1 tbd Sun Microsystems, Inc.
4 2.6.16.54-0.2.5 (SLES 10),
5 2.6.18-92.1.6.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).
15 Severity : enhancement
17 Description: Update to RHEL5 kernel-2.6.18-92.1.6.el5.
22 Description: Using wrong pointer in osc_brw_prep_request
23 Details : Access to array[-1] can produce panic if kernel compiled with
24 CONFIG_PAGE_ALLOC enabled
28 Description: Client runs out of low memory
29 Details : Consider only lowmem when counting initial number of llap pages
33 Description: *optional* service tags registration
34 Details : if the "service tags" package is installed on a Lustre node
35 When the filesystem is mounted, a local-node service tag will
36 be created. See http://inventory.sun.com/ for more information
37 about the Service Tags asset management system.
41 Description: Kernel BUG tries to release flock
42 Details : Lustre does not destroy flock lock before last reference goes
43 away. So always drop flock locks when client is evicted and
44 perform unlock regardless of successfulness of speaking to MDS.
48 Description: add recount protection for osc callbacks, so avoid panic on shutdown
52 Description: sanity test 65a fails if stripecount of -1 is set
53 Details : handle -1 striping on filesystem in ll_dirstripe_verify
58 Description: ASSERTION(CheckWriteback(page,cmd)) failed
59 Details : badly clear PG_Writeback bit in ll_ap_completion can produce false
62 Severity : enhancement
64 Description: Update to RHEL5 kernel-2.6.18-53.1.21.el5.
68 Description: do not process already freed flock
69 Details : flock can possibly be freed by another thread before it reaches
70 to ldlm_flock_completion_ast.
74 Description: LBUG during stress test
75 Details : Need properly lock accesses the flock deadlock detection list.
79 Description: oops in page fault handler
80 Details : kernel page fault handler can return two special 'pages' in error case, don't
81 try dereference NOPAGE_SIGBUS and NOPAGE_OMM.
85 Description: timeout with invalidate import.
86 Details : ptlrpcd_check call obd_zombie_impexp_cull and wait request which should be
87 handled by ptlrpcd. This produce long age waiting and -ETIMEOUT
88 ptlrpc_invalidate_import and as result LASSERT.
90 Severity : enhancement
92 Description: Update to RHEL5 kernel-2.6.18-53.1.19.el5.
96 Description: enable MGS and MDT services start separately
97 Details : add a 'nomgs' option in mount.lustre to enable start a MDT with
98 a co-located MGS without starting the MGS, which is a complement
99 to 'nosvc' mount option.
103 Frequency : after recovery
104 Description: precreate to many object's after del orphan.
105 Details : del orphan st in oscc last_id == next_id and this triger growing
106 count of precreated objects. Set flag LOW to skip increase count
107 of precreated objects.
111 Frequency : rare, on clear nid stats
112 Description: ASSERTION(client_stat->nid_exp_ref_count == 0)
113 Details : when clean nid stats sometimes try destroy live entry,
114 and this produce panic in free.
118 Description: Stack overflow during MDS log replay
119 ease stack pressure by using a thread dealing llog_process.
123 Description: wait until IO finished before start new when do lock cancel.
124 Details : VM protocol want old IO finished before start new, in this case
125 need wait until PG_writeback is cleared until check dirty flag and
126 call writepages in lock cancel callback.
128 Severity : enhancement
130 Description: using special macro for print time and cleanup in includes.
134 Description: mds_mfd_close() ASSERTION(rc == 0)
135 Details : In mds_mfd_close(), we need protect inode's writecount change
136 within its orphan write semaphore to prevent possible races.
140 Description: Obsolete CURRENT_SECONDS and use cfs_time_current_sec() instead.
144 Frequency : rare, on shutdown ost
145 Description: don't hit live lock with umount ost.
146 Details : shrink_dcache_parent can be in long loop with destroy dentries,
147 use shrink_dcache_sb instead.
151 Description: don't panic with use echo client
152 Details : echo client pass NULL as client nid pointer and this produce null
157 Description: fix build on ppc32
158 Details : compile code with -m64 flag produce wrong object file for ppc32.
162 Description: add message levels for liblustreapi
166 Description: fix for occasional failure case of -ENOSPC in recovery-small tests
167 Details : Move the 'good_osts' check before the 'total_bavail' check. This
168 will result in an -EAGAIN and in the exit call path we call
169 alloc_rr() which will with increasing aggressiveness attempt to
170 aquire precreated objects on the minimum number of required OSCs.
174 Description: Use old size assignment to avoid deadlock
175 Details : This reverts the changes in bugs 2369 and bug 14138 that introduced
176 the scheduling while holding a spinlock. We do not need locking
177 for size in ll_update_inode() because size is only updated from
178 the MDS for directories or files without objects, so there is no
179 other place to do the update, and concurrent access to such inodes
180 are protected by the inode lock.
184 Description: resolve "_IOWR redefined" build error on SLES10
188 Description: dump the memory debugging after all modules are unloaded to
189 suppress false negative in conf_sanity test 39
191 Severity : enhancement
193 Description: build kernel-ib packages for OFED 1.3 in our release cycle
198 Description: fix SLES kernel versioning
199 Details : the kernel version for our SLES 10 kernel did not include a "-"
200 before the "smp" at the end. while this was not a problem in
201 general, it did mean that software trying to use the kernel
202 version to try to detect a vendor specific kernel would fail.
203 this was most evident by the OFED build scripts.
207 Description: Don't update lov_desc members until making sure they are valid
208 Details : When updating lov_desc members via proc fs, need fix their
209 validities before doing the real update.
213 Description: don't put request into delay list while invalidate in flight.
214 Details : ptlrpc_delay_request sometimes put in delay list while invalidate
215 import in flight. this produce timeout for invalidate and sometimes
216 can cause stale data.
220 Frequency : on ppc only
221 Description: not convert ost objects for directory because it's not exist.
222 Details : ll_dir_getstripe assume dirrectory has ost objects but this wrong.
226 Description: Add FMODE_EXEC file flag for SLES10 SP1 kernel.
228 Severity : enhancement
230 Description: Update to support 2.6.22.14 vanilla kernel.
234 Frequency : rare, on recovery
235 Description: read procfs can produce deadlock in some situation
236 Details : Holding lprocfs lock which send rpc can produce block for destroy
237 obd objects and this also block reconnect with -EALREADY. This isn't
238 fix all lprocfs bugs - but make it rare.
240 Severity : enhancement
242 Description: Update kernel to RHEL5 2.6.18-53.1.14.el5.
245 Frequency : frequent on X2 node
247 Description: mdc_set_open_replay_data LBUG
248 Details : Set replay data for requests that are eligible for replay.
252 Description: lustre_mgs: operation 101 on unconnected MGS
253 Details : When MGC is disconnected from MGS long enough, MGS will evict the
254 MGC, and late on MGC cannot successfully connect to MGS and a lot
255 of the error messages complaining that MGS is not connected.
259 Frequency : on network error
260 Description: panic with double free request if network error
261 Details : mdc_finish_enqueue is finish request if any network error ocuring,
262 but it's true only for synchronus enqueue, for async enqueue
263 (via ptlrpcd) this incorrect and ptlrpcd want finish request
266 Severity : enhancement
268 Description: client-side metadata stat-ahead during readdir(directory readahead)
269 Details : perform client-side metadata stat-ahead when the client detects
270 readdir and sequential stat of dir entries therein
273 Frequency : on start mds
275 Description: Implement get_info(last_id) in obdfilter.
278 Frequency : occasional
280 Description: Correctly check stale fid, not start epoch if ost not support SOM
281 Details : open with flag O_CREATE need set old fid in op_fid3 because op_fid2
282 overwrited with new generated fid, but mds can anwer with one of these
283 two fids and both is not stale. setattr incorectly start epoch and
284 assume will be called done_writeting, but without SOM done_writing
288 Frequency : rare, depends on device drivers and load
290 Description: MDS or OSS nodes crash due to stack overflow
291 Details : Code changes in 1.8.0 increased the stack usage of some functions.
292 In some cases, in conjunction with device drivers that use a lot
293 of stack the MDS (or possibly OSS) service threads could overflow
294 the stack. One change which was identified to consume additional
295 stack has been reworked to avoid the extra stack usage.
298 Frequency : occasional
300 Description: Do not fail import if osc_interpret_create gets -EAGAIN
301 Details : If osc_interpret_create got -EAGAIN it immediately exits and
302 wakeup oscc_waitq. After wakeup oscc_wait_for_objects call
303 oscc_has_objects and see OSC has no objests and call
304 oscc_internal_create to resend create request.
306 Severity : enhancement
308 Description: Update to SLES10 SP1 latest kernel-2.6.16.54-0.2.5.
310 Severity : enhancement
312 Description: Update to RHEL5 latest kernel-2.6.18-53.1.13.el5.
315 Frequency : very rare
317 Description: Fix replay if there is an un-replied request and open
318 Details : In some cases, older replay request will revert the
319 mcd->mcd_last_xid on MDS which is used to record the client's
322 Severity : enhancement
324 Description: Update to RHEL5 latest kernel-2.6.18-53.1.6.el5.
326 Severity : enhancement
328 Description: Add rhel5 support to HEAD.
330 Serverity : enhancement
332 Description: Update RHEL4 kernel to 2.6.9-67.0.4.
337 Description: Don't allow skipping OSTs if index has been specified.
338 Details : Don't allow skipping OSTs if index has been specified, make locking
339 in internal create lots better.
343 Description: LBUG in ptlrpc_check_set() bad phase ebc0de00
344 Details : access to bitfield in structure is always rounded to long
345 and this produce problem with not atomic change any bit.
349 Description: Lustre make rpms failed.
350 Details : Remove ldiskfs spec file to avoids rpmbuild be confused when
351 builds Lustre rpms from tarball.
354 Frequency : rare on shutdown ost
356 Description: If llog cancel was not send before clean_exports phase, this can
357 produce deadlock in llog code.
358 Details : If llog thread has last reference to obd and call class_import_put
359 this produce deadlock because llog_cleanup_commit_master wait when
360 last llog_commit_thread exited, but this never success because was
361 called from llog_commit_thread.
365 Description: allow userland application know is lost one of stripes.
366 Details : fill lvb_blocks with error code on ost and return it to
367 application if error flag found.
371 Description: NULL lov_tgts causing MDS oops
372 Details : more safe checks for NULL lov_tgts for avoid oops.
374 Severity : enhancement
376 Description: Update to RHEL4 latest kernel-2.6.9-67.0.1.EL.
380 Descriptoin: make lov_create() will not stuck in obd_statfs_rqset()
381 Details : If an OST is down the MDS will hang indefinitely in
382 obd_statfs_rqset() waiting for the statfs data. While for
383 MDS QOS usage of statfs, it should not stuck in waiting.
385 Severity : enhancement
387 Description: remote_acl support
388 Details : Support ACL-based permission check for remote user.
389 Support setfacl/getfacl for remote user with the utils
390 "lfs {l,r}{s,g}etfacl" which follow the same parameter format as
391 the system "{s,g}etfacl" utils.
393 Severity : enhancement
395 Description: Update to RHEL4 U6 kernel-2.6.9-67.EL.
397 Severity : enhancement
399 Description: Update to RHEL5 latest kernel-2.6.18-53.1.4.el5.
403 Description: make mgs_setparam() handle fsname containing dash
404 Details : fsname containing a dash does not work with lctl conf_param
406 Severity : enhancement
408 Description: Update to SLES10 SP1 latest kernel-2.6.16.54-0.2.3.
410 Severity : enhancement
412 Description: Update to RHEL5 Update-1 kernel 2.6.18-53.el5.
416 Frequency : rare, at shutdown
417 Description: access already free / zero obd_namespace.
418 Details : if client_disconnect_export was called without force flag set,
419 and exist connect request in flight, this can produce access to
420 NULL pointer (or already free pointer) when connect_interpret
421 store ocd flags in obd_namespace.
425 Frequency : only at startup
426 Description: not alloc memory with spinlock held.
427 Details : allocation memory with GFP_KERNEL can produce sleep deadlock,
428 if any spinlock held.
430 Severity : enhancement
432 Description: make lustre randomly fail allocating memory
433 Details : Make lustre randomly failed allocating memory for testing purpose.
435 Severity : enhancement
437 Description: lost problems with lov objid file
438 Details : Fixes some scability and access to not inited memory problems
439 in work with lov objdid file.
444 Description: lfs find does not continue on file error
445 Details : Continue other files processing when a file/dir is absent.
449 Description: Inconsistent usage of lustre_pack_reply()
450 Details : Standardize the usage of lustre_pack_reply() such that it
451 always generate a CERROR on failure.
454 Frequency : occasional
456 Description: MDS hang or stay in waiting lock
457 Details : If client receive lock with CBPENDING flag ldlm need send lock
458 cancel as separate rpc, to avoid situation when cancel request
459 can't processed due all i/o threads stay in wait lock.
463 Description: Update to RHEL5 kernel 2.6.18-8.1.15.el5.
467 Description: Update to SLES10 SP1 kernel 2.6.16.53-0.16
471 Description: Update to SLES9 kernel-2.6.5-7.287.3.
475 Description: Update to RHEL4 latest kernel.
477 Severity : enhancement
479 Description: Build SLES10 patchless client fails
480 Details : The configure was broken by run ./configure with
481 --with-linux-obj=.... argument for patchless client. When the
482 configure use --with-linux-obj, the LINUXINCLUDE= -Iinclude
483 can't search header adequately. Use absolute path such as
484 -I($LINUX)/include instead.
488 Description: interrupt oig_wait produce painc on resend.
489 Details : brw_redo_request can be used for resend requests from ptlrpcd and
490 private set, and this produce situation when rq_ptlrpcd_data not
491 copyed to new allocated request and triggered LBUG on assert
492 req->rq_ptlrpcd_data != NULL. But this member used only for wakeup
493 ptlrpcd set if request is changed and can be safety changed to use
498 Description: LASSERT_{REQ,REP}SWAB macros are buggy
499 Details : If SWAB_PARANOIA is disabled, the LASSERT_REQSWAB and
500 LASSERT_REPSWAB macros become no-ops, which is incorrect. Drop
501 these macros and replace them with their difinitions instead.
505 Description: Update kernel patches for SLES10 2.6.16.53-0.8.
506 Details : Update which_patch & target file for SLES10 latest kernel.
509 Description: Remove client patches from SLES 10 kernel.
510 Details : This causes SLES 10 clients to behave as patchless clients
511 even on a Lustre-patched (server) kernel.
513 Severity : enhancement
515 Description: self-adjustable client's lru lists
516 Details : use adaptive algorithm for managing client cached locks lru
517 lists according to current server load, other client's work
518 pattern, memory activities, etc. Both, server and client
519 side namespaces provide number of proc tunables for controlling
522 Severity : enhancement
524 Description: light-weight GSS support
525 Details : Support krb5n and krb5a mode, which keep Kerberos 5 authentication
526 and reduce performance overhead.
528 Severity : enhancement
530 Description: Linux keyring support
531 Details : Support using service of Linux keyring for Lustre GSS internal
532 context refresh/cache mechanism.
536 Description: Fix errors in lfs documentation
537 Details : Fixes man pages
541 Description: don't use GFP_* in generic Lustre code.
542 Details : Use cfs_alloc_* functions and CFS_* flags for code portability.
546 Description: obdclass is limited by single OBD_ALLOC(idarray)
547 Details : replace OBD_ALLOC/OBD_FREE with OBD_VMALLOC/OBD_VFREE
551 Description: warnings with build patchless client with vanila 2.6.19 and up
552 Details : change old ctl_table style and replace ctl_table/ctl_table_header
553 with cfs_sysctl_table_t/cfs_sysctl_table_header_t
557 Frequency : Only for SLES
558 Description: sanity_quota fail test_1
559 Details : There are multiple occurences of $TSTUSR in SLES's /etc/group
560 file, which makes TSTID[2] inunique.
564 Frequency : Only for SLES9
565 Description: Kernel patches for SLES9 2.6.5-7.286 kernel
566 Details : Update target/ChangeLog/which_patch .
570 Frequency : Only for test_10 in sanity-quota.sh
571 Description: a bug in quota test code
572 Details : checking a test flag in a wrong place
576 Frequency : Only for test_13 in sanity-quota.sh
577 Description: a bug in quota test script
578 Details : shouldn't check overfull allocated quota
582 Description: jbd statistics
583 Details : Port older jbd statistics patch for sles10
585 Severity : enhancement
587 Description: Add printing inode info into message about error in writepage.
592 Description: reply_lock_interpret crash due to race with it and lock cancel.
593 Details : Do not replay locks that are being cancelled. Do not reference
594 locks by their address during replay, just by their handle.
598 Frequency : When flocks are used.
599 Description: assertion failure in ldlm_cli_enquque_fini for non NULL lock.
600 Details : Flock locks might destroy just granted lock if it could be merged
601 with another existing flock, this is done in completion handler,
602 so teach ldlm_cli_enquque_fini that this is a valid case for
608 Description: Oops in read and write path when failing to allocate lock.
609 Details : Check if lock allocation failed and return error back.
613 Description: lstripe command fails for valid OST index
614 Details : The stripe offset is compared to lov->desc.ld_tgt_count
615 instead of lov->desc.ld_active_tgt_count.
619 Description: sanity.sh failed test 103
620 Details : RHEL mis-interpret setfacl "-X" param, so we won't test setfacl
625 Description: df doesn't work properly if diskfs blocksize != 4K
626 Details : Choose biggest blocksize of OST's as the LOV's blocksize.
630 Description: lfs find on -1 stripe looping in lsm_lmm_verify_common()
631 Details : Avoid lov_verify_lmm_common() on directory with -1 stripe count.
633 Severity : enhancement
635 Description: Adaptive timeouts
636 Details : RPC timeouts adapt to changing server load and network
637 conditions to reduce resend attempts and improve recovery time.
641 Description: llapi_file_create() does not allow some changes
642 Details : add llapi_file_open() that allows specifying the mode and
643 open flags, and also returns an open file handle.
647 Description: merge and cleanup kernel patches.
648 Details : Remove mnt_lustre_list in vfs_intent-2.6-rhel4.patch.
652 Description: Add journal checksum support.(Kernel part)
653 Details : The journal checksum feature adds two new flags i.e
654 JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT and
655 JBD2_FEATURE_COMPAT_CHECKSUM. JBD2_FEATURE_CHECKSUM flag
656 indicates that the commit block contains the checksum for
657 the blocks described by the descriptor blocks. Now commit
658 record can be sent to disk without waiting for descriptor
659 blocks to be written to disk. This behavior is controlled
660 using JBD2_FEATURE_ASYNC_COMMIT flag.
664 Description: OSS needs mutliple precreate threads
665 Details : Add ability to start more than one create thread per OSS.
669 Description: Sanity.sh test_65k failed: setstripe should have succeeded
670 Details : execute lfs setstripe on client
674 Description: mds_obd_create error creating tmp object
675 Details : When the user sets quota on root, llog will be affected and can't
676 create files and write files.
679 Frequency : Always on ia64 patchless client, and possibly others.
681 Description: Add EXPORT_SYMBOL check for node_to_cpumask symbol.
682 Details : This allows the patchless client to be loaded on architectures
687 Description: RedHat Update kernel for RHEL5
688 Details : Add the kernel config file for RHEL5.
692 Description: RedHat Update kernel for RHEL5
693 Details : Modify the kernel config file more closer RHEL5.
697 Description: Build failure against Centos5 (RHEL5)
698 Details : Define PAGE_SIZE when it isn't present.
702 Description: test_7 on sanity-quota.sh can't be used on separate machines
703 Details : add facet to handle it
707 Description: "ll_intent_file_open()) lock enqueue: err: -13" with nfs
708 Details : with NFS, the anon dentry's parent was set to itself in
709 d_alloc_anon(), so in MDS, we use rec->ur_fid1 to find the
710 corresponding dentry other than use rec->ur_name.
712 Severity : enhancement
714 Description: lfs setstripe enhancement
715 Details : Make lfs setstripe understand 'k', 'm' and 'g' for stripe size.
719 Description: enable data checksumming by default
720 Details : enable checksum by default, allow --disable-checksum
721 configure option and "-o nochecksum" mount option. Checksums
722 can also be disabled at runtime via $LPROC/osc/*/checksum_pages.
726 Description: lustre support for RHEL5
727 Details : Add support for RHEL5.
731 Description: Client eviction due to failover config
732 Details : after a connection loss, the lustre client should attempt to
733 reconnect to the last active server first before trying the
734 other potential connections.
738 Description: when mds and osts use different quota unit(32bit and 64bit),
739 quota will be released repeatly.
740 Details : void sending multiple quota reqs to mds, which will keep the
741 status between the reqs.
745 Description: osts not allocated evenly to files
746 Details : change the condition to increase offset_idx
749 Frequency : Always for filesystems larger than 2TB on 32-bit systems.
751 Description: Data corruption for OSTs that are formatted larger than 2TB
753 Details : When generating the bio request for lustre file writes the
754 sector number would overflow a temporary variable before being
755 used for the IO. The data reads correctly from Lustre (which
756 will overflow in a similar manner) but other file data or
757 filesystem metadata may be corrupted in some cases.
761 Description: Tune the kernel for good SCSI performance.
762 Details : Set the value of /sys/block/{dev}/queue/max_sectors_kb
763 to the value of /sys/block/{dev}/queue/max_hw_sectors_kb
768 Description: rewrite ext2-derived code in obdclass/uuid.c
769 Details : rewrite inherited code (uuid parsing code from ext2 utils)
770 from scratch preserving functionality.
774 Description: "lfs find -obd UUID" prints directories
775 Details : "lfs find -obd UUID" will return all directory names instead
776 of just file names. It is incorrect because the directories
777 do not reside on the OSTs.
781 Description: use i_size_read and i_size_write in 2.6 port
782 Details : replace inode->i_size access with i_size_read/write()
785 Frequency : when removing large files
787 Description: scheduling issue during removal of large Lustre files
788 Details : Don't take the BKL in fsfilt_ext3_setattr() for 2.6 kernels.
789 It causes scheduling issues when removing large files (17TB in the
793 Frequency : only with liblustre clients on XT3
795 Description: evictions taking too long
796 Details : allow llrd to evict clients directly on OSTs
799 Frequency : only on ppc
801 Description: /proc/fs/lustre/devices broken on ppc
802 Details : The patch as applied to 1.6.2 doesn't look correct for all arches.
803 We should make sure the type of 'index' is loff_t and then cast
804 explicitly as needed below. Do not assign an explicitly cast
808 Frequency : only for rhel5
810 Description: Kernel patches update for RHEL5 2.6.18-8.1.10.el5.
811 Details : Modify the target file & which_kernel.
813 Severity : enhancement
815 Description: omit set fsid for export NFS
816 Details : fix set/restore device id for avoid EMFILE error and mark lustre fs
817 as FS_REQUIRES_DEV for avoid problems with generate fsid.
821 Frequency : Always, for kernels after 2.6.16
822 Description: Fix warning idr_remove called for id=.. which is not allocated.
823 Details : Last kernels save old s_dev before kill super and not allow
824 to restore from callback - restore it before call kill_anon_super.
