1 tbd Sun Microsystems, Inc.
4 2.6.16.54-0.2.5 (SLES 10),
5 2.6.18-53.1.21.el5 (RHEL 5),
6 2.6.22.14 vanilla (kernel.org).
7 * Client support for unpatched kernels:
8 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
9 2.6.16 - 2.6.21 vanilla (kernel.org)
10 * Recommended e2fsprogs version: 1.40.7-sun3
11 * Note that reiserfs quotas are disabled on SLES 10 in this kernel.
12 * RHEL 4 and RHEL 5/SLES 10 clients behaves differently on 'cd' to a
13 removed cwd "./" (refer to Bugzilla 14399).
15 Severity : enhancement
17 Description: Update to RHEL5 kernel-2.6.18-53.1.21.el5.
21 Description: do not process already freed flock
22 Details : flock can possibly be freed by another thread before it reaches
23 to ldlm_flock_completion_ast.
27 Description: LBUG during stress test
28 Details : Need properly lock accesses the flock deadlock detection list.
32 Description: oops in page fault handler
33 Details : kernel page fault handler can return two special 'pages' in error case, don't
34 try dereference NOPAGE_SIGBUS and NOPAGE_OMM.
38 Description: timeout with invalidate import.
39 Details : ptlrpcd_check call obd_zombie_impexp_cull and wait request which should be
40 handled by ptlrpcd. This produce long age waiting and -ETIMEOUT
41 ptlrpc_invalidate_import and as result LASSERT.
43 Severity : enhancement
45 Description: Update to RHEL5 kernel-2.6.18-53.1.19.el5.
49 Description: enable MGS and MDT services start separately
50 Details : add a 'nomgs' option in mount.lustre to enable start a MDT with
51 a co-located MGS without starting the MGS, which is a complement
52 to 'nosvc' mount option.
56 Frequency : after recovery
57 Description: precreate to many object's after del orphan.
58 Details : del orphan st in oscc last_id == next_id and this triger growing
59 count of precreated objects. Set flag LOW to skip increase count
60 of precreated objects.
64 Frequency : rare, on clear nid stats
65 Description: ASSERTION(client_stat->nid_exp_ref_count == 0)
66 Details : when clean nid stats sometimes try destroy live entry,
67 and this produce panic in free.
71 Description: Stack overflow during MDS log replay
72 ease stack pressure by using a thread dealing llog_process.
76 Description: wait until IO finished before start new when do lock cancel.
77 Details : VM protocol want old IO finished before start new, in this case
78 need wait until PG_writeback is cleared until check dirty flag and
79 call writepages in lock cancel callback.
81 Severity : enhancement
83 Description: using special macro for print time and cleanup in includes.
87 Description: mds_mfd_close() ASSERTION(rc == 0)
88 Details : In mds_mfd_close(), we need protect inode's writecount change
89 within its orphan write semaphore to prevent possible races.
93 Description: Obsolete CURRENT_SECONDS and use cfs_time_current_sec() instead.
97 Frequency : rare, on shutdown ost
98 Description: don't hit live lock with umount ost.
99 Details : shrink_dcache_parent can be in long loop with destroy dentries,
100 use shrink_dcache_sb instead.
104 Description: don't panic with use echo client
105 Details : echo client pass NULL as client nid pointer and this produce null
110 Description: fix build on ppc32
111 Details : compile code with -m64 flag produce wrong object file for ppc32.
115 Description: add message levels for liblustreapi
119 Description: fix for occasional failure case of -ENOSPC in recovery-small tests
120 Details : Move the 'good_osts' check before the 'total_bavail' check. This
121 will result in an -EAGAIN and in the exit call path we call
122 alloc_rr() which will with increasing aggressiveness attempt to
123 aquire precreated objects on the minimum number of required OSCs.
127 Description: Use old size assignment to avoid deadlock
128 Details : This reverts the changes in bugs 2369 and bug 14138 that introduced
129 the scheduling while holding a spinlock. We do not need locking
130 for size in ll_update_inode() because size is only updated from
131 the MDS for directories or files without objects, so there is no
132 other place to do the update, and concurrent access to such inodes
133 are protected by the inode lock.
137 Description: resolve "_IOWR redefined" build error on SLES10
141 Description: dump the memory debugging after all modules are unloaded to
142 suppress false negative in conf_sanity test 39
144 Severity : enhancement
146 Description: build kernel-ib packages for OFED 1.3 in our release cycle
151 Description: fix SLES kernel versioning
152 Details : the kernel version for our SLES 10 kernel did not include a "-"
153 before the "smp" at the end. while this was not a problem in
154 general, it did mean that software trying to use the kernel
155 version to try to detect a vendor specific kernel would fail.
156 this was most evident by the OFED build scripts.
160 Description: Don't update lov_desc members until making sure they are valid
161 Details : When updating lov_desc members via proc fs, need fix their
162 validities before doing the real update.
166 Description: don't put request into delay list while invalidate in flight.
167 Details : ptlrpc_delay_request sometimes put in delay list while invalidate
168 import in flight. this produce timeout for invalidate and sometimes
169 can cause stale data.
173 Frequency : on ppc only
174 Description: not convert ost objects for directory because it's not exist.
175 Details : ll_dir_getstripe assume dirrectory has ost objects but this wrong.
179 Description: Add FMODE_EXEC file flag for SLES10 SP1 kernel.
181 Severity : enhancement
183 Description: Update to support 2.6.22.14 vanilla kernel.
187 Frequency : rare, on recovery
188 Description: read procfs can produce deadlock in some situation
189 Details : Holding lprocfs lock which send rpc can produce block for destroy
190 obd objects and this also block reconnect with -EALREADY. This isn't
191 fix all lprocfs bugs - but make it rare.
193 Severity : enhancement
195 Description: Update kernel to RHEL5 2.6.18-53.1.14.el5.
198 Frequency : frequent on X2 node
200 Description: mdc_set_open_replay_data LBUG
201 Details : Set replay data for requests that are eligible for replay.
205 Description: lustre_mgs: operation 101 on unconnected MGS
206 Details : When MGC is disconnected from MGS long enough, MGS will evict the
207 MGC, and late on MGC cannot successfully connect to MGS and a lot
208 of the error messages complaining that MGS is not connected.
212 Frequency : on network error
213 Description: panic with double free request if network error
214 Details : mdc_finish_enqueue is finish request if any network error ocuring,
215 but it's true only for synchronus enqueue, for async enqueue
216 (via ptlrpcd) this incorrect and ptlrpcd want finish request
219 Severity : enhancement
221 Description: client-side metadata stat-ahead during readdir(directory readahead)
222 Details : perform client-side metadata stat-ahead when the client detects
223 readdir and sequential stat of dir entries therein
226 Frequency : on start mds
228 Description: Implement get_info(last_id) in obdfilter.
231 Frequency : occasional
233 Description: Correctly check stale fid, not start epoch if ost not support SOM
234 Details : open with flag O_CREATE need set old fid in op_fid3 because op_fid2
235 overwrited with new generated fid, but mds can anwer with one of these
236 two fids and both is not stale. setattr incorectly start epoch and
237 assume will be called done_writeting, but without SOM done_writing
241 Frequency : rare, depends on device drivers and load
243 Description: MDS or OSS nodes crash due to stack overflow
244 Details : Code changes in 1.8.0 increased the stack usage of some functions.
245 In some cases, in conjunction with device drivers that use a lot
246 of stack the MDS (or possibly OSS) service threads could overflow
247 the stack. One change which was identified to consume additional
248 stack has been reworked to avoid the extra stack usage.
251 Frequency : occasional
253 Description: Do not fail import if osc_interpret_create gets -EAGAIN
254 Details : If osc_interpret_create got -EAGAIN it immediately exits and
255 wakeup oscc_waitq. After wakeup oscc_wait_for_objects call
256 oscc_has_objects and see OSC has no objests and call
257 oscc_internal_create to resend create request.
259 Severity : enhancement
261 Description: Update to SLES10 SP1 latest kernel-2.6.16.54-0.2.5.
263 Severity : enhancement
265 Description: Update to RHEL5 latest kernel-2.6.18-53.1.13.el5.
268 Frequency : very rare
270 Description: Fix replay if there is an un-replied request and open
271 Details : In some cases, older replay request will revert the
272 mcd->mcd_last_xid on MDS which is used to record the client's
275 Severity : enhancement
277 Description: Update to RHEL5 latest kernel-2.6.18-53.1.6.el5.
279 Severity : enhancement
281 Description: Add rhel5 support to HEAD.
283 Serverity : enhancement
285 Description: Update RHEL4 kernel to 2.6.9-67.0.4.
290 Description: Don't allow skipping OSTs if index has been specified.
291 Details : Don't allow skipping OSTs if index has been specified, make locking
292 in internal create lots better.
296 Description: LBUG in ptlrpc_check_set() bad phase ebc0de00
297 Details : access to bitfield in structure is always rounded to long
298 and this produce problem with not atomic change any bit.
302 Description: Lustre make rpms failed.
303 Details : Remove ldiskfs spec file to avoids rpmbuild be confused when
304 builds Lustre rpms from tarball.
307 Frequency : rare on shutdown ost
309 Description: If llog cancel was not send before clean_exports phase, this can
310 produce deadlock in llog code.
311 Details : If llog thread has last reference to obd and call class_import_put
312 this produce deadlock because llog_cleanup_commit_master wait when
313 last llog_commit_thread exited, but this never success because was
314 called from llog_commit_thread.
318 Description: allow userland application know is lost one of stripes.
319 Details : fill lvb_blocks with error code on ost and return it to
320 application if error flag found.
324 Description: NULL lov_tgts causing MDS oops
325 Details : more safe checks for NULL lov_tgts for avoid oops.
327 Severity : enhancement
329 Description: Update to RHEL4 latest kernel-2.6.9-67.0.1.EL.
333 Descriptoin: make lov_create() will not stuck in obd_statfs_rqset()
334 Details : If an OST is down the MDS will hang indefinitely in
335 obd_statfs_rqset() waiting for the statfs data. While for
336 MDS QOS usage of statfs, it should not stuck in waiting.
338 Severity : enhancement
340 Description: remote_acl support
341 Details : Support ACL-based permission check for remote user.
342 Support setfacl/getfacl for remote user with the utils
343 "lfs {l,r}{s,g}etfacl" which follow the same parameter format as
344 the system "{s,g}etfacl" utils.
346 Severity : enhancement
348 Description: Update to RHEL4 U6 kernel-2.6.9-67.EL.
350 Severity : enhancement
352 Description: Update to RHEL5 latest kernel-2.6.18-53.1.4.el5.
356 Description: make mgs_setparam() handle fsname containing dash
357 Details : fsname containing a dash does not work with lctl conf_param
359 Severity : enhancement
361 Description: Update to SLES10 SP1 latest kernel-2.6.16.54-0.2.3.
363 Severity : enhancement
365 Description: Update to RHEL5 Update-1 kernel 2.6.18-53.el5.
369 Frequency : rare, at shutdown
370 Description: access already free / zero obd_namespace.
371 Details : if client_disconnect_export was called without force flag set,
372 and exist connect request in flight, this can produce access to
373 NULL pointer (or already free pointer) when connect_interpret
374 store ocd flags in obd_namespace.
378 Frequency : only at startup
379 Description: not alloc memory with spinlock held.
380 Details : allocation memory with GFP_KERNEL can produce sleep deadlock,
381 if any spinlock held.
383 Severity : enhancement
385 Description: make lustre randomly fail allocating memory
386 Details : Make lustre randomly failed allocating memory for testing purpose.
388 Severity : enhancement
390 Description: lost problems with lov objid file
391 Details : Fixes some scability and access to not inited memory problems
392 in work with lov objdid file.
397 Description: lfs find does not continue on file error
398 Details : Continue other files processing when a file/dir is absent.
402 Description: Inconsistent usage of lustre_pack_reply()
403 Details : Standardize the usage of lustre_pack_reply() such that it
404 always generate a CERROR on failure.
407 Frequency : occasional
409 Description: MDS hang or stay in waiting lock
410 Details : If client receive lock with CBPENDING flag ldlm need send lock
411 cancel as separate rpc, to avoid situation when cancel request
412 can't processed due all i/o threads stay in wait lock.
416 Description: Update to RHEL5 kernel 2.6.18-8.1.15.el5.
420 Description: Update to SLES10 SP1 kernel 2.6.16.53-0.16
424 Description: Update to SLES9 kernel-2.6.5-7.287.3.
428 Description: Update to RHEL4 latest kernel.
430 Severity : enhancement
432 Description: Build SLES10 patchless client fails
433 Details : The configure was broken by run ./configure with
434 --with-linux-obj=.... argument for patchless client. When the
435 configure use --with-linux-obj, the LINUXINCLUDE= -Iinclude
436 can't search header adequately. Use absolute path such as
437 -I($LINUX)/include instead.
441 Description: interrupt oig_wait produce painc on resend.
442 Details : brw_redo_request can be used for resend requests from ptlrpcd and
443 private set, and this produce situation when rq_ptlrpcd_data not
444 copyed to new allocated request and triggered LBUG on assert
445 req->rq_ptlrpcd_data != NULL. But this member used only for wakeup
446 ptlrpcd set if request is changed and can be safety changed to use
451 Description: LASSERT_{REQ,REP}SWAB macros are buggy
452 Details : If SWAB_PARANOIA is disabled, the LASSERT_REQSWAB and
453 LASSERT_REPSWAB macros become no-ops, which is incorrect. Drop
454 these macros and replace them with their difinitions instead.
458 Description: Update kernel patches for SLES10 2.6.16.53-0.8.
459 Details : Update which_patch & target file for SLES10 latest kernel.
462 Description: Remove client patches from SLES 10 kernel.
463 Details : This causes SLES 10 clients to behave as patchless clients
464 even on a Lustre-patched (server) kernel.
466 Severity : enhancement
468 Description: self-adjustable client's lru lists
469 Details : use adaptive algorithm for managing client cached locks lru
470 lists according to current server load, other client's work
471 pattern, memory activities, etc. Both, server and client
472 side namespaces provide number of proc tunables for controlling
475 Severity : enhancement
477 Description: light-weight GSS support
478 Details : Support krb5n and krb5a mode, which keep Kerberos 5 authentication
479 and reduce performance overhead.
481 Severity : enhancement
483 Description: Linux keyring support
484 Details : Support using service of Linux keyring for Lustre GSS internal
485 context refresh/cache mechanism.
489 Description: Fix errors in lfs documentation
490 Details : Fixes man pages
494 Description: don't use GFP_* in generic Lustre code.
495 Details : Use cfs_alloc_* functions and CFS_* flags for code portability.
499 Description: obdclass is limited by single OBD_ALLOC(idarray)
500 Details : replace OBD_ALLOC/OBD_FREE with OBD_VMALLOC/OBD_VFREE
504 Description: warnings with build patchless client with vanila 2.6.19 and up
505 Details : change old ctl_table style and replace ctl_table/ctl_table_header
506 with cfs_sysctl_table_t/cfs_sysctl_table_header_t
510 Frequency : Only for SLES
511 Description: sanity_quota fail test_1
512 Details : There are multiple occurences of $TSTUSR in SLES's /etc/group
513 file, which makes TSTID[2] inunique.
517 Frequency : Only for SLES9
518 Description: Kernel patches for SLES9 2.6.5-7.286 kernel
519 Details : Update target/ChangeLog/which_patch .
523 Frequency : Only for test_10 in sanity-quota.sh
524 Description: a bug in quota test code
525 Details : checking a test flag in a wrong place
529 Frequency : Only for test_13 in sanity-quota.sh
530 Description: a bug in quota test script
531 Details : shouldn't check overfull allocated quota
535 Description: jbd statistics
536 Details : Port older jbd statistics patch for sles10
538 Severity : enhancement
540 Description: Add printing inode info into message about error in writepage.
545 Description: reply_lock_interpret crash due to race with it and lock cancel.
546 Details : Do not replay locks that are being cancelled. Do not reference
547 locks by their address during replay, just by their handle.
551 Frequency : When flocks are used.
552 Description: assertion failure in ldlm_cli_enquque_fini for non NULL lock.
553 Details : Flock locks might destroy just granted lock if it could be merged
554 with another existing flock, this is done in completion handler,
555 so teach ldlm_cli_enquque_fini that this is a valid case for
561 Description: Oops in read and write path when failing to allocate lock.
562 Details : Check if lock allocation failed and return error back.
566 Description: lstripe command fails for valid OST index
567 Details : The stripe offset is compared to lov->desc.ld_tgt_count
568 instead of lov->desc.ld_active_tgt_count.
572 Description: sanity.sh failed test 103
573 Details : RHEL mis-interpret setfacl "-X" param, so we won't test setfacl
578 Description: df doesn't work properly if diskfs blocksize != 4K
579 Details : Choose biggest blocksize of OST's as the LOV's blocksize.
583 Description: lfs find on -1 stripe looping in lsm_lmm_verify_common()
584 Details : Avoid lov_verify_lmm_common() on directory with -1 stripe count.
588 Description: obd_health_check_timeout too short
589 Details : set obd_health_check_timeout as 1.5x of obd_timeout
593 Description: llapi_file_create() does not allow some changes
594 Details : add llapi_file_open() that allows specifying the mode and
595 open flags, and also returns an open file handle.
599 Description: merge and cleanup kernel patches.
600 Details : Remove mnt_lustre_list in vfs_intent-2.6-rhel4.patch.
604 Description: Add journal checksum support.(Kernel part)
605 Details : The journal checksum feature adds two new flags i.e
606 JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT and
607 JBD2_FEATURE_COMPAT_CHECKSUM. JBD2_FEATURE_CHECKSUM flag
608 indicates that the commit block contains the checksum for
609 the blocks described by the descriptor blocks. Now commit
610 record can be sent to disk without waiting for descriptor
611 blocks to be written to disk. This behavior is controlled
612 using JBD2_FEATURE_ASYNC_COMMIT flag.
616 Description: OSS needs mutliple precreate threads
617 Details : Add ability to start more than one create thread per OSS.
621 Description: Sanity.sh test_65k failed: setstripe should have succeeded
622 Details : execute lfs setstripe on client
626 Description: mds_obd_create error creating tmp object
627 Details : When the user sets quota on root, llog will be affected and can't
628 create files and write files.
631 Frequency : Always on ia64 patchless client, and possibly others.
633 Description: Add EXPORT_SYMBOL check for node_to_cpumask symbol.
634 Details : This allows the patchless client to be loaded on architectures
639 Description: RedHat Update kernel for RHEL5
640 Details : Add the kernel config file for RHEL5.
644 Description: RedHat Update kernel for RHEL5
645 Details : Modify the kernel config file more closer RHEL5.
649 Description: Build failure against Centos5 (RHEL5)
650 Details : Define PAGE_SIZE when it isn't present.
654 Description: test_7 on sanity-quota.sh can't be used on separate machines
655 Details : add facet to handle it
659 Description: "ll_intent_file_open()) lock enqueue: err: -13" with nfs
660 Details : with NFS, the anon dentry's parent was set to itself in
661 d_alloc_anon(), so in MDS, we use rec->ur_fid1 to find the
662 corresponding dentry other than use rec->ur_name.
664 Severity : enhancement
666 Description: lfs setstripe enhancement
667 Details : Make lfs setstripe understand 'k', 'm' and 'g' for stripe size.
671 Description: enable data checksumming by default
672 Details : enable checksum by default, allow --disable-checksum
673 configure option and "-o nochecksum" mount option. Checksums
674 can also be disabled at runtime via $LPROC/osc/*/checksum_pages.
678 Description: lustre support for RHEL5
679 Details : Add support for RHEL5.
683 Description: Client eviction due to failover config
684 Details : after a connection loss, the lustre client should attempt to
685 reconnect to the last active server first before trying the
686 other potential connections.
690 Description: when mds and osts use different quota unit(32bit and 64bit),
691 quota will be released repeatly.
692 Details : void sending multiple quota reqs to mds, which will keep the
693 status between the reqs.
697 Description: osts not allocated evenly to files
698 Details : change the condition to increase offset_idx
701 Frequency : Always for filesystems larger than 2TB on 32-bit systems.
703 Description: Data corruption for OSTs that are formatted larger than 2TB
705 Details : When generating the bio request for lustre file writes the
706 sector number would overflow a temporary variable before being
707 used for the IO. The data reads correctly from Lustre (which
708 will overflow in a similar manner) but other file data or
709 filesystem metadata may be corrupted in some cases.
713 Description: Tune the kernel for good SCSI performance.
714 Details : Set the value of /sys/block/{dev}/queue/max_sectors_kb
715 to the value of /sys/block/{dev}/queue/max_hw_sectors_kb
720 Description: rewrite ext2-derived code in obdclass/uuid.c
721 Details : rewrite inherited code (uuid parsing code from ext2 utils)
722 from scratch preserving functionality.
726 Description: "lfs find -obd UUID" prints directories
727 Details : "lfs find -obd UUID" will return all directory names instead
728 of just file names. It is incorrect because the directories
729 do not reside on the OSTs.
733 Description: use i_size_read and i_size_write in 2.6 port
734 Details : replace inode->i_size access with i_size_read/write()
737 Frequency : when removing large files
739 Description: scheduling issue during removal of large Lustre files
740 Details : Don't take the BKL in fsfilt_ext3_setattr() for 2.6 kernels.
741 It causes scheduling issues when removing large files (17TB in the
745 Frequency : only with liblustre clients on XT3
747 Description: evictions taking too long
748 Details : allow llrd to evict clients directly on OSTs
751 Frequency : only on ppc
753 Description: /proc/fs/lustre/devices broken on ppc
754 Details : The patch as applied to 1.6.2 doesn't look correct for all arches.
755 We should make sure the type of 'index' is loff_t and then cast
756 explicitly as needed below. Do not assign an explicitly cast
760 Frequency : only for rhel5
762 Description: Kernel patches update for RHEL5 2.6.18-8.1.10.el5.
763 Details : Modify the target file & which_kernel.
765 Severity : enhancement
767 Description: omit set fsid for export NFS
768 Details : fix set/restore device id for avoid EMFILE error and mark lustre fs
769 as FS_REQUIRES_DEV for avoid problems with generate fsid.
773 Frequency : Always, for kernels after 2.6.16
774 Description: Fix warning idr_remove called for id=.. which is not allocated.
775 Details : Last kernels save old s_dev before kill super and not allow
776 to restore from callback - restore it before call kill_anon_super.
780 Description: buffer overruns could theoretically occur
781 Details : llapi_semantic_traverse() modifies the "path" argument by
782 appending values to the end of the origin string, and a
783 overrun may occur. Adding buffer overrun check in liblustreapi.
