 2.6.5 kernels
4357 - limit stripes per file to 160 (the maximum EA size) (6093)
4359 2005-03-22 Cluster File Systems, Inc. <info@clusterfs.com>
4362 - don't LASSERT in ll_release on NULL lld with NFS export (4655, 5760)
4363 - hold NS lock when calling handle_ast_error->del_waiting_lock (5746)
4364 - fix setattr mtime regression from lovcleanup merge (4829, 5669)
4365 - workaround for 2.6 crash in ll_unhash_aliases (5687, 5210)
4366 - small ext3 extents cleanups and fixes (5733)
4367 - improved mballoc code, several small races and bugs fixed (5733, 5638)
4368 - kernel version 43 - fix remove_suid bugs in both 2.4 and 2.6 (5695)
4369 - avoid needless client->OST connect, fix handle mismatch (5317)
4370 - fix DLM error path that led to out-of-sync client, long delays (5779)
4371 - support common vfs-enforced mount options (nodev,nosuid,noexec) (5637)
4372 - fix several locking issues related to i_size (5492,5624,5654,5672)
4373 - don't move pending lock onto export if it is already evicted (5683)
4374 - fix kernel oops when creating .foo in unlinked directory (5548)
4375 - fix deadlock in obdfilter statistics vs. object create (5811)
4376 - use time_{before,after} to avoid timer jiffies wrap (5882)
4377 - shutdown --force/--failover stability (3607,3651,4797,5203,4834)
4378 - Do not leak request if server was not able to process it (5154)
4379 - If mds_open unable to find parent dir, make that negative lookup(5154)
4380 - don't create new directories with extent-mapping (5909, 5936)
4382 - fix lustre/lustrefs init scripts for SuSE (patch from Scali, 5702)
4383 - don't hold the pinger_sem in ptlrpc_pinger_sending_on_import
4384 - change obd_increase_kms to obd_adjust_kms (up or down) (5654)
4385 - lconf, lmc search both /usr/lib and /usr/lib64 for Python libs (5800)
4386 - support for RHEL4 kernel on i686 (5773)
4387 - provide error messages when incompatible logs are encountered (5898)
4389 2005-02-18 Cluster File Systems, Inc. <info@clusterfs.com>
4390 * version 1.4.0.10 (1.4.1 release candidate 1)
4392 - don't keep a lock reference when lock is not granted (4238)
4393 - unsafe list practices (rarely) led to infinite eviction loop (4908)
4394 - add per-fs limit of Lustre pages in page cache, avoid OOM (4699)
4395 - drop import inflight refcount on signal_completed_replay error (5255)
4396 - unlock page after async write error during send (3677)
4397 - handle missing objects in filter_preprw_read properly (5265)
4398 - no transno return for symlink open, don't save no-trasno open (3440)
4399 - don't try to complete elan receive that already failed (4012)
4400 - free RPC server reply state on error (5406)
4401 - clean up thread from ptlrpc_start_thread() on error (5160)
4402 - readahead could read extra page into cache that wasn't ejected (5388)
4403 - prevent races in class_attach/setup/cleanup/detach (5260)
4404 - don't dereference de->d_inode after l_dput of de (5458)
4405 - use "int" for stripe value returned from lock_to_stripe (5544)
4406 - mballoc allocation and error-checking fixes in 2.6 (5504)
4407 - block device patches to fix I/O request sizes in 2.6 (5482)
4408 - look up hostnames for IB nals (5602)
4409 - 2.6 changed lock ordering of 2 semaphores, caused deadlock (5654)
4410 - don't start multiple acceptors for the same port (5277)
4411 - fix incorrect LASSERT in mds_getattr_name (5635)
4412 - export a proc file for general "ping" checking (5628)
4413 - fix "lfs check" to not block when the MDS is down (5628)
4415 - service request history (4965)
4416 - put {ll,lov,osc}_async_page structs in a single slab (4699)
4417 - create an "evict_client" /proc entry on OSTs, like the MDS has
4418 - fix mount usage message, return errors per mount(8) (5168)
4419 - change grep [] to grep "[]" in tests so they work in more UMLs
4420 - fix ppc64/x86_64 spec to use %{_libdir} instead of /usr/lib (5389)
4421 - remove ancient LOV_MAGIC_V0 EA support (5047)
4422 - add "disk I/Os in flight" and "I/O req time" stats in obdfilter
4423 - align r/w RPCs to PTLRPC_MAX_BRW_SIZE boundary for performance (3451)
4424 - allow readahead allocations to fail when low on memory (5383)
4425 - mmap locking landed again, after considerable improvement (2828)
4426 - add get_hostaddr() to lustreDB.py for LDAP support (5459)
4428 2004-11-23 Cluster File Systems, Inc. <info@clusterfs.com>
4431 - send OST transaction number in read/write reply to free req (4966)
4432 - don't ASSERT in ptl_send_rpc() if we run out of memory (5119)
4433 - lock /proc/sys/portals/routes internal state, avoiding oops (4827)
4434 - the watchdog thread now runs as interruptible (5246)
4435 - flock/lockf fixes (but it's still disabled, pending 5135)
4436 - don't use EXT3 constants in llite code (5094)
4437 - memory shortage at startup could cause assertion (5176)
4439 - reorganization of lov code
4440 - single portals codebase
4442 - add extents/mballoc support (5025)
4443 - direct I/O reads in the obdfilter (4048)
4444 - kernel patches from LNXI for 2.6 (bluesmoke, perfctr, mtd, kexec)
4446 tbd Cluster File Systems, Inc. <info@clusterfs.com>
4449 - send OST transaction number in read/write reply to free req (4966)
4450 - don't ASSERT in ptl_send_rpc() if we run out of memory (5119)
4451 - lock /proc/sys/portals/routes internal state, avoiding oops (4827)
4452 - the watchdog thread now runs as interruptible (5246)
4453 - handle missing objects in filter_preprw_read properly (5265)
4454 - unsafe list practices (rarely) led to infinite eviction loop (4908)
4455 - drop import inflight refcount on signal_completed_replay error (5255)
4456 - unlock page after async write error during send (3677)
4457 - return original error code on reconstructed replies (3761)
4458 - no transno return for symlink open, don't save no-trasno open (3440)
4460 - add pid to ldlm debugging output (4922)
4461 - bump the watchdog timeouts -- we can't handle 30sec yet
4462 - extra debugging for orphan dentry/inode bug (5259)
4464 2004-11-16 Cluster File Systems, Inc. <info@clusterfs.com>
4467 - fix TCP_NODELAY bug, which caused extreme perf regression (5134)
4468 - allocate qswnal tx descriptors singly to avoid fragmentation (4504)