828 Description: buffer overruns could theoretically occur
829 Details : llapi_semantic_traverse() modifies the "path" argument by
830 appending values to the end of the origin string, and a
831 overrun may occur. Adding buffer overrun check in liblustreapi.
835 Description: Fix error on 'ls .' at the top of the Lustre mount.
836 Details : Don't revalidate dentry if it is a root dentry.
840 Description: Kernel patches update for RHEL4 2.6.9-55.0.6.
841 Details : Modify vm-tunables-rhel4.patch.
845 Description: Kernel config for 2.6.18-vanilla.
846 Details : Modify targets/2.6-vanilla.target.in.
847 Add config file kernel-2.6.18-2.6-vanilla-i686.config.
848 Add config file kernel-2.6.18-2.6-vanilla-i686-smp.config.
849 Add config file kernel-2.6.18-2.6-vanilla-x86_64.config.
850 Add config file kernel-2.6.18-2.6-vanilla-x86_64-smp.config.
854 Description: improve handling recoverable errors
855 Details : if request processig with error which can be recoverable on server
856 request should be resend, otherwise page released from cache and
861 Description: Kernel patches update for RHEL5 2.6.18-8.1.14.el5.
862 Details : Modify target file & which_patch.
863 A flaw was found in the IA32 system call emulation provided
864 on AMD64 and Intel 64 platforms. An improperly validated 64-bit
865 value could be stored in the %RAX register, which could trigger an
866 out-of-bounds system call table access. An untrusted local user
867 could exploit this flaw to run code in the kernel
868 (ie a root privilege escalation). (CVE-2007-4573).
872 Description: O_DIRECT bypasses client statistics.
873 Details : When running with O_DIRECT I/O, neither the client rpc_stats nor
874 read_ahead_stats were updated. copied the stats section from
875 osc_send_oap_rpc() into async_internal().
879 Description: Add jbd statistics patch for RHEL5 and 2.6.18-vanilla
883 Description: change order of libsysio includes
884 Details : '#include sysio.h' should always come before '#include xtio.h'
888 Description: handle "serious error: objid * already exists" more gracefully
889 Details : If LAST_ID value on disk is smaller than the objects existing in
890 the O/0/d* directories, it indicates disk corruption and causes an
891 LBUG(). If the object is 0-length, then we should use the existing
892 object. This will help to avoid a full fsck in most cases.
894 Severity : enhancement
896 Description: adapt the lustre_config script to support the upgrade case
897 Details : Add "-u" option for lustre_config script to support upgrading 1.4
898 server targets to 1.6 in parallel.
902 Description: To avoid grant space > avaible space when the disk is almost
903 full. Without this patch you might see the error "grant XXXX >
904 available" or some LBUG about grant, when the disk is almost
906 Details : In filter_check_grant, for non_grant cache write, we should
907 check the left space by if (*left > ungranted + bytes), instead
908 of (*left > ungranted), because only we are sure the left space
909 is enough for another "bytes", then the ungrant space should be
910 increase. In client, we should update cl_avail_grant only there
911 is OBD_MD_FLGRANT in the reply.
915 Description: Update RHEL 4 kernel to fix local root privilege escalation.
916 Details : Update to the latest RHEL 4 kernel to fix the vulnerability
917 described in CVE-2007-4573. This problem could allow untrusted
918 local users to gain root access.
921 Frequency : when using O_DIRECT and quotas
923 Description: Incorrect file ownership on O_DIRECT output files
924 Details : block usage reported by 'lfs quota' does not take into account
925 files that have been written with O_DIRECT.
930 Description: touch file failed when fs is not full
931 Details : OST in recovery should not be discarded by MDS in alloc_qos(),
932 otherwise we can get ENOSP while fs is not full.
936 Description: parallel lock callbacks
937 Details : Instead of sending blocking and completion callbacks as separated
938 requests, adding them to a set and sending in parallel.
941 Frequency : only for Cray XT3
942 Bugzilla : 12829/13455
943 Description: Changing primary group doesn't change the group lustre assigns to
945 Details : When CRAY_XT3 is defined, the fsgid supplied by the client is
946 overridden with the primary group provided by the group upcall,
947 whereas the supplied fsgid can be trusted if it is in the list of
948 supplementary groups returned by the group upcall.
950 Severity : enhancement
952 Description: Allow masking D_WARNING, D_ERROR messages from console
953 Details : Console messages can now be disabled via lnet.printk.
957 Description: User code with malformed file open parameter crashes client node
958 Details : Before packing join_file req, all the related reference should be
959 checked carefully in case some malformed flags cause fake
960 join_file req on client.
964 Description: LDLM_ENQUEUE races with LDLM_CP_CALLBACK
965 Details : ldlm_completion_ast() assumes that a lock is granted when the req
966 mode is equal to the granted mode. However, it should also check
967 that LDLM_FL_CP_REQD is not set.
971 Description: Heavy nfs access might result in deadlocks
972 Details : After ELC code landed, it is now improper to enqueue any mds
973 locks under och_sem, because enqueue might want to decide to
974 cancel open locks for same inode we are holding och_sem for.
978 Description: Client eviction while running blogbench
979 Details : A lot of unlink operations with concurrent I/O can lead to a
980 deadlock causing evictions. To address the problem, the number of
981 oustanding OST_DESTROY requests is now throttled to
982 max_rpcs_in_flight per OSC and LDLM_FL_DISCARD_DATA blocking
983 callbacks are processed in priority.
987 Description: enable ACLs on MDS by default
988 Details : ACLs must be enabled on MDS by default.
991 Frequency : PPC/PPC64 only
993 Description: conflicts between asm-ppc64/types.h and lustre_types.h
994 Details : fix duplicated definitions between asm-ppc64/types.h and
995 lustre_types.h on PPC.
998 Frequency : PPC/PPC64 only
1000 Description: asm-ppc/segment.h does not exist
1001 Details : fix compile issue on PPC.
1005 Description: better handle error messages in extents code
1008 Frequency : RHEL4 only
1010 Description: mkfs is very slow on IA64/RHEL4
1011 Details : A performance regression has been discovered in the MPT Fusion
1012 driver between versions 3.02.73rh and 3.02.99.00rh. As a
1013 consequence, we have downgraded the MPT Fusion driver in the RHEL4
1014 kernel from 3.02.99.00 to 3.02.73 until this problem is fixed.
1016 Severity : enhancement
1018 Description: SNMP support enhancement
1019 Details : Adding total number of sampled request for an MDS node in snmp
1022 Severity : enhancement
1024 Description: Optimize ldlm waiting list processing for PR extent locks
1025 Details : When processing waiting list for read extent lock and meeting read
1026 lock that is same or wider to it that is not contended, skip
1027 processing rest of the list and immediatelly return current
1028 status of conflictness, since we are guaranteed there are no
1029 conflicting locks in the rest of the list.
1033 Description: Time out and refuse to reconnect
1034 Details : When the failover node is the primary node, it is possible
1035 to have two identical connections in imp_conn_list. We must
1036 compare not conn's pointers but NIDs, otherwise we can defeat
1037 connection throttling.
1041 Description: port llog fixes from b1_6 into HEAD
1042 Details : Port llog reference couting and some llog cleanups from b1_6
1043 (bug 10800) into HEAD, for protect from panic and access to already
1044 free llog structures.
1048 Description: Detect stride IO mode in read-ahead
1049 Details : When a client does stride read, read-ahead should detect that and
1050 read-ahead pages according to the detected stride pattern.
1054 Description: data checksumming impacts single node performance
1055 Details : add support for several checksum algorithm. Currently, only CRC32
1056 and Adler-32 are supported. The checksum type can be changed on
1057 the fly via /proc/fs/lustre/osc/*/checksum_type.
1061 Description: use adler32 for page checksums
1062 Details : when available, use the Adler-32 algorithm instead of CRC32 for
1067 Description: build for x2 fails
1068 Details : fix compile issue on Cray systems.
1072 Description: Properly match for duplicate locks
1073 Details : Due to different lock order from skiplists code, we need to
1074 traverse entire list for now
1077 Frequency : only on PPC/SLES10
1079 Description: "BITS_PER_LONG is not 32 or 64" in linux/idr.h
1080 Details : On SLES10/PPC, fs.h includes idr.h which requires BITS_PER_LONG to
1081 be defined. Add a hack in mkfs_lustre.c to work around this compile
1086 Description: LASSERT on MDS when client holding flock lock dies
1087 Details : ldlm pool logic depends on number of granted locks equal to
1088 number of released locks which is not true for flock locks, so
1089 just exclude such locks from consideration.
1093 Description: MDS deadlock with many ll_sync_lov threads and I/O stalled
1094 Details : Use fsfilt_sync() for both the whole filesystem sync and
1095 individual file sync to eliminate dangerous inode locking
1096 with I_LOCK that can lead to a deadlock.
1100 Description: performance in 1.6.3
1101 Details : Force q->max_phys_segments to MAX_PHYS_SEGMENTS on SLES10 to be
1102 sure that 1MB requests are not fragmented by the block layer.
1104 Severity : enhancement
1106 Description: organize the server-side client stats on per-nid basis
1107 Details : Change the structure of stats under obdfilter and mds to
1115 The "uuid"s file would list the uuids of _active_ exports.
1116 And the clear entry is to clear all stats and stale nids.
1118 Severity : enhancement
1120 Description: eliminate client locks in face of contention
1121 Details : file contention detection and lockless i/o implementation
1122 for contended files.
1126 Description: Reinitialize optind to 0 so that interactive lfs works in all cases
1129 Frequency : very rare, if additional xattrs are used on kernels >= 2.6.12
1131 Description: MDS may lose file striping (and hence file data) in some cases
1132 Details : If there are additional extended attributes stored on the MDS,
1133 in particular ACLs, SELinux, or user attributes (if user_xattr
1134 is specified for the client mount options) then there is a risk
1135 of attribute loss. Additionally, the Lustre file striping
1136 needs to be larger than default (e.g. striped over all OSTs),
1137 and an additional attribute must be stored initially in the
1138 inode and then increase in size enough to be moved to the
1139 external attribute block (e.g. ACL growing in size) for file
1144 Description: skiplist implementation simplification
1145 Details : skiplists are used to group compatible locks on granted list
1146 that was implemented as tracking first and last lock of each lock group
1147 the patch changes that to using doubly linked lists
1151 Description: MDS LBUG: ASSERTION(!IS_ERR(dchild))
1152 Details : Change LASSERTs to client eviction (i.e. abort client's recovery)
1153 because LASSERT on both the data supplied by a client, and the data
1154 on disk is dangerous and incorrect.
1156 Severity : enhancement
1158 Description: Slow truncate/writes to huge files at high offsets.
1159 Details : Directly associate cached pages to lock that protect those pages,
1160 this allows us to quickly find what pages to write and remove
1161 once lock callback is received.
1165 Description: more ldlm soft lockups
1166 Details : In ldlm_resource_add_lock(), call to ldlm_resource_dump()
1167 starve other threads from the resource lock for a long time in
1168 case of long waiting queue, so change the debug level from
1169 D_OTHER to the less frequently used D_INFO.
1171 Severity : enhancement
1173 Description: add -gid, -group, -uid, -user options to lfs find
1177 Description: Hung threads in invalidate_inode_pages2_range
1178 Details : The direct IO path doesn't call check_rpcs to submit a new RPC once
1179 one is completed. As a result, some RPCs are stuck in the queue
1184 Description: filter threads hungs on waiting journal commit
1185 Details : Cleanup filter group llog code, then only filter group llog will
1186 be only created in the MDS/OST syncing process.
1190 Description: Procfs and llog threads access destoryed import sometimes.
1191 Details : Sync the import destoryed process with procfs and llog threads by
1192 the import refcount and semaphore.
1194 Severity : enhancement
1196 Description: openlock cache of b1_6 port to HEAD
1201 Description: kernel BUG at ldiskfs2_ext_new_extent_cb
1202 Details : If insertion of an extent fails, then discard the inode
1203 preallocation and free data blocks else it can lead to duplicate
1208 Description: don't always update ctime in ext3_xattr_set_handle()
1209 Details : Current xattr code updates the inode ctime in ext3_xattr_set_handle.
1210 In some cases the ctime should not be updated, for example for
1211 2.0->1.8 compatibility it is necessary to delete an xattr and it
1212 should not update the ctime.
1214 --------------------------------------------------------------------------------
1216 2007-08-10 Cluster File Systems, Inc. <info@clusterfs.com>
1218 * Support for kernels:
1219 2.6.5-7.283 (SLES 9),
1220 2.6.9-55.EL (RHEL 4),
1221 2.6.16.46-0.14 (SLES 10),
1222 2.6.18.8 vanilla (kernel.org)
1223 * Client support for unpatched kernels:
1224 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
1225 2.6.16 - 2.6.22 vanilla (kernel.org)
1226 * Due to recently discovered recovery problems, we do not recommend
1227 using patchless RHEL 4 clients with this or any earlier release.
1228 * Recommended e2fsprogs version: 1.39.cfs8
1229 * Note that reiserfs quotas are disabled on SLES 10 in this kernel.
1230 * Starting with this release, the ldiskfs backing filesystem required
1231 by Lustre is now in its own package, lustre-ldiskfs. This package
1232 should be installed. It is versioned separately from Lustre and
1233 may be released separately in future.
1237 Description: block reactivating mgc import until all deactivates complete
1238 Details : Fix race when failing back MDT/MGS to itself (testing)
1241 Frequency : at statup only
1243 Description: mds_lov_synchronize race leads to various problems
1244 Details : simultaneous MDT->OST connections at startup can cause the
1245 sync to abort, leaving the OSC in a bad state.
1247 Severity : enhancement
1249 Description: add optional extra BUILD_VERSION info
1250 Details : add a new environment variable (namely LUSTRE_VERS) which allows
1251 to override the lustre version.
1254 Frequency : 2.6.18 servers only
1256 Description: ll_kern_mount() doesn't release the module reference
1257 Details : The ldiskfs module reference count never drops down to 0
1258 because ll_kern_mount() doesn't release the module reference.
1263 Description: server LBUG when using old ost_num_threads parameter
1264 Details : Accept the old ost_num_threads parameter but warn that it
1265 is deprecated, and fix an off-by-one error that caused an LBUG.
1270 Description: Transient SCSI error results in persistent IO issue
1271 Details : iobuf->dr_error is not reinitialized to 0 between two
1275 Frequency : sometimes when underlying device returns I/O errors
1277 Description: OSTs not going read-only during write failures
1278 Details : OSTs are not remounted read-only when the journal commit threads
1279 get I/O errors because fsfilt_ext3 calls journal_start/stop()
1280 instead of the ext3 wrappers.
1284 Description: poor connect scaling with increasing client count
1285 Details : Don't run filter_grant_sanity_check for more than 100 exports
1286 to improve scaling for large numbers of clients.
1289 Frequency : SLES10 only
1291 Description: sanity-quota.sh quotacheck failed: rc = -22
1292 Details : Quotas cannot be enabled on SLES10.
1295 Frequency : liblustre clients only
1297 Description: getdirentries does not give error when run on compute nodes
1298 Details : getdirentries does not fail when the size specified as an argument
1299 is too small to contain at least one entry
1301 Severity : enhancement
1303 Description: Add LNET router traceability for debug purposes
1304 Details : If a checksum failure occurs with a router as part of the
1305 IO path, the NID of the last router that forwarded the bulk data
1306 is printed so it can be identified.
1311 Description: OST "spontaneously" evicts client; client has imp_pingable == 0
1312 Details : Due to a race condition, liblustre clients were occasionally
1313 evicted incorrectly.
1315 Severity : enhancement
1317 Description: lfs setstripe use optional parameters instead of postional
1320 Severity : enhancement
1322 Description: Nanosecond timestamp support for ldiskfs
1323 Details : The on-disk ldiskfs filesystem has added support for nanosecond
1324 resolution timestamps. There is not yet support for this at
1325 the Lustre filesystem level.
1328 Frequency : during server recovery
1330 Description: MDS failing to send precreate requests due to OSCC_FLAG_RECOVERING
1331 Details : request with rq_no_resend flag not awake l_wait_event if they get a
1335 Frequency : nfs export on patchless client
1337 Description: connectathon hang when test nfs export over patchless client
1338 Details : Disconnected dentry cannot be found with lookup, so we do not need
1339 to unhash it or make it invalid
1342 Description: fix llapi_lov_get_uuids() to allow many OSTs to be returned
1343 Details: : Change llapi_lov_get_uuids() to read the UUIDs from /proc instead
1344 of using an ioctl. This allows lfsck for > 160 OSTs to succeed.
1349 Description: open req refcounting wrong on reconnect
1350 Details : If reconnect happened between getting open reply from server and
1351 call to mdc_set_replay_data in ll_file_open, we will schedule
1352 replay for unreferenced request that we are about to free.
1353 Subsequent close will crash in variety of ways.
1354 Check that request is still eligible for replay in
1355 mdc_set_replay_data().
1360 Description: disable writes to filesystem when reading health_check file
1361 Details : the default for reading the health_check proc file has changed
1362 to NOT do a journal transaction and write to disk, because this
1363 can cause reads of the /proc file to hang and block HA state
1364 checking on a healthy but otherwise heavily loaded system. It
1365 is possible to return to the previous behaviour during configure
1366 with --enable-health-write.
1368 Severity : enhancement
1370 Description: 64-bit inode version
1371 Details: : Add a on-disk 64-bit inode version for ext3 to track changes made
1372 to the inode. This will be required for version-based recovery.
1377 Description: MDS fails to start if a duplicate client export is detected
1378 Details : in some rare cases it was possible for a client to connect to
1379 an MDS multiple times. Upon recovery the MDS would detect this
1380 and fail during startup. Handle this more gracefully.
1382 Severity : enhancement
1384 Description: Add -o localflock option to simulate old noflock
1386 Details : This will achieve local-only flock/fcntl locks
1392 Description: log_commit_thread vs filter_destroy race leads to crash
1393 Details : Take import reference before releasing llog record semaphore
1398 Description: Wrong request locking in request set processing
1399 Details : ptlrpc_check_set wrongly uses req->rq_lock for proctect add to
1400 imp_delayed_list, in this place should be used imp_lock.
1403 Frequency : when reconnection
1405 Description: Grant Leak when osc reconnect to OST
1406 Details : When osc reconnect ost, OST(filter) should check whether it
1407 should grant more space to client by comparing fed_grant and
1408 cl_avail_grant, and return the granted space to client instead
1409 of "new granted" space, because client will call osc_init_grant
1410 to update the client grant space info.
1413 Frequency : when client reconnect to OST
1415 Description: Grant Leak when osc do resend and replay bulk write
1416 Details : When osc reconnect to OST, OST(filter)should clear grant info of
1417 bulk write request, because the grant info while be sync between
1418 OSC and OST when reconnect, and we should ignore the grant info
1419 these of resend/replay write req.
1424 Description: Grant space more than avaiable left space sometimes.
1425 Details : When then OST is about to be full, if two bulk writing from
1426 different clients came to OST. Accord the avaliable space of the
1427 OST, the first req should be permitted, and the second one
1428 should be denied by ENOSPC. But if the seconde arrived before
1429 the first one is commited. The OST might wrongly permit second
1430 writing, which will cause grant space > avaiable space.
1433 Frequency : when client is evicted
1435 Description: Grant might be wrongly erased when osc is evicted by OST
1436 Details : when the import is evicted by server, it will fork another
1437 thread ptlrpc_invalidate_import_thread to invalidate the
1438 import, where the grant will be set to 0. While the original
1439 thread will update the grant it got when connecting. So if
1440 the former happened latter, the grant will be wrongly errased
1441 because of this race.
1446 Description: Checking Stale with correct fid
1447 Details : ll_revalidate_it should uses de_inode instead of op_data.fid2
1448 to check whether it is stale, because sometimes, we want the
1449 enqueue happened anyway, and op_data.fid2 will not be initialized.
1451 Severity : enhancement
1453 Description: update patchless client
1454 Details : Add support for patchless client with 2.6.20, 2.6.21 and RHEL 5
1457 Frequency : only with 2.4 kernel
1459 Description: random memory corruption
1460 Details : size of struct ll_inode_info is to big for union inode.u and this
1461 can be cause of random memory corruption.
1466 Description: Memory leak in recovery
1467 Details : Lov_mds_md was not free in an error handler in mds_create_object.
1468 It should also check obd_fail before fsfilt_start, otherwise if
1469 fsfilt_start return -EROFS,(failover mds during mds recovery).
1470 then the req will return with repmsg->transno = 0 and rc = EROFS.
1471 and we met hit the assert LASSERT(req->rq_reqmsg->transno ==
1472 req->rq_repmsg->transno) in ptlrpc_replay_interpret. Fcc should
1473 be freed no matter whether fsfilt_commit success or not.
1476 Frequency : only with huge count clients
1478 Description: Prevents from taking the superblock lock in llap_from_page for
1480 Details : using LL_ORIGIN_REMOVEPAGE origin flag instead of LL_ORIGIN_UNKNOW
1481 for llap_from_page call in ll_removepage prevents from taking the
1482 superblock lock for a soon died page.
1487 Description: Not check open intent error before release open handle
1488 Details : in some rare cases, the open intent error is not checked before
1489 release open handle, which may cause
1490 ASSERTION(open_req->rq_transno != 0), because it tries to release
1491 the failed open handle.
1496 Description: Set cat log bitmap only after create log success.
1497 Details : in some rare cases, the cat log bitmap is set too early. and it
1498 should be set only after create log success.
1502 Description: Accessing a block bevice can re-enable I/O when Lustre is
1503 tearing down a device.
1504 Details : dev_clear_rdonly(bdev) must be called in kill_bdev() instead of
1509 Description: service threads may hog cpus when there are a lot of requests
1510 Details : Insert cond_resched to give other threads a chance to use some CPU
1515 Description: the cat log was not initialized in recovery
1516 Details : When mds(mgs) do recovery, the tgt_count might be zero, so the
1517 unlink log on mds will not be initialized until mds post
1518 recovery. And also in mds post recovery, the unlink log will
1519 initialization will be done asynchronausly, so there will be race
1520 between add unlink log and unlink log initialization.
1524 Description: brw_stats were being printed incorrectly
1525 Details : brw_stats were being printed as log2 but all of them were not
1526 recorded as log2. Also remove some code duplication arising from
1527 filter_tally_{read,write}.
1531 Frequency : rare, only in recovery.
1532 Description: ASSERTION(req->rq_type != LI_POISON) failed
1533 Details : imp_lock should be held while iterating over imp_sending_list for
1534 prevent destroy request after get timeout in ptlrpc_queue_wait.
1538 Description: replay-single.sh test 52 fails
1539 Details : A lock's skiplist need to be cleanup when it being unlinked
1540 from its resource list.
1544 Description: Short directio read returns full requested size rather than
1546 Details : Direct I/O operations should return actual amount of bytes
1547 transferred rather than requested size.
1549 Severity : enhancement
1551 Description: metadata RPC reduction (e.g. for rm performance)
1552 Details : decrease the amount of synchronous RPC between clients and servers
1553 by canceling conflicing lock before the operation on the client
1554 and packing thier handles into the main operation RPC to server.
1556 Severity : enhancement
1558 Description: Async OSC create to avoid the blocking unnecessarily.
1559 Details : If a OST has no remain object, system will block on the creating
1560 when need to create a new object on this OST. Now, ways use
1561 pre-created objects when available, instead of blocking on an
1562 empty osc while others are not empty. If we must block, we block
1563 for the shortest possible period of time.