787 Description: Fix error on 'ls .' at the top of the Lustre mount.
788 Details : Don't revalidate dentry if it is a root dentry.
792 Description: Kernel patches update for RHEL4 2.6.9-55.0.6.
793 Details : Modify vm-tunables-rhel4.patch.
797 Description: Kernel config for 2.6.18-vanilla.
798 Details : Modify targets/2.6-vanilla.target.in.
799 Add config file kernel-2.6.18-2.6-vanilla-i686.config.
800 Add config file kernel-2.6.18-2.6-vanilla-i686-smp.config.
801 Add config file kernel-2.6.18-2.6-vanilla-x86_64.config.
802 Add config file kernel-2.6.18-2.6-vanilla-x86_64-smp.config.
806 Description: improve handling recoverable errors
807 Details : if request processig with error which can be recoverable on server
808 request should be resend, otherwise page released from cache and
813 Description: Kernel patches update for RHEL5 2.6.18-8.1.14.el5.
814 Details : Modify target file & which_patch.
815 A flaw was found in the IA32 system call emulation provided
816 on AMD64 and Intel 64 platforms. An improperly validated 64-bit
817 value could be stored in the %RAX register, which could trigger an
818 out-of-bounds system call table access. An untrusted local user
819 could exploit this flaw to run code in the kernel
820 (ie a root privilege escalation). (CVE-2007-4573).
824 Description: O_DIRECT bypasses client statistics.
825 Details : When running with O_DIRECT I/O, neither the client rpc_stats nor
826 read_ahead_stats were updated. copied the stats section from
827 osc_send_oap_rpc() into async_internal().
831 Description: Add jbd statistics patch for RHEL5 and 2.6.18-vanilla
835 Description: change order of libsysio includes
836 Details : '#include sysio.h' should always come before '#include xtio.h'
840 Description: handle "serious error: objid * already exists" more gracefully
841 Details : If LAST_ID value on disk is smaller than the objects existing in
842 the O/0/d* directories, it indicates disk corruption and causes an
843 LBUG(). If the object is 0-length, then we should use the existing
844 object. This will help to avoid a full fsck in most cases.
846 Severity : enhancement
848 Description: adapt the lustre_config script to support the upgrade case
849 Details : Add "-u" option for lustre_config script to support upgrading 1.4
850 server targets to 1.6 in parallel.
854 Description: To avoid grant space > avaible space when the disk is almost
855 full. Without this patch you might see the error "grant XXXX >
856 available" or some LBUG about grant, when the disk is almost
858 Details : In filter_check_grant, for non_grant cache write, we should
859 check the left space by if (*left > ungranted + bytes), instead
860 of (*left > ungranted), because only we are sure the left space
861 is enough for another "bytes", then the ungrant space should be
862 increase. In client, we should update cl_avail_grant only there
863 is OBD_MD_FLGRANT in the reply.
867 Description: Update RHEL 4 kernel to fix local root privilege escalation.
868 Details : Update to the latest RHEL 4 kernel to fix the vulnerability
869 described in CVE-2007-4573. This problem could allow untrusted
870 local users to gain root access.
873 Frequency : when using O_DIRECT and quotas
875 Description: Incorrect file ownership on O_DIRECT output files
876 Details : block usage reported by 'lfs quota' does not take into account
877 files that have been written with O_DIRECT.
882 Description: touch file failed when fs is not full
883 Details : OST in recovery should not be discarded by MDS in alloc_qos(),
884 otherwise we can get ENOSP while fs is not full.
888 Description: parallel lock callbacks
889 Details : Instead of sending blocking and completion callbacks as separated
890 requests, adding them to a set and sending in parallel.
893 Frequency : only for Cray XT3
894 Bugzilla : 12829/13455
895 Description: Changing primary group doesn't change the group lustre assigns to
897 Details : When CRAY_XT3 is defined, the fsgid supplied by the client is
898 overridden with the primary group provided by the group upcall,
899 whereas the supplied fsgid can be trusted if it is in the list of
900 supplementary groups returned by the group upcall.
902 Severity : enhancement
904 Description: Allow masking D_WARNING, D_ERROR messages from console
905 Details : Console messages can now be disabled via lnet.printk.
909 Description: User code with malformed file open parameter crashes client node
910 Details : Before packing join_file req, all the related reference should be
911 checked carefully in case some malformed flags cause fake
912 join_file req on client.
916 Description: LDLM_ENQUEUE races with LDLM_CP_CALLBACK
917 Details : ldlm_completion_ast() assumes that a lock is granted when the req
918 mode is equal to the granted mode. However, it should also check
919 that LDLM_FL_CP_REQD is not set.
923 Description: Heavy nfs access might result in deadlocks
924 Details : After ELC code landed, it is now improper to enqueue any mds
925 locks under och_sem, because enqueue might want to decide to
926 cancel open locks for same inode we are holding och_sem for.
930 Description: Client eviction while running blogbench
931 Details : A lot of unlink operations with concurrent I/O can lead to a
932 deadlock causing evictions. To address the problem, the number of
933 oustanding OST_DESTROY requests is now throttled to
934 max_rpcs_in_flight per OSC and LDLM_FL_DISCARD_DATA blocking
935 callbacks are processed in priority.
939 Description: enable ACLs on MDS by default
940 Details : ACLs must be enabled on MDS by default.
943 Frequency : PPC/PPC64 only
945 Description: conflicts between asm-ppc64/types.h and lustre_types.h
946 Details : fix duplicated definitions between asm-ppc64/types.h and
947 lustre_types.h on PPC.
950 Frequency : PPC/PPC64 only
952 Description: asm-ppc/segment.h does not exist
953 Details : fix compile issue on PPC.
957 Description: better handle error messages in extents code
960 Frequency : RHEL4 only
962 Description: mkfs is very slow on IA64/RHEL4
963 Details : A performance regression has been discovered in the MPT Fusion
964 driver between versions 3.02.73rh and 3.02.99.00rh. As a
965 consequence, we have downgraded the MPT Fusion driver in the RHEL4
966 kernel from 3.02.99.00 to 3.02.73 until this problem is fixed.
968 Severity : enhancement
970 Description: SNMP support enhancement
971 Details : Adding total number of sampled request for an MDS node in snmp
974 Severity : enhancement
976 Description: Optimize ldlm waiting list processing for PR extent locks
977 Details : When processing waiting list for read extent lock and meeting read
978 lock that is same or wider to it that is not contended, skip
979 processing rest of the list and immediatelly return current
980 status of conflictness, since we are guaranteed there are no
981 conflicting locks in the rest of the list.
985 Description: Time out and refuse to reconnect
986 Details : When the failover node is the primary node, it is possible
987 to have two identical connections in imp_conn_list. We must
988 compare not conn's pointers but NIDs, otherwise we can defeat
989 connection throttling.
993 Description: port llog fixes from b1_6 into HEAD
994 Details : Port llog reference couting and some llog cleanups from b1_6
995 (bug 10800) into HEAD, for protect from panic and access to already
996 free llog structures.
1000 Description: Detect stride IO mode in read-ahead
1001 Details : When a client does stride read, read-ahead should detect that and
1002 read-ahead pages according to the detected stride pattern.
1006 Description: data checksumming impacts single node performance
1007 Details : add support for several checksum algorithm. Currently, only CRC32
1008 and Adler-32 are supported. The checksum type can be changed on
1009 the fly via /proc/fs/lustre/osc/*/checksum_type.
1013 Description: use adler32 for page checksums
1014 Details : when available, use the Adler-32 algorithm instead of CRC32 for
1019 Description: build for x2 fails
1020 Details : fix compile issue on Cray systems.
1024 Description: Properly match for duplicate locks
1025 Details : Due to different lock order from skiplists code, we need to
1026 traverse entire list for now
1029 Frequency : only on PPC/SLES10
1031 Description: "BITS_PER_LONG is not 32 or 64" in linux/idr.h
1032 Details : On SLES10/PPC, fs.h includes idr.h which requires BITS_PER_LONG to
1033 be defined. Add a hack in mkfs_lustre.c to work around this compile
1038 Description: LASSERT on MDS when client holding flock lock dies
1039 Details : ldlm pool logic depends on number of granted locks equal to
1040 number of released locks which is not true for flock locks, so
1041 just exclude such locks from consideration.
1045 Description: MDS deadlock with many ll_sync_lov threads and I/O stalled
1046 Details : Use fsfilt_sync() for both the whole filesystem sync and
1047 individual file sync to eliminate dangerous inode locking
1048 with I_LOCK that can lead to a deadlock.
1052 Description: performance in 1.6.3
1053 Details : Force q->max_phys_segments to MAX_PHYS_SEGMENTS on SLES10 to be
1054 sure that 1MB requests are not fragmented by the block layer.
1056 Severity : enhancement
1058 Description: organize the server-side client stats on per-nid basis
1059 Details : Change the structure of stats under obdfilter and mds to
1067 The "uuid"s file would list the uuids of _active_ exports.
1068 And the clear entry is to clear all stats and stale nids.
1070 Severity : enhancement
1072 Description: eliminate client locks in face of contention
1073 Details : file contention detection and lockless i/o implementation
1074 for contended files.
1078 Description: Reinitialize optind to 0 so that interactive lfs works in all cases
1081 Frequency : very rare, if additional xattrs are used on kernels >= 2.6.12
1083 Description: MDS may lose file striping (and hence file data) in some cases
1084 Details : If there are additional extended attributes stored on the MDS,
1085 in particular ACLs, SELinux, or user attributes (if user_xattr
1086 is specified for the client mount options) then there is a risk
1087 of attribute loss. Additionally, the Lustre file striping
1088 needs to be larger than default (e.g. striped over all OSTs),
1089 and an additional attribute must be stored initially in the
1090 inode and then increase in size enough to be moved to the
1091 external attribute block (e.g. ACL growing in size) for file
1096 Description: skiplist implementation simplification
1097 Details : skiplists are used to group compatible locks on granted list
1098 that was implemented as tracking first and last lock of each lock group
1099 the patch changes that to using doubly linked lists
1103 Description: MDS LBUG: ASSERTION(!IS_ERR(dchild))
1104 Details : Change LASSERTs to client eviction (i.e. abort client's recovery)
1105 because LASSERT on both the data supplied by a client, and the data
1106 on disk is dangerous and incorrect.
1108 Severity : enhancement
1110 Description: Slow trucate/writes to huge files at high offsets.
1111 Details : Directly associate cached pages to lock that protect those pages,
1112 this allows us to quickly find what pages to write and remove
1113 once lock callback is received.
1117 Description: more ldlm soft lockups
1118 Details : In ldlm_resource_add_lock(), call to ldlm_resource_dump()
1119 starve other threads from the resource lock for a long time in
1120 case of long waiting queue, so change the debug level from
1121 D_OTHER to the less frequently used D_INFO.
1123 --------------------------------------------------------------------------------
1125 2007-08-10 Cluster File Systems, Inc. <info@clusterfs.com>
1127 * Support for kernels:
1128 2.6.5-7.283 (SLES 9),
1129 2.6.9-55.EL (RHEL 4),
1130 2.6.16.46-0.14 (SLES 10),
1131 2.6.18.8 vanilla (kernel.org)
1132 * Client support for unpatched kernels:
1133 (see http://wiki.lustre.org/index.php?title=Patchless_Client)
1134 2.6.16 - 2.6.22 vanilla (kernel.org)
1135 * Due to recently discovered recovery problems, we do not recommend
1136 using patchless RHEL 4 clients with this or any earlier release.
1137 * Recommended e2fsprogs version: 1.39.cfs8
1138 * Note that reiserfs quotas are disabled on SLES 10 in this kernel.
1139 * Starting with this release, the ldiskfs backing filesystem required
1140 by Lustre is now in its own package, lustre-ldiskfs. This package
1141 should be installed. It is versioned separately from Lustre and
1142 may be released separately in future.
1146 Description: block reactivating mgc import until all deactivates complete
1147 Details : Fix race when failing back MDT/MGS to itself (testing)
1150 Frequency : at statup only
1152 Description: mds_lov_synchronize race leads to various problems
1153 Details : simultaneous MDT->OST connections at startup can cause the
1154 sync to abort, leaving the OSC in a bad state.
1156 Severity : enhancement
1158 Description: add optional extra BUILD_VERSION info
1159 Details : add a new environment variable (namely LUSTRE_VERS) which allows
1160 to override the lustre version.
1163 Frequency : 2.6.18 servers only
1165 Description: ll_kern_mount() doesn't release the module reference
1166 Details : The ldiskfs module reference count never drops down to 0
1167 because ll_kern_mount() doesn't release the module reference.
1172 Description: server LBUG when using old ost_num_threads parameter
1173 Details : Accept the old ost_num_threads parameter but warn that it
1174 is deprecated, and fix an off-by-one error that caused an LBUG.
1179 Description: Transient SCSI error results in persistent IO issue
1180 Details : iobuf->dr_error is not reinitialized to 0 between two
1184 Frequency : sometimes when underlying device returns I/O errors
1186 Description: OSTs not going read-only during write failures
1187 Details : OSTs are not remounted read-only when the journal commit threads
1188 get I/O errors because fsfilt_ext3 calls journal_start/stop()
1189 instead of the ext3 wrappers.
1193 Description: poor connect scaling with increasing client count
1194 Details : Don't run filter_grant_sanity_check for more than 100 exports
1195 to improve scaling for large numbers of clients.
1198 Frequency : SLES10 only
1200 Description: sanity-quota.sh quotacheck failed: rc = -22
1201 Details : Quotas cannot be enabled on SLES10.
1204 Frequency : liblustre clients only
1206 Description: getdirentries does not give error when run on compute nodes
1207 Details : getdirentries does not fail when the size specified as an argument
1208 is too small to contain at least one entry
1210 Severity : enhancement
1212 Description: Add LNET router traceability for debug purposes
1213 Details : If a checksum failure occurs with a router as part of the
1214 IO path, the NID of the last router that forwarded the bulk data
1215 is printed so it can be identified.
1220 Description: OST "spontaneously" evicts client; client has imp_pingable == 0
1221 Details : Due to a race condition, liblustre clients were occasionally
1222 evicted incorrectly.
1224 Severity : enhancement
1226 Description: lfs setstripe use optional parameters instead of postional
1229 Severity : enhancement
1231 Description: Nanosecond timestamp support for ldiskfs
1232 Details : The on-disk ldiskfs filesystem has added support for nanosecond
1233 resolution timestamps. There is not yet support for this at
1234 the Lustre filesystem level.
1237 Frequency : during server recovery
1239 Description: MDS failing to send precreate requests due to OSCC_FLAG_RECOVERING
1240 Details : request with rq_no_resend flag not awake l_wait_event if they get a
1244 Frequency : nfs export on patchless client
1246 Description: connectathon hang when test nfs export over patchless client
1247 Details : Disconnected dentry cannot be found with lookup, so we do not need
1248 to unhash it or make it invalid
1251 Description: fix llapi_lov_get_uuids() to allow many OSTs to be returned
1252 Details: : Change llapi_lov_get_uuids() to read the UUIDs from /proc instead
1253 of using an ioctl. This allows lfsck for > 160 OSTs to succeed.
1258 Description: open req refcounting wrong on reconnect
1259 Details : If reconnect happened between getting open reply from server and
1260 call to mdc_set_replay_data in ll_file_open, we will schedule
1261 replay for unreferenced request that we are about to free.
1262 Subsequent close will crash in variety of ways.
1263 Check that request is still eligible for replay in
1264 mdc_set_replay_data().
1269 Description: disable writes to filesystem when reading health_check file
1270 Details : the default for reading the health_check proc file has changed
1271 to NOT do a journal transaction and write to disk, because this
1272 can cause reads of the /proc file to hang and block HA state
1273 checking on a healthy but otherwise heavily loaded system. It
1274 is possible to return to the previous behaviour during configure
1275 with --enable-health-write.
1277 Severity : enhancement
1279 Description: 64-bit inode version
1280 Details: : Add a on-disk 64-bit inode version for ext3 to track changes made
1281 to the inode. This will be required for version-based recovery.
1286 Description: MDS fails to start if a duplicate client export is detected
1287 Details : in some rare cases it was possible for a client to connect to
1288 an MDS multiple times. Upon recovery the MDS would detect this
1289 and fail during startup. Handle this more gracefully.
1291 Severity : enhancement
1293 Description: Add -o localflock option to simulate old noflock
1295 Details : This will achieve local-only flock/fcntl locks
1301 Description: log_commit_thread vs filter_destroy race leads to crash
1302 Details : Take import reference before releasing llog record semaphore
1307 Description: Wrong request locking in request set processing
1308 Details : ptlrpc_check_set wrongly uses req->rq_lock for proctect add to
1309 imp_delayed_list, in this place should be used imp_lock.
1312 Frequency : when reconnection
1314 Description: Grant Leak when osc reconnect to OST
1315 Details : When osc reconnect ost, OST(filter) should check whether it
1316 should grant more space to client by comparing fed_grant and
1317 cl_avail_grant, and return the granted space to client instead
1318 of "new granted" space, because client will call osc_init_grant
1319 to update the client grant space info.
1322 Frequency : when client reconnect to OST
1324 Description: Grant Leak when osc do resend and replay bulk write
1325 Details : When osc reconnect to OST, OST(filter)should clear grant info of
1326 bulk write request, because the grant info while be sync between
1327 OSC and OST when reconnect, and we should ignore the grant info
1328 these of resend/replay write req.
1333 Description: Grant space more than avaiable left space sometimes.
1334 Details : When then OST is about to be full, if two bulk writing from
1335 different clients came to OST. Accord the avaliable space of the
1336 OST, the first req should be permitted, and the second one
1337 should be denied by ENOSPC. But if the seconde arrived before
1338 the first one is commited. The OST might wrongly permit second
1339 writing, which will cause grant space > avaiable space.
1342 Frequency : when client is evicted
1344 Description: Grant might be wrongly erased when osc is evicted by OST
1345 Details : when the import is evicted by server, it will fork another
1346 thread ptlrpc_invalidate_import_thread to invalidate the
1347 import, where the grant will be set to 0. While the original
1348 thread will update the grant it got when connecting. So if
1349 the former happened latter, the grant will be wrongly errased
1350 because of this race.
1355 Description: Checking Stale with correct fid
1356 Details : ll_revalidate_it should uses de_inode instead of op_data.fid2
1357 to check whether it is stale, because sometimes, we want the
1358 enqueue happened anyway, and op_data.fid2 will not be initialized.
1360 Severity : enhancement
1362 Description: update patchless client
1363 Details : Add support for patchless client with 2.6.20, 2.6.21 and RHEL 5
1366 Frequency : only with 2.4 kernel
1368 Description: random memory corruption
1369 Details : size of struct ll_inode_info is to big for union inode.u and this
1370 can be cause of random memory corruption.
1375 Description: Memory leak in recovery
1376 Details : Lov_mds_md was not free in an error handler in mds_create_object.
1377 It should also check obd_fail before fsfilt_start, otherwise if
1378 fsfilt_start return -EROFS,(failover mds during mds recovery).
1379 then the req will return with repmsg->transno = 0 and rc = EROFS.
1380 and we met hit the assert LASSERT(req->rq_reqmsg->transno ==
1381 req->rq_repmsg->transno) in ptlrpc_replay_interpret. Fcc should
1382 be freed no matter whether fsfilt_commit success or not.
1385 Frequency : only with huge count clients
1387 Description: Prevents from taking the superblock lock in llap_from_page for
1389 Details : using LL_ORIGIN_REMOVEPAGE origin flag instead of LL_ORIGIN_UNKNOW
1390 for llap_from_page call in ll_removepage prevents from taking the
1391 superblock lock for a soon died page.
1396 Description: Not check open intent error before release open handle
1397 Details : in some rare cases, the open intent error is not checked before
1398 release open handle, which may cause
1399 ASSERTION(open_req->rq_transno != 0), because it tries to release
1400 the failed open handle.
1405 Description: Set cat log bitmap only after create log success.
1406 Details : in some rare cases, the cat log bitmap is set too early. and it
1407 should be set only after create log success.
1411 Description: Accessing a block bevice can re-enable I/O when Lustre is
1412 tearing down a device.
1413 Details : dev_clear_rdonly(bdev) must be called in kill_bdev() instead of
1418 Description: service threads may hog cpus when there are a lot of requests
1419 Details : Insert cond_resched to give other threads a chance to use some CPU
1424 Description: the cat log was not initialized in recovery
1425 Details : When mds(mgs) do recovery, the tgt_count might be zero, so the
1426 unlink log on mds will not be initialized until mds post
1427 recovery. And also in mds post recovery, the unlink log will
1428 initialization will be done asynchronausly, so there will be race
1429 between add unlink log and unlink log initialization.
1433 Description: brw_stats were being printed incorrectly
1434 Details : brw_stats were being printed as log2 but all of them were not
1435 recorded as log2. Also remove some code duplication arising from
1436 filter_tally_{read,write}.
1440 Frequency : rare, only in recovery.
1441 Description: ASSERTION(req->rq_type != LI_POISON) failed
1442 Details : imp_lock should be held while iterating over imp_sending_list for
1443 prevent destroy request after get timeout in ptlrpc_queue_wait.
1447 Description: replay-single.sh test 52 fails
1448 Details : A lock's skiplist need to be cleanup when it being unlinked
1449 from its resource list.
1453 Description: Short directio read returns full requested size rather than
1455 Details : Direct I/O operations should return actual amount of bytes
1456 transferred rather than requested size.
1458 Severity : enhancement
1460 Description: metadata RPC reduction (e.g. for rm performance)
1461 Details : decrease the amount of synchronous RPC between clients and servers
1462 by canceling conflicing lock before the operation on the client
1463 and packing thier handles into the main operation RPC to server.
1465 Severity : enhancement
1467 Description: Async OSC create to avoid the blocking unnecessarily.
1468 Details : If a OST has no remain object, system will block on the creating
1469 when need to create a new object on this OST. Now, ways use
1470 pre-created objects when available, instead of blocking on an
1471 empty osc while others are not empty. If we must block, we block
1472 for the shortest possible period of time.
1474 Severity : enhancement
1476 Description: refine locking for avoid write wrong info into lov_objid file
1477 Details : fix possible races with add new target and write/update data in
1480 --------------------------------------------------------------------------------
1482 2007-05-03 Cluster File Systems, Inc. <info@clusterfs.com>
1487 Frequency : on some architectures
1489 Description: 1.6 client sometimes fails to mount from a 1.4 MDT
1490 Details : Uninitialized flags sometimes cause configuration commands to
1494 Frequency : patchless clients only
1496 Description: missing __iget() symbol export
1497 Details : The __iget() symbol export is missing. To avoid the need for
1498 this on patchless clients the deathrow inode reaper is turned
1499 off, and we depend on the VM to clean up old inodes. This
1500 dependency was during via the fix for bug 12181.
1502 --------------------------------------------------------------------------------
1504 2007-04-19 Cluster File Systems, Inc. <info@clusterfs.com>
1506 * CONFIGURATION CHANGE. This version of Lustre WILL NOT
1507 INTEROPERATE with older versions automatically. In many cases a
1508 special upgrade step is needed. Please read the
1509 user documentation before upgrading any part of a live system.
1510 * WIRE PROTOCOL CHANGE from previous 1.6 beta versions. This
1511 version will not interoperate with 1.6 betas before beta5 (1.5.95).