4469 - don't LBUG on obdo_alloc() failure, use OBD_SLAB_ALLOC() (4800)
4470 - fix NULL dereference in /proc/sys/portals/routes (4827)
4471 - allow failed mdc_close() operations to be interrupted (4561)
4472 - stop precreate on OST before MDS would time out on it (4778)
4473 - don't send partial-page writes before EOF from client (4410)
4474 - discard client grant for sub-page writes on large-page clients (4520)
4475 - don't free dentries not owned by NFS code, check generation (4806)
4476 - fix lsm leak if mds_create_objects() fails (4801)
4477 - limit debug_daemon file size, always print CERROR messages (4789)
4478 - use transno after validating reply (3892)
4479 - process timed out requests if import state changes (3754)
4480 - update mtime on OST during writes, return in glimpse (4829)
4481 - add mkfsoptions to LDAP (4679)
4482 - use ->max_readahead method instead of zapping global ra (5039)
4483 - don't interrupt __l_wait_event() during strace
4485 - add software watchdogs to catch hung threads quickly (4941)
4486 - make lustrefs init script start after nfs is mounted
4487 - fix CWARN/ERROR duplication (4930)
4488 - return async write errors to application if possible (2248)
4489 - add /proc/sys/portal/memused (bytes allocated by PORTALS_ALLOC)
4490 - print NAL number in %x format (4645)
4491 - update barely-supported suse-2.4.21-171 series (4842)
4492 - support for sles 9 %post scripts
4493 - support for building 2.6 kernel-source packages
4494 - support for sles km_* packages
4496 2004-10-07 Cluster File Systems, Inc. <info@clusterfs.com>
4499 - ignore -ENOENT errors in osc_destroy (3639)
4500 - notify osc create thread that OSC is being cleaned up (4600)
4501 - add nettype argument for llmount in #5d in conf-sanity.sh (3936)
4502 - reconstruct ost_handle() like mds_handle() (4657)
4503 - create a new thread to do import eviction to avoid deadlock (3969)
4504 - let lconf resolve symlinked-to devices (4629)
4505 - don't unlink "objects" from directory with default EA (4554)
4506 - hold socknal file ref over connect in case target is down (4394)
4507 - allow more than 32000 subdirectories in a single directory (3244)
4508 - fix blocks count for O_DIRECT writes (3751)
4509 - OST returns ENOSPC from object create when no space left (4539)
4510 - don't send truncate RPC if file size isn't changing (4410)
4511 - limit OSC precreate to 1/2 of value OST considers bogus (4778)
4512 - bind to privileged port in socknal and tcpnal (3689)
4514 - rate limit CERROR/CWARN console message to avoid overload (4519)
4515 - GETFILEINFO dir ioctl returns LOV EA + MDS stat in 1 call (3327)
4516 - basic mmap support (3918)
4517 - kernel patch series update from b1_4 (4711)
4519 2004-09-16 Cluster File Systems, Inc. <info@clusterfs.com>
4522 - avoid crash during MDS cleanup with OST shut down (2775)
4523 - fix loi_list_lock/oig_lock inversion on interrupted IO (4136)
4524 - don't use bad inodes on the MDS (3744)
4525 - dynamic object preallocation to improve recovery speed (4236)
4526 - don't hold spinlock over lock dumping or change debug flags (4401)
4527 - don't zero obd_dev when it is force cleaned (3651)
4528 - print grants to console if they go negative (4431)
4529 - "lctl deactivate" will stop automatic recovery attempts (3406)
4530 - look for existing locks in ldlm_handle_enqueue() (3764)
4531 - don't resolve lock handle twice in recovery avoiding race (4401)
4532 - revalidate should check working dir is a directory (4134)
4534 - don't always mark "slow" obdfilter messages as errors (4418)
4536 2004-08-24 Cluster File Systems, Inc. <info@clusterfs.com>
4539 - don't close LustreDB during write_conf until it is done (3860)
4540 - fix typo in lconf for_each_profile (3821)
4541 - allow dumping logs from multiple threads at one time (3820)
4542 - don't allow multiple threads in OSC recovery (3812)
4543 - fix debug_size parameters (3864)
4544 - fix mds_postrecov to initialize import for llog ctxt (3121)
4545 - replace config semaphore with spinlock (3306)
4546 - be sure to send a reply for a CANCEL rpc with bad export (3863)
4547 - don't allow enqueue to complete on a destroyed export (3822)
4548 - down write_lock before checking llog header bitmap (3825)
4549 - recover from lock replay timeout (3764)
4550 - up llog sem before sending rpc (3652)
4551 - reduce ns lock hold times when setting kms (3267)
4552 - change a dlm LBUG to LASSERTF, to maybe learn something (4228)
4553 - fix NULL deref and obd_dev leak on setup error (3312)
4554 - replace some LBUG about llog ops with error handling (3841)
4555 - don't match INVALID dentries from d_lookup and spin (3784)
4556 - hold dcache_lock while marking dentries INVALID and hashing (4255)
4557 - fix invalid assertion in ptlrpc_set_wait (3880)
4559 - add libwrap support for the TCP acceptor (3996)
4560 - add /proc/sys/portals/routes for non-root route listing (3994)
4561 - allow setting MDS UUID in .xml (2580)
4562 - print the stack of a process that LBUGs (4228)
4564 2004-07-14 Cluster File Systems, Inc. <info@clusterfs.com>
4567 - don't cleanup request in ll_file_open() on failed MDS open (3430)
4568 - make sure to unset replay flag from failed open requests (3440)
4569 - if default stripe count is 0, use OST count for inode size (3636)
4570 - update parent mtime/ctime on client for create/unlink (2611)
4571 - drop dentry ref in ext3_add_link from open_connect_dentry (3266)
4572 - free recovery state on server during a forced cleanup (3571)
4573 - unregister_reply for resent reqs (3063)
4574 - loop back devices mounting and status check on 2.6 (3563)
4575 - fix resource-creation race that can provoke i_size == 0 (3513)
4576 - don't try to use bad inodes returned from MDS/OST fs lookup (3688)
4577 - more debugging for page-accounting assertion (3746)
4578 - return -ENOENT instead of asserting if ost getattr+unlink race (3558)
4579 - avoid deadlock after precreation failure (3758)
4580 - fix race and lock order deadlock in orphan handling (3450, 3750)
4581 - add validity checks when grabbing inodes from l_ast_data (3599)
4583 - add /proc/.../recovery_status to obdfilter (3428)