1565 Severity : enhancement
1567 Description: refine locking for avoid write wrong info into lov_objid file
1568 Details : fix possible races with add new target and write/update data in
1571 --------------------------------------------------------------------------------
1573 2007-05-03 Cluster File Systems, Inc. <info@clusterfs.com>
1578 Frequency : on some architectures
1580 Description: 1.6 client sometimes fails to mount from a 1.4 MDT
1581 Details : Uninitialized flags sometimes cause configuration commands to
1585 Frequency : patchless clients only
1587 Description: missing __iget() symbol export
1588 Details : The __iget() symbol export is missing. To avoid the need for
1589 this on patchless clients the deathrow inode reaper is turned
1590 off, and we depend on the VM to clean up old inodes. This
1591 dependency was during via the fix for bug 12181.
1593 --------------------------------------------------------------------------------
1595 2007-04-19 Cluster File Systems, Inc. <info@clusterfs.com>
1597 * CONFIGURATION CHANGE. This version of Lustre WILL NOT
1598 INTEROPERATE with older versions automatically. In many cases a
1599 special upgrade step is needed. Please read the
1600 user documentation before upgrading any part of a live system.
1601 * WIRE PROTOCOL CHANGE from previous 1.6 beta versions. This
1602 version will not interoperate with 1.6 betas before beta5 (1.5.95).
1603 * WARNING: Lustre configuration and startup changes are required with
1604 this release. See https://mail.clusterfs.com/wikis/lustre/MountConf
1610 Severity : enhancement
1612 Description: MountConf
1613 Details : Lustre configuration is now managed via mkfs and mount
1614 commands instead of lmc and lconf. New obd types (MGS, MGC)
1615 are added for dynamic configuration management. See
1616 https://mail.clusterfs.com/wikis/lustre/MountConf for
1619 Severity : enhancement
1621 Description: dynamic OST addition
1622 Details : OSTs can now be added to a live filesystem
1624 Severity : enhancement
1626 Description: startup order invariance
1627 Details : MDTs and OSTs can be started in any order. Clients only
1628 require the MDT to complete startup.
1630 Severity : enhancement
1632 Description: parallel, asynchronous orphan cleanup
1633 Details : orphan cleanup is now performed in separate threads for each
1634 OST, allowing parallel non-blocking operation.
1636 Severity : enhancement
1638 Description: optimized stripe assignment
1639 Details : stripe assignments are now made based on ost space available,
1640 ost previous usage, and OSS previous usage, in order to try
1641 to optimize storage space and networking resources.
1643 Severity : enhancement
1645 Description: Permanently set tunables
1646 Details : All writable /proc/fs/lustre tunables can now be permanently
1647 set on a per-server basis, at mkfs time or on a live system.
1649 Severity : enhancement
1651 Description: Lustre message v2
1652 Details : Add lustre message format v2.
1654 Severity : enhancement
1656 Description: client OST exclusion list
1657 Details : Clients can be started with a list of OSTs that should be
1658 declared "inactive" for known non-responsive OSTs.
1662 Description: SPEC SFS validation failure on NFS v2 over lustre.
1663 Details : Changes the blocksize for regular files to be 2x RPC size,
1664 and not depend on stripe size.
1666 Severity : enhancement
1668 Description: Multiple MD RPCs in flight.
1669 Details : Further unserialise some read-only MDS RPCs - learn about intents.
1670 To avoid overly-overloading MDS, introduce a limit on number of
1671 MDS RPCs in flight for a single client and add /proc controls
1672 to adjust this limit.
1674 Severity : enhancement
1676 Description: client read/write statistics
1677 Details : Add client read/write call usage stats for performance
1678 analysis of user processes.
1679 /proc/fs/lustre/llite/*/offset_stats shows non-sequential
1680 file access. extents_stats shows chunk size distribution.
1681 extents_stats_per_process show chunk size distribution per
1684 Severity : enhancement
1686 Description: mds statistics
1687 Details : Add detailed mds operations statistics in
1688 /proc/fs/lustre/mds/*/stats.
1692 Description: Failure of copying files with lustre special EAs.
1693 Details : Client side always return success for setxattr call for lustre
1694 special xattr (currently only "trusted.lov").
1699 Description: Refcount LNET uuids
1700 Details : The global LNET uuid list grew linearly with every startup;
1701 refcount repeated list entries instead of always adding to
1704 Severity : enhancement
1706 Description: Dynamic service threads
1707 Details : Within a small range, start extra service threads
1708 automatically when the request queue builds up.
1711 Frequency : mixed-endian client/server environments
1713 Description: mixed-endian crashes
1714 Details : The new msg_v2 system had some failures in mixed-endian
1717 Severity : enhancement
1719 Description: Easy OST removal
1720 Details : OSTs can be permanently deactivated with e.g. 'lctl
1721 conf_param lustre-OST0001.osc.active=0'
1723 Severity : enhancement
1725 Description: MGS proc entries
1726 Details : Added basic proc entries for the MGS showing what filesystems
1729 Severity : enhancement
1731 Description: provide MGS failover
1732 Details : Added config lock reacquisition after MGS server failover.
1734 Severity : enhancement
1736 Description: add Linux 2.4 support
1737 Details : Added support for RHEL 2.4.21 kernel for 1.6 servers and clients
1741 Description: a large application tries to do I/O to the same resource and dies
1742 in the middle of it.
1743 Details : Check the req->rq_arrival time after the call to
1744 ost_brw_lock_get(), but before we do anything about
1745 processing it & sending the BULK transfer request. This
1746 should help move old stale pending locks off the queue as
1747 quickly as obd_timeout.
1750 Frequency : when an incorrect nid is specified during startup
1752 Description: ptlrpc connect to non-existant node causes kernel crash
1753 Details : LNET can't be re-entered from an event callback, which
1754 happened when we expire a message after the export has been
1755 cleaned up. Instead, hand the zombie cleanup off to another
1758 Severity : enhancement
1760 Description: plain/inodebits lock performance improvement
1761 Details : Grouping plain/inodebits in granted list by their request modes
1762 and bits policy, thus improving the performance of search through
1766 Frequency : only if OST filesystem is corrupted
1768 Description: client incorrectly hits assertion in ptlrpc_replay_req()
1769 Details : for a short time RPCs with bulk IO are in the replay list,
1770 but replay of bulk IOs is unimplemented. If the OST filesystem
1771 is corrupted due to disk cache incoherency and then replay is
1772 started it is possible to trip an assertion. Avoid putting
1773 committed RPCs into the replay list at all to avoid this issue.
1776 Frequency : liblustre (e.g. catamount) on a large cluster with >= 8 OSTs/OSS
1778 Description: System hang on startup
1779 Details : This bug allowed the liblustre (e.g. catamount) client to
1780 return to the app before handling all startup RPCs. This
1781 could leave the node unresponsive to lustre network traffic
1782 and manifested as a server ptllnd timeout.
1784 Severity : enhancement
1786 Description: Add "/proc/sys/lustre/debug_peer_on_timeout"
1787 Details : liblustre envirable: LIBLUSTRE_DEBUG_PEER_ON_TIMEOUT
1788 boolean to control whether to print peer debug info when a
1789 client's RPC times out.
1792 Frequency : only for kernels with patches from Lustre below 1.4.3
1794 Description: Remove old rdonly API
1795 Details : Remove old rdonly API which unused from at least lustre 1.4.3
1798 Frequency : only for devices with external journals
1800 Description: Set external device read-only also
1801 Details : During a commanded failover stop, we set the disk device
1802 read-only while the server shuts down. We now also set any
1803 external journal device read-only at the same time.
1806 Frequency : when upgrading from 1.4 while trying to change parameters
1808 Description: The wrong (new) MDC name was used when setting parameters for
1809 upgraded MDT's. Also allows changing of OSC (and MDC)
1810 parameters if --writeconf is specified at tunefs upgrade time.
1813 Frequency : when setting specific ost indicies
1815 Description: QOS code breaks on skipped indicies
1816 Details : Add checks for missing OST indicies in the QOS code, so OSTs
1817 created with --index need not be sequential.
1819 Severity : enhancement
1821 Description: Add uninit_groups feature to ldiskfs2 to speed up e2fsck
1822 Details : The uninit_groups feature works in conjunction with the kernel
1823 filesystem code (ldiskfs2 only) and e2fsprogs-1.39-cfs6 to speed
1824 up the pass1 processing of e2fsck. This is a read-only feature
1825 in ldiskfs2 only, so older kernels and current ldiskfs cannot
1826 mount filesystems that have had this feature enabled.
1828 Severity : enhancement
1830 Description: Improve multi-block allocation algorithm to avoid fragmentation
1831 Details : The mballoc3 code (ldiskfs2 only) adds new mechanisms to improve
1832 allocation locality and avoid filesystem fragmentation.
1834 ------------------------------------------------------------------------------
1836 2007-04-01 Cluster File Systems, Inc. <info@clusterfs.com>
1838 * Support for kernels:
1839 2.4.21-47.0.1.EL (RHEL 3)
1840 2.6.5-7.283 (SLES 9)
1841 2.6.9-42.0.10.EL (RHEL 4)
1842 2.6.12.6 vanilla (kernel.org)
1843 2.6.16.27-0.9 (SLES 10)
1844 * Recommended e2fsprogs version: 1.39.cfs5
1846 * Note that reiserfs quotas are disabled on SLES 10 in this kernel
1850 Frequency : occasional, depends on client load and configuration
1851 Bugzilla : 12181, 12203
1852 Description: data loss for recently-modified files
1854 Details : In some cases it is possible that recently written or created
1855 files may not be written to disk in a timely manner (this should
1856 normally be within 30s unless client IO load is very high).
1857 The problem appears as zero-length files or files that are a
1858 multiple of 1MB in size after a client crash or client eviction
1859 that are missing data at the end of the file.
1861 This problem is more likely to be hit on clients where files are
1862 repeatedly created and unlinked in the same directory, clients
1863 have a large amount of RAM, have many CPUs, the filesystem has
1864 many OSTs, the clients are rebooted frequently, and/or the files
1865 are not accessed by other nodes after being written.
1867 The presence of the problem can be detected by looking at
1868 /proc/sys/fs/inode-state. If the first number (nr_inodes) is
1869 smaller than the second (nr_unused) then dirty files will not
1870 be flushed automatically to disk. "sync; sleep 10" should be
1871 run several times on the node before unmounting it to update
1872 Lustre (this is also safe to run on nodes without this problem).
1874 There is also a related kernel bug in the RHEL4 4 2.6.9 kernel
1875 that can cause this same problem, so customers using that kernel
1876 also need to update the kernel in addition to Lustre. In order
1877 to properly fix this bug, the RHEL3 2.4.21 kernel is also updated.
1879 It is normal that files written just before a client crash (less
1880 than 30s) may not yet have been flushed to disk, even for local
1884 Frequency : frequent on thin XT3 nodes
1886 Description: UUID collision on thin XT3 Linux nodes
1887 Details : UUIDs on Compute Node Linux XT3 nodes were not generated
1888 randomly, since we relied on an insufficiently-seeded PRNG.
1893 Description: OSS hangs after "All ost request buffers busy"
1894 Details : A deadlock between quota and journal operations caused OSS
1895 hangs after printing "All ost request buffers busy."
1898 Frequency : always on liblustre builds
1900 Description: Cleanup compiler warnings on liblustre
1903 Frequency : always on liblustre builds on XT3
1905 Description: LC_CONFIG_CDEBUG don't run while build liblustre on XT3.
1909 Description: Addition of EXT3_FEATURE_RO_COMPAT_DIR_NLINKS flag for
1910 > 32000 subdirectories
1911 Details : Add EXT3_FEATURE_RO_COMPAT_DIR_NLINK flag to
1912 EXT3_FEATURE_RO_COMPAT_SUPP. This flag will be set whenever
1913 subdirectory count crosses 32000. This will aid e2fsck to
1914 correctly handle more than 32000 subdirectories.
1917 Frequency : liblustre (e.g. catamount) on a large cluster with >= 8 OSTs/OSS
1919 Description: System hang on startup
1920 Details : This bug allowed the liblustre (e.g. catamount) client to
1921 return to the app before handling all startup RPCs. This
1922 could leave the node unresponsive to lustre network traffic
1923 and manifested as a server ptllnd timeout.
1925 Severity : enhancement
1927 Description: Add "/proc/sys/lustre/debug_peer_on_timeout"
1928 (liblustre envirable: LIBLUSTRE_DEBUG_PEER_ON_TIMEOUT)
1929 boolean to control whether to print peer debug info when a
1930 client's RPC times out.
1935 Description: make O_SYNC working on 2.6 kernels
1936 Details : 2.6 kernels use different method for mark pages for write,
1937 so need add a code to lustre for O_SYNC work.
1942 Description: Failure to close file and release space on NFS
1943 Details : Put inode details into lock acquired in ll_intent_file_open.
1944 Use mdc_intent_lock in ll_intent_open to properly
1945 detect all kind of errors unhandled by mdc_enqueue.
1950 Description: proc file read during shutdown sometimes raced obd removal,
1952 Details : Add lock to prevent obd access after proc file removal.
1955 Frequency : Only for files larger than 4GB on 32-bit clients.
1957 Description: improperly doing page alignment of locks
1958 Details : Modify lustre core code to use CFS_PAGE_* defines instead of
1959 PAGE_*. Make CFS_PAGE_MASK a 64-bit mask.
1964 Description: RPCs being resent when they shouldn't be
1965 Details : Some RPCs that should not be resent are being resent. This
1966 can cause inconsistencies in the RPC state machine. Do not
1967 resend such requests.
1970 Frequency : rare, only with NFS export
1972 Description: Crash on NFS re-export node
1973 Details : under very unusual load conditions an assertion is hit in
1974 ll_intent_file_open()
1977 Frequency : only if OST filesystem is corrupted
1979 Description: client incorrectly hits assertion in ptlrpc_replay_req()
1980 Details : for a short time RPCs with bulk IO are in the replay list,
1981 but replay of bulk IOs is unimplemented. If the OST filesystem
1982 is corrupted due to disk cache incoherency and then replay is
1983 started it is possible to trip an assertion. Avoid putting
1984 committed RPCs into the replay list at all to avoid this issue.
1989 Description: large O_DIRECT requests fail under memory pressure/fragmentation
1990 Details : Large single O_DIRECT read and write calls can fail to allocate
1991 a sufficiently large buffer to process the request. In case of
1992 allocation failure the allocation is retried with a smaller
1993 buffer and broken into smaller requests.
1995 Severity : enhancement
1997 Description: Add -o localflock option to simulate old noflock behaviour.
1998 Details : This will achieve local-only flock/fcntl locks coherentness.
2003 Description: versioning check is incomplete
2004 Details : Checking the version difference of client vs. server, report
2005 error if the gap is too big.
2010 Description: add support PG_writeback bit
2011 Details : add support for PG_writeback bit for Lustre, for more carefull
2012 work with page cache in 2.6 kernel. This also fix some deadlocks
2013 and remove hack for work O_SYNC with 2.6 kernel.
2015 Severity : enhancement
2017 Description: Add uninit_groups feature to ldiskfs2 to speed up e2fsck
2018 Details : The uninit_groups feature works in conjunction with the kernel
2019 filesystem code (ldiskfs2 only) and e2fsprogs-1.39-cfs6 to speed
2020 up the pass1 processing of e2fsck. This is a read-only feature
2021 in ldiskfs2 only, so older kernels and current ldiskfs cannot
2022 mount filesystems that have had this feature enabled.
2024 Severity : enhancement
2026 Description: Improve multi-block allocation algorithm to avoid fragmentation
2027 Details : The mballoc3 code (ldiskfs2 only) adds new mechanisms to improve
2028 allocation locality and avoid filesystem fragmentation.
2030 ------------------------------------------------------------------------------
2032 2007-02-09 Cluster File Systems, Inc. <info@clusterfs.com>
2034 * Support for kernels:
2035 2.6.9-42.0.3.EL (RHEL 4)
2036 2.6.5-7.276 (SLES 9)
2037 2.4.21-47.0.1.EL (RHEL 3)
2038 2.6.12.6 vanilla (kernel.org)
2039 2.6.16.21-0.8 (SLES10)
2040 * Recommended e2fsprogs version: 1.39.cfs2-0
2042 * The backwards-compatible /proc/sys/portals symlink has been removed
2043 in this release. Before upgrading, please ensure that you change
2044 any configuration scripts or /etc/sysctl.conf files that access
2045 /proc/sys/portals/* or sysctl portals.* to use the corresponding
2046 entry in /proc/sys/lnet or sysctl lnet.*. This change can be made
2047 in advance of the upgrade on any system running Lustre 1.4.6 or
2048 newer, since /proc/sys/lnet was added in that version.
2049 * Note that reiserfs quotas are disabled on SLES 10 in this kernel
2053 Frequency : only when quota is used
2055 Description: avoid scanning export list for quota master
2056 Details : Change the algorithms to avoid scanning export list in order
2057 to improve the efficiency.
2060 Frequency : MDS failover only, very rarely
2062 Description: "went back in time" messages on mds failover
2063 Details : The greatest transno may be lost when the current operation
2064 finishes with an error (transno==0) and the client's last_rcvd
2065 record is over-written. Save the greatest transno in the
2066 mds_last_transno for this case.
2069 Frequency : always for specific kernels and striping counts
2071 Description: client may get "Matching packet too big" without ACL support
2072 Details : Clients compiled without CONFIG_FS_POSIX_ACL get an error message
2073 when trying to access files in certain configurations. The
2074 clients should in fact be denied when mounting because they do
2075 not understand ACLs.
2078 Frequency : Cray XT3 with more than 4000 clients and multiple jobs
2080 Description: many clients connecting with IO in progress causes connect timeouts
2081 Details : Avoid synchronous journal commits to avoid delays caused by many
2082 clients connecting/disconnecting when bulk IO is in progress.
2083 Queue liblustre connect requests on OST_REQUEST_PORTAL instead of
2084 OST_IO_PORTAL to avoid delays behind potentially many pending
2088 Frequency : occasionally with multiple writers to a single file
2090 Description: shared writes to file may result in wrong size reported by stat()
2091 Details : Allow growing of kms when extent lock is cancelled
2094 Frequency : always with random mmap IO to multi-striped file
2096 Description: mmap write might be lost if we are writing to a 'hole' in stripe
2097 Details : Only if the hole is at the end of OST object so that kms is too
2098 small. Fix is to increase kms accordingly in ll_nopage.
2101 Frequency : rare, only if OST filesystem is inconsistent with MDS filesystem
2103 Description: writes to a missing object would leak memory on the OST
2104 Details : If there is an inconsistency between the MDS and OST filesystems,
2105 such that the MDS references an object that doesn't exist, writes
2106 to that object will leak memory due to incorrect cleanup in the
2107 error handling path, eventually running out of memory on the OST.
2112 Description: Creating too long symlink causes lustre errors
2113 Details : Check symlink and name lengths before sending requests to MDS.
2116 Frequency : only if flock is enabled (not on by default)
2118 Description: posix locks not released on fd closure on 2.6.9+
2119 Details : We failed to add posix locks to list of inode locks on 2.6.9+
2120 kernels, this caused such locks not to be released on fd close and
2121 then assertions on fs unmount about still used locks.
2124 Frequency : MDS failover only, very rarely
2126 Description: clients may get ASSERTION(granted_lock != NULL)
2127 Details : When request was taking a long time, and a client was resending
2128 a getattr by name lock request. The were multiple lock requests
2129 with the same client lock handle and
2130 mds_getattr_name->fixup_handle_for_resent_request found one of the
2131 lock handles but later failed with ASSERTION(granted_lock != NULL).
2136 Description: handle->h_buffer_credits > 0, assertion failure
2137 Details : h_buffer_credits is zero after truncate, causing assertion
2138 failure. This patch extends the transaction or creates a new
2142 Frequency : NFS re-export or patchless client
2143 Bugzilla : 11179, 10796
2144 Description: Crash on NFS re-export node (__d_move)
2145 Details : We do not want to hash the dentry if we don't have a lock.
2146 But if this dentry is later used in d_move, we'd hit uninitialised
2147 list head d_hash, so we just do this to init d_hash field but
2148 leave dentry unhashed.
2151 Frequency : NFS re-export or patchless client
2153 Description: NFS exports has problem with symbolic link
2154 Details : lustre client didn't properly install dentry when re-exported
2155 to NFS or running patchless client.
2158 Frequency : NFS re-export or patchless client
2160 Description: Various nfs/patchless fixes.
2161 Details : fixes reuse disconected alias for lookup process - this fixes
2162 warning "find_exported_dentry: npd != pd",
2163 fix permission error with open files at nfs.
2164 fix apply umask when do revalidate.
2167 Frequency : occasional
2169 Description: Crash on NFS re-export node
2170 Details : calling clear_page() on the wrong pointer triggered oops in
2171 generic_mapping_read().
2174 Frequency : rarely, using O_DIRECT IO
2176 Description: unaligned directio crashes client with LASSERT
2177 Details : check for unaligned buffers before trying any requests.
2180 Frequency : rarely, using CFS RAID5 patches in non-standard kernel series
2182 Description: stale data returned from RAID cache
2183 Details : If only a small amount of IO is done to the RAID device before
2184 reading it again it is possible to get stale data from the RAID
2185 cache instead of reading it from disk.
2188 Frequency : always for sles10 kernel
2190 Description: sles10 support
2191 Details : ll_follow_link: compile fixes and using of nd_set_link
2192 under newer kernels.
2195 Frequency : depends on arch, kernel and compiler version, always on sles10
2198 Description: recursive or deep enough symlinks cause stack overflow
2199 Details : getting rid of large stack-allocated variable in
2203 Frequency : depends on hardware
2205 Description: lustre write performance loss in the SLES10 kernel
2206 Details : the performance loss is caused by using of write barriers in the
2207 ext3 code. The SLES10 kernel turns barrier support on by
2208 default. The fix is to undo that change for ldiskfs.
2210 ------------------------------------------------------------------------------
2212 2006-12-09 Cluster File Systems, Inc. <info@clusterfs.com>
2214 * Support for kernels:
2215 2.6.9-42.0.3EL (RHEL 4)
2216 2.6.5-7.276 (SLES 9)
2217 2.4.21-47.0.1.EL (RHEL 3)
2218 2.6.12.6 vanilla (kernel.org)
2222 Frequency : quota enabled and large files being deleted
2224 Description: releasing more than 4GB of quota at once hangs OST
2225 Details : If a user deletes more than 4GB of files on a single OST it
2226 will cause the OST to spin in an infinite loop. Release
2227 quota in < 4GB chunks, or use a 64-bit value for 1.4.7.1+.
2232 Description: statfs data retrieved from /proc may be stale or zero
2233 Details : When reading per-device statfs data from /proc, in the
2234 {kbytes,files}_{total,free,avail} files, it may appear
2235 as zero or be out of date.
2238 Frequency : systems with MD RAID1 external journal devices
2240 Description: lconf's call to blkid is confused by RAID1 journal devices
2241 Details : Use the "blkid -l" flag to locate the MD RAID device instead
2242 of returning all block devices that match the journal UUID.