1512 * WARNING: Lustre configuration and startup changes are required with
1513 this release. See https://mail.clusterfs.com/wikis/lustre/MountConf
1519 Severity : enhancement
1521 Description: MountConf
1522 Details : Lustre configuration is now managed via mkfs and mount
1523 commands instead of lmc and lconf. New obd types (MGS, MGC)
1524 are added for dynamic configuration management. See
1525 https://mail.clusterfs.com/wikis/lustre/MountConf for
1528 Severity : enhancement
1530 Description: dynamic OST addition
1531 Details : OSTs can now be added to a live filesystem
1533 Severity : enhancement
1535 Description: startup order invariance
1536 Details : MDTs and OSTs can be started in any order. Clients only
1537 require the MDT to complete startup.
1539 Severity : enhancement
1541 Description: parallel, asynchronous orphan cleanup
1542 Details : orphan cleanup is now performed in separate threads for each
1543 OST, allowing parallel non-blocking operation.
1545 Severity : enhancement
1547 Description: optimized stripe assignment
1548 Details : stripe assignments are now made based on ost space available,
1549 ost previous usage, and OSS previous usage, in order to try
1550 to optimize storage space and networking resources.
1552 Severity : enhancement
1554 Description: Permanently set tunables
1555 Details : All writable /proc/fs/lustre tunables can now be permanently
1556 set on a per-server basis, at mkfs time or on a live system.
1558 Severity : enhancement
1560 Description: Lustre message v2
1561 Details : Add lustre message format v2.
1563 Severity : enhancement
1565 Description: client OST exclusion list
1566 Details : Clients can be started with a list of OSTs that should be
1567 declared "inactive" for known non-responsive OSTs.
1571 Description: SPEC SFS validation failure on NFS v2 over lustre.
1572 Details : Changes the blocksize for regular files to be 2x RPC size,
1573 and not depend on stripe size.
1575 Severity : enhancement
1577 Description: Multiple MD RPCs in flight.
1578 Details : Further unserialise some read-only MDS RPCs - learn about intents.
1579 To avoid overly-overloading MDS, introduce a limit on number of
1580 MDS RPCs in flight for a single client and add /proc controls
1581 to adjust this limit.
1583 Severity : enhancement
1585 Description: client read/write statistics
1586 Details : Add client read/write call usage stats for performance
1587 analysis of user processes.
1588 /proc/fs/lustre/llite/*/offset_stats shows non-sequential
1589 file access. extents_stats shows chunk size distribution.
1590 extents_stats_per_process show chunk size distribution per
1593 Severity : enhancement
1595 Description: mds statistics
1596 Details : Add detailed mds operations statistics in
1597 /proc/fs/lustre/mds/*/stats.
1601 Description: Failure of copying files with lustre special EAs.
1602 Details : Client side always return success for setxattr call for lustre
1603 special xattr (currently only "trusted.lov").
1608 Description: Refcount LNET uuids
1609 Details : The global LNET uuid list grew linearly with every startup;
1610 refcount repeated list entries instead of always adding to
1613 Severity : enhancement
1615 Description: Dynamic service threads
1616 Details : Within a small range, start extra service threads
1617 automatically when the request queue builds up.
1620 Frequency : mixed-endian client/server environments
1622 Description: mixed-endian crashes
1623 Details : The new msg_v2 system had some failures in mixed-endian
1626 Severity : enhancement
1628 Description: Easy OST removal
1629 Details : OSTs can be permanently deactivated with e.g. 'lctl
1630 conf_param lustre-OST0001.osc.active=0'
1632 Severity : enhancement
1634 Description: MGS proc entries
1635 Details : Added basic proc entries for the MGS showing what filesystems
1638 Severity : enhancement
1640 Description: provide MGS failover
1641 Details : Added config lock reacquisition after MGS server failover.
1643 Severity : enhancement
1645 Description: add Linux 2.4 support
1646 Details : Added support for RHEL 2.4.21 kernel for 1.6 servers and clients
1650 Description: a large application tries to do I/O to the same resource and dies
1651 in the middle of it.
1652 Details : Check the req->rq_arrival time after the call to
1653 ost_brw_lock_get(), but before we do anything about
1654 processing it & sending the BULK transfer request. This
1655 should help move old stale pending locks off the queue as
1656 quickly as obd_timeout.
1659 Frequency : when an incorrect nid is specified during startup
1661 Description: ptlrpc connect to non-existant node causes kernel crash
1662 Details : LNET can't be re-entered from an event callback, which
1663 happened when we expire a message after the export has been
1664 cleaned up. Instead, hand the zombie cleanup off to another
1667 Severity : enhancement
1669 Description: plain/inodebits lock performance improvement
1670 Details : Grouping plain/inodebits in granted list by their request modes
1671 and bits policy, thus improving the performance of search through
1675 Frequency : only if OST filesystem is corrupted
1677 Description: client incorrectly hits assertion in ptlrpc_replay_req()
1678 Details : for a short time RPCs with bulk IO are in the replay list,
1679 but replay of bulk IOs is unimplemented. If the OST filesystem
1680 is corrupted due to disk cache incoherency and then replay is
1681 started it is possible to trip an assertion. Avoid putting
1682 committed RPCs into the replay list at all to avoid this issue.
1685 Frequency : liblustre (e.g. catamount) on a large cluster with >= 8 OSTs/OSS
1687 Description: System hang on startup
1688 Details : This bug allowed the liblustre (e.g. catamount) client to
1689 return to the app before handling all startup RPCs. This
1690 could leave the node unresponsive to lustre network traffic
1691 and manifested as a server ptllnd timeout.
1693 Severity : enhancement
1695 Description: Add "/proc/sys/lustre/debug_peer_on_timeout"
1696 Details : liblustre envirable: LIBLUSTRE_DEBUG_PEER_ON_TIMEOUT
1697 boolean to control whether to print peer debug info when a
1698 client's RPC times out.
1701 Frequency : only for kernels with patches from Lustre below 1.4.3
1703 Description: Remove old rdonly API
1704 Details : Remove old rdonly API which unused from at least lustre 1.4.3
1707 Frequency : only for devices with external journals
1709 Description: Set external device read-only also
1710 Details : During a commanded failover stop, we set the disk device
1711 read-only while the server shuts down. We now also set any
1712 external journal device read-only at the same time.
1715 Frequency : when upgrading from 1.4 while trying to change parameters
1717 Description: The wrong (new) MDC name was used when setting parameters for
1718 upgraded MDT's. Also allows changing of OSC (and MDC)
1719 parameters if --writeconf is specified at tunefs upgrade time.
1722 Frequency : when setting specific ost indicies
1724 Description: QOS code breaks on skipped indicies
1725 Details : Add checks for missing OST indicies in the QOS code, so OSTs
1726 created with --index need not be sequential.
1728 Severity : enhancement
1730 Description: Add uninit_groups feature to ldiskfs2 to speed up e2fsck
1731 Details : The uninit_groups feature works in conjunction with the kernel
1732 filesystem code (ldiskfs2 only) and e2fsprogs-1.39-cfs6 to speed
1733 up the pass1 processing of e2fsck. This is a read-only feature
1734 in ldiskfs2 only, so older kernels and current ldiskfs cannot
1735 mount filesystems that have had this feature enabled.
1737 Severity : enhancement
1739 Description: Improve multi-block allocation algorithm to avoid fragmentation
1740 Details : The mballoc3 code (ldiskfs2 only) adds new mechanisms to improve
1741 allocation locality and avoid filesystem fragmentation.
1743 ------------------------------------------------------------------------------
1745 2007-04-01 Cluster File Systems, Inc. <info@clusterfs.com>
1747 * Support for kernels:
1748 2.4.21-47.0.1.EL (RHEL 3)
1749 2.6.5-7.283 (SLES 9)
1750 2.6.9-42.0.10.EL (RHEL 4)
1751 2.6.12.6 vanilla (kernel.org)
1752 2.6.16.27-0.9 (SLES 10)
1753 * Recommended e2fsprogs version: 1.39.cfs5
1755 * Note that reiserfs quotas are disabled on SLES 10 in this kernel
1759 Frequency : occasional, depends on client load and configuration
1760 Bugzilla : 12181, 12203
1761 Description: data loss for recently-modified files
1763 Details : In some cases it is possible that recently written or created
1764 files may not be written to disk in a timely manner (this should
1765 normally be within 30s unless client IO load is very high).
1766 The problem appears as zero-length files or files that are a
1767 multiple of 1MB in size after a client crash or client eviction
1768 that are missing data at the end of the file.
1770 This problem is more likely to be hit on clients where files are
1771 repeatedly created and unlinked in the same directory, clients
1772 have a large amount of RAM, have many CPUs, the filesystem has
1773 many OSTs, the clients are rebooted frequently, and/or the files
1774 are not accessed by other nodes after being written.
1776 The presence of the problem can be detected by looking at
1777 /proc/sys/fs/inode-state. If the first number (nr_inodes) is
1778 smaller than the second (nr_unused) then dirty files will not
1779 be flushed automatically to disk. "sync; sleep 10" should be
1780 run several times on the node before unmounting it to update
1781 Lustre (this is also safe to run on nodes without this problem).
1783 There is also a related kernel bug in the RHEL4 4 2.6.9 kernel
1784 that can cause this same problem, so customers using that kernel
1785 also need to update the kernel in addition to Lustre. In order
1786 to properly fix this bug, the RHEL3 2.4.21 kernel is also updated.
1788 It is normal that files written just before a client crash (less
1789 than 30s) may not yet have been flushed to disk, even for local
1793 Frequency : frequent on thin XT3 nodes
1795 Description: UUID collision on thin XT3 Linux nodes
1796 Details : UUIDs on Compute Node Linux XT3 nodes were not generated
1797 randomly, since we relied on an insufficiently-seeded PRNG.
1802 Description: OSS hangs after "All ost request buffers busy"
1803 Details : A deadlock between quota and journal operations caused OSS
1804 hangs after printing "All ost request buffers busy."
1807 Frequency : always on liblustre builds
1809 Description: Cleanup compiler warnings on liblustre
1812 Frequency : always on liblustre builds on XT3
1814 Description: LC_CONFIG_CDEBUG don't run while build liblustre on XT3.
1818 Description: Addition of EXT3_FEATURE_RO_COMPAT_DIR_NLINKS flag for
1819 > 32000 subdirectories
1820 Details : Add EXT3_FEATURE_RO_COMPAT_DIR_NLINK flag to
1821 EXT3_FEATURE_RO_COMPAT_SUPP. This flag will be set whenever
1822 subdirectory count crosses 32000. This will aid e2fsck to
1823 correctly handle more than 32000 subdirectories.
1826 Frequency : liblustre (e.g. catamount) on a large cluster with >= 8 OSTs/OSS
1828 Description: System hang on startup
1829 Details : This bug allowed the liblustre (e.g. catamount) client to
1830 return to the app before handling all startup RPCs. This
1831 could leave the node unresponsive to lustre network traffic
1832 and manifested as a server ptllnd timeout.
1834 Severity : enhancement
1836 Description: Add "/proc/sys/lustre/debug_peer_on_timeout"
1837 (liblustre envirable: LIBLUSTRE_DEBUG_PEER_ON_TIMEOUT)
1838 boolean to control whether to print peer debug info when a
1839 client's RPC times out.
1844 Description: make O_SYNC working on 2.6 kernels
1845 Details : 2.6 kernels use different method for mark pages for write,
1846 so need add a code to lustre for O_SYNC work.
1851 Description: Failure to close file and release space on NFS
1852 Details : Put inode details into lock acquired in ll_intent_file_open.
1853 Use mdc_intent_lock in ll_intent_open to properly
1854 detect all kind of errors unhandled by mdc_enqueue.
1859 Description: proc file read during shutdown sometimes raced obd removal,
1861 Details : Add lock to prevent obd access after proc file removal.
1864 Frequency : Only for files larger than 4GB on 32-bit clients.
1866 Description: improperly doing page alignment of locks
1867 Details : Modify lustre core code to use CFS_PAGE_* defines instead of
1868 PAGE_*. Make CFS_PAGE_MASK a 64-bit mask.
1873 Description: RPCs being resent when they shouldn't be
1874 Details : Some RPCs that should not be resent are being resent. This
1875 can cause inconsistencies in the RPC state machine. Do not
1876 resend such requests.
1879 Frequency : rare, only with NFS export
1881 Description: Crash on NFS re-export node
1882 Details : under very unusual load conditions an assertion is hit in
1883 ll_intent_file_open()
1886 Frequency : only if OST filesystem is corrupted
1888 Description: client incorrectly hits assertion in ptlrpc_replay_req()
1889 Details : for a short time RPCs with bulk IO are in the replay list,
1890 but replay of bulk IOs is unimplemented. If the OST filesystem
1891 is corrupted due to disk cache incoherency and then replay is
1892 started it is possible to trip an assertion. Avoid putting
1893 committed RPCs into the replay list at all to avoid this issue.
1898 Description: large O_DIRECT requests fail under memory pressure/fragmentation
1899 Details : Large single O_DIRECT read and write calls can fail to allocate
1900 a sufficiently large buffer to process the request. In case of
1901 allocation failure the allocation is retried with a smaller
1902 buffer and broken into smaller requests.
1904 Severity : enhancement
1906 Description: Add -o localflock option to simulate old noflock behaviour.
1907 Details : This will achieve local-only flock/fcntl locks coherentness.
1912 Description: versioning check is incomplete
1913 Details : Checking the version difference of client vs. server, report
1914 error if the gap is too big.
1919 Description: add support PG_writeback bit
1920 Details : add support for PG_writeback bit for Lustre, for more carefull
1921 work with page cache in 2.6 kernel. This also fix some deadlocks
1922 and remove hack for work O_SYNC with 2.6 kernel.
1924 Severity : enhancement
1926 Description: Add uninit_groups feature to ldiskfs2 to speed up e2fsck
1927 Details : The uninit_groups feature works in conjunction with the kernel
1928 filesystem code (ldiskfs2 only) and e2fsprogs-1.39-cfs6 to speed
1929 up the pass1 processing of e2fsck. This is a read-only feature
1930 in ldiskfs2 only, so older kernels and current ldiskfs cannot
1931 mount filesystems that have had this feature enabled.
1933 Severity : enhancement
1935 Description: Improve multi-block allocation algorithm to avoid fragmentation
1936 Details : The mballoc3 code (ldiskfs2 only) adds new mechanisms to improve
1937 allocation locality and avoid filesystem fragmentation.
1939 ------------------------------------------------------------------------------
1941 2007-02-09 Cluster File Systems, Inc. <info@clusterfs.com>
1943 * Support for kernels:
1944 2.6.9-42.0.3.EL (RHEL 4)
1945 2.6.5-7.276 (SLES 9)
1946 2.4.21-47.0.1.EL (RHEL 3)
1947 2.6.12.6 vanilla (kernel.org)
1948 2.6.16.21-0.8 (SLES10)
1949 * Recommended e2fsprogs version: 1.39.cfs2-0
1951 * The backwards-compatible /proc/sys/portals symlink has been removed
1952 in this release. Before upgrading, please ensure that you change
1953 any configuration scripts or /etc/sysctl.conf files that access
1954 /proc/sys/portals/* or sysctl portals.* to use the corresponding
1955 entry in /proc/sys/lnet or sysctl lnet.*. This change can be made
1956 in advance of the upgrade on any system running Lustre 1.4.6 or
1957 newer, since /proc/sys/lnet was added in that version.
1958 * Note that reiserfs quotas are disabled on SLES 10 in this kernel
1962 Frequency : only when quota is used
1964 Description: avoid scanning export list for quota master
1965 Details : Change the algorithms to avoid scanning export list in order
1966 to improve the efficiency.
1969 Frequency : MDS failover only, very rarely
1971 Description: "went back in time" messages on mds failover
1972 Details : The greatest transno may be lost when the current operation
1973 finishes with an error (transno==0) and the client's last_rcvd
1974 record is over-written. Save the greatest transno in the
1975 mds_last_transno for this case.
1978 Frequency : always for specific kernels and striping counts
1980 Description: client may get "Matching packet too big" without ACL support
1981 Details : Clients compiled without CONFIG_FS_POSIX_ACL get an error message
1982 when trying to access files in certain configurations. The
1983 clients should in fact be denied when mounting because they do
1984 not understand ACLs.
1987 Frequency : Cray XT3 with more than 4000 clients and multiple jobs
1989 Description: many clients connecting with IO in progress causes connect timeouts
1990 Details : Avoid synchronous journal commits to avoid delays caused by many
1991 clients connecting/disconnecting when bulk IO is in progress.
1992 Queue liblustre connect requests on OST_REQUEST_PORTAL instead of
1993 OST_IO_PORTAL to avoid delays behind potentially many pending
1997 Frequency : occasionally with multiple writers to a single file
1999 Description: shared writes to file may result in wrong size reported by stat()
2000 Details : Allow growing of kms when extent lock is cancelled
2003 Frequency : always with random mmap IO to multi-striped file
2005 Description: mmap write might be lost if we are writing to a 'hole' in stripe
2006 Details : Only if the hole is at the end of OST object so that kms is too
2007 small. Fix is to increase kms accordingly in ll_nopage.
2010 Frequency : rare, only if OST filesystem is inconsistent with MDS filesystem
2012 Description: writes to a missing object would leak memory on the OST
2013 Details : If there is an inconsistency between the MDS and OST filesystems,
2014 such that the MDS references an object that doesn't exist, writes
2015 to that object will leak memory due to incorrect cleanup in the
2016 error handling path, eventually running out of memory on the OST.
2021 Description: Creating too long symlink causes lustre errors
2022 Details : Check symlink and name lengths before sending requests to MDS.
2025 Frequency : only if flock is enabled (not on by default)
2027 Description: posix locks not released on fd closure on 2.6.9+
2028 Details : We failed to add posix locks to list of inode locks on 2.6.9+
2029 kernels, this caused such locks not to be released on fd close and
2030 then assertions on fs unmount about still used locks.
2033 Frequency : MDS failover only, very rarely
2035 Description: clients may get ASSERTION(granted_lock != NULL)
2036 Details : When request was taking a long time, and a client was resending
2037 a getattr by name lock request. The were multiple lock requests
2038 with the same client lock handle and
2039 mds_getattr_name->fixup_handle_for_resent_request found one of the
2040 lock handles but later failed with ASSERTION(granted_lock != NULL).
2045 Description: handle->h_buffer_credits > 0, assertion failure
2046 Details : h_buffer_credits is zero after truncate, causing assertion
2047 failure. This patch extends the transaction or creates a new
2051 Frequency : NFS re-export or patchless client
2052 Bugzilla : 11179, 10796
2053 Description: Crash on NFS re-export node (__d_move)
2054 Details : We do not want to hash the dentry if we don't have a lock.
2055 But if this dentry is later used in d_move, we'd hit uninitialised
2056 list head d_hash, so we just do this to init d_hash field but
2057 leave dentry unhashed.
2060 Frequency : NFS re-export or patchless client
2062 Description: NFS exports has problem with symbolic link
2063 Details : lustre client didn't properly install dentry when re-exported
2064 to NFS or running patchless client.
2067 Frequency : NFS re-export or patchless client
2069 Description: Various nfs/patchless fixes.
2070 Details : fixes reuse disconected alias for lookup process - this fixes
2071 warning "find_exported_dentry: npd != pd",
2072 fix permission error with open files at nfs.
2073 fix apply umask when do revalidate.
2076 Frequency : occasional
2078 Description: Crash on NFS re-export node
2079 Details : calling clear_page() on the wrong pointer triggered oops in
2080 generic_mapping_read().
2083 Frequency : rarely, using O_DIRECT IO
2085 Description: unaligned directio crashes client with LASSERT
2086 Details : check for unaligned buffers before trying any requests.
2089 Frequency : rarely, using CFS RAID5 patches in non-standard kernel series
2091 Description: stale data returned from RAID cache
2092 Details : If only a small amount of IO is done to the RAID device before
2093 reading it again it is possible to get stale data from the RAID
2094 cache instead of reading it from disk.
2097 Frequency : always for sles10 kernel
2099 Description: sles10 support
2100 Details : ll_follow_link: compile fixes and using of nd_set_link
2101 under newer kernels.
2104 Frequency : depends on arch, kernel and compiler version, always on sles10
2107 Description: recursive or deep enough symlinks cause stack overflow
2108 Details : getting rid of large stack-allocated variable in
2112 Frequency : depends on hardware
2114 Description: lustre write performance loss in the SLES10 kernel
2115 Details : the performance loss is caused by using of write barriers in the
2116 ext3 code. The SLES10 kernel turns barrier support on by
2117 default. The fix is to undo that change for ldiskfs.
2119 ------------------------------------------------------------------------------
2121 2006-12-09 Cluster File Systems, Inc. <info@clusterfs.com>
2123 * Support for kernels:
2124 2.6.9-42.0.3EL (RHEL 4)
2125 2.6.5-7.276 (SLES 9)
2126 2.4.21-47.0.1.EL (RHEL 3)
2127 2.6.12.6 vanilla (kernel.org)
2131 Frequency : quota enabled and large files being deleted
2133 Description: releasing more than 4GB of quota at once hangs OST
2134 Details : If a user deletes more than 4GB of files on a single OST it
2135 will cause the OST to spin in an infinite loop. Release
2136 quota in < 4GB chunks, or use a 64-bit value for 1.4.7.1+.
2141 Description: statfs data retrieved from /proc may be stale or zero
2142 Details : When reading per-device statfs data from /proc, in the
2143 {kbytes,files}_{total,free,avail} files, it may appear
2144 as zero or be out of date.
2147 Frequency : systems with MD RAID1 external journal devices
2149 Description: lconf's call to blkid is confused by RAID1 journal devices
2150 Details : Use the "blkid -l" flag to locate the MD RAID device instead
2151 of returning all block devices that match the journal UUID.
2154 Frequency : always, for aggregate stripe size over 4GB
2156 Description: "lfs setstripe" fails assertion when setting 4GB+ stripe width
2157 Details : Using "lfs setstripe" to set stripe size * stripe count over 4GB
2158 will fail the kernel with "ASSERTION(lsm->lsm_xfersize != 0)"
2161 Frequency : always if "lfs find" used on a local file/directory
2163 Description: "lfs find" segfaults if used on a local file/directory
2164 Details : The case where a directory component was not specified wasn't
2165 handled correctly. Handle this properly.
2168 Frequency : always on ppc64
2170 Description: the write to an ext3 filesystem mounted with mballoc got stuck
2171 Details : ext3_mb_generate_buddy() uses find_next_bit() which does not
2172 perform endianness conversion.
2175 Frequency : rarely (truncate to non-zero file size after write under load)
2176 Bugzilla : 10730, 10687
2177 Description: Files padded with zeros to next 4K multiple
2178 Details : With filesystems mounted using the "extents" option (2.6 kernels)
2179 it is possible that files that are truncated to a non-zero size
2180 immediately after being written are filled with zero bytes beyond
2181 the truncated size. No file data is lost.