4584 - lightweight CDEBUG infrastructure, debug daemon (3668)
4585 - change default OSC RPC parameters to be better on small clusters
4586 - turn off OST read cache for files smaller than 32MB
4587 - install man pages and include them in rpms (3100)
4588 - add new init script for (un)mounting lustre filesystems (2593)
4589 - run chkconfig in %post for init scripts (3701)
4590 - drop scimac NAL (unmaintained)
4592 2004-06-17 Cluster File Systems, Inc. <info@clusterfs.com>
4595 - clean kiobufs before and after use (3485)
4596 - strip trailing '/'s before comparing paths with /proc/mounts (3486)
4597 - remove assertions to work around "in-flight rpcs" recovery bug (3063)
4598 - change init script to fail more clearly if not run as root (1528)
4599 - allow clients to reconnect during replay (1742)
4600 - fix ns_lock/i_sem lock ordering deadlock for kms update (3477)
4601 - don't do DNS lookups on NIDs too small for IP addresses (3442)
4602 - re-awaken ptlrpcd if new requests arrive during check_set (3554)
4603 - fix cond_resched (3554)
4604 - only evict unfinished clients after recovery (3515)
4605 - allow bulk resend, prevent data loss (3570)
4606 - dynamic ptlrpc request buffer allocation (2102)
4607 - don't allow unlinking open directory if it isn't empty (2904)
4608 - set MDS/OST threads to umask 0 to not clobber client modes (3359)
4609 - remove extraneous obd dereference causing LASSERT failure (3334)
4610 - don't use get_cycles() when creating temp. files on the mds (3156)
4611 - hold i_sem when setting i_size in ll_extent_lock() (3564)
4612 - handle EEXIST for set-stripe, set proper directory name (3336)
4614 - servers can dump a log evicting a client - lustre.dump_on_timeout=1
4615 - fix ksocknal_fmb_callback() error messages (2918)
4617 2004-05-27 Cluster File Systems, Inc. <info@clusterfs.com>
4620 - don't copy lvb into (possibly NULL) reply on error (2983)
4621 - don't deref dentry after dput, don't free lvb on error (2922)
4622 - use the kms to determine writeback rpc length (2947)
4623 - increment oti_logcookies when osc is inactive (2948)
4624 - update client's i_blocks count via lvb messages (2543)
4625 - handle intent open/close of special files properly (1557)
4626 - mount MDS with errors=remount-ro, like obdfilter (2009)
4627 - initialize lock handle to avoid ASSERT on error cleanup (3057)
4628 - don't use cancelling-locks' kms values (2947)
4629 - use highest lock extent for kms, not last one (2925)
4630 - don't dereference ERR_PTR() dentry in error handling path (3107)
4631 - fix thread race in portals_debug_dumplog() (3122)
4632 - create lprocfs device entries at setup instead of at attach (1519)
4633 - common AST error handler, don't evict client on completion race (3145)
4634 - zero nameidata in detach_mnt in 2.6 (3118)
4635 - verify d_inode after revalidate_special is valid in 2.6 (3116)
4636 - use lustre_put_super() to handle zconf unmounts in 2.6 (3064)
4637 - initialize RPC timeout timer earlier for 2.6 (3219)
4638 - don't dereference NULL reply buffer if mdc_close was never sent (2410)
4639 - print nal/nid for unknown nid (3258)
4640 - additional checks for oscc recovery before doing precreate (3284)
4641 - fix ll_extent_lock() error return code for 64-bit systems (3043)
4642 - don't crash in mdc_close for bad permissions on open (3285)
4643 - zero i_rdev for non-device files (3147)
4644 - clear page->private before handing to FS, better assertion (3119)
4645 - tune the read pipeline (3236)
4646 - fix incorrect decref of invalidated dentry (2350)
4647 - provide read-ahead stats and refine rpc in flight stats (3328)
4648 - don't hold journal transaction open across create RPC (3313)
4649 - update atime on MDS at close time (3265)
4650 - close LDAP connection when recovering to avoid server load (3315)
4651 - update iopen-2.6 patch with fixes from 2399,2517,2904 (3301)
4652 - don't leak open file on MDS after open resend (3325)
4653 - serialize filter_precreate and filter_destroy_precreated (3329)
4654 - loop device shouldn't call sync_dev() for nul device (3092)
4655 - clear page cache after eviction (2766)
4656 - resynchronize MDS->OST in background (2824)
4657 - refuse to mount the same filesystem twice on same mountpoint (3394)
4658 - allow llmount to create routes for mounting behind routers (3320)
4659 - push lock cancellation to blocking thread for glimpse ASTs (3409)
4660 - don't call osc_set_data_with_check() for TEST_LOCK matches (3159)
4661 - fix rare problem with rename on htree directories (3417)
4663 - allow default OST striping configuration per directory (1414)
4664 - fix compilation for qswnal for 2.6 kernels (3125)
4665 - increase maximum number of MDS request buffers for large systems
4666 - change liblustreapi to be useful for external progs like lfsck (3098)
4667 - increase local configuration timeout for slow disks (3353)
4668 - allow configuring ldlm AST timeout - lustre.ldlm_timeout=<seconds>
4670 2004-03-22 Cluster File Systems, Inc. <info@clusterfs.com>
4673 - fixes for glimpse AST timeouts / incorrectly 0-sized files (2818)
4674 - don't overwrite extent policy data in reply if lock was blocked (2901)
4675 - drop filter export grants atomically with removal from device (2663)
4676 - del obd_self_export from work_list in class_disconnect_exports (2908)
4677 - don't LBUG if MDS recovery times out during orphan cleanup (2530)
4678 - swab reply message in mdc_close, other PPC fixes (2464)
4679 - fix destroying of named logs (2325)
4680 - overwrite old logs when running lconf --write_conf (2264)
4681 - bump LLOG_CHUNKSIZE to 8k to allow for larger clusters (2306)
4682 - fix race in target_handle_connect (2898)
4683 - mds_reint_create() should take same inode create lock (2926)
4684 - correct journal credits calculated for CANCEL_UNLINK_LOG (2931)
4685 - don't close files for self_export to avoid uninitialized obd (2936)
4686 - allow MDS with the same name as client node (2939)
4687 - hold dentry reference for closed log files for unlink (2325)
4688 - reserve space for all logs during transactions (2059)
4689 - don't evict page beyond end of stripe extent (2925)