2245 Frequency : always, for aggregate stripe size over 4GB
2247 Description: "lfs setstripe" fails assertion when setting 4GB+ stripe width
2248 Details : Using "lfs setstripe" to set stripe size * stripe count over 4GB
2249 will fail the kernel with "ASSERTION(lsm->lsm_xfersize != 0)"
2252 Frequency : always if "lfs find" used on a local file/directory
2254 Description: "lfs find" segfaults if used on a local file/directory
2255 Details : The case where a directory component was not specified wasn't
2256 handled correctly. Handle this properly.
2259 Frequency : always on ppc64
2261 Description: the write to an ext3 filesystem mounted with mballoc got stuck
2262 Details : ext3_mb_generate_buddy() uses find_next_bit() which does not
2263 perform endianness conversion.
2266 Frequency : rarely (truncate to non-zero file size after write under load)
2267 Bugzilla : 10730, 10687
2268 Description: Files padded with zeros to next 4K multiple
2269 Details : With filesystems mounted using the "extents" option (2.6 kernels)
2270 it is possible that files that are truncated to a non-zero size
2271 immediately after being written are filled with zero bytes beyond
2272 the truncated size. No file data is lost.
2274 Severity : enhancement
2276 Description: Allow recovery/failover for liblustre clients.
2277 Details : liblustre clients were unaware of failover configurations until
2280 Severity : enhancement
2282 Description: user file locks should fail when not mounting with flock option
2283 Details : Set up an error-returning stub in ll_file_operations.lock field
2284 to prevent incorrect behaviour when client is mounted without
2285 flock option. Also, set up properly f_op->flock field for
2289 Frequency : always on ia64
2291 Description: "lfs df" loops on printing out MDS statfs information
2292 Details : The obd_ioctl_data was not initialized and in some systems
2293 this caused a failure during the ioctl that did not return
2294 an error. Initialize the struct and return an error on failure.
2297 Frequency : SLES 9 only
2299 Description: Error of copying files with lustre special EAs as root
2300 Details : Client side always return success for setxattr call for lustre
2301 special xattr (currently only "trusted.lov").
2304 Frequency : rarely on clusters with both ia64+i386 clients
2306 Description: ia64+i686 clients doing shared IO on the same file may LBUG
2307 Details : In rare cases when both ia64+i686 (or other mixed-PAGE_SIZE)
2308 clients are doing concurrent writes to the same file it is
2309 possible that the ia64 clients may LASSERT because the OST
2310 extent locks are not PAGE_SIZE aligned. Ensure that grown
2311 locks are always aligned on the request boundary.
2314 Frequency : specific use, occasional
2316 Description: Overwriting in use executable truncates on-disk binary image
2317 Details : If one node attempts to overwrite an executable in use by
2318 another node, we now correctly return ETXTBSY instead of
2319 truncating the file.
2321 Severity : enhancement
2323 Description: Async OSC create to avoid the blocking unnecessarily.
2324 Details : If a OST has no remain object, system will block on the creating
2325 when need to create a new object on this OST. Now, ways use
2326 pre-created objects when available, instead of blocking on an
2327 empty osc while others are not empty. If we must block, we block
2328 for the shortest possible period of time.
2333 Description: chmod on Lustre root is propagated to other clients
2334 Details : Re-validate root's dentry in ll_lookup_it to avoid having it
2335 invalid by the follow_mount time.
2338 Frequency : liblustre clients only
2340 Description: Race in 'instant cancel' lock handling could lead to such locks
2341 never to be granted in case of SMP MDS
2342 Details : Do not destroy not yet granted but cbpending locks in
2346 Frequency : replay/resend of open
2348 Description: non null lock assetion failure in mds_intent_policy
2349 Details : Trying to replay/resend lockless open requests resulted in
2350 mds_open() returning 0 with no lock. Now it sets a flag if
2351 a lock is going to be returned.
2353 Severity : enhancement
2355 Description: Checksum enhancements
2356 Details : New checksum enhancements allow for resending RPCs that failed
2359 Severity : enhancement
2361 Description: Tunables on number of dirty pages in cacche
2362 Details : Allow to set limit on number of dirty pages cached.
2367 Description: client crash on unmount - lock still has references
2368 Details : In some error handling cases it was possible to leak a lock
2369 reference on a client while accessing a file. This was not
2370 harmful to the client during operation, but would cause the
2371 client to crash when the filesystem is unmounted.
2374 Frequency : specific case, rare
2376 Description: ETXTBSY on mds though file not in use by client
2377 Details : ETXTBSY is no longer incorrectly returned when attempting to
2378 chmod or chown a directory that the user previously tried to
2379 execute or a currently-executing binary.
2382 Frequency : extremely rare except on liblustre-based clients
2384 Description: Lustre space not freed when files are deleted
2385 Details : Clean up open-unlinked files after client eviction. Previously
2386 the unlink was skipped and the files remained as orphans.
2391 Description: OST failure "would be an LBUG" in waiting_locks_callback()
2392 Details : In some cases it was possible to send a blocking callback to
2393 a client doing a glimpse, even though that client didn't get
2394 a lock granted. When the glimpse lock is cancelled on the OST
2395 the freed lock is left on the waiting list and corrupted the list.
2398 Frequency : all core dumps
2400 Description: Broke core dumps to lustre
2401 Details : Negative dentry may be unhashed if parent does not have UPDATE
2402 lock, but some callers, e.g. do_coredump, expect dentry to be
2403 hashed after successful create, hash it in ll_create_it.
2405 ------------------------------------------------------------------------------
2407 2006-09-13 Cluster File Systems, Inc. <info@clusterfs.com>
2409 * Support for kernels:
2410 2.6.9-42.0.2.EL (RHEL 4)
2411 2.6.5-7.276 (SLES 9)
2412 2.4.21-40.EL (RHEL 3)
2413 2.6.12.6 vanilla (kernel.org)
2417 Frequency : always on RHEL 3
2419 Description: Number of open files grows over time
2420 Details : The number of open files grows over time, whether or not
2421 Lustre is started. This was due to a filp leak introduced
2422 by one of our kernel patches.
2424 ------------------------------------------------------------------------------
2426 08-20-2006 Cluster File Systems, Inc. <info@clusterfs.com>
2428 * Support for kernels:
2429 2.6.9-42.EL (RHEL 4)
2430 2.6.5-7.276 (SLES 9)
2431 2.4.21-40.EL (RHEL 3)
2432 2.6.12.6 vanilla (kernel.org)
2437 Bugzilla : 5719, 9635, 9792, 9684
2438 Description: OST (or MDS) trips assertions in (re)connection under heavy load
2439 Details : If a server is under heavy load and cannot reply to new
2440 connection requests before the client resends the (re)connect,
2441 the connection handling code can behave badly if two service
2442 threads are concurrently handing separate (re)connections from
2443 the same client. Add better locking to the connection handling
2444 code, and ensure that only a single connection will be processed
2445 for a given client UUID, even if the lock is dropped.
2447 Severity : enhancement
2449 Description: add TCP zero-copy support to kernel
2450 Details : Add support to the kernel TCP stack to allow zero-copy bulk
2451 sends if the hardware supports scatter-gather and checksumming.
2452 This allows socklnd to do client-write and server-read more
2453 efficiently and reduce CPU utilization from skbuf copying.
2456 Frequency : only if NFS exporting from client
2458 Description: NULL pointer deref in ll_iocontrol() if chattr mknod file
2459 Details : If setting attributes on a file created under NFS that had
2460 never been opened it would be possible to oops the client
2461 if the file had no objects.
2464 Frequency : always for liblustre
2466 Description: liblustre client does MDS+OSTs setattr RPC for each write
2467 Details : When doing a write from a liblustre client, the client
2468 incorrectly issued an RPC to the MDS and each OST the file was
2469 striped over in order to update the timestamps. When writing
2470 with small chunks and many clients this could overwhelm the MDS
2471 with RPCs. In all cases it would slow down the write because
2472 these RPCs are unnecessary.
2474 Severity : enhancement
2476 Description: allow number of MDS service threads to be changed at module load
2477 Details : It is now possible to change the number of MDS service threads
2478 running. Adding "options mds mds_num_threads={N}" to the MDS's
2479 /etc/modprobe.conf will set the number of threads for the next
2480 time Lustre is restarted (assuming the "mds" module is also
2481 reloaded at that time). The default number of threads will
2482 stay the same, 32 for most systems.
2487 Description: OST crash if filesystem is unformatted or corrupt
2488 Details : If an OST is started on a device that has never been formatted
2489 or if the filesystem is corrupt and cannot even mount then the
2490 error handling cleanup routines would dereference a NULL pointer.
2495 Description: NULL pointer deref in llap_from_page.
2496 Details : get_cache_page_nowait can return a page with NULL (or otherwise
2497 incorrect) mapping if the page was truncated/reclaimed while it was
2498 searched for. Check for this condition and skip such pages when
2499 doing readahead. Introduce extra check to llap_from_page() to
2500 verify page->mapping->host is non-NULL (so page is not anonymous).
2503 Frequency : Sometimes when using sys_sendfile
2505 Description: "page not covered by a lock" warnings from ll_readpage
2506 Details : sendfile called ll_readpage without right page locks present.
2507 Now we introduced ll_file_sendfile that does necessary locking
2508 around call to generic_file_sendfile() much like we do in
2512 Frequency : with certain MDS communication failures at client mount time
2514 Description: NULL pointer deref after failed client mount
2515 Details : a client connection request may delayed by the network layer
2516 and not be sent until after the PTLRPC layer has timed out the
2517 request. If the client fails the mount immediately it will try
2518 to clean up before the network times out the request. Add a
2519 reference from the request import to the obd device and delay
2520 the cleanup until the network drops the request.
2523 Frequency : occasionally during client (re)connect
2525 Description: assertion failure during client (re)connect
2526 Details : processing a client connection request may be delayed by the
2527 client or server longer than the client connect timeout. This
2528 causes the client to resend the connection request. If the
2529 original connection request is replied in this interval, the
2530 client may trip an assertion failure in ptlrpc_connect_interpret()
2531 which thought it would be the only running connect process.
2534 Frequency : only with obd_echo servers and clients that are rebooted
2536 Description: kernel BUG accessing uninitialized data structure
2537 Details : When running an obd_echo server it did not start the ping_evictor
2538 thread, and when a client was evicted an uninitialized data
2539 structure was accessed. Start the ping_evictor in the RPC
2540 service startup instead of the OBD startup.
2542 Severity : enhancement
2543 Bugzilla : 10193 (patchless)
2544 Description: Remove dependency on various unexported kernel interfaces.
2545 Details : No longer need reparent_to_init, exit_mm, exit_files,
2546 sock_getsockopt, filemap_populate, FMODE_EXEC, put_filp.
2549 Frequency : rare (only users of deprecated and unsupported LDAP config)
2551 Description: write_conf for zeroconf mount queried LDAP incorrectly for client
2552 Details : LDAP apparently contains 'lustreName' attributes instead of
2553 'name'. A simple remapping of the name is sufficient.
2556 Frequency : rare (only with non-default dump_on_timeout debug enabled)
2558 Description: waiting_locks_callback trips kernel BUG if client is evicted
2559 Details : Running with the dump_on_timeout debug flag turned on makes
2560 it possible that the waiting_locks_callback() can try to dump
2561 the Lustre kernel debug logs from an interrupt handler. Defer
2562 this log dumping to the expired_lock_main() thread.
2564 Severity : enhancement
2566 Description: Support NFS exporting on 2.6 kernels.
2567 Details : Implement non-rawops metadata methods for NFS server to use without
2568 changing NFS server code.
2571 Frequency : very rare (synthetic metadata workload only)
2573 Description: two racing renames might cause an MDS thread to deadlock
2574 Details : Running the "racer" program may cause one MDS thread to rename
2575 a file from being the source of a rename to being the target of
2576 a rename at exactly the same time that another thread is doing
2577 so, and the second thread has already enqueued these locks after
2578 doing a lookup of the target and is trying to relock them in
2579 order. Ensure that we don't try to re-lock the same resource.
2582 Frequency : only very large systems with liblustre clients
2584 Description: slow eviction of liblustre clients with the "evict_by_nid" RPC
2585 Details : Use asynchronous set_info RPCs to send the "evict_by_nid" to
2586 all OSTs in parallel. This allows the eviction of stale liblustre
2587 clients to proceed much faster than if they were done in series,
2588 and also offers similar improvements for other set_info RPCs.
2593 Description: excessive CPU usage during initial read phase on client
2594 Details : During the initial read phase on a client, it would agressively
2595 retry readahead on the file, consuming too much CPU and impacting
2596 performance (since 1.4.5.8). Improve the readahead algorithm
2597 to avoid this, and also improve some other common cases (read
2598 of small files in particular, where "small" is files smaller than
2599 /proc/fs/lustre/llite/*/max_read_ahead_whole_mb, 2MB by default).
2604 Description: MDS crash when receiving packet with unknown intent.
2605 Details : Do not LBUG in unknown intent case, just return -EFAULT
2607 Severity : enhancement
2608 Bugzilla : 9293, 9385
2609 Description: MDS RPCs are serialised on client. This is unnecessary for some.
2610 Details : Do not serialize getattr (non-intent version) and statfs.
2613 Frequency : occasional, when OST network is overloaded/intermittent
2615 Description: client evicted by OST after bulk IO timeout
2616 Details : If a client sends a bulk IO request (read or write) the OST
2617 may evict the client if it is unresposive to its data GET/PUT
2618 request. This is incorrect if the network is overloaded (takes
2619 too long to transfer the RPC data) or dropped the OST GET/PUT
2620 request. There is no need to evict the client at all, since
2621 the pinger and/or lock callbacks will handle this, and the
2622 client can restart the bulk request.
2625 Frequency : Always when mmapping file with no objects
2627 Description: client crashes when mmapping file with no objects
2628 Details : Check that we actually have objects in a file before doing any
2629 operations on objects in ll_vm_open, ll_vm_close and
2635 Description: Request leak when working with deleted CWD
2636 Details : Introduce advanced request refcount tracking for requests
2637 referenced from lustre intent.
2639 Severity : Enhancement
2641 Description: Cache open file handles on client.
2642 Details : MDS now will return special lock along with openhandle, if
2643 requested and client is allowed to hold openhandle, even if unused,
2644 until such a lock is revoked. Helps NFS a lot, since NFS is opening
2645 closing files for every read/write openration.
2647 Severity : Enhancement
2649 Description: Cache open negative dentries on client when possible.
2650 Details : Guard negative dentries with UPDATE lock on parent dir, drop
2651 negative dentries on lock revocation.
2656 Description: Remounting a client read-only wasn't possible with a zconf mount
2657 Details : It wasn't possible to remount a client read-only with llmount.
2659 Severity : enhancement
2660 Description: Include MPICH 1.2.6 Lustre ADIO interface patch
2661 Details : In lustre/contrib/ or /usr/share/lustre in RPM a patch for
2662 MPICH is included to add Lustre-specific ADIO interfaces.
2663 This is based closely on the UFS ADIO layer and only differs
2664 in file creation, in order to allow the OST striping to be set.
2665 This is user-contributed code and not supported by CFS.
2670 Description: extended inode attributes (immutable, append-only) work improperly
2671 when 2.4 and 2.6 kernels are used on client/server or vice versa
2672 Details : Introduce kernel-independent values for these flags.
2674 Severity : enhancement
2677 Description: Allow fractional MB tunings for lustre in /proc/ filesystem.
2678 Details : Many of the /proc/ tunables can only be tuned at a megabyte
2679 granularity. Now, Fractional MB granularity is be supported,
2680 this is very useful for low memory system.
2682 Severity : enhancement
2684 Description: Getattr by fid
2685 Details : Getting a file attributes by its fid, obtaining UPDATE|LOOKUP
2686 locks, avoids extra getattr rpc requests to MDS, allows '/' to
2687 have locks and avoids getattr rpc requests for it on every stat.
2690 Frequency : Always, for filesystems larger than 2TB
2692 Description: ldiskfs crash at mount for filesystem larger than 2TB with mballoc
2693 Details : Kenrel kmalloc limits allocations to 128kB and this prevents
2694 filesystems larger than 2TB to be mounted with mballoc enabled.
2697 Frequency : Always, for 32-bit kernel without CONFIG_LBD and filesystem > 2TB
2699 Description: ldiskfs crash at mount for filesystem larger than 2TB with mballoc
2700 Details : If a 32-bit kernel is compiled without CONFIG_LBD enabled and a
2701 filesystems larger than 2TB is mounted then the kernel will
2702 silently corrupt the start of the filesystem. CONFIG_LBD is
2703 enabled for all CFS-supported kernels, but the possibility of
2704 this happening with a modified kernel config exists.
2706 Severity : enhancement
2708 Description: add client O_DIRECT support for 2.6 kernels
2709 Details : It is now possible to do O_DIRECT reads and writes to files
2710 in the Lustre client mountpoint on 2.6 kernel clients.
2712 Severity : enhancement
2714 Description: parallel glimpse, setattr, statfs, punch, destroy requests
2715 Details : Sends glimpse, setattr, statfs, punch, destroy requests to OSTs in
2716 parallel, not waiting for response from every OST before sending
2717 a rpc to the next OST.
2722 Description: setattr vs write race when updating file timestamps
2723 Details : Client processes that update a file timestamp into the past
2724 right after writing to the file (e.g. tar) it is possible that
2725 the updated file modification time can be reset to the current
2726 time due to a race between processing the setattr and write RPC.
2728 Severity : enhancement
2730 Description: Bring 'lfs find' closer in line with regular Linux find.
2731 Details : lfs find util supports -atime, -mtime, -ctime, -maxdepth, -print,
2732 -print0 options and obtains all the needed info through the lustre
2735 Severity : enhancement
2737 Description: support up to 1024 configured devices on one node
2738 Details : change obd_dev array from statically allocated to dynamically
2739 allocated structs as they are first used to reduce memory usage
2744 Description: Flush dirty partially truncated pages during truncate
2745 Details : Immediatelly flush partially truncated pages in filter_setattr,
2746 this way we completely avoid having any pages in page cache on OST
2747 and can retire ugly workarounds during writes to flush such pages.
2752 Description: i_sem vs transaction deadlock in mds_obd_destroy during unlink.
2753 Details : protect inode from truncation within vfs_unlink() context
2754 just take a reference before calling vfs_unlink() and release it
2755 when parent's i_sem is free.
2760 Description: last_id value checked outside lock on OST caused LASSERT failure
2761 Details : If there were multiple MDS->OST object precreate requests in
2762 flight, it was possible that the OST's last object id was checked
2763 outside a lock and incorrectly tripped an assertion. Move checks
2764 inside locks, and discard old precreate requests.
2767 Frequency : always, if extents are used on OSTs
2769 Description: index ei_leaf_hi (48-bit extension) is not zeroed in extent index
2770 Details : OSTs using the extents format would not zero the high 16 bits of
2771 the index physical block number. This is not a problem for any
2772 OST filesystems smaller than 16TB, and no kernels support ext3
2773 filesystems larger than 16TB yet. This is fixed in 1.4.7 (all
2774 new/modified files) and can be fixed for existing filesystems
2775 with e2fsprogs-1.39-cfs1.
2780 Description: import connection selection may be incorrect if timer wraps
2781 Details : Using a 32-bit jiffies timer with HZ=1000 may cause backup
2782 import connections to be ignored if the 32-bit jiffies counter
2783 wraps. Use a 64-bit jiffies counter.
2786 Frequency : very large clusters immediately after boot
2788 Description: LNET request buffers exhausted under heavy short-term load
2789 Details : If a large number of client requests are generated on a service
2790 that has previously never seen so many requests it is possible
2791 that the request buffer growth cannot keep up with the spike in
2792 demand. Instead of dropping incoming requests, they are held in
2793 the LND until the RPC service can accept more requests.
2796 Frequency : Sometimes during replay
2798 Description: Assertion failure in ll_local_open after replay.
2799 Details : If replay happened on an open request reply before we were able
2800 to set replay handler, reply will become not swabbed tripping the
2801 assertion in ll_local_open. Now we set the handler right after
2802 recognising of open request
2805 Frequency : very rare
2807 Description: kernel reports "badness in vsnprintf"
2808 Details : Reading from the "recovery_status" /proc file in small chunks
2809 may cause a negative length in lprocfs_obd_rd_recovery_status()
2810 call to vsnprintf() (which is otherwise harmless). Exit early
2811 if there is no more space in the output buffer.
2813 Severity : enhancement
2815 Description: clear OBD RPC statistics by writing to them
2816 Details : It is now possible to clear the OBD RPC statistics by writing
2817 to the "stats" file.
2822 Description: Client mtime is not the same on different clients after utimes
2823 Details : In some cases, the client was using the utimes() syscall on
2824 a file cached on another node. The clients now validate the
2825 ctime from the MDS + OSTs to determine which one is right.
2830 Description: Inability to activate failout mode
2831 Details : lconf script incorrectly assumed that in pythong string's numeric
2832 value is used in comparisons.
2835 Frequency : always with multiple stripes per file
2837 Description: Inefficient object allocation for mutli-stripe files
2838 Details : When selecting which OSTs to stripe files over, for files with
2839 a stripe count that divides evenly into the number of OSTs,
2840 the MDS is always picking the same starting OST for each file.
2841 Return the OST selection heuristic to the original design.
2846 Description: mount failures may take full timeout to return an error
2847 Details : Under some heavy load conditions it is possible that a
2848 failed mount can wait for the full obd_timeout interval,
2849 possibly several minutes, before reporting an error.
2850 Instead return an error as soon as the status is known.
2852 Frequency : quota enabled and large files being deleted
2854 Description: releasing more than 4GB of quota at once hangs OST
2855 Details : If a user deletes more than 4GB of files on a single OST it
2856 will cause the OST to spin in an infinite loop. Release
2857 quota in < 4GB chunks, or use a 64-bit value for 1.4.7.1+.
2862 Description: statfs data retrieved from /proc may be stale or zero
2863 Details : When reading per-device statfs data from /proc, in the
2864 {kbytes,files}_{total,free,avail} files, it may appear
2865 as zero or be out of date.
2868 Frequency : systems with MD RAID1 external journal devices
2870 Description: lconf's call to blkid is confused by RAID1 journal devices
2871 Details : Use the "blkid -l" flag to locate the MD RAID device instead
2872 of returning all block devices that match the journal UUID.
2875 Frequency : always, for aggregate stripe size over 4GB
2877 Description: assertion fails when trying to use 4GB stripe size
2878 Details : Use "setstripe" to set stripe size over 4GB will fail the kernel,
2879 complaining "ASSERTION(lsm->lsm_xfersize != 0)"
2882 Frequency : always on ppc64
2884 Description: the first write on an ext3 filesystem with mballoc got stuck
2885 Details : ext3_mb_generate_buddy() uses find_next_bit() which does not
2886 perform endianness conversion.
2888 ------------------------------------------------------------------------------
2890 02-14-2006 Cluster File Systems, Inc. <info@clusterfs.com>
2892 * WIRE PROTOCOL CHANGE. This version of Lustre networking WILL NOT
2893 INTEROPERATE with older versions automatically. Please read the
2894 user documentation before upgrading any part of a live system.