2183 Severity : enhancement
2185 Description: Allow recovery/failover for liblustre clients.
2186 Details : liblustre clients were unaware of failover configurations until
2189 Severity : enhancement
2191 Description: user file locks should fail when not mounting with flock option
2192 Details : Set up an error-returning stub in ll_file_operations.lock field
2193 to prevent incorrect behaviour when client is mounted without
2194 flock option. Also, set up properly f_op->flock field for
2198 Frequency : always on ia64
2200 Description: "lfs df" loops on printing out MDS statfs information
2201 Details : The obd_ioctl_data was not initialized and in some systems
2202 this caused a failure during the ioctl that did not return
2203 an error. Initialize the struct and return an error on failure.
2206 Frequency : SLES 9 only
2208 Description: Error of copying files with lustre special EAs as root
2209 Details : Client side always return success for setxattr call for lustre
2210 special xattr (currently only "trusted.lov").
2213 Frequency : rarely on clusters with both ia64+i386 clients
2215 Description: ia64+i686 clients doing shared IO on the same file may LBUG
2216 Details : In rare cases when both ia64+i686 (or other mixed-PAGE_SIZE)
2217 clients are doing concurrent writes to the same file it is
2218 possible that the ia64 clients may LASSERT because the OST
2219 extent locks are not PAGE_SIZE aligned. Ensure that grown
2220 locks are always aligned on the request boundary.
2223 Frequency : specific use, occasional
2225 Description: Overwriting in use executable truncates on-disk binary image
2226 Details : If one node attempts to overwrite an executable in use by
2227 another node, we now correctly return ETXTBSY instead of
2228 truncating the file.
2230 Severity : enhancement
2232 Description: Async OSC create to avoid the blocking unnecessarily.
2233 Details : If a OST has no remain object, system will block on the creating
2234 when need to create a new object on this OST. Now, ways use
2235 pre-created objects when available, instead of blocking on an
2236 empty osc while others are not empty. If we must block, we block
2237 for the shortest possible period of time.
2242 Description: chmod on Lustre root is propagated to other clients
2243 Details : Re-validate root's dentry in ll_lookup_it to avoid having it
2244 invalid by the follow_mount time.
2247 Frequency : liblustre clients only
2249 Description: Race in 'instant cancel' lock handling could lead to such locks
2250 never to be granted in case of SMP MDS
2251 Details : Do not destroy not yet granted but cbpending locks in
2255 Frequency : replay/resend of open
2257 Description: non null lock assetion failure in mds_intent_policy
2258 Details : Trying to replay/resend lockless open requests resulted in
2259 mds_open() returning 0 with no lock. Now it sets a flag if
2260 a lock is going to be returned.
2262 Severity : enhancement
2264 Description: Checksum enhancements
2265 Details : New checksum enhancements allow for resending RPCs that failed
2268 Severity : enhancement
2270 Description: Tunables on number of dirty pages in cacche
2271 Details : Allow to set limit on number of dirty pages cached.
2276 Description: client crash on unmount - lock still has references
2277 Details : In some error handling cases it was possible to leak a lock
2278 reference on a client while accessing a file. This was not
2279 harmful to the client during operation, but would cause the
2280 client to crash when the filesystem is unmounted.
2283 Frequency : specific case, rare
2285 Description: ETXTBSY on mds though file not in use by client
2286 Details : ETXTBSY is no longer incorrectly returned when attempting to
2287 chmod or chown a directory that the user previously tried to
2288 execute or a currently-executing binary.
2291 Frequency : extremely rare except on liblustre-based clients
2293 Description: Lustre space not freed when files are deleted
2294 Details : Clean up open-unlinked files after client eviction. Previously
2295 the unlink was skipped and the files remained as orphans.
2300 Description: OST failure "would be an LBUG" in waiting_locks_callback()
2301 Details : In some cases it was possible to send a blocking callback to
2302 a client doing a glimpse, even though that client didn't get
2303 a lock granted. When the glimpse lock is cancelled on the OST
2304 the freed lock is left on the waiting list and corrupted the list.
2307 Frequency : all core dumps
2309 Description: Broke core dumps to lustre
2310 Details : Negative dentry may be unhashed if parent does not have UPDATE
2311 lock, but some callers, e.g. do_coredump, expect dentry to be
2312 hashed after successful create, hash it in ll_create_it.
2314 ------------------------------------------------------------------------------
2316 2006-09-13 Cluster File Systems, Inc. <info@clusterfs.com>
2318 * Support for kernels:
2319 2.6.9-42.0.2.EL (RHEL 4)
2320 2.6.5-7.276 (SLES 9)
2321 2.4.21-40.EL (RHEL 3)
2322 2.6.12.6 vanilla (kernel.org)
2326 Frequency : always on RHEL 3
2328 Description: Number of open files grows over time
2329 Details : The number of open files grows over time, whether or not
2330 Lustre is started. This was due to a filp leak introduced
2331 by one of our kernel patches.
2333 ------------------------------------------------------------------------------
2335 08-20-2006 Cluster File Systems, Inc. <info@clusterfs.com>
2337 * Support for kernels:
2338 2.6.9-42.EL (RHEL 4)
2339 2.6.5-7.276 (SLES 9)
2340 2.4.21-40.EL (RHEL 3)
2341 2.6.12.6 vanilla (kernel.org)
2346 Bugzilla : 5719, 9635, 9792, 9684
2347 Description: OST (or MDS) trips assertions in (re)connection under heavy load
2348 Details : If a server is under heavy load and cannot reply to new
2349 connection requests before the client resends the (re)connect,
2350 the connection handling code can behave badly if two service
2351 threads are concurrently handing separate (re)connections from
2352 the same client. Add better locking to the connection handling
2353 code, and ensure that only a single connection will be processed
2354 for a given client UUID, even if the lock is dropped.
2356 Severity : enhancement
2358 Description: add TCP zero-copy support to kernel
2359 Details : Add support to the kernel TCP stack to allow zero-copy bulk
2360 sends if the hardware supports scatter-gather and checksumming.
2361 This allows socklnd to do client-write and server-read more
2362 efficiently and reduce CPU utilization from skbuf copying.
2365 Frequency : only if NFS exporting from client
2367 Description: NULL pointer deref in ll_iocontrol() if chattr mknod file
2368 Details : If setting attributes on a file created under NFS that had
2369 never been opened it would be possible to oops the client
2370 if the file had no objects.
2373 Frequency : always for liblustre
2375 Description: liblustre client does MDS+OSTs setattr RPC for each write
2376 Details : When doing a write from a liblustre client, the client
2377 incorrectly issued an RPC to the MDS and each OST the file was
2378 striped over in order to update the timestamps. When writing
2379 with small chunks and many clients this could overwhelm the MDS
2380 with RPCs. In all cases it would slow down the write because
2381 these RPCs are unnecessary.
2383 Severity : enhancement
2385 Description: allow number of MDS service threads to be changed at module load
2386 Details : It is now possible to change the number of MDS service threads
2387 running. Adding "options mds mds_num_threads={N}" to the MDS's
2388 /etc/modprobe.conf will set the number of threads for the next
2389 time Lustre is restarted (assuming the "mds" module is also
2390 reloaded at that time). The default number of threads will
2391 stay the same, 32 for most systems.
2396 Description: OST crash if filesystem is unformatted or corrupt
2397 Details : If an OST is started on a device that has never been formatted
2398 or if the filesystem is corrupt and cannot even mount then the
2399 error handling cleanup routines would dereference a NULL pointer.
2404 Description: NULL pointer deref in llap_from_page.
2405 Details : get_cache_page_nowait can return a page with NULL (or otherwise
2406 incorrect) mapping if the page was truncated/reclaimed while it was
2407 searched for. Check for this condition and skip such pages when
2408 doing readahead. Introduce extra check to llap_from_page() to
2409 verify page->mapping->host is non-NULL (so page is not anonymous).
2412 Frequency : Sometimes when using sys_sendfile
2414 Description: "page not covered by a lock" warnings from ll_readpage
2415 Details : sendfile called ll_readpage without right page locks present.
2416 Now we introduced ll_file_sendfile that does necessary locking
2417 around call to generic_file_sendfile() much like we do in
2421 Frequency : with certain MDS communication failures at client mount time
2423 Description: NULL pointer deref after failed client mount
2424 Details : a client connection request may delayed by the network layer
2425 and not be sent until after the PTLRPC layer has timed out the
2426 request. If the client fails the mount immediately it will try
2427 to clean up before the network times out the request. Add a
2428 reference from the request import to the obd device and delay
2429 the cleanup until the network drops the request.
2432 Frequency : occasionally during client (re)connect
2434 Description: assertion failure during client (re)connect
2435 Details : processing a client connection request may be delayed by the
2436 client or server longer than the client connect timeout. This
2437 causes the client to resend the connection request. If the
2438 original connection request is replied in this interval, the
2439 client may trip an assertion failure in ptlrpc_connect_interpret()
2440 which thought it would be the only running connect process.
2443 Frequency : only with obd_echo servers and clients that are rebooted
2445 Description: kernel BUG accessing uninitialized data structure
2446 Details : When running an obd_echo server it did not start the ping_evictor
2447 thread, and when a client was evicted an uninitialized data
2448 structure was accessed. Start the ping_evictor in the RPC
2449 service startup instead of the OBD startup.
2451 Severity : enhancement
2452 Bugzilla : 10193 (patchless)
2453 Description: Remove dependency on various unexported kernel interfaces.
2454 Details : No longer need reparent_to_init, exit_mm, exit_files,
2455 sock_getsockopt, filemap_populate, FMODE_EXEC, put_filp.
2458 Frequency : rare (only users of deprecated and unsupported LDAP config)
2460 Description: write_conf for zeroconf mount queried LDAP incorrectly for client
2461 Details : LDAP apparently contains 'lustreName' attributes instead of
2462 'name'. A simple remapping of the name is sufficient.
2465 Frequency : rare (only with non-default dump_on_timeout debug enabled)
2467 Description: waiting_locks_callback trips kernel BUG if client is evicted
2468 Details : Running with the dump_on_timeout debug flag turned on makes
2469 it possible that the waiting_locks_callback() can try to dump
2470 the Lustre kernel debug logs from an interrupt handler. Defer
2471 this log dumping to the expired_lock_main() thread.
2473 Severity : enhancement
2475 Description: Support NFS exporting on 2.6 kernels.
2476 Details : Implement non-rawops metadata methods for NFS server to use without
2477 changing NFS server code.
2480 Frequency : very rare (synthetic metadata workload only)
2482 Description: two racing renames might cause an MDS thread to deadlock
2483 Details : Running the "racer" program may cause one MDS thread to rename
2484 a file from being the source of a rename to being the target of
2485 a rename at exactly the same time that another thread is doing
2486 so, and the second thread has already enqueued these locks after
2487 doing a lookup of the target and is trying to relock them in
2488 order. Ensure that we don't try to re-lock the same resource.
2491 Frequency : only very large systems with liblustre clients
2493 Description: slow eviction of liblustre clients with the "evict_by_nid" RPC
2494 Details : Use asynchronous set_info RPCs to send the "evict_by_nid" to
2495 all OSTs in parallel. This allows the eviction of stale liblustre
2496 clients to proceed much faster than if they were done in series,
2497 and also offers similar improvements for other set_info RPCs.
2502 Description: excessive CPU usage during initial read phase on client
2503 Details : During the initial read phase on a client, it would agressively
2504 retry readahead on the file, consuming too much CPU and impacting
2505 performance (since 1.4.5.8). Improve the readahead algorithm
2506 to avoid this, and also improve some other common cases (read
2507 of small files in particular, where "small" is files smaller than
2508 /proc/fs/lustre/llite/*/max_read_ahead_whole_mb, 2MB by default).
2513 Description: MDS crash when receiving packet with unknown intent.
2514 Details : Do not LBUG in unknown intent case, just return -EFAULT
2516 Severity : enhancement
2517 Bugzilla : 9293, 9385
2518 Description: MDS RPCs are serialised on client. This is unnecessary for some.
2519 Details : Do not serialize getattr (non-intent version) and statfs.
2522 Frequency : occasional, when OST network is overloaded/intermittent
2524 Description: client evicted by OST after bulk IO timeout
2525 Details : If a client sends a bulk IO request (read or write) the OST
2526 may evict the client if it is unresposive to its data GET/PUT
2527 request. This is incorrect if the network is overloaded (takes
2528 too long to transfer the RPC data) or dropped the OST GET/PUT
2529 request. There is no need to evict the client at all, since
2530 the pinger and/or lock callbacks will handle this, and the
2531 client can restart the bulk request.
2534 Frequency : Always when mmapping file with no objects
2536 Description: client crashes when mmapping file with no objects
2537 Details : Check that we actually have objects in a file before doing any
2538 operations on objects in ll_vm_open, ll_vm_close and
2544 Description: Request leak when working with deleted CWD
2545 Details : Introduce advanced request refcount tracking for requests
2546 referenced from lustre intent.
2548 Severity : Enhancement
2550 Description: Cache open file handles on client.
2551 Details : MDS now will return special lock along with openhandle, if
2552 requested and client is allowed to hold openhandle, even if unused,
2553 until such a lock is revoked. Helps NFS a lot, since NFS is opening
2554 closing files for every read/write openration.
2556 Severity : Enhancement
2558 Description: Cache open negative dentries on client when possible.
2559 Details : Guard negative dentries with UPDATE lock on parent dir, drop
2560 negative dentries on lock revocation.
2565 Description: Remounting a client read-only wasn't possible with a zconf mount
2566 Details : It wasn't possible to remount a client read-only with llmount.
2568 Severity : enhancement
2569 Description: Include MPICH 1.2.6 Lustre ADIO interface patch
2570 Details : In lustre/contrib/ or /usr/share/lustre in RPM a patch for
2571 MPICH is included to add Lustre-specific ADIO interfaces.
2572 This is based closely on the UFS ADIO layer and only differs
2573 in file creation, in order to allow the OST striping to be set.
2574 This is user-contributed code and not supported by CFS.
2579 Description: extended inode attributes (immutable, append-only) work improperly
2580 when 2.4 and 2.6 kernels are used on client/server or vice versa
2581 Details : Introduce kernel-independent values for these flags.
2583 Severity : enhancement
2586 Description: Allow fractional MB tunings for lustre in /proc/ filesystem.
2587 Details : Many of the /proc/ tunables can only be tuned at a megabyte
2588 granularity. Now, Fractional MB granularity is be supported,
2589 this is very useful for low memory system.
2591 Severity : enhancement
2593 Description: Getattr by fid
2594 Details : Getting a file attributes by its fid, obtaining UPDATE|LOOKUP
2595 locks, avoids extra getattr rpc requests to MDS, allows '/' to
2596 have locks and avoids getattr rpc requests for it on every stat.
2599 Frequency : Always, for filesystems larger than 2TB
2601 Description: ldiskfs crash at mount for filesystem larger than 2TB with mballoc
2602 Details : Kenrel kmalloc limits allocations to 128kB and this prevents
2603 filesystems larger than 2TB to be mounted with mballoc enabled.
2606 Frequency : Always, for 32-bit kernel without CONFIG_LBD and filesystem > 2TB
2608 Description: ldiskfs crash at mount for filesystem larger than 2TB with mballoc
2609 Details : If a 32-bit kernel is compiled without CONFIG_LBD enabled and a
2610 filesystems larger than 2TB is mounted then the kernel will
2611 silently corrupt the start of the filesystem. CONFIG_LBD is
2612 enabled for all CFS-supported kernels, but the possibility of
2613 this happening with a modified kernel config exists.
2615 Severity : enhancement
2617 Description: add client O_DIRECT support for 2.6 kernels
2618 Details : It is now possible to do O_DIRECT reads and writes to files
2619 in the Lustre client mountpoint on 2.6 kernel clients.
2621 Severity : enhancement
2623 Description: parallel glimpse, setattr, statfs, punch, destroy requests
2624 Details : Sends glimpse, setattr, statfs, punch, destroy requests to OSTs in
2625 parallel, not waiting for response from every OST before sending
2626 a rpc to the next OST.
2631 Description: setattr vs write race when updating file timestamps
2632 Details : Client processes that update a file timestamp into the past
2633 right after writing to the file (e.g. tar) it is possible that
2634 the updated file modification time can be reset to the current
2635 time due to a race between processing the setattr and write RPC.
2637 Severity : enhancement
2639 Description: Bring 'lfs find' closer in line with regular Linux find.
2640 Details : lfs find util supports -atime, -mtime, -ctime, -maxdepth, -print,
2641 -print0 options and obtains all the needed info through the lustre
2644 Severity : enhancement
2646 Description: support up to 1024 configured devices on one node
2647 Details : change obd_dev array from statically allocated to dynamically
2648 allocated structs as they are first used to reduce memory usage
2653 Description: Flush dirty partially truncated pages during truncate
2654 Details : Immediatelly flush partially truncated pages in filter_setattr,
2655 this way we completely avoid having any pages in page cache on OST
2656 and can retire ugly workarounds during writes to flush such pages.
2661 Description: i_sem vs transaction deadlock in mds_obd_destroy during unlink.
2662 Details : protect inode from truncation within vfs_unlink() context
2663 just take a reference before calling vfs_unlink() and release it
2664 when parent's i_sem is free.
2669 Description: last_id value checked outside lock on OST caused LASSERT failure
2670 Details : If there were multiple MDS->OST object precreate requests in
2671 flight, it was possible that the OST's last object id was checked
2672 outside a lock and incorrectly tripped an assertion. Move checks
2673 inside locks, and discard old precreate requests.
2676 Frequency : always, if extents are used on OSTs
2678 Description: index ei_leaf_hi (48-bit extension) is not zeroed in extent index
2679 Details : OSTs using the extents format would not zero the high 16 bits of
2680 the index physical block number. This is not a problem for any
2681 OST filesystems smaller than 16TB, and no kernels support ext3
2682 filesystems larger than 16TB yet. This is fixed in 1.4.7 (all
2683 new/modified files) and can be fixed for existing filesystems
2684 with e2fsprogs-1.39-cfs1.
2689 Description: import connection selection may be incorrect if timer wraps
2690 Details : Using a 32-bit jiffies timer with HZ=1000 may cause backup
2691 import connections to be ignored if the 32-bit jiffies counter
2692 wraps. Use a 64-bit jiffies counter.
2695 Frequency : very large clusters immediately after boot
2697 Description: LNET request buffers exhausted under heavy short-term load
2698 Details : If a large number of client requests are generated on a service
2699 that has previously never seen so many requests it is possible
2700 that the request buffer growth cannot keep up with the spike in
2701 demand. Instead of dropping incoming requests, they are held in
2702 the LND until the RPC service can accept more requests.
2705 Frequency : Sometimes during replay
2707 Description: Assertion failure in ll_local_open after replay.
2708 Details : If replay happened on an open request reply before we were able
2709 to set replay handler, reply will become not swabbed tripping the
2710 assertion in ll_local_open. Now we set the handler right after
2711 recognising of open request
2714 Frequency : very rare
2716 Description: kernel reports "badness in vsnprintf"
2717 Details : Reading from the "recovery_status" /proc file in small chunks
2718 may cause a negative length in lprocfs_obd_rd_recovery_status()
2719 call to vsnprintf() (which is otherwise harmless). Exit early
2720 if there is no more space in the output buffer.
2722 Severity : enhancement
2724 Description: clear OBD RPC statistics by writing to them
2725 Details : It is now possible to clear the OBD RPC statistics by writing
2726 to the "stats" file.
2731 Description: Client mtime is not the same on different clients after utimes
2732 Details : In some cases, the client was using the utimes() syscall on
2733 a file cached on another node. The clients now validate the
2734 ctime from the MDS + OSTs to determine which one is right.
2739 Description: Inability to activate failout mode
2740 Details : lconf script incorrectly assumed that in pythong string's numeric
2741 value is used in comparisons.
2744 Frequency : always with multiple stripes per file
2746 Description: Inefficient object allocation for mutli-stripe files
2747 Details : When selecting which OSTs to stripe files over, for files with
2748 a stripe count that divides evenly into the number of OSTs,
2749 the MDS is always picking the same starting OST for each file.
2750 Return the OST selection heuristic to the original design.
2755 Description: mount failures may take full timeout to return an error
2756 Details : Under some heavy load conditions it is possible that a
2757 failed mount can wait for the full obd_timeout interval,
2758 possibly several minutes, before reporting an error.
2759 Instead return an error as soon as the status is known.
2761 Frequency : quota enabled and large files being deleted
2763 Description: releasing more than 4GB of quota at once hangs OST
2764 Details : If a user deletes more than 4GB of files on a single OST it
2765 will cause the OST to spin in an infinite loop. Release
2766 quota in < 4GB chunks, or use a 64-bit value for 1.4.7.1+.
2771 Description: statfs data retrieved from /proc may be stale or zero
2772 Details : When reading per-device statfs data from /proc, in the
2773 {kbytes,files}_{total,free,avail} files, it may appear
2774 as zero or be out of date.
2777 Frequency : systems with MD RAID1 external journal devices
2779 Description: lconf's call to blkid is confused by RAID1 journal devices
2780 Details : Use the "blkid -l" flag to locate the MD RAID device instead
2781 of returning all block devices that match the journal UUID.
2784 Frequency : always, for aggregate stripe size over 4GB
2786 Description: assertion fails when trying to use 4GB stripe size
2787 Details : Use "setstripe" to set stripe size over 4GB will fail the kernel,
2788 complaining "ASSERTION(lsm->lsm_xfersize != 0)"
2791 Frequency : always on ppc64
2793 Description: the first write on an ext3 filesystem with mballoc got stuck
2794 Details : ext3_mb_generate_buddy() uses find_next_bit() which does not
2795 perform endianness conversion.
2797 ------------------------------------------------------------------------------
2799 02-14-2006 Cluster File Systems, Inc. <info@clusterfs.com>
2801 * WIRE PROTOCOL CHANGE. This version of Lustre networking WILL NOT
2802 INTEROPERATE with older versions automatically. Please read the
2803 user documentation before upgrading any part of a live system.
2804 * WARNING: Lustre networking configuration changes are required with
2805 this release. See https://bugzilla.clusterfs.com/show_bug.cgi?id=10052
2808 * Support for kernels:
2809 2.6.9-22.0.2.EL (RHEL 4)
2810 2.6.5-7.244 (SLES 9)
2811 2.6.12.6 vanilla (kernel.org)
2814 Severity : enhancement
2815 Bugzilla : 7981/8208
2816 Description: Introduced Lustre Networking (LNET)
2817 Details : LNET is new networking infrastructure for Lustre, it includes
2818 a reorganized network configuration mode (see the user
2819 documentation for full details) as well as support for routing
2820 between different network fabrics. Lustre Networking Devices
2821 (LNDs) for the supported network fabrics have also been
2822 created for this new infrastructure.
2824 Severity : enhancement
2825 Description: Introduced Access control lists
2826 Details : clients can set ACLs on files and directories in order to have
2827 more fine-grained permissions than the standard Unix UGO+RWX.