4690 - don't oops on a deleted current working directory (2399)
4691 - handle hard links to targets without a parent properly (2517)
4692 - don't dereference NULL lock when racing during eviction (2867)
4693 - don't grow lock extents when lots of conflicting locks (2919)
4695 2004-03-04 Cluster File Systems, Inc. <info@clusterfs.com>
4698 - account for cache space usage on clients to avoid data loss (974)
4699 - lfsck support in lustre kernel code (2349)
4700 - reduce journal credits needed for BRW writes (2370)
4701 - orphan handling to avoid losing space on client/server crashes
4702 - ptlrpcd can be blocked, stopping ALL progress (2477)
4703 - use lock value blocks to assist in proper KMS, faster stat (1021)
4704 - takes i_sem instead of DLM locks internally on obdfilter (2720)
4705 - recovery for initial connections (2355)
4706 - fixes for mds_cleanup_orphans (1934)
4707 - abort_recovery crashes MDS in b_eq (mds_unlink_orphan) (2584)
4708 - block all file creations until orphan recovery completes (1901)
4709 - client remove rq_connection from request struct (2423)
4710 - conf-sanity test_5, proper cleanup in umount log not availale (2640)
4711 - recovery timer race (2670)
4712 - mdc_close recovey bug (2532)
4713 - ptlrpc cleanup bug (2710)
4714 - mds timeout on local locks (2588)
4715 - namespace lock held during RPCs (2431)
4716 - handle interrupted sync write properly (2503)
4717 - don't try to handle a message that hasn't been replied to (2699)
4718 - client assert failure during cleanup after abort recovery (2701)
4719 - leak mdc device after failed mount (2712)
4720 - ptlrpc_check_set allows timedout requests to complete (2714)
4721 - wait for inflight reqs when ptlrpcd finishes (2710)
4722 - make sure unregistered services are removed from the srv_list
4723 - reset bulk XID's when resending them (caught by 1138 test)
4724 - unregister_bulk after timeout
4725 - fix lconf error (2694)
4726 - handle write after unfinished setstripe, stripe-only getstripe (2388)
4727 - readahead locks pages, leaves pending causing memory pressure (2673)
4728 - increase OST request buffers to 4096 on large machines (2729)
4729 - fix up permission of existing directories in simple_mkdir (2661)
4730 - init deleted item, add assertions ptlrpc_abort_inflight() (2725)
4731 - don't assign transno to errored transactions (2742)
4732 - don't delete objects on OST if given a bogus objid from MDS (2751)
4733 - handle large client PAGE_SIZE readdir on small PAGE_SIZE MDS (2777)
4734 - if rq_no_resend, then timeout request after recovery (2432)
4735 - fix MDS llog_logid record size, 64-bit array alignment (2733)
4736 - don't call usermode_helper from ptlrpcd, DEFAULT upcall (2773)
4737 - put magic in mount.lustre data, check for bad/NULL mount data (2529)
4738 - MDS recovery shouldn't delete objects that it has given out (2730)
4739 - if enqueue arrives after completion, don't clobber LVB (2819)
4740 - don't unlock pages twice when trigger_group_io returns error (2814)
4741 - don't deref NULL rq_repmsg if ldlm_handle_enqueue failed (2822)
4742 - don't write pages to disk if there was an error (1450)
4743 - don't ping imports that have recovery disabled (2676)
4744 - take buffered bytes into account when balancing socknal conn (2817)
4745 - hold a DLM lock over readdir always, use truncate_inode_pages (2706)
4746 - reconnect unlink llog connection after MDS reconnects to OST (2816)
4747 - remove little-endian swabbing of llog records (1987)
4748 - set/limit i_blksize to LL_MAX_BLKSIZE on client (2884)
4749 - retry reposting request buffers if they fail (1191)
4750 - grow extent at grant time to avoid granting a revoked lock (2809)
4751 - lock revoke doesn't evict page if covered by a second lock (2765)
4752 - disable VM readahead to avoid reading outside lock extents (2805)
4754 - return LL_SUPER_MAGIC from statfs for the filesystem type (1972)
4755 - updated kernel patches for hp-2.4.20 kernel (2681)
4757 2004-02-07 Cluster File Systems, Inc. <info@clusterfs.com>
4760 - fix truncated write corruption (2366)
4761 - fix for failed assertion in iopen_connect_dentry (1792,2517)
4763 - don't flag the ptlrpcd thread with PF_MEMALLOC (2636)
4764 - ensure len(uuid) < 37 in lmc (1171)
4765 - fix ia64 OOPS in llog_test (2255)
4766 - zero end of page at obdfilter for partial page writes (2648)
4767 - don't leave stale dentries around after renames (bug 2428)
4768 - fix timeouts when evicting a client with a single lock held (2642)
4769 - set deadline for the initial HELLO message to drain (2634)
4770 - print out dotted-quad IP addresses in the socknal (2302)
4772 - additional debugging for MDS client eviction problem (2443)
4773 - fix mkfsoptions support for osts (2603, 2604)
4775 2004-01-27 Cluster File Systems, Inc. <info@clusterfs.com>
4778 - add series for the vanilla 2.6.0 kernel
4779 - add series for the vanilla 2.4.24 kernel
4780 - add series for a cray x86/64 UL kernel drop
4781 - fix xattr patches for the vanilla 2.4.19 series
4783 - generate true UUIDs in lmc (1171)
4784 - have portals stack dumping break in UML (2466)
4785 - avoid bad dchild deref; avoid inum lock w/o creation (2362)
4786 - allocate with _NOFS in ldlm to avoid deadlock (1933)
4787 - wake callback waiting threads on client eviction (2460)
4788 - Add --ptldebug and --subsystem to lmc (1719)
4789 - update assertion to allow safe interrupt allocation
4790 - set rq_no_resend for cancel requests (2432)
4791 - recalculate ptlrpcd timeout after resend (2494)
4792 - call vfs_rmdir when removing pending directories (2368)
4793 - fix renaming a file to itself (2429)
4794 - lmc creates a default one-stripe lov (2454)
4795 - expand procfs space to handle large clusters (2326)
4796 - increase UML stack to avoid overflow
4797 - update lconf's list of debug and subsystem masks
4798 - fix lfs find --obd (2510)
4799 - /proc tunable for disabling filter read caching (2591)
4800 - stop rpm packages from altering slapd.conf (2301)
4801 - disable nagle in the socknal under 0conf (2578)
4802 - choose mds inode size based on stripe count (2572)
4803 - fix kernel-source rpm problems (2516)