2895 * WARNING: Lustre networking configuration changes are required with
2896 this release. See https://bugzilla.clusterfs.com/show_bug.cgi?id=10052
2899 * Support for kernels:
2900 2.6.9-22.0.2.EL (RHEL 4)
2901 2.6.5-7.244 (SLES 9)
2902 2.6.12.6 vanilla (kernel.org)
2905 Severity : enhancement
2906 Bugzilla : 7981/8208
2907 Description: Introduced Lustre Networking (LNET)
2908 Details : LNET is new networking infrastructure for Lustre, it includes
2909 a reorganized network configuration mode (see the user
2910 documentation for full details) as well as support for routing
2911 between different network fabrics. Lustre Networking Devices
2912 (LNDs) for the supported network fabrics have also been
2913 created for this new infrastructure.
2915 Severity : enhancement
2916 Description: Introduced Access control lists
2917 Details : clients can set ACLs on files and directories in order to have
2918 more fine-grained permissions than the standard Unix UGO+RWX.
2919 The MDS must be started with the "-o acl" mount option.
2921 Severity : enhancement
2922 Description: Introduced filesystem quotas
2923 Details : Administrators may now establish per-user quotas on the
2926 Severity : enhancement
2928 Description: Configuration change for the XT3
2929 The PTLLND is now used to run Lustre over Portals on the XT3
2930 The configure option(s) --with-cray-portals are no longer used.
2931 Rather --with-portals=<path-to-portals-includes> is used to
2932 enable building on the XT3. In addition to enable XT3 specific
2933 features the option --enable-cray-xt3 must be used.
2938 Description: Running on many-way SMP OSTs can trigger oops in llcd_send()
2939 Details : A race between allocating a new llcd and re-getting the llcd_lock
2940 allowed another thread to grab newly-allocated llcd.
2942 Severity : enhancement
2944 Description: 2.6 OST async journal commit and locking fix to improve performance
2945 Details : The filter_direct_io()+filter_commitrw_write() journal commits for
2946 2.6 kernels are now async as they already were in 2.4 kernels so
2947 that they can commit concurrently with the network bulk transfer.
2948 For block-allocated files the filter allocation semaphore is held
2949 to avoid filesystem fragmentation during allocation. BKL lock
2950 removed for 2.6 xattr operations where it is no longer needed.
2955 Description: lconf incorrectly determined whether two IP networks could talk
2956 Details : In some more complicated routing and multiple-network
2957 configurations, lconf will avoid trying to make a network
2958 connection to a disjoint part of the IP space. It was doing the
2959 math incorrectly for one set of cases.
2964 Description: Fix for potential infinite loop processing records in an llog.
2965 Details : If an llog record is corrupted/zeroed, it is possible to loop
2966 forever in llog_process(). Validate the llog record length
2967 and skip the remainder of the block on error.
2970 Frequency : occasional (liblustre only)
2972 Description: liblustre could not open files whose last component is a symlink
2973 Details : sysio_path_walk() would incorrectly pass the open intent to
2974 intermediate path components.
2977 Frequency : rare (liblustre only with non-standard tuning)
2978 Bugzilla : 7201 (7350)
2979 Description: Tuning the MDC DLM LRU size to zero triggers client LASSERT
2980 Details : llu_lookup_finish_locks() tries to set lock data on a lock
2981 after it has been released, only do this for referenced locks
2983 Severity : enhancement
2985 Description: specifying an (invalid) directory default stripe_size of -1
2986 would reset the directory default striping
2987 Details : stripe_size -1 was used internally to signal directory stripe
2988 removal, now use "all default" to signal dir stripe removal
2989 as a directory striping of "all default" is not useful
2992 Frequency : common for large clusters running liblustre clients
2994 Description: doing an ls when liblustre clients are running is slow
2995 Details : sending a glimpse AST to a liblustre client waits for every AST
2996 to time out, as liblustre clients will not respond. Since they
2997 cannot cache data we refresh the OST lock LVB from disk instead.
2999 Severity : enhancement
3001 Description: doing an ls at the same time as file IO can be slow
3002 Details : enqueue and other "small" requests can be blocked behind many
3003 large IO requests. Create a new OST IO portal for non-IO
3004 requests so they can be processed faster.
3007 Frequency : rare (only HPUX clients mounting unsupported re-exported NFS vol)
3009 Description: an HPUX NFS client would get -EACCESS when ftruncate()ing a newly
3010 created file with mode 000
3011 Details : the Linux NFS server relies on an MDS_OPEN_OWNEROVERRIDE hack to
3012 allow an ftruncate() as a non-root user to a file with mode 000.
3013 Lustre now respects this flag to disable mode checks when
3014 truncating a file owned by the user
3017 Frequency : liblustre-only, when liblustre client dies unexpectedly or becomes
3020 Description: Revoking locks from clients that went dead or catatonic might take
3022 Details : New lock flags FL_CANCEL_ON_BLOCK used by liblustre makes
3023 cancellation of such locks instant on servers without waiting for
3024 any reply from clients. Clients drops these locks when cancel
3025 notification from server is received without replying.
3028 Frequency : liblustre-only, when liblustre client dies or becomes busy
3030 Description: Doing ls on Linux clients can take a long time with active
3032 Details : Liblustre client cannot handle ASTs in timely manner, so avoid
3033 granting such locks to it in the first place if possible. Locks
3034 are taken by proxy on the OST during the read or write and
3035 dropped immediately afterward. Add connect flags handling, do
3036 not grant locks to liblustre clients for glimpse ASTs.
3038 Severity : enhancement
3040 Description: Improve read-ahead algorithm to avoid excessive IO for random reads
3041 Details : Existing read-ahead algorithm is tuned for the case of streamlined
3042 sequential reads and behaves badly with applications doing random
3043 reads. Improve it by reading ahead at least read region, and
3044 avoiding excessive large RPC for small reads.
3046 Severity : enhancement
3048 Description: Creating more than 1000 files for a single job may cause a load
3049 imbalance on the OSTs if there are also a large number of OSTs.
3050 Details : qos_prep_create() uses an OST index reseed value that is an
3051 even multiple of the number of available OSTs so that if the
3052 reseed happens in the middle of the object allocation it will
3053 still utilize the OSTs as uniformly as possible.
3058 Description: OST or MDS may oops in ping_evictor_main()
3059 Details : ping_evictor_main() drops obd_dev_lock if deleting a stale export
3060 but doesn't restart at beginning of obd_exports_timed list
3063 Severity : enhancement
3065 Description: improve by-nid export eviction on the MDS and OST
3066 Details : allow multiple exports with the same NID to be evicted at one
3067 time without re-searching the exports list.
3070 Frequency : rare, only with supplementary groups enabled on SMP 2.6 kernels
3072 Description: MDS may oops in groups_free()
3073 Details : in rare race conditions a newly allocated group_info struct is
3074 freed again, and this can be NULL. The 2.4 compatibility code
3075 for groups_free() checked for a NULL pointer, but 2.6 did not.
3078 Frequency : common for liblustre clients doing little filesystem IO
3079 Bugzilla : 9352, 7313
3080 Description: server may evict liblustre clients accessing contended locks
3081 Details : if a client is granted a lock or receives a completion AST
3082 with a blocking AST already set it would not reply to the AST
3083 for LDLM_FL_CANCEL_ON_BLOCK locks. It now replies to such ASTs.
3086 Frequency : lfs setstripe, only systems with more than 160 OSTs
3088 Description: unable to set striping with a starting offset beyond OST 160
3089 Details : llapi_create_file() incorrectly limited the starting stripe
3090 index to the maximum single-file stripe count.
3093 Frequency : LDAP users only
3095 Description: lconf did not handle in-kernel recovery with LDAP properly
3096 Details : lconf/LustreDB get_refs() is searching the wrong namespace
3098 Severity : enhancement
3100 Description: bind OST threads to NUMA nodes to improve performance
3101 Details : all OST threads are uniformly bound to CPUs on a single NUMA
3102 node and do their allocations there to localize memory access
3104 Severity : enhancement
3106 Description: llmount can determine client NID directly from Myrinet (GM)
3107 Details : the client NID code from gmnalnid was moved directly into
3108 llmount, removing the need to use this or specifying the
3109 client NID explicitly when mounting GM clients with zeroconf
3112 Frequency : if client is started with down MDS
3114 Description: if client is started with down MDS mount hangs in ptlrpc_queue_wait
3115 Details : Having an LWI_INTR() wait event (interruptible, but no timeout)
3116 will wait indefinitely in ptlrpc_queue_wait->l_wait_event() after
3117 ptlrpc_import_delayed_req() because we didn't check if the
3118 request was interrupted, and we also didn't break out of the
3119 event loop if there was no timeout
3124 Description: data loss during non-page-aligned writes to a single file from
3125 both multiple nodes and multiple threads on one node at same time
3126 Details : updates to KMS and lsm weren't protected by common lock. Resulting
3127 inconsistency led to false short-reads, that were cached and later
3128 used by ->prepare_write() to fill in partially written page,
3129 leading to data loss.
3132 Frequency : always, if lconf --abort_recovery used
3134 Description: lconf --abort_recovery fails with 'Operation not supported'
3135 Details : lconf was attempting to abort recovery on the MDT device and not
3138 Severity : enhancement
3140 Description: remove cleanup logs
3141 Details : replace lconf-generated cleanup logs with lustre internal
3142 cleanup routines. Eliminates the need for client-cleanup and
3145 Severity : enhancement
3147 Description: add support for EAs (user and system) on lustre filesystems
3148 Details : it is now possible to store extended attributes in the Lustre
3149 client filesystem, and with the user_xattr mount option it
3150 is possible to allow users to store EAs on their files also
3152 Severity : enhancement
3154 Description: Add possibility (config option) to show minimal available OST free
3156 Details : When compiled with --enable-mindf configure option, statfs(2)
3157 (and so, df) will return least minimal free space available from
3158 all OSTs as amount of free space on FS, instead of summary of
3159 free spaces of all OSTs.
3161 Severity : enhancement
3163 Description: do not expand extent locks acquired on OST-side
3164 Details : Modify ldlm_extent_policy() to not expand local locks, acquired
3165 by server: they are not cached anyway.
3168 Frequency : when mmap is used/binaries executed from Lustre
3170 Description: Unmmap pages before throwing them away from read cache.
3171 Details : llap_shrink cache now attempts to unmap pages before discarding
3172 them (if unmapping failed - do not discard). SLES9 kernel has
3173 extra checks that trigger if this unmapping is not done first.
3178 Description: lconf didn't resolve symlinks before checking to see whether a
3179 given mountpoint was already in use
3182 Frequency : when migrating failover services
3183 Bugzilla : 6395, 9514
3184 Description: When migrating a subset of services from a node (e.g. failback
3185 from a failover service node) the remaining services would
3186 time out and evict clients.
3187 Details : lconf --force (implied by --failover) sets the global obd_timeout
3188 to 5 seconds in order to quickly disconnect, but this caused
3189 other RPCs to time out too quickly. Do not change the global
3190 obd_timeout for force cleanup, only set it for DISCONNECT RPCs.
3192 Severity : enhancement
3193 Frequency : if MDS is started with down OST
3194 Bugzilla : 9439,5706
3195 Description: Allow startup/shutdown of an MDS without depending on the
3196 availability of the OSTs.
3197 Details : Asynchronously call mds_lov_synchronize during MDS startup.
3198 Add appropriate locking and lov-osc refcounts for safe
3199 cleaning. Add osc abort_inflight calls in case the
3200 synchronize never started.
3203 Frequency : occasional (Cray XT3 only)
3205 Description: root not authorized to access files in CRAY_PORTALS environment
3206 Details : The client process capabilities were not honoured on the MDS in
3207 a CRAY_PORTALS/CRAY_XT3 environment. If the file had previously
3208 been accessed by an authorized user then root was able to access
3209 the file on the local client also. The root user capabilities
3210 are now allowed on the MDS, as this environment has secure UID.
3213 Frequency : occasional
3215 Description: ldiskfs "too long searching" message happens too often
3216 Details : A debugging message (otherwise harmless) prints too often on
3217 the OST console. This has been reduced to only happen when
3218 there are fragmentation problems on the filesystem.
3223 Description: Division by zero in statfs when all OSCs are inactive
3224 Details : lov_get_stripecnt() returns zero due to incorrect order of checks,
3225 lov_statfs divides by value returned by lov_get_stripecnt().
3229 Bugzilla : 9489, 3273
3230 Description: First write from each client to each OST was only 4kB in size,
3231 to initialize client writeback cache, which caused sub-optimal
3232 RPCs and poor layout on disk for the first writen file.
3233 Details : Clients now request an initial cache grant at (re)connect time
3234 and so that they can start streaming writes to the cache right
3235 away and always do full-sized RPCs if there is enough data.
3236 If the OST is rebooted the client also re-establishes its grant
3237 so that client cached writes will be honoured under the grant.
3242 Description: Slow ls (and stat(2) syscall) on files residing on IO-loaded OSTs
3243 Details : Now I/O RPCs go to different portal number and (presumably) fast
3244 lock requests (and glimses) and other RPCs get their own service
3245 threads pool that should be able to service those RPCs
3248 Severity : enhancement
3250 Description: Ability to exchange lustre version between client and servers and
3251 issue warnings at client side if client is too old. Also for
3252 liblustre clients there is ability to refuse connection of too old
3254 Details : New 'version' field is added to connect data structure that is
3255 filled with version info. That info is later checked by server and
3259 Frequency : rare, liblustre only.
3260 Bugzilla : 9296, 9581
3261 Description: Two simultaneous writes from liblustre at offset within same page
3262 might proceed at the same time overwriting eachother with stale
3264 Details : I/O lock withing llu_file_prwv was released too early, before data
3265 actually was hitting the wire. Extended lock-holding time until
3266 server acknowledges receiving data.
3269 Frequency : extremely rare. Never observed in practice.
3271 Description: avoid generating lustre_handle cookie of 0.
3272 Details : class_handle_hash() generates handle cookies by incrementing
3273 global counter, and can hit 0 occasionaly (this is unlikely, but
3274 not impossible, because initial value of cookie counter is
3275 selected randonly). Value of 0 is used as a sentinel meaning
3276 "unassigned handle" --- avoid it. Also coalesce two critical
3277 sections in this function into one.
3279 Severity : enhancement
3281 Description: allow liblustre clients to delegate truncate locking to OST
3282 Details : To avoid overhead of locking, liblustre client instructs OST to
3283 take extent lock in ost_punch() on client's behalf. New connection
3284 flag is added to handle backward compatibility.
3286 Severity : enhancement
3287 Bugzilla : 4928, 7341, 9758
3288 Description: allow number of OST service threads to be specified
3289 Details : a module parameter allows the number of OST service threads
3290 to be specified via "options ost ost_num_threads={N}" in the
3291 OSS's /etc/modules.conf or /etc/modprobe.conf.
3295 Bugzilla : 6146, 9635, 9895
3296 Description: servers crash with bad pointer in target_handle_connect()
3297 Details : In rare cases when a client is reconnecting it was possible that
3298 the connection request was the last reference for that export.
3299 We would temporarily drop the export reference and get a new
3300 one, but this may have been the last reference and the export
3301 was just destroyed. Get new reference before dropping old one.
3303 Severity : enhancement
3304 Frequency : if client is started with failover MDS
3306 Description: Allow multiple MDS hostnames in the mount command
3307 Details : Try to read the configuration from all specified MDS
3308 hostnames during a client mount in case the "primary"
3311 Severity : enhancement
3313 Description: Stop sending data to evicted clients as soon as possible.
3314 Details : Check if the client we are about to send or are sending data to
3315 was evicted already. (Check is done every second of waiting,
3316 for which l_wait_event interface was extended to allow checking
3317 of exit condition at specified intervals).
3320 Frequency : rare, normally only when NFS exporting is done from client
3322 Description: 'bad disk LOV MAGIC: 0x00000000' error when chown'ing files
3324 Details : Make mds_get_md() recognise empty md case and set lmm size to 0.
3327 Frequency : always, if srand() is called before liblustre initialization
3329 Description: Liblustre uses system PRNG disturbing its usage by user application
3330 Details : Introduce internal to lustre fast and high-quality PRNG for
3331 lustre usage and make liblustre and some other places in generic
3332 lustre code to use it.
3334 Severity : enhancement
3335 Bugzilla : 9477, 9557, 9870
3336 Description: Verify that the MDS configuration logs are updated when xml is
3337 Details : Check if the .xml configuration logs are newer than the config
3338 logs stored on the MDS and report an error if this is the case.
3339 Request --write-conf, or allow starting with --old_conf.
3341 Severity : enhancement
3343 Description: Handle symlinks in the path when checking if Lustre is mounted.
3344 Details : Resolve intermediate symlinks when checking if a client has
3345 mounted a filesystem to avoid duplicate client mounts.
3350 Description: lconf can hit an error exception but still return success.
3351 Details : The lconf command catches the Command error exception at the top
3352 level script context and will exit with the associated exit
3353 status, but doesn't ensure that this exit status is non-zero.
3358 Description: failure of ptlrpc thread startup can cause oops
3359 Details : Starting a ptlrpc service thread can fail if there are a large
3360 number of threads or the server memory is very fragmented.
3361 Handle this without oopsing.
3364 Frequency : always, only if liblustre and non-default acceptor port was used
3366 Description: liblustre cannot connect to servers with non-default acceptor port
3367 Details : tcpnal_set_default_params() was not called and was therefore
3368 ignoring the environment varaible TCPNAL_PORT, as well as other
3369 TCPNAL_ environment variables
3374 Description: two objects could be created on the same OST for a single file
3375 Details : If an OST is down, in some cases it was possible to create two
3376 objects on a single OST for a single file. No problems other
3377 than potential performance impact and spurious error messages.
3381 Bugzilla : 5681, 9562
3382 Description: Client may oops in ll_unhash_aliases
3383 Details : Client dcache may become inconsistent in race condition.
3384 In some cases "getcwd" can fail if the current directory is
3390 Description: Inode refcounting problems in NFS export code
3391 Details : link_raw functions used to call d_instantiate without obtaining
3392 extra inode reference first.
3396 Bugzilla : 9942, 9903
3397 Description: Referencing freed requests leading to crash, memleaks with NFS.
3398 Details : We used to require that call to ll_revalidate_it was always
3399 followed by ll_lookup_it. Also with revalidate_special() it is
3400 possible to call ll_revalidate_it() twice for the same dentry
3401 even if first occurence returned success. This fix changes semantic
3402 between DISP_ENQ_COMPLETE disposition flag to mean there is extra
3403 reference on a request referred from the intent.
3404 ll_intent_release() then releases such a request.
3407 Frequency : rare, normally benchmark loads only
3409 Description: unlinked inodes were kept in memory on the client
3410 Details : If a client is repeatedly creating and unlinking files it
3411 can accumulate a lot of stale inodes in the inode slab cache.
3412 If there is no other client load running this can cause the
3413 client node to run out of memory. Instead flush old inodes
3414 from client cache that have the same inode number as a new inode.
3417 Frequency : SLES9 2.6.5 kernel and long filenames only
3418 Bugzilla : 9969, 10379
3419 Description: utime reports stale NFS file handle
3420 Details : SLES9 uses out-of-dentry names in some cases, which confused
3421 the lustre dentry revalidation. Change it to always use the
3425 Frequency : rare, unless heavy write-truncate concurrency is continuous
3426 Bugzilla : 4180, 6984, 7171, 9963, 9331
3427 Description: OST becomes very slow and/or deadlocked during object unlink
3428 Details : filter_destroy() was holding onto the parent directory lock
3429 while truncating+unlinking objects. For very large objects this
3430 may block other threads for a long time and slow overall OST
3431 responsiveness. It may also be possible to get a lock ordering
3432 deadlock in this case, or run out of journal credits because of
3433 the combined truncate+unlink. Solution is to do object truncate
3434 first in one transaction without parent lock, and then do the
3435 final unlink in a new transaction with the parent lock. This
3436 reduces the lock hold time dramatically.
3439 Frequency : rare, 2.4 kernels only
3441 Description: MDS or OST cleanup may trip kernel BUG when dropping kernel lock
3442 Details : mds_cleanup() and filter_cleanup() need to drop the kernel lock
3443 before unmounting their filesystem in order to avoid deadlock.
3444 The kernel_locked() function in 2.4 kernels only checks whether
3445 the kernel lock is held, not whether it is this process that is
3446 holding it as 2.6 kernels do.
3451 Description: MDS or OST may oops/LBUG if a client is connecting multiple times
3452 Details : The client ptlrpc code may be trying to reconnect to a down
3453 server before a previous connection attempt has timed out.
3454 Increase the reconnect interval to be longer than the connection
3455 timeout interval to avoid sending duplicate connections to
3459 Frequency : echo_client brw_test command
3461 Description: fix echo_client to work with OST preallocated code
3462 Details : OST preallocation code (5137) didn't take echo_client IO path
3463 into account: echo_client calls filter methods outside of any
3464 OST thread and, hence, there is no per-thread preallocated
3465 pages and buffers to use. Solution: hijack pga pages for IO. As
3466 a byproduct, this avoids unnecessary data copying.
3470 Bugzilla : 3555, 5962, 6025, 6155, 6296, 9574
3471 Description: Client can oops in mdc_commit_close() after open replay
3472 Details : It was possible for the MDS to return an open request with no
3473 transaction number in mds_finish_transno() if the client was
3474 evicted, but without actually returning an error. Clients
3475 would later try to replay that open and may trip an assertion
3476 Simplify the client close codepath, and always return an error
3477 from the MDS in case the open is not successful.
3480 Frequency : rare, 2.6 OSTs only
3482 Description: OST may deadlock under high load on fragmented files
3483 Details : If there was a heavy load and highly-fragmented OST filesystems
3484 it was possible to have all the OST threads deadlock waiting on
3485 allocation of biovecs, because the biovecs were not released
3486 until the entire RPC IO was completed. Instead, release biovecs
3487 as soon as they are complete to ensure forward IO progress.
3489 Severity : enhancement
3491 Description: Support for specifying external journal device at mount
3492 Details : If an OST or MDS device is formatted with an external journal
3493 device, this device major/minor is stored in the ext3 superblock
3494 and may not be valid for failover. Allow detecting and
3495 specifying the external journal at mount time.
3500 Description: Mounting an MDS with pending unlinked files may cause oops
3501 Details : target_finish_recovery() calls mds_postrecov() which returned
3502 the number of orphans unlinked. mds_lov_connect->mds_postsetup()
3503 considers this an error and immediately begins cleaning up the
3504 lov, just after starting the mds_lov process
3506 Severity : enhancement
3508 Description: Implement 'lfs df' to report actual free space on per-OST basis
3509 Details : Add sub-command 'df' on 'lfs' to report the disk space usage of
3510 MDS/OSDs. Usage: lfs df [-i][-h]. Command Options: '-i' to report
3511 usage of objects; '-h' to report in human readable format.
3513 ------------------------------------------------------------------------------
3515 08-26-2005 Cluster File Systems, Inc. <info@clusterfs.com>
3522 Description: Mounting an ldiskfs file system with mballoc may crash OST node.
3523 Details : ldiskfs mballoc code may reference an uninitialized buddy struct
3524 at startup during orphan unlinking. Instead, skip buddy update
3525 before setup, as it will be regenerated after recovery is complete.