2828 The MDS must be started with the "-o acl" mount option.
2830 Severity : enhancement
2831 Description: Introduced filesystem quotas
2832 Details : Administrators may now establish per-user quotas on the
2835 Severity : enhancement
2837 Description: Configuration change for the XT3
2838 The PTLLND is now used to run Lustre over Portals on the XT3
2839 The configure option(s) --with-cray-portals are no longer used.
2840 Rather --with-portals=<path-to-portals-includes> is used to
2841 enable building on the XT3. In addition to enable XT3 specific
2842 features the option --enable-cray-xt3 must be used.
2847 Description: Running on many-way SMP OSTs can trigger oops in llcd_send()
2848 Details : A race between allocating a new llcd and re-getting the llcd_lock
2849 allowed another thread to grab newly-allocated llcd.
2851 Severity : enhancement
2853 Description: 2.6 OST async journal commit and locking fix to improve performance
2854 Details : The filter_direct_io()+filter_commitrw_write() journal commits for
2855 2.6 kernels are now async as they already were in 2.4 kernels so
2856 that they can commit concurrently with the network bulk transfer.
2857 For block-allocated files the filter allocation semaphore is held
2858 to avoid filesystem fragmentation during allocation. BKL lock
2859 removed for 2.6 xattr operations where it is no longer needed.
2864 Description: lconf incorrectly determined whether two IP networks could talk
2865 Details : In some more complicated routing and multiple-network
2866 configurations, lconf will avoid trying to make a network
2867 connection to a disjoint part of the IP space. It was doing the
2868 math incorrectly for one set of cases.
2873 Description: Fix for potential infinite loop processing records in an llog.
2874 Details : If an llog record is corrupted/zeroed, it is possible to loop
2875 forever in llog_process(). Validate the llog record length
2876 and skip the remainder of the block on error.
2879 Frequency : occasional (liblustre only)
2881 Description: liblustre could not open files whose last component is a symlink
2882 Details : sysio_path_walk() would incorrectly pass the open intent to
2883 intermediate path components.
2886 Frequency : rare (liblustre only with non-standard tuning)
2887 Bugzilla : 7201 (7350)
2888 Description: Tuning the MDC DLM LRU size to zero triggers client LASSERT
2889 Details : llu_lookup_finish_locks() tries to set lock data on a lock
2890 after it has been released, only do this for referenced locks
2892 Severity : enhancement
2894 Description: specifying an (invalid) directory default stripe_size of -1
2895 would reset the directory default striping
2896 Details : stripe_size -1 was used internally to signal directory stripe
2897 removal, now use "all default" to signal dir stripe removal
2898 as a directory striping of "all default" is not useful
2901 Frequency : common for large clusters running liblustre clients
2903 Description: doing an ls when liblustre clients are running is slow
2904 Details : sending a glimpse AST to a liblustre client waits for every AST
2905 to time out, as liblustre clients will not respond. Since they
2906 cannot cache data we refresh the OST lock LVB from disk instead.
2908 Severity : enhancement
2910 Description: doing an ls at the same time as file IO can be slow
2911 Details : enqueue and other "small" requests can be blocked behind many
2912 large IO requests. Create a new OST IO portal for non-IO
2913 requests so they can be processed faster.
2916 Frequency : rare (only HPUX clients mounting unsupported re-exported NFS vol)
2918 Description: an HPUX NFS client would get -EACCESS when ftruncate()ing a newly
2919 created file with mode 000
2920 Details : the Linux NFS server relies on an MDS_OPEN_OWNEROVERRIDE hack to
2921 allow an ftruncate() as a non-root user to a file with mode 000.
2922 Lustre now respects this flag to disable mode checks when
2923 truncating a file owned by the user
2926 Frequency : liblustre-only, when liblustre client dies unexpectedly or becomes
2929 Description: Revoking locks from clients that went dead or catatonic might take
2931 Details : New lock flags FL_CANCEL_ON_BLOCK used by liblustre makes
2932 cancellation of such locks instant on servers without waiting for
2933 any reply from clients. Clients drops these locks when cancel
2934 notification from server is received without replying.
2937 Frequency : liblustre-only, when liblustre client dies or becomes busy
2939 Description: Doing ls on Linux clients can take a long time with active
2941 Details : Liblustre client cannot handle ASTs in timely manner, so avoid
2942 granting such locks to it in the first place if possible. Locks
2943 are taken by proxy on the OST during the read or write and
2944 dropped immediately afterward. Add connect flags handling, do
2945 not grant locks to liblustre clients for glimpse ASTs.
2947 Severity : enhancement
2949 Description: Improve read-ahead algorithm to avoid excessive IO for random reads
2950 Details : Existing read-ahead algorithm is tuned for the case of streamlined
2951 sequential reads and behaves badly with applications doing random
2952 reads. Improve it by reading ahead at least read region, and
2953 avoiding excessive large RPC for small reads.
2955 Severity : enhancement
2957 Description: Creating more than 1000 files for a single job may cause a load
2958 imbalance on the OSTs if there are also a large number of OSTs.
2959 Details : qos_prep_create() uses an OST index reseed value that is an
2960 even multiple of the number of available OSTs so that if the
2961 reseed happens in the middle of the object allocation it will
2962 still utilize the OSTs as uniformly as possible.
2967 Description: OST or MDS may oops in ping_evictor_main()
2968 Details : ping_evictor_main() drops obd_dev_lock if deleting a stale export
2969 but doesn't restart at beginning of obd_exports_timed list
2972 Severity : enhancement
2974 Description: improve by-nid export eviction on the MDS and OST
2975 Details : allow multiple exports with the same NID to be evicted at one
2976 time without re-searching the exports list.
2979 Frequency : rare, only with supplementary groups enabled on SMP 2.6 kernels
2981 Description: MDS may oops in groups_free()
2982 Details : in rare race conditions a newly allocated group_info struct is
2983 freed again, and this can be NULL. The 2.4 compatibility code
2984 for groups_free() checked for a NULL pointer, but 2.6 did not.
2987 Frequency : common for liblustre clients doing little filesystem IO
2988 Bugzilla : 9352, 7313
2989 Description: server may evict liblustre clients accessing contended locks
2990 Details : if a client is granted a lock or receives a completion AST
2991 with a blocking AST already set it would not reply to the AST
2992 for LDLM_FL_CANCEL_ON_BLOCK locks. It now replies to such ASTs.
2995 Frequency : lfs setstripe, only systems with more than 160 OSTs
2997 Description: unable to set striping with a starting offset beyond OST 160
2998 Details : llapi_create_file() incorrectly limited the starting stripe
2999 index to the maximum single-file stripe count.
3002 Frequency : LDAP users only
3004 Description: lconf did not handle in-kernel recovery with LDAP properly
3005 Details : lconf/LustreDB get_refs() is searching the wrong namespace
3007 Severity : enhancement
3009 Description: bind OST threads to NUMA nodes to improve performance
3010 Details : all OST threads are uniformly bound to CPUs on a single NUMA
3011 node and do their allocations there to localize memory access
3013 Severity : enhancement
3015 Description: llmount can determine client NID directly from Myrinet (GM)
3016 Details : the client NID code from gmnalnid was moved directly into
3017 llmount, removing the need to use this or specifying the
3018 client NID explicitly when mounting GM clients with zeroconf
3021 Frequency : if client is started with down MDS
3023 Description: if client is started with down MDS mount hangs in ptlrpc_queue_wait
3024 Details : Having an LWI_INTR() wait event (interruptible, but no timeout)
3025 will wait indefinitely in ptlrpc_queue_wait->l_wait_event() after
3026 ptlrpc_import_delayed_req() because we didn't check if the
3027 request was interrupted, and we also didn't break out of the
3028 event loop if there was no timeout
3033 Description: data loss during non-page-aligned writes to a single file from
3034 both multiple nodes and multiple threads on one node at same time
3035 Details : updates to KMS and lsm weren't protected by common lock. Resulting
3036 inconsistency led to false short-reads, that were cached and later
3037 used by ->prepare_write() to fill in partially written page,
3038 leading to data loss.
3041 Frequency : always, if lconf --abort_recovery used
3043 Description: lconf --abort_recovery fails with 'Operation not supported'
3044 Details : lconf was attempting to abort recovery on the MDT device and not
3047 Severity : enhancement
3049 Description: remove cleanup logs
3050 Details : replace lconf-generated cleanup logs with lustre internal
3051 cleanup routines. Eliminates the need for client-cleanup and
3054 Severity : enhancement
3056 Description: add support for EAs (user and system) on lustre filesystems
3057 Details : it is now possible to store extended attributes in the Lustre
3058 client filesystem, and with the user_xattr mount option it
3059 is possible to allow users to store EAs on their files also
3061 Severity : enhancement
3063 Description: Add possibility (config option) to show minimal available OST free
3065 Details : When compiled with --enable-mindf configure option, statfs(2)
3066 (and so, df) will return least minimal free space available from
3067 all OSTs as amount of free space on FS, instead of summary of
3068 free spaces of all OSTs.
3070 Severity : enhancement
3072 Description: do not expand extent locks acquired on OST-side
3073 Details : Modify ldlm_extent_policy() to not expand local locks, acquired
3074 by server: they are not cached anyway.
3077 Frequency : when mmap is used/binaries executed from Lustre
3079 Description: Unmmap pages before throwing them away from read cache.
3080 Details : llap_shrink cache now attempts to unmap pages before discarding
3081 them (if unmapping failed - do not discard). SLES9 kernel has
3082 extra checks that trigger if this unmapping is not done first.
3087 Description: lconf didn't resolve symlinks before checking to see whether a
3088 given mountpoint was already in use
3091 Frequency : when migrating failover services
3092 Bugzilla : 6395, 9514
3093 Description: When migrating a subset of services from a node (e.g. failback
3094 from a failover service node) the remaining services would
3095 time out and evict clients.
3096 Details : lconf --force (implied by --failover) sets the global obd_timeout
3097 to 5 seconds in order to quickly disconnect, but this caused
3098 other RPCs to time out too quickly. Do not change the global
3099 obd_timeout for force cleanup, only set it for DISCONNECT RPCs.
3101 Severity : enhancement
3102 Frequency : if MDS is started with down OST
3103 Bugzilla : 9439,5706
3104 Description: Allow startup/shutdown of an MDS without depending on the
3105 availability of the OSTs.
3106 Details : Asynchronously call mds_lov_synchronize during MDS startup.
3107 Add appropriate locking and lov-osc refcounts for safe
3108 cleaning. Add osc abort_inflight calls in case the
3109 synchronize never started.
3112 Frequency : occasional (Cray XT3 only)
3114 Description: root not authorized to access files in CRAY_PORTALS environment
3115 Details : The client process capabilities were not honoured on the MDS in
3116 a CRAY_PORTALS/CRAY_XT3 environment. If the file had previously
3117 been accessed by an authorized user then root was able to access
3118 the file on the local client also. The root user capabilities
3119 are now allowed on the MDS, as this environment has secure UID.
3122 Frequency : occasional
3124 Description: ldiskfs "too long searching" message happens too often
3125 Details : A debugging message (otherwise harmless) prints too often on
3126 the OST console. This has been reduced to only happen when
3127 there are fragmentation problems on the filesystem.
3132 Description: Division by zero in statfs when all OSCs are inactive
3133 Details : lov_get_stripecnt() returns zero due to incorrect order of checks,
3134 lov_statfs divides by value returned by lov_get_stripecnt().
3138 Bugzilla : 9489, 3273
3139 Description: First write from each client to each OST was only 4kB in size,
3140 to initialize client writeback cache, which caused sub-optimal
3141 RPCs and poor layout on disk for the first writen file.
3142 Details : Clients now request an initial cache grant at (re)connect time
3143 and so that they can start streaming writes to the cache right
3144 away and always do full-sized RPCs if there is enough data.
3145 If the OST is rebooted the client also re-establishes its grant
3146 so that client cached writes will be honoured under the grant.
3151 Description: Slow ls (and stat(2) syscall) on files residing on IO-loaded OSTs
3152 Details : Now I/O RPCs go to different portal number and (presumably) fast
3153 lock requests (and glimses) and other RPCs get their own service
3154 threads pool that should be able to service those RPCs
3157 Severity : enhancement
3159 Description: Ability to exchange lustre version between client and servers and
3160 issue warnings at client side if client is too old. Also for
3161 liblustre clients there is ability to refuse connection of too old
3163 Details : New 'version' field is added to connect data structure that is
3164 filled with version info. That info is later checked by server and
3168 Frequency : rare, liblustre only.
3169 Bugzilla : 9296, 9581
3170 Description: Two simultaneous writes from liblustre at offset within same page
3171 might proceed at the same time overwriting eachother with stale
3173 Details : I/O lock withing llu_file_prwv was released too early, before data
3174 actually was hitting the wire. Extended lock-holding time until
3175 server acknowledges receiving data.
3178 Frequency : extremely rare. Never observed in practice.
3180 Description: avoid generating lustre_handle cookie of 0.
3181 Details : class_handle_hash() generates handle cookies by incrementing
3182 global counter, and can hit 0 occasionaly (this is unlikely, but
3183 not impossible, because initial value of cookie counter is
3184 selected randonly). Value of 0 is used as a sentinel meaning
3185 "unassigned handle" --- avoid it. Also coalesce two critical
3186 sections in this function into one.
3188 Severity : enhancement
3190 Description: allow liblustre clients to delegate truncate locking to OST
3191 Details : To avoid overhead of locking, liblustre client instructs OST to
3192 take extent lock in ost_punch() on client's behalf. New connection
3193 flag is added to handle backward compatibility.
3195 Severity : enhancement
3196 Bugzilla : 4928, 7341, 9758
3197 Description: allow number of OST service threads to be specified
3198 Details : a module parameter allows the number of OST service threads
3199 to be specified via "options ost ost_num_threads={N}" in the
3200 OSS's /etc/modules.conf or /etc/modprobe.conf.
3204 Bugzilla : 6146, 9635, 9895
3205 Description: servers crash with bad pointer in target_handle_connect()
3206 Details : In rare cases when a client is reconnecting it was possible that
3207 the connection request was the last reference for that export.
3208 We would temporarily drop the export reference and get a new
3209 one, but this may have been the last reference and the export
3210 was just destroyed. Get new reference before dropping old one.
3212 Severity : enhancement
3213 Frequency : if client is started with failover MDS
3215 Description: Allow multiple MDS hostnames in the mount command
3216 Details : Try to read the configuration from all specified MDS
3217 hostnames during a client mount in case the "primary"
3220 Severity : enhancement
3222 Description: Stop sending data to evicted clients as soon as possible.
3223 Details : Check if the client we are about to send or are sending data to
3224 was evicted already. (Check is done every second of waiting,
3225 for which l_wait_event interface was extended to allow checking
3226 of exit condition at specified intervals).
3229 Frequency : rare, normally only when NFS exporting is done from client
3231 Description: 'bad disk LOV MAGIC: 0x00000000' error when chown'ing files
3233 Details : Make mds_get_md() recognise empty md case and set lmm size to 0.
3236 Frequency : always, if srand() is called before liblustre initialization
3238 Description: Liblustre uses system PRNG disturbing its usage by user application
3239 Details : Introduce internal to lustre fast and high-quality PRNG for
3240 lustre usage and make liblustre and some other places in generic
3241 lustre code to use it.
3243 Severity : enhancement
3244 Bugzilla : 9477, 9557, 9870
3245 Description: Verify that the MDS configuration logs are updated when xml is
3246 Details : Check if the .xml configuration logs are newer than the config
3247 logs stored on the MDS and report an error if this is the case.
3248 Request --write-conf, or allow starting with --old_conf.
3250 Severity : enhancement
3252 Description: Handle symlinks in the path when checking if Lustre is mounted.
3253 Details : Resolve intermediate symlinks when checking if a client has
3254 mounted a filesystem to avoid duplicate client mounts.
3259 Description: lconf can hit an error exception but still return success.
3260 Details : The lconf command catches the Command error exception at the top
3261 level script context and will exit with the associated exit
3262 status, but doesn't ensure that this exit status is non-zero.
3267 Description: failure of ptlrpc thread startup can cause oops
3268 Details : Starting a ptlrpc service thread can fail if there are a large
3269 number of threads or the server memory is very fragmented.
3270 Handle this without oopsing.
3273 Frequency : always, only if liblustre and non-default acceptor port was used
3275 Description: liblustre cannot connect to servers with non-default acceptor port
3276 Details : tcpnal_set_default_params() was not called and was therefore
3277 ignoring the environment varaible TCPNAL_PORT, as well as other
3278 TCPNAL_ environment variables
3283 Description: two objects could be created on the same OST for a single file
3284 Details : If an OST is down, in some cases it was possible to create two
3285 objects on a single OST for a single file. No problems other
3286 than potential performance impact and spurious error messages.
3290 Bugzilla : 5681, 9562
3291 Description: Client may oops in ll_unhash_aliases
3292 Details : Client dcache may become inconsistent in race condition.
3293 In some cases "getcwd" can fail if the current directory is
3299 Description: Inode refcounting problems in NFS export code
3300 Details : link_raw functions used to call d_instantiate without obtaining
3301 extra inode reference first.
3305 Bugzilla : 9942, 9903
3306 Description: Referencing freed requests leading to crash, memleaks with NFS.
3307 Details : We used to require that call to ll_revalidate_it was always
3308 followed by ll_lookup_it. Also with revalidate_special() it is
3309 possible to call ll_revalidate_it() twice for the same dentry
3310 even if first occurence returned success. This fix changes semantic
3311 between DISP_ENQ_COMPLETE disposition flag to mean there is extra
3312 reference on a request referred from the intent.
3313 ll_intent_release() then releases such a request.
3316 Frequency : rare, normally benchmark loads only
3318 Description: unlinked inodes were kept in memory on the client
3319 Details : If a client is repeatedly creating and unlinking files it
3320 can accumulate a lot of stale inodes in the inode slab cache.
3321 If there is no other client load running this can cause the
3322 client node to run out of memory. Instead flush old inodes
3323 from client cache that have the same inode number as a new inode.
3326 Frequency : SLES9 2.6.5 kernel and long filenames only
3327 Bugzilla : 9969, 10379
3328 Description: utime reports stale NFS file handle
3329 Details : SLES9 uses out-of-dentry names in some cases, which confused
3330 the lustre dentry revalidation. Change it to always use the
3334 Frequency : rare, unless heavy write-truncate concurrency is continuous
3335 Bugzilla : 4180, 6984, 7171, 9963, 9331
3336 Description: OST becomes very slow and/or deadlocked during object unlink
3337 Details : filter_destroy() was holding onto the parent directory lock
3338 while truncating+unlinking objects. For very large objects this
3339 may block other threads for a long time and slow overall OST
3340 responsiveness. It may also be possible to get a lock ordering
3341 deadlock in this case, or run out of journal credits because of
3342 the combined truncate+unlink. Solution is to do object truncate
3343 first in one transaction without parent lock, and then do the
3344 final unlink in a new transaction with the parent lock. This
3345 reduces the lock hold time dramatically.
3348 Frequency : rare, 2.4 kernels only
3350 Description: MDS or OST cleanup may trip kernel BUG when dropping kernel lock
3351 Details : mds_cleanup() and filter_cleanup() need to drop the kernel lock
3352 before unmounting their filesystem in order to avoid deadlock.
3353 The kernel_locked() function in 2.4 kernels only checks whether
3354 the kernel lock is held, not whether it is this process that is
3355 holding it as 2.6 kernels do.
3360 Description: MDS or OST may oops/LBUG if a client is connecting multiple times
3361 Details : The client ptlrpc code may be trying to reconnect to a down
3362 server before a previous connection attempt has timed out.
3363 Increase the reconnect interval to be longer than the connection
3364 timeout interval to avoid sending duplicate connections to
3368 Frequency : echo_client brw_test command
3370 Description: fix echo_client to work with OST preallocated code
3371 Details : OST preallocation code (5137) didn't take echo_client IO path
3372 into account: echo_client calls filter methods outside of any
3373 OST thread and, hence, there is no per-thread preallocated
3374 pages and buffers to use. Solution: hijack pga pages for IO. As
3375 a byproduct, this avoids unnecessary data copying.
3379 Bugzilla : 3555, 5962, 6025, 6155, 6296, 9574
3380 Description: Client can oops in mdc_commit_close() after open replay
3381 Details : It was possible for the MDS to return an open request with no
3382 transaction number in mds_finish_transno() if the client was
3383 evicted, but without actually returning an error. Clients
3384 would later try to replay that open and may trip an assertion
3385 Simplify the client close codepath, and always return an error
3386 from the MDS in case the open is not successful.
3389 Frequency : rare, 2.6 OSTs only
3391 Description: OST may deadlock under high load on fragmented files
3392 Details : If there was a heavy load and highly-fragmented OST filesystems
3393 it was possible to have all the OST threads deadlock waiting on
3394 allocation of biovecs, because the biovecs were not released
3395 until the entire RPC IO was completed. Instead, release biovecs
3396 as soon as they are complete to ensure forward IO progress.
3398 Severity : enhancement
3400 Description: Support for specifying external journal device at mount
3401 Details : If an OST or MDS device is formatted with an external journal
3402 device, this device major/minor is stored in the ext3 superblock
3403 and may not be valid for failover. Allow detecting and
3404 specifying the external journal at mount time.
3409 Description: Mounting an MDS with pending unlinked files may cause oops
3410 Details : target_finish_recovery() calls mds_postrecov() which returned
3411 the number of orphans unlinked. mds_lov_connect->mds_postsetup()
3412 considers this an error and immediately begins cleaning up the
3413 lov, just after starting the mds_lov process
3415 Severity : enhancement
3417 Description: Implement 'lfs df' to report actual free space on per-OST basis
3418 Details : Add sub-command 'df' on 'lfs' to report the disk space usage of
3419 MDS/OSDs. Usage: lfs df [-i][-h]. Command Options: '-i' to report
3420 usage of objects; '-h' to report in human readable format.
3422 ------------------------------------------------------------------------------
3424 08-26-2005 Cluster File Systems, Inc. <info@clusterfs.com>
3431 Description: Mounting an ldiskfs file system with mballoc may crash OST node.
3432 Details : ldiskfs mballoc code may reference an uninitialized buddy struct
3433 at startup during orphan unlinking. Instead, skip buddy update
3434 before setup, as it will be regenerated after recovery is complete.
3439 Description: If an OST is inactive, its locks might reference stale inodes.
3440 Details : lov_change_cbdata() must iterate over all namespaces, even if
3441 they are inactive to clear inode references from the lock.
3443 Severity : enhancement
3444 Frequency : occasional, if non-standard max_dirty_mb used
3446 Description: Client will block write RPCs if not enough grant
3447 Details : If a client has max_dirty_mb smaller than max_rpcs_in_flight,
3448 then the client will block writes while waiting for another RPC
3449 to complete instead of consuming its dirty limit. With change
3450 we get improved performance when max_dirty_mb is small.
3452 Severity : enhancement
3453 Bugzilla : 3389, 6253
3454 Description: Add support for supplementary groups on the MDS.
3455 Details : The MDS has an upcall /proc/fs/lustre/mds/{mds}/group_upcall
3456 (set to /usr/sbin/l_getgroups if enabled) which will do MDS-side
3457 lookups for user supplementary groups into a cache.