4805 - add --disable-doc to avoid pdf generation (2421)
4806 - update documentation, tests, type-os, comments
4807 - avoid format warnings on ia64
4808 - remove the TOE NAL
4809 - tiny code cleanups by removing unused fields
4811 2004-01-07 Cluster File Systems, Inc. <info@clusterfs.com>
4814 - fix obvious semaphore misuse in as-yet-unused setattr path (2348)
4815 - remove the most blatant lies from BUILDING file (2371)
4816 - change default debug level to reasonable production setting
4817 - reduce client side cache size to reduce cache flush time
4818 - reduce max RPCs in flight to avoid unnecessary file fragmentation
4819 - make TCP zerocopy and pinger support enabled by default (2476)
4820 - sync writes completed after process exits caused crashes (2319)
4821 - maintain correct mount count on the MDS (2356)
4822 - backout 1557, because 2316 wasn't really fixed
4823 - better file I/O statistics gathering in /proc
4824 - don't take unnecessary, deadlock-inducing bug in readpage (2383)
4825 - another kernel patch to fix zero-copy TCP function export
4826 - don't take duplicate lock when processing re-sent getattr (2420)
4827 - lctl uses obd_self_export instead of creating new conn (2353)
4828 - MDS/OST recovery case which requires object creation asserted (2425)
4829 - move lfs from /usr/sbin to /usr/bin in packages
4830 - fix race between mds_client_add and mds_client_free (2417)
4831 - use kmalloc instead of slabs in portals (2430)
4832 - don't create duplicate records when a failover MDS is present (2442)
4833 - remove unnecessary mount age check (2332)
4834 - don't remove directory inodes from locks prematurely (2451)
4835 - don't break if MDS service name is the same as hostname (2103)
4836 - fix races in client write RPC generation when cache full (2482)
4838 2003-12-13 Cluster File Systems, Inc. <info@clusterfs.com>
4841 - remove now-unused request->rq_obd (278)
4842 - if an allocation fails, print out how much memory we've used (1933)
4843 - use PORTAL_SLAB_ALLOC for structures, to get GFP_MEMALLOC (1933)
4844 - add the "configurable stack size" patch to most series files (1256)
4845 - ability to write large log records, for 100+ OST configs (2306)
4846 - fix NULL deref when filter_prep fails (2314)
4847 - fix operator precedence error in filter_sync
4848 - dynamic allocation of socknal TX descriptors (2315)
4849 - fix a missed case in the GFP_MEMALLOC patch, can cause deadlock (2310)
4850 - fix gcc 2.96 compilation problem in xattr kernel patch (2294)
4851 - ensure that CWARN messages in Portals always get to the syslog
4852 - __init/__exit are not for prototype decls (ldlm_init/exit)
4853 - x86-64 compile warning fixes
4854 - fix gateway LMC keyword conflict (2318)
4855 - fix MDS lock inversions in getattr/reint paths (1844)
4856 - fix a rare lock re-ordering bug, which caused deadlock (2322)
4857 - fix i_sem/journal inversion in fsfilt_ext3_write_record (2306)
4858 - DLM race condition prevented some lock evictions (2328)
4859 - ENOMEM detection and retry on socknal sends (2230)
4860 - use GFP_NOFS throughout Lustre, to combat ENOMEM (2230)
4861 - move osc_rpcd into ptlrpc, for use in MDC and others (2329)
4862 - protect MDS inode fsdata with stronger locking; fixes assertion (2313)
4863 - better error messages when a client is rejected during recovery (1505)
4864 - avoid cancelling locks which were never granted, after failure (2330)
4865 - fix i_sem/journal inversion in mds_client_add (2333)
4866 - fix truncate/getattr lock cycle deadlock (2334)
4867 - use rpcd to send close; allows resend after timeout, avoid leak (1897)
4868 - fix two rare exit paths which could leak an l_lock() ref (2321)
4869 - fencepost error in MDS/OST orphan recovery (2226)
4870 - make log record alignment 8 bytes (1988)
4871 - lstripe now fails when requested offset > ost_count (2237)
4872 - ensure that all kernel series have a complete list.h (1607)
4873 - fix crashes in special-file operations (2316)
4874 - lctl create/brw OID mismatch, caused by obsolete filter loop (2339)
4876 - allow configurable automake binary, for testing new versions
4877 - small update to the lfs documentation
4879 2003-12-03 Cluster File Systems, Inc. <info@clusterfs.com>
4881 * fix negative export reference count in fsfilt_sync (2312)
4883 2003-12-01 Cluster File Systems, Inc. <info@clusterfs.com>
4884 * release candidate 0.9.1
4886 - orphans are moved into the PENDING directory for possible recovery
4887 - replayed opens now open by fid for orphan/rename safety (1042)
4888 - last close of an orphan inode generates a transno (683)
4889 - chdir() and mount() now pin the directory entry (1020)
4890 - avoid CERROR in normal ll_setattr_raw() error case (1500)
4891 - discard very old requests without processing them (1502)
4892 - remove some common, well-understood CERRORs (1505)
4893 - require O_DIRECT I/O to be page-sized to workaround IA64 crash (1609)
4894 - clear "grant" flags in OST replies until OST grant code lands (1644)
4895 - fix read performance by not clobbering i_blksize on client (1598)
4896 - fix __ldlm_handle2lock oops by not dereferencing lock after PUT (1625)
4897 - make LRU size a /proc tunable, clears locks when reduced (707)
4898 - fix some lprocfs rot that prevented ptlbd from loading (1732)
4899 - server locks take references on exports now (1558)
4900 - build fixes for 2.4.20-rh trees (1663)
4901 - return an error from lov_create if all OSCs are inactive (1751)
4902 - fix import levels when a reconnect happens without a timeout (1597)
4903 - exit early from mds_open if we get a lookup error (1749)
4904 - partial page read at EOF wouldn't wait for disk before sending (1642)
4905 - avoid NULL deref in obdfilter when reading page past EOF (1592)
4906 - avoid LASSERT in ll_intent_lock if server failed very early (1090)
4907 - fix LBUG in ll_it_open_error with rc = -2 (1861)
4908 - write/truncate lock inversion (1639)
4909 - Don't auto-load obdclass, portals modules during cleanup (1495)
4910 - fix timestamps from jumping to "now" (1763)
4911 - extra journal assertions (1648)
4912 - add an extra multiunlink test (1771)
4913 - fix read_record/write_record API (1776)