3530 Description: If an OST is inactive, its locks might reference stale inodes.
3531 Details : lov_change_cbdata() must iterate over all namespaces, even if
3532 they are inactive to clear inode references from the lock.
3534 Severity : enhancement
3535 Frequency : occasional, if non-standard max_dirty_mb used
3537 Description: Client will block write RPCs if not enough grant
3538 Details : If a client has max_dirty_mb smaller than max_rpcs_in_flight,
3539 then the client will block writes while waiting for another RPC
3540 to complete instead of consuming its dirty limit. With change
3541 we get improved performance when max_dirty_mb is small.
3543 Severity : enhancement
3544 Bugzilla : 3389, 6253
3545 Description: Add support for supplementary groups on the MDS.
3546 Details : The MDS has an upcall /proc/fs/lustre/mds/{mds}/group_upcall
3547 (set to /usr/sbin/l_getgroups if enabled) which will do MDS-side
3548 lookups for user supplementary groups into a cache.
3552 Description: O_CREAT|O_EXCL open flags in liblustre always return -EEXIST
3553 Details : Make libsysio to not enforce O_EXCL by clearing the flag,
3554 for liblustre O_EXCL is enforced by MDS.
3558 Description: readdir never returns NULL in liblustre.
3559 Details : Corrected llu_iop_getdirentries logic, to return offset of next
3560 dentry in struct dirent.
3564 Frequency : liblustre only, depends on application IO pattern
3565 Description: liblustre clients evicted if not contacting servers
3566 Details : Don't put liblustre clients into the ping_evictor list, so
3567 they will not be evicted by the pinger ever.
3569 Severity : enhancement
3571 Description: Add ability to evict clients by NID from MDS.
3572 Details : By echoing "nid:$NID" string into
3573 /proc/fs/lustre/mds/.../evict_client client with nid that equals to
3574 $NID would be instantly evicted from this MDS and from all active
3575 OSTs connected to it.
3579 Description: Do not query file size twice, somewhat slowing stat(2) calls.
3580 Details : lookup_it_finish() used to query file size from OSTs that was not
3585 Description: service threads change working directory to that of init
3586 Details : Starting lustre service threads may pin the working directory
3587 of the parent thread, making that filesystem busy. Threads
3588 now change to the working directory of init to avoid this.
3592 Frequency : during shutdown only
3593 Description: shutdown with a failed MDS or OST can cause unmount to hang
3594 Details : Don't resend DISCONNECT messages in ptlrpc_disconnect_import()
3599 Frequency : 2.6 only
3600 Description: chmod/chown may include an extra supplementary group
3601 Details : ll{,u}_mdc_pack_op_data() does not properly initialize the
3602 supplementary group and if none is specified this is used.
3605 Bugzilla : 5479 (6816)
3607 Description: Racing open + rm can assert client in mdc_set_open_replay_data()
3608 Details : If lookup is in progress on a file that is unlinked we might try
3609 to revalidate the inode and fail in revalidate after lookup is
3610 complete and ll_file_open() enqueues the open again but
3611 it_open_error() was not checking DISP_OPEN_OPEN errors correctly.
3614 Frequency : always, if lconf --abort_recovery used
3616 Description: lconf --abort_recovery fails with 'Operation not supported'
3617 Details : lconf was attempting to abort recovery on the MDT device and not
3620 ------------------------------------------------------------------------------
3622 2005-08-08 Cluster File Systems, Inc. <info@clusterfs.com>
3627 Frequency : rare (only unsupported configurations with a node running as an
3629 Bugzilla : 6514, 5137
3630 Description: Mounting a Lustre file system on a node running as an OST could
3632 Details : OSTs now preallocates memory needed to write out data at
3633 startup, instead of when needed, to avoid having to
3634 allocate memory in possibly low memory situations.
3635 Specifically, if the file system is mounted on on OST,
3636 memory pressure could force it to try to write out data,
3637 which it needed to allocate memory to do. Due to the low
3638 memory, it would be unable to do so and the node would
3639 become unresponsive.
3641 Severity : enhancement
3643 Description: Addition of lconf --service command line option
3644 Details : lconf now accepts a '--service <arg>' option, which is
3645 shorthand for 'lconf --group <arg> --select <arg>=<hostname>'
3647 Severity : enhancement
3649 Description: Failover mode is now the default for OSTs.
3650 Details : By default, OSTs will now run in failover mode. To return to
3651 the old behaviour, add '--failout' to the lmc line for OSTs.
3653 Severity : enhancement
3655 Description: Health checks are now provided for MDS and OSTs
3656 Details : Additional detailed health check information on MSD and OSTs
3657 is now provided through the procfs health_check value.
3660 Frequency : occasional, depends on IO load
3662 Description: Disk fragmentation on the OSTs could eventually cause slowdowns
3663 after numerous create/delete cycles
3664 Details : The ext3 inode allocation policy would not allocate new inodes
3665 very well on the OSTs because there are no new directories
3666 being created. Instead we look for groups with free space if
3667 the parent directories are nearly full.
3672 Description: Network or server problems during mount may cause partially
3673 mounted clients instead of returning an error.
3674 Details : The config llog parsing code may overwrite the error return
3675 code during mount error handling, returning success instead
3681 Description: MDS can fail to allocate large reply buffers
3682 Details : After long uptimes the MDS can fail to allocate large reply
3683 buffers (e.g. zconf client mount config records) due to memory
3684 fragmentation or consumption by the buffer cache. Preallocate
3685 some large reply buffers so that these replies can be sent even
3686 under memory pressure.
3690 Frequency : rare (liblustre)
3691 Description: fsx running with liblustre complained that using truncate() to
3692 extend the file doesn't work. This patch corrects that issue.
3693 Details : This is the liblustre equivalent of the fix for bug 6196. Fixes
3694 ATTR_SIZE and lsm use in llu_setattr_raw.
3698 Frequency : rare, only 2.6 kernels
3699 Description: Unusual file access patterns on the MDS may result in inode
3700 data being lost in very rare circumstances.
3701 Details : Bad interaction between the ea-in-inode patch and the "no-read"
3702 code in the 2.6 kernel caused the inode and/or EA data not to
3703 be read from disk, causing single-file corruption.
3707 Frequency : rare, only 2.6 filesystems using extents
3708 Description: Heavy concurrent write and delete load may cause data corruption.
3709 Details : It was possible under high-load situations to have an extent
3710 metadata block in the block device cache from a just-unlinked
3711 file overwrite a newly-allocated data block. We now unmap any
3712 metadata buffers that alias just-allocated data blocks.
3716 Frequency : filesystems with default stripe_count larger than 77
3717 Description: lconf+mke2fs fail when formatting filesystem with > 77 stripes
3718 Details : lconf specifies an inode size of 4096 bytes when the default
3719 stripe_count is larger than 77. This conflicts with the default
3720 inode density of 1 per 4096 bytes. Allocate smaller inodes in
3721 this case to avoid pinning too much memory for large EAs.
3723 ------------------------------------------------------------------------------
3725 2005-07-07 Cluster File Systems, Inc. <info@clusterfs.com>
3730 Frequency : rare (extremely heavy IO load with hundreds of clients)
3732 Description: Client is evicted, gets IO error writing to file
3733 Details : lock ordering changes for bug 5492 reintroduced bug 3267 and
3734 caused clients to be evicted for AST timeouts. The fixes in
3735 bug 5192 mean we no longer need to have such short AST timeouts
3736 so ldlm_timeout has been increased.
3739 Frequency : occasional during --force or --failover shutdown under load
3740 Bugzilla : 5949, 4834
3741 Description: Server oops/LBUG if stopped with --force or --failover under load
3742 Details : a collection of import/export refcount and cleanup ordering
3743 issues fixed for safer force cleanup
3746 Frequency : only filesystems larger than 120 OSTs
3747 Bugzilla : 5990, 6223
3748 Description: lfs getstripe would oops on a very large filesystem
3749 Details : lov_getconfig used kfree on vmalloc'd memory
3752 Frequency : only filesystems exporting via NFS to Solaris 10 clients
3753 Bugzilla : 6242, 6243
3754 Description: reading from files that had been truncated to a non-zero size
3755 but never opened returned no data
3756 Details : ll_file_read() reads zeros from no-object files to EOF
3761 Description: A bug in MDS/OSS recovery could cause the OSS to fail an assertion
3762 Details : There's little harm in aborting MDS/OSS recovery and letting it
3763 try again, so I removed the LASSERT and return an error instead.
3765 Severity : enhancement
3767 Description: New debugging infrastructure for tracking down data corruption
3768 Details : The I/O checksum code was replaced to: (a) control it at runtime,
3769 (b) cover more of the client-side code path, and (c) try to narrow
3770 down where problems occurred
3774 Bugzilla : 3819, 4364, 4397, 6313
3775 Description: Racing close and eviction MDS could cause assertion in mds_close
3776 Details : It was possible to get multiple mfd references during close and
3777 client eviction, leading to one thread referencing a freed mfd.
3779 Severity: : enhancement
3780 Bugzilla : 3262, 6359
3781 Description: Attempts to reconnect to servers are now more aggressive.
3782 Details : This builds on the enhanced upcall-less recovery that was added
3783 in 1.4.2. When trying to reconnect to servers, clients will
3784 now try each server in the failover group every 10 seconds. By
3785 default, clients would previously try one server every 25 seconds.
3790 Description: After recovery, certain operations trigger a failed
3791 assertion on a client.
3792 Details : Failing over an mds, using lconf -d --failover, while a
3793 client was doing a readdir() call would cause the client to
3794 LBUG after recovery completed and the readdir() was resent.
3796 Severity : enhancement
3798 Description: Default groups are now added by lconf
3799 Details : You can now run lconf --group <servicename> without having to
3800 manually add groups with lmc.
3803 Frequency : occasional
3805 Description: Nodes with an elan id of 0 trigger a failed assertion
3808 Frequency : always when accessing e.g. tty/console device nodes
3810 Description: tty and some other devices nodes cannot be used on lustre
3811 Details : file's private_data field is used by device data and lustre
3812 values in there got lost. New field was added to struct file to
3813 store fs-specific private data.
3816 Frequency : when exporting Lustre via NFS
3818 Description: NFSD failed occasionally when looking up a path component
3819 Details : NFSD is looking up ".." which was broken in ext3 directories
3820 that had grown large enough to become hashed.
3823 Frequency : Clusters with multiple interfaces not on the same subnet
3825 Description: Nodes will repeatedly try to reconnect to an interface which it
3826 cannot reach and report an error to the log.
3827 Details : Extra peer list entries will be created by lconf with some peers
3828 unreachable. lconf now validates the peer before adding it.
3831 Frequency : Only if a default stripe is set on the filesystem root.
3833 Description: Setting a default stripe on the filesystem root prevented the
3834 filesystem from being remounted.
3835 Details : The client was sending extra request flags in the root getattr
3836 request and did not allocate a reply buffer for the dir EA.
3839 Frequency : occasional, higher if lots of files are accessed by one client
3840 Bugzilla : 6159, 6097
3841 Description: Client trips assertion regarding lsm mismatch/magic
3842 Details : While revalidating inodes the VFS looks up inodes with ifind()
3843 and in rare cases can find an inode that is being freed.
3844 The ll_test_inode() code will free the lsm during ifind()
3845 when it finds an existing inode and then the VFS later attaches
3846 this free lsm to a new inode.
3850 Bugzilla : 6422, 7030
3851 Description: MDS deadlock between mkdir and client eviction
3852 Details : Creating a new file via mkdir or mknod (starting a transaction
3853 and getting the ns lock) can deadlock with client eviction
3854 (gets ns lock and trying to finish a synchronous transaction).
3857 Frequency : occasional
3858 Description: While starting a server, the fsfilt_ext3 module could not be
3860 Details : CFS's improved ext3 filesystem is named ldiskfs for 2.6
3861 kernels. Previously, lconf would still use the ext3 name
3862 when trying to load modules. Now, it will correctly use
3863 ext3 on 2.4 and ldiskfs on 2.6.
3865 Severity : enhancement
3866 Description: The default stripe count has been changed to 1
3867 Details : The interpretation of the default stripe count (0, to lfs
3868 or lmc) has been changed to mean striping across a single
3869 OST, rather than all available. For general usage we have
3870 found a stripe count of 1 or 2 works best.
3872 Severity : enhancement
3873 Description: Add support for compiling against Cray portals.
3874 Details : Conditional compiling for some areas that are different
3878 Frequency : occasional
3879 Bugzilla : 6409, 6834
3880 Description: Creating files with an explicit stripe count may lead to
3881 a failed assertion on the MDS
3882 Details : If some OSTs are full or unavailable, creating files may
3883 trigger a failed assertion on the MDS. Now, Lustre will
3884 try to use other servers or return an error to the
3888 Frequency : occasional
3890 Description: Multiple concurrent overlapping read+write on multiple SMP nodes
3891 caused lock timeout during readahead (since 1.4.2).
3892 Details : Processes doing readahead might match a lock that hasn't been
3893 granted yet if there are overlapping and conflicting lock
3894 requests. The readahead process waits on ungranted lock
3895 (original lock is CBPENDING), while OST waits for that process
3896 to cancel CBPENDING read lock and eventually evicts client.
3898 Severity : enhancement
3900 Description: Initial enabling of flock support for clients
3901 Details : Implements fcntl advisory locking and file status functions.
3902 This feature is provided as an optional mount flag (default
3903 off), and is NOT CURRENTLY SUPPORTED. Not all types of record
3904 locking are implemented yet, and those that are are not guaranteed
3905 to be completely correct in production environments.
3906 mount -t lustre -o [flock|noflock] ...
3909 Frequency : occasional
3911 Description: OSTs running 2.4 kernels but with extents enabled might trip an
3912 assertion in the ext3 JBD (journaling) layer.
3913 Details : The b_committed_data struct is protected by the big kernel lock
3914 in 2.4 kernels, serializing journal_commit_transaction() and
3915 ext3_get_block_handle->ext3_new_block->find_next_usable_block()
3916 access to this struct. In 2.6 kernels there is finer grained
3917 locking to improve SMP performance of the JBD layer.
3921 Description: Changes the "SCSI I/O Stats" kernel patch to default to "enabled"
3926 Description: merge and cleanup kernel patches.
3929 -----------------------------------------------------------------------------
3931 2005-05-05 Cluster File Systems, Inc. <info@clusterfs.com>
3933 NOTE: Lustre 1.4.2 uses an incompatible network protocol than previous
3934 versions of Lustre. Please update all servers and clients to
3935 version 1.4.2 or later at the same time. You must also run
3936 "lconf --write-conf {config}.xml" on the MDS while it is stopped
3937 to update the configuration logs.
3939 - fix for HPUX NFS client breakage when NFS exporting Lustre (5781)
3940 - mdc_enqueue does not need max_mds_easize request buffer on send (5707)
3941 - swab llog records of type '0' so we get proper header size/idx (5861)
3942 - send llog cancel req to DLM cancel portal instead of cb portal (5515)
3943 - fix rename of one directory over another leaking an inode (5953)
3944 - avoid SetPageDirty on 2.6 (5981)
3945 - don't re-add just-being-destroyed locks to the waiting list (5653)
3946 - when creating new directories, inherit the parent's custom
3947 striping settings if present parent (3048)
3948 - flush buffers from cache before direct IO in 2.6 obdfilter (4982)
3949 - don't hold i_size_sem in ll_nopage() and ll_ap_refresh_count (6077)
3950 - don't hold client locks on temporary worklist from l_lru (5666)
3951 - handle IO errors in 2.6 obdfilter bio completion routine (6046)
3952 - automatically evict dead clients (5921)
3953 - Update file size properly in create+truncate+fstat case (6196)
3954 - Do not unhash mountpoint dentries, do not allow removal of
3956 - Avoid lock ordering deadlock issue with write/truncate (6203,5654)
3957 - reserve enough journal credits in fsfilt_start_log for setattr (4554)
3958 - ldlm_enqueue freed-export error path would always LBUG (6149,6184)
3959 - don't reference lr_lvb_data until after we hold lr_lvb_sem (6170)
3960 - don't overwrite last_rcvd if there is a *_client_add() error (6086)
3961 - Correctly handle reads of files with no objects (6243)
3962 - lctl recover will also mark a device active if deactivate used (5933)
3964 - by default create 1 inode per 4kB space on MDS, per 16kB on OSTs
3965 - allow --write-conf on an MDS with different nettype than client (5619)
3966 - don't write config llogs to MDS for mounts not from that MDS (5617)
3967 - lconf should create multiple TCP connections from a client (5201)
3968 - init scripts are now turned off by default; run chkconfig --on
3969 lustre and chkconfig --on lustrefs to use them
3970 - upcalls are no longer needed for clients to recover to failover
3972 - add --abort-recovery option to lconf to abort recovery on device
3974 - add support for an arbitrary number of OSTs (3026)
3975 - Quota support protocol changes.