3461 Description: O_CREAT|O_EXCL open flags in liblustre always return -EEXIST
3462 Details : Make libsysio to not enforce O_EXCL by clearing the flag,
3463 for liblustre O_EXCL is enforced by MDS.
3467 Description: readdir never returns NULL in liblustre.
3468 Details : Corrected llu_iop_getdirentries logic, to return offset of next
3469 dentry in struct dirent.
3473 Frequency : liblustre only, depends on application IO pattern
3474 Description: liblustre clients evicted if not contacting servers
3475 Details : Don't put liblustre clients into the ping_evictor list, so
3476 they will not be evicted by the pinger ever.
3478 Severity : enhancement
3480 Description: Add ability to evict clients by NID from MDS.
3481 Details : By echoing "nid:$NID" string into
3482 /proc/fs/lustre/mds/.../evict_client client with nid that equals to
3483 $NID would be instantly evicted from this MDS and from all active
3484 OSTs connected to it.
3488 Description: Do not query file size twice, somewhat slowing stat(2) calls.
3489 Details : lookup_it_finish() used to query file size from OSTs that was not
3494 Description: service threads change working directory to that of init
3495 Details : Starting lustre service threads may pin the working directory
3496 of the parent thread, making that filesystem busy. Threads
3497 now change to the working directory of init to avoid this.
3501 Frequency : during shutdown only
3502 Description: shutdown with a failed MDS or OST can cause unmount to hang
3503 Details : Don't resend DISCONNECT messages in ptlrpc_disconnect_import()
3508 Frequency : 2.6 only
3509 Description: chmod/chown may include an extra supplementary group
3510 Details : ll{,u}_mdc_pack_op_data() does not properly initialize the
3511 supplementary group and if none is specified this is used.
3514 Bugzilla : 5479 (6816)
3516 Description: Racing open + rm can assert client in mdc_set_open_replay_data()
3517 Details : If lookup is in progress on a file that is unlinked we might try
3518 to revalidate the inode and fail in revalidate after lookup is
3519 complete and ll_file_open() enqueues the open again but
3520 it_open_error() was not checking DISP_OPEN_OPEN errors correctly.
3523 Frequency : always, if lconf --abort_recovery used
3525 Description: lconf --abort_recovery fails with 'Operation not supported'
3526 Details : lconf was attempting to abort recovery on the MDT device and not
3529 ------------------------------------------------------------------------------
3531 2005-08-08 Cluster File Systems, Inc. <info@clusterfs.com>
3536 Frequency : rare (only unsupported configurations with a node running as an
3538 Bugzilla : 6514, 5137
3539 Description: Mounting a Lustre file system on a node running as an OST could
3541 Details : OSTs now preallocates memory needed to write out data at
3542 startup, instead of when needed, to avoid having to
3543 allocate memory in possibly low memory situations.
3544 Specifically, if the file system is mounted on on OST,
3545 memory pressure could force it to try to write out data,
3546 which it needed to allocate memory to do. Due to the low
3547 memory, it would be unable to do so and the node would
3548 become unresponsive.
3550 Severity : enhancement
3552 Description: Addition of lconf --service command line option
3553 Details : lconf now accepts a '--service <arg>' option, which is
3554 shorthand for 'lconf --group <arg> --select <arg>=<hostname>'
3556 Severity : enhancement
3558 Description: Failover mode is now the default for OSTs.
3559 Details : By default, OSTs will now run in failover mode. To return to
3560 the old behaviour, add '--failout' to the lmc line for OSTs.
3562 Severity : enhancement
3564 Description: Health checks are now provided for MDS and OSTs
3565 Details : Additional detailed health check information on MSD and OSTs
3566 is now provided through the procfs health_check value.
3569 Frequency : occasional, depends on IO load
3571 Description: Disk fragmentation on the OSTs could eventually cause slowdowns
3572 after numerous create/delete cycles
3573 Details : The ext3 inode allocation policy would not allocate new inodes
3574 very well on the OSTs because there are no new directories
3575 being created. Instead we look for groups with free space if
3576 the parent directories are nearly full.
3581 Description: Network or server problems during mount may cause partially
3582 mounted clients instead of returning an error.
3583 Details : The config llog parsing code may overwrite the error return
3584 code during mount error handling, returning success instead
3590 Description: MDS can fail to allocate large reply buffers
3591 Details : After long uptimes the MDS can fail to allocate large reply
3592 buffers (e.g. zconf client mount config records) due to memory
3593 fragmentation or consumption by the buffer cache. Preallocate
3594 some large reply buffers so that these replies can be sent even
3595 under memory pressure.
3599 Frequency : rare (liblustre)
3600 Description: fsx running with liblustre complained that using truncate() to
3601 extend the file doesn't work. This patch corrects that issue.
3602 Details : This is the liblustre equivalent of the fix for bug 6196. Fixes
3603 ATTR_SIZE and lsm use in llu_setattr_raw.
3607 Frequency : rare, only 2.6 kernels
3608 Description: Unusual file access patterns on the MDS may result in inode
3609 data being lost in very rare circumstances.
3610 Details : Bad interaction between the ea-in-inode patch and the "no-read"
3611 code in the 2.6 kernel caused the inode and/or EA data not to
3612 be read from disk, causing single-file corruption.
3616 Frequency : rare, only 2.6 filesystems using extents
3617 Description: Heavy concurrent write and delete load may cause data corruption.
3618 Details : It was possible under high-load situations to have an extent
3619 metadata block in the block device cache from a just-unlinked
3620 file overwrite a newly-allocated data block. We now unmap any
3621 metadata buffers that alias just-allocated data blocks.
3625 Frequency : filesystems with default stripe_count larger than 77
3626 Description: lconf+mke2fs fail when formatting filesystem with > 77 stripes
3627 Details : lconf specifies an inode size of 4096 bytes when the default
3628 stripe_count is larger than 77. This conflicts with the default
3629 inode density of 1 per 4096 bytes. Allocate smaller inodes in
3630 this case to avoid pinning too much memory for large EAs.
3632 ------------------------------------------------------------------------------
3634 2005-07-07 Cluster File Systems, Inc. <info@clusterfs.com>
3639 Frequency : rare (extremely heavy IO load with hundreds of clients)
3641 Description: Client is evicted, gets IO error writing to file
3642 Details : lock ordering changes for bug 5492 reintroduced bug 3267 and
3643 caused clients to be evicted for AST timeouts. The fixes in
3644 bug 5192 mean we no longer need to have such short AST timeouts
3645 so ldlm_timeout has been increased.
3648 Frequency : occasional during --force or --failover shutdown under load
3649 Bugzilla : 5949, 4834
3650 Description: Server oops/LBUG if stopped with --force or --failover under load
3651 Details : a collection of import/export refcount and cleanup ordering
3652 issues fixed for safer force cleanup
3655 Frequency : only filesystems larger than 120 OSTs
3656 Bugzilla : 5990, 6223
3657 Description: lfs getstripe would oops on a very large filesystem
3658 Details : lov_getconfig used kfree on vmalloc'd memory
3661 Frequency : only filesystems exporting via NFS to Solaris 10 clients
3662 Bugzilla : 6242, 6243
3663 Description: reading from files that had been truncated to a non-zero size
3664 but never opened returned no data
3665 Details : ll_file_read() reads zeros from no-object files to EOF
3670 Description: A bug in MDS/OSS recovery could cause the OSS to fail an assertion
3671 Details : There's little harm in aborting MDS/OSS recovery and letting it
3672 try again, so I removed the LASSERT and return an error instead.
3674 Severity : enhancement
3676 Description: New debugging infrastructure for tracking down data corruption
3677 Details : The I/O checksum code was replaced to: (a) control it at runtime,
3678 (b) cover more of the client-side code path, and (c) try to narrow
3679 down where problems occurred
3683 Bugzilla : 3819, 4364, 4397, 6313
3684 Description: Racing close and eviction MDS could cause assertion in mds_close
3685 Details : It was possible to get multiple mfd references during close and
3686 client eviction, leading to one thread referencing a freed mfd.
3688 Severity: : enhancement
3689 Bugzilla : 3262, 6359
3690 Description: Attempts to reconnect to servers are now more aggressive.
3691 Details : This builds on the enhanced upcall-less recovery that was added
3692 in 1.4.2. When trying to reconnect to servers, clients will
3693 now try each server in the failover group every 10 seconds. By
3694 default, clients would previously try one server every 25 seconds.
3699 Description: After recovery, certain operations trigger a failed
3700 assertion on a client.
3701 Details : Failing over an mds, using lconf -d --failover, while a
3702 client was doing a readdir() call would cause the client to
3703 LBUG after recovery completed and the readdir() was resent.
3705 Severity : enhancement
3707 Description: Default groups are now added by lconf
3708 Details : You can now run lconf --group <servicename> without having to
3709 manually add groups with lmc.
3712 Frequency : occasional
3714 Description: Nodes with an elan id of 0 trigger a failed assertion
3717 Frequency : always when accessing e.g. tty/console device nodes
3719 Description: tty and some other devices nodes cannot be used on lustre
3720 Details : file's private_data field is used by device data and lustre
3721 values in there got lost. New field was added to struct file to
3722 store fs-specific private data.
3725 Frequency : when exporting Lustre via NFS
3727 Description: NFSD failed occasionally when looking up a path component
3728 Details : NFSD is looking up ".." which was broken in ext3 directories
3729 that had grown large enough to become hashed.
3732 Frequency : Clusters with multiple interfaces not on the same subnet
3734 Description: Nodes will repeatedly try to reconnect to an interface which it
3735 cannot reach and report an error to the log.
3736 Details : Extra peer list entries will be created by lconf with some peers
3737 unreachable. lconf now validates the peer before adding it.
3740 Frequency : Only if a default stripe is set on the filesystem root.
3742 Description: Setting a default stripe on the filesystem root prevented the
3743 filesystem from being remounted.
3744 Details : The client was sending extra request flags in the root getattr
3745 request and did not allocate a reply buffer for the dir EA.
3748 Frequency : occasional, higher if lots of files are accessed by one client
3749 Bugzilla : 6159, 6097
3750 Description: Client trips assertion regarding lsm mismatch/magic
3751 Details : While revalidating inodes the VFS looks up inodes with ifind()
3752 and in rare cases can find an inode that is being freed.
3753 The ll_test_inode() code will free the lsm during ifind()
3754 when it finds an existing inode and then the VFS later attaches
3755 this free lsm to a new inode.
3759 Bugzilla : 6422, 7030
3760 Description: MDS deadlock between mkdir and client eviction
3761 Details : Creating a new file via mkdir or mknod (starting a transaction
3762 and getting the ns lock) can deadlock with client eviction
3763 (gets ns lock and trying to finish a synchronous transaction).
3766 Frequency : occasional
3767 Description: While starting a server, the fsfilt_ext3 module could not be
3769 Details : CFS's improved ext3 filesystem is named ldiskfs for 2.6
3770 kernels. Previously, lconf would still use the ext3 name
3771 when trying to load modules. Now, it will correctly use
3772 ext3 on 2.4 and ldiskfs on 2.6.
3774 Severity : enhancement
3775 Description: The default stripe count has been changed to 1
3776 Details : The interpretation of the default stripe count (0, to lfs
3777 or lmc) has been changed to mean striping across a single
3778 OST, rather than all available. For general usage we have
3779 found a stripe count of 1 or 2 works best.
3781 Severity : enhancement
3782 Description: Add support for compiling against Cray portals.
3783 Details : Conditional compiling for some areas that are different
3787 Frequency : occasional
3788 Bugzilla : 6409, 6834
3789 Description: Creating files with an explicit stripe count may lead to
3790 a failed assertion on the MDS
3791 Details : If some OSTs are full or unavailable, creating files may
3792 trigger a failed assertion on the MDS. Now, Lustre will
3793 try to use other servers or return an error to the
3797 Frequency : occasional
3799 Description: Multiple concurrent overlapping read+write on multiple SMP nodes
3800 caused lock timeout during readahead (since 1.4.2).
3801 Details : Processes doing readahead might match a lock that hasn't been
3802 granted yet if there are overlapping and conflicting lock
3803 requests. The readahead process waits on ungranted lock
3804 (original lock is CBPENDING), while OST waits for that process
3805 to cancel CBPENDING read lock and eventually evicts client.
3807 Severity : enhancement
3809 Description: Initial enabling of flock support for clients
3810 Details : Implements fcntl advisory locking and file status functions.
3811 This feature is provided as an optional mount flag (default
3812 off), and is NOT CURRENTLY SUPPORTED. Not all types of record
3813 locking are implemented yet, and those that are are not guaranteed
3814 to be completely correct in production environments.
3815 mount -t lustre -o [flock|noflock] ...
3818 Frequency : occasional
3820 Description: OSTs running 2.4 kernels but with extents enabled might trip an
3821 assertion in the ext3 JBD (journaling) layer.
3822 Details : The b_committed_data struct is protected by the big kernel lock
3823 in 2.4 kernels, serializing journal_commit_transaction() and
3824 ext3_get_block_handle->ext3_new_block->find_next_usable_block()
3825 access to this struct. In 2.6 kernels there is finer grained
3826 locking to improve SMP performance of the JBD layer.
3830 Description: Changes the "SCSI I/O Stats" kernel patch to default to "enabled"
3835 Description: merge and cleanup kernel patches.
3838 -----------------------------------------------------------------------------
3840 2005-05-05 Cluster File Systems, Inc. <info@clusterfs.com>
3842 NOTE: Lustre 1.4.2 uses an incompatible network protocol than previous
3843 versions of Lustre. Please update all servers and clients to
3844 version 1.4.2 or later at the same time. You must also run
3845 "lconf --write-conf {config}.xml" on the MDS while it is stopped
3846 to update the configuration logs.
3848 - fix for HPUX NFS client breakage when NFS exporting Lustre (5781)
3849 - mdc_enqueue does not need max_mds_easize request buffer on send (5707)
3850 - swab llog records of type '0' so we get proper header size/idx (5861)
3851 - send llog cancel req to DLM cancel portal instead of cb portal (5515)
3852 - fix rename of one directory over another leaking an inode (5953)
3853 - avoid SetPageDirty on 2.6 (5981)
3854 - don't re-add just-being-destroyed locks to the waiting list (5653)
3855 - when creating new directories, inherit the parent's custom
3856 striping settings if present parent (3048)
3857 - flush buffers from cache before direct IO in 2.6 obdfilter (4982)
3858 - don't hold i_size_sem in ll_nopage() and ll_ap_refresh_count (6077)
3859 - don't hold client locks on temporary worklist from l_lru (5666)
3860 - handle IO errors in 2.6 obdfilter bio completion routine (6046)
3861 - automatically evict dead clients (5921)
3862 - Update file size properly in create+truncate+fstat case (6196)
3863 - Do not unhash mountpoint dentries, do not allow removal of
3865 - Avoid lock ordering deadlock issue with write/truncate (6203,5654)
3866 - reserve enough journal credits in fsfilt_start_log for setattr (4554)
3867 - ldlm_enqueue freed-export error path would always LBUG (6149,6184)
3868 - don't reference lr_lvb_data until after we hold lr_lvb_sem (6170)
3869 - don't overwrite last_rcvd if there is a *_client_add() error (6086)
3870 - Correctly handle reads of files with no objects (6243)
3871 - lctl recover will also mark a device active if deactivate used (5933)
3873 - by default create 1 inode per 4kB space on MDS, per 16kB on OSTs
3874 - allow --write-conf on an MDS with different nettype than client (5619)
3875 - don't write config llogs to MDS for mounts not from that MDS (5617)
3876 - lconf should create multiple TCP connections from a client (5201)
3877 - init scripts are now turned off by default; run chkconfig --on
3878 lustre and chkconfig --on lustrefs to use them
3879 - upcalls are no longer needed for clients to recover to failover
3881 - add --abort-recovery option to lconf to abort recovery on device
3883 - add support for an arbitrary number of OSTs (3026)
3884 - Quota support protocol changes.
3885 - forward compatibility changes to wire structs (6007)
3886 - rmmod NALs that might be loaded because of /etc/modules.conf (6133)
3887 - support for mountfsoptions and clientoptions to the Lustre LDAP (5873)
3888 - improved "lustre status" script
3889 - initialize blocksize for non-regular files (6062)
3890 - added --disable-server and --disable-client configure options (5782)
3891 - introduce a lookup cache for lconf to avoid repeated DB scans (6204)
3892 - Vanilla 2.4.29 support
3893 - increase maximum number of obd devices to 520 (6242)
3894 - remove the tcp-zero-copy patch from the suse-2.4 series (5902)
3895 - Quadrics Elan drivers are now included for the RHEL 3 2.4.21 and
3896 SLES 9 2.6.5 kernels
3897 - limit stripes per file to 160 (the maximum EA size) (6093)
3899 2005-03-22 Cluster File Systems, Inc. <info@clusterfs.com>
3902 - don't LASSERT in ll_release on NULL lld with NFS export (4655, 5760)
3903 - hold NS lock when calling handle_ast_error->del_waiting_lock (5746)
3904 - fix setattr mtime regression from lovcleanup merge (4829, 5669)
3905 - workaround for 2.6 crash in ll_unhash_aliases (5687, 5210)
3906 - small ext3 extents cleanups and fixes (5733)
3907 - improved mballoc code, several small races and bugs fixed (5733, 5638)
3908 - kernel version 43 - fix remove_suid bugs in both 2.4 and 2.6 (5695)
3909 - avoid needless client->OST connect, fix handle mismatch (5317)
3910 - fix DLM error path that led to out-of-sync client, long delays (5779)
3911 - support common vfs-enforced mount options (nodev,nosuid,noexec) (5637)
3912 - fix several locking issues related to i_size (5492,5624,5654,5672)
3913 - don't move pending lock onto export if it is already evicted (5683)
3914 - fix kernel oops when creating .foo in unlinked directory (5548)
3915 - fix deadlock in obdfilter statistics vs. object create (5811)
3916 - use time_{before,after} to avoid timer jiffies wrap (5882)
3917 - shutdown --force/--failover stability (3607,3651,4797,5203,4834)
3918 - Do not leak request if server was not able to process it (5154)
3919 - If mds_open unable to find parent dir, make that negative lookup(5154)
3920 - don't create new directories with extent-mapping (5909, 5936)
3922 - fix lustre/lustrefs init scripts for SuSE (patch from Scali, 5702)
3923 - don't hold the pinger_sem in ptlrpc_pinger_sending_on_import
3924 - change obd_increase_kms to obd_adjust_kms (up or down) (5654)
3925 - lconf, lmc search both /usr/lib and /usr/lib64 for Python libs (5800)
3926 - support for RHEL4 kernel on i686 (5773)
3927 - provide error messages when incompatible logs are encountered (5898)
3929 2005-02-18 Cluster File Systems, Inc. <info@clusterfs.com>
3930 * version 1.4.0.10 (1.4.1 release candidate 1)
3932 - don't keep a lock reference when lock is not granted (4238)
3933 - unsafe list practices (rarely) led to infinite eviction loop (4908)
3934 - add per-fs limit of Lustre pages in page cache, avoid OOM (4699)
3935 - drop import inflight refcount on signal_completed_replay error (5255)
3936 - unlock page after async write error during send (3677)
3937 - handle missing objects in filter_preprw_read properly (5265)
3938 - no transno return for symlink open, don't save no-trasno open (3440)
3939 - don't try to complete elan receive that already failed (4012)
3940 - free RPC server reply state on error (5406)
3941 - clean up thread from ptlrpc_start_thread() on error (5160)
3942 - readahead could read extra page into cache that wasn't ejected (5388)
3943 - prevent races in class_attach/setup/cleanup/detach (5260)
3944 - don't dereference de->d_inode after l_dput of de (5458)
3945 - use "int" for stripe value returned from lock_to_stripe (5544)
3946 - mballoc allocation and error-checking fixes in 2.6 (5504)
3947 - block device patches to fix I/O request sizes in 2.6 (5482)
3948 - look up hostnames for IB nals (5602)
3949 - 2.6 changed lock ordering of 2 semaphores, caused deadlock (5654)
3950 - don't start multiple acceptors for the same port (5277)
3951 - fix incorrect LASSERT in mds_getattr_name (5635)
3952 - export a proc file for general "ping" checking (5628)
3953 - fix "lfs check" to not block when the MDS is down (5628)
3955 - service request history (4965)
3956 - put {ll,lov,osc}_async_page structs in a single slab (4699)
3957 - create an "evict_client" /proc entry on OSTs, like the MDS has
3958 - fix mount usage message, return errors per mount(8) (5168)
3959 - change grep [] to grep "[]" in tests so they work in more UMLs
3960 - fix ppc64/x86_64 spec to use %{_libdir} instead of /usr/lib (5389)
3961 - remove ancient LOV_MAGIC_V0 EA support (5047)
3962 - add "disk I/Os in flight" and "I/O req time" stats in obdfilter
3963 - align r/w RPCs to PTLRPC_MAX_BRW_SIZE boundary for performance (3451)
3964 - allow readahead allocations to fail when low on memory (5383)
3965 - mmap locking landed again, after considerable improvement (2828)
3966 - add get_hostaddr() to lustreDB.py for LDAP support (5459)
3968 2004-11-23 Cluster File Systems, Inc. <info@clusterfs.com>
3971 - send OST transaction number in read/write reply to free req (4966)
3972 - don't ASSERT in ptl_send_rpc() if we run out of memory (5119)
3973 - lock /proc/sys/portals/routes internal state, avoiding oops (4827)
3974 - the watchdog thread now runs as interruptible (5246)
3975 - flock/lockf fixes (but it's still disabled, pending 5135)
3976 - don't use EXT3 constants in llite code (5094)
3977 - memory shortage at startup could cause assertion (5176)
3979 - reorganization of lov code
3980 - single portals codebase
3982 - add extents/mballoc support (5025)
3983 - direct I/O reads in the obdfilter (4048)