4914 - fix leak of offset_extent, possible incorrect i_size later (1772)
4915 - fix lasserts in mis-matched transnos during open-unlink testing (1541)
4916 - Debugging for the kqswnal_get_idle_tx problems (1820)
4917 - Allow recovery to be attempted multiple times (1536)
4918 - Write out MDS last_rcvd file after it is first created (1600)
4919 - Fix tx_descriptor leak in failed transmit situations (1827)
4920 - ext3 journaling fixes for assertion failure after IO error (1871)
4921 - class_export_put() on freed export after completion AST error (1896)
4922 - Fix revalidate looping in VFS (1322)
4923 - Don't access a freed export during MDS_REINT timeout (1521)
4924 - Add open-unlink recovery support on the MDS (1673,1764)
4925 - Return an error if no MDS data was read from last_rcvd (1946)
4926 - Fix for lookup "." or ".." crash on error (1932,1931,1935)
4927 - Don't setup a disk device that doesn't match exported UUID (317)
4928 - Reduce bulk RPC timeout to avoid cascading client/OST failures (1845)
4929 - avoid committing NULL handle in force close
4930 - local.sh is now a one-stripe LOV configuration
4931 - POSIX utime.4 -EPERM on FIFO not owned by user (56)
4932 - fix ext3 htree duplicate directory entry corruption (1516)
4933 - POSIX creat.13, fstat.1, open.18, stat.3 new file atime/mtime (2020)
4934 - update to new LOV EA format (2097)
4935 - interoperability for different PAGE_SIZE/wordsize (686,1821,1343,2042)
4937 2003-06-15 Phil Schwan <phil@clusterfs.com>
4940 - imports and exports cleanup too early, need refcounts (349, 879, 1045)
4941 - per-import/export recovery handling (958, 931, 959)
4942 - multiple last-rcvd slots, for serving multiple FSes (949)
4943 - connections are again shared between multiple imp/exports (963, 964)
4944 - "umount -f" would hang if any requests needed to be sent (393, 978)
4945 - avoid pinning large req buffer by copying for queued messages (989)
4946 - add "uuid" to "lctl device" command to help upcalls (991)
4947 - "open" RPCs with transnos would confuse recovery counters (1037)
4948 - do proper endian conversion of all wire messages (288, 340, 891)
4949 - remove OST bulk get LBUGs, fix ost_brw_write cleanup (1126)
4950 - call waiting locks callback from LDLM recovery thread (1127, 1151)
4951 - fix ptlrpc_connection leak in target_handle_connect (1174)
4952 - fix import refcounting bug in OST and MDS cleanup (1134)
4953 - if an invalid-at-open-time OSC returned before close(), LBUG (1150)
4954 - fix very unlikely obd_types race condition (501)
4955 - remove osc_open hack for echo_client (1187)
4956 - we leaked exports/dlmimps for forcibly disconnected clients (1143)
4957 - a failure in read_inode2 leads to deadlock (1139)
4958 - cancel ack-locks as soon as transaction is committed (1072)
4959 - fix major leaks and crashes in the bulk I/O path (937, 1057)
4960 - make sure to commitrw after any preprw to avoid deadlock (1162)
4961 - failing to execute a file in a lustre FS would lock inode (1203)
4962 - small DEBUG_REQ fix to avoid dereferencing a NULL (1227)
4963 - don't ASSERT while cleaning up an incompletely-setup obd (1248)
4964 - obd_uuid2tgt would walk off the end of the list (1255)
4965 - on IA64 the osc would give portals incorrect bulk size (1258)
4966 - fix debug daemon ioctl interface; allows daemon on ia64 (1274)
4967 - fix lock inversion caused by new llite matching code (1282)
4968 - limit the number of dirty pages on a client to 10MB (1286)
4969 - timed out locks were not being corrected cancelled (1289)
4970 - fix O_DIRECT above 4GB on IA-32 (1292)
4971 * major user-visible changes
4972 - fail out/fail over policy now controlled by the upcall (993)
4974 - add OBD_PING to check server availability and failure (954)
4975 - lustre messages are now sent in sending host order (288, 340, 891)
4976 - add eadatalen to MDS getattr reply (340)
4977 - OST read replies may contain second buffer, with per-page status (593)
4979 2003-03-11 Phil Schwan <phil@clusterfs.com>
4982 - LDLM_DEBUG macro fix, for gcc 3.2 (850)
4983 - failed open()s could cause deadlock; fixed (867, 869)
4984 - stop cancelling OST locks when files are closed (481)
4985 - overlapping XID spaces caused network corruption (851, 853)
4986 - fix unsafe fsfilt counter arithmetic; change to atomic_t
4987 - setattr_raw added, to do single-RPC, server-side setattrs
4988 - lmc/lconf syntax change for OST UUIDs
4989 - fix crashy race condition between ptlrpc_free_req and osc_close
4990 - don't use request in mdc_enqueue if we hit a timeout (889)
4991 - don't set the inode i_size for regular files from the MDS (896)
4992 - handle out of order completion AST (842)
4993 - don't LBUG if a lock request times out after receiving AST (913)
4994 - avoid d_rehash race in ll_find_alias by rehashing inside dcache_lock
4995 - if a bad lock AST arrives, send an error instead of dropping entirely
4996 - return 0 from revalidate2 if ll_intent_lock returns -EINTR (912)
4997 - fix leak in bulk IO when only partially completed (899, 900, 926)
4998 - fix O_DIRECT for ia64 (55)
4999 - (almost) eliminate Lustre-kernel-thread effects on load average (722)
5000 - C-z after timeout could hang a process forever; fixed (977)
5002 - client-side I/O cache (678, 924, 929, 941, 970)
5004 - READPAGE and SETATTRs which don't take server-side locks get
5007 2003-02-11 Phil Schwan <phil@clusterfs.com>
5010 - Fix ldlm_lock_match on the MDS to avoid matching remote locks (592)
5011 - Fix fsfilt_extN_readpage() to read a full page of directory
5012 entries, or fake the remainder if PAGE_SIZE != blocksize (500)
5013 - Avoid extra mdc_getattr() in ll_intent_lock when possible (534, 604)
5014 - Fix imbalanced LOV object allocation and out-of-bound access (469)
5015 - Most intent operations were removed, in favour of a new RPC mode
5016 that does a single RPC to the server and bypasses most of the VFS
5017 - All LDLM resource ID arrays were removed in favour of ldlm_res_id
5018 - Aggressively cancel local locks on DLM servers
5019 - mds_reint_unlink sends EA to the client if it's the last nlink.
5020 client uses that EA to unlink OST objects.