3976 - forward compatibility changes to wire structs (6007)
3977 - rmmod NALs that might be loaded because of /etc/modules.conf (6133)
3978 - support for mountfsoptions and clientoptions to the Lustre LDAP (5873)
3979 - improved "lustre status" script
3980 - initialize blocksize for non-regular files (6062)
3981 - added --disable-server and --disable-client configure options (5782)
3982 - introduce a lookup cache for lconf to avoid repeated DB scans (6204)
3983 - Vanilla 2.4.29 support
3984 - increase maximum number of obd devices to 520 (6242)
3985 - remove the tcp-zero-copy patch from the suse-2.4 series (5902)
3986 - Quadrics Elan drivers are now included for the RHEL 3 2.4.21 and
3987 SLES 9 2.6.5 kernels
3988 - limit stripes per file to 160 (the maximum EA size) (6093)
3990 2005-03-22 Cluster File Systems, Inc. <info@clusterfs.com>
3993 - don't LASSERT in ll_release on NULL lld with NFS export (4655, 5760)
3994 - hold NS lock when calling handle_ast_error->del_waiting_lock (5746)
3995 - fix setattr mtime regression from lovcleanup merge (4829, 5669)
3996 - workaround for 2.6 crash in ll_unhash_aliases (5687, 5210)
3997 - small ext3 extents cleanups and fixes (5733)
3998 - improved mballoc code, several small races and bugs fixed (5733, 5638)
3999 - kernel version 43 - fix remove_suid bugs in both 2.4 and 2.6 (5695)
4000 - avoid needless client->OST connect, fix handle mismatch (5317)
4001 - fix DLM error path that led to out-of-sync client, long delays (5779)
4002 - support common vfs-enforced mount options (nodev,nosuid,noexec) (5637)
4003 - fix several locking issues related to i_size (5492,5624,5654,5672)
4004 - don't move pending lock onto export if it is already evicted (5683)
4005 - fix kernel oops when creating .foo in unlinked directory (5548)
4006 - fix deadlock in obdfilter statistics vs. object create (5811)
4007 - use time_{before,after} to avoid timer jiffies wrap (5882)
4008 - shutdown --force/--failover stability (3607,3651,4797,5203,4834)
4009 - Do not leak request if server was not able to process it (5154)
4010 - If mds_open unable to find parent dir, make that negative lookup(5154)
4011 - don't create new directories with extent-mapping (5909, 5936)
4013 - fix lustre/lustrefs init scripts for SuSE (patch from Scali, 5702)
4014 - don't hold the pinger_sem in ptlrpc_pinger_sending_on_import
4015 - change obd_increase_kms to obd_adjust_kms (up or down) (5654)
4016 - lconf, lmc search both /usr/lib and /usr/lib64 for Python libs (5800)
4017 - support for RHEL4 kernel on i686 (5773)
4018 - provide error messages when incompatible logs are encountered (5898)
4020 2005-02-18 Cluster File Systems, Inc. <info@clusterfs.com>
4021 * version 1.4.0.10 (1.4.1 release candidate 1)
4023 - don't keep a lock reference when lock is not granted (4238)
4024 - unsafe list practices (rarely) led to infinite eviction loop (4908)
4025 - add per-fs limit of Lustre pages in page cache, avoid OOM (4699)
4026 - drop import inflight refcount on signal_completed_replay error (5255)
4027 - unlock page after async write error during send (3677)
4028 - handle missing objects in filter_preprw_read properly (5265)
4029 - no transno return for symlink open, don't save no-trasno open (3440)
4030 - don't try to complete elan receive that already failed (4012)
4031 - free RPC server reply state on error (5406)
4032 - clean up thread from ptlrpc_start_thread() on error (5160)
4033 - readahead could read extra page into cache that wasn't ejected (5388)
4034 - prevent races in class_attach/setup/cleanup/detach (5260)
4035 - don't dereference de->d_inode after l_dput of de (5458)
4036 - use "int" for stripe value returned from lock_to_stripe (5544)
4037 - mballoc allocation and error-checking fixes in 2.6 (5504)
4038 - block device patches to fix I/O request sizes in 2.6 (5482)
4039 - look up hostnames for IB nals (5602)
4040 - 2.6 changed lock ordering of 2 semaphores, caused deadlock (5654)
4041 - don't start multiple acceptors for the same port (5277)
4042 - fix incorrect LASSERT in mds_getattr_name (5635)
4043 - export a proc file for general "ping" checking (5628)
4044 - fix "lfs check" to not block when the MDS is down (5628)
4046 - service request history (4965)
4047 - put {ll,lov,osc}_async_page structs in a single slab (4699)
4048 - create an "evict_client" /proc entry on OSTs, like the MDS has
4049 - fix mount usage message, return errors per mount(8) (5168)
4050 - change grep [] to grep "[]" in tests so they work in more UMLs
4051 - fix ppc64/x86_64 spec to use %{_libdir} instead of /usr/lib (5389)
4052 - remove ancient LOV_MAGIC_V0 EA support (5047)
4053 - add "disk I/Os in flight" and "I/O req time" stats in obdfilter
4054 - align r/w RPCs to PTLRPC_MAX_BRW_SIZE boundary for performance (3451)
4055 - allow readahead allocations to fail when low on memory (5383)
4056 - mmap locking landed again, after considerable improvement (2828)
4057 - add get_hostaddr() to lustreDB.py for LDAP support (5459)
4059 2004-11-23 Cluster File Systems, Inc. <info@clusterfs.com>
4062 - send OST transaction number in read/write reply to free req (4966)
4063 - don't ASSERT in ptl_send_rpc() if we run out of memory (5119)
4064 - lock /proc/sys/portals/routes internal state, avoiding oops (4827)
4065 - the watchdog thread now runs as interruptible (5246)
4066 - flock/lockf fixes (but it's still disabled, pending 5135)
4067 - don't use EXT3 constants in llite code (5094)
4068 - memory shortage at startup could cause assertion (5176)
4070 - reorganization of lov code
4071 - single portals codebase
4073 - add extents/mballoc support (5025)
4074 - direct I/O reads in the obdfilter (4048)
4075 - kernel patches from LNXI for 2.6 (bluesmoke, perfctr, mtd, kexec)
4077 tbd Cluster File Systems, Inc. <info@clusterfs.com>
4080 - send OST transaction number in read/write reply to free req (4966)
4081 - don't ASSERT in ptl_send_rpc() if we run out of memory (5119)
4082 - lock /proc/sys/portals/routes internal state, avoiding oops (4827)
4083 - the watchdog thread now runs as interruptible (5246)
4084 - handle missing objects in filter_preprw_read properly (5265)
4085 - unsafe list practices (rarely) led to infinite eviction loop (4908)
4086 - drop import inflight refcount on signal_completed_replay error (5255)
4087 - unlock page after async write error during send (3677)
4088 - return original error code on reconstructed replies (3761)
4089 - no transno return for symlink open, don't save no-trasno open (3440)
4091 - add pid to ldlm debugging output (4922)
4092 - bump the watchdog timeouts -- we can't handle 30sec yet
4093 - extra debugging for orphan dentry/inode bug (5259)
4095 2004-11-16 Cluster File Systems, Inc. <info@clusterfs.com>
4098 - fix TCP_NODELAY bug, which caused extreme perf regression (5134)
4099 - allocate qswnal tx descriptors singly to avoid fragmentation (4504)
4100 - don't LBUG on obdo_alloc() failure, use OBD_SLAB_ALLOC() (4800)
4101 - fix NULL dereference in /proc/sys/portals/routes (4827)
4102 - allow failed mdc_close() operations to be interrupted (4561)
4103 - stop precreate on OST before MDS would time out on it (4778)
4104 - don't send partial-page writes before EOF from client (4410)
4105 - discard client grant for sub-page writes on large-page clients (4520)
4106 - don't free dentries not owned by NFS code, check generation (4806)
4107 - fix lsm leak if mds_create_objects() fails (4801)
4108 - limit debug_daemon file size, always print CERROR messages (4789)
4109 - use transno after validating reply (3892)
4110 - process timed out requests if import state changes (3754)
4111 - update mtime on OST during writes, return in glimpse (4829)
4112 - add mkfsoptions to LDAP (4679)
4113 - use ->max_readahead method instead of zapping global ra (5039)
4114 - don't interrupt __l_wait_event() during strace
4116 - add software watchdogs to catch hung threads quickly (4941)
4117 - make lustrefs init script start after nfs is mounted
4118 - fix CWARN/ERROR duplication (4930)
4119 - return async write errors to application if possible (2248)
4120 - add /proc/sys/portal/memused (bytes allocated by PORTALS_ALLOC)
4121 - print NAL number in %x format (4645)
4122 - update barely-supported suse-2.4.21-171 series (4842)
4123 - support for sles 9 %post scripts
4124 - support for building 2.6 kernel-source packages
4125 - support for sles km_* packages
4127 2004-10-07 Cluster File Systems, Inc. <info@clusterfs.com>
4130 - ignore -ENOENT errors in osc_destroy (3639)
4131 - notify osc create thread that OSC is being cleaned up (4600)
4132 - add nettype argument for llmount in #5d in conf-sanity.sh (3936)
4133 - reconstruct ost_handle() like mds_handle() (4657)
4134 - create a new thread to do import eviction to avoid deadlock (3969)
4135 - let lconf resolve symlinked-to devices (4629)
4136 - don't unlink "objects" from directory with default EA (4554)
4137 - hold socknal file ref over connect in case target is down (4394)
4138 - allow more than 32000 subdirectories in a single directory (3244)
4139 - fix blocks count for O_DIRECT writes (3751)
4140 - OST returns ENOSPC from object create when no space left (4539)
4141 - don't send truncate RPC if file size isn't changing (4410)
4142 - limit OSC precreate to 1/2 of value OST considers bogus (4778)
4143 - bind to privileged port in socknal and tcpnal (3689)
4145 - rate limit CERROR/CWARN console message to avoid overload (4519)
4146 - GETFILEINFO dir ioctl returns LOV EA + MDS stat in 1 call (3327)
4147 - basic mmap support (3918)
4148 - kernel patch series update from b1_4 (4711)
4150 2004-09-16 Cluster File Systems, Inc. <info@clusterfs.com>
4153 - avoid crash during MDS cleanup with OST shut down (2775)
4154 - fix loi_list_lock/oig_lock inversion on interrupted IO (4136)
4155 - don't use bad inodes on the MDS (3744)
4156 - dynamic object preallocation to improve recovery speed (4236)
4157 - don't hold spinlock over lock dumping or change debug flags (4401)
4158 - don't zero obd_dev when it is force cleaned (3651)
4159 - print grants to console if they go negative (4431)
4160 - "lctl deactivate" will stop automatic recovery attempts (3406)
4161 - look for existing locks in ldlm_handle_enqueue() (3764)
4162 - don't resolve lock handle twice in recovery avoiding race (4401)
4163 - revalidate should check working dir is a directory (4134)
4165 - don't always mark "slow" obdfilter messages as errors (4418)
4167 2004-08-24 Cluster File Systems, Inc. <info@clusterfs.com>
4170 - don't close LustreDB during write_conf until it is done (3860)
4171 - fix typo in lconf for_each_profile (3821)
4172 - allow dumping logs from multiple threads at one time (3820)
4173 - don't allow multiple threads in OSC recovery (3812)
4174 - fix debug_size parameters (3864)
4175 - fix mds_postrecov to initialize import for llog ctxt (3121)
4176 - replace config semaphore with spinlock (3306)
4177 - be sure to send a reply for a CANCEL rpc with bad export (3863)
4178 - don't allow enqueue to complete on a destroyed export (3822)
4179 - down write_lock before checking llog header bitmap (3825)
4180 - recover from lock replay timeout (3764)
4181 - up llog sem before sending rpc (3652)
4182 - reduce ns lock hold times when setting kms (3267)
4183 - change a dlm LBUG to LASSERTF, to maybe learn something (4228)
4184 - fix NULL deref and obd_dev leak on setup error (3312)
4185 - replace some LBUG about llog ops with error handling (3841)
4186 - don't match INVALID dentries from d_lookup and spin (3784)
4187 - hold dcache_lock while marking dentries INVALID and hashing (4255)
4188 - fix invalid assertion in ptlrpc_set_wait (3880)
4190 - add libwrap support for the TCP acceptor (3996)
4191 - add /proc/sys/portals/routes for non-root route listing (3994)
4192 - allow setting MDS UUID in .xml (2580)
4193 - print the stack of a process that LBUGs (4228)
4195 2004-07-14 Cluster File Systems, Inc. <info@clusterfs.com>
4198 - don't cleanup request in ll_file_open() on failed MDS open (3430)
4199 - make sure to unset replay flag from failed open requests (3440)
4200 - if default stripe count is 0, use OST count for inode size (3636)
4201 - update parent mtime/ctime on client for create/unlink (2611)
4202 - drop dentry ref in ext3_add_link from open_connect_dentry (3266)
4203 - free recovery state on server during a forced cleanup (3571)
4204 - unregister_reply for resent reqs (3063)
4205 - loop back devices mounting and status check on 2.6 (3563)
4206 - fix resource-creation race that can provoke i_size == 0 (3513)
4207 - don't try to use bad inodes returned from MDS/OST fs lookup (3688)
4208 - more debugging for page-accounting assertion (3746)
4209 - return -ENOENT instead of asserting if ost getattr+unlink race (3558)
4210 - avoid deadlock after precreation failure (3758)
4211 - fix race and lock order deadlock in orphan handling (3450, 3750)
4212 - add validity checks when grabbing inodes from l_ast_data (3599)
4214 - add /proc/.../recovery_status to obdfilter (3428)
4215 - lightweight CDEBUG infrastructure, debug daemon (3668)
4216 - change default OSC RPC parameters to be better on small clusters
4217 - turn off OST read cache for files smaller than 32MB
4218 - install man pages and include them in rpms (3100)
4219 - add new init script for (un)mounting lustre filesystems (2593)
4220 - run chkconfig in %post for init scripts (3701)
4221 - drop scimac NAL (unmaintained)
4223 2004-06-17 Cluster File Systems, Inc. <info@clusterfs.com>
4226 - clean kiobufs before and after use (3485)
4227 - strip trailing '/'s before comparing paths with /proc/mounts (3486)
4228 - remove assertions to work around "in-flight rpcs" recovery bug (3063)
4229 - change init script to fail more clearly if not run as root (1528)
4230 - allow clients to reconnect during replay (1742)
4231 - fix ns_lock/i_sem lock ordering deadlock for kms update (3477)
4232 - don't do DNS lookups on NIDs too small for IP addresses (3442)
4233 - re-awaken ptlrpcd if new requests arrive during check_set (3554)
4234 - fix cond_resched (3554)
4235 - only evict unfinished clients after recovery (3515)
4236 - allow bulk resend, prevent data loss (3570)
4237 - dynamic ptlrpc request buffer allocation (2102)
4238 - don't allow unlinking open directory if it isn't empty (2904)
4239 - set MDS/OST threads to umask 0 to not clobber client modes (3359)
4240 - remove extraneous obd dereference causing LASSERT failure (3334)
4241 - don't use get_cycles() when creating temp. files on the mds (3156)
4242 - hold i_sem when setting i_size in ll_extent_lock() (3564)
4243 - handle EEXIST for set-stripe, set proper directory name (3336)
4245 - servers can dump a log evicting a client - lustre.dump_on_timeout=1
4246 - fix ksocknal_fmb_callback() error messages (2918)
4248 2004-05-27 Cluster File Systems, Inc. <info@clusterfs.com>
4251 - don't copy lvb into (possibly NULL) reply on error (2983)
4252 - don't deref dentry after dput, don't free lvb on error (2922)
4253 - use the kms to determine writeback rpc length (2947)
4254 - increment oti_logcookies when osc is inactive (2948)
4255 - update client's i_blocks count via lvb messages (2543)
4256 - handle intent open/close of special files properly (1557)
4257 - mount MDS with errors=remount-ro, like obdfilter (2009)
4258 - initialize lock handle to avoid ASSERT on error cleanup (3057)
4259 - don't use cancelling-locks' kms values (2947)
4260 - use highest lock extent for kms, not last one (2925)
4261 - don't dereference ERR_PTR() dentry in error handling path (3107)
4262 - fix thread race in portals_debug_dumplog() (3122)
4263 - create lprocfs device entries at setup instead of at attach (1519)
4264 - common AST error handler, don't evict client on completion race (3145)
4265 - zero nameidata in detach_mnt in 2.6 (3118)
4266 - verify d_inode after revalidate_special is valid in 2.6 (3116)
4267 - use lustre_put_super() to handle zconf unmounts in 2.6 (3064)
4268 - initialize RPC timeout timer earlier for 2.6 (3219)
4269 - don't dereference NULL reply buffer if mdc_close was never sent (2410)
4270 - print nal/nid for unknown nid (3258)
4271 - additional checks for oscc recovery before doing precreate (3284)
4272 - fix ll_extent_lock() error return code for 64-bit systems (3043)
4273 - don't crash in mdc_close for bad permissions on open (3285)
4274 - zero i_rdev for non-device files (3147)
4275 - clear page->private before handing to FS, better assertion (3119)
4276 - tune the read pipeline (3236)
4277 - fix incorrect decref of invalidated dentry (2350)
4278 - provide read-ahead stats and refine rpc in flight stats (3328)
4279 - don't hold journal transaction open across create RPC (3313)
4280 - update atime on MDS at close time (3265)
4281 - close LDAP connection when recovering to avoid server load (3315)
4282 - update iopen-2.6 patch with fixes from 2399,2517,2904 (3301)
4283 - don't leak open file on MDS after open resend (3325)
4284 - serialize filter_precreate and filter_destroy_precreated (3329)
4285 - loop device shouldn't call sync_dev() for nul device (3092)
4286 - clear page cache after eviction (2766)
4287 - resynchronize MDS->OST in background (2824)
4288 - refuse to mount the same filesystem twice on same mountpoint (3394)
4289 - allow llmount to create routes for mounting behind routers (3320)
4290 - push lock cancellation to blocking thread for glimpse ASTs (3409)
4291 - don't call osc_set_data_with_check() for TEST_LOCK matches (3159)
4292 - fix rare problem with rename on htree directories (3417)
4294 - allow default OST striping configuration per directory (1414)
4295 - fix compilation for qswnal for 2.6 kernels (3125)
4296 - increase maximum number of MDS request buffers for large systems
4297 - change liblustreapi to be useful for external progs like lfsck (3098)
4298 - increase local configuration timeout for slow disks (3353)
4299 - allow configuring ldlm AST timeout - lustre.ldlm_timeout=<seconds>
4301 2004-03-22 Cluster File Systems, Inc. <info@clusterfs.com>
4304 - fixes for glimpse AST timeouts / incorrectly 0-sized files (2818)
4305 - don't overwrite extent policy data in reply if lock was blocked (2901)
4306 - drop filter export grants atomically with removal from device (2663)
4307 - del obd_self_export from work_list in class_disconnect_exports (2908)
4308 - don't LBUG if MDS recovery times out during orphan cleanup (2530)
4309 - swab reply message in mdc_close, other PPC fixes (2464)
4310 - fix destroying of named logs (2325)
4311 - overwrite old logs when running lconf --write_conf (2264)
4312 - bump LLOG_CHUNKSIZE to 8k to allow for larger clusters (2306)
4313 - fix race in target_handle_connect (2898)
4314 - mds_reint_create() should take same inode create lock (2926)
4315 - correct journal credits calculated for CANCEL_UNLINK_LOG (2931)
4316 - don't close files for self_export to avoid uninitialized obd (2936)
4317 - allow MDS with the same name as client node (2939)
4318 - hold dentry reference for closed log files for unlink (2325)
4319 - reserve space for all logs during transactions (2059)
4320 - don't evict page beyond end of stripe extent (2925)
4321 - don't oops on a deleted current working directory (2399)
4322 - handle hard links to targets without a parent properly (2517)
4323 - don't dereference NULL lock when racing during eviction (2867)
4324 - don't grow lock extents when lots of conflicting locks (2919)
4326 2004-03-04 Cluster File Systems, Inc. <info@clusterfs.com>
4329 - account for cache space usage on clients to avoid data loss (974)
4330 - lfsck support in lustre kernel code (2349)
4331 - reduce journal credits needed for BRW writes (2370)
4332 - orphan handling to avoid losing space on client/server crashes
4333 - ptlrpcd can be blocked, stopping ALL progress (2477)
4334 - use lock value blocks to assist in proper KMS, faster stat (1021)
4335 - takes i_sem instead of DLM locks internally on obdfilter (2720)
4336 - recovery for initial connections (2355)
4337 - fixes for mds_cleanup_orphans (1934)
4338 - abort_recovery crashes MDS in b_eq (mds_unlink_orphan) (2584)
4339 - block all file creations until orphan recovery completes (1901)
4340 - client remove rq_connection from request struct (2423)
4341 - conf-sanity test_5, proper cleanup in umount log not availale (2640)
4342 - recovery timer race (2670)
4343 - mdc_close recovey bug (2532)
4344 - ptlrpc cleanup bug (2710)
4345 - mds timeout on local locks (2588)
4346 - namespace lock held during RPCs (2431)
4347 - handle interrupted sync write properly (2503)
4348 - don't try to handle a message that hasn't been replied to (2699)
4349 - client assert failure during cleanup after abort recovery (2701)
4350 - leak mdc device after failed mount (2712)
4351 - ptlrpc_check_set allows timedout requests to complete (2714)
4352 - wait for inflight reqs when ptlrpcd finishes (2710)
4353 - make sure unregistered services are removed from the srv_list
4354 - reset bulk XID's when resending them (caught by 1138 test)
4355 - unregister_bulk after timeout
4356 - fix lconf error (2694)
4357 - handle write after unfinished setstripe, stripe-only getstripe (2388)
4358 - readahead locks pages, leaves pending causing memory pressure (2673)
4359 - increase OST request buffers to 4096 on large machines (2729)
4360 - fix up permission of existing directories in simple_mkdir (2661)
4361 - init deleted item, add assertions ptlrpc_abort_inflight() (2725)
4362 - don't assign transno to errored transactions (2742)
4363 - don't delete objects on OST if given a bogus objid from MDS (2751)
4364 - handle large client PAGE_SIZE readdir on small PAGE_SIZE MDS (2777)
4365 - if rq_no_resend, then timeout request after recovery (2432)
4366 - fix MDS llog_logid record size, 64-bit array alignment (2733)
4367 - don't call usermode_helper from ptlrpcd, DEFAULT upcall (2773)
4368 - put magic in mount.lustre data, check for bad/NULL mount data (2529)
4369 - MDS recovery shouldn't delete objects that it has given out (2730)
4370 - if enqueue arrives after completion, don't clobber LVB (2819)
4371 - don't unlock pages twice when trigger_group_io returns error (2814)
4372 - don't deref NULL rq_repmsg if ldlm_handle_enqueue failed (2822)
4373 - don't write pages to disk if there was an error (1450)
4374 - don't ping imports that have recovery disabled (2676)
4375 - take buffered bytes into account when balancing socknal conn (2817)
4376 - hold a DLM lock over readdir always, use truncate_inode_pages (2706)
4377 - reconnect unlink llog connection after MDS reconnects to OST (2816)
4378 - remove little-endian swabbing of llog records (1987)
4379 - set/limit i_blksize to LL_MAX_BLKSIZE on client (2884)
4380 - retry reposting request buffers if they fail (1191)
4381 - grow extent at grant time to avoid granting a revoked lock (2809)
4382 - lock revoke doesn't evict page if covered by a second lock (2765)
4383 - disable VM readahead to avoid reading outside lock extents (2805)
4385 - return LL_SUPER_MAGIC from statfs for the filesystem type (1972)
4386 - updated kernel patches for hp-2.4.20 kernel (2681)
4388 2004-02-07 Cluster File Systems, Inc. <info@clusterfs.com>
4391 - fix truncated write corruption (2366)
4392 - fix for failed assertion in iopen_connect_dentry (1792,2517)
4394 - don't flag the ptlrpcd thread with PF_MEMALLOC (2636)
4395 - ensure len(uuid) < 37 in lmc (1171)
4396 - fix ia64 OOPS in llog_test (2255)
4397 - zero end of page at obdfilter for partial page writes (2648)
4398 - don't leave stale dentries around after renames (bug 2428)
4399 - fix timeouts when evicting a client with a single lock held (2642)
4400 - set deadline for the initial HELLO message to drain (2634)
4401 - print out dotted-quad IP addresses in the socknal (2302)
4403 - additional debugging for MDS client eviction problem (2443)
4404 - fix mkfsoptions support for osts (2603, 2604)
4406 2004-01-27 Cluster File Systems, Inc. <info@clusterfs.com>
4409 - add series for the vanilla 2.6.0 kernel
4410 - add series for the vanilla 2.4.24 kernel
4411 - add series for a cray x86/64 UL kernel drop
4412 - fix xattr patches for the vanilla 2.4.19 series
4414 - generate true UUIDs in lmc (1171)
4415 - have portals stack dumping break in UML (2466)
4416 - avoid bad dchild deref; avoid inum lock w/o creation (2362)
4417 - allocate with _NOFS in ldlm to avoid deadlock (1933)
4418 - wake callback waiting threads on client eviction (2460)
4419 - Add --ptldebug and --subsystem to lmc (1719)
4420 - update assertion to allow safe interrupt allocation
4421 - set rq_no_resend for cancel requests (2432)
4422 - recalculate ptlrpcd timeout after resend (2494)
4423 - call vfs_rmdir when removing pending directories (2368)
4424 - fix renaming a file to itself (2429)
4425 - lmc creates a default one-stripe lov (2454)
4426 - expand procfs space to handle large clusters (2326)
4427 - increase UML stack to avoid overflow
4428 - update lconf's list of debug and subsystem masks
4429 - fix lfs find --obd (2510)
4430 - /proc tunable for disabling filter read caching (2591)
4431 - stop rpm packages from altering slapd.conf (2301)
4432 - disable nagle in the socknal under 0conf (2578)
4433 - choose mds inode size based on stripe count (2572)
4434 - fix kernel-source rpm problems (2516)
4436 - add --disable-doc to avoid pdf generation (2421)
4437 - update documentation, tests, type-os, comments
4438 - avoid format warnings on ia64
4439 - remove the TOE NAL
4440 - tiny code cleanups by removing unused fields
4442 2004-01-07 Cluster File Systems, Inc. <info@clusterfs.com>
4445 - fix obvious semaphore misuse in as-yet-unused setattr path (2348)
4446 - remove the most blatant lies from BUILDING file (2371)