3984 - kernel patches from LNXI for 2.6 (bluesmoke, perfctr, mtd, kexec)
3986 tbd Cluster File Systems, Inc. <info@clusterfs.com>
3989 - send OST transaction number in read/write reply to free req (4966)
3990 - don't ASSERT in ptl_send_rpc() if we run out of memory (5119)
3991 - lock /proc/sys/portals/routes internal state, avoiding oops (4827)
3992 - the watchdog thread now runs as interruptible (5246)
3993 - handle missing objects in filter_preprw_read properly (5265)
3994 - unsafe list practices (rarely) led to infinite eviction loop (4908)
3995 - drop import inflight refcount on signal_completed_replay error (5255)
3996 - unlock page after async write error during send (3677)
3997 - return original error code on reconstructed replies (3761)
3998 - no transno return for symlink open, don't save no-trasno open (3440)
4000 - add pid to ldlm debugging output (4922)
4001 - bump the watchdog timeouts -- we can't handle 30sec yet
4002 - extra debugging for orphan dentry/inode bug (5259)
4004 2004-11-16 Cluster File Systems, Inc. <info@clusterfs.com>
4007 - fix TCP_NODELAY bug, which caused extreme perf regression (5134)
4008 - allocate qswnal tx descriptors singly to avoid fragmentation (4504)
4009 - don't LBUG on obdo_alloc() failure, use OBD_SLAB_ALLOC() (4800)
4010 - fix NULL dereference in /proc/sys/portals/routes (4827)
4011 - allow failed mdc_close() operations to be interrupted (4561)
4012 - stop precreate on OST before MDS would time out on it (4778)
4013 - don't send partial-page writes before EOF from client (4410)
4014 - discard client grant for sub-page writes on large-page clients (4520)
4015 - don't free dentries not owned by NFS code, check generation (4806)
4016 - fix lsm leak if mds_create_objects() fails (4801)
4017 - limit debug_daemon file size, always print CERROR messages (4789)
4018 - use transno after validating reply (3892)
4019 - process timed out requests if import state changes (3754)
4020 - update mtime on OST during writes, return in glimpse (4829)
4021 - add mkfsoptions to LDAP (4679)
4022 - use ->max_readahead method instead of zapping global ra (5039)
4023 - don't interrupt __l_wait_event() during strace
4025 - add software watchdogs to catch hung threads quickly (4941)
4026 - make lustrefs init script start after nfs is mounted
4027 - fix CWARN/ERROR duplication (4930)
4028 - return async write errors to application if possible (2248)
4029 - add /proc/sys/portal/memused (bytes allocated by PORTALS_ALLOC)
4030 - print NAL number in %x format (4645)
4031 - update barely-supported suse-2.4.21-171 series (4842)
4032 - support for sles 9 %post scripts
4033 - support for building 2.6 kernel-source packages
4034 - support for sles km_* packages
4036 2004-10-07 Cluster File Systems, Inc. <info@clusterfs.com>
4039 - ignore -ENOENT errors in osc_destroy (3639)
4040 - notify osc create thread that OSC is being cleaned up (4600)
4041 - add nettype argument for llmount in #5d in conf-sanity.sh (3936)
4042 - reconstruct ost_handle() like mds_handle() (4657)
4043 - create a new thread to do import eviction to avoid deadlock (3969)
4044 - let lconf resolve symlinked-to devices (4629)
4045 - don't unlink "objects" from directory with default EA (4554)
4046 - hold socknal file ref over connect in case target is down (4394)
4047 - allow more than 32000 subdirectories in a single directory (3244)
4048 - fix blocks count for O_DIRECT writes (3751)
4049 - OST returns ENOSPC from object create when no space left (4539)
4050 - don't send truncate RPC if file size isn't changing (4410)
4051 - limit OSC precreate to 1/2 of value OST considers bogus (4778)
4052 - bind to privileged port in socknal and tcpnal (3689)
4054 - rate limit CERROR/CWARN console message to avoid overload (4519)
4055 - GETFILEINFO dir ioctl returns LOV EA + MDS stat in 1 call (3327)
4056 - basic mmap support (3918)
4057 - kernel patch series update from b1_4 (4711)
4059 2004-09-16 Cluster File Systems, Inc. <info@clusterfs.com>
4062 - avoid crash during MDS cleanup with OST shut down (2775)
4063 - fix loi_list_lock/oig_lock inversion on interrupted IO (4136)
4064 - don't use bad inodes on the MDS (3744)
4065 - dynamic object preallocation to improve recovery speed (4236)
4066 - don't hold spinlock over lock dumping or change debug flags (4401)
4067 - don't zero obd_dev when it is force cleaned (3651)
4068 - print grants to console if they go negative (4431)
4069 - "lctl deactivate" will stop automatic recovery attempts (3406)
4070 - look for existing locks in ldlm_handle_enqueue() (3764)
4071 - don't resolve lock handle twice in recovery avoiding race (4401)
4072 - revalidate should check working dir is a directory (4134)
4074 - don't always mark "slow" obdfilter messages as errors (4418)
4076 2004-08-24 Cluster File Systems, Inc. <info@clusterfs.com>
4079 - don't close LustreDB during write_conf until it is done (3860)
4080 - fix typo in lconf for_each_profile (3821)
4081 - allow dumping logs from multiple threads at one time (3820)
4082 - don't allow multiple threads in OSC recovery (3812)
4083 - fix debug_size parameters (3864)
4084 - fix mds_postrecov to initialize import for llog ctxt (3121)
4085 - replace config semaphore with spinlock (3306)
4086 - be sure to send a reply for a CANCEL rpc with bad export (3863)
4087 - don't allow enqueue to complete on a destroyed export (3822)
4088 - down write_lock before checking llog header bitmap (3825)
4089 - recover from lock replay timeout (3764)
4090 - up llog sem before sending rpc (3652)
4091 - reduce ns lock hold times when setting kms (3267)
4092 - change a dlm LBUG to LASSERTF, to maybe learn something (4228)
4093 - fix NULL deref and obd_dev leak on setup error (3312)
4094 - replace some LBUG about llog ops with error handling (3841)
4095 - don't match INVALID dentries from d_lookup and spin (3784)
4096 - hold dcache_lock while marking dentries INVALID and hashing (4255)
4097 - fix invalid assertion in ptlrpc_set_wait (3880)
4099 - add libwrap support for the TCP acceptor (3996)
4100 - add /proc/sys/portals/routes for non-root route listing (3994)
4101 - allow setting MDS UUID in .xml (2580)
4102 - print the stack of a process that LBUGs (4228)
4104 2004-07-14 Cluster File Systems, Inc. <info@clusterfs.com>
4107 - don't cleanup request in ll_file_open() on failed MDS open (3430)
4108 - make sure to unset replay flag from failed open requests (3440)
4109 - if default stripe count is 0, use OST count for inode size (3636)
4110 - update parent mtime/ctime on client for create/unlink (2611)
4111 - drop dentry ref in ext3_add_link from open_connect_dentry (3266)
4112 - free recovery state on server during a forced cleanup (3571)
4113 - unregister_reply for resent reqs (3063)
4114 - loop back devices mounting and status check on 2.6 (3563)
4115 - fix resource-creation race that can provoke i_size == 0 (3513)
4116 - don't try to use bad inodes returned from MDS/OST fs lookup (3688)
4117 - more debugging for page-accounting assertion (3746)
4118 - return -ENOENT instead of asserting if ost getattr+unlink race (3558)
4119 - avoid deadlock after precreation failure (3758)
4120 - fix race and lock order deadlock in orphan handling (3450, 3750)
4121 - add validity checks when grabbing inodes from l_ast_data (3599)
4123 - add /proc/.../recovery_status to obdfilter (3428)
4124 - lightweight CDEBUG infrastructure, debug daemon (3668)
4125 - change default OSC RPC parameters to be better on small clusters
4126 - turn off OST read cache for files smaller than 32MB
4127 - install man pages and include them in rpms (3100)
4128 - add new init script for (un)mounting lustre filesystems (2593)
4129 - run chkconfig in %post for init scripts (3701)
4130 - drop scimac NAL (unmaintained)
4132 2004-06-17 Cluster File Systems, Inc. <info@clusterfs.com>
4135 - clean kiobufs before and after use (3485)
4136 - strip trailing '/'s before comparing paths with /proc/mounts (3486)
4137 - remove assertions to work around "in-flight rpcs" recovery bug (3063)
4138 - change init script to fail more clearly if not run as root (1528)
4139 - allow clients to reconnect during replay (1742)
4140 - fix ns_lock/i_sem lock ordering deadlock for kms update (3477)
4141 - don't do DNS lookups on NIDs too small for IP addresses (3442)
4142 - re-awaken ptlrpcd if new requests arrive during check_set (3554)
4143 - fix cond_resched (3554)
4144 - only evict unfinished clients after recovery (3515)
4145 - allow bulk resend, prevent data loss (3570)
4146 - dynamic ptlrpc request buffer allocation (2102)
4147 - don't allow unlinking open directory if it isn't empty (2904)
4148 - set MDS/OST threads to umask 0 to not clobber client modes (3359)
4149 - remove extraneous obd dereference causing LASSERT failure (3334)
4150 - don't use get_cycles() when creating temp. files on the mds (3156)
4151 - hold i_sem when setting i_size in ll_extent_lock() (3564)
4152 - handle EEXIST for set-stripe, set proper directory name (3336)
4154 - servers can dump a log evicting a client - lustre.dump_on_timeout=1
4155 - fix ksocknal_fmb_callback() error messages (2918)
4157 2004-05-27 Cluster File Systems, Inc. <info@clusterfs.com>
4160 - don't copy lvb into (possibly NULL) reply on error (2983)
4161 - don't deref dentry after dput, don't free lvb on error (2922)
4162 - use the kms to determine writeback rpc length (2947)
4163 - increment oti_logcookies when osc is inactive (2948)
4164 - update client's i_blocks count via lvb messages (2543)
4165 - handle intent open/close of special files properly (1557)
4166 - mount MDS with errors=remount-ro, like obdfilter (2009)
4167 - initialize lock handle to avoid ASSERT on error cleanup (3057)
4168 - don't use cancelling-locks' kms values (2947)
4169 - use highest lock extent for kms, not last one (2925)
4170 - don't dereference ERR_PTR() dentry in error handling path (3107)
4171 - fix thread race in portals_debug_dumplog() (3122)
4172 - create lprocfs device entries at setup instead of at attach (1519)
4173 - common AST error handler, don't evict client on completion race (3145)
4174 - zero nameidata in detach_mnt in 2.6 (3118)
4175 - verify d_inode after revalidate_special is valid in 2.6 (3116)
4176 - use lustre_put_super() to handle zconf unmounts in 2.6 (3064)
4177 - initialize RPC timeout timer earlier for 2.6 (3219)
4178 - don't dereference NULL reply buffer if mdc_close was never sent (2410)
4179 - print nal/nid for unknown nid (3258)
4180 - additional checks for oscc recovery before doing precreate (3284)
4181 - fix ll_extent_lock() error return code for 64-bit systems (3043)
4182 - don't crash in mdc_close for bad permissions on open (3285)
4183 - zero i_rdev for non-device files (3147)
4184 - clear page->private before handing to FS, better assertion (3119)
4185 - tune the read pipeline (3236)
4186 - fix incorrect decref of invalidated dentry (2350)
4187 - provide read-ahead stats and refine rpc in flight stats (3328)
4188 - don't hold journal transaction open across create RPC (3313)
4189 - update atime on MDS at close time (3265)
4190 - close LDAP connection when recovering to avoid server load (3315)
4191 - update iopen-2.6 patch with fixes from 2399,2517,2904 (3301)
4192 - don't leak open file on MDS after open resend (3325)
4193 - serialize filter_precreate and filter_destroy_precreated (3329)
4194 - loop device shouldn't call sync_dev() for nul device (3092)
4195 - clear page cache after eviction (2766)
4196 - resynchronize MDS->OST in background (2824)
4197 - refuse to mount the same filesystem twice on same mountpoint (3394)
4198 - allow llmount to create routes for mounting behind routers (3320)
4199 - push lock cancellation to blocking thread for glimpse ASTs (3409)
4200 - don't call osc_set_data_with_check() for TEST_LOCK matches (3159)
4201 - fix rare problem with rename on htree directories (3417)
4203 - allow default OST striping configuration per directory (1414)
4204 - fix compilation for qswnal for 2.6 kernels (3125)
4205 - increase maximum number of MDS request buffers for large systems
4206 - change liblustreapi to be useful for external progs like lfsck (3098)
4207 - increase local configuration timeout for slow disks (3353)
4208 - allow configuring ldlm AST timeout - lustre.ldlm_timeout=<seconds>
4210 2004-03-22 Cluster File Systems, Inc. <info@clusterfs.com>
4213 - fixes for glimpse AST timeouts / incorrectly 0-sized files (2818)
4214 - don't overwrite extent policy data in reply if lock was blocked (2901)
4215 - drop filter export grants atomically with removal from device (2663)
4216 - del obd_self_export from work_list in class_disconnect_exports (2908)
4217 - don't LBUG if MDS recovery times out during orphan cleanup (2530)
4218 - swab reply message in mdc_close, other PPC fixes (2464)
4219 - fix destroying of named logs (2325)
4220 - overwrite old logs when running lconf --write_conf (2264)
4221 - bump LLOG_CHUNKSIZE to 8k to allow for larger clusters (2306)
4222 - fix race in target_handle_connect (2898)
4223 - mds_reint_create() should take same inode create lock (2926)
4224 - correct journal credits calculated for CANCEL_UNLINK_LOG (2931)
4225 - don't close files for self_export to avoid uninitialized obd (2936)
4226 - allow MDS with the same name as client node (2939)
4227 - hold dentry reference for closed log files for unlink (2325)
4228 - reserve space for all logs during transactions (2059)
4229 - don't evict page beyond end of stripe extent (2925)
4230 - don't oops on a deleted current working directory (2399)
4231 - handle hard links to targets without a parent properly (2517)
4232 - don't dereference NULL lock when racing during eviction (2867)
4233 - don't grow lock extents when lots of conflicting locks (2919)
4235 2004-03-04 Cluster File Systems, Inc. <info@clusterfs.com>
4238 - account for cache space usage on clients to avoid data loss (974)
4239 - lfsck support in lustre kernel code (2349)
4240 - reduce journal credits needed for BRW writes (2370)
4241 - orphan handling to avoid losing space on client/server crashes
4242 - ptlrpcd can be blocked, stopping ALL progress (2477)
4243 - use lock value blocks to assist in proper KMS, faster stat (1021)
4244 - takes i_sem instead of DLM locks internally on obdfilter (2720)
4245 - recovery for initial connections (2355)
4246 - fixes for mds_cleanup_orphans (1934)
4247 - abort_recovery crashes MDS in b_eq (mds_unlink_orphan) (2584)
4248 - block all file creations until orphan recovery completes (1901)
4249 - client remove rq_connection from request struct (2423)
4250 - conf-sanity test_5, proper cleanup in umount log not availale (2640)
4251 - recovery timer race (2670)
4252 - mdc_close recovey bug (2532)
4253 - ptlrpc cleanup bug (2710)
4254 - mds timeout on local locks (2588)
4255 - namespace lock held during RPCs (2431)
4256 - handle interrupted sync write properly (2503)
4257 - don't try to handle a message that hasn't been replied to (2699)
4258 - client assert failure during cleanup after abort recovery (2701)
4259 - leak mdc device after failed mount (2712)
4260 - ptlrpc_check_set allows timedout requests to complete (2714)
4261 - wait for inflight reqs when ptlrpcd finishes (2710)
4262 - make sure unregistered services are removed from the srv_list
4263 - reset bulk XID's when resending them (caught by 1138 test)
4264 - unregister_bulk after timeout
4265 - fix lconf error (2694)
4266 - handle write after unfinished setstripe, stripe-only getstripe (2388)
4267 - readahead locks pages, leaves pending causing memory pressure (2673)
4268 - increase OST request buffers to 4096 on large machines (2729)
4269 - fix up permission of existing directories in simple_mkdir (2661)
4270 - init deleted item, add assertions ptlrpc_abort_inflight() (2725)
4271 - don't assign transno to errored transactions (2742)
4272 - don't delete objects on OST if given a bogus objid from MDS (2751)
4273 - handle large client PAGE_SIZE readdir on small PAGE_SIZE MDS (2777)
4274 - if rq_no_resend, then timeout request after recovery (2432)
4275 - fix MDS llog_logid record size, 64-bit array alignment (2733)
4276 - don't call usermode_helper from ptlrpcd, DEFAULT upcall (2773)
4277 - put magic in mount.lustre data, check for bad/NULL mount data (2529)
4278 - MDS recovery shouldn't delete objects that it has given out (2730)
4279 - if enqueue arrives after completion, don't clobber LVB (2819)
4280 - don't unlock pages twice when trigger_group_io returns error (2814)
4281 - don't deref NULL rq_repmsg if ldlm_handle_enqueue failed (2822)
4282 - don't write pages to disk if there was an error (1450)
4283 - don't ping imports that have recovery disabled (2676)
4284 - take buffered bytes into account when balancing socknal conn (2817)
4285 - hold a DLM lock over readdir always, use truncate_inode_pages (2706)
4286 - reconnect unlink llog connection after MDS reconnects to OST (2816)
4287 - remove little-endian swabbing of llog records (1987)
4288 - set/limit i_blksize to LL_MAX_BLKSIZE on client (2884)
4289 - retry reposting request buffers if they fail (1191)
4290 - grow extent at grant time to avoid granting a revoked lock (2809)
4291 - lock revoke doesn't evict page if covered by a second lock (2765)
4292 - disable VM readahead to avoid reading outside lock extents (2805)
4294 - return LL_SUPER_MAGIC from statfs for the filesystem type (1972)
4295 - updated kernel patches for hp-2.4.20 kernel (2681)
4297 2004-02-07 Cluster File Systems, Inc. <info@clusterfs.com>
4300 - fix truncated write corruption (2366)
4301 - fix for failed assertion in iopen_connect_dentry (1792,2517)
4303 - don't flag the ptlrpcd thread with PF_MEMALLOC (2636)
4304 - ensure len(uuid) < 37 in lmc (1171)
4305 - fix ia64 OOPS in llog_test (2255)
4306 - zero end of page at obdfilter for partial page writes (2648)
4307 - don't leave stale dentries around after renames (bug 2428)
4308 - fix timeouts when evicting a client with a single lock held (2642)
4309 - set deadline for the initial HELLO message to drain (2634)
4310 - print out dotted-quad IP addresses in the socknal (2302)
4312 - additional debugging for MDS client eviction problem (2443)
4313 - fix mkfsoptions support for osts (2603, 2604)
4315 2004-01-27 Cluster File Systems, Inc. <info@clusterfs.com>
4318 - add series for the vanilla 2.6.0 kernel
4319 - add series for the vanilla 2.4.24 kernel
4320 - add series for a cray x86/64 UL kernel drop
4321 - fix xattr patches for the vanilla 2.4.19 series
4323 - generate true UUIDs in lmc (1171)
4324 - have portals stack dumping break in UML (2466)
4325 - avoid bad dchild deref; avoid inum lock w/o creation (2362)
4326 - allocate with _NOFS in ldlm to avoid deadlock (1933)
4327 - wake callback waiting threads on client eviction (2460)
4328 - Add --ptldebug and --subsystem to lmc (1719)
4329 - update assertion to allow safe interrupt allocation
4330 - set rq_no_resend for cancel requests (2432)
4331 - recalculate ptlrpcd timeout after resend (2494)
4332 - call vfs_rmdir when removing pending directories (2368)
4333 - fix renaming a file to itself (2429)
4334 - lmc creates a default one-stripe lov (2454)
4335 - expand procfs space to handle large clusters (2326)
4336 - increase UML stack to avoid overflow
4337 - update lconf's list of debug and subsystem masks
4338 - fix lfs find --obd (2510)
4339 - /proc tunable for disabling filter read caching (2591)
4340 - stop rpm packages from altering slapd.conf (2301)
4341 - disable nagle in the socknal under 0conf (2578)
4342 - choose mds inode size based on stripe count (2572)
4343 - fix kernel-source rpm problems (2516)
4345 - add --disable-doc to avoid pdf generation (2421)
4346 - update documentation, tests, type-os, comments
4347 - avoid format warnings on ia64
4348 - remove the TOE NAL
4349 - tiny code cleanups by removing unused fields
4351 2004-01-07 Cluster File Systems, Inc. <info@clusterfs.com>
4354 - fix obvious semaphore misuse in as-yet-unused setattr path (2348)
4355 - remove the most blatant lies from BUILDING file (2371)
4356 - change default debug level to reasonable production setting
4357 - reduce client side cache size to reduce cache flush time
4358 - reduce max RPCs in flight to avoid unnecessary file fragmentation
4359 - make TCP zerocopy and pinger support enabled by default (2476)
4360 - sync writes completed after process exits caused crashes (2319)
4361 - maintain correct mount count on the MDS (2356)
4362 - backout 1557, because 2316 wasn't really fixed
4363 - better file I/O statistics gathering in /proc
4364 - don't take unnecessary, deadlock-inducing bug in readpage (2383)
4365 - another kernel patch to fix zero-copy TCP function export
4366 - don't take duplicate lock when processing re-sent getattr (2420)
4367 - lctl uses obd_self_export instead of creating new conn (2353)
4368 - MDS/OST recovery case which requires object creation asserted (2425)
4369 - move lfs from /usr/sbin to /usr/bin in packages
4370 - fix race between mds_client_add and mds_client_free (2417)
4371 - use kmalloc instead of slabs in portals (2430)
4372 - don't create duplicate records when a failover MDS is present (2442)
4373 - remove unnecessary mount age check (2332)
4374 - don't remove directory inodes from locks prematurely (2451)
4375 - don't break if MDS service name is the same as hostname (2103)
4376 - fix races in client write RPC generation when cache full (2482)
4378 2003-12-13 Cluster File Systems, Inc. <info@clusterfs.com>
4381 - remove now-unused request->rq_obd (278)
4382 - if an allocation fails, print out how much memory we've used (1933)
4383 - use PORTAL_SLAB_ALLOC for structures, to get GFP_MEMALLOC (1933)
4384 - add the "configurable stack size" patch to most series files (1256)
4385 - ability to write large log records, for 100+ OST configs (2306)
4386 - fix NULL deref when filter_prep fails (2314)
4387 - fix operator precedence error in filter_sync
4388 - dynamic allocation of socknal TX descriptors (2315)
4389 - fix a missed case in the GFP_MEMALLOC patch, can cause deadlock (2310)
4390 - fix gcc 2.96 compilation problem in xattr kernel patch (2294)
4391 - ensure that CWARN messages in Portals always get to the syslog
4392 - __init/__exit are not for prototype decls (ldlm_init/exit)
4393 - x86-64 compile warning fixes
4394 - fix gateway LMC keyword conflict (2318)
4395 - fix MDS lock inversions in getattr/reint paths (1844)
4396 - fix a rare lock re-ordering bug, which caused deadlock (2322)
4397 - fix i_sem/journal inversion in fsfilt_ext3_write_record (2306)
4398 - DLM race condition prevented some lock evictions (2328)
4399 - ENOMEM detection and retry on socknal sends (2230)
4400 - use GFP_NOFS throughout Lustre, to combat ENOMEM (2230)
4401 - move osc_rpcd into ptlrpc, for use in MDC and others (2329)
4402 - protect MDS inode fsdata with stronger locking; fixes assertion (2313)
4403 - better error messages when a client is rejected during recovery (1505)