5021 - mds_reint_{rename,unlink,link} were rewritten to take ordered locks
5022 - recursive symlinks were fixed (439)
5023 - fixed NULL deref in DEBUG_REQ
5024 - filter_update_lastobjid no longer calls sync, which annoyed extN
5025 - fixed multi-client small-writes to a single file problem (445)
5026 - fixed mtime updates during file writes (607)
5027 - fixed vector writes on obdfilter causing problems when ENOSPC (670)
5028 - fixed bug in obd_brw_read/write() (under guise of testing 367)
5029 - fixed Linux OST size reporting problem (444, 656)
5030 - OST now updates object mtime with writes or setattr (607, 619)
5031 - client verifies file size before zeroing page past EOF (445)
5032 - OST now writes last allocated objid to disk with allocation (108)
5033 - LOV on echo now works (409)
5035 - mds_reint_unlink sends a new buffer, with the EA included. this
5036 buffer is only valid if body->valid & OBD_MD_FLEASIZE, which is only
5037 set if a regular file was being unlinked, and it was the last link
5038 - use PtlGet from the target for bulk writes (315)
5039 - OST now updates object mtime with writes or setattr (607, 619)
5040 - LDLM now has a grant-time callback to revalidate locked items, if
5042 - Many MDS operations were reorganized to combat race conditions
5044 - Merge b_intel branch (updated lprocfs code) - now at /proc/fs/lustre
5045 - configure check to avoid gcc version 2.96 20000731-2.96-98) (606)
5047 2003-01-06 Andreas Dilger <adilger@clusterfs.com>
5050 - Fully reactivate OST imports after reconnection (512, others)
5051 - Make sure client sees our -ENOTCONN from mds_handle (513 - partial)
5052 - More graceful error handling for truncating on dead OST (515)
5053 - Don't error out unless we're actually accessing dead stripes (474)
5054 - Fix garbage sizes when stripes are missing (410)
5055 - LRU counters were broken, causing constant lock purge (433, 432)
5056 - garbage on read from stripes with failed OSTs (441)
5057 - mark OSCs as active before reconnecting during recovery (438)
5058 - lov_enqueue and lov_cancel need to handle inactive OSTs (403)
5059 - lfind did not preserve OST order in output (443)
5060 - symlinks cause hung clients, incorrect data (439)
5061 - stop dereferencing request after dropping refcount (457)
5062 - don't LASSERT(spin_is_locked) on non-SMP (455)
5063 - fixes for many rename() bugs
5064 - fstat didn't correctly synchronize attributes (399)
5065 - server must handle lock cancellation during blocking AST prep (487)
5066 - bulk descriptors were free()d too soon (511)
5067 - fix paths in lconf, which would load incorrect modules (451, 507)
5068 - fix confusing lconf 'host not found' error message (386)
5069 - fix lock order deadlock on OST (O/R i_sem before journal ops, 478)
5070 - fix race condition in mdc_blocking_ast() for inode access (526)
5071 - fix lov_unpackmd() unpacking wrong number of stripes (537)
5072 - fix lov_set_osc_active() marking wrong OSC inactive (440)
5073 - fix bad lstripe lov_unpackmd() assertion (fix layering too) (527)
5074 - fix multiple writes of stripe MD to MDS (358, maybe 519)
5075 - fix lstripe in several ways (kernel side) (527)
5076 - fix request leak in ldlm_cli_enqueue (262)
5077 - incorrect OSC was marked inactive after OST failure
5078 - call mds_fs_cleanup before unmounting filesystem (524)
5079 - fix races between taking ns_lock and ldlm_lock_change_resource
5080 - fix races updating LOV export open file list
5081 - fix lov_enqueue error path, avoid decref-ing bad lock handle
5082 - fix recovery NULL deref in ldlm_cli_cancel_unused
5083 - fix some DLM races by using new hash table for lock handles (419)
5084 - permit the client to specify desired inodes, at replay
5085 - duplicate requests when we queue them for replay reintegration
5086 - fix last_rcvd offset calculation
5087 - sync after each recovered transaction, so we always make progress
5088 - never, not always, ERESTART requests without transnos
5089 - store the lov_desc in the MDS, so we don't depend on getlovinfo to
5091 - skip replay if the MDS says that the client is already connected
5092 - don't check for a recovery-enabled export to match lctl's UUID
5093 - don't INC_USE_COUNT for phantom exports
5094 - don't crash when cleaning up phantom exports (567)
5095 - don't double-finish or set replay data for errored mdc_open requests
5096 - abort requests when they time out, so we don't get old replies
5097 - send/receive replies for AST messages again
5098 - if the client says that it doesn't have the lock, cancel it on the
5100 - if we timeout during I/O, don't try to cancel an in-use lock; instead
5101 mark it as destroyed, it will all work out when decref is called
5102 - fix module use counts (22, 581)
5104 - ASTs now expect a reply (server cancels lock on error reply)
5106 2002-12-02 Andreas Dilger <adilger@clusterfs.com>
5109 - fix many simultaneous client startup (392)
5110 - fix dentry->d_it clobbering
5111 - credentials weren't being shipped for readdir/getattr operations
5112 - remove invalid assertions triggered during some concurrent MD
5114 - proper Lustre versions added (336, 389)
5115 - fix memory leak for create error case (398)
5116 - fix LOV locking bug that would get cli/srv out of sync
5117 - fix echo client over LOV (409)
5118 - fix dbench 2, extN refcount problem (170, 258, 356, 418)
5119 - fix double-O_EXCL intent crash (424)
5120 - avoid sending multiple lock CANCELs (352)
5122 - MDS can do multi-client recovery (modulo bugs in new code)
5124 - many updates, edits, cleanups
5126 2002-11-18 Phil Schwan <phil@clusterfs.com>
5129 - fix null d_it dereference (346)
5130 - fix full OST/dbench hang (333)
5131 - fix permission problem with file removal (286)
5132 - fix removal of OSCs from LOV when they fail
5133 - fix NULL deref during bulk timeout (214)
5134 - fix problems related to multiple filesystems on one MDS (241)
5135 - fixed serious subtle metadata locking bugs
5136 - free locks on clients when inodes are removed due to memory
5138 - fix inode pointer in lock data (285)
5139 - partial support for multiple MDS on a single host (241)
5140 - data locks weren't cancelled at clear_inode time (290, 311)
5141 - intent locks could lead to unbounded lock growth (205)
5142 - added a maximum lock count, an LRU list, and a flusher
5143 - fix multiple rename (365)
5144 - properly abstracted the echo client
5145 - OSC locked 1 byte too many; fixed
5146 - rewrote brw callback code:
5147 - fixed recovery bugs related to LOVs (306)
5148 - fixed too-many-pages-in-one-write crash (191)
5149 - fixed (again) crash in sync_io_timeout (214)
5150 - probably fixed callback-related race (385)
5152 - Add capability to MDS protocol
5153 - LDLM cancellations and callbacks on different portals
5155 2002-10-28 Andreas Dilger <adilger@clusterfs.com>
5158 - limit client IOV size to PTL_MD_MAX_IOV (611336, 191)
5159 - defer open object destruction to close time (601981, 138)