4447 - change default debug level to reasonable production setting
4448 - reduce client side cache size to reduce cache flush time
4449 - reduce max RPCs in flight to avoid unnecessary file fragmentation
4450 - make TCP zerocopy and pinger support enabled by default (2476)
4451 - sync writes completed after process exits caused crashes (2319)
4452 - maintain correct mount count on the MDS (2356)
4453 - backout 1557, because 2316 wasn't really fixed
4454 - better file I/O statistics gathering in /proc
4455 - don't take unnecessary, deadlock-inducing bug in readpage (2383)
4456 - another kernel patch to fix zero-copy TCP function export
4457 - don't take duplicate lock when processing re-sent getattr (2420)
4458 - lctl uses obd_self_export instead of creating new conn (2353)
4459 - MDS/OST recovery case which requires object creation asserted (2425)
4460 - move lfs from /usr/sbin to /usr/bin in packages
4461 - fix race between mds_client_add and mds_client_free (2417)
4462 - use kmalloc instead of slabs in portals (2430)
4463 - don't create duplicate records when a failover MDS is present (2442)
4464 - remove unnecessary mount age check (2332)
4465 - don't remove directory inodes from locks prematurely (2451)
4466 - don't break if MDS service name is the same as hostname (2103)
4467 - fix races in client write RPC generation when cache full (2482)
4469 2003-12-13 Cluster File Systems, Inc. <info@clusterfs.com>
4472 - remove now-unused request->rq_obd (278)
4473 - if an allocation fails, print out how much memory we've used (1933)
4474 - use PORTAL_SLAB_ALLOC for structures, to get GFP_MEMALLOC (1933)
4475 - add the "configurable stack size" patch to most series files (1256)
4476 - ability to write large log records, for 100+ OST configs (2306)
4477 - fix NULL deref when filter_prep fails (2314)
4478 - fix operator precedence error in filter_sync
4479 - dynamic allocation of socknal TX descriptors (2315)
4480 - fix a missed case in the GFP_MEMALLOC patch, can cause deadlock (2310)
4481 - fix gcc 2.96 compilation problem in xattr kernel patch (2294)
4482 - ensure that CWARN messages in Portals always get to the syslog
4483 - __init/__exit are not for prototype decls (ldlm_init/exit)
4484 - x86-64 compile warning fixes
4485 - fix gateway LMC keyword conflict (2318)
4486 - fix MDS lock inversions in getattr/reint paths (1844)
4487 - fix a rare lock re-ordering bug, which caused deadlock (2322)
4488 - fix i_sem/journal inversion in fsfilt_ext3_write_record (2306)
4489 - DLM race condition prevented some lock evictions (2328)
4490 - ENOMEM detection and retry on socknal sends (2230)
4491 - use GFP_NOFS throughout Lustre, to combat ENOMEM (2230)
4492 - move osc_rpcd into ptlrpc, for use in MDC and others (2329)
4493 - protect MDS inode fsdata with stronger locking; fixes assertion (2313)
4494 - better error messages when a client is rejected during recovery (1505)
4495 - avoid cancelling locks which were never granted, after failure (2330)
4496 - fix i_sem/journal inversion in mds_client_add (2333)
4497 - fix truncate/getattr lock cycle deadlock (2334)
4498 - use rpcd to send close; allows resend after timeout, avoid leak (1897)
4499 - fix two rare exit paths which could leak an l_lock() ref (2321)
4500 - fencepost error in MDS/OST orphan recovery (2226)
4501 - make log record alignment 8 bytes (1988)
4502 - lstripe now fails when requested offset > ost_count (2237)
4503 - ensure that all kernel series have a complete list.h (1607)
4504 - fix crashes in special-file operations (2316)
4505 - lctl create/brw OID mismatch, caused by obsolete filter loop (2339)
4507 - allow configurable automake binary, for testing new versions
4508 - small update to the lfs documentation
4510 2003-12-03 Cluster File Systems, Inc. <info@clusterfs.com>
4512 * fix negative export reference count in fsfilt_sync (2312)
4514 2003-12-01 Cluster File Systems, Inc. <info@clusterfs.com>
4515 * release candidate 0.9.1
4517 - orphans are moved into the PENDING directory for possible recovery
4518 - replayed opens now open by fid for orphan/rename safety (1042)
4519 - last close of an orphan inode generates a transno (683)
4520 - chdir() and mount() now pin the directory entry (1020)
4521 - avoid CERROR in normal ll_setattr_raw() error case (1500)
4522 - discard very old requests without processing them (1502)
4523 - remove some common, well-understood CERRORs (1505)
4524 - require O_DIRECT I/O to be page-sized to workaround IA64 crash (1609)
4525 - clear "grant" flags in OST replies until OST grant code lands (1644)
4526 - fix read performance by not clobbering i_blksize on client (1598)
4527 - fix __ldlm_handle2lock oops by not dereferencing lock after PUT (1625)
4528 - make LRU size a /proc tunable, clears locks when reduced (707)
4529 - fix some lprocfs rot that prevented ptlbd from loading (1732)
4530 - server locks take references on exports now (1558)
4531 - build fixes for 2.4.20-rh trees (1663)
4532 - return an error from lov_create if all OSCs are inactive (1751)
4533 - fix import levels when a reconnect happens without a timeout (1597)
4534 - exit early from mds_open if we get a lookup error (1749)
4535 - partial page read at EOF wouldn't wait for disk before sending (1642)
4536 - avoid NULL deref in obdfilter when reading page past EOF (1592)
4537 - avoid LASSERT in ll_intent_lock if server failed very early (1090)
4538 - fix LBUG in ll_it_open_error with rc = -2 (1861)
4539 - write/truncate lock inversion (1639)
4540 - Don't auto-load obdclass, portals modules during cleanup (1495)
4541 - fix timestamps from jumping to "now" (1763)
4542 - extra journal assertions (1648)
4543 - add an extra multiunlink test (1771)
4544 - fix read_record/write_record API (1776)
4545 - fix leak of offset_extent, possible incorrect i_size later (1772)
4546 - fix lasserts in mis-matched transnos during open-unlink testing (1541)
4547 - Debugging for the kqswnal_get_idle_tx problems (1820)
4548 - Allow recovery to be attempted multiple times (1536)
4549 - Write out MDS last_rcvd file after it is first created (1600)
4550 - Fix tx_descriptor leak in failed transmit situations (1827)
4551 - ext3 journaling fixes for assertion failure after IO error (1871)
4552 - class_export_put() on freed export after completion AST error (1896)
4553 - Fix revalidate looping in VFS (1322)
4554 - Don't access a freed export during MDS_REINT timeout (1521)
4555 - Add open-unlink recovery support on the MDS (1673,1764)
4556 - Return an error if no MDS data was read from last_rcvd (1946)
4557 - Fix for lookup "." or ".." crash on error (1932,1931,1935)
4558 - Don't setup a disk device that doesn't match exported UUID (317)
4559 - Reduce bulk RPC timeout to avoid cascading client/OST failures (1845)
4560 - avoid committing NULL handle in force close
4561 - local.sh is now a one-stripe LOV configuration
4562 - POSIX utime.4 -EPERM on FIFO not owned by user (56)
4563 - fix ext3 htree duplicate directory entry corruption (1516)
4564 - POSIX creat.13, fstat.1, open.18, stat.3 new file atime/mtime (2020)
4565 - update to new LOV EA format (2097)
4566 - interoperability for different PAGE_SIZE/wordsize (686,1821,1343,2042)
4568 2003-06-15 Phil Schwan <phil@clusterfs.com>
4571 - imports and exports cleanup too early, need refcounts (349, 879, 1045)
4572 - per-import/export recovery handling (958, 931, 959)
4573 - multiple last-rcvd slots, for serving multiple FSes (949)
4574 - connections are again shared between multiple imp/exports (963, 964)
4575 - "umount -f" would hang if any requests needed to be sent (393, 978)
4576 - avoid pinning large req buffer by copying for queued messages (989)
4577 - add "uuid" to "lctl device" command to help upcalls (991)
4578 - "open" RPCs with transnos would confuse recovery counters (1037)
4579 - do proper endian conversion of all wire messages (288, 340, 891)
4580 - remove OST bulk get LBUGs, fix ost_brw_write cleanup (1126)
4581 - call waiting locks callback from LDLM recovery thread (1127, 1151)
4582 - fix ptlrpc_connection leak in target_handle_connect (1174)
4583 - fix import refcounting bug in OST and MDS cleanup (1134)
4584 - if an invalid-at-open-time OSC returned before close(), LBUG (1150)
4585 - fix very unlikely obd_types race condition (501)
4586 - remove osc_open hack for echo_client (1187)
4587 - we leaked exports/dlmimps for forcibly disconnected clients (1143)
4588 - a failure in read_inode2 leads to deadlock (1139)
4589 - cancel ack-locks as soon as transaction is committed (1072)
4590 - fix major leaks and crashes in the bulk I/O path (937, 1057)
4591 - make sure to commitrw after any preprw to avoid deadlock (1162)
4592 - failing to execute a file in a lustre FS would lock inode (1203)
4593 - small DEBUG_REQ fix to avoid dereferencing a NULL (1227)
4594 - don't ASSERT while cleaning up an incompletely-setup obd (1248)
4595 - obd_uuid2tgt would walk off the end of the list (1255)
4596 - on IA64 the osc would give portals incorrect bulk size (1258)
4597 - fix debug daemon ioctl interface; allows daemon on ia64 (1274)
4598 - fix lock inversion caused by new llite matching code (1282)
4599 - limit the number of dirty pages on a client to 10MB (1286)
4600 - timed out locks were not being corrected cancelled (1289)
4601 - fix O_DIRECT above 4GB on IA-32 (1292)
4602 * major user-visible changes
4603 - fail out/fail over policy now controlled by the upcall (993)
4605 - add OBD_PING to check server availability and failure (954)
4606 - lustre messages are now sent in sending host order (288, 340, 891)
4607 - add eadatalen to MDS getattr reply (340)
4608 - OST read replies may contain second buffer, with per-page status (593)
4610 2003-03-11 Phil Schwan <phil@clusterfs.com>
4613 - LDLM_DEBUG macro fix, for gcc 3.2 (850)
4614 - failed open()s could cause deadlock; fixed (867, 869)
4615 - stop cancelling OST locks when files are closed (481)
4616 - overlapping XID spaces caused network corruption (851, 853)
4617 - fix unsafe fsfilt counter arithmetic; change to atomic_t
4618 - setattr_raw added, to do single-RPC, server-side setattrs
4619 - lmc/lconf syntax change for OST UUIDs
4620 - fix crashy race condition between ptlrpc_free_req and osc_close
4621 - don't use request in mdc_enqueue if we hit a timeout (889)
4622 - don't set the inode i_size for regular files from the MDS (896)
4623 - handle out of order completion AST (842)
4624 - don't LBUG if a lock request times out after receiving AST (913)
4625 - avoid d_rehash race in ll_find_alias by rehashing inside dcache_lock
4626 - if a bad lock AST arrives, send an error instead of dropping entirely
4627 - return 0 from revalidate2 if ll_intent_lock returns -EINTR (912)
4628 - fix leak in bulk IO when only partially completed (899, 900, 926)
4629 - fix O_DIRECT for ia64 (55)
4630 - (almost) eliminate Lustre-kernel-thread effects on load average (722)
4631 - C-z after timeout could hang a process forever; fixed (977)
4633 - client-side I/O cache (678, 924, 929, 941, 970)
4635 - READPAGE and SETATTRs which don't take server-side locks get
4638 2003-02-11 Phil Schwan <phil@clusterfs.com>
4641 - Fix ldlm_lock_match on the MDS to avoid matching remote locks (592)
4642 - Fix fsfilt_extN_readpage() to read a full page of directory
4643 entries, or fake the remainder if PAGE_SIZE != blocksize (500)
4644 - Avoid extra mdc_getattr() in ll_intent_lock when possible (534, 604)
4645 - Fix imbalanced LOV object allocation and out-of-bound access (469)
4646 - Most intent operations were removed, in favour of a new RPC mode
4647 that does a single RPC to the server and bypasses most of the VFS
4648 - All LDLM resource ID arrays were removed in favour of ldlm_res_id
4649 - Aggressively cancel local locks on DLM servers
4650 - mds_reint_unlink sends EA to the client if it's the last nlink.
4651 client uses that EA to unlink OST objects.
4652 - mds_reint_{rename,unlink,link} were rewritten to take ordered locks
4653 - recursive symlinks were fixed (439)
4654 - fixed NULL deref in DEBUG_REQ
4655 - filter_update_lastobjid no longer calls sync, which annoyed extN
4656 - fixed multi-client small-writes to a single file problem (445)
4657 - fixed mtime updates during file writes (607)
4658 - fixed vector writes on obdfilter causing problems when ENOSPC (670)
4659 - fixed bug in obd_brw_read/write() (under guise of testing 367)
4660 - fixed Linux OST size reporting problem (444, 656)
4661 - OST now updates object mtime with writes or setattr (607, 619)
4662 - client verifies file size before zeroing page past EOF (445)
4663 - OST now writes last allocated objid to disk with allocation (108)
4664 - LOV on echo now works (409)
4666 - mds_reint_unlink sends a new buffer, with the EA included. this
4667 buffer is only valid if body->valid & OBD_MD_FLEASIZE, which is only
4668 set if a regular file was being unlinked, and it was the last link
4669 - use PtlGet from the target for bulk writes (315)
4670 - OST now updates object mtime with writes or setattr (607, 619)
4671 - LDLM now has a grant-time callback to revalidate locked items, if
4673 - Many MDS operations were reorganized to combat race conditions
4675 - Merge b_intel branch (updated lprocfs code) - now at /proc/fs/lustre
4676 - configure check to avoid gcc version 2.96 20000731-2.96-98) (606)
4678 2003-01-06 Andreas Dilger <adilger@clusterfs.com>
4681 - Fully reactivate OST imports after reconnection (512, others)
4682 - Make sure client sees our -ENOTCONN from mds_handle (513 - partial)
4683 - More graceful error handling for truncating on dead OST (515)
4684 - Don't error out unless we're actually accessing dead stripes (474)
4685 - Fix garbage sizes when stripes are missing (410)
4686 - LRU counters were broken, causing constant lock purge (433, 432)
4687 - garbage on read from stripes with failed OSTs (441)
4688 - mark OSCs as active before reconnecting during recovery (438)
4689 - lov_enqueue and lov_cancel need to handle inactive OSTs (403)
4690 - lfind did not preserve OST order in output (443)
4691 - symlinks cause hung clients, incorrect data (439)
4692 - stop dereferencing request after dropping refcount (457)
4693 - don't LASSERT(spin_is_locked) on non-SMP (455)
4694 - fixes for many rename() bugs
4695 - fstat didn't correctly synchronize attributes (399)
4696 - server must handle lock cancellation during blocking AST prep (487)
4697 - bulk descriptors were free()d too soon (511)
4698 - fix paths in lconf, which would load incorrect modules (451, 507)
4699 - fix confusing lconf 'host not found' error message (386)
4700 - fix lock order deadlock on OST (O/R i_sem before journal ops, 478)
4701 - fix race condition in mdc_blocking_ast() for inode access (526)
4702 - fix lov_unpackmd() unpacking wrong number of stripes (537)
4703 - fix lov_set_osc_active() marking wrong OSC inactive (440)
4704 - fix bad lstripe lov_unpackmd() assertion (fix layering too) (527)
4705 - fix multiple writes of stripe MD to MDS (358, maybe 519)
4706 - fix lstripe in several ways (kernel side) (527)
4707 - fix request leak in ldlm_cli_enqueue (262)
4708 - incorrect OSC was marked inactive after OST failure
4709 - call mds_fs_cleanup before unmounting filesystem (524)
4710 - fix races between taking ns_lock and ldlm_lock_change_resource
4711 - fix races updating LOV export open file list
4712 - fix lov_enqueue error path, avoid decref-ing bad lock handle
4713 - fix recovery NULL deref in ldlm_cli_cancel_unused
4714 - fix some DLM races by using new hash table for lock handles (419)
4715 - permit the client to specify desired inodes, at replay
4716 - duplicate requests when we queue them for replay reintegration
4717 - fix last_rcvd offset calculation
4718 - sync after each recovered transaction, so we always make progress
4719 - never, not always, ERESTART requests without transnos
4720 - store the lov_desc in the MDS, so we don't depend on getlovinfo to
4722 - skip replay if the MDS says that the client is already connected
4723 - don't check for a recovery-enabled export to match lctl's UUID
4724 - don't INC_USE_COUNT for phantom exports
4725 - don't crash when cleaning up phantom exports (567)
4726 - don't double-finish or set replay data for errored mdc_open requests
4727 - abort requests when they time out, so we don't get old replies
4728 - send/receive replies for AST messages again
4729 - if the client says that it doesn't have the lock, cancel it on the
4731 - if we timeout during I/O, don't try to cancel an in-use lock; instead
4732 mark it as destroyed, it will all work out when decref is called
4733 - fix module use counts (22, 581)
4735 - ASTs now expect a reply (server cancels lock on error reply)
4737 2002-12-02 Andreas Dilger <adilger@clusterfs.com>
4740 - fix many simultaneous client startup (392)
4741 - fix dentry->d_it clobbering
4742 - credentials weren't being shipped for readdir/getattr operations
4743 - remove invalid assertions triggered during some concurrent MD
4745 - proper Lustre versions added (336, 389)
4746 - fix memory leak for create error case (398)
4747 - fix LOV locking bug that would get cli/srv out of sync
4748 - fix echo client over LOV (409)
4749 - fix dbench 2, extN refcount problem (170, 258, 356, 418)
4750 - fix double-O_EXCL intent crash (424)
4751 - avoid sending multiple lock CANCELs (352)
4753 - MDS can do multi-client recovery (modulo bugs in new code)
4755 - many updates, edits, cleanups
4757 2002-11-18 Phil Schwan <phil@clusterfs.com>
4760 - fix null d_it dereference (346)
4761 - fix full OST/dbench hang (333)
4762 - fix permission problem with file removal (286)
4763 - fix removal of OSCs from LOV when they fail
4764 - fix NULL deref during bulk timeout (214)
4765 - fix problems related to multiple filesystems on one MDS (241)
4766 - fixed serious subtle metadata locking bugs
4767 - free locks on clients when inodes are removed due to memory
4769 - fix inode pointer in lock data (285)
4770 - partial support for multiple MDS on a single host (241)
4771 - data locks weren't cancelled at clear_inode time (290, 311)
4772 - intent locks could lead to unbounded lock growth (205)
4773 - added a maximum lock count, an LRU list, and a flusher
4774 - fix multiple rename (365)
4775 - properly abstracted the echo client
4776 - OSC locked 1 byte too many; fixed
4777 - rewrote brw callback code:
4778 - fixed recovery bugs related to LOVs (306)
4779 - fixed too-many-pages-in-one-write crash (191)
4780 - fixed (again) crash in sync_io_timeout (214)
4781 - probably fixed callback-related race (385)
4783 - Add capability to MDS protocol
4784 - LDLM cancellations and callbacks on different portals
4786 2002-10-28 Andreas Dilger <adilger@clusterfs.com>
4789 - limit client IOV size to PTL_MD_MAX_IOV (611336, 191)
4790 - defer open object destruction to close time (601981, 138)
4791 - open/close OST file handle in obdo (OBD_MD_FLHANDLE) (601981, 138)
4792 - move LDLM_ENQUEUE/CONVERT back to MDS portal (625069)
4793 - abstract ll_lookup2, fix ll_revalidate2 to use abstraction (256)
4794 - don't call obd_setattr in ll_file_release for destroyed objects
4795 * protocol change to lustre_msg: move |version| and add |flags|
4796 * protocol change to osc_punch: "start" in "o_size", "end" in "o_blocks"
4797 * lock replay: for LDLM_FL_REPLAY trust client to do right thing
4798 * added replay of create, unlink, link and rename operations during
4799 MDS failover; recovery should be much more robust now
4800 * remove failed OSCs from LOVs (only lov_create uses this so far)
4801 * the lustre-HOWTO was brought (more) up to date (582544)
4803 2002-10-23 Phil Schwan <phil@clusterfs.com>
4806 - in-use dentries weren't being reused properly (617851)
4807 - prevent multiple LDLM setup (599178)
4808 - fix LOV size calculations for truncate (617853)
4809 - fix client handling of MDS intent errors (POSIX)
4810 - fix permission bug in lovstripe.c test (624321)
4811 - fix MDS thread deadlock - move LDLM handler to DLM portal (625069)
4812 - truncate past end of file could corrupt data
4813 - proper cleanup after timeouts, crashes, etc (592524, 550815)
4814 - a race in recovery could return ETIMEDOUT to apps (623947)
4815 - building outside the source directory was fixed
4816 * the lustre-HOWTO was brought (more) up to date (582544)
4817 * major progress was made on recovery functionality
4819 2002-10-10 Phil Schwan <phil@clusterfs.com>
4822 - recovery deadlock fix
4823 - rm -rf causes LBUG fix (617817)
4824 - file open by multiple tasks fix (618962)
4825 - directory permissions bugs (602707 and 620007)
4826 - journal_stop fixed with locking (611313)
4827 - O_APPEND failures resolved (618273, perhaps 614459)
4828 - lconf PATH fix (619770)
4829 - IA64 build fix (621450)
4830 - RPC buffer sizes scale with amount of memory
4832 2002-10-01 Phil Schwan <phil@clusterfs.com>
4835 - locks would be cancelled without throwing away data pages,
4836 resulting in inconsistent data (605627)
4837 - inode attributes were not always being refreshed (605627, 612449)
4838 - lconf now continues to cleanup after lctl reports an error
4839 - MDS now enforces user permissions (602707)
4840 - lprocfs cleanup fixed, but not yet enabled (614157)
4841 - fixed infinite server hang, should a client not respond to an AST
4842 - avoid going into recovery if user calls readlink() with a buffer
4843 that's too small (613941)
4844 - AST RPCs no longer require replies (614867) -- this may be changed
4845 - don't crash server if client sends an IOV that's too big (611336)
4846 - fixed lock conversion deadlock (611892)
4847 - fixed the following of symlinks (614622)
4848 * recovery: the server can remove locks from a client that dies, other
4849 clients can make progress
4850 * more extN patch fixes
4851 * compile-time configurable ptlrpc buffer allocations
4853 - collaborative read cache document
4854 - Lustre Lite Performance CDR document-in-progress
4856 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
4859 - fix typo in patch-2.4.18
4861 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
4864 - clear ptlrpc request each time in handle_incoming_request()
4865 - unlink of files now destroys the object on the OST
4867 2002-09-19 Peter Braam <braam@clusterfs.com>
4869 * add hard link support
4870 * change obdfile creation method
4871 * kernel patch changed
4873 2002-09-19 Peter Braam <braam@clusterfs.com>
4876 - stack overflow bug in extN fixed
4878 2002-09-18 Andreas Dilger <adilger@clusterfs.com>
4880 * documentation updates
4881 - add man pages for config tools
4882 - update tests/README to describe testing with new config tools
4883 - finish metadata API descriptions
4884 * bug fixes and cleanups
4885 - statfs workaround for 16TB limit
4886 - LOV stripe allocation improved, can stripe on subset of OSTs
4887 - LOV file size/IO offset was wrong for files > 4GB in size
4888 - object EA data was being dropped, caused files to be unreadable
4889 - memory overflow with non-LOV OST caused memory corruption
4890 - fixed regression tests to work with new config tools, obdfilter
4891 - fixed bug when directory size became larger than 1 block
4892 - fixed bug (for single client case) when PWD was deleted
4893 - invalidate local directory pages when doing intent-based ops
4894 - avoid LDLM oops when lock callback contained bad data
4896 2002-09-09 Andreas Dilger <adilger@clusterfs.com>
4898 * documentation updates
4899 * bug fixes and cleanups
4900 - configuration tools
4903 - 64-bit compile warnings
4904 - 64-bit internal statfs data
4906 * test_brw on persistent OST devices
4908 * lprocfs (disabled)
4910 2002-09-04 Andreas Dilger <adilger@clusterfs.com>
4912 * documentation updates
4913 * bug fixes and cleanups
4914 * configuration tools
4916 2002-08-30 Peter J. Braam <braam@clusterfs.com>
4919 * many small fixes to 0_5_4
4920 * io/network handling
4921 * thinkos in MDS operations
4923 2002-08-24 Peter J. Braam <braam@clusterfs.com>
4926 * crucial basic fixes to 0.5.3
4927 * IOR, Iozone work over Elan
4930 2002-08-07 Phil Schwan <phil@clusterfs.com>
4931 * version 0_5_3, our first alpha
4932 * we use the new Portals iovs
4933 * documentation updates
4934 * bug fixes and cleanups
4935 * small changes in the DLM wire protocol
4937 2002-07-25 Peter J. Braam <braam@clusterfs.com>
4938 * version 0_5_1 with some initial stability,
4939 * locking on MD and file I/O.
4940 * documentation updates
4941 * several bug fixes since 0.5.0
4942 * small changes in wire protocol
4944 2002-07-18 Phil Schwan <phil@clusterfs.com>
4946 * delivered as Lustre Light Alpha
4947 * fixed a crash after handling invalid MDS requests
4948 * fixed directory pages for architectures with non-4k pages sizes
4950 2002-07-11 Andreas Dilger <adilger@clusterfs.com>
4951 * release version v0_4_4
4952 * Moves TCP acceptor to be on port 2432 (unused Coda port) instead
4954 * Fixes a number of interruption problems with OST operations.
4955 * Update documentation for portals header changes
4956 * Move all wire protocol structs/defines to lustre_idl.h
4957 * Fixes symlink length bug.
4958 * Add tcpdump to repository.
4960 2002-07-05 Andreas Dilger <adilger@clusterfs.com>
4961 * release version v0_4_3
4962 * Fixes statfs for inodes on extN.
4963 * Fixes bug in runtests which would delete /etc/hosts.
4964 * Use 64-bit object IDs wherever possible (not into VFS though)
4965 Remove ost_get_info, which is unused by lustre, and out of date.
4967 2002-07-03 Peter Braam <braam@clusterfs.com>
4968 * release version v0_4_2 Fixes a lookup error (type not passed)
4969 * move forward to head of Portals
4970 * move forward to latest Lustre kernel
4972 2002-06-25 Peter Braam <braam@clusterfs.com>
4973 * release version v0_4_1. Hopefully stable on single node use.