4404 - avoid cancelling locks which were never granted, after failure (2330)
4405 - fix i_sem/journal inversion in mds_client_add (2333)
4406 - fix truncate/getattr lock cycle deadlock (2334)
4407 - use rpcd to send close; allows resend after timeout, avoid leak (1897)
4408 - fix two rare exit paths which could leak an l_lock() ref (2321)
4409 - fencepost error in MDS/OST orphan recovery (2226)
4410 - make log record alignment 8 bytes (1988)
4411 - lstripe now fails when requested offset > ost_count (2237)
4412 - ensure that all kernel series have a complete list.h (1607)
4413 - fix crashes in special-file operations (2316)
4414 - lctl create/brw OID mismatch, caused by obsolete filter loop (2339)
4416 - allow configurable automake binary, for testing new versions
4417 - small update to the lfs documentation
4419 2003-12-03 Cluster File Systems, Inc. <info@clusterfs.com>
4421 * fix negative export reference count in fsfilt_sync (2312)
4423 2003-12-01 Cluster File Systems, Inc. <info@clusterfs.com>
4424 * release candidate 0.9.1
4426 - orphans are moved into the PENDING directory for possible recovery
4427 - replayed opens now open by fid for orphan/rename safety (1042)
4428 - last close of an orphan inode generates a transno (683)
4429 - chdir() and mount() now pin the directory entry (1020)
4430 - avoid CERROR in normal ll_setattr_raw() error case (1500)
4431 - discard very old requests without processing them (1502)
4432 - remove some common, well-understood CERRORs (1505)
4433 - require O_DIRECT I/O to be page-sized to workaround IA64 crash (1609)
4434 - clear "grant" flags in OST replies until OST grant code lands (1644)
4435 - fix read performance by not clobbering i_blksize on client (1598)
4436 - fix __ldlm_handle2lock oops by not dereferencing lock after PUT (1625)
4437 - make LRU size a /proc tunable, clears locks when reduced (707)
4438 - fix some lprocfs rot that prevented ptlbd from loading (1732)
4439 - server locks take references on exports now (1558)
4440 - build fixes for 2.4.20-rh trees (1663)
4441 - return an error from lov_create if all OSCs are inactive (1751)
4442 - fix import levels when a reconnect happens without a timeout (1597)
4443 - exit early from mds_open if we get a lookup error (1749)
4444 - partial page read at EOF wouldn't wait for disk before sending (1642)
4445 - avoid NULL deref in obdfilter when reading page past EOF (1592)
4446 - avoid LASSERT in ll_intent_lock if server failed very early (1090)
4447 - fix LBUG in ll_it_open_error with rc = -2 (1861)
4448 - write/truncate lock inversion (1639)
4449 - Don't auto-load obdclass, portals modules during cleanup (1495)
4450 - fix timestamps from jumping to "now" (1763)
4451 - extra journal assertions (1648)
4452 - add an extra multiunlink test (1771)
4453 - fix read_record/write_record API (1776)
4454 - fix leak of offset_extent, possible incorrect i_size later (1772)
4455 - fix lasserts in mis-matched transnos during open-unlink testing (1541)
4456 - Debugging for the kqswnal_get_idle_tx problems (1820)
4457 - Allow recovery to be attempted multiple times (1536)
4458 - Write out MDS last_rcvd file after it is first created (1600)
4459 - Fix tx_descriptor leak in failed transmit situations (1827)
4460 - ext3 journaling fixes for assertion failure after IO error (1871)
4461 - class_export_put() on freed export after completion AST error (1896)
4462 - Fix revalidate looping in VFS (1322)
4463 - Don't access a freed export during MDS_REINT timeout (1521)
4464 - Add open-unlink recovery support on the MDS (1673,1764)
4465 - Return an error if no MDS data was read from last_rcvd (1946)
4466 - Fix for lookup "." or ".." crash on error (1932,1931,1935)
4467 - Don't setup a disk device that doesn't match exported UUID (317)
4468 - Reduce bulk RPC timeout to avoid cascading client/OST failures (1845)
4469 - avoid committing NULL handle in force close
4470 - local.sh is now a one-stripe LOV configuration
4471 - POSIX utime.4 -EPERM on FIFO not owned by user (56)
4472 - fix ext3 htree duplicate directory entry corruption (1516)
4473 - POSIX creat.13, fstat.1, open.18, stat.3 new file atime/mtime (2020)
4474 - update to new LOV EA format (2097)
4475 - interoperability for different PAGE_SIZE/wordsize (686,1821,1343,2042)
4477 2003-06-15 Phil Schwan <phil@clusterfs.com>
4480 - imports and exports cleanup too early, need refcounts (349, 879, 1045)
4481 - per-import/export recovery handling (958, 931, 959)
4482 - multiple last-rcvd slots, for serving multiple FSes (949)
4483 - connections are again shared between multiple imp/exports (963, 964)
4484 - "umount -f" would hang if any requests needed to be sent (393, 978)
4485 - avoid pinning large req buffer by copying for queued messages (989)
4486 - add "uuid" to "lctl device" command to help upcalls (991)
4487 - "open" RPCs with transnos would confuse recovery counters (1037)
4488 - do proper endian conversion of all wire messages (288, 340, 891)
4489 - remove OST bulk get LBUGs, fix ost_brw_write cleanup (1126)
4490 - call waiting locks callback from LDLM recovery thread (1127, 1151)
4491 - fix ptlrpc_connection leak in target_handle_connect (1174)
4492 - fix import refcounting bug in OST and MDS cleanup (1134)
4493 - if an invalid-at-open-time OSC returned before close(), LBUG (1150)
4494 - fix very unlikely obd_types race condition (501)
4495 - remove osc_open hack for echo_client (1187)
4496 - we leaked exports/dlmimps for forcibly disconnected clients (1143)
4497 - a failure in read_inode2 leads to deadlock (1139)
4498 - cancel ack-locks as soon as transaction is committed (1072)
4499 - fix major leaks and crashes in the bulk I/O path (937, 1057)
4500 - make sure to commitrw after any preprw to avoid deadlock (1162)
4501 - failing to execute a file in a lustre FS would lock inode (1203)
4502 - small DEBUG_REQ fix to avoid dereferencing a NULL (1227)
4503 - don't ASSERT while cleaning up an incompletely-setup obd (1248)
4504 - obd_uuid2tgt would walk off the end of the list (1255)
4505 - on IA64 the osc would give portals incorrect bulk size (1258)
4506 - fix debug daemon ioctl interface; allows daemon on ia64 (1274)
4507 - fix lock inversion caused by new llite matching code (1282)
4508 - limit the number of dirty pages on a client to 10MB (1286)
4509 - timed out locks were not being corrected cancelled (1289)
4510 - fix O_DIRECT above 4GB on IA-32 (1292)
4511 * major user-visible changes
4512 - fail out/fail over policy now controlled by the upcall (993)
4514 - add OBD_PING to check server availability and failure (954)
4515 - lustre messages are now sent in sending host order (288, 340, 891)
4516 - add eadatalen to MDS getattr reply (340)
4517 - OST read replies may contain second buffer, with per-page status (593)
4519 2003-03-11 Phil Schwan <phil@clusterfs.com>
4522 - LDLM_DEBUG macro fix, for gcc 3.2 (850)
4523 - failed open()s could cause deadlock; fixed (867, 869)
4524 - stop cancelling OST locks when files are closed (481)
4525 - overlapping XID spaces caused network corruption (851, 853)
4526 - fix unsafe fsfilt counter arithmetic; change to atomic_t
4527 - setattr_raw added, to do single-RPC, server-side setattrs
4528 - lmc/lconf syntax change for OST UUIDs
4529 - fix crashy race condition between ptlrpc_free_req and osc_close
4530 - don't use request in mdc_enqueue if we hit a timeout (889)
4531 - don't set the inode i_size for regular files from the MDS (896)
4532 - handle out of order completion AST (842)
4533 - don't LBUG if a lock request times out after receiving AST (913)
4534 - avoid d_rehash race in ll_find_alias by rehashing inside dcache_lock
4535 - if a bad lock AST arrives, send an error instead of dropping entirely
4536 - return 0 from revalidate2 if ll_intent_lock returns -EINTR (912)
4537 - fix leak in bulk IO when only partially completed (899, 900, 926)
4538 - fix O_DIRECT for ia64 (55)
4539 - (almost) eliminate Lustre-kernel-thread effects on load average (722)
4540 - C-z after timeout could hang a process forever; fixed (977)
4542 - client-side I/O cache (678, 924, 929, 941, 970)
4544 - READPAGE and SETATTRs which don't take server-side locks get
4547 2003-02-11 Phil Schwan <phil@clusterfs.com>
4550 - Fix ldlm_lock_match on the MDS to avoid matching remote locks (592)
4551 - Fix fsfilt_extN_readpage() to read a full page of directory
4552 entries, or fake the remainder if PAGE_SIZE != blocksize (500)
4553 - Avoid extra mdc_getattr() in ll_intent_lock when possible (534, 604)
4554 - Fix imbalanced LOV object allocation and out-of-bound access (469)
4555 - Most intent operations were removed, in favour of a new RPC mode
4556 that does a single RPC to the server and bypasses most of the VFS
4557 - All LDLM resource ID arrays were removed in favour of ldlm_res_id
4558 - Aggressively cancel local locks on DLM servers
4559 - mds_reint_unlink sends EA to the client if it's the last nlink.
4560 client uses that EA to unlink OST objects.
4561 - mds_reint_{rename,unlink,link} were rewritten to take ordered locks
4562 - recursive symlinks were fixed (439)
4563 - fixed NULL deref in DEBUG_REQ
4564 - filter_update_lastobjid no longer calls sync, which annoyed extN
4565 - fixed multi-client small-writes to a single file problem (445)
4566 - fixed mtime updates during file writes (607)
4567 - fixed vector writes on obdfilter causing problems when ENOSPC (670)
4568 - fixed bug in obd_brw_read/write() (under guise of testing 367)
4569 - fixed Linux OST size reporting problem (444, 656)
4570 - OST now updates object mtime with writes or setattr (607, 619)
4571 - client verifies file size before zeroing page past EOF (445)
4572 - OST now writes last allocated objid to disk with allocation (108)
4573 - LOV on echo now works (409)
4575 - mds_reint_unlink sends a new buffer, with the EA included. this
4576 buffer is only valid if body->valid & OBD_MD_FLEASIZE, which is only
4577 set if a regular file was being unlinked, and it was the last link
4578 - use PtlGet from the target for bulk writes (315)
4579 - OST now updates object mtime with writes or setattr (607, 619)
4580 - LDLM now has a grant-time callback to revalidate locked items, if
4582 - Many MDS operations were reorganized to combat race conditions
4584 - Merge b_intel branch (updated lprocfs code) - now at /proc/fs/lustre
4585 - configure check to avoid gcc version 2.96 20000731-2.96-98) (606)
4587 2003-01-06 Andreas Dilger <adilger@clusterfs.com>
4590 - Fully reactivate OST imports after reconnection (512, others)
4591 - Make sure client sees our -ENOTCONN from mds_handle (513 - partial)
4592 - More graceful error handling for truncating on dead OST (515)
4593 - Don't error out unless we're actually accessing dead stripes (474)
4594 - Fix garbage sizes when stripes are missing (410)
4595 - LRU counters were broken, causing constant lock purge (433, 432)
4596 - garbage on read from stripes with failed OSTs (441)
4597 - mark OSCs as active before reconnecting during recovery (438)
4598 - lov_enqueue and lov_cancel need to handle inactive OSTs (403)
4599 - lfind did not preserve OST order in output (443)
4600 - symlinks cause hung clients, incorrect data (439)
4601 - stop dereferencing request after dropping refcount (457)
4602 - don't LASSERT(spin_is_locked) on non-SMP (455)
4603 - fixes for many rename() bugs
4604 - fstat didn't correctly synchronize attributes (399)
4605 - server must handle lock cancellation during blocking AST prep (487)
4606 - bulk descriptors were free()d too soon (511)
4607 - fix paths in lconf, which would load incorrect modules (451, 507)
4608 - fix confusing lconf 'host not found' error message (386)
4609 - fix lock order deadlock on OST (O/R i_sem before journal ops, 478)
4610 - fix race condition in mdc_blocking_ast() for inode access (526)
4611 - fix lov_unpackmd() unpacking wrong number of stripes (537)
4612 - fix lov_set_osc_active() marking wrong OSC inactive (440)
4613 - fix bad lstripe lov_unpackmd() assertion (fix layering too) (527)
4614 - fix multiple writes of stripe MD to MDS (358, maybe 519)
4615 - fix lstripe in several ways (kernel side) (527)
4616 - fix request leak in ldlm_cli_enqueue (262)
4617 - incorrect OSC was marked inactive after OST failure
4618 - call mds_fs_cleanup before unmounting filesystem (524)
4619 - fix races between taking ns_lock and ldlm_lock_change_resource
4620 - fix races updating LOV export open file list
4621 - fix lov_enqueue error path, avoid decref-ing bad lock handle
4622 - fix recovery NULL deref in ldlm_cli_cancel_unused
4623 - fix some DLM races by using new hash table for lock handles (419)
4624 - permit the client to specify desired inodes, at replay
4625 - duplicate requests when we queue them for replay reintegration
4626 - fix last_rcvd offset calculation
4627 - sync after each recovered transaction, so we always make progress
4628 - never, not always, ERESTART requests without transnos
4629 - store the lov_desc in the MDS, so we don't depend on getlovinfo to
4631 - skip replay if the MDS says that the client is already connected
4632 - don't check for a recovery-enabled export to match lctl's UUID
4633 - don't INC_USE_COUNT for phantom exports
4634 - don't crash when cleaning up phantom exports (567)
4635 - don't double-finish or set replay data for errored mdc_open requests
4636 - abort requests when they time out, so we don't get old replies
4637 - send/receive replies for AST messages again
4638 - if the client says that it doesn't have the lock, cancel it on the
4640 - if we timeout during I/O, don't try to cancel an in-use lock; instead
4641 mark it as destroyed, it will all work out when decref is called
4642 - fix module use counts (22, 581)
4644 - ASTs now expect a reply (server cancels lock on error reply)
4646 2002-12-02 Andreas Dilger <adilger@clusterfs.com>
4649 - fix many simultaneous client startup (392)
4650 - fix dentry->d_it clobbering
4651 - credentials weren't being shipped for readdir/getattr operations
4652 - remove invalid assertions triggered during some concurrent MD
4654 - proper Lustre versions added (336, 389)
4655 - fix memory leak for create error case (398)
4656 - fix LOV locking bug that would get cli/srv out of sync
4657 - fix echo client over LOV (409)
4658 - fix dbench 2, extN refcount problem (170, 258, 356, 418)
4659 - fix double-O_EXCL intent crash (424)
4660 - avoid sending multiple lock CANCELs (352)
4662 - MDS can do multi-client recovery (modulo bugs in new code)
4664 - many updates, edits, cleanups
4666 2002-11-18 Phil Schwan <phil@clusterfs.com>
4669 - fix null d_it dereference (346)
4670 - fix full OST/dbench hang (333)
4671 - fix permission problem with file removal (286)
4672 - fix removal of OSCs from LOV when they fail
4673 - fix NULL deref during bulk timeout (214)
4674 - fix problems related to multiple filesystems on one MDS (241)
4675 - fixed serious subtle metadata locking bugs
4676 - free locks on clients when inodes are removed due to memory
4678 - fix inode pointer in lock data (285)
4679 - partial support for multiple MDS on a single host (241)
4680 - data locks weren't cancelled at clear_inode time (290, 311)
4681 - intent locks could lead to unbounded lock growth (205)
4682 - added a maximum lock count, an LRU list, and a flusher
4683 - fix multiple rename (365)
4684 - properly abstracted the echo client
4685 - OSC locked 1 byte too many; fixed
4686 - rewrote brw callback code:
4687 - fixed recovery bugs related to LOVs (306)
4688 - fixed too-many-pages-in-one-write crash (191)
4689 - fixed (again) crash in sync_io_timeout (214)
4690 - probably fixed callback-related race (385)
4692 - Add capability to MDS protocol
4693 - LDLM cancellations and callbacks on different portals
4695 2002-10-28 Andreas Dilger <adilger@clusterfs.com>
4698 - limit client IOV size to PTL_MD_MAX_IOV (611336, 191)
4699 - defer open object destruction to close time (601981, 138)
4700 - open/close OST file handle in obdo (OBD_MD_FLHANDLE) (601981, 138)
4701 - move LDLM_ENQUEUE/CONVERT back to MDS portal (625069)
4702 - abstract ll_lookup2, fix ll_revalidate2 to use abstraction (256)
4703 - don't call obd_setattr in ll_file_release for destroyed objects
4704 * protocol change to lustre_msg: move |version| and add |flags|
4705 * protocol change to osc_punch: "start" in "o_size", "end" in "o_blocks"
4706 * lock replay: for LDLM_FL_REPLAY trust client to do right thing
4707 * added replay of create, unlink, link and rename operations during
4708 MDS failover; recovery should be much more robust now
4709 * remove failed OSCs from LOVs (only lov_create uses this so far)
4710 * the lustre-HOWTO was brought (more) up to date (582544)
4712 2002-10-23 Phil Schwan <phil@clusterfs.com>
4715 - in-use dentries weren't being reused properly (617851)
4716 - prevent multiple LDLM setup (599178)
4717 - fix LOV size calculations for truncate (617853)
4718 - fix client handling of MDS intent errors (POSIX)
4719 - fix permission bug in lovstripe.c test (624321)
4720 - fix MDS thread deadlock - move LDLM handler to DLM portal (625069)
4721 - truncate past end of file could corrupt data
4722 - proper cleanup after timeouts, crashes, etc (592524, 550815)
4723 - a race in recovery could return ETIMEDOUT to apps (623947)
4724 - building outside the source directory was fixed
4725 * the lustre-HOWTO was brought (more) up to date (582544)
4726 * major progress was made on recovery functionality
4728 2002-10-10 Phil Schwan <phil@clusterfs.com>
4731 - recovery deadlock fix
4732 - rm -rf causes LBUG fix (617817)
4733 - file open by multiple tasks fix (618962)
4734 - directory permissions bugs (602707 and 620007)
4735 - journal_stop fixed with locking (611313)
4736 - O_APPEND failures resolved (618273, perhaps 614459)
4737 - lconf PATH fix (619770)
4738 - IA64 build fix (621450)
4739 - RPC buffer sizes scale with amount of memory
4741 2002-10-01 Phil Schwan <phil@clusterfs.com>
4744 - locks would be cancelled without throwing away data pages,
4745 resulting in inconsistent data (605627)
4746 - inode attributes were not always being refreshed (605627, 612449)
4747 - lconf now continues to cleanup after lctl reports an error
4748 - MDS now enforces user permissions (602707)
4749 - lprocfs cleanup fixed, but not yet enabled (614157)
4750 - fixed infinite server hang, should a client not respond to an AST
4751 - avoid going into recovery if user calls readlink() with a buffer
4752 that's too small (613941)
4753 - AST RPCs no longer require replies (614867) -- this may be changed
4754 - don't crash server if client sends an IOV that's too big (611336)
4755 - fixed lock conversion deadlock (611892)
4756 - fixed the following of symlinks (614622)
4757 * recovery: the server can remove locks from a client that dies, other
4758 clients can make progress
4759 * more extN patch fixes
4760 * compile-time configurable ptlrpc buffer allocations
4762 - collaborative read cache document
4763 - Lustre Lite Performance CDR document-in-progress
4765 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
4768 - fix typo in patch-2.4.18
4770 2002-09-20 Andreas Dilger <adilger@clusterfs.com>
4773 - clear ptlrpc request each time in handle_incoming_request()
4774 - unlink of files now destroys the object on the OST
4776 2002-09-19 Peter Braam <braam@clusterfs.com>
4778 * add hard link support
4779 * change obdfile creation method
4780 * kernel patch changed
4782 2002-09-19 Peter Braam <braam@clusterfs.com>
4785 - stack overflow bug in extN fixed
4787 2002-09-18 Andreas Dilger <adilger@clusterfs.com>
4789 * documentation updates
4790 - add man pages for config tools
4791 - update tests/README to describe testing with new config tools
4792 - finish metadata API descriptions
4793 * bug fixes and cleanups
4794 - statfs workaround for 16TB limit
4795 - LOV stripe allocation improved, can stripe on subset of OSTs
4796 - LOV file size/IO offset was wrong for files > 4GB in size
4797 - object EA data was being dropped, caused files to be unreadable
4798 - memory overflow with non-LOV OST caused memory corruption
4799 - fixed regression tests to work with new config tools, obdfilter
4800 - fixed bug when directory size became larger than 1 block
4801 - fixed bug (for single client case) when PWD was deleted
4802 - invalidate local directory pages when doing intent-based ops
4803 - avoid LDLM oops when lock callback contained bad data
4805 2002-09-09 Andreas Dilger <adilger@clusterfs.com>
4807 * documentation updates
4808 * bug fixes and cleanups
4809 - configuration tools
4812 - 64-bit compile warnings
4813 - 64-bit internal statfs data
4815 * test_brw on persistent OST devices
4817 * lprocfs (disabled)
4819 2002-09-04 Andreas Dilger <adilger@clusterfs.com>
4821 * documentation updates
4822 * bug fixes and cleanups
4823 * configuration tools
4825 2002-08-30 Peter J. Braam <braam@clusterfs.com>
4828 * many small fixes to 0_5_4
4829 * io/network handling
4830 * thinkos in MDS operations
4832 2002-08-24 Peter J. Braam <braam@clusterfs.com>
4835 * crucial basic fixes to 0.5.3
4836 * IOR, Iozone work over Elan
4839 2002-08-07 Phil Schwan <phil@clusterfs.com>
4840 * version 0_5_3, our first alpha
4841 * we use the new Portals iovs
4842 * documentation updates
4843 * bug fixes and cleanups
4844 * small changes in the DLM wire protocol
4846 2002-07-25 Peter J. Braam <braam@clusterfs.com>
4847 * version 0_5_1 with some initial stability,
4848 * locking on MD and file I/O.
4849 * documentation updates
4850 * several bug fixes since 0.5.0
4851 * small changes in wire protocol
4853 2002-07-18 Phil Schwan <phil@clusterfs.com>
4855 * delivered as Lustre Light Alpha
4856 * fixed a crash after handling invalid MDS requests
4857 * fixed directory pages for architectures with non-4k pages sizes
4859 2002-07-11 Andreas Dilger <adilger@clusterfs.com>
4860 * release version v0_4_4
4861 * Moves TCP acceptor to be on port 2432 (unused Coda port) instead
4863 * Fixes a number of interruption problems with OST operations.
4864 * Update documentation for portals header changes
4865 * Move all wire protocol structs/defines to lustre_idl.h
4866 * Fixes symlink length bug.
4867 * Add tcpdump to repository.
4869 2002-07-05 Andreas Dilger <adilger@clusterfs.com>
4870 * release version v0_4_3
4871 * Fixes statfs for inodes on extN.
4872 * Fixes bug in runtests which would delete /etc/hosts.
4873 * Use 64-bit object IDs wherever possible (not into VFS though)
4874 Remove ost_get_info, which is unused by lustre, and out of date.
4876 2002-07-03 Peter Braam <braam@clusterfs.com>
4877 * release version v0_4_2 Fixes a lookup error (type not passed)
4878 * move forward to head of Portals
4879 * move forward to latest Lustre kernel
4881 2002-06-25 Peter Braam <braam@clusterfs.com>
4882 * release version v0_4_1. Hopefully stable on single node use.
git://git.whamcloud.com / fs / lustre-release.git / blob