5160 - open/close OST file handle in obdo (OBD_MD_FLHANDLE) (601981, 138)
5161 - move LDLM_ENQUEUE/CONVERT back to MDS portal (625069)
5162 - abstract ll_lookup2, fix ll_revalidate2 to use abstraction (256)
5163 - don't call obd_setattr in ll_file_release for destroyed objects
5164 * protocol change to lustre_msg: move |version| and add |flags|
5165 * protocol change to osc_punch: "start" in "o_size", "end" in "o_blocks"
5166 * lock replay: for LDLM_FL_REPLAY trust client to do right thing
5167 * added replay of create, unlink, link and rename operations during
5168 MDS failover; recovery should be much more robust now
5169 * remove failed OSCs from LOVs (only lov_create uses this so far)
5170 * the lustre-HOWTO was brought (more) up to date (582544)
5172 2002-10-23 Phil Schwan <phil@clusterfs.com>
5175 - in-use dentries weren't being reused properly (617851)
5176 - prevent multiple LDLM setup (599178)
5177 - fix LOV size calculations for truncate (617853)
5178 - fix client handling of MDS intent errors (POSIX)
5179 - fix permission bug in lovstripe.c test (624321)
5180 - fix MDS thread deadlock - move LDLM handler to DLM portal (625069)
5181 - truncate past end of file could corrupt data
5182 - proper cleanup after timeouts, crashes, etc (592524, 550815)
5183 - a race in recovery could return ETIMEDOUT to apps (623947)
5184 - building outside the source directory was fixed
5185 * the lustre-HOWTO was brought (more) up to date (582544)
5186 * major progress was made on recovery functionality
5188 2002-10-10 Phil Schwan <phil@clusterfs.com>
5191 - recovery deadlock fix
5192 - rm -rf causes LBUG fix (617817)
5193 - file open by multiple tasks fix (618962)
5194 - directory permissions bugs (602707 and 620007)
5195 - journal_stop fixed with locking (611313)
5196 - O_APPEND failures resolved (618273, perhaps 614459)
5197 - lconf PATH fix (619770)
5198 - IA64 build fix (621450)
5199 - RPC buffer sizes scale with amount of memory
5201 2002-10-01 Phil Schwan <phil@clusterfs.com>
5204 - locks would be cancelled without throwing away data pages,
5205 resulting in inconsistent data (605627)
5206 - inode attributes were not always being refreshed (605627, 612449)
5207 - lconf now continues to cleanup after lctl reports an error
5208 - MDS now enforces user permissions (602707)
5209 - lprocfs cleanup fixed, but not yet enabled (614157)
5210 - fixed infinite server hang, should a client not respond to an AST
5211 - avoid going into recovery if user calls readlink() with a buffer
5212 that's too small (613941)
5213 - AST RPCs no longer require replies (614867) -- this may be changed
5214 - don't crash server if client sends an IOV that's too big (611336)
5215 - fixed lock conversion deadlock (611892)
5216 - fixed the following of symlinks (614622)
5217 * recovery: the server can remove locks from a client that dies, other
5218 clients can make progress
5219 * more extN patch fixes
5220 * compile-time configurable ptlrpc buffer allocations
5222 - collaborative read cache document
5223 - Lustre Lite Performance CDR document-in-progress
5225 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
5228 - fix typo in patch-2.4.18
5230 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
5233 - clear ptlrpc request each time in handle_incoming_request()
5234 - unlink of files now destroys the object on the OST
5236 2002-09-19 Peter Braam <braam@clusterfs.com>
5238 * add hard link support
5239 * change obdfile creation method
5240 * kernel patch changed
5242 2002-09-19 Peter Braam <braam@clusterfs.com>
5245 - stack overflow bug in extN fixed
5247 2002-09-18 Andreas Dilger <adilger@clusterfs.com>
5249 * documentation updates
5250 - add man pages for config tools
5251 - update tests/README to describe testing with new config tools
5252 - finish metadata API descriptions
5253 * bug fixes and cleanups
5254 - statfs workaround for 16TB limit
5255 - LOV stripe allocation improved, can stripe on subset of OSTs
5256 - LOV file size/IO offset was wrong for files > 4GB in size
5257 - object EA data was being dropped, caused files to be unreadable
5258 - memory overflow with non-LOV OST caused memory corruption
5259 - fixed regression tests to work with new config tools, obdfilter
5260 - fixed bug when directory size became larger than 1 block
5261 - fixed bug (for single client case) when PWD was deleted
5262 - invalidate local directory pages when doing intent-based ops
5263 - avoid LDLM oops when lock callback contained bad data
5265 2002-09-09 Andreas Dilger <adilger@clusterfs.com>
5267 * documentation updates
5268 * bug fixes and cleanups
5269 - configuration tools
5272 - 64-bit compile warnings
5273 - 64-bit internal statfs data
5275 * test_brw on persistent OST devices
5277 * lprocfs (disabled)
5279 2002-09-04 Andreas Dilger <adilger@clusterfs.com>
5281 * documentation updates
5282 * bug fixes and cleanups
5283 * configuration tools
5285 2002-08-30 Peter J. Braam <braam@clusterfs.com>
5288 * many small fixes to 0_5_4
5289 * io/network handling
5290 * thinkos in MDS operations
5292 2002-08-24 Peter J. Braam <braam@clusterfs.com>
5295 * crucial basic fixes to 0.5.3
5296 * IOR, Iozone work over Elan
5299 2002-08-07 Phil Schwan <phil@clusterfs.com>
5300 * version 0_5_3, our first alpha
5301 * we use the new Portals iovs
5302 * documentation updates
5303 * bug fixes and cleanups
5304 * small changes in the DLM wire protocol
5306 2002-07-25 Peter J. Braam <braam@clusterfs.com>
5307 * version 0_5_1 with some initial stability,
5308 * locking on MD and file I/O.
5309 * documentation updates
5310 * several bug fixes since 0.5.0
5311 * small changes in wire protocol
5313 2002-07-18 Phil Schwan <phil@clusterfs.com>
5315 * delivered as Lustre Light Alpha
5316 * fixed a crash after handling invalid MDS requests
5317 * fixed directory pages for architectures with non-4k pages sizes
5319 2002-07-11 Andreas Dilger <adilger@clusterfs.com>
5320 * release version v0_4_4
5321 * Moves TCP acceptor to be on port 2432 (unused Coda port) instead
5323 * Fixes a number of interruption problems with OST operations.
5324 * Update documentation for portals header changes
5325 * Move all wire protocol structs/defines to lustre_idl.h
5326 * Fixes symlink length bug.
5327 * Add tcpdump to repository.
5329 2002-07-05 Andreas Dilger <adilger@clusterfs.com>
5330 * release version v0_4_3
5331 * Fixes statfs for inodes on extN.
5332 * Fixes bug in runtests which would delete /etc/hosts.
5333 * Use 64-bit object IDs wherever possible (not into VFS though)
5334 Remove ost_get_info, which is unused by lustre, and out of date.
5336 2002-07-03 Peter Braam <braam@clusterfs.com>
5337 * release version v0_4_2 Fixes a lookup error (type not passed)
5338 * move forward to head of Portals
5339 * move forward to latest Lustre kernel
5341 2002-06-25 Peter Braam <braam@clusterfs.com>
5342 * release version v0_4_1. Hopefully stable on single node use.